CONTRIBUTORS
Kevin M. Carlsmith (193) Department of Psychology, Colgate University, Hamilton, NY 13346 Serena Chen (151...
75 downloads
1114 Views
3MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
CONTRIBUTORS
Kevin M. Carlsmith (193) Department of Psychology, Colgate University, Hamilton, NY 13346 Serena Chen (151) Department of Psychology, 3210 Tolman Hall, University of California, Berkeley USA 94720–1650 Amy J. C. Cuddy (61) Kellogg School of Management, Northwestern University, Evanston, IL 60208 John M. Darley (193) Department of Psychology, Princeton University, Princeton, NJ 08540 Susan T. Fiske (61) Department of Psychology, Princeton University, Princeton, NJ 08540 Peter Glick (61) Department of Psychology, Lawrence University, Appleton, WI 54912 Miles Hewstone (237) Department of Experimental Psychology, Oxford University, Oxford, OX1 3UD, United Kingdom John G. Holmes (1) Department of Psychology, University of Waterloo, Waterloo, Ontario, N2L 3G1, Canada Dacher Keltner (151) Department of Psychology, 3210 Tolman Hall, University of California, Berkeley USA 94720–1650 Michael W. Kraus (151) Department of Psychology, 3210 Tolman Hall, University of California, Berkeley USA 94720–1650 Robin Martin (237) Work and Organisational Psychology Group, Aston Business School, Aston University, Birmingham, B4 7ET, United Kingdom
vii
viii
Contributors
Sandra L. Murray (1) Department of Psychology, University of Buffalo, State University of New York, New York 14260–4110 Gerben A. Van Kleef (151) Department of Social Psychology, University of Amsterdam, Roetersstraat 15, 1018 WB Amsterdam, The Netherlands
C H A P T E R
O N E
The Commitment-Insurance System: Self-Esteem and the Regulation of Connection in Close Relationships Sandra L. Murray* and John G. Holmes† Contents 1. The Commitment-Insurance System 1.1. The ‘‘If–then’’ contingencies of exchange 1.2. If partner committed, then pursue connectedness 1.3. Revisiting the sociometer 2. The Commitment-Insurance System: The Empirical Progression 2.1. If equal to partner, then partner is committed 2.2. The alarm and repair contingencies 2.3. If partner committed, then pursue connectedness 3. Theoretical Innovations, Applications, and New Directions 3.1. The historical context 3.2. Dyadic strength: A motivational perspective on interdependence theory 3.3. Judging equivalence: What counts? 3.4. The duality of interdependent life 3.5. Taking levels of processing perspectives one step further 4. Conclusion Acknowledgments References
3 4 9 11 11 12 21 33 38 38 40 46 48 54 55 55 55
Abstract A levels of processing model of the commitment-insurance system is described to explain how low and high self-esteem people cope with the interdependence dilemma posed by ensuring that a partner’s commitment is commensurate with their own. Two levels to this system are detailed: (1) the procedural rules that govern perception and behavior without conscious awareness and regardless of self-esteem and (2) the controlled or deliberated rules that govern perception and behavior differentially as a function of self-esteem. In the automatic rule
* {
Department of Psychology, University at Buffalo, State University of New York, New York 14260-4110 Department of Psychology, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
Advances in Experimental Social Psychology, Volume 40 ISSN 0065-2601, DOI: 10.1016/S0065-2601(07)00001-9
#
2008 Elsevier Inc. All rights reserved.
1
2
Sandra L. Murray and John G. Holmes
system, feeling as valuable as the partner signals the partner’s intrinsic motivation to be responsive, and thus the safety of seeking connection. In contrast, feeling inferior activates an exchange script specifying that partners need to match in desirability to avoid being replaced. Once activated, exchange concerns then motivate reparative behavioral efforts to secure the partner’s dependence, thereby soliciting a supplementary, instrumental form of commitment-insurance against rejection and nonresponsiveness to need. In the controlled rule system, optimistic chronic expectations of acceptance allow high, but now low, selfesteem people to override the application of these automatic contingencies. Research supporting the model is reviewed and provides the basis for delineating a novel formulation of interdependence theory. It’s a discrimination that’s widespread, but largely unspoken, causing pain and stress to the affected couples, who often find it hard to talk about, even to each other. I’m talking, of course, about marrying outside your looks. Marrying a few degrees up or down the hotness scale. Refusing to stay within your cute-gory.
Though tongue-in-cheek, Belinda Luscombe’s comments in an essay for Time Magazine (May 7, 2007, p. 86) illustrate a ubiquitous principle of social life—the need to perceive a matched or equivalent exchange (Thibaut and Kelley, 1959). The implicit assumption that partners need to be ‘‘matched’’ in terms of basic value is so engrained that violations of this equivalence principle prompt a search for explanation. After all, who has not implicitly assumed that an unattractive man paired with a beautiful woman is either rich, powerful, or both? This chapter examines how the motivation to perceive an equivalent exchange affects the internal dynamics of close romantic relationships. We argue that four implicit and interconnected ‘‘if–then’’ exchange contingencies solve a basic problem of interdependence—ensuring that a partner’s commitment is commensurate with one’s own (Kelley, 1979). To avoid costly rejection experiences, people rely on a dynamic and dyadic barometer that estimates their worth relative to the partner. In this metric, feeling as valuable as the partner functions as an appetitive or approach cue. It provides the primary basis for trusting in the partner’s intrinsic commitment to respond to one’s needs (i.e., ‘‘if equal, then committed’’). Such confidence in a partner’s commitment then signals the safety of seeking connection oneself (i.e., ‘‘if committed, then seek connection’’). In contrast, feeling inferior to the partner functions as an aversive or avoidance cue. Specifically, feeling inferior activates the exchange script (i.e., ‘‘if inferior, then exchange concerns’’). This script specifies that partners need to match in desirability to avoid being replaced by a more deserving alternative (White, 1980). Once activated, exchange concerns then motivate reparative behavioral efforts to secure the partner’s dependence (i.e., ‘‘if exchange concerns, then promote dependence’’). Through efforts as varied as finding lost keys, organizing schedules, and limiting outside friends, people make
The Commitment-Insurance System
3
themselves indispensable, and secure a supplementary, instrumental form of commitment-insurance against rejection and nonresponsiveness to needs. In the first section of this chapter, we outline the normative operation of the commitment-insurance system and describe how its sensitivity shifts for people who differ in chronic concerns about rejection—specifically, people low and high in self-esteem (Leary et al., 1995). In the second section of this chapter, we draw on daily diary, experimental, cross-sectional, and longitudinal studies from our ongoing research programs to illustrate the operation of this system. In the final section of this chapter, we integrate the current model with classic perspectives on interdependence to present a new look on interdependence theory, detail the applications of the commitment-insurance model, and point to directions for future research.
1. The Commitment-Insurance System Situations of interdependence are fundamental to romantic life. One partner’s actions constrain the other partner’s capacity to meet important needs and goals (Kelley, 1979). Compounding this vulnerability, partners’ interests and goals typically diverge in at least some important respects. After all, people rarely even choose a partner whose basic personality dimensions merge easily with their own (Lykken and Tellegen, 1993). Interdependence thus necessitates risk: people need to depend on partners who may not always be responsive to their needs and might one day decide to dissolve the relationship (Murray et al., 2006b). The risk of nonresponsiveness inherent in interdependence puts people’s basic goal to seek connection to others in conflict with the need to protect against rejection (Murray et al., 2006b, 2008). In soliciting Harry’s support after a difficult day at work, Sally can satisfy connectedness goals, but in so doing, she also risks his criticism or rebuff, compromising selfprotection goals. To function effectively, people need regulatory systems in place to solve the goal conflicts interdependence poses (Murray et al., 2006b). Within such a system, Sally needs a flexible means of ensuring that seeking Harry’s support, and more generally allowing herself to feel connected to him, is sufficiently safe. In particular, people need a system in place to gauge and ensure their partner’s commitment, and thus minimize the chance of experiencing rejection or nonresponsiveness to needs. This regulatory system must track whether a partner is motivated to respond to one’s needs when times are difficult and elicit compensatory behaviors to secure responsiveness when it is in question. Attachment and evolutionary theorists assume that securing a partner who is motivated to care for one’s needs is the sine qua non of satisfying interpersonal relationships. Perceiving responsiveness is thought to be so critical because partner
4
Sandra L. Murray and John G. Holmes
responsiveness minimizes the likelihood of rejections that could ultimately threaten survival and the transmission of one’s genes (Bowlby, 1969; Gilbert, 2005; Reis et al., 2004; Tooby and Cosmides, 1996). Accordingly, evolutionary theorists argue that specialized cognitive mechanisms evolved to help people to detect which intimates perceive qualities in them that could not be obtained in alternate relationships—thereby ensuring loyalty in costly circumstances (Tooby and Cosmides, 1996).
1.1. The ‘‘If–then’’ contingencies of exchange The commitment-insurance system depicted in Fig. 1.1 assumes that people gauge the status of their partner’s current commitment by monitoring whether their partner is intrinsically motivated to care for them. It assumes that attributions of intrinsic motivation afford the primary basis for reasoning that a partner will tend to one’s needs come what may. Consistent with this logic, people report greater trust in their partner’s continuing love when they more strongly believe that their partner sees them as intrinsically valuable (Rempel et al., 1985). In fact, priming extrinsic factors, such as gains in social status, that could explain their partner’s commitment actually makes people feel less loved (Seligman et al., 1980)! It seems that people implicitly recognize that such benefits could easily be provided by others, and thus cannot provide a secure basis for guaranteeing a partner’s continued responsiveness. The model stipulates that people gauge partner motivations by relying on a dynamic and dyadic barometer—one that assesses their worth relative to the partner. In this metric, seeing oneself as just as good a person as the partner provides a critical platform for reasoning that the partner is likely to
Equal to partner?
Yes
A
G
Perceived commitment
X
Connectedness goals H
Chronic perceived regard No E
Alarm contingencies
B
C
F Chronic perceived regard
X Exchange concerns Repair contingency
D
F
Promote partner dependence
Figure 1.1
The commitment-insurance system in close relationships.
The Commitment-Insurance System
5
see one’s own qualities as intrinsically valuable (Path A in Fig. 1.1). Thus, to feel cherished, and essentially irreplaceable, Harry needs to believe that he brings as many desirable qualities to the relationship as Sally does (Murray et al., 2005). He also needs to believe that Sally shares this appraisal of his worth (Derrick and Murray, 2007). The perception that he possesses such valued qualities gives Harry reason to believe that Sally wants to care for him because he is special to her—an attribution of intrinsic motivation that means her commitment is not likely to waver with costly circumstances. Why would feeling equal to their partner have this effect on people’s motivational assessments? First, people model images of an ideal partner after their own self-image (Murray et al., 1996a). Given people’s assumption that others see the world just as they do (Griffin and Ross, 1991), they may believe that other people, including their partner, also hope to find their own personally cherished qualities in an ideal partner. Thus, feeling equal to Sally gives Harry reason to believe that Sally wants to care for him because he meets her ideals. Second, people may assume that their partner is less likely to notice (or even have) alternatives when their partner (and alternative suitors) perceives them as worthy of their partner’s affections. Thus, feeling equal allows Harry to believe that Sally’s head will not be turned by potential overtures of attractive others because he ‘‘knows’’ that he deserves her affection (Simpson, 1987). 1.1.1. Seeking supplemental insurance: The alarm and repair contingencies The alarm (Path B) and repair contingencies (Path D) depicted in Fig. 1.1 function to pre-empt the problems of nonresponsiveness to need that could arise if a partner’s intrinsic motivation was to fade. Together, these implicit ‘‘if–then’’ procedural rules serve a preventive function by ensuring that Harry at least needs to care for Sally because he cannot afford to let his commitment waver—a supplementary and instrumental form of commitment-insurance that trades feeling cherished or irreplaceable for feeling indispensable. The costs of rejection are sufficiently serious that the commitment-insurance model assumes that these contingencies are an implicit and normative feature of people’s relationship representations (Bowlby, 1969; Gilbert, 2005; Tooby and Cosmides, 1996)—automatically regulating behavior without awareness or intention for low and high self-esteem people alike (Baldwin, 1992). How might this occur? Occasions where one partner feels inferior to the other are inevitable. Such occasions might be triggered by a specific event, such as one partner receiving a major promotion when the other partner is experiencing work difficulties. They might also be triggered by specific circumstances, such as managing a challenging child or negotiating a major purchase, that reveal one partner’s greater aptitude. Some people, such as those low in self-esteem, are also more likely to feel inferior to their partner
6
Sandra L. Murray and John G. Holmes
regardless of the circumstance because they chronically doubt the worth of their own qualities (Murray et al., 2005). Regardless of its source, felt inferiority undermines people’s perception that their partner perceives uniquely valuable qualities in them—compromising a primary source of commitment-insurance. Drops in the perception of equivalence automatically trigger the alarm contingency (Path B in Fig. 1.1). In this metric, feeling inferior activates the exchange script. This script contains the implicit knowledge that partner qualities need to match in terms of their desirability for one partner to merit or deserve the other. Failing such a match, this script further stipulates that a more deserving partner might be traded for oneself (Thibaut and Kelley, 1959). Comparisons of worth may often center around obvious social commodities, such as attractiveness, intelligence, and popularity (Buss and Barnes, 1986; Rubin, 1973). Qualities with less immediately obvious currency, such as warmth or patience, may also factor into such desirability calculations if people are truly consumed by issues of balance (Gangestad and Simpson, 2000). The model further assumes that the ‘‘if–then’’ contingency between inferiority and exchange is so fundamental that even unconsciously activating the exchange script triggers reciprocal concerns about inferiority and one’s deservingness and substitutability (Path C in Fig. 1.1). Such anxieties are basic to the exchange script because ways in which one falls short of a partner’s standards are likely to surface for everyone once a comparative accounting process is activated. Once activated, the exchange script (and its ensuing concerns about balance) then trigger the repair contingency. This compensatory contingency functions to secure a partner’s responsiveness and commitment by prompting behaviors meant to increase the partner’s dependence on the relationship (Path D in Fig. 1.1). By eliciting greater partner need or dependence, people equalize the perceived exchange in ways that increases confidence in their partner’s commitment (Path E in Fig. 1.1). Imagine that Sally, like the author of the Time Magazine article, assumes that her husband Harry falls outside her ‘‘cute-gory.’’ Feeling diminished by Harry’s greater physical appeal, Sally is unlikely to believe that Harry finds her irreplaceable in this regard because he could readily find a more attractive partner (given the abundance of such fish in the sea). Consequently, she needs to find additional or supplementary ways to make herself indispensable to Harry. As one means of making herself needed or indispensable, she might try to specialize their roles in ways that increase Harry’s dependence on her. For instance, she might strive to be highly proficient in those domains, such as cooking or financial management, where Harry is weaker. She might also take disproportionate responsibility for instrumental or practical matters. For instance, she might manage Harry’s schedule or coordinate his relationships with his family, making herself indispensable to his successful completion of his daily activities. She might further secure Harry’s responsiveness and
The Commitment-Insurance System
7
commitment by making sure that all of Harry’s friends come to like her, thereby narrowing his social network. Through such (automatic) behavioral reactions, Sally can better ensure that Harry needs to be responsive to her needs in even those situations where he might not want to do so precisely because he cannot afford to lose her.1 The idea that people may sometimes need additional reason to believe that their partner is committed and motivated to respond to their needs resonates with cross-cultural and interdependence analyses. In interdependent cultures, believing that a partner perceives qualities worth valuing in the self is necessary, but not sufficient, to instill confidence in the partner’s acceptance. Instead, people in such cultures need to believe that the partner’s family also values the qualities one brings to the relationship (MacDonald and Jessica, 2006). Interdependence analyses similarly emphasize the importance of structural constraints on commitment (Levinger, 1976; Rusbult and Van Lange, 2003). In such analyses, a person’s commitment to the relationship is not just governed by the intrinsic factors that draw them toward it, such as attraction to a partner’s especially gregarious nature or unique wit. Instead, commitment is constrained by the instrumental or extrinsic factors, such as need satisfaction (Drigotas and Rusbult, 1992), the absence of viable alternatives (Rusbult et al., 1998), religious prohibitions, or family disapproval (Levinger, 1976) that make dissolving the relationship difficult. Our assumption that people are compelled to seek commitment-insurance in certain circumstances also helps explain a paradoxical real life phenomenon. Much to the chagrin of conservatives who warned no-fault divorce would destroy the nuclear family, the incidence of divorce has actually decreased since no-fault divorce laws were introduced (Matouschek and Rasul, in press). Though not using our terms, political scientists explain this phenomenon by using the logic that marriage provides a form of commitment-insurance. They reason that couples in more tenuous relationships should be hesitant to commit. However, the formal barrier to dissolution provided by the marriage contract gives such vulnerable people the extra insurance they need to commit. Without this supplemental form of insurance, vulnerable couples are less likely to marry in the first place, contributing to an overall decrease in divorce rates. 1.1.2. The moderating role of perceived regard: Self-esteem as an exemplar Although the alarm and repair contingencies operate seamlessly and automatically for low and high self-esteem people in most circumstances, certain situations provoke considered social comparisons to the partner and more deliberate or conscious consideration of exchange norms. When people 1
In the third section of this chapter, we present a typology that delineates these means of dependence-promotion into specific categories.
8
Sandra L. Murray and John G. Holmes
explicitly consider the exchange script, its consequences may diverge as a function of people’s global self-esteem. This assumption is captured by the moderating Path F in Fig. 1.1. Consistent with this logic, models of attitudes, impression formation, and stereotyping assume that automatically-activated associations control behavior unless people have the motivation, opportunity, and capacity to over-ride them (Fiske and Neuberg, 1990; Olson and Fazio, in press). Indeed, different implicit and explicit contingency rules govern low and high self-esteem people’s willingness to risk dependence on a romantic partner (Murray et al., 2008). Imagine that Harry has noticed that he has put on a lot of weight, whereas Sally’s waistline has benefited from several recent trips to the gym. As he contemplates finishing the last slice of pizza, he comes across a popular magazine article detailing how mismatched couples experience greater conflict and thoughts of straying. In such circumstances, Harry might more readily counter applying exchange contingencies to his own relationship if he has high rather than low self-esteem. Why would this be the case? People gauge their partner’s regard for them by assessing the worth of their own qualities (Murray et al., 1998, 2000). Because people with high self-esteem possess more positive and more certain beliefs about themselves than do people with low self-esteem (Campbell, 1990), they correctly assume that their partner sees many special traits worth valuing in them. In contrast, people with low self-esteem incorrectly assume that their partner sees few qualities in them worth valuing (Murray et al., 2000). Such differing appraisals foster similarly optimistic (or pessimistic) appraisals of the partner’s ongoing responsiveness and commitment (Murray et al., 2001). The moderating paths (Paths F in Fig. 1.1) linking perceived regard to the ‘‘if–then’’ exchange contingencies reflect the assumption that their strength (and likelihood of being suppressed) differs as a function of people’s chronic expectations about their partner’s regard. Feeling confident that their partner values them should largely inoculate high self-esteem people against making social comparisons that might activate doubts about their partner’s commitment in situations where they reflect on something bad about themselves or something good about their partner (Murray et al., 1998, 2005). However, if circumstances did somehow provoke explicit deliberations about inferiority and exchange, feeling uniquely valued should still provide high self-esteem people sufficient reason to trust in a partner’s ongoing commitment. In fact, they might suppress the tendency to apply the exchange script to their own relationship, thereby protecting a sense of faith in their partner’s intrinsic motivation (Rempel et al., 1985). Supporting this analysis, Leary and Baumeister (2000) argue that people with high self-esteem can override the automatic tendency to internalize rejection experiences. Similarly, people who are motivated to avoid prejudice can resist applying automatically activated stereotypes to specific group members (Kunda and Spencer, 2003). For people who are less confident of their partner’s responsiveness, such as people low in self-esteem, the commitment-insurance system is not likely
The Commitment-Insurance System
9
to be so easily suppressed. People with low self-esteem generally do not believe that their partner perceives qualities worth valuing in them (Murray et al., 2000). Unsure of their value to their partner, they should scrutinize all possible sources of social comparison information for what they might mean about the partner’s likely responsiveness. Failure at work, success at school, or praise from a friend thus have inflated meaning because of what such experiences might signal about relative comparisons to the partner. When exchange contingencies rise to the level of conscious deliberation for low self-esteem people, they should also exert just as powerful an influence. As low self-esteem people lack the primary form of insurance against risk, exchange considerations (whether implicit or explicit) may routinely and habitually prompt substitute forms of insurance-seeking. Consequently, they may react to even the salience of exchange by seeking to increase their partner’s dependence—because they do not believe they possess the kinds of qualities that could otherwise secure or bolster their partner’s commitment.
1.2. If partner committed, then pursue connectedness The link between perceived commitment and connection (Path G in Fig. 1.1) links the output of the system—the partner’s perceived commitment—to the activation of connectedness goals. This ‘‘if–then’’ contingency or procedural rule reflects the assumption that connection is safest, and most sought, when people are confident of the partner’s reciprocated affections and commitment (Murray et al., 2000, 2006b). By making connection contingent, the system functions to minimize the likelihood and pain of rejection or nonresponsiveness to need by ensuring that people do not allow their own sense of connection to the relationship to outstrip their partner’s level of commitment (Holmes, 1981, 1991). Consistent with this logic, considerable evidence suggests that felt acceptance is a relatively automatic cue to the safety of connection. Unconsciously primed thoughts of security heighten empathy for others (Mikulincer et al., 2001), diminish the tendency to derogate out-group members (Mikulincer and Shaver, 2001), and increase the desire to seek support from others in dealing with a personal crisis (Pierce and Lydon, 1998). In contrast, felt rejection automatically triggers the perception of risk and the desire to distance oneself from a relationship partner. MacDonald and Leary (2005) argued that rejection elicits a social pain akin to a physical pain to trigger increased distance between oneself and the source of the pain. For instance, people respond to ostracism from strangers by aggressing against those who ostracized them, which suggests that the need to distance oneself from painful interactions is a relatively basic and automatic one (Twenge et al., 2001).
10
Sandra L. Murray and John G. Holmes
1.2.1. The moderating role of self-esteem The link between perceived commitment and connectedness does not assume that all sources of confidence in a partner’s commitment are equal in promoting connectedness goals. Instead, confidence that is rooted in feeling cherished or irreplaceable (i.e., the primary form of commitmentinsurance) may have qualitatively different effects than confidence that stems from feeling indispensable (i.e., the secondary or supplemental form of commitment-insurance). Such differential consequences are captured by the moderating Path H illustrated in Fig. 1.1. Perceiving a partner as intrinsically motivated should afford a stronger and more stable level of conviction in a partner’s commitment than perceiving one’s partner as instrumentally motivated. If coordinating Sally’s schedule and cooking her meals gives Harry confidence that Sally needs to be with him, he might reasonably conclude such motivations would fade if he failed to keep up his (perceived) end of this exchange-bargain. Accordingly, feeling cherished or irreplaceable likely affords a resilience that allows people to approach their partner in situations of real conflict (Murray et al., 2003a) and actively promote the partner’s autonomy and independent activities (Feeney, 2007). However, feeling indispensable may offer an unstable and tenuous trust in partner’s commitment, one that motivates people to constantly gauge their partner’s acceptance and to seek out only those situations where the risk of rejection seems minimal (Murray et al., 2008). This logic suggests that low and high self-esteem may differ in how cautiously or intrepidly they pursue connectedness goals within their relationships. The moderating path linking chronic perceived regard to the strength of the ‘‘if partner committed, then connect’’ contingency captures this assumption (Path H in Fig. 1.1). This connection–regulation contingency guides people’s selection of situations within their relationships. It allows people to identify which specific situations should be avoided and which should be approached by signaling the likelihood of their partner’s responsiveness in each instance (Kelley et al., 2003; Murray et al., 2006b, 2008). Imagine that Sally has just criticized Harry for eating that pizza in front of her and undermining her own weight loss attempts. Such criticism signals rejection, automatically prompting Harry’s need to selfprotect and perhaps lash out at Sally in return (Murray et al., 2006b, 2008, Experiment 7; Twenge et al., 2001). By contrast, Sally’s praise signals acceptance, prompting Harry’s desire to seek connection. For people with high self-esteem, the faith in the partner’s intrinsic motivation afforded by feeling cherished and valued should offer the kind of resilience that effectively desensitizes, and may even reverse, this contingency. In situations where they feel acutely rejected, high self-esteem people should actually draw closer to their partner and seek still greater interdependence. For people with low self-esteem, however, feeling
The Commitment-Insurance System
11
needed rather than wanted should polarize such contingencies. For lows, the slightest of criticisms or the most minor of conflicts may signal that a partner has not recognized their many efforts—putting the partner’s need or dependence on the relationship in question. In situations where they feel acutely rejected, low self-esteem should distance, diminishing their partner’s value and avoiding those situations that might leave them vulnerable to their partner’s decisions and choices. In so doing, they can effectively dull the pain of the rejection they anticipate in advance (Murray et al., 1998).
1.3. Revisiting the sociometer Our model of how the commitment-insurance system implicitly regulates people’s pursuit of connectedness goals resonates with the sociometer model of self-esteem. Leary and his colleagues argue that state self-esteem functions as a ‘‘sociometer’’ that measures a person’s perceived likelihood of being accepted by others and regulates appropriate approach behavior (Leary and Baumeister, 2000; Leary et al., 1995). In this metric, painful drops in self-esteem signal the possibility of rejection and motivate people to direct approach behaviors toward potentially accepting others (MacDonald and Leary, 2005). Our model extends the sociometer model in three important respects. First, the sociometer is calibrated to gauge the acceptance of generic ‘‘others’’ who will be uniformly accepting or rejecting based on internal assessments of one’s own qualities. However, the exchange-alarm is sensitive to the inherently interdependent nature of any potential relationship bond. To gauge Sally’s acceptance, it is not enough for Harry to know that he possesses generally desirable qualities; he needs to know how he measures up against Sally’s views of herself. Second, our model stipulates the ‘‘if–then’’ contingencies that promote behaviors likely to secure a partner’s dependence in circumstances where their intrinsic motivation to be responsive might wane. Third, our model stipulates the contingencies that govern people’s willingness to risk different types of connection to their partner.
2. The Commitment-Insurance System: The Empirical Progression In this next section, we outline the empirical support for the operation of the commitment-insurance system. In so doing, we draw heavily on results from our research programs, but we also point to supporting evidence from other laboratories. We first review the evidence for the assumption that perceiving a matched or equivalent exchange supports optimistic inferences about a partner’s continued acceptance and commitment
12
Sandra L. Murray and John G. Holmes
(Path A in Fig. 1.1). We then review recent empirical data illustrating the implicit operation of the exchange and repair systems (Paths B, C, D, E, and F in Fig. 1.1). Next, we detail research linking perceptions of a partner’s acceptance and continued commitment to the pursuit of connectedness goals (Path G in Fig. 1.1). We conclude this section by examining how individual differences in self-esteem affect the relation between perceived commitment and connectedness goals (Path H in Fig. 1.1).
2.1. If equal to partner, then partner is committed Is perceiving a matched or equivalent exchange necessary for supporting optimistic and stable perceptions of a partner’s acceptance and commitment (Path A in Fig. 1.1)? If equivalence in worth signals likely partner responsiveness, at least three observations follow. First, people should implicitly set romantic aspirations in ways that ensure that they stay within their league— setting aside hopes of the most desirable partner for a retainable one. Second, once a relationship is established, people should think and behave in ways that protect desired perceptions of equivalence in worth. Third, situations that elevate people’s relative worth, such as receiving a promotion at work or being the object of another’s flirtatious glance, should heighten confidence in the partner’s acceptance and commitment. In contrast, situations that question their relative worth, such as getting a poor evaluation at work or being the recipient of a friend’s scorn or criticism, should decrease confidence in the partner’s acceptance and commitment. Such comparative effects should be most evident for people, such as those low in self-esteem, who have chronic difficulty believing that they are as good a person as their partner. We now examine the evidence for each of these observations in turn. 2.1.1. Observation one: Setting aspirations We first came upon evidence for the importance of the principle of ‘‘matching’’ or ‘‘equivalence’’ in structuring romantic aspirations quite serendipitously. At the time, we were immersed in a program of research on positive illusions (Murray et al., 1996a). In this research program, we asked a sample of dating couples and a sample of married couples to describe themselves, their current partner, and their ideal partner on a series of interpersonally oriented positive and negative qualities (e.g., warm, responsive, intelligent, critical, demanding). We then tracked changes in these perceptions over the course of a year in the dating sample (Murray et al., 1996b). This research program revealed that people’s hopes for an ideal partner did indeed shape their impressions of their own partner. Though it might seem obvious in retrospect, at the time, we doubted that ideals would have such an assimilative effect—precisely because we assumed that people uniformly would want the warmest and smartest and
The Commitment-Insurance System
13
least critical and demanding partner possible. Much to our surprise, that was not the case. Instead, people’s impressions of their own qualities or worth seriously constrained their aspirations (see also Campbell et al., 2001). People who expressed more doubt about their own status on these qualities expected less from their ideal partner. In contrast, people with more positive self-images on these traits expected significantly more in an ideal partner (Murray et al., 1996a). These ‘‘deservingness’’ effects also emerged in a longitudinal study of dating couples (Murray et al., 1996b). People who initially appraised themselves less positively expected relatively less of an ideal partner over time. The opposite was the case for people who made more positive assessments of their own worth. The hope-limiting effects of self-doubt suggest that people implicitly know that there is little point aspiring to a partner they do not deserve (and whose affections and commitment they would then have trouble sustaining). Consistent with this logic, people typically express more interest in pursuing relationships with dating partners whose level of physical attractiveness, and general social desirability, matches their own (Berscheid et al., 1971). The perception of similarity generally may be such a powerful trigger of interpersonal attraction (Berscheid and Reis, 1998) because it functions as a heuristic cue that signals which specific prospective partners one merits or deserves (and, therefore, might be more likely to obtain). 2.1.2. Observation two: Maintaining perceived matches Our research on positive illusions revealed a second, serendipitous effect. Namely, people’s views of their own worth constrained perceptions of their partner’s worth. It seems Sally would rather not believe that Harry is brilliant, hilarious, and gorgeous if she worries about her own intelligence, wit, and attractiveness. This projection effect has replicated in several independent samples of dating and marital couples. People who see more positive interpersonal qualities in themselves also attribute more positive qualities to their partner (Murray et al., 1996a, 2000). In contrast, people troubled by greater uncertainty about their own worth perceive their partner in a comparably less positive light. Couples in established dating and marital relationships also tend to believe their level of physical attractiveness merits or matches their partner’s (Feingold, 1988). Furthermore, such ‘‘deservingness’’ effects emerge longitudinally. When people in dating relationships doubt the worth of their own qualities, they come to see their partner less positively over the course of a year, perhaps reflecting motivated attempts to keep them within their grasp (Murray et al., 1996b). Subsequent research suggested that perceiving a ‘‘match’’ or equivalence is so important that dating and married intimates even exaggerate perceptions of similarity in domains that should have little bearing on one’s worth as a romantic partner (Murray et al., 2002a). In this study, we asked people to describe themselves and their partners in three different
14
Sandra L. Murray and John G. Holmes
domains—interpersonal qualities, values, and day-to-day feelings. Not surprisingly, these dating and married intimates overestimated the extent to which their partner possessed the same profile of qualities as they did. If Harry perceived himself as more intelligent than ambitious, and more lazy than critical, he overestimated the extent to which Sally perceived herself as possessing the same profile of traits. However, this presumption of similarity also extended to domains—daily emotional experiences and values—that had less obvious bearing for one’s interpersonal worth or currency. Of course, such assimilation effects might reflect people’s inability to correct for the biasing effects of their own self-views in making judgments (Kruger, 1999)—a cognitive bias, rather than a motivated need to perceive equivalence. However, people actually seem to go out of their way to structure their relationship experiences in ways that allow them to avoid potentially threatening comparisons. Satisfied dating and married intimates react to the potential threat posed by one partner outperforming the other by perceiving such imbalances as complementary (Beach et al., 1998, 2001). People are sufficiently concerned about being outperformed by close others that they actually provide less performance-enhancing information to close than distant others (Pemberton and Sedikides, 2001). Such motivated attempts to keep a partner within one’s league reflect realistic fears. In situations where such perceptions of balance are pre-empted, and Harry outperforms Sally in a domain that is important to Sally, but not Harry, communication patterns are most negative (O’Mahen et al., 2000). The real world pressure toward matching on social commodities is sufficiently powerful that imbalances in dating partners’ physical attractiveness even forecast relationship dissolution (White, 1980). 2.1.3. Observation three: Comparisons shift perceived commitment No matter how strong people’s motivation to perceive equivalence, reality is not always conducive to such desires (Kunda, 1990). Think again of our increasingly rotund Harry and his newly svelte wife Sally. Having just loosened his belt after finishing the last of the pizza, Harry might not be able to escape comparing his girth to Sally’s slim physique. How might such an upward social comparison affect Harry’s confidence that Sally still finds him attractive (and that she’s not eyeing the trainer at her gym with more than a passing or professional interest)? Over the past several years, we conducted a series of experiments to examine how such upward (and downward) social comparisons affect inferences about a partner’s long-term caring and commitment (Derrick and Murray, 2007; Murray et al., 1998, 2002b, 2005). Most perspectives on the perception of partner responsiveness assume that focusing on a partner’s positive qualities would foster the greatest optimism (e.g., Mikulincer and Shaver, 2003). Our dyadic perspective on the nature of the ‘‘commitment-gauge’’ makes the opposite prediction for people, such as those low
15
The Commitment-Insurance System
in self-esteem, who often feel inferior to their partner. In such circumstances, seeing a partner more negatively should actually instill greater optimism than seeing them more positively! We first found preliminary evidence of such an ironic effect in a sample of dating couples (Murray et al., 2005). To create an index of people’s ongoing concerns about inferiority, we compared people’s ratings of themselves on the interpersonal qualities scale (e.g., warm, responsive, intelligent, critical, moody) to their ratings of their partner. This allowed us to create two groups of people—those who rated their partner more positively than themselves (i.e., the inferior group) and those who rated themselves at least as positively as their partner (i.e., the equal group). We then looked at the association between perceptions of the partner’s interpersonal qualities and inferences about the partner’s acceptance and commitment. Figure 1.2 presents the strength of this association for women in the inferior and equal groups. (We found parallel effects for men.) It compellingly illustrates how the perception of equivalence controls inferences about a partner’s commitment in these long-standing dating relationships: the tendency to link more desirable qualities to greater responsiveness and commitment was significantly weaker for people who felt inferior. Specifically, for people who felt equal to their partner, perceiving the partner’s traits more positively strongly predicted greater confidence of the partner’s ongoing love and commitment. However, this inference was not drawn as readily for people who felt inferior, suggesting that the commitment-gauge does indeed rely on a dyadic metric.
Perceptions of the partner’s love
9 Inferior footing Equal/better footing 8
7
6
−1
−0.5
0 0.5 Women’s perceptions of partner
1
1.5
Figure 1.2 The relation between perceptions of the partner and perceived love as a function of feeling on inferior or equal/better footing. Source: Murray et al. (2005). Reprinted with permission, American Psychological Assoiation.
16
Sandra L. Murray and John G. Holmes
2.1.3.1. Manipulating the comparison gap A series of experiments where we manipulated people’s relative standing to their partner and then measured confidence in the partner’s long-term acceptance and commitment further supported this conclusion. We expected relative comparisons to the partner to control inferences about caring and commitment and provide causal evidence for the operation of the commitment-gauge (Path A in Fig. 1.1). Experiences that accentuate the gap should decrease confidence in the partner’s acceptance and commitment; experiences that close the comparison gap should increase it. Such effects should be most evident for people low in self-esteem because they are engaged in an active and ongoing assessment of their partner’s likely responsiveness and commitment (Murray et al., 2001). Table 1.1 summarizes the general strategies we employed to increase and decrease the comparison gap. In our first set of experiments, we attempted to increase the comparison gap between the self and the partner. We did this in two ways: we either led people to perceive a new fault in themselves (Murray et al., 1998) or we led people to believe their partner perceived such a fault (Murray et al., 2002b). In two of the ‘‘own fault’’ experiments (Murray et al., 1998), we highlighted an old fault by asking experimental participants to describe a time when they had transgressed or let their partner down. In a further experiment, we introduced a new fault by leading people to believe that they were not treating their partner especially considerately. We did this by having experimental participants describe their behavior on an inventory that disproportionately sampled minor instances of inconsiderateness (e.g., ‘‘How often do you feel a little bit impatient with your partner?’’). In a final experiment, we had experimental participants complete a difficult, if not impossible, version of a purported intelligence test. In all the experiments, we then measured perceptions of the partner’s acceptance and love and projections for the partner’s continued commitment to the relationship (e.g., ‘‘I couldn’t do
Table 1.1 Manipulating the comparison gap Increasing the Comparison Gap
Decreasing the Comparison Gap
Seeing new fault in self
Seeing new virtue in self
Seeing new fault in partner
High demand partner
Behaving inconsiderately Conflict impaired
Partner seeing new fault in self
Recounting Will discover past sins secret self Failing conside- Unexpressed rateness test complaint Failing Laundry list of intelligence test criticisms
The Commitment-Insurance System
17
anything that would make my partner think less of me’’; ‘‘I am confident that my partner will always want to look beyond my faults and see the best in me’’). As our dyadic perspective on felt acceptance would anticipate, those people who already felt on inferior footing to their partner—people low in self-esteem—reacted to self-doubt by doubting their partner’s long-term acceptance and commitment. The deleterious effect that contemplated faults had on commitment-appraisals was just as evident when the fault had little obvious bearing on the relationship (i.e., lack of integrative ability) as when it had obvious bearing (i.e., lack of considerateness)! Consider the results from the intellectual failure experiment as an example (Experiment 4, Murray et al., 1998). When led to doubt their own intelligence, low selfesteem participants questioned their partner’s long-term acceptance and commitment (relative to controls). In contrast, people high in self-esteem reacted to the same self-doubts by embellishing their partner’s acceptance and love. The latter result suggests that global feelings of self-worth can indeed function as a resource or buffer against felt inferiority. We observed similarly compelling evidence for the operation of a commitment-gauge in a second series of experiments where we gave people direct reason to believe that their partner might perceive them as inferior (Murray et al., 2002b). In Experiment 1, we asked experimental participants to focus on those dark, seamy, or secret sides of themselves they would rather their partner not see. We then told them their partner was likely to find them out. In Experiment 2, we led experimental participants to believe their partner had a complaint about their behavior or personality they had not yet expressed. In both experiments, low self-esteem participants reported greater global doubts about their partner’s ongoing commitment than did control participants. However, high self-esteem participants did not react to these relatively future-oriented threats by doubting their partner’s acceptance and commitment. High self-esteem people are not unassailable, though. In a third experiment, we upped the psychological ante by creating an immediate threat. We led experimental participants to believe that their partner (who was physically present) had a long list of complaints about their personality. To do this, we seated the couple back to back in the experimental room. We then gave the experimental participant a one-page questionnaire that asked them to list important qualities in their partner that they disliked. The instructions clearly stated that they only needed to list one fault if that was all that quickly came to mind. The experimental participants believed that their partner received identical instructions. However, the partner actually received a one-page questionnaire that asked them to list at least 25 items occupying the contents of their bedroom or apartment. So experimental participants quickly listed one quality they disliked in their partner and then they spent several long, anxious minutes listening to their partner write, and write, and
18
Sandra L. Murray and John G. Holmes
write. Believing that their partner had a laundry list of complaints had a significant and unsettling effect on people’s confidence in their partner’s acceptance and commitment regardless of self-esteem. Thus, when the circumstances are right, even high self-esteem people can be vulnerable to the pernicious effects of feeling on uneven footing. This suggests that the ‘‘if–then’’ contingency linking perceptions of equivalence to confidence in a partner’s acceptance and commitment is indeed a normative one. In our third series of experiments, we attempted to decrease or close the comparison gap between the self and partner. In Experiment 1, we did this by giving people in the experimental condition reason to feel better about the merit of their own qualities. We led experimental participants to believe that they possessed the kind of trait profile that was in high demand by romantic partners (Murray et al., 2005). We then measured perceptions of the partner’s acceptance and commitment. As we expected, knowledge of this new virtue increased state self-esteem for low self-esteem people, effectively bringing their partner more within their grasp. In turn, this new appreciation of their own relational strengths increased low self-esteem people’s confidence in their partner’s long-term responsiveness and commitment (relative to control participants). In contrast, this manipulation had no effect on high self-esteem people, presumably because they already perceived themselves as equal to their partner.2 In our next two experiments, we attempted to close the comparison gap by humanizing the partner—that is, by pointing out a serious, but not fatal, flaw. In the first experiment, we led participants to believe their partner was not behaving considerately. We created this perception by having experimental participants complete a biased partner behavior inventory. Items on this inventory focused on minor, but typically frequent, inconsiderate partner behaviors (e.g., ‘‘Have you ever had to say the same thing twice to your partner because he/she wasn’t listening the first time?’’; ‘‘Have you ever felt like your partner was being selfish?’’). As we expected they would, experimental participants readily agreed that such events had happened. We then reinforced this message by informing them their partner was not behaving particularly considerately. Control participants completed a neutral inventory. We then measured confidence in the partner’s long-term acceptance and commitment using a combination of measures that tapped
2
We tried, fruitlessly, in our earlier research (Murray et al., 1998) to increase low self-esteem people’s confidence in their partner’s acceptance by leading them to believe that they were either exceptionally considerate or exceptionally intelligent. Much to our chagrin, these intended affirmations actually undermined low selfesteem people’s confidence in their partner’s acceptance. With the benefits of hindsight, we realized that these effects might have emerged because activating thoughts of such virtues also activated the contingency that the partner’s acceptance may depend on their continued possession of a variety of such qualities, a standard that lows may feel incapable of meeting (Baldwin and Sinclair, 1996). We took this message to heart in designing the ‘‘high demand partner’’ feedback by making the acceptance of others an explicit part of the self-esteem enhancing message (see also Rudich and Vallacher, 1999).
19
The Commitment-Insurance System
Confidence in the partner’s commitment
overall perceptions of the partner’s commitment (e.g., ‘‘I am confident my partner will always want to stay in our relationship’’), perceptions of the partner’s closeness (e.g., ‘‘My partner feels extremely attached to me’’), and predictions for the partner’s future commitment-relevant behaviors (e.g., ‘‘My partner will want more independence and pull away from me,’’ reverse-scored). Figure 1.3 presents the results for this composite dependent measure. As expected, low, but not high, self-esteem participants actually reported greater confidence in their partner’s commitment when their partner was behaving badly (Murray et al., 2005; Experiment 2)! In a further experiment, we led experimental participants to believe that their partner was at fault for a past conflict. We first asked all participants to recount such an incident. We then had experimental participants complete a scale designed to bias their memory of this incident in ways that heightened their partner’s culpability (e.g., ‘‘My partner cut me off during the conflict, not always allowing me to speak when I wanted to speak,’’ ‘‘My partner raised his/her voice with me, speaking forcefully and impatiently’’). Participants responded to these items on 5-point scales, anchored so as to elicit reasonably high scores (1 ¼ not true, 5 ¼ true). Then they received computerized feedback stating that their partner had behaved more negatively during that conflict than most partners behave. Control participants completed a neutral scale. We then administered an implicit measure of felt acceptance to determine whether finding fault in the partner automatically increases felt acceptance (as we assumed). In this task, participants discriminated between person (e.g., ‘‘lovable’’) and object words (e.g., ‘‘truck’’), indicating whether each word that appeared could ever be used to describe them (Dodgson and Wood, 1998). The target words of interest involved positive
0.25
−0.25
Badly behaved partner Control −0.75 Low self-esteem
High self-esteem
Figure 1.3 Confidence in a partner’s commitment as a function of global self-esteem and a badly behaving partner.
20
Sandra L. Murray and John G. Holmes
Reaction times to implicit acceptance words
trait terms tapping perceptions of the self as lovable (e.g., ‘‘appealing,’’ ‘‘lovable,’’ ‘‘kind,’’ ‘‘compassionate,’’ ‘‘warm’’) and perceptions that others care and accept the self (e.g., ‘‘connected,’’ ‘‘cherished,’’ ‘‘accepted,’’ ‘‘loved,’’ ‘‘supported’’). Figure 1.4 illustrates the self-esteem by condition interaction we obtained. When primed to see a significant fault in their partner, low, but not high, self-esteem participants were actually quicker to associate acceptance words with the self. This suggests that the commitment-gauge is so sensitive to relative comparison gains for people low in self-esteem that it even operates automatically (Murray et al., 2005, Experiment 3). We recently replicated the effects of shifting the comparison gap among people insecure in attachment style (Derrick and Murray, 2007) and in the day-to-day lives of married couples (Murray et al., 2006a). Specifically, people high in attachment anxiety report greater confidence in their partner’s acceptance when they think about domains where they exceed or outperform the standard set by their partner (Derrick and Murray, 2007, Experiment (1)). Even people who typically avoid closeness report greater confidence in their partner’s acceptance and commitment when they are led to believe their partner looks up to them in several respects (Derrick and Murray, 2007, Experiment (2)). In a daily diary study, low self-esteem people based their daily assessment of their partner’s caring and commitment on their own performance at work that day (and the comparison to the partner it presumably induced). On days when they closed the comparison gap (because they were more successful at work), low self-esteem people perceived their spouse as more accepting and loving.
950 940
Conflict-impaired partner
930
Control
920 910 900 890 880 870 860 850 840 830 Low self-esteem
High self-esteem
Figure 1.4 Implicit acceptance associations as a function of global self-esteem and a conflict-impaired partner.
The Commitment-Insurance System
21
However, low self-esteem married women perceived their husband as being more rejecting on days when the comparison gap widened (because these women had performed poorly at work). 2.1.4. Summary As the model of the commitment-insurance system stipulates, feeling equal to a partner signals a partner’s acceptance and ongoing commitment. When given reason to feel equal to their partner, low self-esteem people report greater confidence in their partner’s ongoing acceptance and commitment. When given reason to question their relative standing, low, and sometimes even high, self-esteem people report greater doubt about their partner’s commitment. Perceiving equivalence is so important that people dare not even hope to possess a partner whose qualities are more desirable than their own. They also structure their ongoing perceptions in ways that help maintain some degree of equivalence. What happens when such motivated efforts fail, as they often do, and feeling inferior prompts concerns that a partner’s commitment might fade?
2.2. The alarm and repair contingencies The alarm and repair systems exist to deal with precisely this set of circumstances. In this metric, drops in the perception of equivalence activate the exchange alarm (Path B in Fig. 1.1)—an exchange script that specifies that partners need to match in desirability and that failing such a match another partner might take one’s place (Thibaut and Kelley, 1959). The ‘‘if–then’’ contingency linking felt inferiority and exchange is so powerful that merely activating the exchange script automatically can trigger concerns about inferiority and substitutability (Path C in Fig. 1.1). Once activated, exchange concerns trigger the repair system (Path D in Fig. 1.1). This contingency prompts behavioral efforts to secure a partner’s dependence, thereby securing at least the partner’s instrumental motivation to be responsive to one’s needs. 2.2.1. Even newlyweds need insurance In providing our first test of the alarm and repair systems, we set out to conduct the most conservative test of our hypotheses we could envision (Murray et al., 2007). Newlyweds have little objective reason to question their partner’s responsiveness. Each partner has just made a dramatic public testament to their commitment—one that actually decreases rejection anxieties that people have carried with them throughout their adult lives (Davila et al., 1999). So, if there is any group that should not need commitmentinsurance, newlyweds are the ones. Nonetheless, we believed that exchange contingencies solve such a fundamental problem of interdependence that they would be activated automatically in even this circumstance.
22
Sandra L. Murray and John G. Holmes
How might we test such an assumption? The Murray lab is currently following a longitudinal sample of newlywed couples as part of a larger project. Each year both members of these newlywed couples complete standardized, electronic daily diaries for 14 days. The first wave of data collection (obtained between 2 and 6 months of marriage) allowed us to test whether feeling inferior did indeed promote behavioral efforts to increase the partner’s dependence. Each day participants rated how inferior they felt to their partner (e.g., ‘‘I’m not good enough for my partner,’’ ‘‘My partner is a better person than I am,’’ ‘‘My partner is more fun to be around than I am’’). They also indicated whether they had engaged in mundane acts, such as looking for a partner’s lost keys or glasses, packing a lunch, doing a partner’s chores, running his/her errands, or picking up his/her clothes, that should at least make one partner feel more indispensable to the other (Clark and Grote, 1998). We conducted multilevel analyses that examined whether Sally’s feelings of inferiority predicted her dependence-promoting behavior on subsequent days. To demonstrate that any effect of felt inferiority on dependencepromotion operated automatically, we estimated the strength of this contingency controlling for Sally’s self-reported feelings of inferiority on Tuesday. The effects of Monday’s felt inferiority thus capture the effects of largely unconscious doubts (i.e., doubts that Sally no longer realized that she was still experiencing Tuesday). Our analyses revealed striking evidence that even newlyweds do indeed need insurance. On days after people worried more about their status relative to their partner, they were more likely to behave in ways that should make them feel more indispensable, seeking a substitute form of commitment-insurance by searching for lost keys, making lunches, and picking up dirty dishes. This contingency was so powerful that it was just as evident for high as for low self-esteem people. It was also just as evident when we controlled for reports of other positive (e.g., giving compliments, being affectionate) and negative behaviors (e.g., being critical, yelling at the partner) that same day. This suggests that feeling inferior does not trigger a generic desire to be good (or avoid being bad). Rather, it selectively elicited the behaviors that should make one partner more needed or indispensable to the other. Amazingly, these newlyweds seemed to have an unconscious appreciation of Waller’s (1938) principle of least interest. This principle stipulates that the less invested partner holds the greatest power over the relationship. In the case of such an imbalance, the more invested (in this case more inferior) partner needs to try harder to win the superior partner’s affections (and equalize the power base). These newlyweds did just that—presumably without knowing exactly why they were doing it. But did such insurance-seeking work? Such efforts could make Sally feel more needed without having any effect on Harry’s actual commitment. Fortunately, the diary data also provided a means of testing the functional
The Commitment-Insurance System
23
consequences of such dependence-promotion efforts. Each day people indicated how much they doubted whether they had chosen the right partner, a telling index of their commitment. This allowed us to examine whether Sally’s dependence-promoting behaviors predicted Harry’s feelings of commitment on subsequent days. We were amazed to see that it did. Doing things as seemingly small as picking up dirty dishes and lost keys actually diminished the partner’s doubts about the marriage on subsequent days! Buoyed by the strength of these results, we conducted a series of seven experiments to examine the precise mechanisms underlying the operation of the alarm and repair systems (Murray et al., 2007). 2.2.2. If feel inferior, then exchange concerns We first took a closer look at the ‘‘if–then’’ contingency linking inferiority to the activation of the exchange alarm (Murray et al., 2007). Examining this contingency presented an immediate problem because people, especially those high in self-esteem, may resist making comparisons that could elicit feelings of inferiority. Accordingly, we induced such comparisons by manipulating the strength of people’s motivation to engage in social comparison and the standard (high or low) their partner’s qualities set. In two experiments (Murray et al., 2007, Experiments 2 and 3), we manipulated social comparison goals using a procedure developed by Stapel and Tesser (2001). Under the guise of completing an impressionistic personality test, our undergraduate participants first answered 15 questions about preferences and philosophy (e.g., ‘‘Spontaneity can be an excuse for irresponsibility,’’ ‘‘Many people feel uneasy when there is little work for them to do’’). Participants in the ‘‘self-uncertain’’ condition then learned that the computer program tabulating their responses was unable to compute a consistent profile of their personality. By inducing uncertainty, such feedback increases the need for social comparison information to define the self (Stapel and Blanton, 2004; Stapel and Tesser, 2001). So, in the present experiment, such feedback should increase people’s tendency to compare themselves to the partner, giving us a means to test our dyadic logic about the source of exchange concerns. Participants in the ‘‘self-certain’’ or control condition learned that the program had assembled a consistent personality profile. Participants in the ‘‘high partner standard’’ condition then described their partner’s two most positive qualities. Participants in the ‘‘low partner standard’’ condition described their partner’s two most negative qualities. We then measured people’s tendency to see their relationships in exchange or ‘‘even trade’’ terms. In one experiment, we utilized a global measure of anxiety about imbalance in the relationship. These items tapped anxiety or concern about meeting the level of contribution owed to the partner (e.g., ‘‘I feel indebted to my partner,’’ ‘‘I worry I don’t do as much for my partner as he/she does for me’’). In a second experiment, we utilized specific or situated measures of exchange concerns and communal goals
24
Sandra L. Murray and John G. Holmes
(i.e., the opposite of exchange concerns). To tap exchange concerns, we asked participants to imagine asking their partner to make certain sacrifices (e.g., going to class to take notes for them when they were sick). Then they rated how much this sacrifice would put them in their partner’s debt. To tap communal goals, we asked participants to respond to items tapping their willingness to meet their partner’s needs without expectation of reciprocation (Mills et al., 2004). We expected participants, especially those low in self-esteem, to worry more about exchange when social comparison goals and the partner’s positive traits were primed. In the first experiment, we found a significant social comparison goals by partner standards interaction that generalized across self-esteem levels. Figure 1.5 illustrates this interaction. When feeling uncertain about the self, and thus in greater need of social comparison information, people reported more anxiety about making an even relationship trade when they were focused on their partner’s two most positive (i.e., higher standard) than two most negative (i.e., lower standard) qualities. However, the partner’s most positive qualities did not induce exchange concerns when people were not motivated to engage in social comparison. These results again powerfully illustrate the importance of dyadic considerations. People’s social comparison goals completely changed the implications of their partner’s most positive qualities! In the face of social comparison goals, a partner’s positive qualities were not so positive after all. Instead, these qualities induce concerns about exchange! Nonetheless, the measure of exchange concerns in this experiment was global in nature. This left open the possibility that the effects might reflect the activation of general relationship insecurities rather than application of exchange contingencies to the relationship.
Exchange concerns
4
3.5
High partner standard Low partner standard
3
2.5
2 Low need for social comparison High need for social comparison (self uncertain condition) (self certain condition)
Figure 1.5 standards.
Exchange concerns as a function of social comparison goals and partner-
25
The Commitment-Insurance System
Exchange concerns
Fortunately, the situated measure of exchange concerns we utilized in our second experiment did not share this limitation. This experiment again revealed that dyadic comparisons control the activation of the exchange script. It also suggested that exchange concerns are more readily activated for low than high self-esteem people. Figures 1.6 and 1.7 present the significant three-way interactions we obtained for the measures of exchange concerns and communal goals, respectively. When primed to engage in social comparison, low self-esteem people expected partner sacrifices to put them more in their partner’s debt when they were focused on positive than negative qualities. The opposite effect emerged when they were not motivated to compare themselves to their partner. In contrast, high self-esteem participants reacted to being in a position of inferiority by denying exchange considerations applied to their relationships. When primed to engage in social comparison, high self-esteem people reported stronger communal goals when focused on positive than negative qualities. Impressively, the effects of upward social comparison in triggering exchange concerns was not a reflection of comparable shifts in people’s feelings about themselves in either experiment. The manipulations did not have any significant effects on mood or state self-esteem.
2.9
2.9
2.8
2.8
2.7
2.7
2.6
2.6
2.5
2.5
2.4
2.4
2.3
2.3
2.2
2.2
2.1
2.1 2
2 Low self-esteem High self-esteem
Low self-esteem High self-esteem
Low need for social comparison (self-certain condition)
High need for social comparison (self-uncertain condition)
High partner standard Low partner standard
High partner standard Low partner standard
Figure 1.6 Exchange concerns as a function of self-esteem, social comparison goals, and partner-standards.
26
Communal goals
Sandra L. Murray and John G. Holmes
8
8
7.8
7.8
7.6
7.6
7.4
7.4
7.2
7.2
7
7
6.8
6.8
6.6
6.6
6.4
6.4
6.2
6.2
6
6 Low self-esteem High self-esteem
Low self-esteem High self-esteem
Low need for social comparison (self-certain condition)
High need for social comparison (self-uncertain condition)
High partner standard Low partner standard
High partner standard Low partner standard
Figure 1.7 Communal goals as a function of self-esteem, social comparison goals, and partner-standards.
2.2.3. If exchange concerns, then promote dependence In our next two experiments (Murray et al., 2007), we examined the implicit operation of the repair system—the automatic ‘‘if–then’’ link between the activation of exchange and reparative behaviors intended to secure the partner’s dependence (Path D in Fig. 1.1). In the first experiment, we activated the exchange script incidentally by asking people to evaluate how appealing others would find purportedly real, personal ads (Murray et al., 2007, Experiment 4). In the experimental condition, these ads happened to emphasize the romantic hopefuls’ expectations of making an even or matched trade. In the control condition, these ads conveyed no such expectations. Thus, the experimental ads activated the exchange script incidental to the ‘‘real’’ purpose of the experiment— evaluating how appealing others would find the ads. As an example, one of the romantic hopefuls in the exchange ad condition conveyed high expectations for a match. This ad read: ‘‘Accomplished man seeking attractive woman for a meaningful relationship. I am looking for someone who is a good fit for me, and can live up to the qualities I bring to the relationship in her own way. My hobbies include sailing, tennis, and going out to restaurants.’’ Another hopeful, one presumably a little less enthralled with herself,
The Commitment-Insurance System
27
wrote: ‘‘Twenty-ish female seeking male for a serious relationship. My friends describe me as attractive in my own way, and as being a great conversationalist, intelligent, and as having a great sense of humor. I’m not interested in aiming too high, but I’m looking for a partner who would bring similarly desirable qualities to the relationship.’’ In a further experiment, we utilized a truly implicit exchange prime by superimposing different pictures on the computer screen that administered the questions. In the experimental condition, we superimposed pictures of US coins. The coins in question were mostly nickels. We expected these coins to activate the general metaphor of an economic exchange—the idea that quantities are bought and sold (Vohs et al., 2006). In the control condition, we superimposed pictures of circles on the computer screen (Murray et al., 2007, Experiment 5). In both experiments, we then measured felt inferiority to the partner (e.g., ‘‘My partner has a more interesting personality than I have,’’ ‘‘I feel inferior to my partner’’), concerns about alternative partners substituting for the self (e.g., ‘‘I worry about somebody taking my partner away from me’’), and behavioral efforts to increase the partner’s dependence. We employed multiple measures of dependence-promotion, including role specialization, life task management, narrowing social networks, and imposing barriers that might keep the partner committed. The measure of role specialization assessed people’s desire to specialize roles within their relationship in ways that heightened their general responsibility for their partner (e.g., ‘‘It bothers me when my partner does things for him/herself that I could do for him/her,’’ ‘‘I want my partner to feel like there are some things that he/she needs me to do for him/her’’). The life task management measure assessed how often people took specific responsibility for mundane, instrumental aspects of the partner’s life (e.g., ‘‘cooking for my partner,’’ ‘‘keeping track of my partner’s school schedule,’’ ‘‘remembering my partner’s important appointments,’’ ‘‘making sure my partner’s bills get paid on time’’). The social network measure tapped people’s efforts to center their partner’s social worlds around themselves (e.g., ‘‘I do the things my partner likes to do so he/she won’t need to do those things with other people,’’ ‘‘I make a lot of effort to make sure my partner’s friends like me’’). The measure of barriers tapped how much people thought their partner stood to lose by ending the relationship (e.g., ‘‘My partner would lose friends we share,’’ ‘‘My partner would have trouble finding a partner who did as much for him/her as I do’’). The results of these experiments dramatically illustrated the normative power of activating the exchange script. First, priming the exchange script automatically activated concerns about being inferior and being replaced for both low and high self-esteem people alike. We found a significant main effect on a composite measure of exchange anxieties in each experiment (i.e., felt inferiority to the partner and concerns about alternative partners
28
0.20
0.20
0.15
0.15
0.10
0.10
Partner-dependence promotion
Worry about being inferior and being replaced
Sandra L. Murray and John G. Holmes
0.05 0 −0.05 −0.10 −0.15
0.05 0 −0.05 −0.10 −0.15 −0.20
−0.20 Exchange (coins)
Control (circles)
Exchange (coins)
Control (circles)
Figure 1.8 The effects of exchange priming (US coins) on felt inferiority to the partner and dependence-promotion.
replacing the self each transformed to a z-score and averaged). The first panel of Fig. 1.8 presents the results in the coin priming experiment. The mere metaphor of a purely economic exchange imparted by the image of US coins prompted stronger fears of being inferior to the partner and being replaced for low and high self-esteem people alike. Evaluating the general appeal of personal ads that happened to mention the romantic hopefuls’ expectations of a matched exchange also heightened worries about being replaced and feelings of inferiority to the partner (relative to controls). Second, priming the exchange script automatically triggered dependencepromoting behaviors that function to make people feel more indispensable. We found a significant main effect of condition on a composite measure of dependence-promotion (i.e., role specialization, life task management, social network narrowing, and imagined barriers each transformed to a z-score and averaged) in each experiment. Evaluating the general appeal of personal ads simply conveying expectations of an even trade increased people’s need to see their partner as dependent on them and, thus, themselves as indispensable. Amazingly, the simple image of US coins had a comparably powerful effect on dependence-promotion! The second panel of Fig. 1.8 presents these results. In sum, these implicit reminders of the exchange script automatically elicited dependence-promotion. They increased people’s sense of responsibility for managing the mundane aspects of their partner’s life, their desire to specialize roles, their efforts to center themselves in their partner’s social world, and their need to impose and perceive barriers that kept their partner committed. As excited as we were by these results, we realized that the effects of the exchange primes on dependence-promotion might reflect the general
The Commitment-Insurance System
29
regulation of positive relationship sentiment rather than functioning to secure the partner’s commitment. We included a dependent measure tapping closeness in each experiment to deal with just this possibility. Priming exchange continued to predict heightened inferiority worries and increased dependence-promotion in these experiments when we controlled for feelings of closeness. This suggests that the alarm and repair systems automatically operate to solve the specific problem posed by ensuring a partner’s commitment rather than functioning simply to make people feel better. 2.2.4. If partner dependent, then equal to partner Do such efforts to prove one’s indispensability to the partner even the exchange and redress feelings of inferiority (Path E in Fig. 1.1)? The newlywed data provided compelling evidence that behavioral efforts to increase a partner’s dependence do actually promote greater partner commitment. We also found some evidence in this study that such efforts also predicted decreased feelings of inferiority for the enactor on the same day. However, this effect did not emerge across days. Accordingly, to further address this causal link in our logic, we designed an experiment that manipulated perceptions of the partner’s dependence (Murray et al., 2007, Experiment 1). We then measured felt inferiority (e.g., ‘‘I feel inferior to my partner’’) and confidence in the partner’s continuing commitment (e.g., ‘‘I am confident that my partner will always want to stay in our relationship’’). When our undergraduate participants arrived at the laboratory, they completed a relationship behavior inventory purportedly developed by researchers at the University of California, Los Angeles. In the high partner-dependence condition, participants indicated how often they engaged in 15 frequent relationship behaviors (e.g., ‘‘providing transportation,’’ ‘‘doing favors for their partner’’) on a scale designed to elicit high scores. In the low partner-dependence condition, participants indicated how often they engaged in 15 infrequent relationship behaviors (e.g., ‘‘managing my partner’s finances,’’ ‘‘organizing time with my partner’s family’’) on a frequency scale designed to elicit low scores. Participants in the high partner-dependence condition then received feedback that their partner needed and depended on them to an unusually high degree—that he/she counted on them for a number of everyday decisions and arrangements and for planning and managing the practical or mundane aspects of his/her life. Participants in the low partner-dependence condition learned that their partner counted on them when he/she needed to, but was comfortable being independent, and taking responsibility for the mundane aspects of his/ her life. In designing this feedback, we were mindful of the fact that people, especially those high in self-esteem, might be annoyed to learn that their partner could not manage the simple aspects of life without assistance.
30
Sandra L. Murray and John G. Holmes
Nonetheless, perceiving a partner as dependent still leveled the perceived playing field. Participants worried less about being inferior to their partner when they were led to see their partner as especially dependent, and thus, themselves as indispensable in their partner’s life. For people with high selfesteem, this state evidence of their partner’s dependence even increased their overall confidence in their partner’s love and commitment! 2.2.5. The moderating role of self-esteem: Making exchange explicit The newlywed study and the personal ads and US coin exchange priming experiments together provide powerful convergent evidence that the contingencies of exchange are implicit, procedural features of people’s relationship representations—features that are normative, and thus, not moderated by global self-esteem. However, when exchange becomes the focus of a person’s conscious deliberations, we expected global self-esteem to moderate the application of exchange contingencies to the relationship (Path F in Fig. 1.1). When consciously deliberating the principle of making a matched exchange, high self-esteem people should resist applying such considerations to their own relationships, seeing themselves as less in need of the instrumental insurance provided by their partner’s dependence in such circumstances. In contrast, low self-esteem people should still be vulnerable to these concerns and redouble efforts to increase their partner’s dependence (because they do not generally believe they possess the kinds of qualities that could otherwise secure their commitment). In the first of these experiments (Murray et al., 2007, Experiment 6), we primed the exchange script by having people play matchmaker within a dating service simulation. Control participants picked magazine articles for others. The experimenter introduced the matchmaker role to experimental participants by stating that the best romantic relationships are ones in which romantic partners are well matched in terms of their traits and qualities and that relationships work best when partners feel they have gotten a good deal. Experimental participants received laminated cards that depicted two male and two female target profiles and six potential dates. Each profile contained a self-description (e.g., ‘‘I’m a fit sociable woman with an active social life, seeking a like-minded man for a long-term relationship. I enjoy movies, hiking, cooking and travel.’’). It also contained numerical ratings (out of 100) of the profiled person on either four social commodities, such as career success, or four communal qualities, such as emotional stability. The averaged ratings for the target and potential date profiles varied in ways that made better (i.e., equal) and worse (i.e., unequal) fits. The experimental participants then selected a potential date for each target. We then measured felt inferiority to the partner, worry about being replaced, and dependencepromotion as we did in the implicit exchange priming experiments.
The Commitment-Insurance System
31
In the second of these experiments (Murray et al., 2007, Experiment 7), we primed the exchange script by having people divine or predict the fate of others’ marriages. Participants in the two experimental conditions read profiles of couples they believed had been involved in a longitudinal study of marriage. Their task was to divine the fate of the marriage (together or divorced) based on the profile. Each profile described a couple that was matched or mismatched on either social commodities, such as attractiveness, or communal qualities, such as gregariousness. An example of a matched and mismatched social commodities couple follow: (Matched). Jim and Janet discovered they shared many of the same interests when they met in high school. They were both popular among their peers, and both worked hard to achieve high academic standards. They chose the same college, and they felt lucky to have found an appropriate match so early on in their lives. Three years into their marriage, they are talking about having children. (Mismatched). Alison and Michael even seemed like an unlikely pairing to themselves. Alison was always very motivated to succeed in her career, and really valued ambition in others. Michael is more content taking a laid-back approach to his work. Although Alison appreciates many of Michael’s qualities, his laid-back approach sometimes makes her wonder whether she settled for someone who was not quite up to her level of professionalism. Participants read each profile and then predicted the marriage’s fate. In the social commodities condition, we further reinforced the principle of exchange by having the computer respond ‘‘correct’’ when they designated a matched couple as together or a mismatched couple as divorced and ‘‘incorrect’’ when they designated a matched couple as divorced and a mismatched couple as together. We then measured felt inferiority to the partner, worry about being replaced, and dependence-promotion. Both experiments yielded strong support for our hypotheses. When playing the role of matchmaker, low, but not high, self-esteem people reported greater worry about being inferior and being replaced as compared to controls. High self-esteem people suppressed these concerns. In fact, when divining marital fates, high self-esteem people actually reported less worry about being inferior and being replaced as compared to controls. As expected, low self-esteem people were not able to escape the need to prove their partner’s dependence when they consciously deliberated the exchange script. Figure 1.9 presents the interaction for the composite measure of dependence-promotion in the marital divination task as an example. In both experiments, low self-esteem people in the exchange priming conditions doubled their behavioral efforts to prove their partner’s dependence by making themselves indispensable. However, high self-esteem people downplayed their partner’s dependence in these circumstances, presumably
32
Sandra L. Murray and John G. Holmes
0.4 0.35 0.3 0.25
Dependence promotion
0.2 0.15 0.1 0.05 0 −0.05 −0.1 −0.15 −0.2 −0.25 −0.3 −0.35 −0.4 High self-esteem
Low self-esteem Social commodities exchange
Control
Communal qualities exchange
Figure 1.9 Partner-dependence promotion as a function of global self-esteem and marital divination.
protecting the perception that their partner’s commitment was intrinsically motivated. The effects of exchange priming all remained significant when we controlled for reports of closeness, suggesting again that exchange contingencies function specifically to regulate the partner’s dependence. 2.2.6. Summary The alarm and repair contingencies function to minimize or avoid the potential interdependence problems posed by a partner’s fading intrinsic motivation. In the alarm metric, drops in the perception of equivalence activate the exchange script (Path B in Fig. 1.1) and the exchange script activates concerns about inferiority (Path C in Fig. 1.1). Once activated, exchange concerns trigger the repair system (Path D in Fig. 1.1). This contingency prompts behavioral efforts to increase the partner’s dependence and make oneself indispensable, thereby evening the exchange (Path E in Fig. 1.1). In combination, the newlywed and implicit exchange priming experiments suggest that ‘‘if–then’’ exchange contingencies are a fundamental aspect of relationship representations, shaping perception and
The Commitment-Insurance System
33
behavior without people’s awareness. When consciously deliberated, the exchange script triggers the alarm and repair contingencies for low selfesteem people. However, high self-esteem people resist applying exchange considerations to their relationships in such circumstances, protecting the desired belief that their partner really wants rather than needs to be with them (Path F in Fig. 1.1).
2.3. If partner committed, then pursue connectedness By gauging a partner’s motivations for responsiveness, the commitmentinsurance system helps people discern which specific situations are likely to produce outcomes that satisfy their needs (Murray et al., 2006b). Such a behavioral guide is needed to negotiate interdependent life. No matter how accommodating Harry generally is to Sally’s needs, there will inevitably be situations where his behavior is capricious, less than generous, or outright selfish. Sally also needs to risk such rebuffs in situations where she is especially vulnerable because her needs for connection are heightened. Imagine that Sally has had a bad day at work and comes to Harry for comfort. In such an instance, Sally might experience anything less than Harry’s undivided attention as nonresponsiveness. In such ways, interdependent situations often put connectedness and self-protection goals in conflict (Murray et al., 2006b, 2008). By making Sally’s willingness to pursue connection contingent on the perception of Harry’s commitment, this system provides a general guide for resolving such conflicts (Path G in Fig. 1.1).3 We found initial evidence of the normative importance of perceived regard in regulating the pursuit of connectedness goals in cross-sectional samples of dating and married couples. As part of our interest in understanding the biases inherent in romantic perception, we asked people to describe how they believed their partner saw them on a variety of positive and negative interpersonal qualities (e.g., warm, responsive, selfish). This meta-perspective served as our measure of perceived regard (Murray et al., 2000). As the commitment-insurance system stipulates, such metaperceptions did indeed constrain optimism about the partner’s level of commitment. In both dating and marital relationships, people who felt less positively regarded on these traits also felt less accepted and loved, and critically, they perceived their partner’s commitment to their relationship as more tenuous and unstable (Murray et al., 2001).
3
We use the term perceived commitment to refer to a constellation of constructs that signal a partner’s willingness to respond to one’s needs. These tightly interconnected constructs include perceptions of the partner’s acceptance and love, perceptions of the partner’s regard for the self on specific traits, perceptions of a partner’s commitment, and perceptions of the partner’s willingness to respond selflessly to one’s needs (Reis et al., 2004; Murray et al., 2006b).
34
Sandra L. Murray and John G. Holmes
Such cross-situational perceptions of the partner’s regard for the self in turn constrained the general willingness to risk a sense of connection to the partner. In terms of the qualities people attribute to a partner—the social inference that is perhaps most critical for establishing secure connections— people in both dating and marital relationships reserved judgment, not letting themselves believe that they had found the right partner until they felt confident of that partner’s positive regard. Specifically, people attributed more positive qualities to their partner when they believed their partner saw them more positively (Murray et al., 2000). In contrast, dating and marital intimates were more likely to find fault in their partner’s traits when they were more uncertain about their partner’s regard. Such commitmentcalibration was also evident over time, pointing to the causal priority of perceived regard in fostering connectedness. Dating intimates who initially felt more positively regarded later reported greater certainty in their commitment. They also came to see their partner’s traits more generously (Murray et al., 2000). Such evidence of a contingency linking confidence in another’s responsiveness to connectedness is not limited to general perceptions. Instead, state perceptions of another’s responsiveness regulate approach behaviors in specific situations. From the perspective of attachment theorists, finding that a caregiver is not consistently available or responsive in specific contexts deactivates proximity-seeking behaviors (Mikulincer and Shaver, 2003). For instance, when infants’ attempts to elicit care go unmet, they develop avoidant behavioral patterns in stressful situations, actively diverting attention away from their caregivers (Ainsworth et al., 1978). In adulthood, people who suffer hurt feelings at the hands of specific others report that the incident caused them to restrict or sever their relationship with the transgressor (Leary et al., 1998). When the general risks of nonresponsiveness and relationship loss are experimentally primed, people also see their partner’s past transgressions as more serious and they become less willing to believe excuses for these transgressions (Boon and Holmes, 1999). In contrast, when people perceive someone to be more accepting of their self-disclosures, they also report greater feelings of closeness to a partner in a specific interaction (Laurenceau et al., 1998). People are also more willing to offer help that puts them at a disadvantage when they believe a prospective relationship partner is available for a deeper, more caring relationship than when they believe a partner is available for only a more superficial relationship (Clark et al., 1986). 2.3.1. The moderating role of self-esteem Earlier we reasoned that perceptions of a partner’s commitment that are rooted in feeling cherished (i.e., the primary form of commitmentinsurance) may have qualitatively different effects than confidence that stems from feeling indispensable (i.e., the secondary form). If that is the
The Commitment-Insurance System
35
case, low and high self-esteem people should differ in the strength of the ‘‘if–then’’ perceived-commitment contingencies that govern their pursuit of connectedness goals (Path H in Fig. 1.1). We found the first evidence that low and high self-esteem people respond differently to situated threats to their partner’s responsiveness in a series of experimental studies examining the effects of acute self-doubt on relationship perceptions. In developing this research, we began with a provocative set of experiments conducted by Baldwin and Sinclair (1996). These experiments revealed that low, but not high, self-esteem people perceive interpersonal acceptance as contingent in nature. Low self-esteem people were quicker to identify rejection-related words in a lexical decision task when they were primed with failure than with success, and they were quicker to identify acceptance-related words when they were primed with success than with failure. Such implicit contingencies were not evident for people high in self-esteem. On the basis of these data, we reasoned that low self-esteem people may use their own qualities as a barometer of a partner’s responsiveness. In turn, we expected such situated changes in their expectations of their partner’s responsiveness resulting from acute self-doubt to govern their willingness to seek connection to their partner. Specifically, for low self-esteem people, questioning a partner’s responsiveness in one specific situation should activate ‘‘if– then’’ contingencies that link such doubts to motivated efforts to decrease connectedness. Why would this be the case? People low in self-esteem already question their overall worthiness of acceptance (Leary and Baumeister, 2000). Consequently, specific experiences with rejection hurt a lot because they pose a large proportional loss to a diminished or impoverished resource (Murray et al., 2002b, 2006b). By reacting to such experiencing by decreasing connectedness, low self-esteem people give their partner less power to hurt them in the future by making the partner a less important source of need satisfaction and less valued informant on their worthiness of love. In two initial experiments, we led experimental participants to believe that they were not particularly considerate (Experiment 3) or particularly intelligent (Experiment 4). For lows, such feedback induced considerable anxiety about their dating partner’s ongoing responsiveness and commitment as compared to controls (Murray et al., 1998). Such anxieties then curtailed connectedness pursuits. Low self-esteem people responded to these induced anxieties by relying less on their partner as a source of selfesteem and comfort (Murray et al., 1998). They also evaluated their partner’s qualities more negatively (Murray et al., 1998). Such distancing efforts were also evident in a further series of experiments where we directly primed anxieties about responsiveness by leading experimental participants to believe their partner was upset with them (Murray et al., 2002b). Such devaluing efforts also surface on implicit measures of partner regard suggesting that they are in part automatic (DeHart et al., 2004).
36
Sandra L. Murray and John G. Holmes
Similarly contingent connectedness pursuits are also evident for people who are likely to be low in self-esteem by virtue of high levels of attachment-related anxiety or chronic rejection sensitivity. Women higher in attachment anxiety display greater anger toward their partner in a situation in which their partner was not as responsive as they hoped (Rholes et al., 1999). After discussing a serious relationship problem, more anxiously attached people also reported greater anger and hostility (as compared to controls who discussed a minor problem), and most critical, they downplayed their commitment (Simpson et al., 1996). Intimates high on attachment-related anxiety also reported feeling less close to their partner in a situation where they accurately inferred their partner’s attraction to opposite-sex others (Simpson et al., 1999). People high on attachment-related anxiety also react to higher levels of daily conflict by minimizing their feelings of closeness to their partner (Campbell et al., 2005). Women chronically high on rejection sensitivity respond to a potential partner’s indifference by evaluating that partner more negatively (Ayduk et al., 1999). They are also more likely to initiate conflicts on days after they felt more rejected by their partner, and simply priming rejection-related words activates hostility-related thoughts for these women (Ayduk et al., 1999). The picture may not be as bleak as it seems for people chronically susceptible to rejection anxieties, however. Perceiving a partner as dependent—though not a fully satisfactory substitute for feeling cherished—may nonetheless pre-empt defensive attempts to distance oneself from the partner in some limited number of situations within the relationship. In these pockets of safety, low self-esteem people may anticipate sufficient acceptance to risk seeking greater connection to their partner. For instance, if Sally has engineered Harry’s social network such that he can come only to her for advice about work, she may feel confident enough of his acceptance in this domain to risk disclosing work-related worries herself. Perceiving Harry as dependent on her support functionally defines this type of situation as low in risk—negating the need to regulate her own dependence. As a result, Sally might more readily escape her usual tendency to respond to the prospect of dependence on Harry with avoidance and distancing behaviors. For people high in self-esteem, questioning a partner’s responsiveness generally activates ‘‘if–then’’ contingencies that link hurtful situations to the goal of increasing connectedness. People with high self-esteem generally anticipate interpersonal acceptance. Consequently, specific rejections pose a smaller proportional loss to a comparably rich resource. Instead, the goal of maintaining the desired level of confidence in the partner’s ongoing love and commitment is likely to prevail and guide behavior in most situations within the relationship (Murray et al., 2003a,b). Consistent with this logic, dating intimates high in self-esteem react to experimentally induced doubts about their considerateness or intelligence
The Commitment-Insurance System
37
by reporting greater reliance on their partner as a source of reassurance and comfort (Murray et al., 1998). People high in self-esteem also responded to induced concerns about their dating partner’s likely annoyance with them by reporting greater feelings of closeness to that same partner (Experiment 2, Murray et al., 2002b). Similarly, people low on attachment-related anxiety (and thus higher in self-esteem) come to value their partner more after discussing a serious than a minor conflict (Simpson et al., 1996). Such contingencies even extend to situations where a partner is attracted to another: people low in attachment-related anxiety feel closer to their partner the more accurate they are in discerning the potential threat conveyed by their partner’s attraction to others (Simpson et al., 1999). 2.3.1.1. Function of feeling uniquely valued? We reasoned that selfesteem (and its correlates) has its effects in setting the calibration of the ‘‘if committed, then connect’’ contingency because high self-esteem affords reason to trust in a partner’s intrinsic motivation to be responsive. More recent evidence offers further, though still indirect, evidence that this is indeed the case. First, people with high self-esteem believe their partner perceives them more positively on desirable interpersonal qualities, such as warmth, responsiveness, attractiveness, and intelligence (Murray et al., 2000). In fact, high self-esteem people are more likely than lows to believe that their partner sees virtues in them that they do not even see in themselves. Second, feeling more valued in a partner’s eyes seems to set the calibration of the ‘‘if committed, then connect’’ contingency in a way that mirrors the effects of global self-esteem. We found this evidence for the moderating power of perceived regard in a recent daily diary study involving a community-based sample of married couples (Murray et al., 2003a,b). In this study, Murray and her colleagues indexed people’s chronic confidence in their partner’s ongoing motivation to be responsive by asking them to describe how their partner perceived them on various desirable (and undesirable) interpersonal qualities. In this metric, believing that a partner sees oneself more positively gives people greater reason to trust that their partner sees them as equal, and thus has more than ample reason to cherish them and want to tend to their needs (Murray et al., 2000). Each member of the couple then completed a standardized daily diary each day for 21 consecutive days. Analyses of these data revealed that generally believing that a partner perceived one’s traits as more intrinsically valuable reversed the ‘‘if committed, then connect’’ contingency. These people actually drew closer to their spouse on days after they worried most about their partner’s rejection and responsiveness—just like people high in self-esteem. For people who believed their partner saw little to value in their traits, however, feeling acutely rejected activated the behavioral contingency ‘‘distance myself from my partner.’’ They responded to such hurts by treating their partner in more
38
Sandra L. Murray and John G. Holmes
cold, critical, and negative ways the next day. These reactions emerged even though the partners of people who felt less valued were not actually upset with them on those days when lows felt most rejected. Instead, people who felt less valued responded to imagined rejections by treating their partner badly. 2.3.2. Summary Perceptions of partner’s ongoing motivation to be caring and responsive do indeed seem to regulate people’s general willingness to risk a sense of connection to their partner (Path G in Fig. 1.1). When confident their partner values them, and is committed to them, dating and married intimates allow themselves to see more to value in their partner. In this way, the normative operation of the ‘‘if–then’’ commitment-connectedness contingency ensures that people do not generally allow their own level of commitment to outstrip their partner’s. Because even the most well-intentioned partner will behave selfishly or inconsiderately on occasion, certain circumstances do merit relaxing this contingency. In such situations, trusting in a partner’s intrinsic motivations for caring seems to afford a kind of resilience that desensitizes this contingency. High self-esteem people react to induced concerns about their partner’s responsiveness by drawing closer to their partner. However, the level of trust that is afforded by confidence in only a partner’s extrinsic motivations supersensitizes this contingency. Low selfesteem people respond to the slightest indication that their partner’s responsiveness might be in question by derogating their partner, thereby diminishing their sense of connection.
3. Theoretical Innovations, Applications, and New Directions In this next section, we detail how the central tenets of the commitment-insurance model originated in classic perspectives on interdependence theory (Kelley, 1979; Levinger, 1976; Thibaut and Kelley, 1959). This historical context then provides a platform for presenting a ‘‘New Look’’ at interdependence theory. In outlining this ‘‘New Look,’’ we detail the applications of the model, integrate further theoretical perspectives, and point to directions for future research.
3.1. The historical context In their classic text, Thibaut and Kelley (1959) specified the beginning propositions of interdependence theory. In conceptualizing the essence of a relationship, they started with the tangible—the practical rewards and
The Commitment-Insurance System
39
costs the partner provided. The rewards Harry might receive from Sally included benefits as abstract as support and as concrete as money. The costs he might incur included the loss of his time or the weathering of Sally’s occasional bad temper. Harry’s overall outcomes in the relationship reflected a subtractive tally of costs from rewards. In stipulating the strength of the dyad, Thibaut and Kelley (1959) drew a distinction between the comparisons that govern satisfaction and the comparisons that govern commitment. Within this comparative calculus, Harry’s level of satisfaction reflects a comparison between the outcomes he receives and the outcomes he expects (i.e., his ‘‘comparison level’’). His happiness thus depends on his outcomes exceeding his expectations. Satisfaction is not enough to keep Harry committed to his relationship, however. Instead, commitment reflects a comparison between the outcomes Harry receives and the ones he expects await him in an alternate relationship or in life alone (i.e., his ‘‘comparison level for alternatives’’). This commitment thus depends on his outcomes exceeding his comparison level of alternatives. Although groundbreaking, the classic model seems strangely nondyadic in retrospect. The main role the partner played in determining the strength of the dyad was as a source of rewards or costs; the comparisons determining dyadic strength all centered around comparisons to one partner’s internal standards or expectations. Kelley (1979) and Levinger (1976) later presented models of dyadic strength or pair cohesiveness that had decidedly more dyadic flavors. In a reformulation of interdependence theory, Kelley (1979) added a critical motivational element to the model by distinguishing between given and effective outcomes. Imagine that Harry and Sally want to go to a movie together, but Harry wants to see a romantic comedy, whereas Sally wants to see a documentary. At the level of Harry’s given outcomes (i.e., the ‘‘given matrix’’), going to the romantic comedy is most desirable. However, Kelley (1979) recognized that Harry’s outcomes also depend on his consideration of Sally’s welfare. Harry can transform the situation into one where going to the documentary is actually more pleasing to him because such a choice optimizes Sally’s happiness (i.e., the ‘‘effective matrix’’). Witnessing such transformations in turn gives Sally reason for trusting in Harry’s commitment to her. In this reformulation, Sally’s outcomes do not depend just on the tangible benefits Harry provides. Instead, they depend on the symbolic outcomes based on the attributions of caring Sally can draw from his behavior. Levinger (1976) presented a still more comprehensive model of pair cohesiveness that incorporated internal and external motivating factors or considerations. As Fig. 1.10 illustrates, this model integrated interdependence theory assumptions with Lewin’s conceptions of force fields (i.e., motivational tensions) within the life space (Deutsch and Krauss, 1965;
40
Sandra L. Murray and John G. Holmes
b +
S
b -
-
b
H
+
b
Figure 1.10 Levinger’s model of pair cohesiveness (adapted from Levinger, 1976).
Hall and Lindzey, 1957). Each circle represents one partner, ‘‘S’’ for Sally and ‘‘H’’ for Harry. The degree of overlap between these circles represents their level of interdependence. For each person, Levinger (1976) recognized that certain factors or attractions, marked by a ‘‘þ,’’ pull a couple together, and certain attractions, marked by a ‘‘-,’’ push a couple apart. Positive attractions might include Harry’s intelligence or Sally’s physical appeal. Negative attractions might include Sally’s impatience or Harry’s laziness. The barriers, marked by ‘‘b,’’ represent restraining forces, such as sunk costs, the absence of alternative partners, or familial pressures that keep the dyad intact even when positive attractions cease to outweigh negative ones. When the net positive and negative attractions operating on Sally and Harry are equally strong, their level of interdependence is greatest, and the strength of their mutual commitment is more secure.
3.2. Dyadic strength: A motivational perspective on interdependence theory The commitment-insurance model expands upon the classic models by stipulating that dyadic strength depends on the attributions people make for their partner’s commitment. Such meta-perspective taking is critical within our model because assessments of a partner’s commitment control one’s own commitment. Within our ‘‘New Look’’ on interdependence theory, a theory of the partner’s mind (or motivations) takes a central or deciding role (Holmes, 2002). We would even go so far as to argue that people may not fully consider issues of their own happiness and actually weigh the benefits provided against the costs incurred until they are certain of their partner’s commitment. Without it, one’s own desires are practically moot. Figure 1.11 presents this ‘‘New Look’’ on dyadic strength. In this model, interdependence takes on multiple flavors, depending on one’s vantage
41
The Commitment-Insurance System
A = Perceived motivational interdependence (shared theory of mind) B = Actual motivational interdependence (actual similarity) C = Perceived similarity in motivations D = Accuracy in motivational perceptions Barriers
+ + −
+
+
+
Harry’s motivations
Sally’s motivations
B
−
−
D C
A H+
Sally’s perception of harry’s motivations H−
A
C S+ Harry’s perception of sally’s motivations S−
S+
Barriers
Figure 1.11 A new look at interdependence theory.
point on Harry’s and Sally’s motivations and goals. The innermost circles capture the most fundamental level of interdependence—the degree of convergence in partners’ theories of mind (Area A). These circles capture the degree of overlap between Sally’s perception of Harry’s motivations and Harry’s perception of Sally’s motivations. We place these meta-perspectives at the core because behaviors that increase interdependence, such as selfdisclosing, seeking support, or giving a partner decision-making power, all require taking on substantial risk of rejection. Taking such risks requires confidence in a partner’s ongoing commitment (Murray et al., 2006b, 2008). Consequently, our model assumes that actual behavioral interdependence is greatest in relationships where each partner perceives the other as intrinsically motivated. Why? Because feeling intrinsically valued desensitizes people to rejection. It then increases their willingness to enter situations where their partner’s actions control their outcomes which then affords opportunities for increased connection and continued relationship growth (Kelley et al., 2003). In contrast, lower levels of risk-taking, and thus lower levels of interdependence, are likely to characterize relationships where one or both partners perceive the other less than intrinsically motivated. Sally’s meta-perspective on the positive and negative attractions most salient to Harry give her initial grounds for attributing his commitment to
42
Sandra L. Murray and John G. Holmes
more or less intrinsic factors. In particular, the relative comparison Sally makes to Harry shapes her appraisal of his major attractions within the relationship. These perceived attractions are captured by H(þ) and H(-) within Fig. 1.11. Feeling equal to Harry gives her reason to believe that Harry sees many intrinsically valuable qualities in her, such as attractiveness, intelligence, and wit, and few intrinsically undesirable ones—presumably the same constellation of qualities she thinks he values in himself. However, feeling inferior to Harry gives Sally ample reason to suspect that other factors must (and need to) govern his commitment. Consequently, she might conclude that he is drawn to her for the instrumental benefits she provides, such as managing his schedule and coordinating his friends (the original focus on interdependence theory analyses). Such calculations of the other’s benefits and costs may diverge across partners as the figure illustrates; in the case diagrammed, Harry believes Sally sees more of value in him than vice versa. Perceptions of a partner’s motivations then may shape subsequent behavioral experiences within the relationship, which then serves to reinforce such expectations. Trusting in Harry’s intrinsic motivations allows Sally to put herself in diagnostic situations where she has the potential to obtain actual behavioral evidence that proves Harry’s caring (Kelley, 1979). For instance, knowing that Harry values her may give Sally the confidence to ask him to go out with her friends when she knows he would rather spend the evening alone. It might also allow her to tell Harry about her fears that she will not succeed at her job. If Harry responds positively to such situations, Sally then has objective grounds for trusting in his intrinsic motivation. Consistent with this analysis, people report greater trust in their partner’s caring when they witness signs of their partner’s commitment, such as selfless responsiveness to their needs or willingness to excuse their transgressions (Wieselquist et al., 1999). People also report greater trust in a partner’s caring when that partner responds appropriately to their selfdisclosures (Collins and Feeney, 2004). Such opportunities for learning may be largely foreclosed if Sally suspects that Harry stays with her only because of the instrumental benefits she provides. In such circumstances, Sally should be hesitant to ask Harry for a sacrifice of any kind because doing subtracts from the very benefits she believes sustain his commitment in the first place. The surrounding circles capture higher order levels of interdependence. One set of circles capture the overlap between partners’ motivations—the degree of motivational interdependence (Area B). At this level, Sally’s perception of the positive (þ) and negative () traits inherent in Harry governs her ongoing assessment of her own commitment. When Sally perceives more inherently valuable than costly qualities in Harry, she is likely to experience her commitment as intrinsically motivated. However, when such implicit calculations fall short, she might attribute her sense of
43
The Commitment-Insurance System
commitment to more instrumental factors—focusing on what Harry does for her. Such assessments may also diverge across partners; in our illustration, Sally perceives an equal balance of positives and negatives, but Harry perceives more positives than negatives. A second set of circles capture the overlap between Sally’s perceptions of Harry’s motivations and her assessment of her own motivations (Area C). These circles capture how Sally’s perceptions of Harry’s motivations control or constrain her attributions for her own commitment. We assume Sally simply cannot afford to believe that Harry is truly wonderful (and hard to replace) unless she believes that he also sees herself as irreplaceable. Failing such a conclusion, Sally is safer assuming she is drawn to the instrumental benefits Harry provides (benefits, incidentally, that she might assume are easily found in other relationships). The remaining circle (Area D) captures the overlap between Sally’s perceptions of Harry’s motivations and Harry’s actual motivation. In the case diagrammed, Sally has some appreciation of Harry’s motivations, but Harry has no appreciation of Sally’s motivations. The outermost oval captures the extrinsic factors or barriers that might continue to maintain the dyad in circumstances where neither intrinsic nor instrumental motivations are sufficient to explain (or sustain) commitment. These real and psychological barriers include factors as subjective as people’s investments or sunk costs within the relationship (Van Lange and Rusbult, 1993), family or religious prohibitions against dissolving the relationship (Levinger, 1976), the presence of children (Levinger, 1976), and the perceived absence of viable alternatives to the relationship (Simpson, 1987; Van Lange and Rusbult, 1993). 3.2.1. Typology of motivations Our ‘‘New Look’’ on interdependence assumes that the attributions people make for their partner’s commitment provide the core of dyadic strength. To this point, we have said relatively little about the nature of these attributions. Table 1.2 presents a possible typology (adapted from Rempel et al., 1985). It also specifies the evidence base for each of these attributions and their insurance value. As we detail below, the attributions that afford Table 1.2
A typology of motivations
Motivation
Evidence base
Insurance value
Intrinsic Instrumental
Traits Behavioral benefits; intangible benefits; pockets of incompetence Engineered; inherent
Greatest Moderate
Extrinsic
Least
44
Sandra L. Murray and John G. Holmes
greater insurance are ones that instill more optimism about a partner’s commitment—the belief that the commitment is generated by attraction to the positive (i.e., an approach motivation) rather than compelled by fear of the negative (i.e., an avoidance motivation). Attributions to a partner’s intrinsic motivations afford the greatest insurance. Such attributions involve Sally’s inference that Harry sees intrinsically valuable traits in her—qualities that have an enduring essence and stay with her from one situation to the next. Thus, attributing Harry’s commitment to the value he puts on her intelligence, warmth, and generosity gives Sally reason to believe that Harry will respond to her needs because he wants to stay with her—precisely because he takes so much joy in having someone like her as his partner. Attributions to a partner’s instrumental motivations afford an intermediate level of insurance. Such attributions involve Sally’s inference that Harry needs to stay with her because he depends on the many behavioral benefits she provides. Such benefits might involve very tangible, practical ones, such as cooking his meals, balancing his checkbook, managing his family relationships, and scheduling his appointments. They might also involve more intangible or socioemotional ones, such as providing a sympathetic ear, laughing at his over-told jokes, and satisfying his needs for physical affection. Sally might even engineer Harry’s dependence by creating pockets of incompetence—perhaps by ensuring he does not have the information he needs to manage his financial life independently or coordinate their children’s activities on his own (Wegner et al., 1985). In so doing, Sally tries to convince herself (and Harry) that he could not manage without her. The confidence that results from such efforts may be tenuous, however. Sally could reasonably assume that Harry’s commitment would waver if the benefits disappeared and that other people could easily provide such benefits (if they so desired). Attributions to a partner’s extrinsic motivations afford the most minimal level of insurance. Such attributions involve Sally’s inference that Harry has to stay with her because certain outside constraints compel it. Partners might engineer some of these barriers; others might be inherent in the context surrounding the marriage (Levinger, 1976). Engineered barriers might include Sally insinuating herself in Harry’s social network in such a way that ensures dissolving their relationship would disrupt his broader social network. She might also derogate Harry’s alternatives to the current relationship (Simpson, 1987), perhaps finding subtle ways to remind him that he’s lucky to have her (and that others would not put up with his faults). Inherent barriers might include religious prohibitions against divorce, the presence of their children, or shared financial debt and obligations (Levinger, 1976). These three sources of commitment-insurance are not likely to be substitutable. Instead, they could interact in either a synergistic or an
The Commitment-Insurance System
45
antagonistic fashion. Imagine that Sally believes that Harry truly values her sense of humor, warmth, and intelligence. In such circumstances, feeling like Harry also appreciates the many practical benefits she provides might only further reinforce her confidence in Harry’s commitment. However, the knowledge that Harry needed or appreciated the many things she did for him might instill less confidence if she could not also point to more enduring qualities he valued in her through a kind of attributional discounting (Kelley, 1967). Moreover, if Sally focused on the reasons why Harry could not leave (e.g., shared children, or shared debt), such a focus might even diminish her confidence that he valued either her traits or the instrumental benefits she provides (Seligman et al., 1980). Such distinctions suggest possible avenues for future research. One immediate question is this: Can too much substitution be bad for one’s relationship health? Striving to provide a partner with enough instrumental benefits to ensure their dependence might make people lose sight of the qualities their partner actually does value in them, diminishing their selfesteem. Over the longer term, continued efforts to prove one’s indispensability to a partner might also create feelings of unfairness or being taken for granted that precipitate conflicts (Grote and Clark, 2001), much as Waller (1938) prophesized in his principle of least interest. Consistent with this logic, people’s particular motivations for sacrificing for a partner constrain the relationship benefits of such kind deeds. Sacrificing to avoid relationship costs, such as a partner’s anger or rejection, predicts daily relationship difficulties; sacrificing to promote relationship benefits, such as a partner’s happiness, does not (Impett et al., 2005). A second immediate question is this: Do the relationship-benefits of perceiving a partner as committed depend on the nature of the attribution made for the commitment? Imagine that Sally trusts in Harry’s commitment to her either because (a) she knows how much he values her or (b) because she knows he does not have the independent financial where-with-all to seek a divorce. Both attributions (one intrinsic; one extrinsic) may foster a high degree of certainty that Harry is committed. However, only the intrinsic attribution is likely to activate Sally’s connectedness goals. Sally might willingly seek out an intrinsically motivated Harry for support. However, she might deliberately limit such interactions with an extrinsically motivated Harry because she is likely to be much more suspicious of his motivation to provide caring. In this way, the kind of empty-shell marriages described by Berscheid (1983) might develop. Our nested interdependence model also suggests that different fates may await relationships where partners’ attributions for commitment are more or less convergent. When Harry and Sally both trust in one another’s intrinsic motivations, they should both be willing to seek out interdependent situations, increasing dyadic strength. If only one partner trusts in the other’s intrinsic motivation, such asymmetry in people’s actual attributions may
46
Sandra L. Murray and John G. Holmes
create relationship difficulties. In such circumstances, a less trusting Harry is likely to engage in compensatory instrumental behaviors to try to secure Sally’s commitment. Sally might then begin to worry about her own level of contributions to the relationship, feeling like she cannot do as much for Harry as he does for her. Harry’s efforts to even the exchange might then have the ironic effect of creating uncertainty in Sally. Different fates might also await relationships where attributions for the partner’s commitment are more or less accurate. Imagine that Harry finds intrinsic value in Sally’s qualities, but Sally believes he stays with her only because of the instrumental benefits she provides (as is likely to be the case with most low self-esteem people). Harry might grow frustrated with Sally in such circumstances, questioning why she is not willing to afford him the level of trust he feels he deserves. Now imagine that Sally is not quite in touch with the real reasons underlying her commitment to Harry as is typically the case when people have been together for a long time (Berscheid, 1983). She might imagine that she stays only because she lacks viable alternatives; once such alternatives arise, she might leave the relationship only to discover how much she took for granted Harry’s gregarious nature. The hypotheses outlined above might have too Western a flavor for some. Linking intrinsic motivation to the perception of valued traits assumes that people do indeed perceive inherent and stable qualities in themselves. This might be less true of people in more Eastern or interdependent cultures (Dweck, 2000; Markus and Kitayama, 1991). For people in such cultures, perceptions of a partner’s instrumental motivations may be more diagnostic of commitment than perceptions of the partner’s intrinsic motivations. Alternately, the reasons for making attributions to caring may themselves differ across cultures. For people in Eastern cultures, the belief that one provides benefits to the partner that cannot be provided by others may foster perceptions of intrinsic motivation—because such provision makes people feel complete and valued in their role as a relationship partner (Markus and Kitayama, 1991).
3.3. Judging equivalence: What counts? To this point, we have said little about the calculations that control people’s assessments of equivalence or similarity in worth. Such calculations are critical within the commitment-insurance system because such assessments constrain people’s inferences about their partner’s motivations; they also control the operation of the alarm and repair systems. Discerning the implicit calculus that controls such assessments is a critical task for the future. Imagine that Harry thinks that Sally is smarter and more articulate than him (and that Harry knows that Sally shares this perception). Can he even the perceived exchange calculus by emphasizing his gregariousness and sense of humor? Can one desirable trait compensate for another? That is,
The Commitment-Insurance System
47
can apples (his gregariousness) make up for oranges (her intelligence)? Or does he need to find some way to convince himself that he is just as smart and articulate as Sally—but in different ways. Could Harry come to believe that his intuitive emotional intelligence is just as valuable to Sally as her more systematic and analytic mind? Alternately, can he make up for his lack of intellectual and verbal dexterity by compensating behaviorally—perhaps by showcasing his ingenuity in fixing mechanical objects around the house, by cooking Sally gourmet dinners, or by compiling her electronic banking records? Attempts at trait- versus behavioral-compensation might be more or less effective given the circumstances. People might be able to compensate across traits only if they believe their partner values the compensatory trait. Harry’s gregariousness might make up for his relative lack of intelligence only if he thinks Sally values his gregariousness as much as her intelligence. Trait compensation might also be more effective when people think of themselves and their partner as an interdependent unit or team because it is easier to bask in the partner’s greater strengths in such a mindset. Consistent with this logic, the tendency to sabotage a friend’s success on a self-relevant task is attenuated when an interdependent selfconstrual is primed (Gardner et al., 2002). People with high self-esteem may also be more likely to perceive equivalence in compensatory traits because they are so adept at convincing themselves of their own merit (Taylor and Brown, 1988). In contrast, people with low self-esteem may have little recourse other than behavioral compensation. However, it may even the exchange only if they can convince themselves that the behavioral benefits they provide are so special (and so numerous) that their partner could not find them in alternate relationships. The small kernel of truth in people’s self-assessments of worth further complicates the equivalence calculus. On even the most obvious of qualities, people evidence shocking inability to recognize their own strengths (or limitations). Self-assessments of critical social commodities, such as their intelligence, physical attractiveness, and popularity, bear little resemblance to reality (Baumeister, 1993). This means that Harry’s assessments of his worth may bear little resemble to his actual currency on the interpersonal marketplace. Harry’s assessments of his worth may also bear little resemble to Sally’s assessments of his worth (Murray et al., 2000). Furthermore, Harry and Sally might possess objectively equivalent worth, but Sally might still feel inferior to Harry (if she has low self-esteem and he does not). The potential for such bias in the perception of equivalence raises a number of research issues. As one pressing example, it becomes critical to examine the effects of actual as opposed to perceived equivalence in worth. Such asymmetries may have different, selective effects on the intrinsic, instrumental, and extrinsic factors that make a relationship bond more or less fragile. Imagine that Sally
48
Sandra L. Murray and John G. Holmes
is considerably less attractive than her husband Harry (like the author of the Time Magazine article that opened this chapter). Other women might notice this asymmetry and it might lead them to suspect that flirting with Harry will yield a favorable response. In such ways, this objective imbalance in worth actually might increase Harry’s alternatives to the current relationship, lessening one potential barrier restraining his commitment. Consistent with this logic, imbalances in partners’ level of physical attractiveness forecasts relationship dissolution (White, 1980). However, perceived imbalances in the equivalence of the exchange may be more likely to affect the instrumental constraints on a partner’s commitment. When Sally feels inferior to Harry, she may up the behavioral ante to secure his commitment by making herself indispensable.
3.4. The duality of interdependent life Interdependence poses fundamental problems: maintaining one’s own commitment in the face of an imperfect partner (Murray, 1999), gauging and maintaining a partner’s dependence and commitment (Murray et al., 2007), discerning which specific relationship situations are safe enough to risk one’s own dependence on a partner, and identifying those situations that should be avoided (Murray et al., 2006b, 2008). Ongoing problems require relatively automatic and effortless solutions (Bargh and Ferguson, 2000). Accordingly, the solutions to these problems should be implicit in the normative procedural rules that comprise people’s general working models of relationships (Baldwin, 1992; Holmes and Murray, 2007; Tooby and Cosmides, 1996). Consistent with this logic, both low and high self-esteem people ‘‘know’’ the implicit contingencies of the exchange rule. Implicitly activating the exchange script prompted behavioral efforts to secure the partner’s dependence regardless of self-esteem. However, these implicit procedural rules did not always govern behavior (Wegner, 2002). When capable of doing so, high self-esteem people overrode the exchange script. Low selfesteem people did not. These findings suggest that interdependent life may look radically different when considered at different levels of consciousness. The implicit contingencies that govern people’s automatic goals, wishes, and behavioral intentions may not always compel complementary controlled behaviors. Instead, competing rules may govern willed behaviors. Such a duality or dissociation could arise when spontaneously activated associations enter consciousness (Kunda and Spencer, 2003). It could also arise as a function of the person’s disposition, the partner’s disposition, and the relationship circumstance they face (Murray et al., 2007, 2008). We recently found further support for these propositions in a series of experiments on the risk regulation system (Murray et al., 2008). This system solves the interdependence problem posed by regulating one’s own
49
The Commitment-Insurance System
dependence—deciding which specific relationship situations are safe and which should be avoided. For instance, this system governs whether Sally chooses to disclose something personally revealing to Harry; it also governs whether Harry allows Sally to make decisions that control his outcomes. Just like the commitment-insurance system, it operates at two levels: one set of procedural rules governs people’s implicit or automatic responses regardless of self-esteem; a second set of procedural rules governs people’s controlled or willed behaviors differentially as a function of self-esteem. Figure 1.12 illustrates the operation of this system. Situations that involve interpersonal risk and vulnerability automatically activate two competing goals—the appetitive goal of seeking connection (Path A) and the aversive goal of protecting against rejection (Paths B and C). The link between risk and connectedness goals reflects the functionalist argument that the best means of managing a capricious and dangerous world is to seek connection to others who will meet one’s needs (Bowlby, 1969; Kelley, 1979; Leary and Baumeister, 2000; Mikulincer and Shaver, 2003; Tooby and Cosmides, 1996). The links between risk and self-protection (Path B) and between connectedness and self-protection goals (Path C) reflect the unfortunate reality that others will not always be motivated to do so. Certain situations, such as conflicts of interest, are also riskier than others. In such situations, even generally responsive others might be unwilling to meet one’s needs. For these reasons, risk and the desire to seek connection simultaneously activate the goal of self-protection (i.e., avoiding rejection). The model further stipulates that the intensity of the conflict between connectedness and self-protection goals depends on the immediacy of the Threat immediacy
X
Risk and vulnerability
B
A
Connectedness goals
D
C
Self-protection goals
E
Executive control system (operation moderated by perceived partner responsiveness)
Uncontrolled processes
F
Approach or avoid dependence
Controlled processes
Figure 1.12 The risk regulation system. Source: Murray et al. (2008). Adopted with permission, American Psychological Association.
50
Sandra L. Murray and John G. Holmes
provoking risk. The moderating path (Path D) linking risk immediacy to the strength of connectedness and self-protection goals captures this assumption. Imagine that Harry’s best friend broke a promise to him—a threat that is less immediate because it arises from the external relationship world. Such risks induce feelings of vulnerability and activate the goal of approaching his spouse Sally for comfort (Bowlby, 1969). Needing her comfort then activates the goal of protecting against the possibility of Sally’s rejection or nonresponsiveness to his need for support (Path C). Now imagine that it is Sally who broke a promise—a threat that is more immediate because it arises from Harry’s internal relationship world. Such an experience with risk directly activates connectedness (Path A) and selfprotection goals (Path B), and thereby intensifies the goal conflict. This happens because risks internal to the relationship activate self-protection goals both directly (Path B) and indirectly (Path C). Consequently, Harry’s desire to approach Sally is likely to be all the more ambivalent because he needs the comfort of the very person who made him feel vulnerable. Once activated, either directly (Path B) or indirectly (Path C), the goal of protecting the self against rejection then triggers a corrective executive control system (Path E). This system functions as a ‘‘stop/go’’ routine—one that resolves the conflict between connectedness and self-protection goals by helping people identify which specific situations should be avoided and which situations are safe to approach. For people high in self-esteem, risk triggers a control system that directs them toward the situations of dependence within their relationship that can fulfill connectedness goals (Path F). Accordingly, for a high self-esteem Harry, the knowledge that others may disappoint him actually motivates him to override self-protection goals— making him want to disclose to Sally, seek her support, and give her decision-making power. Harry can afford to take such risks precisely because his resilient sense of confidence in Sally’s responsiveness protects him from the implications of rejection in advance. In contrast, for people low in self-esteem, risk triggers a control system that directs them away from situations where they need to trust or depend on their partner. Accordingly, for a low self-esteem Sally, the knowledge that others may disappoint her motivates her to override connectedness goals and suppress any inclination to seek a greater sense of connection to Harry. Unsure of Harry’s responsiveness, a low self-esteem Sally simply cannot afford to risk being hurt. In designing the experiments to test these hypotheses (Murray et al., 2008), we began with three assumptions in mind. Our model stipulates that resolving goal conflict (i.e., correction) requires greater cognitive resources than initial goal activation. Accordingly, the goal priorities of low and high self-esteem people should converge on implicit measures that tap initial goal activation (i.e., uncontrolled responses) and diverge on explicit measures that tap correction (i.e., controlled responses). For people low in selfesteem, risk should prompt implicit associations that reflect the desire to
The Commitment-Insurance System
51
connect, but deliberative intentions that reflect the need to distance (because their control system prioritizes self-protection goals). However, for people high in self-esteem, risk should prompt implicit associations that reflect the desire to connect and deliberative intentions to enact such goals (because their control system overrides self-protection goals). We first looked for dissociations between uncontrolled and controlled processing by looking for dissociations between implicit and explicit responses to risk as a function of self-esteem. In one experiment (Murray et al., 2008, Experiment 1), we primed risks external to the relationship by asking risk-primed participants to describe a time when someone important to them (i.e., a significant other such as a best friend, parent, romantic partner) seriously disappointed them. We then measured the implicit activation of connectedness goals using a lexical decision task. Risk activated connectedness goals for both low and high self-esteem people. Risk-primed participants were quicker to identity words such as vow, hug, kiss, join, want, grow, rely, hope, faith, and trust in the lexical decision task as compared to control participants (who imagined their commute to school). We then examined people’s controlled responses to risk on explicit measures of connectedness-seeking in a follow-up experiment (Murray et al., 2008, Experiment 2). We expected to see differences between low and high self-esteem people on explicit reports (because such measures allow the control system to operate). We again asked experimental participants to describe a time when a significant other seriously disappointed them. We then measured people’s explicit behavioral intentions—namely, their willingness to enter dependent situations in their current dating relationship (e.g., seeking a partner’s support, giving a partner decisionmaking power, and asking a partner to make sacrifices). These types of situations have the most potential to fulfill needs for connectedness because they offer partners a way to prove their caring and responsiveness (Kelley, 1979). As we expected, high self-esteem people pursued connectedness goals in response to risk: they reported stronger intentions to seek out situations of dependence within their current relationship. Low self-esteem people did not. Our second assumption concerned executive control. Our model stipulates that resolving the conflict between connectedness and self-protection goals requires executive control. If that is the case, the controlled or deliberate responses of low and high self-esteem people should be similar when the executive strength needed to correct and suppress connectedness goals is compromised. In response to risk, low and high self-esteem people should both seek greater interdependence with their partner, thereby fulfilling needs for connection. We conducted two more experiments to examine this hypothesis (Murray et al., 2008, Experiments 5 and 6). We manipulated both the salience of external interpersonal risk (by priming a time when a significant other seriously
52
Sandra L. Murray and John G. Holmes
Willingness to risk interdependence
disappointed them) and executive strength (by either depleting self-regulatory strength or imposing a cognitive load). We then measured people’s willingness to enter risky situations of dependence within their dating relationship. In the self-regulatory strength experiment, we depleted executive strength by asking people to ignore words that flashed at the bottom of a screen as they watched a videotape (Baumeister et al., 1998). In the cognitive load experiment, we diminished executive strength by asking people to rehearse and report a long alphanumeric string (Gilbert et al., 1988). Both experiments revealed risk salience by executive strength interactions. Figure 1.13 presents the results for the cognitive load experiment. When under cognitive load, participants reported greater willingness to seek out interdependent situations in response to risk. Thus, when deprived of the executive strength to correct and pursue self-protection goals, low self-esteem people act on connectedness goals just like high self-esteem people. Our third assumption concerned the sensitivity of the risk regulation system to variations in the strength of connectedness and self-protection goals. This assumption stipulates that risk that stems from the partner’s actions (i.e., an internal threat) should strengthen the activation of selfprotection goals. Consequently, even people normally inclined to trust might reduce dependence in situations where rejection risks are more immediate. Imagine that Sally routinely criticizes Harry’s productivity and organizational skills. After a stressful day at work, even a high self-esteem Harry might fear approaching Sally for support. In an experiment designed to test this hypothesis, we focused on the goals that are activated automatically when one partner hurts the other (Murray et al., 2008, Experiment 7). We asked experimental participants to describe a time when their current romantic partner seriously 0.55 0.45 0.35
Load No load
0.25 0.15 0.05 −0.05 −0.15 −0.25 Control
Interpersonal risk prime
Figure 1.13 Willingness to risk interdependence as a function of risk and cognitive load. Source: Murray et al. (2008). Reprinted with permission, American Psychological Association.
53
The Commitment-Insurance System
disappointed them. We then measured the implicit activation of connectedness and self-protection goals using a lexical decision task. We also measured the explicit pursuit of connectedness goals by asking participants whether or not they had forgiven their partner for the transgression (greater forgiveness conveys stronger connectedness goals). Control participants provided a brief description of a disappointment and whether or not they had forgiven their partners after they completed the dependent measures. People’s implicit and explicit responses to risk diverged strongly as a function of self-esteem. Figure 1.14 presents reaction times to self-protection words (e.g., ‘‘retreat,’’ ‘‘protect,’’ ‘‘defense,’’ ‘‘caution,’’ ‘‘careful,’’ ‘‘prevent,’’ ‘‘watchful’’) as these data make our point most compellingly. Look first at the reaction times of high self-esteem people. The responses of high self-esteem people revealed implicit associations to risk that mirrored deliberative considerations. Unforgiven transgressions compelled more caution than forgiven ones. When high self-esteem people had not forgiven their partner’s transgression, reminders of this hurt activated self-protection goals. They were quicker to identify self-protection words in the partnerdisappointed than control condition. However, being reminded of this hurt
860
Reaction times to self-protection words
860 840 840 820 820 800
800
780
780
760
760 740
740
720
720 700
700 Forgiven transgression
Unforgiven transgression
Low self-esteem
Forgiven transgression
Unforgiven transgression
High self-esteem
Partner disappointed
Partner disappointed
Control
Control
Figure 1.14 Self-protection goal activation as a function of self-esteem, forgiveness, and partner disappointed.
54
Sandra L. Murray and John G. Holmes
actually suppressed self-protection goals when high self-esteem people had forgiven the transgression. They were slower to identify self-protection words in the partner-disappointed than control condition. Now look at the reaction times of low self-esteem people. For lows, it was actually forgiving the transgression (and the specter of being closer to the partner) that activated self-protection goals. When low self-esteem people had forgiven the transgression, they were actually quicker to identify selfprotection words in the partner-disappointed than control condition. The responses of low self-esteem people reveal implicit associations to risk that belied their more deliberative judgments. For them, forgiving the transgression seemed to make the prospect of rejection of greater concern, prompting greater defensiveness in automatic goals.
3.5. Taking levels of processing perspectives one step further The potential for a disjunction between the implicit and explicit operation of the commitment-insurance and risk regulation systems points to many potential directions for further research. The existence of implicit procedural rules about exchange suggests that even high self-esteem people may have spots of vulnerability within their relationships. In circumstances where a high self-esteem Harry is not aware that he is feeling intimidated by Sally, he may automatically fall into behaviors meant to increase her dependence. The newlywed data attests to this possibility. However, in one of our social comparison experiments, high self-esteem people reported stronger communal goals when they compared themselves to their partner’s most positive qualities (see Fig. 1.7). This suggests that high self-esteem people might become highly practiced at defending themselves against worrisome social comparisons to their partner, automatically suppressing the exchange script. Delineating the conditions when people can and cannot override the implicit procedural rules that govern their general working models of relationships remains an important goal for further research. Future research is also needed to determine whether overriding automatically activated procedural rules is actually adaptive over the longer term. In the newlywed study, behaving in ways that promote a partner’s dependence actually worked! When Harry went out of his way to meet Sally’s needs—finding her keys, cooking her meals, and fixing her I-pod—she became all the more certain that he was the right partner for her. Consequently, high self-esteem people may consciously override the basic procedural rule linking felt inferiority to dependence-promotion at their long-term peril. Low self-esteem people cannot seem to escape this contingency, suggesting that they might be less likely to end up with partners who feel taken for granted. The capacity to consciously override connectedness goals in regulating risk may forecast more deleterious relationship consequences
The Commitment-Insurance System
55
for low self-esteem people, however. Such corrective processes prevent them from entering into the very situations that might give their partner the actual opportunity to prove their responsiveness (Kelley, 1979).
4. Conclusion What happens when one partner falls outside the other’s ‘‘cute-gory?’’ As the attractiveness-impaired author of the Time article lamented, such imbalances create an attribution dilemma for intimate partners and quizzical observers. Why would a seemingly superior partner want to stay with a seemingly inferior one? The commitment-insurance system solves this interdependence problem: if people’s standing relative to their partner does not easily support the inference that they are irreplaceable, the consequent activation of the implicit ‘‘if–then’’ contingencies of exchange can at least afford the inference that they are indispensable. Further untangling the long-term relationship consequences of such dynamics for low and high self-esteem people remains an important goal for future research.
ACKNOWLEDGMENTS We thank Ellen Berscheid for her insightful comments on how our model of risk regulation could be extended into a more comprehensive theory of interdependence. The preparation of this chapter was supported by a grant from the National Institute of Mental Health (MH 60105–02) awarded to S. Murray.
REFERENCES Ainsworth, M. D. S., Belha, M. C., Waters, E., and Wall, S. (1978). ‘‘Patterns of Attachment: Assessed in the Strange Situation and at Home.’’ Erlbaum, Hillsdale, NJ. Ayduk, O., Downey, G., Testa, A., Yen, Y., and Shoda, Y. (1999). Does rejection elicit hostility in rejection sensitive women? Soc. Cogn. 17, 245–271. Baldwin, M. W. (1992). Relational schemas and the processing of social information. Psychol. Bull. 112, 461–484. Baldwin, M. W., and Sinclair, L. (1996). Self-esteem and ‘‘if. . .then’’ contingencies of interpersonal acceptance. J. Pers. Soc. Psychol. 71, 1130–1141. Bargh, J. A., and Ferguson, M. J. (2000). Beyond behaviorism: On the automaticity of higher mental processes. Psychol. Bull. 126, 925–945. Baumeister, R. F. (1993). ‘‘Self-esteem: The Puzzle of Low Self-Regard.’’ Plenum Press, New York. Baumeister, R. F., Bratslavsky, E., Muraven, M., and Tice, D. M. (1998). Ego-depletion: Is the active self a limited resource? J. Pers. Soc. Psychol. 74, 1252–1265. Beach, S. R. H., Tesser, A., Fincham, F. D., Jones, D. J., Johnson, D., and Whitaker, D. J. (1998). Pleasure and pain in doing well together: An investigation of performance-related affect in close relationships. J. Pers. Soc. Psychol. 74, 923–938.
56
Sandra L. Murray and John G. Holmes
Beach, S. R. H., Whitaker, D. J., Jones, D. J., and Tesser, A. (2001). When does performance feedback prompt complementarity in romantic relationships? Pers. Relat. 8, 231–248. Berscheid, E. (1983). Emotion. In ‘‘Close relationships’’ (H. H. Kelley, E. Berscheid, A. Christensen, J. H. Harvey, T. L. Huston, G. Levinger, E. McClintock, L. A. Peplau, and D. R. Peterson, eds.), pp. 110–168. W. H. Freeman, New York. Berscheid, E., and Reis, H. T. (1998). Attraction and close relationships. In ‘‘The Handbook of Social Psychology’’ (D. T. Gilbert, S. T. Fiske, and G. Lindsey, eds.), Vol. 2, pp. 193–281. Mcgraw-Hill, Boston. Berscheid, E., Dion, K., Walster, E., and Walster, G. W. (1971). Physical attractiveness and dating choice: A test of the matching hypothesis. J. Exp. Soc. Psychol. 7, 173–189. Boon, S. D., and Holmes, J. G. (1999). Interpersonal risk and the evaluation of transgressions in close relationships. Pers. Relat. 6, 151–168. Bowlby, J. (1969). ‘‘Attachment and Loss’’ (Vol. 1: Attachment). Hogarth Press, London. Buss, D. M., and Barnes, M. (1986). Preferences in human mate selection. J. Pers. Soc. Psychol. 50, 559–570. Campbell, J. D. (1990). Self-esteem and clarity of the self-concept. J. Pers. Soc. Psychol. 59, 538–549. Campbell, L., Simpson, J. A., Kashy, D. A., and Fletcher, G. J. O. (2001). Ideal standards, the self, and flexibility of ideals in close relationships. Pers. Soc. Psychol. Bull. 27, 447–462. Campbell, L., Simpson, J. A., Boldry, J., and Kashy, D. A. (2005). Perceptions of conflict and support in romantic relationships: The role of attachment anxiety. J. Pers. Soc. Psychol. 88, 510–531. Clark, M. S., and Grote, N. K. (1998). Why aren’t indices of relationship costs always negatively related to indices of relationship quality? Pers. Soc. Psychol. Rev. 2, 2–17. Clark, M. S., Mills, J., and Powell, M. C. (1986). Keeping track of needs in communal and exchange relationships. J. Pers. Soc. Psychol. 51, 333–338. Collins, N. L., and Feeney, B. C. (2004). Working models of attachment shape perceptions of social support: Evidence from experimental and observational studies. J. Pers. Soc. Psychol. 87, 363–383. Davila, J., Karney, B. R., and Bradbury, T. N. (1999). Attachment change processes in the early years of marriage. J. Pers. Soc. Psychol. 76, 783–802. DeHart, T., Pelham, B., and Murray, S. L. (2004). Regulating dependency in close relationships: Implicit evaluations of significant others. Soc. Cognition 22, 126–146. Derrick, J. L., and Murray, S. L. (2007). Enhancing relationship perceptions by reducing felt inferiority: The moderating role of attachment style. Pers. Relat. 14, 531–549. Deutsch, M., and Krauss, R. M. (1965). ‘‘Theories in Social Psychology.’’ Basic Books, Inc., New York, NY. Dodgson, P. G., and Wood, J. V. (1998). Self-esteem and the cognitive accessibility of strengths and weaknesses after failure. J. Pers. Soc. Psychol. 75, 178–197. Drigotas, S. M., and Rusbult, C. E. (1992). Should I stay or should I go? A dependence model of breakups. J. Pers. Soc. Psychol. 62, 62–87. Dweck, C. S. (2000). ‘‘Self-Theories: Their Role in Motivation, Personality and Development.’’ Psychology Press, Philadelphia, PA. Feeney, B. C. (2007). The dependency paradox in close relationships: Accepting dependence promotes independence. J. Pers. Soc. Psychol. 92, 268–285. Feingold, A. (1988). Matching for attractiveness in romantic partners and same-sex friends: A meta-analysis and theoretical critique. Psychol. Bull. 104, 226–235. Fiske, S. T., and Neuberg, S. L. (1990). A continuum of impression formation, from category-based to individuating processes: Influences of information and motivation on attention and interpretation. In ‘‘Advances in experimental social psychology’’ (M. P. Zanna, ed.), Vol. 23, pp. 1–74. Academic Press, NY.
The Commitment-Insurance System
57
Gangestad, S. W., and Simpson, J. A. (2000). The evolution of human mating: Tradeoffs and strategic pluralism. Behav. Brain Sci. 23, 573–587. Gardner, W. L., Gabriel, S., and Hochschild, L. (2002). When you and I are ‘‘We,’’ You are not threatening: The role of self-expansion in social comparison. J. Pers. Soc. Psychol. 82, 239–251. Gilbert, P. (2005). Social mentalities: A biopsychosocial and evolutionary approach to social relationships. In ‘‘Interpersonal Cognition’’ (M. Baldwin, ed.), pp. 299–333. Guilford Press, New York, NY. Gilbert, D. T., Pelham, B. W., and Krull, D. S. (1988). On cognitive busyness: When person perceivers meet persons perceived. J. Pers. Soc. Psychol. 54, 733–740. Griffin, D. W., and Ross, L. (1991). Subjective construal, social inference and human misunderstanding. Adv. Exp. Soc. Psychol. 24, 319–359. Grote, N. K., and Clark, M. S. (2001). Perceiving unfairness in the family: Cause or consequence of marital distress. J. Pers. Soc. Psychol. 80, 281–293. Hall, C. S., and Lindzey, G. (1957). ‘‘Theories of Personality.’’ John Wiley and Sons, Inc., New York, NY. Holmes, J. G. (1981). The exchange process in close relationships: Microbehavior and macromotives. In ‘‘The Justice Motive in Social Behavior’’ (M. L. Lerner and S. Lerner, eds.), pp. 261–284. Plenum, New York. Holmes, J. G. (1991). Trust and the appraisal process in close relationships. In ‘‘Advances in Personal Relationships’’ (W. H. Jones and D. Perlman, eds.), Vol. 2, pp. 57–104. Jessica Kingsley, London. Holmes, J. G. (2002). Interpersonal expectations as the building blocks of social cognition: An interdependence theory perspective. Pers. Relat. 9, 1–26. Holmes, J. G., and Murray, S. L. (2007). Felt security as a normative resource: Evidence for an elemental risk regulation system? Psychol. Inquiry 18, 163–167. Impett, E. A., Gable, S. L., and Peplau, L. A. (2005). Giving up and giving in: The costs and benefits of daily sacrifice in intimate relationships. J. Pers. Soc. Psychol. 89, 327–344. Kelley, H. H. (1967). Attribution theory in social psychology. In ‘‘Nebraska Symposium on Motivation’’ (D. Levine, ed.), Vol. 15, pp. 192–240. University of Nebraska Press, Lincoln. Kelley, H. H. (1979). ‘‘Personal Relationships: Their Structures and Processes.’’ Erlbaum, Hillsdale, NJ. Kelley, H. H., Holmes, J. G., Kerr, N., Reis, H., Rusbult, C., and Van Lange, P. A. (2003). ‘‘An Atlas of Interpersonal Situations.’’ Cambridge Press, Cambridge, UK. Kruger, J. (1999). Lake Wobegon Be Gone! The ‘‘Below-Average Effect’’ and the egocentric nature of comparative ability judgments. J. Pers. Soc. Psychol. 77, 221–232. Kunda, Z. (1990). The case for motivated reasoning. Psychol. Bull. 108, 480–498. Kunda, Z., and Spencer, S. J. (2003). When do stereotypes come to mind and when do they color judgment? A goal-based theoretical framework for stereotype activation and application. Psychol. Bull. 129, 522–544. Laurenceau, J. P., Barrett, L. F., and Pietromonaco, P. R. (1998). Intimacy as an interpersonal process: The importance of self-disclosure, partner disclosure, and perceived partner responsiveness in interpersonal exchanges. J. Pers. Soc. Psychol. 74, 1238–1251. Leary, M. R., and Baumeister, R. F. (2000). The nature and function of self-esteem: Sociometer theory. In ‘‘Advances in Experimental Social Psychology’’ (M. P. Zanna, ed.), Vol. 32, pp. 2–51. Academic Press, San Diego, CA. Leary, M. R., Tambor, E. S., Terdal, S. K., and Downs, D. L. (1995). Self-esteem as an interpersonal monitor: The sociometer hypothesis. J. Pers. Soc. Psychol. 68, 518–530. Leary, M. R., Springer, C., Negel, L., Ansell, E., and Evans, K. (1998). The causes, phenomenology, and consequences of hurt feelings. J. Pers. Soc. Psychol. 74, 1225–1237.
58
Sandra L. Murray and John G. Holmes
Levinger, G. (1976). A social psychological perspective on marital dissolution. J. Soc. Issues 32, 21–49. Lykken, D. T., and Tellegen, A. (1993). Is human mating adventitious or the result of lawful choice? A twin study of mate selection. J. Pers. Soc. Psychol. 65, 56–68. MacDonald, G., and Jessica, M. (2006). Family approval as a constraint in dependency regulation: Evidence from Australia and Indonesia. Pers. Relat. 13, 183–194. MacDonald, G., and Leary, M. R. (2005). Why does social exclusion hurt? The relationship between social and physical pain. Psychol. Bull. 131, 202–223. Markus, H., and Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychol. Rev. 98, 224–253. Matouschek, N., and Rasul, I. (in press). The economics of the marriage contract: Theories and evidence. J. Law Econ. Mikulincer, M., and Shaver, P. R. (2001). Attachment theory and intergroup bias: Evidence that priming the secure base schema attenuates negative reactions to outgroups. J. Pers. Soc. Psychol. 81, 97–115. Mikulincer, M., and Shaver, P. R. (2003). The attachment behavioral system in adulthood: Activation, psychodynamics, and interpersonal processes. In ‘‘Advances in Experimental Social Psychology’’ (M. Zanna, ed.), Vol. 35, pp. 52–153. Academic Press, New York. Mikulincer, M., Gillath, O., Halevy, V., Avihou, N., Avidan, S., and Eshkoli, N. (2001). Attachment theory and reactions to others’ needs: Evidence that activation of the sense of attachment security promotes empathic responses. J. Pers. Soc. Psychol. 81, 1205–1224. Mills, J., Clark, M. S., Ford, T. E., and Johnson, M. (2004). Measurement of communal strength. Pers. Relat. 11, 213–230. Murray, S. L. (1999). The quest for conviction: Motivated cognition in romantic relationship. Psychol. Inquiry 10, 23–34. Murray, S. L., Holmes, J. G., and Griffin, D. (1996a). The benefits of positive illusions: Idealization and the construction of satisfaction in close relationships. J. Pers. Soc. Psychol. 70, 79–98. Murray, S. L., Holmes, J. G., and Griffin, D. W. (1996b). The self-fulfilling nature of positive illusions in romantic relationship: Love is not blind, but prescient. J. Pers. Soc. Psychol. 71, 1155–1180. Murray, S. L., Holmes, J. G., MacDonald, G., and Ellsworth, P. (1998). Through the looking glass darkly? When self-doubts turn into relationship insecurities. J. Pers. Soc. Psychol. 75, 1459–1480. Murray, S. L., Holmes, J. G., and Griffin, D. W. (2000). Self-esteem and the quest for felt security: How perceived regard regulates attachment processes. J. Pers. Soc. Psychol. 78, 478–498. Murray, S. L., Holmes, J. G., Griffin, D. W., Bellavia, G., and Rose, P. (2001). The mismeasure of love: How self-doubt contaminates relationship beliefs. Pers. Soc. Psychol. Bull. 27, 423–436. Murray, S. L., Holmes, J. G., Bellavia, G., Griffin, D. W., and Dolderman, D. (2002a). Kindred spirits? The benefits of egocentrism in close relationships. J. Pers. Soc. Psychol. 82, 563–581. Murray, S. L., Rose, P., Bellavia, G., Holmes, J., and Kusche, A. (2002b). When rejection stings: How self-esteem constrains relationship-enhancement processes. J. Pers. Soc. Psychol. 83, 556–573. Murray, S. L., Bellavia, G., Rose, P., and Griffin, D. (2003a). Once hurt, twice hurtful: How perceived regard regulates daily marital interaction. J. Pers. Soc. Psychol. 84, 126–147. Murray, S. L., Griffin, D. W., Rose, P., and Bellavia, G. (2003b). Calibrating the sociometer: The relational contingencies of self-esteem. J. Pers. Soc. Psychol. 85, 63–84.
The Commitment-Insurance System
59
Murray, S. L., Rose, P., Holmes, J. G., Derrick, J., Podchaski, E., Bellavia, G., and Griffin, D. (2005). Putting the partner within reach: A dyadic perspective on felt security in close relationships. J. Pers. Soc. Psychol. 88, 327–347. Murray, S. L., Griffin, D. W., Rose, P., and Bellavia, G. (2006a). For better or worse? Selfesteem and the contingencies of acceptance in marriage. Pers. Soc. Psychol. Bull. 32, 866–882. Murray, S. L., Holmes, J. G., and Collins, N. L. (2006b). Optimizing assurance: The risk regulation system in relationships. Psychol. Bull. 132, 641–666. Murray, S. L., Aloni, A., Holmes, J. G., Derrick, J. L., Anthony, D. B., and Leder, S. (2007). Fostering partner dependence as commitment insurance: The implicit contingencies of exchange in close relationships. Manuscript under editorial review. State University of New York, Buffalo. Murray, S. L., Derrick, J. L., Leder, S., and Holmes, J. G. (2008). Balancing connectedness and self-protection goals in close relationships: A levels of processing perspective on risk regulation. J. Pers. Soc. Psychol. 94, 429–459. Olson, M. A., and Fazio, R. H. (in press). Implicit and explicit measures of attitudes: The perspective of the MODE model. In ‘‘Attitudes: Insights from the new implicit measures’’ (R. E. Petty, R. H. Fazio, and P. Brinol, eds.). Mahwah, Erlbaum, NJ. O’Mahen, H. A., Beach, S. R. H., and Tesser, A. (2000). Relationship ecology and negative communication in romantic relationships: A self-evaluation maintenance perspective. Pers. Soc. Psychol. Bull. 26, 1343–1352. Pemberton, M., and Sedikides, C. (2001). When do individuals help close others improve? The role of information diagnosticity. J. Pers. Soc. Psychol. 81, 234–246. Pierce, T., and Lydon, J. (1998). Priming relational schemas: Effects of contextually activated and chronically accessible interpersonal expectations on responses to a stressful event. J. Pers. Soc. Psychol. 75, 1441–1448. Reis, H. T., Clark, M. S., and Holmes, J. G. (2004). Perceived partner responsiveness as an organizing construct in the study of intimacy and closeness. In ‘‘Handbook of Closeness and Intimacy’’ (D. Mashek and A. P. Aron, eds.), pp. 201–225. Lawrence Erlbaum, Mahweh, NJ. Rempel, J. K., Holmes, J. G., and Zanna, M. P. (1985). Trust in close relationships. J. Pers. Soc. Psychol. 49, 95–112. Rholes, W. S., Simpson, J. A., and Orina, M. M. (1999). Attachment and anger in an anxiety-provoking situation. J. Pers. Soc. Psychol. 76, 940–957. Rubin, Z. (1973). ‘‘Liking and Loving: An Invitation to Social Psychology.’’ Holt, Rinehart and Winston, New York. Rudich, E. A., and Vallacher, R. R. (1999). To belong or to self-enhance? Motivational bases for choosing interaction partners. Pers. Soc. Psychol. Bull. 25, 1387–1404. Rusbult, C. E., and Van Lange, P. A. M. (2003). Interdependence, interaction, and relationships. Annu. Rev. Psychol. 54, 351–375. Rusbult, C. E., Martz, J. M., and Agnew, C. R. (1998). The investment model scale: Measuring commitment level, satisfaction level, quality of alternatives, and investment size. Pers. Relat. 5, 357–391. Seligman, C., Fazio, R. H., and Zanna, M. P. (1980). Effects of salience of external rewards on liking and loving. J. Pers. Soc. Psychol. 38, 453–460. Simpson, J. A. (1987). The dissolution of romantic relationships: Factors involved in relationship stability and emotional distress. J. Pers. Soc. Psychol. 53, 683–692. Simpson, J. A., Rholes, W. S., and Phillips, D. (1996). Conflict in close relationships: An attachment perspective. J. Pers. Soc. Psychol. 71, 899–914. Simpson, J. A., Ickes, W., and Grich, J. (1999). When accuracy hurts: Reactions of anxiousambivalent dating partners to a relationship-threatening situation. J. Pers. Soc. Psychol. 76, 754–769.
60
Sandra L. Murray and John G. Holmes
Stapel, D. A., and Blanton, H. (2004). From seeing to being: Subliminal social comparisons affect implicit and explicit self-evaluations. J. Pers. Soc. Psychol. 87, 468–481. Stapel, D. A., and Tesser, A. (2001). Self-activation increases social comparison. J. Pers. Soc. Psychol. 81, 742–750. Taylor, S. E., and Brown, J. D. (1988). Illusion and well-being: A social psychological perspective on mental health. Psychol. Bull. 103, 193–210. Thibaut, J. W., and Kelley, H. H. (1959). ‘‘The Social Psychology of Groups.’’ Wiley, New York. Tooby, J., and Cosmides, L. (1996). Friendship and the banker’s paradox: Other pathways to the evolution of adaptations for altruism. Proc. Br. Acad. 88, 119–143. Twenge, J. M., Baumeister, R. F., Tice, D. M., and Stucke, T. S. (2001). If you can’t join them, beat them: Effects of social exclusion on aggressive behavior. J. Pers. Soc. Psychol. 81, 1058–1069. Vohs, K. D., Mead, N. L., and Goode, M. R. (2006). The psychological consequences of money. Science 314, 1154–1156. Waller, W. (1938). ‘‘The Family: A Dynamic Interpretation.’’ Cordon, Co, New York. Wegner, D. M. (2002). ‘‘The Illusion of Conscious Will.’’ MIT Press, Cambridge, MA. Wegner, D. M., Giuliano, T., and Hertel, P. T. (1985). Cognitive interdependence in close relationships. In ‘‘Compatible and Incompatible Relationships’’ (W. Ickes, ed.), pp. 253–276. Springer-Verlag, New York. White, G. L. (1980). Physical attractiveness and courtship progress. J. Pers. Soc. Psychol. 39, 660–668. Wieselquist, J., Rusbult, C. E., Foster, C. A., and Agnew, C. R. (1999). Commitment, prorelationship behavior, and trust in close relationships. J. Pers. Soc. Psychol. 77, 942–966.
C H A P T E R
T W O
Warmth and Competence as Universal Dimensions of Social Perception: The Stereotype Content Model and the BIAS Map Amy J. C. Cuddy,* Susan T. Fiske,† and Peter Glick‡ Contents 1. Introduction 1.1. Defining warmth and competence 1.2. The stereotype content model and the BIAS map 2. Warmth and Competence as Fundamental Dimensions of Social Perception 2.1. Interpersonal perception 2.2. Intergroup perception 2.3. Primacy of warmth 3. Social Structural Roots of Warmth and Competence Judgments 3.1. Why social status should predict competence judgments 3.2. Why competition should predict warmth judgments 3.3. US tests of SCM structural hypotheses 3.4. Cross-cultural tests of SCM structural hypotheses 3.5. Converging theory and evidence 4. Emotional, Behavioral, and Attributional Consequences 4.1. Prejudiced emotions: Admiration, contempt, envy, and pity 4.2. Behaviors: Active and passive, facilitation and harm 5. Spotlight on Ambivalent Combinations: Warm–Incompetent and Competent–Cold 5.1. Pitying prejudice: Warm but incompetent 5.2. Envious prejudice: Competent but cold 6. Current and Future Directions and Summary 6.1. Current and future directions 6.2. Summary
* { {
62 65 66 71 71 74 89 92 94 94 96 98 99 102 102 107 119 119 124 129 130 137
Kellogg School of Management, Northwestern University, Evanston, IL 60208 Department of Psychology, Princeton University, Princeton, NJ 08540 Department of Psychology, Lawrence University, Appleton, WI 54912
Advances in Experimental Social Psychology, Volume 40 ISSN 0065-2601, DOI: 10.1016/S0065-2601(07)00002-0
#
2008 Elsevier Inc. All rights reserved.
61
62
Amy J. C. Cuddy et al.
Acknowledgments References
137 137
Abstract The stereotype content model (SCM) defines two fundamental dimensions of social perception, warmth and competence, predicted respectively by perceived competition and status. Combinations of warmth and competence generate distinct emotions of admiration, contempt, envy, and pity. From these intergroup emotions and stereotypes, the behavior from intergroup affect and stereotypes (BIAS) map predicts distinct behaviors: active and passive, facilitative and harmful. After defining warmth/communion and competence/agency, the chapter integrates converging work documenting the centrality of these dimensions in interpersonal as well as intergroup perception. Structural origins of warmth and competence perceptions result from competitors judged as not warm, and allies judged as warm; high status confers competence and low status incompetence. Warmth and competence judgments support systematic patterns of cognitive, emotional, and behavioral reactions, including ambivalent prejudices. Past views of prejudice as a univalent antipathy have obscured the unique responses toward groups stereotyped as competent but not warm or warm but not competent. Finally, the chapter addresses unresolved issues and future research directions.
1. Introduction Sit in any airport, train station, or bus depot, then watch and listen. The sheer ethnic variety reflected in the visual and aural parade staggers the mind. This variety requires adjustments on all sides, as the world’s peoples encounter each other. The circumstances of migration generate particular images of distinct groups, which in turn create distinct feelings and impulses. Mapping this new geography of intergroup and interpersonal contact is the business of this chapter. Migration and ethnic stereotypes are not new, of course,but the circumstances of ethnic groups shift with history, and with these structural changes come changes in patterns of stereotypes, prejudices, and discrimination. For example, ethnic Chinese who migrated to the United States in the mid-19th century to help build railroads were seen as animal laborers, neither especially competent nor especially trustworthy. After they were expelled and new Chinese migrants appeared in the 20th century, stereotypes changed accordingly to reflect entrepreneurs and technical experts, who are viewed now as perhaps excessively competent, but still not warm. As a result of modern globalization, encounters among people from different social categories are increasingly common. What is more, as the number and range of social categories in society has increased, so has
Warmth and Competence as Universal Dimensions of Social Perception
63
the gap between groups at the top and the bottom, creating further categorical divides. This increase is especially dramatic in the United States (Massey, 2007). North American prejudice researchers have long focused on Black–White intergroup relations, but this model does not apply to all the varieties of differences that people encounter daily, on personal and societal levels. People’s ordinary lives require forming efficient and effective impressions of incredible numbers of other individuals. In examining how people make sense of each other, both as individuals and as group members, we have discovered two dimensions that differentiate groups and individuals. These dimensions appear to be both fundamental and universal, as we will argue here. This chapter presents a framework synthesizing research on perceptions of individuals and groups. The core of this synthesis is the observation that judgments of warmth and competence underlie perceptions of others, driving perceivers’ emotional and behavioral reactions, all resulting from social structural relationships. We argue that these dimensions are universal because they assess questions about others that are both basic and adaptive. Further, we show how judgments of warmth and competence follow from the structure of relations between individuals or between groups: specifically, their form of interdependence (cooperative vs competitive) and status relations. These insights are framed in terms of the stereotype content model (SCM; Fiske et al., 2002b) and a recent extension of this theory, the behavior from intergroup affect and stereotypes (BIAS) map (Cuddy et al., 2007). Although both the SCM and BIAS map are oriented toward explaining intergroup relations, we show here how they extend to interpersonal relations. The functional significance and universality of the warmth and competence dimensions result from their correspondence to two critical questions basic to surviving and thriving in a social world. First, actors need to anticipate others’ intentions toward them; the warmth dimension—comprising such traits as morality, trustworthiness, sincerity, kindness, and friendliness— assesses the other’s perceived intent in the social context. Second, both in importance and temporal sequence, actors need to know others’ capability to pursue their intentions; the competence dimension—comprising such traits as efficacy, skill, creativity, confidence, and intelligence—relates to perceived capability to enact intent. Motivationally, warmth represents an accommodating orientation that profits others more than the self, whereas competence represents self-profitable traits related to the ability to bring about desired events (Peeters, 1983). In short, actors distinguish individuals and groups according to their likely impact on the self or ingroup as determined by perceived intentions and capabilities. Warmth and competence dimensions have consistently emerged in both classic and contemporary studies of person perception (Asch, 1946; Rosenberg et al., 1968; Wojciszke et al., 1998), social-value orientations
64
Amy J. C. Cuddy et al.
(e.g., self- and other-profitability; Peeters, 2002), construals of others’ behaviors (Wojciszke, 1994), and voters’ ratings of political candidates in the United States (Abelson et al., 1982; Kinder and Sears, 1981) and Poland (Wojciszke and Klusek, 1996). Although often under the guise of different labels, which we review below, the warmth and competence dimensions also describe national stereotypes (e.g., morality and competence, Alexander et al., 1999; Phalet and Poppe, 1997; Poppe, 2001; Poppe and Linssen, 1999), characterize the four poles of Wiggins’s interpersonal circumplex of behaviors (i.e., agreeableness and extroversion; Wiggins, 1979), and surface in numerous in-depth analyses of prejudices toward specific social groups (e.g., Clausell and Fiske, 2005; Eckes, 2002; Glick, 2002; Glick and Fiske, 1996; Hurh and Kim, 1989; Kitano and Sue, 1973; Lin et al., 2005; Spence and Helmreich, 1979; cf., Altermatt et al., 2003). More recently, work on the SCM and the BIAS map has documented the centrality of warmth and competence as dimensions of group stereotypes, identified their origins in social structural relations, and delineated their emotional and behavioral consequences. By integrating the SCM and BIAS map with other relevant theory and research from interpersonal and intergroup perception, we hope to achieve three overarching goals in this chapter: (1) to show the centrality of warmth and competence as dimensions of social judgment across varied targets (both individuals and groups), perceivers, and cultures; (2) to locate the origins of warmth and competence judgments in social structural variables; and (3) to assess the emotional and behavioral consequences of warmth and competence judgments. Throughout the chapter, we review our research on this topic, including more than three dozen correlational and experimental studies from seventeen nations. Ultimately, we aim not only to present an integrative review of the overwhelming evidence of the universality of these two dimensions in social perception, but also to provide a common framework for identifying the origins and predicting the social consequences of warmth and competence judgments. We organize this chapter into the following sections. First, we address definitional issues and provide a brief background and summary of the SCM and the BIAS map. Second, we review and integrate converging theory and evidence documenting the centrality of warmth and competence dimensions in interpersonal as well as intergroup perception, including our own research, which has provided extensive evidence of their significance in intergroup relations (Cuddy et al., 2007, in press; Fiske et al., 1999, 2002b, 2007; Glick and Fiske, 2001b). Third, we turn to a discussion of the structural origins of warmth and competence perceptions. We propose that people viewed as competitors are judged as lacking warmth, whereas people viewed as allies are judged as warm; people viewed as high status are judged as competent, whereas people viewed as low status are judged as incompetent. Fourth, we examine the social outcomes of warmth and
Warmth and Competence as Universal Dimensions of Social Perception
65
competence judgments, proposing and reviewing empirical support for the existence of systematic patterns of cognitive, emotional, and behavioral reactions to perceiving others as competent versus incompetent, and warm versus cold. In the fifth section, we discuss ambivalent patterns of prejudice in greater detail. Built upon the paradigm of prejudice as a univalent antipathy, past research has obscured the unique patterns of social responses toward groups that are stereotyped as competent but not warm or as warm but not competent. In the sixth and final section, we summarize and discuss unresolved issues and future research directions.
1.1. Defining warmth and competence Many different labels describe what boil down to virtually the same two dimensions. Our warmth scales have included good-natured, trustworthy, tolerant, friendly, and sincere. Our competence scales have included capable, skillful, intelligent, and confident. Wojciszke et al.’s (1998) terms are morality and competence, but the moral traits include fair, generous, helpful, honest, righteous, sincere, tolerant, and understanding, which overlap entirely with the warmth-trustworthiness dimension identified elsewhere. (There is no dispute about the competence label, but those traits include clever, competent, creative, efficient, foresighted, ingenious, intelligent, and knowledgeable.) Peeters’s (1983, 2002) distinction between self-profitable traits—those that directly benefit or harm the trait possessor (e.g., intelligence, inefficiency)— versus other-profitable traits—those that directly benefit or harm others in the trait possessor’s social world (e.g., trustworthy, hostile)—set the precedent for Wojciszke’s work and essentially agrees with our usage of competence and warmth. Slightly different but still compatible are the communion and agency dimensions originated in personality psychology by Bakan (1956) who noted, in a philosophical context, two fundamental modalities in the existence of living beings, agency for the existence of the organism as an individual, and communion of the individual with belonging to some larger organism. The gender literature picked up this distinction because the dimensions related respectively to femininity and masculinity (Abele, 2003; Carlson, 1971; Spence et al., 1979; White, 1979). Communion and agency since have been frequently linked to gender stereotypes (e.g., Eagly and Steffen, 1984), social motives (e.g., McAdams et al., 1984), sex differences (Buss, 1981), and more (see Rudman and Glick, 2008, for a review). Although acknowledging the important gendered flavor of these two dimensions, and their relations to warmth and competence, we go beyond gender in the SCM. What’s more, whereas communion closely resembles our warmth dimension, agency does not fully capture competence, because agency focuses more on taking effective action. Competence entails the possession of skills, talents, and capability, but it can take the form of
66
Amy J. C. Cuddy et al.
potential action as well as actual action, so we prefer to emphasize competence rather than agency. We use the terms warmth and competence for simplicity, but we view them as closely related to communion and agency. To demonstrate the high redundancy across these variously named dimensions, Abele and Wojciszke (2007) asked participants to rate a list of 300 trait terms, which were selected to represent all of the above-named dimensions in addition to the collectivism/individualism and the Big Five, on the related constructs of agency/communion, morality/competence, collectivism/individualism, and femininity/masculinity. They found that a two factor-solution, with one factor comprising the traits representing agency, individualism, masculinity, and competence and the other dimension comprising the traits representing communion, collectivism, femininity, and morality, accounted for almost 90% of the variance. Participants in the same study also rated these traits on the extent to which they reflected selfinterest (i.e., does possessing the trait benefit or harm the self ) or otherinterest (i.e., does possessing the trait benefit or harm others). As expected, ratings of other-interest positively correlated with morality and ratings of self-interest positively correlated with agency (Abele and Wojciszke, 2007). Finally, Osgood et al.’s semantic differential (1957) defined dimensions that might seem similar to warmth and competence. They identified evaluation, potency, and activity (EPA) as central dimensions of language and attitudes. Commentators have often wondered whether evaluation corresponds to warmth and potency to competence. Our answer is no, not exactly. First, both warmth and competence have evaluative components; it is better to be warm, trustworthy, and helpful than not. Similarly, it is better to be competent and skilled than not. Second, both warmth and competence can be more or less potent as well: one can be strong and warm or weak and warm, and the same for competence. Third, as for activity, it collapses with potency into a single factor, called ‘‘dynamism,’’ at least in person perception (Osgood et al., 1957). Fourth, recent data comparing the evaluation, potency, and agency (EPA) dimensions and the SCM show that they are not redundant (Capozza et al., 2007). Overall, we would suggest that the evaluation x potency/activity space probably operates at a 45 rotation to our space. We return to this point in our discussion of future directions (section 6.1).
1.2. The stereotype content model and the BIAS map 1.2.1. Guiding principles of intergroup BIAS The SCM and the BIAS map integrate several interrelated broad principles of intergroup bias, derived from work on its functional, motivational, and social-cognitive roots. The first principle is that many groups do not receive a one-dimensional, hostile type of prejudice. Recent work by us and others converges on the view that prejudice is both group- and context-dependent and can simultaneously include both negative and subjectively positive
Warmth and Competence as Universal Dimensions of Social Perception
67
responses. For example, according to Cottrell and Neuberg’s (2005) sociofunctional approach, different groups (e.g., gay men versus MexicanAmericans) elicit distinct classes of perceived threats (e.g., to health versus property, respectively), which evoke functionally relevant, distinct emotion profiles (e.g., disgust and pity versus fear and anger, respectively; see also Esses et al., 2001; Stephan and Renfro, 2002). On the basis of internation biases, Alexander and colleagues proposed a functional model called Image Theory (1999, 2005; Brewer and Alexander, 2002), which asserts that actors make three appraisals of outgroups: intergroup goal compatibility, relative status, and power to attain goals. For each outgroup, these appraisals (e.g., incompatible goals, equal status, equal power) induce specific action tendencies (attack) and emotions (anger), generating distinct outgroup ‘‘images’’ (e.g., hostile, opportunistic enemy). Like our model, these other approaches also suggest that the contents of biases vary across groups and situations in ways that cannot be explained by a view of prejudice as an undifferentiated antipathy (Esses et al., 2001; Mackie et al., 2000; Stephan and Stephan, 2000). Second, the contents of the three psychological components of bias—cognitions (stereotypes), affect (emotional prejudices), and behavior (discrimination)—operate in synchrony with one another, an idea that is firmly grounded, for example, in appraisal theories of emotion (Frijda, 1986; Roseman, 1984; Scherer, 1988; Smith and Ellsworth, 1985). Lazarus and Folkman (1984) define cognitive appraisals as assessments of the implications of the others’ behavior for the self (or ingroup). Situations and their corresponding cognitive appraisals elicit discrete patterns of emotions, which in turn trigger specific behavioral responses (e.g., offensive action) adapted to cope with the potential threat the other individual or group poses (Frijda et al., 1989; Izard, 1991; Izard et al., 1993; Roseman et al., 1994), a view that is even supported by evidence of neuroanatomical emotion pathways linked to specific behaviors (Panksepp, 2000). For example, according to intergroup emotions theory (IET), an appraisal-based approach to intergroup relations, appraising the ingroup as stronger than a hostile outgroup elicits anger, which leads to offensive action tendencies, whereas appraising the outgroup as stronger results in fear (Mackie et al., 2000). These patterns of relationships have been documented at both the interpersonal and intergroup levels (Devos et al., 2002; Dijker, 1987; Dijker et al., 1996b; Mackie and Smith, 1998; Mackie et al., 2000; cf. Fiske et al., 2002b). Attitude theories also posit that the affective, cognitive, and behavioral correlates of evaluation tend to converge, depending on circumstances (e.g., Ajzen, 2001). Third, emotions mediate the effects of cognitions on discrimination. Emotions can be viewed as the engines that drive behavior (Tomkins, as cited in Zajonc, 1998) and ‘‘changes in action readiness’’ (Zajonc, 1998, p. 466). Affect often mediates the effects of cognition on behavior, which is a central tenet of appraisal theories of emotion, including IET, which propose a
68
Amy J. C. Cuddy et al.
cognitive appraisal ! emotion ! behavior sequence (Frijda et al., 1989; Mackie et al., 2000; Roseman et al., 1994). Because of this more direct link to behavior, past research suggests that affect often predicts discriminatory behavior better than stereotypes (Dovidio et al., 1996, 2002; Esses and Dovidio, 2002; Esses et al., 1993; Schu¨tz and Six, 1996; Stangor et al., 1991; Talaska et al., 2007). Although we agree that emotions mediate the effects of cognitions (including stereotyped beliefs) on behavior, our theoretical perspective and findings suggest that past research has underestimated the effects of cognition by failing to appreciate how the content of stereotypes on warmth and competence dimensions together create distinct patterns of bias (cognitively, emotionally, and behaviorally). 1.2.2. Basic tenets of the SCM 1.2.2.1. Centrality of warmth and competence The SCM’s first tenet is that perceived warmth and competence underlie and differentiate group stereotypes. Although specific group stereotypes have some idiosyncratic content (e.g., the notion that Black people are ‘‘rhythmic’’), underlying such beliefs are more general themes organized along warmth and competence dimensions. Although we do not discount the importance of specific, historically conditioned beliefs about groups, we suggest that much of the variance in stereotypes of groups is accounted for by the more basic warmth and competence dimensions. As we review in detail below, our research consistently reveals differentiated clusters of high versus low warmth and competence stereotypes across widely varied target groups, such as occupations, nationalities, ethnicities, socioeconomic groups, religions, and gender subtypes. Moreover, these patterns appear to be (a) universal features of social perception, supported in diverse US samples (Fiske et al., 2002b, 1999), including a representative national sample (Cuddy et al., 2007), and in 17 other nations (Cuddy et al., in press) and (b) predicted by the structural relations between groups (Cuddy et al., 2007, in press; Fiske and Cuddy, 2006; Fiske et al., 1999, 2002b). 1.2.2.2. Ambivalent stereotypes The SCM posits that many groups will receive ambivalent stereotypes, comprising a positive evaluation on one dimension and a negative evaluation on the other. In other words, many outgroups are viewed as competent but not warm (e.g., Asians, Jews, the rich), or as warm but not competent (e.g., the disabled, the elderly, housewives). Importantly, subjectively positive stereotypes on one dimension typically do not contradict prejudice or reduce discrimination but reinforce unflattering stereotypes on the other dimension and justify unequal treatment. Although some groups (homeless, poor, welfare recipients) are stereotyped as low on both warmth and competence, only reference groups—ingroups (e.g., students) and societal prototype groups (e.g., Whites, middle-class)— are perceived to be both warm and competent (at least in Western cultures;
69
Warmth and Competence as Universal Dimensions of Social Perception
High
Elderly
Housewives Christians HC-HW Middle-class Americans Black professionals Irish
Warmth
Retarded LC-HW Disabled
Whites British
Poor Welfare recipients
LC-LW Turks Feminists
Arabs
Jews
HC-LW Asians
Rich
Homeless Low Low
High Competence
Figure 2.1 Stereotype content model warmth competence space mapping social groups, in a representative sample survey of American adults. Source: Cuddy et al. (2007). Reproduced by permission.
Cuddy et al., in press). (See Fig. 2.1 for the relative locations of various groups.) 1.2.2.3. Social structural origins of perceived warmth and competence According to the SCM, the origins of perceived warmth and competence lie in social structural variables, namely competition and status, such that non-competitive others are judged to be warm, whereas competitive others are not; and high-status others are judged to be competent, whereas lowstatus others are not. These relationships have been replicated in virtually all of the studies of actual groups cited above (Cuddy et al., 2007, in press; Fiske et al., 1999, 2002b). Further, the same principles hold for experimentally constructed groups (Caprariello et al., 2007; Oldmeadow and Fiske, 2007) and in perceptions of individuals (Russell and Fiske, 2007), which we discuss in more detail below. 1.2.2.4. Warmth and competence judgments elicit signature emotions The SCM proposes that the four combinations of high versus low warmth and competence judgments elicit four unique emotional responses: admiration, contempt, envy, and pity (Fiske et al., 2002a,b). Specifically, groups stereotyped as warm and competent (e.g., ingroups)—elicit admiration.
70
Amy J. C. Cuddy et al.
Groups stereotyped as incompetent and cold (e.g., homeless people) elicit contempt. Groups stereotyped as competent but not warm (e.g., Asians) elicit envy. Groups stereotyped as warm but not competent (e.g., elderly people) elicit pity. These proposals have been supported using both correlational and experimental methods, as well as cross-cultural comparisons (Cuddy et al., 2004, in press; Fiske et al., 2002a,b). We discuss the theoretical underpinnings of these relationships later. 1.2.3. Basic tenets of the BIAS map The BIAS map (Cuddy et al., 2007; Fig. 2.2) extends the SCM by considering the behavioral outcomes of warmth and competence evaluations in social interactions. It proposes that the four combinations of high versus low warmth and competence elicit four unique patterns of behavioral responses: active facilitation (e.g., helping), active harm (e.g., harassing), passive facilitation (e.g., convenient cooperation), and passive harm (e.g., neglecting). 1.2.3.1. Warmth and competence judgments elicit active and passive behaviors Because the warmth dimension is primary (due to its perceived link to others’ intentions), perceived warmth predicts active behaviors: groups judged as warm elicit active facilitation (i.e., help), whereas those judged as lacking warmth elicit active harm (i.e., attack). The competence dimension, being secondary (because it assesses others’ capability to carry out intentions), predicts passive behaviors: groups judged as competent High
Active facilitation
Passive facilitation
pt
em
vy
nt Co
Low Low
Active harm Competence
En
Warmth
n
tio
ira
Pi ty
m
Ad
Passive harm
High
Figure 2.2 Stereotype content model predictions for emotions and BIAS map predictions for behaviors in the warmth by competence space. Source: Cuddy et al. (2007). Reproduced by permission.
Warmth and Competence as Universal Dimensions of Social Perception
71
elicit passive facilitation (i.e., obligatory association, convenient cooperation), whereas those judged as lacking competence elicit passive harm (i.e., neglect, ignoring). In short, distinct types of discrimination follow each warmth-by-competence combination. 1.2.3.2. Discrete emotions elicit specific behavior patterns The BIAS map also connects the four kinds of emotions—corresponding to the four warmth–competence combinations—to predicted behaviors. Specifically, admired (i.e., competent and warm) groups elicit both active and passive facilitation, that is, both helping and associating. Resented, envied (i.e., incompetent and cold) groups elicit both kinds of harm: active attack and passive neglect. The ambivalent combinations are more volatile: pitied groups elicit both active helping and passive neglect, aptly describing patronizing behavior toward older and disabled people, who may sometimes be overhelped and other times neglected. Being institutionalized also can combine active help and passive neglect. In contrast, envied groups elicit both passive association and active harm. For instance people may shop at the stores of entrepreneurial outsiders, ‘‘going along to get along,’’ but under societal breakdown may attack and loot these same shops. Koreans in Los Angeles, Tutsis in Rwanda, Chinese in Indonesia, and Jews in Europe have each experienced such treatment. Consistent with appraisal theories of emotion, the BIAS map predicts that emotions are the proximal cause of social behaviors, a finding reflected in meta-analyses of emotional prejudices and cognitive stereotypes as predictors of discrimination (Dovidio et al., 1996; Talaska et al., 2007). The BIAS map predicts that emotions more strongly and directly predict behaviors because they mediate the link from warmth and competence judgments to behaviors. We later present both correlational and experimental support for these patterns at both the intergroup (e.g., Cuddy et al., 2004, 2007) and interpersonal levels (Asbrock and Cuddy, 2008; Talaska et al., 2007).
2. Warmth and Competence as Fundamental Dimensions of Social Perception 2.1. Interpersonal perception 2.1.1. Centrality of warmth and competence The centrality of warmth and competence is well documented in the area of interpersonal perception, going back over half a century. Perhaps the first empirical suggestion of the importance of these two dimensions came from Asch’s (1946) classic studies, in which the inclusion of the trait ‘‘warm’’ versus ‘‘cold’’ shaped people’s ‘‘Gestalt impressions’’ of a person described by a list of competence-related characteristics. When the target person
72
Amy J. C. Cuddy et al.
described as competent and capable was also described as ‘‘warm,’’ participants perceived him as wise. If the same competent, capable person was instead described as ‘‘cold,’’ participants perceived him as sly. Although Asch’s work intuited the centrality of these two dimensions, extensions of his work more clearly demonstrated their centrality. Rosenberg et al. (1968) asked participants to sort 64 traits into categories that were likely to be found in an individual person. Multidimensional scaling of the results pointed to two nearly orthogonal dimensions as best representing the general trait structure of person judgments. Intellectual good/bad, akin to competence, included such traits as determined, skillful, industrious, intelligent, and scientific; social good/bad, akin to warmth, included such traits as warm, honest, helpful, good-natured, sincere, and tolerant. As Fig. 2.3 illustrates, Asch’s results could be explained as a result of varying the social good/bad dimension, while holding constant the intellectual good/bad description. The power of the warm–cold manipulation to alter a Gestalt impression of an individual was that it tapped a
Good-intellectual
Persistent Imaginative
Scientific Determined Skillful Industrious Intelligent
Serious Shrewd Important Discriminating Cautious Practical Dominating Meditative Daring Reserved Artistic Stern
Unsociable Unpopular
Badsocial
Cold Humorless Pessimistic Irritable Moody
Critical
Reliable
Unhappy Vain Finicky Boring
Unimaginative
Dishonest Squeamish Impulsive Insignificant Submissive Superfical Clumsy Naive Wavering Irresponsible
Wasteful
Unintelligent
Frivolous
Honest Modest Tolerant Helpful Sincere Sentimental Humorous Happy Popular Good natured Sociable Warm
Goodsocial
Foolish
Bad-intellectual
Figure 2.3 Trait adjectives in a multidimensional scaling solution of social (warmth) and intellectual (competence) dimensions. Source: Rosenberg et al. (1968). Reproduced by permission.
Warmth and Competence as Universal Dimensions of Social Perception
73
separate and primary fundamental dimension of person perception (see also Hamilton and Zanna, 1974; Zanna and Hamilton, 1972, 1977). Nevertheless, the implications of these basic dimensions of person perception had not yet reached total consensus. Additionally, calling trait lists ‘‘person perception’’ was empirically tractable but ecologically problematic. Some studies (e.g., Chemers, 1997; Fiske, 1980) addressed ecological validity by providing pictures of stimulus persons engaged in personality-revealing behaviors on two cognate dimensions, such as sociability and responsibility. But these laboratory studies still entailed experimenter-chosen traits, capitalizing on the apparent distinction between the two dimensions, but begging the question of perceivers’ spontaneously used dimensions. Related social (warmth) and task (competence) orientations also describe interactions in small groups (Bales, 1950). Bales coded interactions in generations of a self-observational small group class at Harvard, in addition to interacting small groups in a variety of organizations; all evidence converged on these two dimensions (Bales, 1999). Bales also included a third dimension in his system, which amounted to the sheer volume of interaction, probably most salient in the live interaction context, but less salient in stored impressions. As noted earlier, Peeters (1983, 1992, 1995) has argued for the dimensions of self-profitability (e.g., confident, ambitious, practical, intelligent)— akin to competence—and other-profitability (e.g., conciliatory, tolerant, trustworthy)—akin to warmth. The Peeters distinction has been applied to evaluations of social behavior (Vonk, 1999). In particular, positive and negative behaviors were judged more extremely when they had consequences for others than when they had consequences only for the self. Normally, other-consequential behaviors fall on the warmth dimensions, but this experiment fully crossed warmth/competence and self/other consequences, finding that interpersonal consequences mattered the most. Similarly, dislikeable and strong behavior (roughly, low warmth, and high competence) each amplify the other, showing again that consequences for others are most salient in interpersonal judgments (Vonk, 1996). We return to this point below (section 2.3). Wojciszke and colleagues’ extensive experimental work on the two dimensions continues to build a strong case for their importance. Together, these two dimensions account for 82% of the variance in global impressions of well-known others (Wojciszke et al., 1998). Three-quarters of over 1000 personally experienced past events are framed in terms of either morality or competence (Wojciszke, 1994), and impressions of work supervisors show a similar pattern (Wojciszke et al., 2007). In sum, when people spontaneously interpret behavior or their impressions of others, warmth and competence form basic dimensions that, by themselves, almost entirely account for how people characterize others. (For a review, see Wojciszke, 2005a,b.)
74
Amy J. C. Cuddy et al.
The warmth–competence dimensions also emerge in voters’ evaluations of political candidates (Abelson, 1982; Kinder and Sears, 1985; Todorov et al., 2005; Wojciszke and Klusek, 1996), as types of leadership styles (relation- and task-oriented; Chemers, 2001), as dimensions of inpersonal attraction ( Jamieson et al., 1987, Lydon et al., 1988), as determinants of social network development (Casciaro and Sousa-Lobo, 2005), among others. These dimensions appear in, for example, spontaneous impressions of presidential candidates, which entail both competence and integrity (warmth and trustworthiness) (Abelson et al., 1982; Kinder et al., 1980; Wojciszke and Klusek, 1996). Impressions of leaders more generally involve these dimensions, with image management (building trust), relationship development (warmth), and resource deployment (competence and efficacy) (Chemers, 1997); although one could quibble over separating or combining trust and warmth, the core difference between the task and social domain consistently appears. In perception of others, the dimensions are negatively correlated in most cases ( Judd et al., 2005). This negative correlation leads to a high representation of others in the mixed combinations, high on one dimension and low on the other, as the SCM predicts in the intergroup domain. Judd and colleagues suggest this negative correlation is especially likely when judging people, groups, or cultures about which norms may discourage uniform disparagement, especially in a comparative context. 2.1.1.1. Summary A venerable history of warmth and competence dimensions emerges in independent lines of research. One could add self-perception to this list (e.g., independent, agentic vs interdependent, communal) as well as work on perceptions of particular social categories (e.g., the distinction between communion and agency in gender stereotypes). The various labels used for these basic dimensions, however, had (until recently) obscured the pervasiveness and power of the fundamental, underlying dimensions of warmth and competence (Abele et al., in press).
2.2. Intergroup perception 2.2.1. Centrality of warmth and competence Numerous in-depth analyses of stereotypes of specific social groups also reveal warmth and competence as central dimensions. These dimensions are evident in stereotypes of older people (Cuddy and Fiske, 2002; Cuddy et al., 2005); Asians and Asian Americans (Kitano and Sue, 1973; Lin et al., 2005; Maddux et al., in press); immigrants (Lee and Fiske, 2006); subgroups of gay men (Clausell and Fiske, 2005); subtypes of women (Cuddy et al., 2004; Cuddy and Frantz, 2007; Eckes, 2002; MacDonald and Zanna, 1998); subgroups of Black Americans (Williams and Fiske, 2006); types of mental illnesses (Russell et al., 2007); European nationalities (Cuddy et al., in press;
Warmth and Competence as Universal Dimensions of Social Perception
75
Linssen and Hagendoorn, 1994; Peeters, 1993; Phalet and Poppe, 1997; Poppe and Linssen, 1999); enemy outgroups (Alexander et al., 1999); volunteer organizations (Cadinu and Cerchioni, 2001); linguistic groups (Ruscher, 2001; Yzerbyt et al., 2005); and Fascist depictions of racial groups (Volpato et al., 2007). Research on the SCM has typically asked an initial sample of participants to generate ‘‘important social groups in society’’ to obtain a list of groups people spontaneously use to classify others (e.g., see Fiske et al., 2002b, for the general paradigm). Later samples are asked to evaluate these groups (as they are stereotyped in society) on warmth and competence traits. Cluster analyses consistently reveal clusters of groups that fit specific warmth– competence combinations, not just in the United States but also in other nations. Further, groups that fit into similar clusters elicit the predicted patterns of emotions and behaviors that the SCM and BIAS map predict. These results suggest that important social groups are characterized in terms of warmth–competence stereotypes that, in turn, guide emotional and behavioral reactions toward those groups. We discuss these studies in greater detail below (sections 3–5). Evidence for the spontaneous use of these dimensions in stereotyping comes from a reanalysis of the Princeton stereotyping series begun by Katz and Braly (1933), Gilbert (1951), Karlins et al. (1969), and Leslie et al. (2007). Using the original list of 100 adjectives, 5 independent judges categorized each trait appearing in any of the stereotypes in any of the 4 studies. Using 60% agreement as a criterion, 17 traits were categorized as warmth traits, 33 traits were categorized as competence traits, and 34 traits were categorized as neither, so 60% of the spontaneously checked adjectives for 10 ethnic groups over 75 years fit the warmth–competence dimensions. Recall too that in Wojciszke’s (2005a,b) work, 75–82% of the adjectives in trait descriptions fit these dimensions. The difference may be that the Katz–Braly studies used a number of unusual adjectives relevant to international relations (e.g., nationalistic) but not typical in other kinds of stereotyping or person perception. 2.2.2. Ambivalent stereotypes The hypothesis that many group stereotypes contain both negative and positive components on the warmth versus competence dimensions has garnered ample empirical support. Across outgroups, stereotypes often include a mix of more and less socially desirable traits, not just the uniform antipathy so often assumed. Pitying stereotypes combine warmth with incompetence, portraying some outgroups as neither inclined to nor capable of harm toward members of the ingroup. Envying stereotypes combine competence with coldness, portraying other outgroups as doing well for themselves, but as having negative intentions toward the ingroup. Several studies locate the majority of a domain’s outgroups in the two ambivalent clusters. In Phalet and Poppe’s (1997) multidimensional scaling
76
Amy J. C. Cuddy et al.
of Central and Eastern European national and ethnic stereotypes, the majority of groups (37 out of 58) landed in these two quadrants: incompetent but moral-social (e.g., Byelorussians, Bulgarians, Czechs) and competent but immoral-unsocial (e.g., Germans, Jews). In the same vein, the Ambivalent Sexism Inventory (ASI; Glick and Fiske, 1996) identifies two types of women as targets of ambivalent prejudice: women who exhibit agentic traits (e.g., feminists, lesbians, and professional women) elicit hostile sexism (HS), a form of envious prejudice. By contrast, women who exhibit communal traits (e.g., housewives) elicit benevolent sexism (BS), a paternalistic prejudice. Traditional stereotypes of men are also ambivalent, casting them as competent, but not warm (Glick and Fiske, 1999; Glick et al., 2004). Stereotypes of successful minority groups (e.g., Asian Americans, Jews) are similarly high in competence but low in warmth (Glick, 2002; Lin et al., 2005). Kay and Jost (2003) have argued that stereotypes often have ‘‘complementary’’ content with positive elements compensating for negative elements. When European Union nations rated each other, all nations fell into the ambivalent combinations; none were uniformly positive or negative (Cuddy et al., 2007). Finally, in SCM research in the United States and elsewhere, the majority of groups are stereotyped as high on one dimension and low on the other, rather than as uniformly high or low on warmth and competence (Cuddy et al., in press; Fiske et al., 2002b). Despite their ambivalent content, envious and paternalistic stereotypes still function to maintain the status quo and defend the position of societal reference groups (also see Jost et al., 2001 for this argument). Further, as a form of cross-dimensional ambivalence (MacDonald and Zanna, 1998), these combinations are psychologically consistent for perceivers. Because the positive and negative traits attributed to a specific group are on orthogonal dimensions, perceivers can imagine a group as being warm but incompetent or as competent but cold without experiencing the psychological tension that is classically assumed (e.g., by Freud) to be integral to ambivalence. Whatever the exact proportion of ambivalent to univalent group stereotypes may be, prior theory and research has neglected these ambivalent combinations by focusing on uniformly negative stereotypes (see Glick and Fiske, 2001b for a more extended discussion). As we argue below, this has obscured the true nature of important forms of prejudice. These include the oldest form of prejudice—sexism, which has long fostered inequality through paternalism (Glick and Fiske, 1996; Jackman, 1994)—and the most severe form of prejudice—genocidal hatred, which is most commonly directed toward successful, envied minorities (Glick, 2002, 2005). 2.2.2.1. Warm but incompetent stereotypes Paternalistic ambivalent stereotypes portray a group who is disrespected but pitied, which carries overtones of compassion, sympathy, and even tenderness, under the right conditions. Such paternalism is evident in race, dialect, age, and gender
Warmth and Competence as Universal Dimensions of Social Perception
77
prejudice. Ambivalent racism (Katz and Hass, 1986) comprises a mix of anti-Black attitudes (e.g., perceived incompetence and laziness, violating the work ethic) and paternalistic pro-Black attitudes (e.g., perceived pitiful disadvantage, need for help). Linguistic outgroups provide another example: speakers of nonstandard dialects (e.g., Scottish accents in Great Britain, Chicano accents in the United States) are perceived as incompetent but friendly (Bradac, 1990; Ruscher, 2001). Ageist stereotypes predominantly characterize older people as kind but incompetent, suggesting a similarly ambivalent dynamic (Cuddy and Fiske, 2002). Support for the ambivalent content of elderly stereotypes is plentiful (see Cuddy and Fiske, 2002; Cuddy et al., 2005). Participants rate older people as intellectually incompetent (Rubin and Brown, 1975) and as less ambitious and responsible than younger people (Andreoletti et al., 2001), but also as friendlier and warmer than younger people (Andreoletti et al., 2001). In the workplace, older people are perceived as less competent in job-performance-related tasks than in interpersonal ones (Avolio and Barrett, 1987; Rosen and Jerdee, 1976a,b; Singer, 1986). Moreover, people are quicker at associating elderly names with warmth traits than with competence traits (Zemore and Cuddy, 2000). Paternalism is most prominent in gender stereotypes. Traditional women (e.g., homemakers) are the paternalistic default when people rate women as a general category (Haddock and Zanna, 1994). This generates the ‘‘women are wonderful’’ effect: highly positive ratings of generic women (Eagly and Mladinic, 1989), because they are viewed as especially warm (or communal), while they are not viewed as particularly competent (or agentic). Traditional women are the objects of BS (Glick and Fiske, 1996, 2001a,b) and paternalistic attitudes (e.g., women require men’s protection and provision). In sum, the paternalistic stereotypes just described (of disadvantaged Blacks, nonstandard speakers, elderly people, and traditional women) apply to groups that inhabit the incompetent-but-warm corner of the warmth competence space. We discuss some consequences of paternalistic prejudice toward these specific groups in greater detail later in this chapter. 2.2.2.2. Competent but cold stereotypes In contrast stands a different set of outgroups stereotyped as highly competent but not warm (Glick and Fiske, 2001a,b). Targets of envious ambivalence include nontraditional women, Jews, and Asians. Nontraditional women (e.g., career women, feminists, lesbians, athletes) are acknowledged to be competent, but are viewed as lacking warmth (see also Eagly, 1987; Glick et al., 1997; MacDonald and Zanna, 1998). Derogatory labels for powerful women, such as Iron Maiden, Ice Queen, and Ball Buster, capture this dynamic. Anti-Semitic notions of an international Jewish conspiracy exaggerate Jews’ stereotypically feared competence, whereas views of them as self-serving portray them as not warm (Glick, 2002, 2005). The modern American
78
Amy J. C. Cuddy et al.
equivalent, Asian Americans—who are viewed as the model minority—are seen as highly competent and hardworking, envied as too ambitious, but are simultaneously characterized as unsociable and aloof (Hurh and Kim, 1989; Kitano and Sue, 1973; Maddux et al., in press; Sue and Kitano, 1973; Sue et al., 1975). The Anti Asian-American Prejudice scale measures dislike for Asians’ perceived lack of sociability along with envious respect for their perceived competence (Lin et al., 2005). Thus, nontraditional women, Jews, and Asians elicit a shared stereotype as being too competent and not at all nice. We discuss the social consequences of some of these specific groups in greater detail later. 2.2.2.3. Univalent stereotypes Not all groups receive ambivalent stereotypes; some groups’ stereotypes are evaluatively consistent, or univalent. Low-status groups viewed as openly parasitic (i.e., opportunistic, freeloading, exploitative) are banished to the not warm, not competent cell. These groups, whose members are perceived as both hostile and indolent, are most likely to elicit the uncomplicated and untempered antipathy that past prejudice theory and research associates with derogated groups. Such groups are rejected both for their perceived negative intent toward the rest of society (not warm) and their inferred inability to succeed on their own (not competent). We will see that this is a particularly virulent form of prejudice. Conversely, who is favored as both warm and competent? We suggest three possible reasons why groups are placed in this quadrant. First, ingroup favoritism may lead to viewing the ingroup as both warm and competent. Second, close allies (whose status is similar to or greater than the ingroup’s and who are viewed as cooperative) should receive a purely positive stereotype. Finally, groups that represent the cultural default (e.g., in the United States, the middle class) may be viewed in a univalent, positive way. We label both ingroups and societal ‘‘default’’ groups as ‘‘reference groups’’ because they tend to be viewed as part of a societal ideal, as when most Americans identify themselves as middle-class (even if qualified by ‘‘lower’’ or ‘‘upper’’). Similarly, Whites and Christians in the United States, even when not a local majority, may be viewed as culturally dominant, societal reference groups. Even groups who acknowledge their own exclusion from the cultural ideal may still identify with aspects of the societal reference group. These notions are consistent with theories of the influence dominant groups exert in constructing and disseminating legitimizing ideologies that diffuse, to a greater or lesser extent, throughout society (e.g., social dominance theory, Sidanius and Pratto, 1999; system justification theory, Jost and Banaji, 1994; and Jackman’s, 1994, theory of paternalism). Hence, people’s understandings of culturally shared stereotypes may generally reflect the perspective of society’s dominant reference groups.
Warmth and Competence as Universal Dimensions of Social Perception
79
2.2.3. US SCM studies Now, we describe some of the specific evidence for the SCM, beginning with US samples. 2.2.3.1. Basic method We conducted five US studies, comprising 10 samples with over 1165 respondents, to test SCM predictions regarding (a) the centrality of warmth and competence as dimensions of stereotypes, (b) the prevalence of ambivalent stereotypes, and (c) the relationship of structural variables (status and interdependence) to stereotype contents (Cuddy et al., 2007; Fiske et al., 1999, 2002b). Research samples have been diverse, including not only undergraduates, but also community members (including both middle-aged and older working and retired adults from various US regions) and a representative national sample, which we discuss in greater detail below. Participants in the SCM correlational studies rated lists of societal groups on competence, warmth, status, and competitiveness. Groups were selected based on frequent nomination by participants in pilot studies who generate lists in response to the question ‘‘What various types of people do you think today’s society categorizes into groups?’’ (Fiske et al., 2002b, Pilot Study 1 and Pilot Study 2). We supplement the groups participants most frequently generate with groups chosen for theoretical reasons or because they had been a focus of past prejudice research, including frequently studied subgroups and subtypes (e.g., feminists and Black professionals) as well as superordinate groups (e.g., women and Black people, respectively). Final lists include widely varied target groups characterized by occupation, nationality, race or ethnicity, socioeconomic status, religion, and gender subtypes, among others. Participants rate the social groups on traits related to warmth (warm, nice, friendly, and sincere) and competence (competent, confident, skillful, able), in addition to status and competitiveness, which we discuss later. In long versions of the questionnaire, participants rate between 17 and 25 groups; in a short version, participants rated 6 groups. The scales were developed and refined over the course of the studies (Cuddy et al., 2007; Fiske et al., 1999, 2002b). The original list of traits included both positive and negative items and many items unrelated to warmth and competence. To minimize social desirability biases and to draw on perceived societal stereotypes as culturally shared knowledge, participants were instructed that, ‘‘We . . . are interested in how different groups are considered by U.S. society. We are not interested in your personal opinions, but in how you believe others view these groups.’’ All variables were rated on a 1 (not at all) to 5 (extremely) scale. Principal components factor analyses consistently point to two trait factors—one reflecting warmth and the other reflecting competence. Negative traits did not consistently load onto one factor, so
80
Amy J. C. Cuddy et al.
they were dropped from the lists. Scale reliabilities were sufficiently high for all scales in all samples. 2.2.3.2. Results To test the utility of warmth and competence in describing groups, we first examine the two-dimensional array on groups on their warmth and competence means via cluster analyses. We decided that if the dimensions were useful in ‘‘sorting’’ groups, an ideal cluster solution would have to include at least four clusters that differed significantly on warmth and competence. Of the five samples that answered the long versions of the questionnaire, one yielded a five-cluster solution, three yielded four-cluster solutions, and one yielded a two-cluster solution. The short version, which included only six groups, also yielded a two-cluster solution (these data were combined from five samples in a single study). Table 2.1 presents these data by study. Figure 2.1 provides an example of one cluster solution. The SCM hypothesizes that a substantial number of outgroups will receive ambivalent stereotypes, defined by low ratings on one dimension, coupled with high ratings on the other. Three analyses test the ambivalent stereotypes hypothesis: (a) independent samples t-tests comparing warmth and competence between clusters, (b) paired t-tests comparing warmth and competence within clusters, and (c) paired t-tests comparing warmth and competence within groups. To be identified as ambivalent (highcompetence/low-warmth or low-competence/high-warmth), clusters had to meet two conditions: (1) warmth and competence means differed significantly; and (2) the mean for their high dimension was higher than that of groups low on that dimension, and the mean for their low dimension was lower than that of groups high on that dimension. We predicted that all samples would include the two ambivalent clusters, and that the majority of groups would be rated significantly higher on one than the other dimension. As predicted, all six samples include both a HC–LW cluster and a LC– HW cluster of groups. Some of the groups that consistently landed in the HC–LW cluster were Asians, Jews, rich people, businesswomen, and feminists (all successful minority groups). Some of the groups that consistently landed in the LC–HW cluster were elderly people, housewives, disabled people, and mentally retarded people. Across samples, on average 83% of the groups were rated significantly higher on one dimension than the other, with a range of 74–100%. So, across varied lists of societal groups, the majority of groups received ambivalent stereotypes. On the questionnaires used in two of the samples, we explicitly included ingroups, which we expected to land in the univalent HC–HW cell (Cuddy et al., 2007, Study 1; Fiske et al., 2002b, Study 2). As expected, these samples produced HC–HW clusters that included the ingroups (e.g., Americans, students, middle-class, Whites). Four of six of the samples produced LC– LW clusters, which included highly marginalized outgroups (e.g., homeless, poor, welfare recipients, drug addicts).
81
Warmth and Competence as Universal Dimensions of Social Perception
Table 2.1 Cluster solutions for SCM samples Cluster Sample
United States Fiske et al. (1999) (n ¼ 42) Fiske et al. (2002b) Study 1: students (n ¼ 73) Study 1: nonstudents (n ¼ 38) Study 2 (n ¼ 148) Study 3 (short version, n ¼ 230) Cuddy et al. (2007) International Combined EU samples rating EU nations (n ¼ 755) Europe Belgium (US groups, n ¼ 40) Belgium (own groups, n ¼ 43) Bulgaria (rating EU nations, n ¼ 95) Italy: students (rating own groups, n ¼ 180) Italy: nonstudents (rating own groups n ¼ 41) Norway (rating EU nations, n ¼ 40) Asia (rating own groups) Hong Kong (n ¼ 60) Japan (n ¼ 82) South Korea (n ¼ 91) Latin America (rating own groups) Costa Rica (n ¼ 122) Mexico (n ¼ 89) Israel (rating own groups) Israel-Jewish (n ¼ 104) Israel-Muslim (rating own groups, n ¼ 100) a
HC– LW
LC– HW
HC– HW
LC– LW
MC– MW
X
X
–a
–
–
X X
X X
–a –a
X X
X X
X X
X X
X –
X –
X
X
X
X
X
–
XX
X
–
–
–
X
X
X
XX
–
X
X
X
XX
–
XX
X
X
–
–
X
X
X
X
–
X
X
X
X
–
X XX XX
X X X
– – –
X XX XX
X – –
X X
X X
– X
XX X
– –
X –
X –
X X
X X
X X
In the first few studies, we did not ask preliminary study respondents to nominate ingroups, so none appeared in these surveys. Note: LC, low competence; HW, high warmth; MC, medium competence, and so on. ‘‘X’’ indicates one cluster; ‘‘XX’’ indicates two in that quadrant (e.g., HC–LW and HHC–LLW).
82
Amy J. C. Cuddy et al.
2.2.3.2.1. Summary In sum, across the 10 US samples, the stereotypes of most groups were ambivalent. That is, most groups were viewed as either competent but not warm, or warm but not competent. A small minority of groups were stereotypically low on both warmth and competence, presumably viewed as both low status and free-riding. Only ingroups and mainstream social groups were perceived as both warm and competent. 2.2.4. Cross-cultural SCM studies: Data from 17 nations Any claim that a phenomenon is universal requires testing across multiple cultures (Bond, 1994). American perceivers might be influenced by unique norms, ideologies, and attribution biases that exclusively support our proposed principles. We therefore tested the SCM in 20 non-US samples, representing 17 nations (Cuddy et al., in press). These include 12 samples from 10 European nations (Belgium, Bulgaria, France, Germany, Italy, Netherlands, Norway, Portugal, Spain, and UK), three East Asian nations (Hong Kong, Japan, and S. Korea), three Latin American nations (Costa Rica, Dominican Republic, and Mexico), and two Israeli samples ( Jewish and Muslim). 2.2.4.1. Basic method In a first study, we used the same target groups as in our US studies, but surveyed a non-US sample of perceivers (Belgium). If the warmth–competence dimensions stem from characteristics specific to the perceivers’ culture, they should fail to generalize to non-US respondents. Next, varying both target groups and respondents’ nationalities, seven EU nations (Belgium, France, Germany, Netherlands, Portugal, Spain, and UK) and two non-EU European nations (Bulgaria and Norway) rated the then-current 15 EU member nations, which constituted an alternative, predetermined set of relevant groups, thereby eliminating concerns about biased selection of target groups and the potentially restricted applicability of the model to US-generated target groups. In the third set of studies, 10 samples from 9 nations (Belgium, Costa Rica, Dominican Republic, Hong Kong, Israel-Jewish, Israel-Muslim, Japan, Mexico, and S. Korea, Cuddy et al., in press; and Italy, Durante and Capozza, 2008) rated lists of relevant groups in their respective societies (generated in pilot studies using participants in the same nation). This combined emic–etic (insider–outsider) approach (Hui and Triandis, 1985) unites culturally indigenous approaches to data collection (i.e., indigenous lists of groups) with imported approaches (i.e., our scales). The goal was to create cross-cultural comparisons using equivalent measures, while simultaneously using ecologically valid targets. In the studies where local participants listed relevant social groups, separate samples of participants from Belgium, Costa Rica, Dominican Republic, Hong Kong, Israel-Jewish, Israel-Muslim, Japan, Mexico, and South Korea answered the following questions: (1) ‘‘Off the top of your
Warmth and Competence as Universal Dimensions of Social Perception
83
head, what various types of people do you think today’s society categorizes into groups (i.e., based on ability, age, ethnicity, gender, occupation, race, religion, etc.)?’’ (2) ‘‘What groups are considered to be of very low status by [Belgian/Costa Rican/Japanese/etc.] society?’’ (3) ‘‘What groups, based on the same criteria used in the first question, do you consider yourself to be a member of ?’’ Question 1 aimed at getting participants to list relevant social groups in the least constrained way. In US studies, however, this question typically yielded lists that failed to include extremely low status outgroups that might fit the pure antipathy model of prejudice, nor did it typically generate ingroups (who, as default groups, may not be listed). Thus, Questions 2 and 3 were intended to insure that all types of groups would be listed. Groups listed by at least 15% of participants were included on the final questionnaire. A total of 1841 respondents from 17 nations completed a version of the SCM questionnaire (Cuddy et al., in press). University students predominated in most samples, which were 60% female with an average age of 21. Different samples rated different groups. A Belgian sample rated groups from American studies (Fiske et al., 2002b, Study 2), but translated into French. Samples from the nine European nations (eight EU members— Belgium, France, Germany, Italy, Netherlands, Portugal, Spain, and the United Kingdom—and two non-EU members—Bulgaria and Norway) rated the 15 then-current member nations of the European Union. Eight samples from seven nations (Belgium, Costa Rica, Dominican Republic, Hong Kong, Israel-Jewish, Israeli-Muslim, Japan, Mexico, and S. Korea) rated their own lists of relevant groups, generated in pilot studies (described above). Like the groups used in American studies, these groups varied on race, gender, socioeconomic status, occupation, religion, immigration history, and so on. Participants rated the social groups on items measuring warmth (warm, nice, friendly, sincere) and competence (competent, confident, skillful, able), which were adapted from the US studies (Fiske et al., 1999, 2002b). For each nation, translators converted the questionnaire to the relevant language, and all independent back-translations were satisfactory. As in the initial work, to reduce social desirability biases and to tap consensual stereotypes, participants were told, ‘‘We are interested in how different groups are considered by [Belgian, German, Hong Kong etc.] society. We are not interested in your personal opinions, but in how you believe others view these groups.’’ All items were rated on a 1 (not at all) to 5 (extremely) scale; reliabilities were sufficiently high for all scales in all samples. 2.2.4.2. Results In all samples, groups spread out fairly equally along the warmth and competence dimensions, producing no fewer than three clusters of stereotypes. Forty-two percent of the samples produced five-cluster
84
Amy J. C. Cuddy et al.
solutions; 33% produced four-cluster solutions; and 25% produced threecluster solutions (collapsing across EU samples rating EU nations). Moreover, in all samples, at least two clusters significantly differed on both warmth and competence, suggesting that participants were using both dimensions to distinguish groups. Table 2.1 presents the cluster solutions for all samples. Remarkably, the two ambivalent clusters appeared in 19 of 20 samples. In some samples, more than one of each ambivalent cluster appeared. On average, 79% (S.D. ¼ 13%) of groups were rated significantly higher on one dimension than the other, with a range of 59% (Hong Kong; Cuddy et al., in press, Study 2) to 93%. 2.2.4.2.1. Europe We combined the 7 EU samples that rated the 15 then-current EU member nations. The combined sample produced a threecluster solution, with only ambivalent clusters and the third cluster representing groups that were viewed as hypercompetent and especially lacking in warmth (in addition to a less extreme competent, but also not warm cluster). The LC–HW cluster included Greece, Ireland, Italy, Portugal, and Spain. The HC–LW cluster included Austria, Belgium, Denmark, Finland, France, Luxembourg, Netherlands, and Sweden. The more extreme HHC–LLW cluster—highest on competence and lowest on warmth—comprised Germany and the United Kingdom. Nearly identical patterns replicated in each of the individual EU samples. Participants rated the majority of groups (87%) higher on one than the other dimension, and this held across samples, ranging from 53% (France) to 100% (Spain). Fig. 2.4 depicts the cluster solution. The emphasis on ambivalence (no one is perfect, but no one is all bad or all good) did not stem from shared EU membership. Results were very similar for Norway and Bulgaria (European nations that were not EU members at the time): the majority of groups differed significantly on warmth and competence—93% in Bulgaria and 87% in Norway. They also produced very similar cluster solutions to the combined EU samples, each with two HC–LW clusters and one LC–HW cluster; Bulgaria also generated a HC–HW cluster that included Italy and Spain. These EU samples produced no LC–LW clusters. Thus, no nation occupied the HC–HW (i.e., favored) cell, indicating that no nation represents an agreed-upon ideal European prototype. In the EU samples, each nation rated itself (ingroup) and six other nations (outgroups), so six outgroups and one ingroup rated each nation, providing a rare opportunity to compare ingroup and outgroup ratings of the same group. Comparing each nation’s self-rated warmth and competence (e.g., Italians rating Italy) to that nation’s average warmth and competence ratings of all the other groups (e.g., Italians rating all other EU nations) revealed that participants on average rated their own nation as marginally more competent, but not as more warm, than other nations.
85
Warmth and Competence as Universal Dimensions of Social Perception
High SPA POR
SPA
Warmth
POR GRC
ITA
LC-HW
FRA
IRE SWE BEL FRA NTH DEN HC-LW FIN LUX AUS
BEL
NTH
GER UK HHC-LLW GER UK
Low Low
Competence
High
Figure 2.4 Stereotype content model warmth by competence mapping of EU member nations rating themselves (bold) and other nations, as seen by the EU community. Source: Cuddy et al. (in press). Reproduced by permission.
We also compared nation’s self-rated warmth and competence (e.g., Italians rating Italy) to the aggregated warmth and competence ratings of that nation made by people in other nations (e.g., all other participating nations rating Italy). This analysis produced mixed results. Some nations rated themselves significantly higher than they were rated by other nations on either competence (Netherlands, Belgium) or warmth (Spain, Portugal) or both (France), whereas others disfavored themselves on either competence (Germany) or warmth (Belgium). Yet, other self-ratings were close to the average other-ratings (Germany, Netherlands on warmth, Portugal, Spain on competence, United Kingdom on both). Apparently, higher-status groups favored their ingroup on dimensions that reflect obvious status differences (i.e., competence), whereas lower-status groups favored their ingroup on dimensions irrelevant to status (i.e., warmth), thus insuring positive differentiation from other groups (Ellemers et al., 1997; Mummendey and Wenzel, 1999; Poppe, 2001; Spears and Manstead, 1989; Van Knippenberg and Van Oers, 1984). These results are similar to those obtained by Jost et al. (2001).
86
Amy J. C. Cuddy et al.
In contrast to stereotypes of European nations, cluster solutions were more similar to those found in the United States when European samples rated social groups within their own societies (Belgium and Italy) or the social groups originally generated by US participants (another sample in Belgium). All three of the samples that rated social groups within their societies produced the predicted four combinations of high and low warmth and competence stereotypes (with no additional clusters). The majority of groups (M ¼ 78%, SD ¼ 9%) in these samples were rated significantly higher on one than the other dimension, ranging from 67% (Italy nonstudents) to 87% (Belgium). Moreover, these samples produced HC–HW clusters that included ingroups or societal prototype groups (e.g., welleducated people and Northerners in Italy; Wallonians and Catholics in Belgium), and LC–LW clusters that included extreme outgroups (e.g., poor, unemployed, and immigrants in both Belgium and Italy). In short, European national stereotypes were dominated by ambivalent stereotypes (i.e., viewed as warm and incompetent or as competent and cold), with univalent stereotypes being conspicuously absent, even when people rated their own nation. By contrast, when rating groups within their own societies (or social groups generated by past research in the United States), all of the expected combinations of warmth and competence occurred. 2.2.4.2.2. East Asia Including East Asian samples is particularly important for assessing claims of universality because Asian cultures tend to differ substantially from the United States on the often-cited individualism–collectivism (IC) dimension, relevant to psychological processes that underlie stereotyping (e.g., Hofstede, 1980; Triandis, 1995). Each of the three Asians samples (Hong Kong, Japan, and South Korea; Cuddy et al., in press, Study 2), which rated their own societal groups, produced at least one HC–LW cluster and at least one LC–HW cluster. The majority of Asian groups (76%) scored significantly higher on one than the other dimension, with a range of 59% (Hong Kong) to 90% (South Korea). All three Asian samples also produced at least one LC–LW cluster. Figure 2.5 depicts data from the Hong Kong sample. However, none of the Asian samples produced a HC–HW cluster, typically reserved for ingroups and societal prototype groups, in spite of the fact that we explicitly asked participants in the preliminary groupslisting studies to list ingroups. Instead, these groups (e.g., college graduates, full members of society, educated, middle-class, and students) migrated toward the middle of the warmth–competence space. Some occupied the HC–LW clusters (e.g., Japan: my university, students); some occupied the LC–HW clusters (e.g., Japan: family, friends, Japanese, my clubs; South Korea: college students, middle-class), and some occupied a MC–MW cluster (e.g., Hong Kong: Asians, Chinese, students).
87
Warmth and Competence as Universal Dimensions of Social Perception
High
Christians Children LC-HW Elderly
Foreigners
Warmth
Students Mentally ill
Janitors Poor
Women AsiansMarried Chinese College grads Youths Adults MC-MW HK Bluelocals Singles collar Men White-collar
Filipino maids Unemployed LC-LW
Professionals HC-LW
Immigrants
Mainlainders Pakistani
Rich
Low Low
Competence
High
Figure 2.5 Stereotype content model warmth by competence mapping of societal groups in Hong King, as rated by Hong King Chinese students. Source: Cuddy et al. (in press). Reproduced by permission.
A more fine-grained analysis isolated one group guaranteed to be an ingroup for all samples, namely students. Table 2.2 compares the three relatively collectivistic East Asian samples with three relatively individualist Western samples that rated similar groups. The individualists rate students significantly higher than the average of other groups ( ps < .01), whereas none of the collectivists did. Collectivists attenuating reference-group favoritism fits well-documented IC modesty concerning self (e.g., Heine et al., 1999; Kitayama et al., 1997). Thus, midward migration of referencegroup clusters fits collectivists being less likely to rate themselves and their ingroups too positively. 2.2.4.2.3. Israel We collected data from one Jewish and one Muslim Israeli sample, in which each rated its own societal groups. The Jewish Israeli sample produced a five-cluster solution, with one cluster per cell and one middle cluster. Participants rated the majority of groups (88%) as significantly higher on either competence or warmth. Reference groups (e.g., educated, Jews, and seculars) landed in the HC–HW cluster. Thus, the Jewish Israeli sample resembles the American and European samples, perhaps not surprising because most of their families emigrated from those countries.
88 Table 2.2
Sample
Amy J. C. Cuddy et al.
Reference-group favoritism ratings
Competence
Warmth
Collectivist samples (Cuddy et al., in press) Hong Kong .06 .39* Japan .10 .07 Korea .19 .17 Individualist samples Belgium (Cuddy .36* .51* et al., in press) United States .39* .17* (Fiske et al., 2002b, Study 2) United States .72* .46* (Cuddy et al., 2007, Study 1)
General positivity (warmth þ competence mean)
.17 .02 .18 .44* .29*
.49*
* p < .01. Note: Values represent difference between the mean for the ingroup ‘‘students’’ and the mean for all other groups in that sample. High numbers indicate higher scores for students (i.e., more referencegroup favoritism).
The Muslim Israeli sample was the only one that deviated from the SCM, producing neither ambivalent cluster and rating just less than half the groups (46%) significantly higher on one dimension. Also unusual, the Muslim Israeli sample located clear outgroups (e.g., Jews, leftists, seculars) in the HC–HW cluster, above reference groups (e.g., Arabs and Palestinians), which landed in the MC–MW cluster. These unusual results warrant further investigation. 2.2.4.2.4. Latin America The Latin American samples (Costa Rica and Mexico) rated their own societal groups. Both of these samples produced the two ambivalent clusters: one HC–LW cluster and one LC–HW cluster. The majority of groups (88% in Costa Rica, 78% in Mexico) were rated significantly higher on competence or warmth. Both samples also produced univalent HC–HW and LC–LW clusters. 2.2.4.2.5. Summary Across culturally varied perceivers and targets, cluster analyses corroborated the claim that perceived warmth and competence universally differentiate stereotypes. Comparing warmth and competence—within groups, within clusters, and between clusters—supported the claim that many outgroups receive ambivalent stereotypes as competent but not warm, or as warm but not competent. Across studies and samples, the majority of groups significantly differed on warmth and
Warmth and Competence as Universal Dimensions of Social Perception
89
competence, and both ambivalent combinations occurred in almost all (19 of 20) samples. The prevalence of ambivalent groups is striking given that past prejudice theory and research has emphasized univalent derogation of out-groups. The weight of the data we have collected, in a variety of nations, clearly contradicts prior assumptions about intergroup relations.
2.3. Primacy of warmth Both warmth and competence consistently emerge as core dimensions of social perception. What’s more, considerable evidence suggests that warmth judgments are primary, both in the sense that warmth is judged before competence and that warmth judgments carry more weight in affective and behavioral reactions. From an evolutionary perspective, the primacy of warmth makes sense because another’s intent for good or ill matters more to survival than whether the other can act on those goals. Similarly, morality (warmth) judgments determine approach-avoidance tendencies, making them the fundamental aspect of evaluation (Cacioppo et al., 1997; Peeters, 2001) and therefore prior to the competence-efficacy judgments. People infer warmth from the perceived motives of the other (Reeder et al., 2002). Research confirms that information about the moral–social dimension is more cognitively accessible, more sought-after by perceivers, more predictive, and more heavily weighted in evaluative judgments. The warmth dimension predicts the valence (i.e., positive or negative) of interpersonal judgments, whereas the competence dimension predicts impression extremity (i.e., how positive or how negative) (Wojciszke et al., 1998, 1993). Note that Asch intuited the priority of the warmth dimension by manipulating it against a background of competence-related traits; as noted above, warm–cold changes the Gestalt impression of a competent person. 2.3.1. Importance of ‘‘other profitable’’ traits Moral-warmth traits facilitate or hinder mainly other people, whereas competence traits facilitate or hinder mainly the self. That is, the moralwarmth, ‘‘other-profitable’’ traits include kind, honest, and aggressive because they immediately bear on people around the judged person. ‘‘Selfprofitable’’ traits include competence, intelligence, and efficiency because they directly and unconditionally bear on the possessors’ chance to achieve their goals (e.g., Peeters, 2001). In a series of studies, Wojciszke et al. (1998) demonstrated the primacy of warmth traits in global evaluations of others. First, participants who were asked to list the most important personality traits listed significantly more warmth traits than competence traits, and the five most frequently listed traits were warmth-related (sincere, honest, cheerful, tolerant, and loyal) (Wojciszke et al., 1998, Study 1). Second, when asked to select traits that would help them decide whether a target person deserved their generally positive opinion, participants selected significantly more
90
Amy J. C. Cuddy et al.
warmth than competence traits. Third, warmth was a significantly stronger predictor (accounting for 59% of the variance) than competence (accounting for 29% of the variance) of global impressions of familiar others. Fourth, the warmth content of a fictitious other’s behaviors was the strongest predictor of global evaluations of the target, whereas competence content only weakly modified the intensity of the evaluation. Fifth, evaluations based on warmth information were strong and stable; evaluations based on competence information were weak and dependent on accompanying warmth information. And sixth, regardless of the valence of the competence information about a target, negative warmth information always elicited negative global evaluations, and positive warmth information always elicited positive global evaluations. Corroborating the importance of otherprofitable traits, recent work reveals that the majority of cultural universals fall into socio-moral (i.e., warmth) domain and that people’s implicit theories of how behaviors imply traits is fairly consensual across cultures in the domain of warmth, but not in the domain of competence (Ybarra et al., in press). Thus, warmth assessments are primary, at least from the observer’s perspective (Wojciszke, 2005b). 2.3.2. Rapidity of warmth judgments and accessibility of warmth information Cognitively, people prove more sensitive to warmth information than to competence information. In lexical decision tasks, social perceivers identified warmth-related trait words faster than competence-related trait words, even when controlling for word length (Ybarra et al., 2001). In rapidly judging faces at 100 ms exposure times, social perceivers judged trustworthiness most reliably, followed by competence (Willis and Todorov, 2006). Reliability was calculated as correlations with time-unconstrained judgments of the same faces, and it is striking that people made these judgments in a fraction of a second, with moral–social judgments occurring first. In another series of studies (Hack et al., 2007), perceivers judged warmth faster than competence in (a) an anticipated interaction paradigm, (b) a photo evaluation task without contextual cues, and a photo evaluation task including SCM groups that varied in status and competition. Converging evidence across investigators supports the priority of warmth in social perception. 2.3.3. Perceivers and situations moderate the primacy of warmth The priority of detecting warmth over competence, though robust, is stronger for some kinds of perceivers. Women, whose gender roles emphasize communal (warmth) over agentic (competence) traits (Abele, 2003) especially show this difference (Wojciszke et al., 1998). Communal traits affect their lives more, whereas competence traits affect men relatively more (Abele, 2003). In parallel, collectivistic orientations emphasize the
Warmth and Competence as Universal Dimensions of Social Perception
91
social–moral dimension, whereas individualistic orientations emphasize the competence dimension (Wojciszke, 1997). The primacy of warmth is also moderated by self–other outcome dependency. The impact of competence on global evaluations of close others is greater than the impact of competence on global evaluations of distant others, although warmth still has a greater impact (Abele and Wojciszke, 2007). For example, competence affects global evaluations of others when it contributes to the perceiver’s well-being, such as when the perceiver’s (e.g., an employee) positive outcome is contingent on the target’s (e.g., the boss) competence. Similarly, the accessibility of the two dimensions depends in part on situational pressures. Depending on the context that is primed, people construe ambiguous social behaviors (e.g., tutoring another student, avoiding a car accident, failing to cheer up a sibling, leaving a meeting) in either competence or warmth terms. When actions are framed from the actor’s (self-related, individualistic) perspective, people interpret the behaviors in competence terms, but when actions are framed from the observer’s (otherrelated, collectivistic) perspective, people interpret them in more warmthmoral terms (Wojciszke et al., 1998). 2.3.4. Diagnosticity of warmth and competence information, positive and negative Social perceivers engage a complex calculus regarding relative diagnosticity of the two fundamental dimensions. They asymmetrically process positivenegative warmth information and positive-negative competence information, but in opposite ways: Perceivers sensitively heed information that confirms others’ competence and that disconfirms others’ warmth. (Kubicka-Daab, 1989; Singh and Teoh, 2000; Skowronski and Carlston, 1987; Tausch et al., 2007; Ybarra and Stephan, 1999). For example, perceivers weight intelligent behaviors more heavily than unintelligent behaviors when evaluating another’s general competence, but weight unsociable behaviors more heavily than sociable behaviors when evaluating another’s general warmth/likeability (Singh and Teoh, 2000; Skowronski and Carlson, 1987). In social networks, perceivers view negative warmth traits and positive competence traits as transitive: If someone is perceived as unfriendly (or intelligent), then others in that person’s social network — even if connected only through indirect ties (i.e., a friend of a friend) — also are perceived as unfriendly (or intelligent; Wang and Cuddy, 2008). Heightened sensitivity to disconfirming warmth information reflects concerns about others’ intentions or motives (Reeder et al., 2002). To be perceived as warm, a person must consistently behave in moral-sociable ways; a negative deviation eliminates the presumption of morality and is attributed to the person’s (apparently unfriendly and untrustworthy) disposition. On the other hand, a person perceived as cold may sometimes behave in
92
Amy J. C. Cuddy et al.
moral-sociable ways, but will continue to be perceived as unfriendly and untrustworthy; positive deviations are easily explained by situational demands. After all, even evil people may be nice when it suits them. In other words, mean and untrustworthy behavior is more diagnostic because it is usually attributed to the other’s disposition, not to social demands. Perceivers interpret warm behavior as highly controllable, and thus non-diagnostic. In contrast, perceivers presume that competent behavior is not under immediate personal control. Hence, competence is asymmetrical in a different way. That is, a person perceived as competent may behave competently most of the time, and a few incompetent behaviors do not undermine the perception of general competence. (Consider the absent-minded professor.) However, a person perceived as incompetent, who presumably lacks the ability, can never behave competently without challenging the perceived incompetence. For competence, positive (as compared to negative) behavior is more diagnostic: Competence is usually attributed to the other’s abilities, not to social demands. Sometimes the dimensions mix: Competent behavior is judged especially as diagnostic when the other is perceived as immoral-unsociable; the competence of an enemy is potentially of greater consequence than the competence of a friend (Peeters, 2001). Thus, asymmetries in processing positive versus negative warmth and competence information boil down to their relative diagnosticity for personality impressions (Fiske, 1980; Skowronski and Carlston, 1987; Tausch et al., 2007; Ybarra and Stephan, 1999). 2.3.4.1. Summary Although both dimensions are fundamental to social perception, warmth judgments appear primary, reflecting the importance of first assessing others’ intentions before determining the other’s ability to carry out those intentions. Negative (versus positive) warmth information and positive (versus negative) competence information more heavily influence social perception due to their perceived diagnosticity of others’ dispositions. These patterns signal sensitivity to potential threats, which aids in survival value for any organism.
3. Social Structural Roots of Warmth and Competence Judgments The SCM suggests that warmth and competence judgments result from the social structural relations between individuals and groups. Because all complex societies are hierarchically organized and have limited resources, we argue that not only the two dimensions but also their social structural predictors ought to generalize across cultures. Specifically, we argue that two variables long identified as important in intergroup relations—competition and status—predict warmth and competence judgments. People viewed as competitors are judged as lacking warmth,
Warmth and Competence as Universal Dimensions of Social Perception
93
whereas people viewed as noncompetitors are judged as warm; people viewed as high status are judged as competent, whereas people viewed as low status are judged as incompetent. In addition to demonstrating these relationships in research on actual groups in the United States (Cuddy et al., 2007; Fiske et al., 2002b, 1999), we will see that they replicate in investigations of experimentally created groups (Caprariello et al., 2007; Oldmeadow and Fiske, 2007) and interpersonal perception (Russell and Fiske, 2007). Status and competition underlie group bias for a number of reasons. By taking into account structural relationships between groups (i.e., perceived competition-cooperation and group status differential), a target’s group membership provides the information necessary to quickly answer the two functional questions we raised at the beginning of this chapter: (a) Does the other intend help or harm? and (b) can the other carry out this intent? If social perception operates in the service of interaction goals (Fiske, 1992), then understanding an outgroup’s intentions and capabilities of carrying out those intentions ought to be a primary motive in perceiving social entities. And though perceiving individuals as social entities differs from perceiving groups (Hamilton and Sherman, 1996; Sedikides et al., 1998), using group membership as a categorization heuristic contributes to the tendency to interpret behavior as confirming a stereotype (Taylor, 1981). Second, linking social structure to traits legitimates unfair social structures over which people feel they have no control (for a review, see Glick and Fiske, 2001a). Legitimizing ideologies relieve people’s compunction about cooperating in a social system that inflicts pain on self or others by justifying the system as fair. For example, self-interest motivates people to believe that those who suffer have brought about their own misery, eliciting just-world beliefs that outcomes are typically deserved (Lerner and Miller, 1978), such as that groups with high-status and well-paying jobs must have earned these outcomes through talent and hard work. The status– competence correlation in fact varies by individual differences in justworld beliefs (Oldmeadow and Fiske, 2007). Similarly, social dominance orientation—the belief that group hierarchies are inevitable, even desirable (Sidanius and Pratto, 1999)—also increases the perceived status–competence correlation (Oldmeadow and Fiske, 2007). In these studies, perceivers saw photographs of more and less expensive houses, judging the probable competence, intelligence, and so on, of the occupants. In a related vein, system-justification legitimates group-level sociopolitical and socioeconomic inequalities ( Jost and Banaji, 1994). Superordinate groups justify their advantage by viewing the status quo as fair, and even subordinate groups may endorse this view because it explains their own outcomes. In a system with clear status differences, high status confers favorable competence stereotypes for perceivers in the dominant group. To lessen the dominant group’s responsibility for inequalities among groups, cooperative subordinate groups are granted warmth stereotypes, which do not challenge (and may
94
Amy J. C. Cuddy et al.
even reinforce current status relations) by pacifying subordinates, who are allocated a ‘‘safe’’ way to positively differentiate themselves (e.g., Jackman, 1994). Also, system-justification theory explains that even disadvantaged groups are motivated to believe in the fairness of structural inequality, in the belief that they might yet succeed, in turn endorsing their own group’s negative stereotypes ( Jost et al., 2001). Thus, group status predicts competence stereotypes because it justifies beliefs in fairness and meritocracy. Competition negatively predicts warmth stereotypes because this excludes groups with goals that conflict with those of the ingroup. When a group is viewed as competing for resources, such as tax dollars, people tend to attribute this behavior to the group’s alleged malice. Given this general theoretical background on legitimating groups’ positions in society, we now examine separately the reasons for the status–competence and competition–warmth predictions.
3.1. Why social status should predict competence judgments SCM predicts that a group’s position on the competence dimension can be predicted from their perceived status relative to other groups in society. High status groups (e.g., rich people) are believed to be competent, whereas low status groups (e.g., poor people) are believed to be incompetent, presumably based on the common but flawed assumption that status invariably derives from ability (as opposed to such factors as opportunity, inheritance, or luck; Fiske et al., 1999, 2002b). In short, status assesses the capability of groups to control resources. High-status groups typically are powerful as well, giving them control over obtaining and providing resources. Indeed, although status is defined by position in society, it often links to power, which is defined by the ability to regulate resources (Fiske, 1993). Therefore recognition of status (and the presumed control over resources) is inherently linked to perceived competence. People may therefore simply infer a group’s traits from their social position. This tendency may be especially strong in Western cultures, in which people tend to overuse internal dispositions, ignoring the influence of the situation or context (Gilbert and Malone, 1995; Jones, 1979; Ross, 1977). Thus, when a group is supposedly overrepresented in high-status jobs and prestigious universities, people may attribute this outcome to the group’s perceived competence.
3.2. Why competition should predict warmth judgments SCM specifically predicts that competition predicts (low) perceived warmth because of the function it serves in the structural relations between groups’ incompatible goals. Compliant, subordinate groups fulfill a convenient role, so they receive paternalistic prejudice. Dominant groups disrespect their competence but simultaneously like the qualities that keep them subordinated, as long as they do not view them as threatening. Warmth-related identities placate subordinates by assigning them socially desirable traits that
Warmth and Competence as Universal Dimensions of Social Perception
95
conveniently also imply deference to others (Glick and Fiske, 2001b; Ridgeway, 2001). Negative intentions will not be attributed to noncompetitive outgroups, whereas attributions of warmth to these groups help to maintain the status quo with a minimum of conflict ( Jackman, 1994). Disabled people and housewives, groups that are not viewed as competing for economic and educational resources, are rated as warmer than virtually all other groups, including most majority groups. The ingroup, its allies, and reference groups do not compete with themselves, so they are acknowledged as warm. The cultural default groups (middleclass, Christian, heterosexual) may not be viewed as competitive, precisely because they possess cultural hegemony. Support for the prediction that competition drives perceptions of warmth also comes from the Phalet and Poppe (1997) and Poppe and Linssen (1999) studies, in which perceived internation conflict negatively predicted socially desirable traits (morality or warmth). By contrast, groups that are perceived as competitive are stereotyped as lacking warmth. This holds true both for both low- and high-status groups. Competitive outgroups elicit frustration and resentment, making it likely that competing goals will be attributed to the outgroup’s negative intent (i.e., lack of warmth). A primary source of negative affect toward outgroups results from perceived incompatibility of their goals with ingroup goals (Fiske and Ruscher, 1993). If successful, outgroups receive grudging respect for their envied control over resources, but are never liked or seen as warm. High-status, competitive groups, such as Asians and Jews in the United States, are viewed as successfully competing for economic and educational resources with mainstream society (Glick and Fiske, 2001b); these groups are rated as significantly less warm than middle-class people and Christians (both majority groups). Low-status groups are not viewed as successful competitors, but nevertheless tend to be seen as exploitative or ‘‘competing’’ in the sense that whatever resources go to them represent a drain on the rest of society. Low-low groups (e.g., welfare recipients) are viewed in a zero-sum system as parasites who contribute little yet suck up resources (e.g., through public assistance subsidized by taxes). Because their goals are perceived as incompatible with the rest of society’s, they are not seen as warm. In short, competition addresses the question of intent. Competition pits the desired resources of one social group against others, and to compete successfully, one must intend to maximize one’s resources over others’ resources. Determining whether other groups have incompatible goals allows groups to recognize potential threats to resources (Fiske, 1993). Knowing that a group intends to compete for resources suggests that group members have negative intentions toward others (making them cold, unfriendly, and untrustworthy); characterizing such groups as cold helps to motivate the ingroup to compete. By contrast, knowing that a group intends to cooperate suggests positive intentions toward others (warmth, friendliness, and trustworthiness); the positive stereotype motivates the ingroup to cooperate.
96
Amy J. C. Cuddy et al.
3.3. US tests of SCM structural hypotheses Having described the general theory behind the structural hypotheses, and the more specific rationale for the specific competition–warmth and status–competence effects, we now review the accumulating evidence. 3.3.1. Correlational evidence In the 5 US studies described previously, comprising 10 samples of diverse participants (including one national random sample), participants also rated the same groups on items measuring perceived status and perceived competitiveness (Cuddy et al., 2007; Fiske et al., 1999, 2002b). Perceived status was assessed by perceptions of the degree to which members of a group generally hold prestigious jobs, have economic success, and attain a high education level. Perceived competitiveness was assessed by the degree to which a group’s perceived resources or power take resources or power away from the rest of society, and the degree to which the group receives ‘‘special breaks.’’ Again, in long versions of the questionnaire, participants rated between 17 and 25 groups; in a short version, participants rated 6 groups (Fiske et al., 2002b, Study 3). The social structural measures were developed and refined over the course of the studies (Cuddy et al., 2007; Fiske et al., 1999, 2002b) and principal components factor analyses consistently recovered two social structure factors, status and competitiveness. Participants made the ratings based on ‘‘. . . how you believe others view these groups.’’ Scale reliabilities were sufficiently high for all scales in all samples. We calculated correlations two ways. At the group level, we averaged ratings across participants for each of the groups, and then calculated the correlation coefficients from the group means. At the individual level, we calculated correlations separately for each individual participant, converted them using Fisher’s r to z, averaged them, and reverted them to rs. The group-level procedure uses a smaller n, but stable means that mask participant-level variation, thus producing larger rs. The individual-level procedure lacks stable means but provides more power. Here, we will present only the individual-level correlations, a more conservative estimate than the group-level analysis. As predicted, status ratings correlated positively with competence ratings in all samples, and the effect size is large: mean r ¼ .81, range ¼ .64–.87 (Cuddy et al., 2007; Fiske et al., 2002b). Competition ratings correlated negatively with warmth ratings in all samples, and the effect size is medium: mean r ¼ -.29, range ¼ -.11 to -.43. Although the competition–warmth correlations were more modest, we suspect this is due to measurement error in the competition variable, which has focused on economic competition, while omitting symbolic, value-driven competition, which may matter more in intergroup structural relations. Table 2.3 presents correlations for all samples.
Warmth and Competence as Universal Dimensions of Social Perception
97
3.3.2. Experimental evidence: Intergroup and interpersonal Of course, correlations remain open to the alternative that perceivers are reasoning in reverse, deducing from stereotypic traits that the groups must have the requisite competitive intent and status. We extended the correlational findings by manipulating the social structural variables of competition and status, then measuring subsequent changes in perceived warmth and competence (as well as emotional reactions, which we will discuss later). We present data at both intergroup and interpersonal levels of analysis. 3.3.2.1. Intergroup experiments We presented participants with a fictitious immigrant group, systematically varying both its competition with other groups and its status (Caprariello et al., 2007). Undergraduate students read a vignette depicting an unfamiliar ethnic group said to be immigrating to the United States in the near future. In a 2 2 between-subjects design, the questionnaires varied the group’s home-country competitiveness (‘‘they take power and resources from’’ versus ‘‘share power and resources with members of other groups’’) and its status (‘‘they typically have prestigious jobs, and are well educated and economically successful’’ versus ‘‘low-status jobs, and are uneducated and economically unsuccessful’’). Participants were then asked ‘‘When members of this ethnic group arrive here, to what extent will people here be likely to view incoming group members in the following ways?’’ Participants then rated the hypothetical ethnic group on warmth and competence adjectives. As the SCM predicts, status affected competence but not warmth ratings, such that high-status groups were rated as more competent than low-status groups. By contrast, competitiveness affected warmth but not competence ratings, such that competitive groups were rated as less warm than cooperative groups. Thus, evidence for these social structural predictors (status and competitive or cooperative interdependence) is not only correlational, but also causal. 3.3.2.2. Interpersonal experiments Here, we report interpersonal analogs to these intergroup findings, strengthening the causal direction from structure to perceived traits. As interpersonal analogs to intergroup relations, interpersonal relationships should similarly determine perceived warmth and competence. We have argued that the SCM’s principles are universal. If the structural, dimensional, emotional, and behavioral predictions generalize from intergroup to interpersonal interactions, then the argument for universality is strengthened. We describe some research that begins to tackle the interpersonal level of analysis (Russell and Fiske, 2007). In a pair of studies, students arrived for an experiment in which they were told that they and another student would play a game for cash rewards (prisoner’s dilemma in Study 1, trivia challenge in Study 2). They learned that they would be competing or cooperating with their partner. In Study 1, the payoff matrix was framed to promote either cooperation (‘‘Team
98
Amy J. C. Cuddy et al.
Game’’) or competition (‘‘Winner Takes All’’), and in Study 2, the reward was contingent either on their joint performance as a team competing with other student teams or based on their dyadic competition. They also learned that their partner’s status was either high or low, through social class information in Study 1 and through random assignment to boss and subordinate roles in Study 2. In Study 1, they rated their partner both before and after the interaction. The ratings beforehand were based on the minimal information contained in the status and interdependence manipulations and on allegedly subliminal information conveyed in rapidly presented sentences that actually contained no information. Cooperative-frame participants rated their partners as reliably warmer than competitive-frame participants. After they played the game (with a computer programmed to respond tit-for-tat), their ratings still reflected the cooperative or competitive framing established before the interaction. Study 2’s manipulation of reward contingency, in an after-only design, also showed significant effects of interdependence on perceived warmth, after an actual, spontaneous interaction with another naı¨ve participant. The status–competence effects also supported the SCM, in both beforeand-after ratings (Study 1) and in post-interaction-only ratings in Study 2. The findings are particularly striking in two respects: first, that they survived both standardized responses (Study 1) and spontaneous responses (Study 2), and second, that they occurred even when their basis was patently arbitrary (Study 2’s status manipulation was accomplished by shuffling the ‘‘boss’’ and ‘‘subordinate’’ cards; both studies’ competition manipulations were simply the experimental instructions in a short-term game). In Study 2, participants also were asked whether situational factors, such as the structure of the game, had influenced their rating; they denied it, but they did think that their partner’s disposition caused their ratings. Finally (Study 3), yoked judges were asked if they thought such manipulations would affect their own reactions, and none of them predicted that the structure of the interpersonal relationships would affect their trait perceptions. The SCM structural predictions apparently operate at the level of interpersonal interactions, as well as intergroup interactions, but lay people do not seem to be aware of this. Instead, they viewed their perceptions as dispositional—the interpersonal version of stereotyping an outgroup instead of understanding that perceptions are shaped by social structure. Most important, SCM predictions hold for personal impressions as well as for societal stereotypes.
3.4. Cross-cultural tests of SCM structural hypotheses As already described, we collected SCM data in 17 nations. As in the US samples, participants rated the same groups on the traits (warmth and competence) and the two social structure scales (status and competitiveness). In all 20
Warmth and Competence as Universal Dimensions of Social Perception
99
samples, exactly as hypothesized, perceived status highly correlated with competence ratings, with a large effect size (average r ¼ .77). This held both for samples from relatively individualistic societies (average r ¼ .80) and from relatively collectivistic societies (average r ¼ .74). Table 2.3 presents the status– competence correlations for all samples. Although these correlations are large enough, raising issues of divergent validity of the two scales, the status and competence items clearly do not measure the same construct: The status measure is wholly demographic, whereas the competence measure comprises traits. These findings suggest that people generally view their societies as meritocratic, inferring that high-status groups must also be highly competent. The relationship between competitiveness and warmth was not as consistent, but was significant in 70% of the samples, and negative (as predicted) in all samples but one (Israel-Muslim). Across samples, the effect size was small but reliable (average r ¼ -.17). Correlations were similar for relatively individualistic cultures (average r ¼ -.22) and relatively collectivistic cultures (average r ¼ -.21). Table 2.3 presents competition-warmth correlations for all samples. Again, we believe the diminished correlations in some samples can be explained by the fact that we had restricted our measure of competition to zero-sum negative economic interdependence; all items measured power and resource tradeoffs (i.e., if this group gains power, other groups in society lose power; resources going to this group take resources away from the rest of society). If we had used value trade-offs, the effects might have been stronger. Also, all but one (Israel-Muslim) of the samples that failed to produce significant competitiveness–warmth correlations were rating fellow E.U. nations. Our construction of competition might have been less relevant for this particular set of groups, given that either (a) participants and groups belonged to the same superordinate economic category, the European Union, repeatedly primed in the instructions (France, Netherlands, United Kingdom), or (b) participants did not belong to the intergroup context at all (Bulgaria and Norway, non-E.U. members). Nonetheless, in most samples and in the aggregate, competition significantly correlated negatively with warmth.
3.5. Converging theory and evidence A parallel effort to predict intergroup images from structural relations arises from enemy images in political psychology (Alexander et al., 1999). Their taxonomy predicts that groups perceived as having incompatible goals (along with information about their status or power) lead to negative perceptions along the warmth dimension: hostile, untrustworthy, ruthless, evil. Low status and power lead to perceived lack of competence and some degree of warmth. Alexander and colleagues’ parsing of the dimensions differs from ours, as they separate status and power (whereas our model presumes that these two typically go together), as well as goal compatibility (which corresponds to interdependence in the SCM). Thus, image theory creates the possibility of a
100
Amy J. C. Cuddy et al.
Table 2.3 Social structure–stereotype correlations, all studies
Study
US data Fiske et al. (2002b) Study 1: students Study 1: nonstudents Study 2 Study 3 Cuddy et al. (2007) International data EU nations rating EU nations Belgium France Germany Netherlands Portugal Spain UK EU combined Europe Belgium (US groups) Belgium (own groups) Italy: students (own groups) Italy: nonstudents (own groups) Bulgaria (rating EU nations) Norway (rating EU nations) Asia (own groups) Hong Kong Japan South Korea Latin America (own groups) Costa Rica Mexico Israel (own groups) Israel-Jewish Israel-Muslim
Status– competence, r
Competition– warmth, r
.83** .64** .88** .87** .83**
.22** .11* .31** .36** .43**
.72** .63** .68** .84** .85** .87** .85** .89**
.48** .02 .15* .05 .17** .15* .04 .25**
.75** .69** .77** .73** .72** .84**
.30** .33** .21** .19** .10 .09
.87** .75** .64**
.15* .17** .39**
.73** .67**
.17* .16*
.83** .55*
.22** .08
* p < .05. ** p < .01. Note: Correlations were calculated at the level of individual participants, not group means. Consistently across studies, off-diagonal (i.e., status-warmth, competition-competence) correlations were either nonsignificant or significantly smaller than the on-diagonal, predicted correlations reported above.
Warmth and Competence as Universal Dimensions of Social Perception
101
2 2 2 taxonomy of group images, though Alexander et al. (1999) focuses on four cells. Image theory and the SCM overlap, independently providing converging predictions about how structural variables affect stereotyping. Although we would argue that the SCM has advantages over image theory,1 the two theories agree about how two social structural variables-status and competition-predict outgroup images, and image theorists’ research has added supporting evidence for these predictions. Although labeled differently, as noted earlier, gender stereotypes have distinguished stereotypically feminine communal traits (matching the warmth dimension) from stereotypically masculine agentic traits (matching the competence dimension). Social role theory (Eagly, 1987; Eagly et al., 2000) posits that gender stereotypes follow from social structure, specifically from a gendered division of labor: homemakers versus employees, as well as sex-typed distribution in paid occupations corresponding to high-status versus low-status roles. Social role theory holds that perceivers infer traits by observing role-constrained behavior, so when groups tend to be concentrated in certain roles, they receive the stereotypes that follow from those roles. As roles shift, gender stereotypes should too (Diekman and Eagly, 2000). In a fictional portrayal of ‘‘city workers’’ and ‘‘child raisers,’’ role-based stereotypes mimicked gender stereotypes, perhaps rationalizing the distribution of the sexes into social roles (Hoffman and Hurst, 1990). Although social role theory provides a more detailed level of analysis linked to specific roles, we view this framework as compatible with the SCM’s broader approach. Note that the traditional gendered division of labor is organized to be interdependent (women relying on men as providers and men on women as homemakers) and status-driven (men’s roles having higher status and women’s lower status). Thus, social role theory’s more detailed analysis of how social structure affects gender stereotypes and the contents of those stereotypes can be viewed as complementary to the SCM. 3.5.1. Summary Social structural variables, competitive-cooperative interdependence and status, respectively predict warmth and competence stereotypes, across a range of targets, respondents, and paradigms. Although the status– competence effects are larger (rivaling reliability coefficients), and the competition–warmth effects are moderate to small, both are reliable. Moreover, correlational data are supplemented by experimental data and parallel findings from related theories. 1
Our model differs from image theory because the SCM (a) distinguishes warmth and competence as fundamental dimensions of stereotype content, (b) addresses how the attribution of positive traits can reinforce some types of prejudice (e.g., perceived competence can be integral to feelings of envy and resentment), (c) integrates predictions about stereotypes, emotions, and behaviors, and (d) encompasses its predictions in a more parsimonious 2 2 model of stereotyping.
102
Amy J. C. Cuddy et al.
4. Emotional, Behavioral, and Attributional Consequences Moving from antecedents to consequences, how do warmth and competence judgments affect how targets are treated? This section focuses on the impact of stereotypes along these fundamental dimensions. We propose that perceptions of high versus low warmth and competence elicit predictable, differentiated patterns of social emotions, behaviors, and attributions (Fig. 2.2). Prior research and theory, dominated by a view of prejudice as a univalent antipathy, has obscured these distinctive patterns and their social consequences, especially for groups that receive ambivalent stereotypes.
4.1. Prejudiced emotions: Admiration, contempt, envy, and pity 4.1.1. Emotions hypotheses Drawing on varied literatures, we review both correlational and experimental evidence that the four combinations of high versus low warmth and competence judgments create four unique emotional responses: admiration, contempt, envy, and pity (Cuddy et al., 2004, 2007; Fiske et al., 2002a,b). Our predictions build on social comparison-based (Smith, 2000) and attributional (e.g., Weiner, 2005) models of emotion. At the two extremes (a) upward assimilative social comparisons—to people perceived as warm and competent (e.g., ingroups)—elicit admiration and pride (Fiske et al., 2002a) and (b) downward contrastive comparisons—to people perceived as incompetent and cold—elicit contempt and disgust (e.g., poor people; Fiske et al., 2002a; Dijker et al., 1996b). In the worst cases, these latter outgroups are severely dehumanized (Harris and Fiske, 2006). The two ambivalent cases include (c) upward contrastive comparisons—to people perceived as competent but not warm (e.g., Asians; Fiske et al., 2002b; Lin et al., 2005; e.g., Jews; Fiske et al., 2002b; Glick, 2002, 2005)—elicit envy and (d) downward assimilative comparisons—to people perceived as warm but not competent—elicit pity (e.g., the elderly; Cuddy et al., 2005; Fiske et al., 2002b). We address each quadrant in turn, focusing on the emotions. 4.1.1.1. Pity Low-status, noncompetitive groups seen as incompetent but warm receive paternalistic prejudice. As we have noted, exemplars of this cluster include elderly people, disabled people, retarded people, and housewives. These groups elicit pity and sympathy, which is directed toward people with negative outcomes whose causes they cannot control (Weiner, 1980, 1985).
Warmth and Competence as Universal Dimensions of Social Perception
103
Physical disabilities (e.g., Alzheimer’s disease, blindness, cancer, heart disease, paraplegia) are perceived as onset-uncontrollable and worthy of sympathy and pity (Weiner et al., 1988). When asked to describe times they pitied others, students’ stories most often referred to people with physical disabilities or victims of environmental circumstances (Weiner et al., 1982). Thus, only when poverty is attributed to external and uncontrollable societal causes does it evoke pity (Zucker and Weiner, 1993). A variety of stigmatizing conditions—blindness, cancer, AIDS, drug abuse, obesity, and homelessness—all can elicit pity when viewed as onset uncontrollable (Rush, 1998). But some of these are more typically viewed as onset controllable (AIDS, drug abuse, obesity, homelessness), compared to others (blindness, cancer). Viewing persons with AIDS as behaviorally responsible for their condition, for example, reduces pity (Dijker et al., 1996a). Generally, physical stigmas are viewed as onset uncontrollable, eliciting pity, whereas mental-behavioral stigmas are viewed as onset-controllable, failing to elicit pity (Stipek et al., 1989; Weiner et al., 1988). Drawing on Weiner’s attributional analysis, Smith’s (2000) theory of social comparison-based emotions describes sympathy and pity as downward assimilative emotions, directed at lower-status groups (as the SCM predicts). The SCM adds the insight that only noncompeting groups are sufficiently ‘‘assimilated’’ into society to evoke sympathetic concern. We have described pity as paternalistic because it presumes that the perceiver has a dominant, though custodial, role. Paternalism implies subjectively positive emotions directed toward less fortunate or lower status others. Only low-status groups that are viewed as having positive intentions (i.e., are cooperative rather than competitive), however, are likely to elicit paternalism due to the perception that their poor outcomes are outside their control (because of their assumed incompetence). Their cooperativeness casts them as warm, with positive intentions, implying that they would surmount their predicament, if they were able (i.e., are not intentionally ‘‘parasitic’’). In sum, LC–HW people are viewed as deserving pity and sympathy for uncontrollable negative outcomes that occur despite their best intentions. 4.1.1.2. Envy Competent but cold groups elicit envy and jealousy, a response we label envious prejudice. The positive side of envious ambivalence is that such groups are perceived as competent and therefore responsible for their own high status. On the negative side, they are viewed as competitors who lack warmth and have hostile intent. When others’ controllable, positive outcomes deprive the self, people feel envy (i.e., envy occurs when one lacks another’s superior, desired outcome; Parrott and Smith, 1993). Envy is by definition, then, directed upward. Moreover, envy generates dislike. Thus, upward contrastive (i.e., competitive) social comparisons elicit envy and resentment (Smith, 2000). Envy focuses on oneself and the other simultaneously, as a comparison in which the self is at a disadvantage.
104
Amy J. C. Cuddy et al.
Envy entails hostility and depression (Smith et al., 1994); the hostility expresses feelings that the outgroup’s superior position is illegitimate, whereas the depression focuses on one’s own inferiority. Because acknowledging envy implies one’s own lack, this emotion is often couched as righteous indignation of the other’s presumably illegitimate gain (Smith, 1991). Thus, although envied groups may also elicit anger under certain conditions (discussed below, section 4.2.2.), anger is not uniquely expressed toward such high-status groups (because it also can be directed downward, toward low-status, competitor groups). In short, envy seems the more appropriate label for attitudes toward high-status, competitive groups than anger or resentment. However, envy in intergroup perceptions may be particularly difficult to measure (Spears and Leach, 2004); people are loath to admit envy because it implies a deficit in the self or ingroup. 4.1.1.3. Contempt The third combination, low-status, ‘‘free-loading’’ groups perceived as incompetent and not warm, receive what we have termed contemptuous prejudice. Groups that elicit such univalent antipathy evoke anger, contempt, disgust, hate, and resentment. Anger is directed toward those whose negative outcomes are perceived as their own fault (Weiner, 1985) and who are viewed as a drain on the rest of society. For example, because conservatives attribute poverty to internal and controllable individual causes, they are more likely to react with blame and anger (Zucker and Weiner, 1993). Dijker’s (1987) and Dijker et al.’s (1996b) work on native Dutch perceptions of Surinamese versus Turkish or Moroccan immigrants illustrates this link. The Dutch are more likely to feel contempt (in Dijker’s measures this emotional dimension is defined as anger, annoyance, aversion, contempt, and antipathy) toward people of Turkish and Moroccan descent in the Netherlands than toward other minorities. Both Turks and Moroccans tend to be Muslims who immigrated to perform lowwage jobs. Their cultural differences (religious beliefs) and status (low-status jobs) are both viewed as choices rather than uncontrollable circumstances. Similarly, voluntary unemployment and poverty caused by gambling elicit anger (Weiner et al., 1982). As other examples, child abuse, drug addiction, obesity, and AIDS are seen as controllable, blameworthy stigmas, eliciting a high degree of anger (Weiner et al., 1988). People who view homosexuality as both immoral and individually controllable react with anger and contempt to individuals with AIDS (Dijker et al., 1996a). When homelessness is attributed to presumably controllable behaviors or traits, such as drug abuse or laziness, it likewise elicits anger (Barnett et al., 1997). In general, when controllability is manipulated for a variety of stigmas, it elicits perceived responsibility, blame, and anger (Rush, 1998). Again, although anger is relevant, contemptuous prejudice involves more specific emotions, such as disgust and a moralistic resentment that
Warmth and Competence as Universal Dimensions of Social Perception
105
includes overtones of injustice and indignation, bitterness toward illegitimate behavior. Injustice powerfully evokes both anger and disgust toward behavior viewed as immoral and as impeding one’s own (or the ingroup’s) goals and plans (Mikula et al., 1998). Outgroups perceived to have interests that detract from ingroups create competition in a zero-sum sense, inciting anger, but contempt and disgust are downward (i.e., target lower status others) contrastive comparisons (Smith, 2000). Anger, contempt, and disgust all express moral outrage, though at different levels—individual, community, and divinity, respectively (Rozin et al., 1999). Violations of individual standards elicit anger; violations of community standards evoke contempt; and violations of divine standards provoke disgust. Groups perceived as low-status, competitive free-loaders with hostile and exploitative intent that impacts others as individuals, as communities, and as religious believers, elicit strong morally justified contempt. 4.1.1.4. Admiration Some groups with high status do not compete with societal ingroups, either because they are dominant, mainstream ingroups and reference groups or their close allies. Because they have high status but also serve as societal reference groups or further the interests of such groups, they elicit admiration and pride. For example, most Americans identify as middle class or aspire to be, so the middle class serves as a societal reference group. These reference groups, receiving univalent positive regard, elicit pride, admiration, and respect. Pride targets others who attain favorable outcomes (e.g., high status) that also have positive implications for the self. Pride results from self-relevant, positive, controllable outcomes (Weiner, 1985). Pride and self-esteem follow positive outcomes attributed to self, and by extension, to one’s group or reference group. People indeed feel positive about the successes of close others, as long as others’ success in a domain, because of its relevance to self-esteem, does not create an unfavorable comparison for the self (Tesser, 1988). Similarly, because one can assimilate the self to the ingroup, close allies, or societal reference groups, the success of these larger entities can be an occasion for pride, rather then envy, such as when one basks in the reflected glory of a group’s (e.g., a team one supports) success (Cialdini et al., 1976) Thus, upward, assimilative social comparisons elicit admiration and inspiration, according to Smith’s (2000) theory. Pride and admiration should therefore be directed toward successful ingroups, reference groups, and close allies. 4.1.1.5. Summary The SCM’s emotional prejudice hypotheses state that: pity targets low-status, noncompetitive groups seen as warm but incompetent, envy targets high-status, competitive groups seen as competent but cold, contempt targets low-status groups seen as competitive (freeloading), and admiration targets mainstream reference groups—high-status, noncompetitive—seen as warm and competent.
106
Amy J. C. Cuddy et al.
4.1.2. Support for emotions hypotheses: Correlational and experimental 4.1.2.1. Correlational evidence In several US studies, the groups rated in our stereotype content surveys (Cuddy et al., 2007, Study 1; Fiske et al., 2002b, Study 4) were also rated on emotion items assessing admiration (admiring, inspired, proud, respectful), contempt (contemptuous, disgusted, hateful, resentful), envy (envious, jealous), and pity (pity, sympathy). Participants received instructions similar to those used in assessing stereotypes, reporting the emotions that people in (e.g., American) society generally feel toward each group. The correlational studies found that emotion ratings differed as predicted within each of the four warmth–competence clusters. The cold–competent cluster (e.g., Asians, Jewish people, and rich people) elicit envy and moderate admiration, but little contempt or pity. Warm–incompetent groups (e.g., disabled people, elderly people, and mentally retarded people) are pitied, receiving much less admiration and contempt, and no envy. The incompetent–cold cluster (e.g., homeless people, poor people, welfare recipients) evoke contempt, but little pity or admiration and no envy. Warm–competent groups (e.g., Americans, middle-class, and students) elicit significantly more admiration and pride that any of the other three emotion factors. Comparing the clusters within each emotion showed that each emotion’s highest rating occurred for the predicted cluster: admiration was highest for the warm–competent cluster, envy was highest for the cold– competent cluster, pity was highest for the warm–incompetent cluster, and contempt was highest for the cold-incompetent cluster. 4.1.2.2. Experimental evidence We also conducted experimental tests of the intergroup emotions hypotheses (Caprariello et al., 2007). In these experiments, we focused on whether the social structural variables—status and competitiveness—elicit the predicted emotions, investigating the claim that each combination of high and low status and competitiveness would actually bring about the predicted emotional responses. In the fictitious immigrant group study described earlier, participants were also asked ‘‘When members of this ethnic group arrive, to what extent will people here be likely to feel each of the following emotions toward them?’’ and made ratings on the four emotions scales. The results supported three of four predictions. Members of low-competition, high-status groups elicited significantly more admiration and pride than members of all other groups. Members of high-status, high-competition groups evoked more envy than members of other groups. Likewise, members of low-competition, low-status groups garnered more pity and sympathy than other group members. Contrary to predictions, however, members of high-competition, lowstatus groups did not evoke higher ratings of contempt and disgust
Warmth and Competence as Universal Dimensions of Social Perception
107
compared to the other groups, and even evoked less contempt than members of high-competition/high-status groups and low-competition/lowstatus groups. The only group that elicited lower contempt ratings were members of low-competition/high-status groups (i.e., stand-ins for ingroups and reference groups), who evoked significantly less contempt than all other groups. Thus, although our prediction about which group would elicit the most contempt was not supported, we did find the lowcompetition/high-status group elicited the least. Participants gave mostly neutral ratings on contempt and disgust (closer to the midpoint than the other emotions), perhaps indicating reluctance to report contempt and disgust toward any group. This may reflect an unwillingness to indicate strong negative emotions toward others (Sears, 1989). The null result for contempt ratings should, however, be placed in the context of other, more supportive data. First, the student samples and national representative sample survey (Cuddy et al., 2007; Fiske et al., 2002b) showed clear reports of disgust/contempt toward low-warmth, low-competence groups. Second, neuroimaging data show that these groups uniquely fail to elicit activation in the medial prefrontal cortex (mPFC), an area that reliably responds to social stimuli. By contrast, lowwarmth, low-competence groups do activate the insula, an area reliably implicated in disgust (Harris and Fiske, 2006). Together, these neuroimaging results suggest that these groups are viewed as less-than-human. Third, ratings of actual (as opposed to fictitious) immigrant groups perceived as low-status and exploitative (e.g., undocumented migrants) do elicit low-warmth and low-competence attributions (Lee and Fiske, 2006). 4.1.3. Summary Thus, experimental findings support the SCM’s predictions that particular combinations of warmth and competence perceptions create distinct emotional reactions toward groups and individuals. In the intergroup context, this stands in sharp contrast to the presumption that intergroup emotions are typically univalent, either wholly positive (e.g., toward allied groups) or negative (toward outgroups). Understanding these distinct emotional profiles—admiration, envy, pity, and contempt—is of special importance for understanding intergroup behavior, to which we now turn.
4.2. Behaviors: Active and passive, facilitation and harm The BIAS map (Cuddy et al., 2007; Fig. 2.2) builds on the SCM, proposing that the four combinations of high versus low warmth and competence elicit not only differentiated emotions, but also four discrete patterns of behavioral responses: active facilitation (e.g., help), active harm (e.g., harassment), passive facilitation (e.g., convenient cooperation), and passive harm (e.g., neglect). We present both correlational and experimental support for
108
Amy J. C. Cuddy et al.
these patterns for both intergroup (e.g., Cuddy et al., 2004, 2007) and interpersonal (Asbrock and Cuddy, 2008) perception. 4.2.1. Behaviors hypotheses 4.2.1.1. Identifying dimensions of behaviors Past work suggests that two dimensions capture a wide range of intergroup behaviors: active–passive concerns intensity; harm-facilitation concerns valence. The active–passive distinction runs through various areas of psychology, showing that behaviors tend to be enacted with relatively more or less effort, directness, engagement, intent, and intensity. This dimension distinguishes more overt and effortful social behaviors, whether positive or negative, such as helping or harassing, from more subtle types that involve less exertion, such as associating or neglecting. The active–passive dimension has been used to classify a range of interpersonal behavior, including aggression (Buss, 1961), romantic relationship behaviors (Sinclair and Fehr, 2005), leadership styles (Eagly et al., 2003), and minority social influence (Kerr, 2002), among others. Ayduk et al. (2003) describe active behaviors as direct, explicit, overt, confrontational, intense, and high risk. By contrast, passive behaviors are indirect, covert, less intense, and avoidant. ‘‘Passive’’ does not imply a completely inert state (which would make ‘‘passive behavior’’ an oxymoron); rather, in psychology, ‘‘passive’’ has often been used to describe behaviors that require less effort, direction, and intention (e.g., passive aggression) relative to behaviors that are unambiguously active and goal-directed (e.g., active aggression). Passive aggression, for example, includes neglecting to do something (e.g., warning someone of danger), whereby the omission then harms another person. For the intergroup domain, we define active behaviors as those conducted with directed effort to affect the target group; they overtly and directly act for or against the target group. We define as passive behaviors those that are conducted or experienced with less directed effort, but still have repercussions for the outgroup; they act with or without the target group. Passive behaviors may reflect a less deliberate or obvious intention on the part of an actor to bring about a specific outcome, but can constitute consequential forms of discrimination (e.g., passive segregation, failure to hire members of a specific group, neglecting an outgroup member’s welfare, not providing service). On the positive side, passive behaviors represent noncommittal rapprochement, as when prejudiced people cooperate with outgroups when convenient, ‘‘go along to get along,’’ patronize businesses owned by disliked outgroups, or tolerate but neither object to nor endorse the outgroup’s presence. A second frequent distinction concerns the valence of behavior as determined by its intended effect on others. We refer to this second dimension as facilitation-harm. This dimension distinguishes prosocial/ helping behavior from antisocial/aggressive behavior (see Batson, 1998 and Geen, 1998 for reviews). Similarly, interdependence theorists focus on how social behavior facilitates or impedes others’ goals (e.g., Thibaut and
Warmth and Competence as Universal Dimensions of Social Perception
109
Kelley, 1959). In the intergroup context, we define facilitation-harm as follows: facilitation leads to ostensibly favorable outcomes or gains for other groups; harm leads to detrimental outcomes or losses for other groups. Combining the two behavioral dimensions creates four classes of behaviors, along two bipolar dimensions: active facilitation (i.e., acting for) explicitly aims to benefit a target. Interpersonally, these behaviors include helping, assisting, and defending others (e.g., opening a door for someone). At the intergroup level, this would include hiring, promoting, and befriending group members. Institutionally, these behaviors include assistance programs for the needy, corporate charitable giving, progressive tax codes, and anti-discrimination policies. Active harm (i.e., acting against) explicitly intends to hurt a target and its interests. Individual insults, bullying, and attack are individual active harms. Using group epithets, sexual harassment, and hate crimes all constitute group-based active harm. Institutionally, active harm can range from discriminatory policies to legalized segregation to mass internment (e.g., Japanese Americans during World War II) to genocide. Passive facilitation (i.e., acting with) accepts obligatory association or convenient cooperation with a target. Such behavior is passive because contact is not desired, but only tolerated in the service of other goals; facilitation of the other is a mere by-product. Interpersonal examples include tolerating obligatory association in educational, commercial, or professional settings, with people one would not otherwise choose as associates. Intergroup examples include hiring the services of outgroup members (e.g., as domestics) or choosing to work with members of a group assumed to be smart (e.g., Asian Americans) on a team project. Institutionally, realpolitik cooperation with a disliked regime illustrates passive facilitation. Passive facilitation acts with the group for one’s own purposes, but simultaneously benefits the other group as a by-product. Passive harm (i.e., acting without) demeans or distances others by diminishing their social worth through excluding, ignoring, or neglecting. Relational or social aggression (e.g., Crick and Grotpeter, 1995) and passive negative coping (e.g., withdrawal of social support; Ayduk et al., 2003) are related concepts. Interpersonal passive harm includes avoiding eye contact, being dismissive, and ignoring another person. The same behaviors applied on the basis of outgroup membership would constitute intergroup passive harm. Institutionally, passive harm involves disregarding the needs of some groups (e.g., by denying assistance) or limiting access to necessary resources such as education, housing, and healthcare. Passive harm acts without the group, denying its existence, harming its members by omission of normal human recognition. Three hypotheses specify how warmth-competence stereotypes and their corresponding social emotions might predict the four classes of behavioral tendencies.
110
Amy J. C. Cuddy et al.
4.2.1.2. Warmth judgments elicit active behavioral responses Because of its primacy in evaluations of others (as reviewed above), we hypothesized that the warmth dimension predicts the valence of active behaviors: perceivers act against (i.e., active harm) stereotypically cold groups and for (i.e., active facilitation) stereotypically warm groups. Warmth stereotypes theoretically derive from the inferred goals of the target group and the potential benefits or harms caused by these goals (Wojciszke, 2005a,b). The SCM supports this link, illustrating how competitive or exploitative groups (whose goals are perceived as harmful) are stereotyped as lacking warmth, whereas noncompetitive groups (perceived as not having harmful goals) are stereotyped as possessing warmth. Because it is self- and ingroup-relevant, warmth information dominates over competence information in person perception. As reviewed above, compared to competence, warmth accounts for twice as much of the variance in global evaluations of others, is chronically more accessible in descriptions of others, leads to stronger and more stable evaluations of others (Wojciszke et al., 1998), and is identified and judged more rapidly (Hack et al., 2007; Ybarra et al., 2001). Warmth information creates a relatively urgent need to react, leading to active facilitation directed at warm groups and active harm directed at cold groups. 4.2.1.3. Competence judgments elicit passive behavioral responses We hypothesized that competence, being secondary, although still important, in evaluations of others (as reviewed above), predicts the valence of passive behaviors: perceivers act without (i.e., passive harm) stereotypically incompetent groups and with (i.e., passive facilitation) stereotypically competent groups. Perceived competence theoretically derives from the inferred efficacy with which the target’s goals are enacted (Wojciszke, 2005a,b). The SCM’s parallel analysis shows that groups high in status (i.e., having resources or power to carry out goals) are stereotyped as competent, whereas lowstatus groups are stereotyped as lacking competence. In contrast to the exigency of warmth information in person perception, the competence or incompetence of the other is less pressing because it is less self- and ingrouprelevant (Wojciszke et al., 1998). Although competence still contributes significantly, albeit less, to the overall variance in global evaluations of others, it contributes significantly less than compared to warmth, and it is chronically less accessible in descriptions of others. It leads to weaker and less stable evaluations of others (Wojciszke et al., 1998), and is identified and judged more slowly than warmth (Hack et al., 2007; Ybarra et al., 2001). The rationale that competence is less pressing than warmth might seem to imply that behavioral responses would be simply weaker, not specifically passive. We hypothesized that competence leads to passivity specifically because competence (and its structural cause, status) imply control over desired resources. Those with status and resources attract others who hope to share some of that status, competence, or resource by
Warmth and Competence as Universal Dimensions of Social Perception
111
proximity or association. Those groups lacking status, competence, and resources will be ignored and neglected because they have nothing to offer. Thus, inferred competence, although still important in perception of others, does not create as immediate a need to react, thus cuing more passive behaviors, which involve acting with (i.e., passive facilitation) or without (i.e., passive harm) others. 4.2.1.3.1. Summary The first hypothesis states that the warmth dimension of stereotypes will predict the valence (i.e., facilitation versus harm) of active behaviors, and the competence dimension of stereotypes will predict the valence of passive behaviors. Specifically, we predict that warmth stereotypes will elicit active facilitation (e.g., helping) and prevent active harm (e.g., attacking), while competence stereotypes will elicit passive facilitation (e.g., associating with) and prevent passive harm (e.g., excluding). Each combination of warmth and competence stereotypes thus relates to a distinct pattern of behavioral tendencies (see Fig. 2.2). 4.2.1.4. Each emotion predicts a unique pattern of behaviors Although judgments of both warmth and competence (cognitions) and emotions affect behaviors, we have suggested that, consistent with past work reviewed earlier in this chapter, emotions more strongly and directly relate to behavior. Assuming that cognitions cue behaviors and emotions activate them (Frijda et al., 1989), we hypothesized that the specific emotion linked to each SCM combination of warmth-competence stereotypes would also predict the hypothesized behavioral tendencies, with two emotions will predicting each behavioral tendency. These specific links are supported by already reviewed theories that conceptualize discrete emotions as outcomes of social comparisons (e.g., Smith, 2000), outcome attributions (e.g., Weiner, 2005), and cognitive appraisals (e.g., Dijker et al., 1996b; Mackie et al., 2000). Figure 2.2 depicts the BIAS map hypotheses.
4.2.1.4.1. Pity (low-competence/high-warmth) Pity is an ambivalent emotion—comprising both compassion, but also sadness and an implicit sense of superiority over the other—that results from appraising another’s negative outcome as unintentional and uncontrollable (Weiner, 2005). Active facilitation includes help-giving elicited by pity (Weiner, 2005). However, to avoid the sadness and depression pity evokes, people may avoid and psychological distance themselves from unfortunate others. Thus, pity does not always activate help, but can lead to avoidance and neglect, such as turning off an appeal to aid starving children (Green and Sedikides, 1999; Roseman et al., 1994). Pity implicitly involves condescension (i.e., disrespect) and can therefore lead to dismissive behaviors, such as the patronizing speech and poor medical treatment often afforded to elderly
112
Amy J. C. Cuddy et al.
people (e.g., Pasupathi and Lockenhoff, 2002). We hypothesized that pity elicits active facilitation (i.e., actively acting for) and passive harm (i.e., passively acting without). 4.2.1.4.2. Envy (high-competence/low-warmth) Envy covets another’s superior outcome and comprises feelings of injustice or inferiority (Smith et al., 1994). Envy is ambivalent, involving both resentment and respect. Because envy implicitly acknowledges that another group has outdone the ingroup, it cues convenient cooperation that might enable the ingroup to acquire some of the coveted outcome. Envy involves begrudging admiration for the other, an ambivalent type of respect that might produce passive facilitation. However, when societies experience widespread misfortunes and instability, envied groups are likely to be scapegoated because they are perceived to have ability (competence) as well as intent to disrupt society (Glick, 2005; Staub, 1996). Scapegoating can then lead to extremely hostile acts against the envied group, including ‘‘ethnic cleansing’’ and genocide. We hypothesized that envy elicits passive facilitation (i.e., passively acting with) under normal conditions, and active harm (actively acting against) when a society is stressed. 4.2.1.4.3. Contempt (low-competence/low-warmth) Contempt and disgust target those with negative outcomes perceived as onset-controllable (Weiner, 2005) and who are viewed as a drain on valuable resources. Contempt-related emotions elicit passively harmful actions such as demeaning, condescending behaviors (Brewer and Alexander, 2002); neglect (Weiner, 2005); and distancing, excluding, or rejecting (Roseman et al., 1994; Rozin et al., 1999). Disgust also motivates attempts to remove a noxious stimulus from one’s perceptual field, eliciting the desire to forcefully expel or obliterate the stimulus (Plutchik, 1980; Roseman et al., 1994). We hypothesized that contempt elicits both active harm (i.e., actively acting against) and passive harm (i.e., passively acting without). 4.2.1.4.4. Admiration (high-competence/high-warmth) Admiration and pride target successful others whose positive outcomes either do not detract from or enhance the self (Tesser and Collins, 1988). Admiration and pride motivate contact (Dijker et al., 1996b) and relate to cooperation (Alexander et al., 1999); happiness, a linked emotion, predicts positive approach behaviors (Neuberg and Cottrell, 2002). We hypothesized that admiration elicits active facilitation (i.e., actively acting for) and passive facilitation (passively acting with). 4.2.1.5. Bias clusters The SCM and BIAS hypotheses imply coordinated ‘‘bias clusters’’ of specific stereotypes, distinct emotions, and pairs of behavioral tendencies. Further, if the specific hypothesized links are supported, ambivalent bias clusters should emerge: groups with ambivalent
Warmth and Competence as Universal Dimensions of Social Perception
113
warmth-competence stereotypes (i.e., high on one dimension, low on the other), and ambivalent emotions (i.e., envy, pity) will be targets of ambivalent patterns of intergroup behaviors—one facilitation behavior and one harm behavior. We predict that high-competence/low-warmth stereotypes will link to passive facilitation and active harm, and low-competence/highwarmth stereotypes will link to active facilitation and passive harm (see Fig. 2.2). Consistent with the third principle presented earlier—that emotions more strongly and directly predict behaviors—the BIAS map also predicts that the relationship between emotions and behavioral tendencies will be stronger than the relationship between stereotypes and behavioral tendencies, and emotions will mediate the stereotypes ! behaviors relationship. 4.2.2. Support for behaviors hypotheses: Correlational and experimental 4.2.2.1. Correlational results The behavior hypotheses were tested by Cuddy et al. (2007) using behavior scales developed in a preliminary study with students. Study 1 used a representative national sample, collected via a random-digit dialing telephone survey (N ¼ 571, 62% female, average age ¼ 43.5) of English-speaking adults, 18 or older, in the 48 contiguous United States. Data were weighted on gender, age, education, census region, and race-ethnicity to match U.S. Census Bureau estimates. The phone survey listed 20 US social groups, chosen from previous studies (Fiske et al., 1999, 2002b; Katz and Braly, 1933), with five groups likely to represent each of the four quadrants of the warmth-competence space. Each participant rated 4 of the 20 groups (1 group randomly chosen from each of the predicted quadrants), using scales that tapped perceptions of competitiveness and status (social structure); warmth and competence (stereotypes) admiration, contempt, envy, and pity (emotions); and the four types of behaviors—active harm, passive harm, active facilitation, and passive facilitation. All but the behavioral tendencies scales were adapted from previous research, and each scale included the two items with the highest average factor loadings across previous studies (Fiske et al., 1999, 2002b). The four behavioral tendencies scales were developed based on data from a preliminary study: attack, fight (active harm); exclude, demean (passive harm); help, protect (active facilitation); and cooperate with, associate with (passive facilitation). Using 5-point scales (1 ¼ not at all; 5 ¼ extremely), participants rated how the groups ‘‘are perceived [treated] by Americans.’’ As before, this instruction was intended to assess perceived societal reactions and to reduce participants’ social desirability concerns. 4.2.2.1.1. Stereotypes ! Behaviors Results of our national sample survey document four hypothesized patterns of discriminatory behavioral tendencies, based on warmth-competence stereotypes and related
114
Amy J. C. Cuddy et al.
emotions. As expected, warmth ratings correlated positively with active facilitation and negatively with active harm. Competence ratings correlated positively with passive facilitation and negatively with passive harm (absolute values ranged from .55 to .77 at the group level of analysis and from .34 to .64 at the individual level; see Table 2.4). We also examined the stereotypes ! behaviors hypothesis by comparing behavioral intentions toward groups in the high-warmth versus low-warmth clusters, and groups in the high-competence versus low-competence clusters. These, too, supported our predictions (see Table 2.5). Groups stereotyped as possessing warmth elicited more active facilitation and less active harm than groups stereotyped as lacking warmth. Groups stereotyped as competent elicited more passive facilitation and less passive harm than groups stereotyped as lacking competence.
Table 2.4 Correlations of behavioral tendencies with stereotypes and emotions (Study 1) Predictor
Behavioral tendency Active facilitation
Stereotypes (group-level) Competence .08 Warmth .73*** Emotions (group-level) Admiration .59** Contempt .63** Envy .06 Pity .51* Stereotypes (participant-level) Competence .17*** Warmth .47*** Emotions (participant-level) Admiration .49*** Contempt .24*** Envy .00 Pity .40***
Active harm
Passive facilitation
Passive harm
.20 .55***
.77*** .45*
.68*** .24
.35 .93*** .22 .10
.95** * .46* .57** .48*
.69** .48* .39 .65**
.10** .34***
.64*** .53***
.50*** .24***
.31*** .54*** .21*** .00
.74*** .33*** .43*** .26***
.58*** .48*** .25*** .41***
* p < .05. ** p < .01. *** p < .001. Note: Bolded correlations were predicted to be significant (23/24 were). But 15/24 others were also significant; although they were theoretically consistent, most appeared in the participant-level analyses, which had especially high power (participant d.f. ¼ 569, group d.f. ¼ 18).
Table 2.5 Behavioral tendencies standardized means (and standard deviations) by competence and warmth stereotypes (Cuddy et al., 2007, Studies 1 and 2) Warmth High
Study 1 (measured stereotypes) Active facilitation .331a (.82) Active harm .305b (.78) Study 2 (manipulated stereotypes) Active facilitation .352a (.98) Active harm .325b (.97)
Competence Low
High
Low
.350b (.61) .323a (.91)
Passive facilitation Passive harm
.277a (.51) .270b (.72)
.333b (.80) .318a (.80)
.371b (.89) .343a (.92)
Passive facilitation Passive harm
.345a (.89) .310b (.98)
.352b (.99) .292a (.93)
Note: Within study, within trait (i.e., warmth, competence), means not sharing a subscript differ at p < .01. All predicted differences are significant.
116
Amy J. C. Cuddy et al.
4.2.2.1.2. Emotions ! Behaviors The national phone survey study also supported the hypothesized relationships between specific positive and negative social emotions (admiration, contempt, envy, and pity) and unique patterns of intergroup behavioral intentions. Correlational data strongly supported seven of eight of the specific predicted links. Although the envy to active harm link was significant only at the individual level, but not at the group level, subsequent analyses clarified that envy operates via anger to predict behavior (see below). At a minimum, the four combinations of warmth-competence stereotypes formed distinct bias clusters, linking with predicted patterns of emotions and behavioral tendencies, including the hypothesized ambivalent clusters. Groups stereotyped as high on competence but low on warmth elicited envy and passive facilitation but active harm. Groups stereotyped as low on competence but high on warmth, on the other hand, elicited pity and active facilitation but passive harm. Consistent with previous research (e.g., Dovidio et al., 1996; Talaska et al., 2007), emotions tended to more strongly and directly predict behavioral tendencies than did stereotypes. Following an appraisal ! emotion ! behavior sequence (e.g., Mackie et al., 2000), for each behavioral tendency, at least one emotion mediated the stereotype ! behavior link. However, some emotions took priority over others. Admiration fully mediated the relationship between warmth stereotypes and active facilitation, and partially mediated the relationship between competence stereotypes and passive facilitation. Contempt fully mediated the relationship between warmth and active harm, and pity fully mediated the relationship between competence stereotypes and passive harm. Figure 2.6 presents media results. Only envy did not mediate any relationships of stereotypes to behavioral tendencies, which led us to conduct a second correlational study to better understand its predicted link to active harm (Cuddy et al., 2007, Study 4). Intergroup envy may elicit active harm only when a society is under great stress and heightened intergroup competition (Glick, 2002, 2005; Staub, 1996). For example, the Nazis viewed Jews as powerful, competent manipulators who had engineered Germany’s defeat in World War I and the subsequent economic crisis. In Rwanda, the Tutsi, also a high-status minority, were similarly blamed for the nation’s economic problems. Active harm (at its most extreme, genocidal attack) can be justified and motivated when an outgroup is viewed as a powerful, capable competitor or exploiter (discussed in greater detail below, section 5.2.). We therefore hypothesized that anger, which can be elicited by the circumstances just described and which leads to offensive actions toward others (e.g., Mackie et al., 2000), would mediate the link between envy and active harm. Participants in this study rated a shorter list of groups on the four SCM emotions scales (admiration, contempt, envy, pity) in addition to two new scales measuring anger and fear. As expected, anger fully mediated the link from envy to active harm.
117
Warmth and Competence as Universal Dimensions of Social Perception
A
B
Admiration .47**
Warmth
.07 (.73**)
Active facilitation
Warmth
Pity
C
Competence
D .96**
.20 (.73**)
Passive facilitation −.17
.76** Envy
.11 (−.53**)
Active harm .01
Envy
Admiration .67**
.96**
−.19
.66**
.40*
Contempt −.64**
.85**
Contempt −.12
Competence
.49* −.14 (−.66**)
−.68**
Passive harm
.67* Pity
Figure 2.6 BIAS map mediation models of stereotypes predicting emotions predicting behaviors, for each quadrant of the stereotype content model space. Source: Cuddy et al. (2007). Reproduced by permission.
4.2.2.2. Experimental results: Intergroup and interpersonal To test proposed causal links to behaviors, we conducted experiments of both intergroup (Cuddy et al., 2007) and interpersonal (Cuddy, unpublished data) perception and behavioral intentions.
4.2.2.2.1. Intergroup We conducted two experiments that varied perceived stereotypes and emotions toward a fictitious immigrant group (adapted from Caprariello et al., 2007) and assessed behavioral intentions (on scales similar to those used in the national phone survey). Both using student samples, one experiment varied warmth and competence stereotypes (Cuddy et al., 2007, Study 2) and the other varied emotions elicited by the group (Cuddy et al., 2007, Study 3). Data from the first experiment strongly supported causal links between warmth and competence stereotypes and, respectively, active and passive behavioral tendencies, perfectly replicating the pattern seen in the phone survey data. Stereotypically warm groups elicited more active facilitation and less active harm than stereotypically cold groups, regardless of stereotypic competence. Stereotypically competent groups elicited more passive
118
Amy J. C. Cuddy et al.
facilitation and less passive harm than stereotypically incompetent groups, regardless of stereotypic warmth. Table 2.5 presents results of this experiment alongside results from the phone survey. These results fit the general notion that cognitive appraisals predict action tendencies (Mackie et al., 2000), but are tailored to the BIAS map’s two-dimensional space. Findings from the second experiment also clearly supported the hypothesized causal links between the typical emotion said to be aroused by the group and behavioral tendencies toward that group, also replicating the four different patterns of behavioral tendencies documented in the national phone survey. Active facilitation was higher for admired and pitied groups, compared with envied and hated groups, who elicited higher active harm. Passive facilitation was higher for admired and envied groups, compared with hated and pitied groups, who elicited higher passive harm. Effect sizes for the ambivalent emotions, envy and pity, were smaller than effect sizes for the univalent emotions, admiration and contempt, but all significantly followed the hypothesized patterns. 4.2.2.2.2. Interpersonal In another experiment using a scenario design (Asbrock and Cuddy, 2008), we asked participants to imagine the following: You are taking a fairly small (i.e., fewer than 30 students) upper-level course in your major. In this course, 60% of your final grade will be determined by your score on a joint project with another student. The project involves conducting research and writing a 20–25-page report of your findings. You cannot choose your partner; the professor makes the assignments.
The next sentence, which included the warmth and competence manipulations, read: The person assigned to work with you is clearly one of the smartest (or least smart) and most competent (or incompetent) students in the class, and (or but), is also friendly (or unfriendly) and warm (or cold) toward other students.
Then we asked participants: Try to imagine how your interactions with this person would unfold as you work to finish the project. Please rate how likely it is that you would behave in each of the following ways toward your partner.
Using a 5-point scale from ‘‘not at all likely’’ to ‘‘very likely,’’ participants made ratings on a series of behaviors, which tapped all four categories of BIAS map behaviors. Active facilitation items were ‘‘share notes with partner’’ and ‘‘do more work to help my partner.’’ Active harm items were ‘‘take undue credit for my partner’s work’’ and ‘‘tell my professor (i.e., tattle) if I don’t think my partner is doing his or her share.’’ Passive facilitation items were, ‘‘accept my partner’s ideas’’ and ‘‘take my partner’s ideas seriously.’’ Passive harm items were ‘‘avoid meeting with my partner,’’ and ‘‘dismiss my partner’s ideas.’’
Warmth and Competence as Universal Dimensions of Social Perception
119
Replicating the intergroup findings, competence affected the passive, but not the active behaviors. Participants endorsed higher passive facilitation and lower passive harm for competent partners than for incompetent partners. Participants also endorsed higher active facilitation and lower active harm for warm partners than for cold partners. 4.2.3. Summary The BIAS map hypotheses concern how specific combinations of warmth and competence stereotypes plus the distinct associated emotions give rise to distinctive patterns of behavior toward groups. These findings also appear to generalize to individuals who are characterized on warmth and competence traits. As predicted, two types of groups (those that are uniformly high or uniformly low on both warmth and competence) arouse consistent behavioral tendencies (i.e., both active and passive facilitation or active and passive harm). By contrast, groups that are stereotyped ambivalently arouse conflicting behavioral tendencies, such as the passive facilitation but active harm elicited by envied groups or the active facilitation but passive harm elicited by pitied groups. Research thus far has generally supported the predicted links, as well as the hypothesis that emotions mediate the effects of warmth and competence stereotypes, and therefore exert a stronger and more direct influence on behavior. However, more research is needed on the moderators that determine when ambivalently stereotyped groups elicit facilitative versus harmful behavior. Because envious and pitying prejudices (the ambivalence specified by the SCM) entail a more complicated pattern of responses and have previously received less attention, the next section focuses more closely on these two types of bias.
5. Spotlight on Ambivalent Combinations: Warm–Incompetent and Competent–Cold A unique feature of our work is its focus on ambivalent combinations of warmth and competence judgments—competent and cold, and warm and incompetent. Our research has demonstrated that groups with ambivalent warmth-competence stereotypes elicit both ambivalent emotions (Cuddy et al., in press; Fiske et al., 2002b) and ambivalent behavioral responses (Cuddy et al., 2007). We discuss findings concerning both types of ambivalent prejudice, pitying and envious, in greater detail and then apply these insights to interpersonal perception.
5.1. Pitying prejudice: Warm but incompetent In this section, we give a more detailed analysis of the two types of pitying, paternalistic prejudice that have received the most research attention: ageism and sexism toward traditional types of women. Both of these
120
Amy J. C. Cuddy et al.
prejudices are prevalent and appear to generalize across cultures (despite past belief that ageist paternalism might be a uniquely Western phenomenon). 5.1.1. Older people Strong evidence suggests that the global category ‘‘elderly people’’ falls squarely into the warm and incompetent cluster of stereotyped groups (Cuddy et al., 2005; Fiske et al., 1999, 2002b; Heckhausen et al., 1989; Hummert et al., 2002; Kite et al., 1991). In the correlational studies (Cuddy et al., 2007; Fiske et al., 1999, 2002b), elderly people reliably fell into the warm–incompetent cluster, alongside disabled and retarded people (see Fig. 2.1). Competence ratings averaged only 2.63 out of 5 (below the scale mid point), and warmth ratings averaged 3.78 (well above the scale mid point). The two scores differed significantly from each other in all samples. As predicted, participants most endorsed items reflecting paternalistic prejudice (pity and sympathy) and least endorsed emotions reflecting envious prejudice (envy and jealousy) toward elderly people. Other researchers have also found that older people are viewed as possessing far fewer competence than warmth traits. Compared to younger people, older people are rated as warmer and friendlier, but also as less ambitious, less responsible (Andreoletti et al., 2001), and less intellectually competent (Rubin and Brown, 1975). When Kite et al. (1991) asked people for their perceptions of young and old women and men, they found that age stereotypes trumped gender stereotypes; regardless of gender, older people were rated as more feminine (i.e., warm) and less masculine (i.e., competent) than younger people. People attribute memory failures of older adults to intellectual incompetence and memory failures of younger adults to lack of attention or effort (Erber and Prager, 1999; Erber et al., 1992, 1993, 1996). In a study of perceptions of life span development, participants predicted that competence-related traits (independent, industrious, intelligent, productive, self-confident, smart) were likely to be lost about nine years earlier (age 72.3) than warmthrelated traits (affectionate, friendly, good-natured, kind, trustworthy; age 81.3), a significant difference (Heckhausen et al., 1989). Finally, several studies have uncovered incompetence stereotypes of older people in the workplace; older employees are believed to be less effective than younger employees in various job-related tasks (Avolio and Barrett, 1987; Rosen and Jerdee, 1976a,b; Singer, 1986). These stereotypes persist even though several studies show that older employees exhibit at least as much, and sometimes more, competence on the job compared to younger employees (McCann and Giles, 2002). There is some reason to believe East Asian collectivistic cultures may be less ageist, given their greater emphasis on filial piety, which is deeply rooted in the Confucian teachings that helped shape some East Asian cultures (Sung, 2001). Our own cross-cultural data (reviewed above) repudiate
Warmth and Competence as Universal Dimensions of Social Perception
121
the assumption that Westerners stand alone in their perception of elderly people as sweet and feeble (Cuddy et al., in press). In all of the samples that rated ‘‘elderly people’’ (Belgium, Costa Rica, Hong Kong, Japan, Israel-Jewish, Israel-Muslim, South Korea), participants viewed them as significantly more warm than competent. Most interestingly, in our three most collectivistic samples—Hong Kong, Japan, and South Korea—this pattern held up, with (a) elderly warmth being significantly higher than elderly competence in all samples, and (b) elderly warmth being significantly higher than the respective overall warmth means and significantly lower than the respective overall competence means. As expected and consistent with the SCM, elderly people were viewed as low status and noncompetitive. Pancultural ageism has been documented in other lines of research, as well. Harwood and colleagues (1996) surveyed participants in six nations, asking people to identify traits they associate with elderly people. Factor analysis uncovered two main dimensions—personal vitality (competence) and benevolence (warmth). Across samples, elderly people were rated as significantly higher on benevolence than on personal vitality. Moreover, the authors were surprised to find that East Asian participants, specifically Hong Kong residents, reported generally negative evaluations of elderly people. In other cross-cultural investigations, participants in China (Tien-Hyatt, 1986/87), Japan (Koyano, 1989), Taiwan (Tien-Hyatt, 1986/87), and Thailand (Sharps et al., 1998) reported even more negative attitudes about older people’s incompetence than their American counterparts. In short, there is mounting evidence that ageism is pancultural. Moreover, the elderly stereotype seems to be somewhat intractable. One study manipulated the competence of an elderly man, then had participants rate the target on warmth and competence (Cuddy et al., 2005). Strikingly, positive competence information did not affect competence ratings, although it did decrease the elderly target’s perceived warmth. Apparently, the negative aspect of the elderly stereotype (incompetence) resists change, whereas the positive aspect (warmth) is more malleable. Unfortunately, the elderly target who behaved more incompetently also gained in warmth. Thus, an incompetent elderly target was rewarded on his group’s positive stereotype (warm) for behaving consistently with his group’s negative stereotype (incompetent). This suggests that when members of ambivalently stereotyped groups behaviorally confirm their negative stereotype (e.g., incompetence), they may inadvertently enhance their positive stereotype (e.g., warmth). This finding resembles the rewards accorded to women by sexists: when women behave incompetently, they are accorded paternalistic BS, being valued on warmth dimensions; but when they behave competently, they receive HS, being devalued on warmth dimensions (Glick and Fiske, 1996, 2001a; Glick et al., 1997).
122
Amy J. C. Cuddy et al.
5.1.2. Traditional women Like elderly people, some women (our second case study in this section), namely those who are assimilated into traditional gender stereotypes, are subjected to paternalism. Eagly and her colleagues (Eagly and Mladinic, 1994; Eagly et al., 1994) established that, contrary to the assumption that subordinate groups receive negative stereotypes, women are stereotyped more positively than men. This has been termed the ‘‘women are wonderful’’ effect and occurs because women are seen as so communal and warm that these characteristics overcome the advantage men have on stereotypes of agency or competence (also positively valued traits). Thus, traditional images of women place them squarely into the paternalized, pitying category in the SCM’s warmth competence space. Ambivalent sexism theory (Glick and Fiske, 1996, 2001a) posits that women are paternalized because they traditionally have lower status and power, yet men are intimately interdependent on women (as mothers, wives, and romantic objects). In other words, women traditionally fit the SCM category of a lower status, but cooperative group, fostering paternalistic sexist attitudes. This subjectively positive aspect of ambivalence toward women, BS, is measured by such items as ‘‘Women should be cherished and protected by men’’ (Glick and Fiske, 1996). Consistent with its paternalistic flavor, BS is associated with subjectively positive stereotypes of women, a finding that holds true in samples from a variety of nations (Glick et al., 2004). At the same time, women are not a homogeneous group and some are viewed as competitors to men, particularly in light of social changes such as the feminist movement and women’s influx into the paid work force. Thus, BS (or paternalism) toward traditional women is accompanied by HS toward nontraditional subtypes of women such as feminists and career women. Contemporary HS represents an envious prejudice toward women who are viewed as competitors with ‘‘unfair advantages’’ (e.g., one HS item complains, ‘‘When women lose to men in a fair competition, they typically complain about being discriminated against’’ and another that women unfairly seek ‘‘special favors, such as hiring policies that favor them over men, under the guise of asking for ‘equality’’’). Because HS and BS are directed toward different subtypes of women, these ideologies are psychologically compatible. That is, it is not psychologically inconsistent to have envious hostility toward feminists and career women, but paternalistic benevolence toward traditional types such as homemakers (Glick et al., 1997; see also Haddock and Zanna, 1994; MacDonald and Zanna, 1998). Thus, different types of women inhabit different quadrants of the SCM space, with career women typically viewed as competent but not warm, whereas homemakers are usually characterized as warm but not competent (Fiske et al., 2002b; though for an anomalous
Warmth and Competence as Universal Dimensions of Social Perception
123
finding, see Cuddy et al., 20072). Consistent with Jackman’s (1994) view of paternalism as a form of social control using an ‘‘iron hand in a velvet glove,’’ BS provides incentives for women to remain in their cooperative, lower-status traditional roles, whereas HS deters them from becoming competitors seeking higher status roles. Despite its positive tenor, the affection BS accords women is contingent on women remaining ‘‘in their place’’ and is itself a potent force for reinforcing gender inequality. Cross-cultural research using the ASI (Glick et al., 2002, 2004) has shown that nations in which people more strongly endorse BS exhibit less structural gender equality, as assessed by nationallevel indicators such as the percentage of women in powerful positions in government (e.g., as representatives in parliaments) and business (e.g., as managers). Although this evidence is correlational, experimental studies show a specific way in which BS can act to reinforce gender inequality— by undermining women’s resistance to it. Jost and Kay (2005) found that women who were merely exposed to items from the BS scale showed increased scores on a measure of system-justification. Compared to women in a control condition and those who were exposed only to HS items, women who saw BS items viewed society as fairer. Collective action to address inequality does not occur unless people feel they are being treated unfairly (e.g., Ellemers, 2001). Jost and Kay’s findings dovetail with research showing that women’s BS scores are especially strongly correlated (more so than men’s BS scores) with endorsement of other traditional gender ideologies, a finding that appears consistently across cultures (Glick et al., 2004). In other words, when women accept BS it is likely that they will also accept their traditional position in society. The soft bigotry of paternalism may be particularly effective because it offers women a subjectively positive (as opposed to hostile) attitude toward their group as a justification for gender inequality. BS also links to women’s romantic fantasies, keeping their agentic ambitions in check (Rudman and Heppen, 2003). Although we have noted that HS tends to be directed toward nontraditional and BS toward traditional types of women, the distinctions between these types of women often break down in contemporary society. In particular, many career women are also mothers. As a result, whether women elicit paternalistic or hostile treatment may depend on which subcategory is salient. It may seem on the surface that being paternalized is less damaging than being treated with hostility, but paternalistic discrimination creates important negative effects. For example, when working women become mothers, activating a traditional role, they lose perceived competence and gain perceived warmth (Cuddy et al., 2004). In a study of 2
We continue to explore the anomaly whereby housewives are rated in the HC-HW cell in this sample, compared to all the others. The most obvious explanation (housewives rating themselves) is not supported. Our best guess is that the meaning of competence can sometimes shift with context.
124
Amy J. C. Cuddy et al.
perceptions of management consultants, people expressed less interest in hiring, promoting, and further training a working mother as compared to a childless female worker. But perhaps most importantly, and as might be expected in a work context, only perceived competence predicted positive behavioral intentions regarding hiring, promotion, and education. Consistent with the BIAS map, such discrimination reflects passive (though severely damaging) harm; the boost in working mothers’ perceived warmth did not help them professionally, whereas their apparent loss in perceived competence hurt them. Like the elderly, women who are assimilated into traditional gender stereotypes are subjected to paternalism. Other research demonstrates how women in work contexts receive both active facilitation, in the form of lavish praise for accomplishments, combined with passive harm, in the form of failure to assign important tasks or leadership roles (Vescio et al., 2005). This paternalistic combination is particularly frustrating as the praise sets up expectations for tangible rewards that are not forthcoming. Of course, praise is cheap, but tangible rewards lead to higher status and power. Thus, although paternalistic treatment of women may come with a smile, it harms women professionally, illustrating an important domain in which paternalism continues to work against equality. 5.1.3. Summary Paternalism’s subjective positivity only masks and renders more effective the subtle ways by which it undermines equality. We have focused here on paternalistic treatment of the elderly and women, but other groups face paternalism as well. Most prominently in prior SCM work, the disabled also appear in the warm but not competent quadrant, a group that deserves further study. Further, Jackman (1994) argues that there are strong elements of paternalism in the treatment of African Americans and in prejudice toward people of low socioeconomic status. Paternalistic treatment’s combination of active facilitation (e.g., being praised for accomplishing easy deeds) and passive harm (e.g., failure to give tangible rewards) is merely a recipe for placating subordinated groups while denying them any rise in status.
5.2. Envious prejudice: Competent but cold The other ambivalent category addresses groups perceived to be competent but not warm, usually directed toward ‘‘model minorities’’ or powerful groups that are viewed as competitors. Here, we focus on envious prejudice toward Asian Americans and Jews. We also consider how, in extreme cases, envious prejudices can motivate the most extreme forms of discrimination, ethnic cleansing and genocide.
Warmth and Competence as Universal Dimensions of Social Perception
125
5.2.1. Asians and Asian Americans In the SCM surveys, Asians appear in the respected-but-disliked envious prejudice cluster, stereotyped as high competence but low warmth. Over time, Euro-Americans’ stereotypes have characterized Asians as intelligent but unsociable. In the classic Katz and Braly (1933) stereotyping study, Japanese were seen as intelligent, industrious, progressive, and shrewd (i.e., competent) but shy and quiet (unsociable); Chinese were sly (implying competence) but conservative, tradition-loving, superstitious, and loyal to family (implying deficient mainstream sociability). The combination of positive and negative stereotypes regarding competence and sociability was an early sign that the Asian outgroup can be perceived relatively favorably on, at most, only one dimension. Similar stereotyping trends held during following decades, with Chinese and Japanese Americans viewed as competent (intelligent, industrious) yet lacking in sociability toward the dominant group (quiet, shy; Gilbert, 1951; Karlins et al., 1969; Maykovich, 1972). Most recently, in a replication of the Princeton Katz–Braly paradigm (Leslie et al., 2007), both Chinese and Japanese were seen as especially intelligent, industrious, and scientifically minded (highly competent) but also loyal to family ties and reserved (still not sociable toward dominant group). Compared to Whites, Asians also have been categorized as more self-disciplined and traditional (again, relatively competent) but as less popular, less sexually loose, and less materialistic (again, relatively unsociable Jackson et al., 1996). Thus, Asian American stereotypes over time demonstrate that the dominant group tends to characterize Asians along the lines of competence and unsociability. The model minority stereotype is the most contemporary view of Asian Americans; it emphasizes their perceived competence by portraying them as diligent and successful in their economic and educational endeavors. In accord with the SCM, the popular stereotype, although seemingly positive, actually carries mixed feelings of simultaneous respect and resentment. Asians may be judged favorably on competence because the White ingroup praises and promotes competence. However, given the tendency for positive attributes to be appreciated as assets only when they reflect well on oneself and the ingroup (Hurh and Kim, 1989), the Asian outgroup’s presumed competence could instead engender group threat and competition (see Insko and Schopler, 1998, for a review of assumed intergroup competition). Prejudiced Whites are most likely to interpret favorable competence characteristics as competitive with the ingroup and the mainstream and therefore subjectively unfavorable. Thus, we can expect racially biased perceivers to consider Asian Americans as excessively and unfairly high in competence. Mixed feelings about the perceived competence of Asian Americans emerges specifically within the context of positive attributes being regarded as negative when the outgroup is believed to possess them. In a series of studies, Maddux et al. (2008) examined the role of realistic threat in explaining how it is that people who hold ‘‘positive’’ stereotypes (i.e.,
126
Amy J. C. Cuddy et al.
competence) about Asian Americans often harbor negative attitudes and emotions toward them. In the first study, in a realistic threat context, attitudes and emotions toward an anonymous group described by only positive, ‘‘model minority’’ attributes were significantly more negative than when the group was described using other positive attributes. A second study demonstrated that realistic threat significantly mediated the relationship between (a) the endorsement of both positive and negative stereotypes of Asian Americans, and (b) subsequent negative attitudes and emotions toward them. Two additional studies conceptually replicated this effect in experimental situations involving interactions with Asian Americans. The representation of Asians as highly competent hard workers does not allow room for corresponding levels of sociability. Consequently, the model minority image reinforces stereotypes of Asian Americans lacking interpersonal skills and not often interacting with others. The low levels of sociability identified with Asians also supports tendencies toward outgroup derogation of Asians. That is, one function of viewing them as competent yet unsociable is to justify a system whereby competence is rewarded but some competent groups are rejected on other grounds, such as lacking sociability (Glick and Fiske, 2001b; Jost et al., 2001). Being stereotyped as competent yet unsociable makes Asians potential targets of a prejudice tinged with envy and discomfort. Anti-Asian American prejudice thus exemplifies envious prejudice, the type directed against outgroups viewed as competent but not warm, according to the SCM. We maintain that the possibility of competitive, threatening relationships between Whites and Asians underlies the tendency to disparage, fear, and discriminate against them. On the basis of these theoretical assumptions, Lin et al. (2005) conducted six studies to demonstrate the viability of a mixed Asian stereotype in which low sociability justifies hostility toward high competence. Excessive competence and deficient sociability factors underlie the Scale of Anti-Asian American Stereotypes (SAAAS), designed to show this envious mixed prejudice. Study 1 began with 131 racial attitude items, using an exploratory factor analysis to examine the factor structure of the SAAAS items and winnowed the SAAAS to 25 items for Studies 2 and 3, which tested 684 respondents on the focused version. Study 2 included a confirmatory factor analysis to confirm the factor structure of the SAAAS obtained in Study 1. Study 3 replicated the results from the Study 2 confirmatory factor analysis in another sample (cross-validation). Study 4 tested the scale’s validity by examining whether extreme scores could predict everyday social behaviors toward Asian Americans. Study 5 replicated Study 4 at another campus, examining the whole spectrum of scores and separating the impact of each hypothesized subscale, sociability and competence. Altogether, Studies 4 and 5 tested the final 25-item SAAAS on 222 respondents at 3 campuses. Scores predicted outgroup friendships, cultural experiences, and (over)
Warmth and Competence as Universal Dimensions of Social Perception
127
estimated campus presence. Study 6 explored whether perceived lack of sociability or perceived excessive competence underlies anti-Asian discrimination in an actual encounter. It showed that allegedly low sociability, rather than excessively high competence, drives rejection of Asian Americans, consistent with the functions of the SCM and with system justification theory ( Jost and Banaji, 1994). Overall, the SAAAS demonstrates mixed, envious anti-Asian American prejudice, contrasting with more-often-studied contemptuous racial prejudices (i.e., against Blacks). 5.2.2. Jews Anti-Semitism has taken the form of an envious prejudice for almost 2000 years. Early Christians needed to differentiate themselves from Judaism, which was the more established religion from which Christianity sprang, and needed to explain why more Jews did not accept Jesus as the messiah. In other words, Judaism initially had greater status and was an ideological competitor, creating the conditions that generate envious prejudice. This prejudice made its way into the New Testament, which includes accusations that Jews were ultimately responsible for killing Jesus, even though it was the Romans who were in charge, tried him, and administered a uniquely Roman form of execution, crucifixion (Sandmel, 1978). Anti-Semitism became deeply engrained in Christendom and beliefs that the Jews were in league with the devil attributed both evil intentions and supernatural powers (i.e., extraordinary competencies) to them. For instance, medieval Christians blamed Jews for engineering the Black Plague and forced them to wear special badges to identify themselves long before Nazis resurrected this idea (Rubenstein, 1966). The Christian prohibition against money lending, a role left to Jews, created fertile grounds for later secular anti-Semitism. ‘‘Middle-man’’ minorities are viewed as competent but cold competitors who profit off the misfortunes of others who need to use their services (Zenner, 1987). Medieval European pogroms and expulsions alternated with periods of tolerating the Jews as a kind of necessary evil given their middle-man role in the economy. This pattern fits the BIAS map’s notion that envied groups elicit passive facilitation when they are viewed as serving a function, but also active harm when a society is under stress and the envy has transformed into anger. The Nazis defined the Jews as unalterable and implacable enemies of the German people, concocting biological rationalizations for anti-Semitism (i.e, that Jews had a different ‘‘racial essence’’). Still, the envious nature of this prejudice remained intact. The Nazis believed that a powerful international Jewish conspiracy of industrialists and financiers (including German Jews) had caused Germany to lose World War I and brought about the collapse of its economy in order to profit from Germany’s misery. Their propaganda consistently exaggerated the power of Jews to shape world events and complained about the Jews’ over-representation in high-status professions
128
Amy J. C. Cuddy et al.
in Germany, such as in the media, academia, medicine, and law, which would otherwise have been occupied by Aryans. According to archival analyses, the Fascists characterized the Jews in similar ways, also fitting the SCM’s competent but cold cluster (Volpato et al., 2007). When an envied (i.e., competent but cold) group is viewed as an intractable enemy that is ruining society, active harm (e.g., violene) will likely result. This is particularly likely when envy turns into anger, an emotion that triggers offensive actions toward others (e.g., Mackie et al., 2000), as discussed above (section 4.2.2.2.1). In particular, Glick (2002, 2005) has suggested that when a society experiences difficult life conditions (Staub, 1996), groups perceived as competent competitors (i.e., envied groups) are most likely to be scapegoated. This occurs because shared social miseries impel people to construct explanations for, as well as possible solutions to, these problems. At base, this is an adaptive tendency—successful societies must band together to diagnose and solve shared problems. However, widespread social problems usually have complex origins, so even experts cannot agree (e.g., not all economists are likely to agree on the causes of a collapsed economy). Blaming disliked human groups for social problems can be psychologically compelling because it locates the source of the problem and suggests a solution: elimination of the enemy group. Although this alone is not a new insight, classic theories of scapegoating (see Allport, 1954) failed to understand which groups are at greatest risk of being blamed and why. An ideological model of scapegoating (Glick, 2002, 2005) posits that scapegoating is not a mere venting of frustration, but rather mediated by shared attributions that gain traction because they offer a culturally plausible explanation for societal problems. But not just any group is perceived as capable of causing such problems, which require not just bad intentions but competence and power. Thus, although the Nazis generally wanted to eliminate other ‘‘inferior races,’’ they were particularly obsessed with the Jews, whom they perceived as especially powerful both inside and outside Germany. The Nazi euthanasia program aimed at eliminating the physically and mentally ill sparked public outrage in Germany, causing the Nazis to halt it (Friedlander, 1995). Although the Nazis saw the ill as parasitical, the general public could not stomach the idea of killing low-status innocents (i.e., people viewed as warm, though incompetent). By contrast, the idea that the Jews—a powerful minority that had, for centuries, been seen first as demonic and later as economically exploitative—were enemies of Germany who had to be eliminated was accepted by many and protested by few. Contrary to the older idea that any weak and vulnerable group is likely to be scapegoated, Glick’s ideological theory proposes that relatively successful minorities, who have long been stereotyped as powerful, as well as ill-intentioned, are at the greatest risk for being blamed. Other historical cases of genocidal attacks seem generally to fit this mold (Glick, 2002). In Turkey, the Armenians had a very similar profile to Jews in
Warmth and Competence as Universal Dimensions of Social Perception
129
Germany in terms of the positions they occupied. During the collapse of the Ottoman Empire, they too were characterized as a powerful enemy within the society and consequently slaughtered. In Cambodia, Pol Pot and his henchmen targeted intellectuals and professionals (i.e., people of high status) to be slain in the killing fields. In Rwanda in the 1990s, the Tutsi, also a high-status minority, were blamed for their nation’s economic and social problems. Anti-Tutsi propaganda during this time bore a striking resemblance to ‘‘Jewish conspiracy’’ tracts, such as the Protocols of the Elders of Zion. Indeed, conspiracy theories about the outgroups’ nefarious machinations against the ingroup may be the hallmark of envious prejudice. Such beliefs invariably ascribe heightened power and influence (i.e., competence) to the targeted outgroup. In short, the BIAS map helps to explain why envied groups are often tolerated but later attacked, particularly under conditions that convert envy into anger. Because they are viewed as skilled, they are also seen as useful during more stable times, but they are extremely vulnerable to blame under social and economic instability. Thus, although pre-Nazi Germany was generally considered less anti-Semitic than many other European nations, the underlying stereotype of Jews as competent and conspiratorial competitors provided ripe conditions for later scapegoating. Ironically, the prior climate in Germany, in which Jews experienced relative success, may have provided greater perceived credibility among Germans for Nazis’ claims that Jews were positioned to ‘‘stab Germany in the back.’’ The dynamics of envious prejudice demand further study because this type of prejudice may help to explain the most extreme forms of intergroup hostility, genocidal attack. 5.2.3. Summary Two envied outgroups present similar profiles: viewed as excessively competent, they allegedly control resources that the mainstream wants. Both Asian Americans and German Jews, as well as other high-status entrepreneurial outsiders, are viewed with a mixture of envy for their accomplishments and status, along with anger for their allegedly not sharing cooperatively with the ingroup and its reference groups. Such groups elicit a ‘‘volatile ambivalence’’ (Harris et al., in press) that obliges association and convenient cooperation when society is stable but sets the stage for attack and even mass killing under extreme social breakdown.
6. Current and Future Directions and Summary This section reviews new directions for research suggested by the SCM and BIAS map, which would appear to have a variety of implications for both intergroup and interpersonal perception and interaction, that have yet to be plumbed, then provides a brief chapter summary.
130
Amy J. C. Cuddy et al.
6.1. Current and future directions 6.1.1. Intergroup attributions Earlier, we examined attributional antecedents, when we applied attributions about people’s responsibility for their social positions (e.g., whether their low status or stigma was ‘‘chosen’’), to understanding the emotions and behaviors groups tend to elicit. In this section, we consider the attributional consequences of membership in groups that occupy different parts of the SCM space. How does the actor’s group membership affect the causal inferences perceivers make about the actor’s behavior? More specifically, when are perceivers likely to view a person’s actions as reflecting underlying dispositions versus excuse away their behavior as, for example, being due to the situation or luck? Prior work has generally assumed that attributions about behavior are less favorable toward outgroup members than ingroup members. For instance, women’s (but not men’s) successes are attributed to luck but their failures are attributed to dispositions (Swim and Sanna, 1996). Ethnic minorities’ (but not ingroups’) prosocial behavior is viewed as disingenuous but their antisocial behavior as dispositional (Hewstone and Ward, 1985). Pettigrew (1979) was confident enough about the generality of this pattern to refer to it as the ‘‘ultimate attribution error’’ (UAE). The UAE assumes a motivational bias, essentially an extension of the self-serving attribution bias (Arkin et al., 1980) to intergroup attribution, such that positive behavior by or outcomes for ‘‘us’’ are viewed as reflecting dispositions and skills, whereas negative behaviors by or outcomes for ‘‘us’’ are excused away. Conversely, the UAE assumes a negative attributional pattern toward outgroups such that negative behavior by or outcomes for ‘‘them’’ are viewed as reflecting harmful dispositions or a persistent lack of skills, whereas positive behaviors by or outcomes for ‘‘them’’ are excused away. On the basis of the SCM, Glick et al. (2007) have theorized that attributional biases toward members of different groups should not uniformly follow the predictions of the UAE, but rather a more complex pattern governed by the match between the actor’s behavior and stereotypes about that actor’s group, labeled here as the Stereotype-Confirming Attribution Bias (SCAB). In other words, rather than a simple motivational bias (ingroup is wholly ‘‘good,’’ and outgroup wholly ‘‘bad’’), the specific content of group stereotypes is likely to create confirmatory cognitive biases that guide attributions about group members’ behaviors. Glick et al. argue that, regardless of their valence (i.e., whether positive or negative), when behaviors or outcomes match stereotypical expectations, perceivers are likely to make dispositional attributions. By contrast, behaviors and outcomes that are stereotype-inconsistent should be excused away (e.g., as a fluke, situationally caused, etc.). In some cases, the SCAB makes similar predictions to the UAE. For instance, a paternalized group member’s success is likely to be attributed to
Warmth and Competence as Universal Dimensions of Social Perception
131
luck or an envied group member’s prosocial behavior to ulterior motives (not prosocial intentions). But consider what the UAE would predict for attributions about a paternalized group member’s prosocial behavior or an envied group member’s achievement behavior. If, for example, the elderly are an ‘‘outgroup,’’ then the UAE would predict that neither an elderly person’s prosocial actions (e.g., hugging someone) nor achievements (e.g., acing a difficult math test) would be seen as dispositional. The UAE would make the same exact predictions for other outgroups, such that members of envied groups (e.g., Asians and Jews) would elicit the same negative attributional pattern for those behaviors. By contrast, the SCAB predicts that the elderly person’s stereotypically consistent prosocial behavior would be seen as dispositional, but his or her stereotype inconsistent achievements would not be. In other words, an elderly person’s hug would be credited to a warm disposition, but the same person’s stellar test performance would be seen as a fluke (not a reflection of underlying competence). The SCAB predicts the opposite for envied groups—an Asian person’s prosocial behavior might be viewed as situationally caused or as manipulative (rather than reflecting a warm disposition), but perceivers are likely to assume that an Asian who aced the math test did so because he or she is smart (i.e., dispositionally competent). In short, the SCAB suggests that the UAE’s predictions will not hold for ambivalently stereotyped groups, but rather that a uniformly positive attribution pattern occurs only for groups that are positively stereotyped on both competence and warmth (admired groups, which are often ingroups) and that a uniformly negative attribution pattern holds only for groups subjected to contemptuous stereotypes that cast them as both incompetent and cold. The SCAB also points out the importance of distinguishing between behavioral domains related to each stereotype dimension. Warmth stereotypes should drive attributions about social behavior, whereas competence stereotypes should drive attributions about achievement behaviors. To test the SCAB’s predictions, Glick et al. (2007) have thus far conducted two experiments. In the first, participants viewed pretested SCMspace pictures of individuals that were each paired with a behavior the individual allegedly had performed (e.g., an elderly person who ‘‘ignored a friend who said hello’’). For each person–behavior pair presented on the computer screen, participants indicated the degree to which they thought behavior reflected a ‘‘trait’’ (defined as ‘‘a stable and long-lasting personal characteristic’’). The 64 pictures portrayed 8 individuals from each of 8 groups. The groups were selected based on prior research to represent the four SCM competence warmth quadrants, with two groups from each quadrant (Americans and students for competent and warm; rich and professionals for competent but cold; elderly and disabled for warm but incompetent;
132
Amy J. C. Cuddy et al.
homeless and drug addicts for incompetent and cold). Because the participants were American college students, the admired groups were also ingroups. Sixty-four behaviors were randomly paired with each picture. The behaviors were either from the social or achievement domain and either positive (prosocial or a success) or negative (antisocial or a failure). Both the pictures and the behaviors were extensively pretested to ensure that they fit the intended categories. The behaviors participants viewed as dispositional indeed depended on the actor’s group membership. For univalent groups, where the UAE and SCAB make similar predictions, results supported both theories. The behaviors of admired group members (whether Americans or students) were seen as more dispositional if positive and less dispositional if negative, across both the social and achievement domains. Conversely, for groups subjected to contempt (whether homeless or drug addicts), behaviors were seen as more dispositional if negative and less dispositional if positive, across both the social and achievement domains. However, for ambivalently stereotyped groups, those for which the theories make incompatible predictions, results supported the SCAB and revealed the shortcomings of the UAE. For envied group members (whether rich or professionals), achievement-related behaviors were viewed as more dispositional if successful (e.g., ‘‘solved an important puzzle’’) and less dispositional if unsuccessful (e.g., ‘‘did a poor job on a work project’’). Indeed, this ‘‘positive’’ attributional pattern in the achievement domain was stronger for envied groups than for admired ingroups (Americans, students), a finding that directly contradicts the UAE. When it came to the social domain, however, envied groups were subjected to a negative attributional bias: relative to the admired groups, their prosocial behaviors were viewed as less dispositional, and their antisocial behaviors were viewed as more dispositional. Also consistent with the SCAB, and problematic for the UAE, was a positive attributional bias toward paternalized groups (whether elderly or disabled) for social behaviors. Participants viewed paternalized group members’ prosocial behaviors as more dispositional, and antisocial behaviors as less dispositional. Further, this bias was again stronger than the bias applied to the admired ingroup members. On the other hand, paternalized group members’ successful achievements were dismissed as less dispositional and their failures as more dispositional, relative to the admired ingroups. In sum, the first study supported the SCAB over the UAE. It also revealed evidence of compensatory stereotyping (Kay and Jost, 2003) of groups that receive ambivalent prejudices. The attributions implied that, relative to admired ingroups, perceivers view envied group members as hypercompetent (even if cold) and paternalized group members as especially warm (even if incompetent). Consistent with work by Harris and Fiske (2006), only groups in the contempt cell of the SCM were treated in a
Warmth and Competence as Universal Dimensions of Social Perception
133
consistently and extremely negative way. These were the only groups that received an absolute negative attributional pattern, not only in the sense that attributional biases disfavored them on both social and achievement dimensions, but also in that negative behaviors by these groups were viewed as significantly more dispositional than positive behaviors. Even though envied and paternalized groups were each attributionally disfavored on one dimension relative to admired ingroups, in neither case was the bias so stark that negative behaviors by these groups were viewed as more dispositional than positive behaviors. A second study used a different method that corrected for potential weaknesses in the first study’s use of pictures of group members. Because it is difficult to find smiling pictures of homeless people and drug addicts, and effects in the first study might have been influenced by facial expressions or other confounds in the pictures. In the second study, participants read a set of four behaviors said to have been performed by an individual. Each set included two social and two achievement behaviors. Within each domain, valence of the behaviors was the same. Between trials, however, all combinations of positive and negative behaviors in the two domains were represented. The behaviors were those used in the prior study. Following each set of four behaviors participants rated ‘‘How likely is it that these behaviors were done by . . .’’ with the remainder of the sentence filled in randomly with the name of one of the groups used in the first study (e.g., ‘‘. . . a homeless person’’). This method therefore used group labels, not pictures. Again, except for the cases where predictions were similar, results supported the SCAB and not the UAE. Participants viewed the behavioral set that paired positive achievement with positive social behaviors as most likely performed by admired ingroup members while they viewed the combination of negative achievement and negative social behaviors as least likely performed by members of admired groups. The opposite was true for groups subject to contempt. Again, it is for ambivalently stereotyped groups that predictions differ for the UAE and SCAB. Consistent with the SCAB, participants rated the set of acts combining positive achievement with negative social behaviors as most likely done by envied groups, and the set combining negative achievement with positive social behaviors as least likely done by envied groups. The opposite happened for paternalized groups, for whom the set combining negative achievement with positive social behaviors was viewed as most likely and the set combining positive achievement with negative social behaviors was rated least likely. In sum, past intergroup attribution research has been blinded by the assumption of straightforward antipathy toward outgroups. The SCM suggests a new way of thinking about how stereotype-behavior consistency (versus inconsistency) drives attributions about individual group members’
134
Amy J. C. Cuddy et al.
behaviors. The SCAB also suggests the need to distinguish between social and achievement behaviors (corresponding, respectively, to warmth and competence stereotypes) to understand the complexities of intergroup biases in attributions. These ideas actually hark back to the earliest days of attribution theory with Heider’s (1958) notions that perceivers seek to discern an actor’s intentions and abilities to determine whether to make dispositional attributions. A Heiderian analysis dovetails nicely with the SCM’s contention that warmth and competence are fundamental dimensions precisely because they address others’ perceived intentions (Heider’s ‘‘try’’) and capabilities (Heider’s ‘‘can’’). Of course, behaviors can combine both social and achievement elements, such as organizing a complex charity drive (on the positive side), or pulling off a complex fraud to steal others’ money (on the negative side). In these cases, perceived warmth and competence should interact to affect attributions. Heider (1958) also anticipated such situations when he noted how ‘‘task difficulty’’ affects attributions. A behavior that has social implications (i.e., helps or harms others) but is difficult to pull off (i.e., can also be seen as involving achievement) should be viewed as beyond the capabilities of incompetent groups, even if they might have the intention to accomplish it. Continuing this line of research, we plan to explore how these biases affect group-level attributions or what Tajfel (1981) called ‘‘social attribution.’’ This refers to attributions about actions of the group as a whole, not just how group categories affect how individual members are perceived. The difference can be illustrated by contrasting such inferences as ‘‘Of course she won the case, she’s a smart Jew’’ (individual) versus ‘‘Of course they caused the economic downturn, that’s what Jews are capable of and interested in doing’’ (social). The latter type of attribution is central to Glick’s (2002, 2005) ideological model of scapegoating (discussed earlier). Only envied groups should be viewed as likely to have the desire and ability to cause large-scale events that harm others (while putatively enriching themselves). 6.1.2. Neural signatures for SCM space? Another future direction also operates at the individual level, but emphasizes even more internal processes within the perceiver. Because we argue that the SCM represents fundamental dimensions of social perceptionnamely, the other’s intent and capacity to enact it—the social brain should be implicated in these canonical forms of perception. As noted earlier, people judge warmth and trustworthiness in a fraction of a second (Hack et al., 2007; Willis and Todorov, 2006; Ybarra et al., 2001). The amygdala, a brain region implicated in vigilance, correlates with judged trustworthiness in face perception (Todorov et al., 2007). People with damaged amygdale
Warmth and Competence as Universal Dimensions of Social Perception
135
cannot reliably judge trustworthiness (Adolphs et al., 1998). Vigilance puts the perceiver on alert, guarding against interpersonal threat. People also judge the competence of faces fairly quickly (with measurable consequences, e.g, for voting, Todorov et al., 2005). Because of the strong link between perceived status and competence, one would predict that high-status, high-competence untrustworthy competitors (i.e., envied groups) would also activate the amygdala’s vigilance system, and they do (Harris et al., in press). Such findings related to two underlying dimensions of the SCM suggest some potential neural signatures for distinct quadrants. So far, Harris, Fiske, and colleagues have studied two quadrants, the envied high-competence/ low-warmth outgroups, and the disgusting, low-low outgroups. In a typical neuroimaging test of the SCM, 60 pretested photographs from each quadrant of the space appear on a screen in the scanner; participants view the photos and either report which SCM emotion they elicit, or merely passively view the photos (it makes no difference to the results). The study of envied outgroups (pictured as rich people and professionals) compared them to groups in the other three quadrants (elderly or disabled, homeless or drug addicts, Americans or students). Envied groups especially activated the amygdala, as noted, suggesting special vigilance. In addition, they activated the mPFC, a brain area reliably associated with dispositional inferences and thinking about other people’s minds (‘‘mentalizing’’). This fits the idea that people watch high-status, high-power others because they control resources, and that people make dispositional inferences about them, in an effort to predict their behavior (Fiske, 1993). The mPFC has a crucial role in social cognition (Amodio and Frith, 2006; Mitchell et al., 2005). For example, it activates uniquely to the exact pattern of consensus, consistency, and distinctiveness information about behavior that elicits a dispositional inference (Harris et al., 2005). Objects viewed behaving according to the same patterns do not activate the mPFC (Harris and Fiske, in press). The idea that the brain has a special sensitivity to social information should gladden the hearts of social psychologists. But the news is not all good. The low-low outgroups, the most negative on the SCM and alone among all the other quadrants, fail to activate the mPFC significantly above baseline (Harris and Fiske, 2006). Instead, homeless people and drug addicts activate the insula, an area implicated in disgust and arousal. These neural patterns fit the dehumanization of extreme outgroups in the SCM space. Indeed, in the SCM surveys, homeless people were rated so negatively on both warmth and competence that often they are three standard deviations out from the mean of the rest of the groups and had to be removed from the cluster analysis so that they did not distort it. This extreme form of prejudice is perhaps reflected in neuroimaging data. Of
136
Amy J. C. Cuddy et al.
course, representation in the brain does not mean such prejudice is hardwired and inevitable; social context affects neural responses, naturally. When perceivers imagine what to feed the homeless and drug-addicted, the mPFC comes back on line (Harris and Fiske, 2007). Ongoing research elaborates on responses to envied groups and examines reactions to pitied groups, to understand the ambivalent mixtures entailed. Although these neuroimaging results do not by themselves establish unique emotional reactions to distinct clusters of outgroups, they do, along with the survey and experimental results already reported, provide a form of converging evidence less open to social desirability biases. 6.1.3. Warmth and competence in self-perception If dimensions of warmth and competence are fundamental in judging other people and other groups, perhaps they are fundamental to judging even oneself. According to Abele and Wojciszke (2007), the main distinction between these dimensions concerns profitability for self (agency) or for others (communion). By this logic, agency is more desirable for self, and communion is more desirable for others. Their research supports this and further that outcome dependency increases importance of another person’s agency. Moreover, people can be differentiated according the degree to which their sense of self-worth (i.e., global self-esteem) derives from appraisals of the self as competent or as warm (Mandisodza et al., 2005). Self-esteem for some individuals is more highly related to appraisals of competence, whereas for others it is more highly related to appraisals of warmth. Given this, Mandisodza et al. suggested that some generality should exist between self-appraisals and appraisals of other individuals and groups. For instance the more important competence differences are in the appraisal of self, the more important they should be in appraisals of other targets, be they groups or individuals. And reciprocally, the more important warmth is in the appraisal of self, the more important it should be in appraisals of other targets. To explore this hypothesis, they asked subjects to complete a self-esteem scale, which yielded the two dimensions of self-evaluation: competence appraisals and warmth appraisals (Tafarodi and Ho, 2006). From this, they computed a difference for each participant (competence self-evaluation minus warmth self-evaluation). Additionally, they asked participants to judge eight target groups and four target individuals on dimensions of competence, warmth, and overall evaluation. They then collapsed across the targets and showed that global evaluations of targets are more highly related to appraisals of their competence for people with higher selfappraisals on competence than on warmth, whereas for those whose self-appraisal is higher on warmth than competence, warmth judgments of targets are more highly related to overall evaluations of those targets than are competence judgments. This research continues.
Warmth and Competence as Universal Dimensions of Social Perception
137
6.2. Summary Synthesizing two models of intergroup bias, the SCM and the BIAS map, with research in interpersonal perception, we have argued that considerable empirical evidence identifies warmth and competence as universal dimensions of social judgment, across perceivers, stimuli, cultures, and time. Evaluations of both individuals and groups on these dimensions follow from social structural relationships. Interdependence predicts perceived warmth, and status predicts perceived competence. Each combination of high and low evaluations on these dimensions has distinct emotional and behavioral consequences. This fact especially matters to group-based prejudices. Group perceived as warm and competent elicit uniformly positive emotions and behavior: admiration, help, and association. Those perceived as lacking both warmth and competence elicit uniform negativity: contempt, neglect, and attack. But most group stereotypes appear high on one dimension and low on the other: the ensuing ambivalent affect and volatile behavior endanger constructive intergroup relations. High warmth wth low competence yields pity and patronizing help or neglect. Low warmth with high competence evokes envy and strategic association or, under threat, attack. Future work focuses on the causes and consequences of warmth and competence judgments in attribution, neuroscience, and self-perception.
ACKNOWLEDGMENTS Preparation of this chapter was facilitated by research funds from the Kellogg School of Management at Northwestern University and from Princeton University. The authors wish to thank their joint and respective research collaborators for stimulating intellectual support.
REFERENCES Abele, A. E. (2003). The dynamics of masculine-agentic and feminine-communal traits: Findings from a prospective study. J. Pers. Soc. Psychol. 85, 768–776. Abele, A. E., and Wojciszke, B. (2007). Agency and communion from the perspective of self versus others. J. Pers. Soc. Psychol. 9, 751–763. Abele, A. E., Cuddy, A. J. C., Yzerbyt, V. Y., and Judd, C. M. (eds.) (in press). Universal dimensions of social perception: Communion and agency. Special issue of European Journal of Social Psychology. Abelson, R. P., Kinder, D. R., Peters, M. D., and Fiske, S. T. (1982). Affective and semantic components in political person perception. J. Pers. Soc. Psychol. 42, 619–630. Adolphs, R., Tranel, D., and Damasio, A. R. (1998). The human amygdala in social judgment. Nature 393, 470–474. Ajzen, I. (2001). Nature and operation of attitudes. Annu. Rev. Psychol. 52, 27–58. Alexander, M. G., Brewer, M. B., and Hermann, R. K. (1999). Images and affect: A functional analysis of outgroup stereotypes. J. Pers. Soc. Psychol. 77, 78–93.
138
Amy J. C. Cuddy et al.
Alexander, M. G., Brewer, M. B., and Livingston, R. W. (2005). Putting stereotype content in context: Image theory and interethnic stereotypes. Pers. Soc. Psychol. Bull. 31, 781–794. Allport, G. W. (1954). ‘‘The Nature of Prejudice’’. Addison-Wesley, Reading, MA. Altermatt, T. W., DeWall, C. N., and Leskinen, E. (2003). Agency and virtue: Dimensions of female stereotypes. Sex Roles 49, 631–641. Amodio, D. M., and Frith, C. D. (2006). Meeting of minds: The medial frontal cortex and social cognition. Nat. Rev. Neurosci. 7, 268–277. Andreoletti, C., Maurice, J. K., and Whalen, H. (2001). Gender, race, and age: Compound stereotypes across the lifespan. Poster presented at the Second Annual Meeting of the Society for Personality and Social Psychology, February, 2001, San Antonio, TX. Arkin, R. M., Cooper, H. M., and Kolditz, T. A. (1980). A statistical review of the literature concerning the self-serving attribution bias in interpersonal influence situations. J. Pers. 48, 435–448. Asbrock, F., and Cuddy, A. J. C. (2008). Interpersonal behavioral outcomes of warmth and competence judgements. Manuscript under review. Asch, S. E. (1946). Forming impressions of personality. J. Abnorm. Soc. Psychol. 41, 258–290. Avolio, B. J., and Barrett, G. J. (1987). Effects of age stereotyping in a simulated interview. Psychol. Aging 2, 56–63. Ayduk, O., May, D., Downey, G., and Higgins, E. T. (2003). Tactical differences in coping with rejection sensitivity: The role of prevention pride. Pers. Soc. Psychol. Bull. 29, 435–448. Bakan, D. (1956). ‘‘The Duality of Human Existence: An Essay on Psychology and Religion.’’ Rand McNally, Oxford, England. Bales, R. F. (1950). A set of categories for the analysis of small group interaction. Am. Sociol. Rev. 15, 257–263. Bales, R. F. ed. (1999). ‘‘Social Interaction Systems: Theory and Measurement.’’ Transaction Press, New Brunswick, NJ. Barnett, M. A., Quackenbush, S. W., and Pierce, S. K. (1997). Perception of and reactions to the homeless: A survey of fourth-grade, high school, and college students in a small Midwestern community. J. Soc. Distress Homel. 6, 238–302. Batson, C. D. (1998). Altrusim and prosocial behavior. In ‘‘Handbook of Social Psychology’’ (D. T. Gilbert, S. T. Fiske, and G. Lindzey, eds.) 4th ed.,Vol. 2, pp. 282–316. McGrawHill, Boston. Bond, M. H. (1994). Trait theory and cross cultural studies of person perception. Psychol. Inq. 2, 114–168. Bradac, J. J. (1990). Language attitudes and impression formation. In ‘‘Handbook of Language and Social Psychology’’ (H. Giles and W. P. Robinson, eds.), pp. 387–412. Wiley, New York. Brewer, M. B., and Alexander, M. G. (2002). Intergroup emotions and images. In ‘‘From Prejudice to Intergroup Emotions: Differentiated Reactions to Social Groups’’ (D. M. Mackie and E. R. Smith, eds.), pp. 209–225. Psychology Press, New York. Buss, A. H. (1961). ‘‘The Psychology of Aggression.’’ Wiley, New York. Buss, D. M. (1981). Sex differences in the evaluation and performance of dominant acts. J. Pers. Soc. Psychol. 40, 147–154. Cacioppo, J. T., Gardner, W. L., and Berntson, G. G. (1997). Beyond bipolar conceptualizations and measures: The case of attitudes and evaluative space. Pers. Soc. Psychol. Rev. 1, 3–25. Cadinu, M. R., and Cerchioni, M. (2001). Compensatory biases after intergroup threat: ‘Yeah, but we have a good personality.’ Eur. J. Soc. Psychol. 31, 353–367. Capozza, D., Trifiletti, E., Pasin, A., and Durante, F. (2007). EPA and warmth-competence dimensions: Are the two universals redundant? Unpublished manuscript. Universita di Padova.
Warmth and Competence as Universal Dimensions of Social Perception
139
Caprariello, P., Cuddy, A. J. C., and Fiske, S. T. (2007). Social structure shapes cultural stereotypes and emotions: A causal test of the stereotype content model. Manuscript submitted for publication. Carlson, R. (1971). Sex differences in ego functioning: Exploratory studies of agency and communion. J. Consult. Clin. Psychol. 37, 267–277. Casciaro, T., and Sousa-Lobo, M. (2005). Competent jerks, lovable fools, and the formation of social networks. Harv. Bus. Rev. 83(6), 92–99. Chemers, M. M. (1997). ‘‘An Integrative Theory of Leadership.’’ Lawrence Erlbaum Associates, Mahwah, NJ. Chemers, M. M. (2001). Leadership effectiveness: An integrative review. In ‘‘Blackwell Handbook of Social Psychology: Group Processes’’ (M. A. Hogg and R. S. Tindale, eds.), pp. 376–399. Blackwell, Oxford. Cialdini, R. B., Borden, R. J., Thorne, R. J., Walker, M. R., Freeman, S., and Sloan, L. R. (1976). Basking in reflected glory: Three (football) field studies. J. Pers. Soc. Psychol. 34, 366–375. Clausell, E., and Fiske, S. T. (2005). When do the parts add up to the whole? Ambivalent stereotype content for gay male subgroups. Soc. Cogn. 23, 157–176. Cottrell, C. A., and Neuberg, S. L. (2005). Different emotional reactions to different groups: A sociofunctional threat-based approach to ‘‘prejudice.’’ J. Pers. Soc. Psychol. 88, 770–789. Crick, N. R., and Grotpeter, J. K. (1995). Relational aggression, gender, and socialpsychological adjustment. Child Dev. 66, 710–722. Cuddy, A. J. C., and Fiske, S. T. (2002). Doddering, but dear: Process, content, and function in stereotyping of older persons. In ‘‘Ageism’’ (T. D. Nelson, ed.), pp. 3–26. MIT Press, Cambridge, MA. Cuddy, A. J. C., and Frantz, C. M. (2007). Legitimating status inequalities: The effect of race on motherhood discrimination Manuscript submitted for publication. Cuddy, A. J. C., Fiske, S. T., and Glick, P. (2004). When professionals become mothers, warmth doesn’t cut the ice. J. Soc. Issues 60, 701–718. Cuddy, A. J. C., Norton, M. I., and Fiske, S. T. (2005). This old stereotype: The pervasiveness and persistence of the elderly stereotype. J. Soc. Issues 61, 265–283. Cuddy, A. J. C., Fiske, S. T., and Glick, P. (2007). The BIAS Map: Behaviors from intergroup affect and stereotypes. J. Pers. Soc. Psychol. 92, 631–648. Cuddy, A. J. C., Fiske, S. T., Kwan, V. S. Y., Glick, P., Demoulin, S., Leyens, J.-Ph., Bond, M. H., Croizet, J.-C., Ellemers, N., Sleebos, E., Htun, T. T., Yamamoto, M., et al. (in press). Is the stereotype content model culture-bound? A cross-cultural comparison reveals systematic similarities and differences. Br. J. Soc. Psychol. Devos, T., Silver, L. A., Mackie, D. M., and Smith, E. R. (2002). Experiencing intergroup emotions. In ‘‘From Prejudice to Intergroup Emotions: Differentiated Reactions to Social Groups’’ (D. M. Mackie and E. R. Smith, eds.), pp. 111–134. Psychology Press, New York. Diekman, A. B., and Eagly, A. H. (2000). Stereotypes as dynamic constructs: Women and men of the past, present, and future. Pers. Soc. Psychol. Bull. 26, 1171–1188. Dijker, A. J. (1987). Emotional reactions to ethnic minorities. Eur. J. Soc. Psychol. 17, 305–325. Dijker, A. J., Kok, G., and Koomen, W. (1996a). Emotional Reactions to People with AIDS. J. Appl. Soc. Psychol. 26, 731–738. Dijker, A. J., Koomen, W., van den Heuvel, H., and Frijda, N. H. (1996b). Perceived ntecedents of emotional reactions in inter-ethnic relations. Br. J. Soc. Psychol. 35, 313–329. Dovidio, J. F., Brigham, J. C., Johnson, B. T., and Gaertner, S. L. (1996). Stereotyping, prejudice, and discrimination: Another look. In ‘‘Stereotypes and Stereotyping’’ (C. N. Macrae, C. Stangor, and M. Hewstone, eds.) pp. 276–319. Guilford, New York.
140
Amy J. C. Cuddy et al.
Dovidio, J. F., Esses, V. M., Beach, K. R., and Gaertner, S. L. (2002). The role of affect in determining intergroup behavior: The case of willingness to engage in intergroup contact. In ‘‘From Prejudice to Intergroup Emotions: Differentiated Reactions to Social Groups’’ (D. M. Mackie and E. R. Smith, eds.), pp. 153–171. Psychology Press, New York. Durante, F., and Capozza, D. (2008). An Italian test of the Stereotype Content Model. Manuscript submitted for publication. Eagly, A. H. (1987). ‘‘Sex Differences in Social Behavior: A Social-Role Interpretation.’’ Erlbaum, Hillsdale, NJ. Eagly, A. H., and Mladinic, A. (1989). Gender stereotypes and attitudes toward women and men. Pers. Soc. Psychol. Bull. 15, 543–558. Eagly, A. H., and Mladinic, A. (1994). Are people prejudiced against women? Some answers from research on attitudes, gender stereotypes, and judgments of competence. In ‘‘European Review of Social Psychology’’ (W. Stroebe and M. Hewstone, eds.), Vol. 5, pp. 1–35. Wiley, New York. Eagly, A. H., and Steffen, V. J. (1984). Gender stereotypes stem from the distribution of women and men into social roles. J. Pers. Soc. Psychol. 46, 735–754. Eagly, A. H., Mladinic, A., and Otto, S. (1994). Cognitive and affective bases of attitudes toward social groups and social policies. J. Exp. Soc. Psychol. 30, 113–137. Eagly, A. H., Wood, W., and Diekman, A. H. (2000). Social role theory of sex differences and similarities: A current appraisal. In ‘‘The Developmental Social Psychology of Gender’’ (T. Eckes and H. M. Trautner, eds.), pp. 123–174. Erlbaum, Mahwah, NJ. Eagly, A. H., Johannesen-Schmidt, M. C., and van Engen, M. L. (2003). Transformational, transactional, and laissez-faire leadership styles: A meta-analyses comparing women and men. Psychol. Bull. 129, 569–591. Eckes, T. (2002). Paternalistic and envious gender stereotypes: Testing predictions from the stereotype content model. Sex Roles 47, 99–114. Ellemers, N. (2001). Social identity, commitment and work behavior. In ‘‘Social Identity Processes in Organizational Contexts’’ (M. A. Hogg and D. J. Terry, eds.). Psychology Press, Hove, Sussex. Ellemers, N., Van Rijswijk, W., Roefs, M., and Simons, C. (1997). Bias in intergroup perceptions: Balancing group identity with social reality. Pers. Soc. Psychol. Bull. 23, 186–198. Erber, J. T., and Prager, I. C. (1999). Age and memory: Perceptions of forgetful young and older adults. In ‘‘Social Cognition and Aging’’ (T. M. Hess and F. Blanchard-Fields, eds.), pp. 198–221. Academic Press, San Diego, CA. Erber, J. T., Etheart, M. E., and Szuchman, L. T. (1992). Age and forgetfulness: Perceivers’ impressions of targets’ capability. Psychol. Aging 7, 479–483. Erber, J. T., Rothberg, S. T., Szuchman, L. T., and Etheart, M. E. (1993). How physicians appraise everyday memory failures of patients across the adult life span. Exp. Aging Res. 19, 195–207. Erber, J. T., Prager, I. G., Williams, M., and Caiola, M. A. (1996). Age and forgetfulness: Confidence in ability and attribution for memory failures. Psychol. Aging 11, 310–315. Esses, V. M., and Dovidio, J. F. (2002). The role of emotions in determining willingness to engage in intergroup contact. Pers. Soc. Psychol. Bull. 28, 1202–1214. Esses, V. M., Haddock, G., and Zanna, M. P. (1993). Values, stereotypes, and emotions as determinants of intergroup attitudes. In ‘‘Affect, Cognition, and Stereotyping: Interactive Processes in Group Perception’’ (D. M. Mackie and D. L. Hamilton, eds.), pp. 137–166. Academic Press, San Diego, CA. Esses, V. M., Dovidio, J. F., Jackson, L. M., and Armstrong, T. L. (2001). The immigration dilemma: The role of perceived group competition, ethnic prejudice, and national identity. J. Soc. Issues 57, 389–412.
Warmth and Competence as Universal Dimensions of Social Perception
141
Fiske, S. T. (1980). Attention and weight in person perception: The impact of negative and extreme behavior. J. Pers. Soc. Psychol. 38, 889–906. Fiske, S. T. (1992). Thinking is for doing: Portraits of social cognition from Daguerreotype to laserphoto. J. Pers. Soc. Psychol. 63, 877–889. Fiske, S. T. (1993). Controlling other people: The impact of power on stereotyping. Am. Psychol. 48, 621–628. Fiske, S. T., and Cuddy, A. J. C. (2006). Stereotype content across cultures as a function of group status. In ‘‘Social Comparison Processes and Levels of Analysis’’ (S. Guimond, ed.), Cambridge University Press, New York. Fiske, S. T., and Ruscher, J. B. (1993). Negative interdependence and prejudice: Whence the affect?. In ‘‘Affect, Cognition, and Stereotyping: Interactive Processes in Group Perception’’ (D. M. Mackie and D. L. Hamilton, eds.), pp. 239–268. Academic Press, London. Fiske, S. T., Xu, J., Cuddy, A. C., and Glick, P. (1999). (Dis)respecting versus (dis)liking: Status and interdependence predict ambivalent stereotypes of competence and warmth. J. Soc. Issues 55, 473–491. Fiske, S. T., Cuddy, A. J., and Glick, P. (2002a). Emotions up and down: Intergroup emotions result from perceived status and competition. In ‘‘From Prejudice to Intergroup Emotions: Differentiated Reactions to Social Groups’’ (D. M. Mackie and E. R. Smith, eds.), pp. 247–264. Psychology Press, Philadelphia, PA. Fiske, S. T., Cuddy, A. J. C., Glick, P., and Xu, J. (2002b). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. J. Pers. Soc. Psychol. 82, 878–902. Fiske, S. T., Thomas, M., and Vescio, T. K. (2007). Universal dimensions of social cognition: warmth and competence. Trends Cogn. Sci. 11, 77–83. Friedlander, H. (1995). ‘‘The Origins of Nazi Genocide: From Euthanasia to the Final Solution.’’ University of North Carolina Press, Chapel Hill, NC. Frijda, N. H. (1986). ‘‘The Emotions.’’ Cambridge University Press, New York. Frijda, N. H., Kuipers, P., and ter Schure, E. (1989). Relations among emotion, appraisal, and emotional action readiness. J. Pers. Soc. Psychol. 57, 212–228. Geen, R. (1998). Aggression and antisocial behavior. In ‘‘Handbook of Social Psychology’’ (D. Gilbert, S. Fiske, and G. Lindzey, eds.) 4th ed., Vol. 2, pp. 317–356. McGraw-Hill, Boston. Gilbert, G. M. (1951). Stereotype persistence and change among college students. J. Abnorm. Soc. Psychol. 46, 245–254. Gilbert, D. T., and Malone, P. S. (1995). The correspondence bias. Psychol. Bull. 117, 21–38. Glick, P. (2002). Sacrificial lambs dressed in wolves’ clothing: Envious prejudice, ideology, and the scapegoating of Jews. In ‘‘Understanding Genocide: The Social Psychology of the Holocaust’’ (L. S. Newman and R. Erber, eds.), pp. 113–142. Oxford University Press, London. Glick, P. (2005). Choice of scapegoats. In ‘‘Reflecting on the Nature of Prejudice’’ ( J. F. Dovidio, P. Glick, and L. Rudman, eds.) pp. 244–261. Blackwell Press, Malden, MA. Glick, P., and Fiske, S. T. (1996). The ambivalent sexism inventory: Differentiating hostile and benevolent sexism. J. Pers. Soc. Psychol. 70, 491–512. Glick, P., and Fiske, S. T. (1999). Sexism and other ‘‘isms’’: Interdependence, status, and the ambivalent content of stereotypes. In ‘‘Sexism and Stereotypes in Modern Society: The Gender Science of Janet Taylor Spence’’ (W. B. Swann, Jr., L. A. Gilbert, and J. Langlois, eds.), pp. 193–221. American Psychological Association, Washington, DC. Glick, P., and Fiske, S. T. (2001a). Ambivalent sexism. In ‘‘Advances in Experimental Social Psychology’’ (M. P. Zanna, ed.), Vol. 33, pp. 115–188. Academic Press, Thousand Oaks, CA.
142
Amy J. C. Cuddy et al.
Glick, P., and Fiske, S. T. (2001b). Ambivalent stereotypes as legitimizing ideologies: Differentiating paternalistic and envious prejudice. In ‘‘The Psychology of Legitimacy: Ideology, Justice, and Intergroup Relations’’ (J. T. Jost and B. Major, eds.), pp. 278–306. Cambridge University Press, New York. Glick, P., Diebold, J., Bailey-Werner, B., and Zhu, L. (1997). The two faces of Adam: Ambivalent sexism and polarized attitudes towards women. Pers. Soc. Psychol. Bull. 23, 1323–1334. Glick, P., Lameiras, M., Fiske, S. T., Eckes, T., Masser, B., Volpato, C., Manganelli, A. M., Pek, J., Huang, L.-l., Sakalli-Ugurlu, N., Rodriguez Castro, Y. R., D’Avila Pereira, M. L., et al. (2004). Men are bad but bold (and women are wonderful but weak): Hostile as well as benevolent attitudes toward men predict gender inequality in 16 nations. J. Pers. Soc. Psychol. 86, 713–728. Glick, P., Thomas, M., Vescio, T. K., and Fiske, S.T. (2007). The stereotype confirming attribution bias. Unpublished data. Green, J. D., and Sedikides, C. (1999). Affect and self-focused attention revisited: The role of affect orientation. Pers. Soc. Psychol. Bull. 25, 104–119. Hack, T. E., Goodwin, S. A., and Fiske, S. T. (2007). The primacy of warmth in social evaluations. Unpublished manuscript, Purdue University. Haddock, G., and Zanna, M. P. (1994). Preferring ‘‘housewives’’ to ‘‘feminists’’: Categorization and the favorability of attitudes toward women. Psychol. Women Q. 18, 25–52. Hamilton, D. L., and Sherman, S. J. (1996). Perceiving persons and groups. Psychol. Rev. 103, 336–355. Hamilton, D. L., and Zanna, M. P. (1974). Context effects in impression formation: Changes in connotative meaning. J. Pers. Soc. Psychol. 29, 649–654. Harris, L. T., and Fiske, S. T. (2006). Dehumanizing the lowest of the low: Neuro-imaging responses to extreme outgroups. Psychol. Sci. 17, 847–853. Harris, L. T., and Fiske, S. T. (2007). Social groups that elicit disgust are differentially processed in mPFC. Soc. Cogn. Affect. Neurosci. 2, 45–51. Harris, L. T., Todorov, A., and Fiske, S. T. (2005). Attributions on the brain: Neuroimaging dispositional inferences, beyond theory of mind. NeuroImage 28, 763–769. Harris, L. T., Fiske, S. T., and Cikara, M. (in press). Envy, as predicted by the stereotype content model: A volatile ambivalence. In ‘‘Envy: Theory and Research’’ (R. Smith, ed.), Oxford, New York. Harwood, J., Giles, H., Ota, H., Pierson, H. D., Gallois, C., Ng, S. H., Lim, T. S., and Somera, L. (1996). College students’ trait ratings of three age groups around the Pacific Rim. J. Cross Cult. Gerontol. 11, 307–317. Heckhausen, J., Dixon, R. A., and Baltes, P. B. (1989). Gains and losses in development throughout adulthood as perceived by different adult age groups. Dev. Psychol. 25, 109–121. Heider, F. (1958). ‘‘The Psychology of Interpersonal Relations.’’ Wiley, Oxford, England. Heine, S. J., Lehman, D. R., Markus, H. R., and Kitayama, S. (1999). Is there a universal need for positive self-regard? Psychol. Rev. 106, 766–794. Hewstone, M., and Ward, C. (1985). Ethnocentrism and causal attribution in Southeast Asia. Pers. Soc. Psychol. Bull. 48, 614–623. Hoffman, C., and Hurst, N. (1990). Gender stereotypes: Perception or rationalization? J. Pers. Soc. Psychol. 58, 197–208. Hofstede, G. (1980). ‘‘Culture’s Consequences: International Differences in Work-Related Values.’’ Sage, Beverly Hills, CA. Hui, C. H., and Triandis, H. C. (1985). Measurement in cross-cultural psychology: A review and comparison of strategies. J. Cross Cult. Psychol. 16, 131–152. Hummert, M. L., Garstka, T. A., O’Brien, L. T., Greenwald, A. G., and Mellott, D. S. (2002). Using the implicit association test to measure age differences in implicit social cognitions. Psychol. Aging 17, 482–495.
Warmth and Competence as Universal Dimensions of Social Perception
143
Hurh, W. M., and Kim, K. C. (1989). The ‘‘success’’ image of Asian Americans: Its validity, and its practical and theoretical implications. Ethnic Racial Studies 12, 512–538. Insko, C. A., and Schopler, J. (1998). Differential distrust of groups and individuals. In ‘‘Intergroup Cognition and Intergroup Behavior’’ (C. Sedikides, J. Schopler, and C. A. Insko, eds.) pp. 75–107. Lawrence Erlbaum, Mahwah, NJ. Izard, C. E. (1991). ‘‘The Psychology of Emotions.’’ Plenum, New York. Izard, C. E., Libero, D. Z., Putnam, P., and Haynes, O. M. (1993). Stability of emotion experiences and their relations to traits of personality. J. Pers. Soc. Psychol. 64, 847–860. Jackman, M. R. (1994). ‘‘The Velvet Glove: Paternalism and Conflict in Gender, Class, and Race Relations.’’ University of California Press, Berkeley. Jackson, L. A., Hodge, C. N., Gerard, D. A., Ingram, J. M., Ervin, K. S., and Sheppard, L. A. (1996). Cognition, affect, and behavior in the prediction of group attitudes. Pers. Soc. Psychol. Bull. 22, 306–316. Jamieson, D. W., Lydon, J. E., and Zanna, M. P. (1987). Attitude and activity preference similarity: Differential bases of interpersonal attraction for low and high self-monitors. J. Pers. Soc. Psychol. 53(6), 1052–1060. Jones, E. E. (1979). The rocky road from acts to dispositions. Am. Psychol. 34, 107–117. Jost, J. T., and Banaji, M. R. (1994). The role of stereotyping in system justification and the production of false consciousness. Br. J. Soc. Psychol. 22, 1–27. Jost, J. T., and Kay, A. C. (2005). Exposure to benevolent sexism and complementary gender stereotypes: Consequences for specific and diffuse forms of system justification. J. Pers. Soc. Psychol. 88, 498–509. Jost, J. T., Burgess, D., and Mosso, C. O. (2001). Conflicts of legitimation among self, group, and system: The integrative potential of system justification theory. In ‘‘The Psychology of Legitimacy’’ ( J. Jost and B. Major, eds.), pp. 363–390. Cambridge University Press, Cambridge, England. Judd, C. M., Hawkins, L. J., Yzerbyt, V., and Kashima, Y. (2005). Fundamental dimensions of social judgment: Understanding the relations between judgments of competence and warmth. J. Pers. Soc. Psychol. 89, 899–913. Karlins, M., Coffman, T. L., and Walters, G. (1969). On the fading of social stereotypes: Studies in three generations of college students. J. Pers. Soc. Psychol. 13, 1–16. Katz, D., and Braly, K. (1933). Racial stereotypes of one hundred college students. J. Abnorm. Soc. Psychol. 28, 280–290. Katz, I., and Hass, R. G. (1986). Racial ambivalence and value conflict: Correlational and priming studies of dual cognitive structures. J. Pers. Soc. Psychol. 55, 893–905. Kay, A. C., and Jost, J. T. (2003). Complementary justice: Effects of ‘‘poor but happy’’ and ‘‘poor but honest’’ stereotype exemplars on system justification and implicit activation of the justice motive. J. Pers. Soc. Psychol. 85, 823–837. Kerr, N. (2002). When is a minority a minority? Active versus passive minority advocacy and social influence. Eur. J. Soc. Psychol. 32, 471–483. Kinder, D. R., and Sears, D. O. (1981). Prejudice and politics: Symbolic racism versus racial threats to the good life. J. Pers. Soc. Psychol. 40, 414–431. Kinder, D. R., and Sears, D. O. (1985). Public opinion and political action. In ‘‘Handbook of Social Psychology’’ (G. Lindzey and E. Aronson, eds.), 3rd ed., Vol. 2, pp. 659–741. Random House, New York. Kinder, D. R., Peters, M. D., Abelson, R. P., and Fiske, S. T. (1980). Presidential prototypes. Pol. Behav. 2, 315–337. Kitano, H. H. L., and Sue, S. (1973). The model minorities. J. Soc. Issues 29, 1–9. Kitayama, S., Markus, H. R., Matsumoto, H., and Norasakkunkit, V. (1997). Individual and collective processes in the construction of the self: Self-enhancement in the United States and self-criticism in Japan. J. Pers. Soc. Psychol. 72, 1245–1267.
144
Amy J. C. Cuddy et al.
Kite, M., Deaux, K., and Miele, M. (1991). Stereotypes of young and old: Does age outweigh gender? Psychol. Aging 6, 19–27. Koyano, W. (1989). Japanese attitudes toward the elderly: A review of research findings. J. Cross Cult. Gerontol. 4, 335–345. Kubicka-Daab, J. (1989). Positivity and negativity effects in impression formation: Differences in processing information about ability and morality dispositions. Polish Psychol. Bull. 20, 295–307. Lazarus, R. S., and Folkman, S. (1984). ‘‘Stress, Appraisal, and Coping.’’ Springer, New York. Lee, T. L., and Fiske, S. T. (2006). Not an outgroup, but not yet an ingroup: Immigrants in the stereotype content model. Int. J. Intercult. Rel. 30, 751–768. Lerner, M. J., and Miller, D. T. (1978). Just world research and the attribution process: Looking back and ahead. Psychol. Bull. 85, 1030–1051. Leslie, L. M., Constantine, V. S., and Fiske, S. T. (2007). Fourth in the Princeton quartet’s analyses of ethnic and national stereotypes: Private ambivalence moderates modern stereotype content 70 years later. Lin, M. H., Kwan, V. S. Y., Cheung, A., and Fiske, S. T. (2005). Stereotype content model explains prejudice for an envied outgroup: Scale of Anti-Asian American stereotypes. Pers. Soc. Psychol. Bull. 31, 34–47. Linssen, H., and Hagendoorn, L. (1994). Social and geographical factors in the explanation of the content of European nationality stereotypes. Br. J. Soc. Psychol. 33, 165–182. Lydon, J. E., Jamieson, D. W., and Zanna, M. P. (1988). Interpersonal similarity and the social and intellectual dimensions of first impressions. Soc. Cogn. 6(4), 269–286. MacDonald, T. K., and Zanna, M. P. (1998). Cross-dimensional ambivalence toward social groups: Can ambivalence affect intentions to hire feminists? Pers. Soc. Psychol. Bull. 24, 427–441. Mackie, D. M., and Smith, E. R. (1998). Intergroup relations: Insights from a theoretically integrative approach. Psychol. Rev. 105, 499–529. Mackie, D. M., Devos, T., and Smith, E. R. (2000). Intergroup emotions: Explaining offensive action tendencies in an intergroup context. J. Pers. Soc. Psychol. 79, 602–616. Maddux, W. W., Galinsky, A. D., Cuddy, A. J. C., and Polifroni, M. (2008). When being a model minority is good . . . and bad: Realistic threat explains negativity toward Asian Americans. Pers. Soc. Psychol. Bull. 34, 74–89. Mandisodza, A. N., Fiske, S. T., and Chaiken, S. (2005). Warmth and competence link intra- and interpersonal evaluations. Massey, D. S. (2007). ‘‘Categorically Unequal: The American Stratification System.’’ Russell Sage, New York. Maykovich, M. (1972). Reciprocity in racial stereotypes: White, black, and yellow. Am. J. Sociol. 77, 876–897. McAdams, D. P., Healy, S., and Krause, S. (1984). Social motives and patterns of friendship. J. Pers. Soc. Psychol. 47, 828–838. McCann, R. M., and Giles, H. (2002). Ageism and the workplace: A communication perspective. In ‘‘Ageism’’ (T. D. Nelson, ed.), pp. 163–199. MIT Press, Cambridge, MA. Mikula, G., Scherer, K. R., and Athenstaedt, U. (1998). The role of injustice in the elicitation of differential emotional reactions. Pers. Soc. Psychol. Bull. 24, 769–783. Mitchell, J. P., Banaji, M. R., and Macrae, C. N. (2005). General and specific contributions of the medial prefrontal cortex to knowledge about mental states. NeuroImage 28, 757–762. Mummendey, A., and Wenzel, M. (1999). Social discrimination and tolerance in intergroup relations: Reactions to intergroup differences. Pers. Soc. Psychol. Rev. 3, 158–175. Neuberg, S. L., and Cottrell, C. A. (2002). Intergroup emotions: A sociofunctional approach. In ‘‘From Prejudice to Intergroup Emotions: Differentiated Reactions to Social Groups’’ (D. Mackie and E. R. Smith, eds.). Psychology Press, New York.
Warmth and Competence as Universal Dimensions of Social Perception
145
Oldmeadow, J., and Fiske, S. T. (2007). Ideology moderates status¼competence stereotypes: Roles for belief in a just world and social dominance orientation. Eur. J. Soc. Psychol. 37, 1135–1148. Osgood, C., Suci, G. J., and Tannenbaum, P. (1957). ‘‘The Measurement of Meaning.’’ University of Illinois Press, Urbana, IL. Panksepp, J. (2000). The neurodynamics of emotions: An evolutionary-neurodevelopmental view. In ‘‘Emotion, Development, and Self-Organization: Dynamic Systems Approaches to Emotional Development’’ (M. D. Lewis and I. Granic, eds.), pp. 236–264. Cambridge University Press, New York. Parrott, W. G., and Smith, R. H. (1993). Distinguishing the experiences of envy and jealousy. J. Pers. Soc. Psychol. 64, 906–920. Pasupathi, M., and Lockenhoff, C. E. (2002). Ageist behavior. In ‘‘Ageism: Stereotyping and Prejudice Against Older Persons’’ (T. Nelson, ed.), pp. 201–246. MIT Press, Cambridge. Peeters, G. (1983). Relational and informational pattern in social cognition. In ‘‘Current Issues in European Social Psychology’’ (W. Doise and S. Moscovici, eds.), pp. 201–237. Cambridge University Press, Cambridge. Peeters, G. (1992). Evaluative meanings of adjectives in vitro and in context: Some theoretical implications and practical consequences of positive-negative asymmetry and behavioraladaptive concepts of evaluation. Psychol. Belgic XXXII-2, 211–231. Peeters, G. (1993). Self-other anchored evaluations of national stereotypes. Unpublished manuscript, Catholic University of Leuven, Leuven, Belgium. Peeters, G. (1995). What’s negative about hatred and positive about love? On negation in cognition, affect, and behavior. In ‘‘Perspectives on Negation’’ (H. C. M. de Swart, and L. J. M. Bergman, eds.), pp. 123–133. Tilburg University Press, Tilburg, The Netherlands. Peeters, G. (2001). From good and bad to can and must: Subjective necessity of acts associated with positively and negatively valued stimuli. Eur. J. Soc. Psychol. 31, 125–136. Peeters, G. (2002). From good and bad to can and must: Subjective necessity of acts associated with positively and negatively valued stimuli. Eur. J. Soc. Psychol. 32, 125–136. Pettigrew, T. F. (1979). The ultimate attribution error: Extending allport’s cognitive analysis of prejudice. Pers. Soc. Psychol. Bull. 5(4), 461–476. Phalet, K., and Poppe, E. (1997). Competence and morality dimensions in national and ethnic stereotypes: A study in six eastern-European countries. Eur. J. Soc. Psychol. 27, 703–723. Plutchik, R. (1980). ‘‘Emotion: A Psychoevolutionary Synthesis.’’ New York, Harper and Row. Poppe, E. (2001). Effects of changes in GNP and perceived group characteristics on national and ethnic stereotypes in central and eastern Europe. J. Appl. Soc. Psychol. 31, 1689–1708. Poppe, E., and Linssen, H. (1999). In-group favouritism and the reflection of realistic dimensions of difference between national states in Central and Eastern European nationality stereotypes. Br. J. Soc. Psychol. 38, 85–102. Reeder, G. D., Kumar, S., Hesson-McInnis, M. S., and Trafimow, D. (2002). Inferences about the morality of an aggressor: The role of perceived motive. J. Pers. Soc. Psychol. 83, 789–803. Ridgeway, C. L. (2001). How do status beliefs develop? The role of resources and interactional experiences. In ‘‘The Psychology of Legitimacy: Emerging Perspectives on Ideology, Justice, and Intergroup Relations’’ ( J. T. Jost and B. Major, eds.), pp. 257–277. Cambridge University Press, Cambridge, England. Roseman, I. J. (1984). Cognitive determinants of emotions: A structural theory. In ‘‘Review of Personality and Social Psychology’’ (P. Shaver, ed.), Vol. 5, pp. 11–36. Sage, Beverly Hills, CA.
146
Amy J. C. Cuddy et al.
Roseman, I., Wiest, C., and Swartz, T. (1994). Phenomenology, behaviors, and goals differentiate discrete emotions. J. Pers. Soc. Psychol. 67, 206–221. Rosen, B., and Jerdee, T. H. (1976a). The nature of job-related age stereotypes on managerial decisions. J. Appl. Soc. Psychol. 62, 180–183. Rosen, B., and Jerdee, T. H. (1976b). Influence of age stereotypes on managerial decisions. Journal of Applied Psychology 61, 428–432. Rosenberg, S., Nelson, C., and Vivekananthan, P. S. (1968). A multidimensional approach to the structure of personality impressions. J. Pers. Soc. Psychol. 9, 283–294. Ross, L. (1977). The intuitive psychologist and his shortcomings: Distortions in the attribution process. In ‘‘Advances in Experimental Social Psychology’’ (L. Berkowitz, ed.), Vol. 10, pp. 174–221. Academic Press, New York. Rozin, P., Lowery, L., Imada, S., and Haidt, J. (1999). The CAD triad hypothesis: A mapping between 3 moral emotions (contempt, anger, disgust) and 3 moral codes (community, autonomy, divinity). J. Pers. Soc. Psychol. 76, 574–586. Rubin, K. H., and Brown, J. (1975). A life-span look at person perception and its relationship to communicative interaction. J. Gerontol. 30, 461–468. Rubenstein, R. L. (1966). ‘‘After Auschwitz: Radical theology and contemporary Judaism.’’ Bobbs-Merrill Educational Publishing, Indianapolis. Rudman, L. A., and Heppen, J. B. (2003). Implicit romantic fantasies and women’s interest in personal power: A glass slipper effect? Pers. Soc. Psychol. Bull. 29, 1357–1370. Rudman, L. A., and Glick, P. (2008). ‘‘The social psychology of gender: How Power and Intimacy Shape Gender Relations.’’ New York, Guilford Press. Ruscher, J. B. (2001). ‘‘The Social Psychology of Prejudiced Communication.’’ Guilford Press, New York. Rush, L. L. (1998). Affective reactions to multiple social stigmas. J. Soc. Psychol. 138, 421–430. Russell, A. M., and Fiske, S. T. (2007). Interpersonal status and competition affect judgments of competence and warmth. Manuscript under review. Russell, A. M., Fiske, S. T., and Moore, G. (2007). Applying the stereotype content model to perceptions of mental illnesses. Manuscript under review. Sandmel, S. (1978). ‘‘Anti-Semitism in the New Testament?’’ Fortress Press, Philadelphia, PA. Scherer, K. R. (1988). ‘‘Facets of emotion: Recent research.’’ Erlbaum, Hillsdale, NJ. Schu¨tz, H., and Six, B. (1996). How strong is the relationship between prejudice and discrimination? A meta-analytic answer. Int. J. Intercult. Rel. 20, 441–462. Sears, D. O. (1989). The person-positivity bias. J. Pers. Soc. Psychol. 44, 233–250. Sedikides, C., Insko, C. A., and Schopler, J.-P., (Eds.). (1998). ‘‘Intergroup Cognition and Intergroup Behavior.’’ Lawrence Erlbaum Associates, Inc., Mahwah, NJ. Sharps, M. J., Price-Sharps, J. L., and Hanson, J. (1998). Attitudes of young adults toward older adults: Evidence from the United States and Thailand. Educ. Gerontol. 24, 655–660. Sidanius, J., and Pratto, F. (1999). ‘‘Social Dominance: An Intergroup Theory of Social Hierarchy and Oppression.’’ Cambridge University Press, New York. Sinclair, L., and Fehr, B. (2005). Voice versus loyalty: Self-construals and responses to dissatisfaction in romantic relationships. J. Exp. Soc. Psychol. 41, 298–304. Singer, M. S. (1986). Age stereotypes as a function of profession. J. Soc. Psychol. 126, 691–692. Singh, R., and Teoh, J. B. P. (2000). Impression formation from intellectual and social traits: Evidence for behavioural adaptation and cognitive processing. Br. J. Soc. Psychol. 39, 537–554. Skowronski, J. J., and Carlston, D. E. (1987). Social judgment and social memory: The role of cue diagnosticity in negativity, positivity, and extremity biases. J. Pers. Soc. Psychol. 52, 689–699.
Warmth and Competence as Universal Dimensions of Social Perception
147
Smith, R. H. (1991). Envy and the sense of injustice. In ‘‘The Psychology of Jealousy and Envy’’ (P. Salovey, ed.), pp. 79–99. Guilford Press, New York. Smith, R. H. (2000). Assimilative and contrastive emotional reactions to upward and downward social comparisons. In ‘‘Handbook of Social Comparison: Theory and Research’’ ( J. Suls and L. Wheeler, eds.), pp. 173–200. Kluwer Academic/Plenum Publishers, New York. Smith, C. A., and Ellsworth, P. C. (1985). Patterns of cognitive appraisal in emotion. J. Pers. Soc. Psychol. 48, 813–838. Smith, R. H., Parrott, W. G., Ozer, D., and Moniz, A. (1994). Subjective injustice and inferiority as predictors of hostile and depressive feelings of envy. Pers. Soc. Psychol. Bull. 20, 705–711. Spears, R., and Manstead, A. S. (1989). The social context of stereotyping and differentiation. Eur. J. Soc. Psychol. 19, 101–121. Spears, R., and Leach, C. W. (2004). Intergroup schadenfreude: Conditions and consequences. In ‘‘The social life of emotions: Studies in emotion and social interaction’’ (L. Z. Tiedens and C. W. Leach, eds.), pp. 336–355. Cambridge University Press, New York, NY. Spence, J. T., and Helmreich, R. L. (1979). The many faces of androgyny: A reply to Locksley and Colten. J. Pers. Soc. Psychol. 37, 1032–1046. Spence, J. T., Helmreich, R. L., and Holahan, C. K. (1979). Negative and positive components of psychological masculinity and femininity and their relationships to selfreports of neurotic and acting out behaviors. J. Pers. Soc. Psychol. 37, 1673–1682. Stangor, C., Sullivan, L. A., and Ford, T. E. (1991). Affective and cognitive determinants of prejudice. Soc. Cogn. 9, 359–380. Staub, E. (1996). Cultural-societal roots of violence: The examples of genocidal violence and of contemporary youth violence in the United States. Am. Psychol. 51, 117–132. Stephan, W. G., and Renfro, C. L. (2002). The role of threat in intergroup relations. In ‘‘From Prejudice to Intergroup Emotions: Differentiated Reactions to Social Groups’’ (D. M. Mackie and E. R. Smith, eds.), pp. 191–207. Psychology Press, New York. Stephan, W. G., and Stephan, C. W. (2000). An integrated threat theory of prejudice. In ‘‘Reducing Prejudice and Discrimination’’ (S. Oskamp, ed.), pp. 23–45. Erlbaum, Mahwah, NJ. Stipek, D. J., Weiner, B., and Li, K.-X. (1989). Testing some attribution–emotion relations in the People’s Republic of China. J. Pers. Soc. Psychol. 56, 109–116. Sue, S., and Kitano, H. H. L. (1973). Stereotypes as a measure of success. J. Soc. Issues 29, 83–98. Sue, S., Sue, D. W., and Sue, D. W. (1975). Asian Americans as a minority group. Am. Psychol. 30, 906–910. Sung, K. T. (2001). Elder respect: Exploration of ideals and forms in East Asia. J. Aging Stud. 15, 13–26. Swim, J. K., and Sanna, L. J. (1996). He’s skilled, she’s lucky: A meta-analysis of observers’ attributions for women’s and men’s successes and failures. Pers. Soc. Psychol. Bull. 22, 507–519. Tafarodi, R. W., and Ho, C. (2006). Moral value, agency, and the measurement of selfesteem. In ‘‘Self-Esteem Issues and Answers: A Sourcebook of Current Perspectives’’ (M. H. Kernis, ed.), pp. 110–116. Psychology Press, New York, NY, US. Tajfel, H. (1981). ‘‘Human Groups and Social Categories.’’ Cambridge University Press, Cambridge, England. Talaska, C. A., Fiske, S. T., and Chaiken, S. (2007). Predicting discrimination: A metaanalysis of the racial attitude-behavior literature. Social Justice Research. Tausch, N., Kenworthy, J., and Hewstone, M. (2007). The confirmability and disconfirmability of trait concepts revisited: Does content matter? J. Pers. Soc. Psychol. 92, 542–556.
148
Amy J. C. Cuddy et al.
Taylor, S. E. (1981). A categorization approach to stereotyping. In ‘‘Cognitive Processes in Stereotyping and Intergroup Behavior’’ (D. L. Hamilton, ed.), pp. 88–114. Elbaum, Hillsdale, NJ. Tesser, A. (1988). Toward a self-evaluation maintenance model of social behavior. In ‘‘Advances in Experimental Social Psychology’’ (L. Berkowitz, ed.), Vol. 21, pp. 181–227. Academic Press, San Diego, CA. Tesser, A., and Collins, J. E. (1988). Emotion in social reflection and comparison situations: Intuitive, systematic, and exploratory approaches. J. Pers. Soc. Psychol. 55, 695–709. Thibaut, J. W., and Kelley, H. H. (1959). ‘‘The Social Psychology of Groups.’’ Transaction Publishers, New Brunswick, NJ. Tien-Hyatt, J. L. (1986/87) Self-perceptions of aging across cultures: Myth or reality? Int J. Aging Hum. Dev. 24, 129–148. Todorov, A., Mandisodza, A. N., Goren, A., and Hall, C. C. (2005). Inferences of competence from faces predict election outcomes. Science 308, 1623–1626. Todorov, A., Gobbini, M. I., Evans, K. K., and Haxby, J. V. (2007). Spontaneous retrieval of affective person knowledge in face perception. Neuropsychologia 445, 163–173. Triandis, H. C. (1995). ‘‘Individualism and Collectivism.’’ Westview Press, Boulder, CO. Van Knippenberg, A., and Van Oers, H. (1984). Social identity and equity concerns in intergroup perceptions. Br. J. Soc. Psychol. 23, 351–361. Vescio, T. K., Gervais, S. J., Snyder, M., and Hoover, A. (2005). Power and the creation of patronizing environments: The stereotype-based behaviors of the powerful and their effects on female performance in masculine domains. J. Pers. Soc. Psychol. 88, 658–672. Volpato, C., Durante, F., and Fiske, S. T. (2007). Using the stereotype content model to examine the evils of fascism: An archival approach. Unpublished manuscript, Universita di Padova. Vonk, R. (1996). Negativity and potency effects in impression formation. Eur. J. Soc. Psychol. 26, 851–865. Vonk, R. (1999). Effects of self-profitability and other-profitability on evaluative judgements of behaviours. Eur. J. Soc. Psychol. 29, 833–842. Wang, J. W., and Cuddy, A. J. C. (2008). Good traits travel: The perceived transitivity of traits across social networks. Paper presented at the 9th Annual Meeting of the Society for Personality and Social Psychology, Albuquerque, NM. Weiner, B. (1980). A cognitive (attribution)-emotion-action model of motivated behavior: An analysis of judgements of help-giving. J. Pers. Soc. Psychol. 39, 186–200. Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psychol. Rev. 92, 548–573. Weiner, B. (2005). ‘‘Social Motivation, Justice, and the Moral Emotions: An Attributional Approach.’’ Lawrence Earlbaum, Mahwah, NJ. Weiner, B., Graham, S., and Chandler, C. (1982). Pity, anger, and guilt: An attributional analysis. Pers. Soc. Psychol. Bull. 8, 226–232. Weiner, B., Perry, R. P., and Magnusson, J. (1988). An attributional analysis of reactions to stigmas. J. Pers. Soc. Psychol. 55, 738–748. White, M. S. (1979). Measuring androgyny in adulthood. Psychol. Women Q. 3, 293–307. Wiggins, J. S. (1979). A psychological taxonomy of trait-descriptive terms: The interpersonal domain. J. Pers. Soc. Psychol. 37, 395–412. Williams, L. N., and Fiske, S. T. (2006). Applying the stereotype content model to intraracial subgroups: Black people’s perceptions about Black people. Unpublished manuscript. Willis, J., and Todorov, A. (2006). First impressions: Making up your mind after a 100-ms. exposure to a face. Psychol. Sci. 17, 592–598.
Warmth and Competence as Universal Dimensions of Social Perception
149
Wojciszke, B. (1994). Multiple meanings of behavior: Construing actions in terms of competence or morality. J. Pers. Soc. Psychol. 67, 222–232. Wojciszke, B. (1997). Parallels between competence- versus morality-related traits and individualistic versus collectivistic values. Eur. J. Soc. Psychol. 27, 245–256. Wojciszke, B. (2005a). Affective concomitants of information on morality and competence. Eur. Psychol. 10, 60–70. Wojciszke, B. (2005b). Morality and competence in person and self perception. Eur. Rev. Social Psychol. 16, 155–188. Wojciszke, B., and Klusek, B. (1996). Moral and competence-related traits in political perception. Polit. Psychol. Bull. 27, 319–324. Wojciszke, B., Brycz, H., and Borkenau, P. (1993). Effects of information content and evaluative extremity on positivity and negativity biases. J. Pers. Soc. Psychol. 64, 327–336. Wojciszke, B., Bazinska, R., and Jaworski, M. (1998). On the dominance of moral categories in impression formation. Pers. Soc. Psychol. Bull. 24, 1245–1257. Wojciszke, B., Abele, A. E., and Baryla, W. (2007). Two dimensions of interpersonal attitudes: Liking depends on communion, respect depends on agency. Ybarra, O., and Stephan, W. G. (1999). Attributional orientations and the prediction of behavior: The attribution-prediction bias. J. Pers. Soc. Psychol. 76, 718–727. Ybarra, O., Chan, E., and Park, D. (2001). Young and old adults’ concerns about morality and competence. Motiv. Emotion 25, 85–100. Ybarra, O., Chan, E., Park, H., Burnstein, E., Monin, B., and Stanik, C. (in press). Life’s recurring challenges and the fundamental dimensions: An integration and its implications for cultural differences and similarities. Eur. J. Soc. Psychol. Yzerbyt, V., Provost, V., and Corneille, O. (2005). Not competent but warm . . . really? Compensatory stereotypes in the French-speaking world. Group Proc. Intergroup Relat. 8, 291–308. Zajonc, R. B. (1998). Emotions. In ‘‘Handbook of Social Psychology’’ (D. T. Gilbert, S. T. Fiske, and G. Lindzey, eds.) 4th ed., Vol. 1, pp. 591–634. McGraw-Hill, Boston. Zanna, M. P., and Hamilton, D. L. (1972). Attribute dimensions and patterns of trait inferences. Psychon. Sci. 27, 353–354. Zanna, M. P., and Hamilton, D. L. (1977). Further evidence for meaning change in impression formation. J. Exp. Soc. Psychol. 13, 224–238. Zemore, S., and Cuddy, A. J. C. (2000). Elderly stereotype contents at the automatic level. Unpublished raw data. Zenner, W. P. (1987). Middleman minorities and genocide. In ‘‘Genocide and the Modern Age: Etiology and Case Studies of Mass Death’’ (I. Walliman and M. N. Dobkowski, eds.), pp. 253–281. Greenwood Press, New York. Zucker, G. S., and Weiner, B. (1993). Conservatism and perceptions of poverty: An attributional analysis. J. Appl. Soc. Psychol. 23, 925–943.
C H A P T E R
T H R E E
A Reciprocal Influence Model of Social Power: Emerging Principles and Lines of Inquiry Dacher Keltner,* Gerben A. Van Kleef,† Serena Chen,* and Michael W. Kraus* Contents 1. 2. 3. 4.
Traditions in the Empirical Study of Power Ultrasociality and Human Hierarchies A Reciprocal Influence Model of Social Power The Acquisition of Social Power 4.1. The disposition to engage socially predicts the acquisition of social power 4.2. Displays of power: Strategic signals of social engagement 5. Power and Social Constraint Processes 5.1. Reputation and gossip as power constraint processes 5.2. Modesty 6. Power as an Interaction Heuristic 7. Power as a Social Affordance 8. Power as a Prioritization Device 8.1. Power and the amplification of preexisting inclinations 8.2. Power and interpersonal responsiveness 9. Future Directions: The Experience of Power and Class and Ideology 9.1. The experience of power: Reciprocal influence dynamics 9.2. Power and the ideologies of agency and obligation 10. Conclusion References
153 154 156 158 159 161 163 164 168 169 170 173 173 174 178 178 180 185 186
Abstract In the present chapter, we advance a reciprocal influence model of social power. Our model is rooted in evolutionist analyses of primate hierarchies, and notions that the capacity for subordinates to form alliances imposes important
* {
Department of Psychology, 3210 Tolman Hall, University of California, Berkeley USA 94720-1650 Department of Social Psychology, University of Amsterdam, Roetersstraat 15, 1018 WB Amsterdam, The Netherlands
Advances in Experimental Social Psychology, Volume 40 ISSN 0065-2601, DOI: 10.1016/S0065-2601(07)00003-2
#
2008 Elsevier Inc. All rights reserved.
151
152
Dacher Keltner et al.
demands upon those in power, and that power heuristically reduces the likelihood of conflicts within groups. Guided by these assumptions, we posit a set of propositions regarding the reciprocal nature of power, and review recent supporting data. With respect to the acquisition of social power, we show that power is afforded to those individuals and strategic behaviors related to advancing the interests of the group. With respect to constraints upon power, we detail how group-based representations (a fellow group member’s reputation), communication (gossip), and self-assessments (an individual’s modest sense of power) constrain the actions of those in power according to how they advance group interests. Finally, with respect to the notion that power acts as a social interaction heuristic, we examine how social power is readily and accurately perceived by group members and gives priority to the emotions, goals, and actions of high-power individuals in shaping interdependent action. We conclude with a discussion of recent studies of the subjective sense of power and class-based ideologies. The fundamental concept in social science is Power, in the same sense that Energy is the fundamental concept in physics . . . The laws of social dynamics are laws which can only be stated in terms of power. (Russell, 1938, p. 10)
Bertrand Russell’s claim that ‘‘the laws of social dynamics are laws that can only be stated in terms of power’’ would have made contact with few empirical findings in social psychology 20 years ago. Since that time, power has become a central area of inquiry, and one with an outpouring of findings that lend credence to Russell’s assertion that to understand the thoughts, feelings, and actions of individuals within social interaction, one must consider their power dynamics (Brauer and Bourhis, 2006; Keltner et al., 2003). In the present chapter, we present a reciprocal influence model of social power. This model is grounded in two assumptions that derive from studies of primate hierarchies. First, power relations are bidirectional, and governed according to the extent to which individuals act in ways that advance the interests of the group. That is, power is acquired by individuals and, just as importantly, granted to others by low-power individuals in affordance and constraint processes that are responsive to how the individual advances the interests of group members. Our second assumption is that power is a heuristic solution to the problem of allocating resources in interdependent relations, and as such, should be a basic dimension of social perception and social behavior. This model helps us frame and address new questions essential to the study of power. How is power acquired and granted to others? What social processes within groups constrain power holders? To what extent do social perceivers reliably perceive others’ capacity for power? How does power influence dyadic exchanges? We rely on our reciprocal influence model of social power and recent empirical studies to provide some initial answers to
Reciprocal Influence and Power
153
these and other questions. We conclude in a more speculative vein, presenting recent evidence concerning the subjective experience of power and how the experience of power might shape class-based ideologies.
1. Traditions in the Empirical Study of Power Social psychological studies of power have concentrated on three broad questions (for reviews, see Keltner et al., 2003; Kipnis, 1976; Ng, 1980; Raven, 1999). First, what are the origins of power? Since French and Raven’s analysis of the bases of power (French and Raven, 1959), investigators have sought to identify the social processes that endow individuals with power. Empirical studies have identified specific behaviors, such as gossip, teasing, and status moves, which influence hierarchy formation in children (e.g., Savin-Williams, 1977), in organizations (Owens and Sutton, 2001), in informal groups (Buss and Craik, 1981), and in the emergence of leaders (Eagly and Johnson, 1990). Other studies have documented how social power derives from membership in demographic groups, such as gender or ethnicity (Berger et al., 1972). A second question concerns the concomitants of power. What does the experience of power correlate with in the phenomenological moment? Studies seeking answers to this question have found that contextual shifts in the individual’s power lead to, for example, variation in cortisol (Ray and Sapolsky, 1992; Sapolsky and Ray, 1989) and testosterone (Bernhardt, 1997; Dabbs, 1997; Gladue et al., 1989; Mazur and Booth, 1998), linguistic and paralinguistic behavior (Dovidio and Ellyson, 1982; Hall et al., 2005; Tiedens and Fragale, 2003), as well as strategic social behavior and mood (Moskowitz, 1994). A third broad question in the empirical literature on power concerns the consequences of power, that is, how power shapes ensuing cognitive, behavioral, and emotional responses (see Bugental, 2000; Kipnis, 1972; Reid and Ng, 1999). Different theoretical models have been advanced to account for how power affects those who have elevated power and those who do not (Keltner et al., 2003). Research within this tradition has examined how the possession (or absence) of power influences, for example, emotion (Langner and Keltner, 2008), approach-related behavior (Galinsky et al., 2003), goal-directed social cognition (Guinote, 2007), the variability of social behavior (Guinote et al., 2002), and the likelihood of condescending behavior (Vescio et al., 2003). These lines of inquiry, both empirical and theoretical, have been characterized by two tendencies, which in part motivated the model we present in this chapter. A first concerns the unit of analysis: almost all studies of social power have focused on the individual as the unit of analysis
154
Dacher Keltner et al.
(for notable recent exceptions, see Copeland, 1994; Guinote et al., 2002; Overbeck and Park, 2001; Tiedens, 2001; Van Kleef et al., 2006; Vescio et al., 2003). Less attention has been paid to the critical question of how power shapes, and is shaped by, dyadic and group processes. How power arises in dyadic and group processes is an important area in need of systematic investigation, for power is inherently relational. An individual’s power (or lack thereof) is shaped by face-to-face dyadic exchanges, group-related processes, and participation in social collectives and ideologies (e.g., Berger et al., 1972; Bourdieu, 1985; Sidanius and Pratto, 2001). Very little is known about how the dyad, the group, and social collective shape the individual’s sense of power. A primary aim of the present chapter will be to fill this lacuna, and to offer a set of theoretical concepts and new findings to clarify how social power is distributed in groups as a result of dyadic exchanges and group-based processes. A second, related tendency in the literature on power is that almost all studies of power to date have conceived of power as a unidirectional phenomenon, originating in the individual, and flowing outward in systematic correlates and consequences. It is now common to study how power determines the individual’s behavior. Or, complementarily, other studies emphasize how power (or status) is an outcome of the individual’s action. There has been little systematic treatment of how social power is actively constructed in processes by which individuals acquire power, and are granted power by others—a longstanding concern in sociological treatments of power, deference, and status (e.g., Emerson, 1962; Goffman, 1967). In the present chapter, we offer a theoretical treatment of the bidirectional nature of power, how it is acquired by individuals, and afforded to them, and how it is regulated within groups. To consider these issues—how power arises in dyads and groups as a result of bidirectional processes—we must first look at the evolution of human hierarchies. Such an analysis sets the stage for our reciprocal influence model of social power and its specific empirical propositions.
2. Ultrasociality and Human Hierarchies Social power reflects the relative influence an individual exerts over his or her interaction partner’s outcomes through the allocation of resources and punishments (De´pret and Fiske, 1999; French and Raven, 1959; Keltner et al., 2003; Lewin, 1951). It translates to the individual’s perceived capacity to modify others’ states by providing or withholding resources or administering punishments, as well as the freedom the individual believes he or she has to deliver resources and punishments (Emerson, 1962; Fiske, 1993; Parker and Rubenstein, 1981; Rusbult et al., 1991; Thibaut and
Reciprocal Influence and Power
155
Kelley, 1959). The individual’s experience and exercise of power occur in dyadic and group-based processes within human hierarchies. The more specific propositions of our reciprocal influence model derive from recent analyses of human ultrasociality, social hierarchies, and different relationships (Boehm, 1999; Caporael and Brewer, 1995; Dunbar, 2004; Fiske, 1991). The most basic assumption emerging from these analyses is that humans are an ultrasocial species, accomplishing most tasks relevant to survival and reproduction, from the provision of resources to the raising of offspring, in highly coordinated, close proximity, face-to-face relationships and groups (Caporael, 1997; Caporael and Brewer, 1995; Keltner and Haidt, 2001). The basic elements of human sociality are relationships, and a central task in human adaptation is to navigate the myriad relationships of human groups effectively (Baumeister and Leary, 1995; Chen and Andersen, 1999; Fiske, 1991). With increases in human sociality and the capacity to communicate and store symbolic information (Dunbar, 2004) came an important property of human social life with profound implications for the distribution and exercise of power: the capacity for subordinates in hierarchies to form alliances and networks. The hierarchical organization of higher primates and early and present-day humans differs from that of other species (Boehm, 1999; de Waal, 1989). Lower status individuals can readily form alliances, most typically dyadic coalitions, which potentially negate any advantages that higher status individuals might enjoy in physical size or power. This development radically shifted how power is acquired and negotiated. The acquisition of power shifted from being based on coercion and assertion to processes by which low-status individuals afford power to high-power individuals (Emerson, 1962). This shift also placed additional importance upon communicative processes in subordinates—e.g., gossip—that can potentially constrain the expression of power of dominant individuals. The capacity for subordinates to form alliances introduced new demands upon individuals in power. An individual’s power depended critically upon that individual’s ability to engage in, and advance, the interests of other group members. Social engagement became the critical ingredient to the acquisition and maintenance of power. For example, in close primate relatives, such as chimps and bonobos, Frans de Waal has shown that social power is based less on sheer strength, coercion, and the unbridled assertion of self-interest, and more on the ability to negotiate conflicts, to enforce group norms, and to allocate resources justly (Aureli and de Waal, 2000). This requirement of those in power to be socially engaged is all the more pronounced in humans. Finally, the centrality and complexity of social relationships in human groups led to a degree of interdependence in human relationships—between parents and offspring, reproductive partners, and same-sex individuals within
156
Dacher Keltner et al.
alliances—that is unprecedented in the primate world (Brewer, 2004; Bugental, 2000; Hrdy, 1999; Rusbult et al., 1991; Sulloway, 1996). Interdependence implies potential competing interests, and the need to establish cooperative mechanisms for negotiating conflicts. Mutually recognized power differences are one such mechanism, serving as a social heuristic that solves more complex problems surrounding the allocation of resources and the coordination of interdependent action. In nonhuman species, well-studied conventionalized status contests— stags locking horns, chimpanzees bearing teeth in threat displays—allow competing group members to establish positions within social hierarchies through signaling, rather than more costly aggressive encounters (Krebs et al., 1993). These status contests make decisions regarding the allocation of resources and the coordination of interdependent action more efficient. As a result, status, or resource holding potential, emerges as a basic property of a repertoire of display behaviors, and as a focus of the social cognition of nonhuman species, which is oriented toward the accurate assessment of conspecifics’ power. Humans rely to an even greater extent upon face-to-face negotiations, rather than violence or territorial arrangements to negotiate competing interests (Boehm, 1999). As a result, power should be an especially potent social heuristic that prioritizes the actions and interests of those with power in situations defined by interdependent action (e.g., Fiske, 1991).
3. A Reciprocal Influence Model of Social Power Human groups, then, are defined by the profound interdependence of their members, and by the capacity for subordinates to form alliances. These properties of human groups place demands upon those in power to act in ways that advance the interests of the group. In addition, they make power a pervasive dimension to social relationships, one that acts in heuristic fashion to pre-empt more costly conflicts and to prioritize the actions and interests of those with power in dyadic exchanges. The above properties of human hierarchies translate to the propositions of a reciprocal influence model of social power, which we summarize in Table 3.1, and which organizes the remainder of this chapter. Our first two propositions concern the acquisition of power. Given the power that subordinates find in forming alliances, we hypothesize that those individuals who actively engage in the interests of others will be afforded power by other group members. We further propose that strategic behaviors that signal the disposition to actively engage in the interests of others will also lead to the affordance of social power. In the formation of hierarchies, the
Reciprocal Influence and Power
157
Table 3.1 A reciprocal influence model of social power: Empirical propositions
The acquisition of social power Proposition 1: Individuals who advance the interests of the group will be afforded social power. Proposition 2: Expressive behaviors associated with social engagement will lead to power affordances. Constraints upon social power Proposition 3: Reputation tracks which group members are worthy of social power. Proposition 4: Gossip is an informal means of identifying individuals who are not deserving of power, and restoring a power balance that serves the interests of the group. Proposition 5: Modesty functions as an internalized regulation mechanism that prevents the abuse of power. Power as an interaction heuristic Proposition 6: Power is readily and accurately perceived in others. Proposition 7: Power prioritizes the emotions, attitudes, and goals of high-power individuals in shaping social interactions.
acquisition of social power is not about manipulation, aggression, or strength; it is based on the ability to act in ways that advance the interests of the group (and that satisfy alliances of subordinates). Our next set of propositions pertains to the constraint processes by which subordinates regulate the power of power holders. With the rise of alliance formations, and the astonishing symbolic and communication capacities of humans, the determination of power within groups increasingly shifted to the actions, communication, and representations of subordinates (e.g., Emerson, 1962). These social and cognitive shifts led to communicative and representational processes by which subordinates afford and constrain the power of those in power. In this chapter we propose, and detail supporting empirical evidence, that the representation of group members’ reputations, and reputation-relevant communication, constrain the actions of those in power. We also suggest that group members constrain their own potential abuses of power through modest self-assessments of power. Our final set of propositions follows from the notion that power acts as a heuristic solution to potential conflict between group members. To the extent that power relations prevent costly aggressive encounters, social power should function something like a social heuristic or relationship model (e.g., Fiske, 1991), quickly and efficiently guiding social perception and behavior. Much as the human mind readily detects baby-like, neotonous cues in others for evolutionary advantage (McArthur and Apatow, 1983), and perceptions of neotony automatically evoke certain reliable patterns of behavior, such as the provision of care, the same should be
158
Dacher Keltner et al.
true of social power. Social power should be readily and accurately identified in other group members, and serve as a guide for how individuals act within dyadic interactions, prioritizing the dispositions, goals, actions, attitudes, and emotions of high-power individuals—our final two propositions.
4. The Acquisition of Social Power Who acquires social power? How is power distributed across group members as hierarchies form? In one well-established line of inquiry, investigators have examined how more static features of the individual— their physical morphology, gender, ethnicity—influence inferences related to power. Here the theoretical notion, supported by numerous studies, is that individuals associated with groups who historically have enjoyed greater economic and political power—for example, European American males or political majorities in U.S. culture—are afforded power as a simple result of their group membership (Berger et al., 1972; Nemeth, 1986). In many group settings and face-to-face interactions, power is negotiated in a more dynamic fashion. On the grammar school playground, in leaderless teams, in groups of friends, in emergent social movements, and on athletic teams, individuals often gravitate to positions of power through processes that are largely independent of more static features of their identity—namely, their patterns of interaction with other group members, their way of being with others. This kind of dynamic acquisition of power has long been of interest to those interested in charisma (Weber, 1947) and the qualities that make for effective leaders (Eagly and Johnson, 1990). In more dynamic settings, who acquires power? In our reciprocal influence model of social power, we reason that the capacity for subordinates within hierarchies to form alliances places demands upon high-power individuals to engage socially and advance the interests of the group. The distribution of power within social groups, therefore, should go preferentially to those individuals who are socially engaged in ways that advance the interests of the group—our social engagement hypothesis. By extension, one would expect social behaviors that are socially engaged, that is, that are oriented toward the interests of others, to prompt attributions of power. As we have already noted, several studies lend credence to the social engagement hypothesis in nonhuman species. Specifically, studies have found that high-status chimpanzees and bonobos acquire and maintain elevated positions of power as a function of their social engagement. Nonhuman primate leadership requires that powerful individuals maintain the social harmony and coherence of relationships and groups through negotiation, reconciliation, and matters of adjudicating the distribution of resources and work (de Waal, 1989).
Reciprocal Influence and Power
159
In humans, select studies provide evidence that could be interpreted as consistent with the social engagement hypothesis. For example, in studies of hierarchy formation in children at a summer camp, Savin-Williams (1977) found that it was the more socially dynamic, outgoing children who rose to positions of leadership. In a study of the social dynamics of members of a fraternity, Keltner et al. (1998) assessed the peer-rated power of 48 members of a fraternity, and examined how they teased one another in a semistructured teasing interaction involving groups of four fraternity members. Consistent with the social engagement hypothesis, the more dynamic, playful, engaging teasers were found to have elevated peer-rated power within the fraternity, independent of whether they were new or older and more established members of the group. More recent tests reported below reveal more rigorous support of the social engagement hypothesis.
4.1. The disposition to engage socially predicts the acquisition of social power In more systematic tests of the social engagement hypothesis, it is necessary to study the formation of hierarchies in emergent groups whose members have no prior history with one another, to identify which group members who all start from similar status positions rise in the social hierarchy. It is also necessary to study the social affordance of power, that is, how group members ascribe power to other group members. Guided by these criteria, Anderson et al. (2001) studied the emergence of hierarchies in three different groups: an all-male fraternity at a Midwestern university; an all female sorority at a Southern university; and in a longitudinal study of the members of a mixed-sex residence hall on a college campus at a Western university. The simple prediction was that individuals, both males and females, who self-report high levels of Extraversion would acquire and maintain elevated levels of power, as afforded by their peers in ratings of influence, prominence, and respect. In the Big Five framework, Extraversion implies an ‘‘energetic approach to the social and material world and includes traits such as sociability, activity, assertiveness, and positive emotionality’’ ( John and Srivastava, 1999, p. 121)—all characteristics that should predispose extraverts to engage in the interest of other group members. This prediction is consistent with findings showing that extraverts report engaging in a variety of social behaviors—conversation, persuasion, conflict resolution, humor— that actively engage with others (Akert and Panter, 1988; Buss, 1996; Buss et al., 1987; Hampson et al., 1987; Kyl-Heku and Buss, 1996; Riggio and Friedman, 1986). Table 3.2 presents the data relevant to our social engagement hypothesis. Consistent with expectation, one can see that highly extraverted males and females in the two same-sex social groups—the fraternity and
160
Dacher Keltner et al.
Table 3.2 Correlations of the Big Five dimensions with social status (summary of three studies)
Dormitory men
Measure
Fraternity Time 2 Time 3 men
Big Five Extraversion .47** Agreeableness .12 Neuroticism .31* Conscientiousness .23 Openness to .05 experience
Sorority women
Dormitory women Time 2 Time 3
.48** .40** .45** .39* .36* .08 .17 .24 .01 .01 .39* .46** .21 .08 .14 .16 .19 .03 .20 .31 .03 .00 .11 .12 .24
Note. * p < .05; ** p < .01.
sorority—were afforded greater status and power by their peers as these two hierarchies formed, as measured in assessments gathered at the beginning of the academic year, when the status hierarchies were still dynamic and forming. The dormitory findings are important for several reasons. First, these longitudinal findings show that the tendency to engage socially (Extraversion) predicts the acquisition of social power at later points in time: in the dormitory study, elevated levels of Extraversion predicted greater influence, prominence, and respect amongst peers for women and men across the course of an academic year. Socially engaged extraverts acquired elevated social status in their peers’ eyes within two weeks of the academic year, as the students were all getting to know one another, and they maintained their positions of elevated status for the next nine months. Social engagement appears to be critical to the stability of a power holder’s position. Second, the sample was diverse with respect to SES and ethnicity and mixed in terms of gender: social engagement predicts the acquisition of social power across groups that are likely to have their own historical legacy with respect to experienced power. In this same investigation, neuroticism for males, which is defined in terms of elevated anxiety, distress, and agitation, was associated with lower peer-related status. This too fits the social engagement hypothesis. Neuroticism has been shown to be associated with an interaction style—self-focus, emotional reactivity, plaintiveness, the over-interpretation of conflict (e.g., Bolger and Schilling, 1991; Keltner, 1996)—that interferes with smooth functioning social relations, and in two studies, cost men in terms of their peer-afforded social status. Interestingly, Agreeableness did not predict the acquisition of power. It is not a pro-social orientation toward others (or an anti-social, Machiavellian orientation) that predicts the acquisition of
Reciprocal Influence and Power
161
power; instead, it is a more general kind of social energy, one that predisposes individuals to engage with other group members in a variety of ways. It is also worth noting that these dynamic dispositional predictors of the acquisition of social status within groups diverge from lay theories of who will rise in social hierarchies. In a survey of undergraduates about which traits would be associated with the acquisition of elevated status, respondents indicated that the more task- and achievement-oriented trait—conscientiousness—would predict the acquisition of elevated status. This lay theory about the acquisition of social power proved to be wrong; conscientiousness, defined by the interest in achievement, task focus, and goal directedness, was unrelated to peer affordances of social power. The acquisition of power appears to hinge more on social engagement than the ability to pursue goals and carry out tasks efficiently. The viability of the social engagement hypothesis rests critically on further research examining different groups, individuals other than college-aged students, and members of different cultures.
4.2. Displays of power: Strategic signals of social engagement Thus far we have seen that in the establishment of male and female hierarchies, the disposition to engage with others—Extraversion—predicts the acquisition of power. In addition, social behaviors that engage others, and presumably advance the interests of the group are rewarded with affordances of power. This association between the disposition toward social engagement and the rise in power has important implications for how people signal social power strategically, a longstanding interest in social psychology (Hall et al., 2005; Henley, 1973; Henley and LaFrance, 1984). More specifically, strategic displays that lead to the acquisition of power should be behaviors that signal social engagement. Social perceivers, furthermore, should prove to be fairly reliable in their detection of the specific behaviors that serve as displays of social power. A recent meta-analytic review by Hall et al. (2005), which included studies published up to 2002, bears upon these two predictions (and in part motivated their review). Hall et al. examined the relation between several nonverbal behaviors (e.g., facial behavior, body movement, interpersonal distance, touch, vocalizations) and power, status, and dominance. Their analysis distinguished between beliefs about the relation of these nonverbal behaviors to power (120 studies) and actually observed relations (91 studies). The results are summarized in Table 3.3. This table reveals several important empirical regularities relevant to the question of how power is signaled strategically. First, reliable relations between nonverbal behaviors and actual power were found only for facial expressiveness, bodily openness, interpersonal distance, loudness of voice, interruptions, and the ability to convey emotions through face and/or voice.
162
Dacher Keltner et al.
Table 3.3 Perceived and actual relations between nonverbal behaviors and power, status, and dominance (adapted from Hall et al., 2005) Nonverbal behavior
Perceived relation
Actual relation
Smiling Gazing Raised brows Facial expressiveness/intensity Nodding Self-touch Other touch Hand/arm gestures Bodily openness Interpersonal distance Loud voice Interruptions Pausing/latency to speak Filled pauses Laughter Rate of speech Vocal pitch Encoding skill/ability to convey emotion
þ þ þ þ þ þ þ þ þ þ x
0 0 0 þ 0 0 0 0 þ þ þ 0 0 0 0 0 þ
Note. þ, –, and 0 denote positive, negative, and nonsignificant (or unreliable) relations, respectively; x denotes a lack of data. Weak and inconsistent effects have been omitted from this table.
As it turns out, however, this constellation of behaviors also tends to be the kind of behaviors that highly extraverted individuals emit reliably in social interactions (e.g., Borkenau and Liebler, 1995; Funder, 1999). Second, the overall correlation between perceived relations and actual relations between power and nonverbal behavior was significant, suggesting that individuals’ beliefs regarding relations between nonverbal behavior and power may be accurate—a theme we examine more directly in the next section. Finally, Table 3.3 reveals that people have quite rich stereotypes of the behaviors associated with power, which are elaborations upon actual behavior to power associations. People associate higher levels of power with less smiling, more gazing, less eyebrow raising, a more expressive face, more nodding, less self-touching, more other-touching, more hand and arm gestures, more bodily openness, smaller interpersonal distances, louder voice, more interruptions, shorter speech latencies, fewer filled pauses, more laughter, higher rate of speech, and a lower voice. It is noteworthy that many of the behaviors individuals erroneously associate with power—for example, increased gazing, nodding, other-touching—are actually agreeable, affiliative behaviors that the data in Table 3.2 reveal to be independent of actual social power.
Reciprocal Influence and Power
163
In addition to these nonverbal behaviors, social engagement (and therefore power) may be signaled by particular patterns of emotional responding. As is the case with the nonverbal behaviors discussed above, in the case of emotional responding, too, well-learned stereotypes may explain at least in part why some patterns of emotional responding lead to the affordance of social power. Tiedens et al. (2000) found that people believe that, in negative situations, high-status individuals are more likely than their lowstatus counterparts to feel anger—a negative emotion that is focused on engaging with others (Keltner et al., 1993)—as opposed to more selffocused emotions such as sadness. In accordance with this belief, people expressing anger are more likely to be seen as high in status and to be granted status by others (Tiedens, 2001). Part of the reason why anger leads to status affordance may be that expressions of anger signal engagement and a desire to change the situation (possibly to the benefit of the group) as opposed to accepting the negative state of affairs. Finally, it is interesting to speculate how responses to the socially engaged style of individuals with power may serve as a self-reinforcing mechanism that perpetuates power structures. Specific responses to the power holder’s socially engaged behavior that are complementary in nature—such as expressions of deference, affirmation, or gratitude—may actually reinforce existing power relations. For example, various politeness tactics—indirect requests, formality, and qualified assertions—tend to systematically characterize how low-power individuals speak to high-power individuals (Brown and Levinson, 1987). This linguistic style of politeness complements the more direct, assertive linguistic style of power holders. In keeping with this thesis, Tiedens and Fragale (2003) showed that complementary responses to dominant nonverbal behaviors produced more comfortable interactions and interpersonal liking. Thus, affording power to individuals who exhibit socially engaged nonverbal behavior by accommodating nonverbally leads to smoother interactions. Over time, people may learn, through repeated reinforcement, that deferential responses to others’ dominant behavior often produce more easy-going, less confrontational interactions. In these ways, low-power individuals afford power to others, thus contributing to the formation and maintenance of status hierarchies.
5. Power and Social Constraint Processes A central assumption of our reciprocal influence model of social power is that power dynamics are bidirectional. Individuals acquire and assert power, as we have seen, through their capacity to engage in the interests of other group members. Strategic displays of socially engaged behaviors are likely to be afforded power. Complementary behaviors on
164
Dacher Keltner et al.
the part of subordinates afford power to other group members. Just as importantly, subordinates readily form face-to-face alliances and fold into more complex networks, and in these social entities have the potential to constrain the actions of those in power. The need to constrain individuals with power is all the more pressing in light of the effects of power upon social behavior. Numerous studies reveal that elevated power leads individuals to act in ways that potentially harm the interests of other group members (for a review, see Keltner et al., 2003). Power renders leaders more likely to act in ways—through disinhibited action, risk taking, and self-serving behavior—that risk damaging the interests of group members, and even, more generally, the cooperative, smooth functioning of the group. This tension between what is needed in leaders and what kind of behavior power tends to produce is likely to be a central motive of social constraint processes by which group members regulate the actions of highpower individuals, and ensure that those who occupy positions of power are likely to act in ways that promote the interests of the group. Most established groups have formalized mechanisms by which low-power individuals regulate the actions of power holders—elections and referenda, evaluation procedures, public forms of accountability, external boards who receive input from group members and provide oversight. These institutionalized checks and balances to the abuses of power are elaborations upon basic communicative processes that accomplish similar functions of regulating the actions of those in power. Group members also rely on more informal means to regulate the actions of power holders. Informal off-record forms of communication, in the form of gossip, teasing, or idle chat, allow group members to comment upon the actions of other group members and, we suggest, constrain power holders and prevent inappropriate individuals (those who do not advance the interests of the group) from rising in power. Here we focus on recent empirical work on two kinds of such informal power constraint processes: reputation-relevant communication (gossip) and modest self-assessments.
5.1. Reputation and gossip as power constraint processes We define reputation as the discussion of an individual’s character by members of a social network. Reputation emerges in communication (Emler, 1994), and is located within the representations of individuals in a social network; it is information about an individual’s engagement in a group as defined by specific norms, values, and needs of the group. The transmission of reputation-related information takes two forms: distributed and discursive (Craik, 2008). Distributive reputation refers to the information about a group member that is stored throughout the group. Distributive reputation is not actively shared, but is accessible by simply
Reciprocal Influence and Power
165
inquiring about a group member (Whitmeyer, 2000). Discursive reputation emerges in active, face-to-face communication amongst group members, in such processes as gossip (Ben-Ze’ev, 1994; Dunbar, 2004; Emler, 1994), teasing (Keltner et al., 2001), and pleasant idle chat. Reputation is a powerful tool of the less powerful and, we suggest, emerged as a mechanism by which group members regulate the distribution of power within their group. The transmission of reputation-relevant information, in distributed and discursive processes, allows low-power individuals to track, evaluate, and comment upon the actions of those individuals in power. Because so many acts of reputation transmission occur when the target of the communication is not present (e.g., gossip), high-power individuals have little or no control over how their reputations are constructed, save by acting in ways that fit the expectations of lowpower individuals. We suggest that distributed and discursive reputational processes will tend to concentrate on the power worthiness of other group members. In light of our reciprocal influence model of social power, we posit that the communication of reputation-relevant information will concentrate on whether other group members are socially engaged, and whether they are trustworthy and oriented to the interests of others. In one recent study guided by this analysis, we examined the content of distributed reputation, that is, group members’ representations of the reputations of their fellow group members (Logli et al., 2008). Ninety-four undergraduates who were members of a residence hall were asked to write about their own reputations and those of two individuals on their hall at two times during an academic year. These narratives were then coded for which personality traits were central to reputation. Importantly, there was a great deal of consensus in group members’ representations of each other’s reputations. And consistent with our foregoing analysis, distributed reputation—their knowledge of others’ reputations—concentrated on traits critical to the status potential of their fellow hall-mate (Extraversion) as well as the individual’s trustworthiness (Agreeableness), but not on other traits like their negative emotionality (Neuroticism) or ability to carry out tasks and goals effectively (Conscientiousness), nor on their idiosyncratic preferences, for example, for particular types of music or forms of recreation (see Table 3.4). This proved to be true of group members’ representations of their own reputations as well as those of other group members. We would further expect forms of discursive reputational processes to center upon the question of whether group members deserve their power and their positions within the social hierarchy. Gossip may be the prototypical kind of discursive communication that transmits reputation-relevant information to other group members. Gossip is a communication of positive or negative reputational information about a group member. Importantly, gossip employs paralinguistic devices, such as indirectness, exaggeration, or humor, which indicate that the claim is not entirely serious, thus providing
166
Dacher Keltner et al.
Table 3.4 Reputation content theme means for self- and peer narratives at Time 1 and Time 3 Self-narratives
Peer narratives
Reputation theme
Time 1
Time 3
Time 1
Time 3
Community Trait Agreeableness Trait Extraversion Idiosyncratic Preferences Energy level Emotion/mood Autonomy Trait Conscientiousness
1.21 1.10
0.82 0.66
0.89 0.93
0.77 0.98
0.64 0.48
0.71 0.40
0.59 0.55
0.53 0.55
0.27 0.24 0.20 0.19
0.25 0.15 0.05 0.16
0.16 0.32 0.09 0.22
0.12 0.22 0.10 0.20
Note. Values refer to the frequencies with which different themes were coded in self- and peer narratives about the individual’s reputation.
low-power individuals a plausible basis for denying any critique contained within the gossip (e.g., Keltner et al., 2001). Gossip also focuses on group members who are not present. These two features of gossip—its off-record quality and its focus on individuals not present—allow group members to comment on the actions of high-power individuals with less fear of conflict or retaliation. Gossip should therefore serve as a means by which a group member’s worthiness of elevated power is evaluated. To test this hypothesis, we investigated the gossip amongst 55 sorority sisters at a West Coast University in the United States (Logli et al., 2008). In an initial phase of the study, sorority sisters completed sociometric ratings of the other sorority sisters, indicating how well-known each sister was, whether she was well-liked, had high status, had deserved status, and had an admirable reputation. All sorority sisters also completed two personality questionnaires relevant to our notion that gossip would target individuals unfit for power—a measure of Agreeableness, which captures the individual’s warmth and kindness, and a measure of Machiavellianism, which captures the individual’s willingness to manipulate and exploit others to rise in status, and cynical expectations that others will do the same. Finally, we gathered sociometric measures of each sister’s self-reported tendency to gossip about other sisters in the sorority. This measure allowed us to identify sisters who were frequent targets of gossip. Our expectations were based on the analysis that gossip is a means by which group members identify sisters who are not deserving of elevated power. The pattern of results observed in Table 3.5, pertaining to the
167
Reciprocal Influence and Power
Table 3.5 Relationships between sociometric ratings and personality measures and the likelihood of being nominated by other group members as a target of gossip Sociometric ratings
Gossip target identification
Well-known Liked Status in house Status deserved Admirable reputation Teased Rater gossips critically about target Target gossips critically about rater Personality measures Agreeableness Machiavellianism
.34* .33* .08 .35* .51* .24þ .84* .74* .39* .28*
* p < .05. Note. Gossip target identification refers to sociometric indications provided by other group members of the likelihood that the individual is gossiped about in the sorority.
profiles of targets of gossip, conform to these expectations. Frequent targets of gossip were well-known but not well-liked. The other group members indicated that these individuals had status that was not well-deserved. They reported that these frequent targets of gossip had poor social reputations. Participants reported frequently teasing and gossiping about these persons— presumably to comment on actions that damage group interests. Moving beyond these sociometric data, other data suggest that gossip identifies group members who are not worthy of positions of power. Given that group members appear to afford power to socially engaged individuals (see Table 3.2), one would expect gossip to target individuals unworthy of power, that is, individuals who are disposed to actually harm the interests of other group members. Consistent with this analysis, one sees in the final two rows of Table 3.5 that frequent targets of gossip self-reported high levels of disagreeableness—they were cold and aggressive. They also reported high levels of Machiavellianism, which captures the tendency to harm others in the pursuit of elevated status and power. In short, gossip targets individuals who are likely to abuse power. Along similar lines, another recent study exploring the antecedents and social functions of gossip indicates that, in addition to the rather banal enjoyment of talking about others who are not present, people often gossip to warn fellow members of a social collective about the defective or normviolating behavior of another member (Beersma and Van Kleef, 2008). Gossipers punish wrongdoers for their deeds by tainting their reputation, thus reducing their social power.
168
Dacher Keltner et al.
5.2. Modesty We have just seen that the informal communication of group members in the form of gossip acts as a constraint upon the actions of those who seek power, targeting individuals who act in ways that harm the interests of other group members. Group members who act in ways that too systematically benefit themselves over the interests of the group, or who strive too explicitly toward the acquisition of power, suffer reputational costs. One interesting possibility is that reputation-related processes such as gossip influence group members’ self-assessments in ways that would lead them to act in ways that subordinate self-interest in the service of advancing the interests of the group. People act in various altruistic ways to enhance their reputations, for example, through gift giving, leaving tips, providing aid, and other forms of generosity (Frank, 1988). Complementarily, group members are likely to internalize reputational discourse processes into selfregulation mechanisms that increase the likelihood that they act in ways that advance the interests of the group. One such candidate is modesty. Modesty, or humility, refers to the tendency to underestimate one’s relative talents and abilities vis-a`-vis other group members, and to underestimate one’s claims to collective resources. Most enduring belief systems have welldeveloped treatments of modesty, which largely focus on cultivating a set of principles that in part constrains the likely excesses of those in power. Within Western thought, the concept of Noblesse Oblige speaks to the necessity for those with power to attend to the needs of those who do not have power. In several Eastern traditions, such as Taoism and Confucianism, there is systematic discussion about the modest leader, as exemplified in the following quote of Lao Tzu: ‘‘To lead the people, walk behind them.’’ Recent work suggests that an ethic of modesty may be built into group members’ self-assessments of their own power, acting as an intrapsychic constraint upon the potential abuses of power (Anderson et al., 2006). In this research, undergraduates formed groups of four to eight participants and visited the laboratory once a week for four weeks. During each visit, the participants engaged in different kinds of group tasks, in one instance disclosing embarrassing experiences to one another, in another engaging in a more competitive collective allocation of resources. After each of the four group interactions, participants rated their own status, as well as the status of each of the other group members, which allowed for the assessment of whether individuals were likely to self-enhance or show modesty in their assessments of their status within the group. Participants also rated how much they liked each of the other group members. This study of the emergence of status in groups yielded two important findings relevant to our present interest in social constraints upon power. The first was a consistent modesty effect in self-assessments of social power: individuals’ self-assessments of status were consistently lower than the status
Reciprocal Influence and Power
169
afforded to them by their peers. This modesty effect diverges from numerous self-enhancement effects documented in the self-literature (e.g., Baumeister, 1998; Dunning et al., 1995; Taylor and Brown, 1988). We would contend that this predilection toward more modest assessments of status within groups acts to constrain the effects of power. The second finding was fitting with this analysis: using cross-lag correlational procedures, Anderson and colleagues found that individuals who selfenhanced in their assessments of their power, holding constant initial levels of being accepted by their group members, were liked less in subsequent interactions of the group. This finding is in keeping with other findings reviewed thus far concerning group representational processes (e.g., that gossip targets Machiavellian group members striving for elevated status). There are costs to having an inflated sense of power in terms of the liking, acceptance, and status affordance of peers. This dynamic is seen in other age groups. For example, children who are prone to bullying often have an inflated sense of their power, and clearly use bullying to assert their dominance in socially unacceptable fashion, but they are systematically shunned and rejected by their peers (Solberg et al., 2007). One intriguing possibility worthy of empirical investigation is that highpower individuals resort to strategic displays of modesty, to afford respect and status to low-power individuals, thus pacifying individuals likely to usurp their power. For example, based on Goffman’s (1957) nuanced analysis of face and deference, investigators have identified several kinds of modest behaviors that afford status to others, including politeness tactics (Brown and Levinson, 1987), acts of etiquette (Elias, 1978), and modest nonverbal displays (e.g., Abu-Lughod, 1986; Keltner and Anderson, 2000). While in general, these modest behaviors tend to be enacted by low-power group members toward high-power individuals, to the extent that effective leadership requires modesty, one might see that powerful individuals rely strategically on these modest acts as a means of maintaining their positions of power. Taken together, the findings we have reviewed in this section speak to the dynamic bidirectional nature of power. We have seen that the communication and social representations of group members systematically focus on the power worthiness of other group members, and whether those individuals tend to act in the interests of the group. We have suggested that these groupbased processes feed back into group members’ own modest self-assessments of their power, a claim in need of more systematic investigation.
6. Power as an Interaction Heuristic Thus far we have examined the distribution of power in social groups. Consistent with our reciprocal influence model, we have seen that individuals who socially engage with other group members acquire power, as do
170
Dacher Keltner et al.
strategic behaviors that are organized according to such intentions. We next examined how power is constrained by those lower in the hierarchy through reputational discourse. We saw that the content of reputation, distributed across group members, focuses on the power worthiness of individuals. And we saw that gossip, a form of discursive reputation transmission, identifies group members who are likely to abuse power. These findings are in keeping with the reciprocal dynamics of social power. Whereas in the first half of this chapter, we have addressed the principles that govern the distribution of power, in this next section we ask how power organizes social perceptions and actions of individuals in dyads and groups. We start from the assumption that social power serves as a heuristic solution to many of the problems of human ultrasociality—increased interdependence, conflict, and the need to negotiate the distribution of resources and work (Chance, 1967). Mutually recognized levels of power between group members pre-empt more costly aggressive encounters over negotiations regarding the distribution of resources and the allocation of work. Mutually recognized levels of power also serve as heuristic solutions to the problems of coordinating interdependent action; more specifically, power prioritizes the actions, intentions, and emotions of high-power individuals in dyadic exchanges.
7. Power as a Social Affordance Within nonhuman species, status contests—stags locking horns, frogs croaking—are designed to provide conspecifics with opportunities for the accurate assessment of one another’s power and, in turn, to pre-empt costly aggressive encounters (Krebs et al., 1993). We likewise assume that there are numerous advantages within human hierarchies to the accurate identification of a group member’s social power. Most obviously, individuals who arrive at accurate appraisals of their own power and that of other group members will avoid competitive status dynamics and conflicts over positions within social hierarchies. Ambiguous appraisals of relative power increase the chances of costly competition and aggressive encounters. It is also plausible that groups who more quickly identify those individuals predisposed to possess power will select more effective leaders—presumably those who will engage in the concerns of group members in ways that benefit the overall functioning of the group. These arguments suggest that power should act like a social affordance (McArthur and Berry, 1987). Social power should have evolved a reliable set of cues (see Table 3.3). Social perceivers, furthermore, should quickly and reliably detect other group members’ social power. Cast within recent advances in the person perception literature (e.g., Funder, 1999), one would
Reciprocal Influence and Power
171
expect group members to achieve impressive levels of consensus and accuracy in judging the power of other group members (see Anderson et al., 2006 for full rationale). A first prediction is that there should be a great deal of consensus in group members’ judgments of one another’s power. To assess this claim, Anderson et al. (2001) had members of dormitory residence halls rate the social status (influence, prominence, and respect) of their peers two weeks into their year together, and then four and nine months later. The question is whether group members agreed in their independent assessments of each other’s social status. Group members showed impressive consensus or agreement in their judgments of who had high and low status the first two weeks of their time together, and these judgments predicted students’ status at the four- and nine-month assessments. Group members appear to form quick and shared impressions of who is and is not worthy of elevated social status, and these perceptions remain stable over time. Importantly, these impressions did not vary significantly according to whether the person being judged was male or female, or according to the individual’s ethnic background. Group members are attuned to the potential others have for social power, beyond ingroup and outgroup social categories. Just as interesting is the question of accuracy: Do group members’ judgments of social power correspond to individuals’ own self-assessments on this important dimension? Both determinants of accuracy—self- and other assessments—are important to consider. Specifically, are individuals aware of their own influence and prominence in groups? And do groups, in effect, pick the right people to lead, the individuals who have a stronger capacity for social power? A positive illusions perspective might suggest that there would be little accuracy in the perception of group members’ status and power. It is well documented that people tend to self-enhance on a variety of dimensions, including their intelligence (Kruger and Dunning, 1999) and personality traits (Messick et al., 1985), and that they do so because of the esteem benefits of selfenhancement (Dunning et al., 1995; Taylor and Brown, 1988). Because one’s social esteem is so intimately tied to self-esteem (Barkow, 1975; Leary et al., 2001), one might expect group members to consistently inflate their own status and power, as a means by which to enhance self-esteem. This line of reasoning is contradicted by the modesty results we presented earlier—that new group members tend to underestimate their degree of power in emergent groups. Furthermore, another line of reasoning would suggest that there are clear costs to such self-enhancement of power, and many benefits to a more accurate correspondence between self- and peer assessments of power and status. More accurate self- and peer assessments of social power clarify individuals’ positions within the social hierarchy. Agreement in self- and peer assessments of social power should enable more smooth functioning social hierarchies within social groups.
172
Dacher Keltner et al.
In two different studies of group dynamics over time, self- and peer ratings of influence, prominence, and respect have been compared, to ascertain the degree of accuracy in the perception of social power (Anderson et al., 2001, 2006). As one can see in Table 3.6, there is a great deal of correspondence between the individual’s own assessment of his or her influence, prominence, and respect, and the ascription of those attributes by peers. The emerging conclusion that people can with immediacy and accuracy judge their own and others’ status is underscored by recent research using ‘‘thin slices’’ of behavior to predict status. This technique is grounded in empirical findings indicating that people are surprisingly accurate at judging others on the basis of very brief observations—or thin slices—of behavior (Ambady et al., 1995). Evidence from studies using this methodology suggests that individuals are able to judge strangers’ status on the basis of brief (90 s) video clips of behavior with considerable accuracy (Dawson and Gilovich, 2004). Interestingly, other research indicates that negotiation outcomes can be reliably predicted from conversational dynamics within the first 5 min of a negotiation, and more importantly, that these dynamics differentially predict outcomes for high as compared to low-status negotiators (Curhan and Pentland, 2007). Finally, there is evidence from neuroscience studies that perceptions of social power and status occur quickly and are represented in specific regions of the brain. In general, the brain is well-equipped to quickly process, store, and retrieve social information, including cues regarding social status and hierarchy (Insel and Fernald, 2004). A recent study showed not only that individuals are able to detect dominance (as signaled through a direct eye gaze and upward head tilt) and submission (averted eye gaze, downward head tilt) in others with as little as 33 ms of exposure to facial cues, but also that differentiation of dominance and submission is represented in the brain (in the mid-superior temporal sulcus, lingual gyrus, and fusiform gyrus) about 200 ms after the face is perceived (Chiao, 2006). All in all, converging evidence from multiple methods points to the immediacy of social power in perception. Table 3.6 Accuracy in perceiving social status: Correlations between self- and peer reports of social status Anderson et al., 2001
Accuracy status ratings
Anderson et al., 2006
Time 1
Time 2
Time 1
Time 2
Time 3
Time 4
.58**
.62**
.59**
.34**
.48**
.42**
Note. * p < .05; ** p < .01.
Reciprocal Influence and Power
173
8. Power as a Prioritization Device In this section, we review two sets of findings consistent with the idea that power may serve as a heuristic solution to the problems of coordinating interdependent action by prioritizing the goals, actions, and emotions of high-power individuals. As a result of this prioritization, those with power tend to play a greater role in shaping dyadic exchanges with the relatively less powerful. Indeed, it is part of the definition of power that those individuals who enjoy elevated power should wield more influence upon the actions of those with less power. The findings we describe below therefore emphasize one direction of influence in our reciprocal influence model—namely, that originating from the powerful to the powerless. In addition, as we shall see, because power differences heuristically prioritize the inclinations of those high relative to low in power, the influence of those with power on the relatively less powerful may at times result in socially inappropriate behavior.
8.1. Power and the amplification of preexisting inclinations Studies documenting that power amplifies individuals’ pursuit of goals and the expression of their states and traits fits the notion that power serves as a prioritization heuristic, leading people in positions of power to exert a greater impact on those with lower power. For example, across several studies, Chen et al. (2001) showed that an experimental manipulation of power increased the tendency for individuals to pursue their chronic relationship goals (i.e., communal-oriented goals vs. exchange-oriented goals) in a dyadic setting. More specifically, individuals who reported the tendency to chronically pursue communally oriented goals, which involve a focus on the needs and welfare of others, behaved in a more socially responsible fashion toward an alleged other participant when primed with high power relative to low power. In contrast, individuals who reported the chronic pursuit of exchange-oriented goals in their relationships, which engender a focus on one’s own needs and interests, responded in a more self-interested fashion when primed with high power relative to low power. Along conceptually related lines, Van Kleef and Coˆte´ (2007) showed that people with higher levels of power are more likely to behave according to their desires than those with low power. In a conflict setting, parties who felt a desire to retaliate in response to inappropriate anger displays by their opponent were more likely to do so when they felt more powerful. Beyond goals and desires, research has shown that elevated power is linked to a greater tendency to express one’s true attitudes and feelings. For example, Anderson and Berdahl (2002) showed that people with high
174
Dacher Keltner et al.
power are more apt to express their attitudes and less inclined to keep their disagreements to themselves in a discussion with another person. In contrast, low-power individuals reported inhibiting the expression of their true attitudes. In essence, high power leads one’s views to be more likely to be ‘‘on the table’’ for discussion. Along related lines, there is evidence that elevated power is associated with greater correspondence between one’s inner experience and outward expression of emotion. For example, research has shown greater correspondence between self-reports of pleasure and smiling behavior among participants randomly assigned to a highpower position relative to their counterparts assigned to a low-power role (Hecht and LaFrance, 1998).
8.2. Power and interpersonal responsiveness From studies of obedience to authority and conformity (Asch, 1951; Milgram, 1963), to studies of conversational dynamics (Brown and Levinson, 1987), it is clear that low-power individuals modify their behavior in deference to high-power individuals, and that the direction of joint or collective action is disproportionately shaped by those in power. Perhaps a more radical possibility is that the actions and representations of people with power exert influences upon more deeply rooted and involuntary processes such as emotional experience and attitude shifts. Humans are a mimetic species—we routinely imitate the actions of those around us (Chartrand and Bargh, 1999)—processes enabled by mirror neurons and regions of the brain that serve basic empathic and imitative responses (Preston and de Waal, 2002). Recent empirical evidence indicates that low-power individuals shift toward the emotions and attitudes of those individuals in power, often without being aware of these shifts. In one study of emotional convergence (the process by which social actors’ emotions come to resemble those of other individuals), friends came to the laboratory at two different times during the year and reported their emotional reactions to different evocative stimuli, such as humorous or disturbing film clips (Anderson et al., 2003). The general pattern observed was that the emotions of friends, as assessed in response to these controlled evocative stimuli, converged over the course of the year: they became more similar in valence and intensity. Just as basic physiological cycles of individuals in close proximity tend to converge and become synchronized (e.g., McClintock, 1971), this study found that young adults’ basic emotions came to resemble one another. With respect to power differences, the central question was whether those individuals with less power within a dyadic bond would make more of the change in the emotional convergence process over the course of the year than high-power individuals. That is, did the emotions of low-power individuals, thought by many to be outside of volitional control, begin to conform to the emotions of high-power friends over time? To address this
175
Reciprocal Influence and Power
question, Anderson and colleagues gathered self-reports of power and influence within the relationship, and identified roommates who felt that they had more power in their relationship, and those who felt they had less power. They then asked whether the emotional style of the high-power individuals, as assessed early in the year, shaped the emergent emotional style of the low power roommate over the course of the year. As can be seen in Table 3.7, this proved to be the case: the emotional styles of low-power individuals shifted over the year to resemble the emotional profiles of high-power individuals. Powerful partners’ own emotions at a later time point were not predicted by their partners’ emotions at Time 1. These null findings emerged for positive and negative emotion. In contrast, the emotions of the less powerful partner at Time 2 were predicted by their partners’ prior emotions. Thus, these findings indicate that lowpower individuals shift their basic emotional reactions to resemble the emotional reactions of high-power individuals within their immediate social environment. These findings fit one of the well-documented properties of the social interactions of high- and low-power individuals: that low-power individuals attend quite carefully to the actions of high-power individuals, who in turn are relatively unaware of the actions of low-power individuals (e.g., Erber and Fiske, 1984; Fiske, 1993; Fiske and De´pret, 1996; Goodwin et al., 2000). This power-related difference in social attention is likely to account, at least in part, for the patterns of emotional convergence. However, as we shall see below, other more motivational processes also appear to play a role Table 3.7 Power in the emotional convergence process: Cross-lagged correlations between participants’ emotions at Time 2 and their partner’s emotions at Time 1 separately for participants high and low in power
Relationship
Romantic relationships Total emotion Positive emotion Negative emotion Dormitory roommates Total emotion Positive emotion Negative emotion
Participants with more power
Participants with less power
.19 .12 .27
.69** .50** .53**
.10 .32 .10
.40** .38* .42**
* p < .05; ** p < .01. Note. The correlations in the first column indicate how well the later emotions of the participants with more power were predicted by the prior emotions of the participant with less power. The correlations in the second column indicate how well the later emotions of the participants with less power were predicted by the prior emotions of the participants with more power.
176
Dacher Keltner et al.
in creating the asymmetrical patterns of responding that are so characteristic of interactions between individuals with different levels of power. A recent study by Van Kleef et al. (2008) examined patterns of emotional responding to others’ emotions as a function of power. Same-sex dyads composed of unacquainted individuals were prompted to talk about instances in their life that had caused them suffering. At several time points during the interaction, participants’ emotions were assessed through selfreport as well as physiological measures. The data showed that participants with a higher subjective sense of power felt less reciprocal distress in response to their partner’s suffering than did those with a lower sense of power. Furthermore, high-power participants responded less compassionately to their partner’s distress. Physiological data (i.e., measures of respiratory sinus arrhythmia) also indicated that high-power individuals were less affected emotionally by their partner’s accounts of distress. These differences could not be attributed to differential attention to the other’s emotion: low- and high-power participants showed similar levels of empathic accuracy. This suggests that the differential responses of high- and low-power individuals may have a more motivational underpinning. In line with such an explanation, high-power individuals reported a weaker inclination to develop a friendly relationship with their interaction partner, suggesting that they were less motivated to invest in them emotionally. Other studies have looked at the role of power in determining responses to others’ emotions in more competitively structured settings. Van Kleef et al. (2004) examined whether behavioral reactions to expressions of anger and happiness in negotiation are moderated by outcome dependency (a proxy of power) and several other variables affecting the motivation to engage in thorough information processing. They found that expressions of anger elicited larger concessions than expressions of happiness, but only in negotiators who were dependent on their opponent for their outcomes. Participants under outcome dependency were strongly affected by the opponent’s emotion, conceding more to an angry than to a happy adversary. Negotiators under low-outcome dependency (and with low information processing motivation), in contrast, were impervious to their counterpart’s emotions. Other studies have documented that this effect generalizes across different power bases (Friedman et al., 2004; Sinaceur and Tiedens, 2006; Van Kleef et al., 2006). For instance, Van Kleef et al. (2006) showed that negotiators with few or poor alternatives to a negotiated agreement, little support from their management, or low legitimate power were strongly affected by their opponent’s anger. In contrast, negotiators with many or highly attractive alternatives, strong support from management, or high legitimate power were immune to their counterpart’s emotional state. It is clear, then, that individuals with high power do not yield to others’ anger. But is this because they attend less to their less powerful counterparts (cf. Fiske, 1993)? Recent evidence suggests that there is more to it.
Reciprocal Influence and Power
177
Van Kleef and Coˆte´ (2007) investigated the interplay between power and the appropriateness of anger expressions in conflict, which was manipulated by having a simulated opponent’s expressions of anger violate a display rule or not. The results showed that the appropriateness of the opponent’s anger did not affect the behavior of low-power negotiators—they conceded more to angry opponents than to nonemotional ones, regardless of whether the anger violated a display rule. In contrast, the appropriateness of the opponent’s anger did matter when the focal negotiator had high power. Highpower negotiators, besides conceding less in general, were found to be especially intransigent when they deemed the opponent’s anger inappropriate. Thus, high-power participants were not insensitive to their opponent’s emotions; instead, they selectively reacted to the other’s emotions when doing so could further their own goals. Together, these studies suggest an important qualification of the widespread idea that powerful individuals pay less attention to their social environment than low-power individuals (e.g., Fiske, 1993). It appears as though high-power people do not necessarily attend less to others in general (see also Chen et al., 2004; Overbeck and Park, 2001, 2006); rather, they attend selectively to emotions and behaviors that they can turn to their own advantage. Powerful people can afford to ignore the individual motivations and desires of others and think of them in stereotypical as opposed to individuating ways. However, high-power people do seem attuned to situations in which others can be used to further their own goals. More generally, these studies of social interaction suggest that highpower individuals evoke in others confirmatory patterns of emotions, attitudes, and behavior (see also Copeland, 1994). An illustrative study by De Dreu and Van Kleef (2004) showed that low-power negotiators who were asked leading questions regarding their future cooperation or competition by a high-power counterpart confirmed the other’s expectations by exhibiting the expected behavior. That is, they behaved more cooperatively after the other had asked leading questions about cooperation (e.g., ‘‘What do you like about cooperating in negotiations?’’), and more competitively after the other had asked about competition (e.g., ‘‘What do you like about winning in negotiations?’’). Low-power negotiators did not initiate a similar self-fulfilling prophecy. Thus it may be that the consensus high-power individuals tend to evoke in those individuals who surround them enhances the certainty and conviction in their views. Overall, the findings discussed in this section fit the proposition that power serves as a prioritization heuristic in dyadic interactions between those with power and those without. This heuristic leads to a disproportionate impact for people in positions of power on those with relatively less power. The wide-ranging research we described suggests that interactions between the powerful and powerless are shaped more by the emotions, attitudes, and goals of the former than the latter.
178
Dacher Keltner et al.
9. Future Directions: The Experience of Power and Class and Ideology Our reciprocal influence model of social power has largely concerned itself with how power is distributed across groups and organizes dyadic exchanges. We have seen that power is afforded to those individuals who advance the interests of the group. Group-based representations (a fellow group member’s reputation), communication (gossip), and self-assessments (an individual’s modest sense of power) reflect the bidirectional nature of power within groups, and how individuals are afforded power by other group members, and constrained in their actions according to how they advance group interests. We have seen that power can be thought of as a social interaction heuristic: it is readily and accurately perceived by group members, and serves as a prioritization device in dyadic interaction, giving priority to the emotions, goals, and actions of high-power individuals in shaping interdependent action. The preliminary nature of the data we have reviewed in support of our reciprocal influence model of social power calls out for further inquiry. For example, it will be interesting to experimentally address whether strategic displays of social engagement lead to shifts in social power (e.g., Tiedens, 2001). Can group members differentiate between sincere and insincere, or feigned, displays of strategic engagement? Does the social engagement hypothesis apply to all manner of groups? Our analysis of the constraint processes that regulate the distribution and conduct of power also warrants further empirical study. Does a group member’s modest self-assessment track his or her reputation in the group? Do immodest leaders suffer reputational costs? Finally, and just as critically, our propositions related to the notion that power serves as an interaction heuristic are in need of further investigation. In this concluding section of our chapter, we draw upon what has been learned in the study of power in dyads and groups from our reciprocal influence model of social power to highlight two new areas of inquiry. A first focuses on the individual, and concerns the experience of power. A second focuses on the social collective, and concerns how the principles of reciprocal power might become elaborated into social ideologies.
9.1. The experience of power: Reciprocal influence dynamics The individual’s sense of power has many well-theorized roots, but has been little studied with social psychological methods. More ‘‘structural’’ perspectives situate the origins of the individual’s social power within institutional contexts. For example, the sense of power is assumed to be amplified in social institutions that selectively distribute power, such as elite preparatory schools and social clubs (e.g., Domhoff, 1998). Broader ideologies that
Reciprocal Influence and Power
179
deem who is worthy of power and who is not are also hypothesized to influence the individual’s sense of power. This kind of analysis has been applied to religion, medical practice, and stereotypes, all social institutions or representations that putatively alter the individual’s own sense of power (Foucault, 1977; Jost and Banaji, 1994). To advance the study of the experience of power, we have developed an eight-item measure of the sense of power, whose items are presented in Table 3.8 (Anderson and Galinsky, 2006; Anderson et al., 2008). These items have satisfactory alpha coefficients and test–retest reliability. Table 3.9 presents a series of preliminary findings related to this 8-item self-report measure. As one might expect, self-reports of the experience of power correlate moderately with a sense of personal control and dominance, which reflects the broad class of strategies used in the acquisition of power, although the magnitude of these correlations suggests that power is distinct from these closely related constructs. The next set of findings indicates, quite intuitively, that the sense of power tracks structural sources of power: MBA students with elevated authority in organizations report having a greater sense of power; people who subjectively report lower SES vis-a`-vis their fellow students also report lower levels of subjective power; and the more leadership positions the individual enjoys in a social group, one sees in the next row, the more subjective power one experiences. Finally, in two different groups—a sorority and a dormitory hall—one sees that the subjective sense of power covaries with peer reports of elevated social status, but not with how well liked the individual is. Power is not simply some facet of a social perceptual halo effect; the individual’s sense of power is independent of being liked. These preliminary findings set the stage for other lines of inquiry. Two conceptual questions stand out. A first concerns the degree of across-context variation that one observes in the experience of power. One model of power holds that it is fairly restricted to hierarchical relationships, and that it is fairly fixed once roles within hierarchies are established. This view Table 3.8 Items to measure the subjective sense of power
I can get people to listen to what I say. Even when I try, I am not able to get my way. I think I have a great deal of power. Even if I voice them, my views have little sway. I can get others to do what I want. My ideas and opinions are often ignored. If I want to, I get to make decisions. My wishes don’t carry much weight. Note. Items are reported in Anderson and Galinsky (2006).
180
Dacher Keltner et al.
Table 3.9 Relations between the sense of power with measures of related constructs and objective and socially afforded origins of power Subjective sense of power
Self-reported traits Internal locus of control Dominance Structural factors Job level in organization Socioeconomic status Leadership positions Peer-measured status In sorority In dormitory Peer-measured popularity In sorority In dormitory
.35** .51** .31** .37** .26* .37** .32** .01 .18
* p < .05; ** p < .01.
might suggest that one would observe little contextual variation in the experience of power. Our approach diverges from this view of social power. We suggest that recognized power differences are a pervasive solution to potential conflict over competing interests in all manner of relationships— for example, between status rivals or parent and child—and the product of bidirectional processes. We would therefore expect the individual’s sense of power to vary according to social context. Indeed, experienced sampling data of Moskowitz (1994) suggest that power-related behaviors vary quite dramatically in different contexts. We would expect similar contextual variation in the experience of power, given power’s pervasiveness and function as a social interaction heuristic. A second conceptual concern pertains to the extent to which the sense of power is reciprocally determined by internal and social processes, in particular power-related actions on the part of the actor, and reputation-related actions on the part of other group members. We have argued that power is reciprocally determined, acquired by the actor, and afforded to the individual by others, according to the extent to which that individual advances group interests. Basic work on how the experience of power tracks these sources of power is needed.
9.2. Power and the ideologies of agency and obligation Social ideologies—broad beliefs systems of a group or culture—are deeply intertwined with conceptions of power (Fiske, 1991; Jost, 2006; Tetlock, 1992). Social ideologies, be they liberal or conservative, involve concepts
Reciprocal Influence and Power
181
regarding who is worthy of power, how it is to be distributed, and how it should influence the allocation of resources and punishments. The recent interest in ideology (e.g., Jost, 2006; Jost et al., 2003) is based on studies of the systematic content of ideologies, and how they may vary according to social cognitive factors, like the need for cognition. One can just as readily ask whether there is an ideological direction to the experience of power. Does an individual’s position within a social hierarchy give rise to certain ideological constructs? We would suggest yes. We suggest that ideologies emerge out of the experience of power and its consequences, and the need to justify these patterns of social response. With respect to the experience of power, several new studies, which we review in Table 3.10, suggest that elevated power is associated with certain social precepts and actions that should feed into the individual’s ideological inclinations. More specifically, these studies reveal that power predisposes individuals to be attuned to potential goals and within situations, to act in unfettered and free fashion, to be optimistic about the pursuit of goals, and to take risks (see summary in Table 3.10). These same studies suggest that experimentally induced low levels of power, in contrast, orient individuals to elevated threat within the environment, as well as to the risks and constraints of pursuing goals and rewards (which of course are greater for those in low-power positions). We propose, in keeping with social intuitionist accounts of morality (Haidt, 2001), that individuals invoke ideological concepts to explain and rationalize their own power-related behavior, as well as the actions of other individuals. In Table 3.11, we summarize a set of predictions regarding how different levels of power might covary with the endorsement of different ideological principles. Given how power gives rise to disinhibited, rewardoriented behavior, and goal-directed cognition, we posit that elevated power will be associated with an ideology of agency. This ideology is in keeping with the effects of power upon the individual, and is defined by an emphasis on freedoms and rights (which enable unfettered action), dispositionist social explanations of people’s places within social hierarchies, and equity as a principle to govern the allocation of resources. These ideological principles readily follow from the high-power individual’s action and reward orientation, and justify his or her elevated standing within social hierarchies. In contrast, reduced power gives rise to perceptions of threat and disinhibition. Ideological principles that fit more readily with the effects of reduced power, and the lives of low-power individuals, are an emphasis on obligations and duties as moral principles (which bind low-power people to others), contextualist explanations of people’s places within hierarchies, and the principles of equality and need as ones that govern the allocation of resources. No experimental studies have tested these predictions, and they certainly warrant exploration. An emergent literature on socio-economic status (SES), defined in terms of education, family wealth, and prestige of occupation,
182 Table 3.10
Recent findings linking power to increased freedom of action, risky choices, and optimism
Source
Power manipulation
Task
Anderson and Berdahl, 2002 (Study 1)
Dispositional power
Dyadic negotiation task
Anderson and Galinsky, 2006 (Study 1)
Dispositional power and power situation prime
Estimates of future personal outcomes
Anderson and Galinsky, 2006 (Study 2) Anderson and Galinsky, 2006 (Study 3) Anderson and Galinsky, 2006 (Study 5)
Dispositional power and power situation prime Nonconscious power prime Structural power
Estimates of fatalities from potential disasters Choice of action in Asian disease problem Dyadic negotiation task
Anderson and Thompson, 2004
Structural power
Dyadic negotiation task
De Dreu and Van Kleef, 2004 (Studies 1 and 2)
Structural power
Dyadic negotiation task
Galinsky et al., 2003 (Study 1)
Structural power
Simulated blackjack game
Effect on emotion, thought, or action
" power–" positive affect, expression of true attitudes, perception of rewards, and # perception of threats " power–" optimism for achieving positive and avoiding negative outcomes. " power–# estimates of fatalities " power–" likelihood of risky choice " power–" sharing information during negotiation " power–positive affect predicts cooperative negotiations # power–" information gathering, and " impression formation goals " power–" likelihood of taking a card
Galinsky et al., 2003 (Study 2)
Power situation prime
Guinote et al., 2006
Minimal group procedure
Guinote et al., 2002
Structural power (group based)
Group task aimed at solving a social dilemma
Magee et al., 2007 (Study 1)
Power situation prime
Magee et al., 2007 (Study 2)
Nonconscious prime
Estimates of likelihood to negotiate Debate competition scenario
Action taken against external environment Perceptions of a potential interaction partner
Magee et al., 2007 (Study 3 and 4) Structural power
Dyadic negotiation task
Snibbe and Markus, 2005
Educational attainment
Vescio et al., 2005 (Study 1)
Gender
CD choice in reactance paradigm Team member assessments when focused on weaknesses or strengths of subordinates
Vescio et al., 2003 (Study 1)
Gender
Assessment of female subordinate’s job performance when focused on weaknesses or strengths
" power–" likelihood of taking action # power–" focus on external environment (dispositions of potential partner) " power–" variability in behavior and selfdescriptions " power–" likelihood to initiate a price negotiation " power–" likelihood to make opening argument " power–" likelihood to make first offer in negotiation " power–" evaluation of chosen objects " power males–" patronizing (assign lower positions and give higher praise) of subordinates when weakness focused " power males–assigned less valued tasks and said had less task relevant attributes when focused on weaknesses
183
184
Dacher Keltner et al.
Table 3.11 Power, social class and the ideologies of freedom and obligation
Moral principles Explanations of hierarchy Resource distribution
Elevated power
Reduced power
Freedom, rights Dispositionist Equity
Obligation, duty Contextualist Equality, need
however, has begun to explore such ideas, and the findings are consistent with the foregoing analysis. For example, in a classic study on culture and morality, lower and upper class participants in Brazil and Philadelphia read a number of vivid scenarios and indicated whether they thought the people in the scenarios should be punished for their actions (Haidt et al., 1993). Some of these situations involved harm, some impure but harmless acts, and others violations of social obligations and duties. Members of both cultures and of both classes were strongly inclined to punish perpetrators of harmful acts. For every other violation, however, the lower SES students expressed a greater sense of moral offense (as indicated by the reported tendency to punish). Whereas lower SES individuals define morality as including issues of obligation, duty, and purity, these are not deemed as moral concerns by upper SES individuals, who instead define morality in terms of freedoms and individual rights. In more recent work, Snibbe and Markus (2005) found that the ideology of agency is systematically related to SES-related differences in preference. Specifically, these authors found that college educated upper SES adults were more likely to listen to music emphasizing self-interest and actualization, and were also more likely to value objects of their own choosing. In contrast, lower SES adults showed no such bias. In our own work on SES and the ideology of agency, we examined how SES influences explanations of inequality (Kraus and Keltner, 2008). In one of the studies, students were presented with a figure representing the rise in income inequality over the past 30 years in the United States. They then were asked to explain this rise in the income gap between rich and poor by attributing this economic condition to dispositional factors (talent, effort, motivation) as well as contextual factors (educational opportunity). Students who indicated higher levels of SES favored more dispositional over contextual factors in explaining the income gap when compared to students from lower SES backgrounds. In evidence that speaks more directly to the ideology of agency, upper SES individuals also indicated a greater sense of personal control over the outcomes in their lives, and this sense of control mediated the relationship between their social class self-reports and dispositional explanations of social inequality (see Fig. 3.1), suggesting that highstatus individuals perceived they had control of their environments, and in turn, judged societal inequality as under individual control.
185
Reciprocal Influence and Power
Subjective SES
.19*
Subjective SES
−.31*
Sense of control
−.22*
Contextual explanations
−.40**
Contextual explanations
Figure 3.1 Model of the relationship between subjective SES and contextual explanations mediated by sense of control after accounting for ethnicity and objective SES.
Similar evidence on SES and the ideology of agency has been found in younger age groups and in different cultures. Higher-status Icelandic 12-year-olds were more likely to describe themselves in terms of psychological factors such as thoughts and feelings compared to lower class children of the same age (Hart and Edelstein, 1992). These disposition biases are also seen in terms of perceptions of the powerful. When information about causes of behavior was ambiguous, actions of powerless actors were seen as situational and those of powerful actors were seen as more dispositional (Overbeck et al., 2006).
10. Conclusion Power is inherently social: its subjective qualities, and effects upon thought, action, and feeling, are shaped by bidirectional processes between individuals in interactions. In the present chapter, we advanced a reciprocal influence model of social power, to begin to understand the bidirectional processes of social power. This model is based on evolutionist analyses of human hierarchies, and notions that the capacity for subordinates to form alliances imposes important demands upon those in power, and that power heuristically reduces the likelihood of conflicts within groups. Guided by these assumptions, we then posited a set of propositions regarding the reciprocal nature of power, and reviewed recent supporting data. With respect to the acquisition of social power, we saw that power is afforded to those individuals who advance the interests of the group. With respect to constraints upon power and the abuses it enables, we reviewed reasoning and evidence showing that group-based representations (a fellow group member’s reputation), communication (gossip), and self-assessments (an individual’s modest sense of power) constrain the actions of those in
186
Dacher Keltner et al.
power according to how they advance group interests. Finally, we reviewed several different kinds of evidence suggesting that power can be thought of as a social interaction heuristic: it is readily and accurately perceived by group members, and serves as a prioritization device in dyadic interaction, giving priority to the emotions, goals, and actions of high-power individuals in shaping interdependent action. The promise of our reciprocal influence model is evident in the clear need for further inquiry in several of the domains we have reviewed. We believe this kind of research will help to further illuminate this most pervasive dimension of human social life—power.
REFERENCES Abu-Lughod, L. (1986). ‘‘Veiled Sentiments: Honor and Poetry in a Bedouin Society.’’ University of California Press, Berkeley, California. Akert, R. M., and Panter, A. T. (1988). Extraversion and the ability to decode non-verbal communication. Pers. Individ. Dif. 9(6), 965–972. Ambady, N., Hallahan, M., and Rosenthal, R. (1995). On judging and being judged accurately in zero-acquaintance situations. J. Pers. Soc. Psychol. 69(3), 518–529. Anderson, C., and Berdahl, J. L. (2002). The experience of power: Examining the effects of power on approach and inhibition tendencies. J. Pers. Soc. Psychol. 83(6), 1362–1377. Anderson, C., and Galinsky, A. D. (2006). Power, optimism, and risk-taking. Eur. J. Soc. Psychol. 36(4), 511–536. Anderson, C., and Thompson, L. (2004). Affect from the top down: How powerful individuals’ positive affect shapes negotiations. Organ. Behav. Hum. Decis. Process. 95(2), 125–139. Anderson, C., John, O. P., Keltner, D., and Kring, A. M. (2001). Who attains social status? Effects of personality and physical attractiveness in social groups. J. Pers. Soc. Psychol. 81, 116–132. Anderson, C., Keltner, D., and John, O. P. (2003). Emotional convergence between people over time. J. Pers. Soc. Psychol. 84(5), 1054–1068. Anderson, C., Srivastava, S., Beer, J. S., Spataro, S. E., and Chatman, J. A. (2006). Knowing your place: Self-perceptions of status in face-to-face groups. J. Pers. Soc. Psychol. 91(6), 1094–1110. Anderson, C., John, O. P., and Keltner, D. (2008). ‘‘The Sense of Power.’’ Manuscript in preparation. Asch, S. E. (1951). ‘‘Effects of Group Pressure Upon the Modification and Distortion of Judgments.’’ Carnegie Press, Oxford, England. Aureli, F., and de Waal, F. B. M. (2000). ‘‘Natural Conflict Resolution.’’ University of California Press, Berkeley, CA. Barkow, J. H. (1975). Prestige and culture: A biosocial interpretation. Curr. Anthropol. 16(4), 553–572. Baumeister, R. F. (1998). The self. In ‘‘Handbook of Social Psychology’’ (D. T. Gilbert, S. T. Fiske, and G Lindzey, eds.), 4th edn., pp. 680–740. McGraw-Hill, New York. Baumeister, R. F., and Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychol. Bull. 117(3), 497–529. Beersma, B., and Van Kleef, G. A. (2008). ‘‘Why People Gossip: Antecedents, Consequences, and Social Functions.’’ Manuscript in preparation.
Reciprocal Influence and Power
187
Ben-Ze’ev, A. (1994). The vindication of gossip. In ‘‘Good Gossip’’ (R. Goodman and A. Ben-Ze’ev, eds.), pp. 11–24. University Press of Kansas, Lawrence, KS. Berger, J., Cohen, B. P., and Zelditch, M., Jr. (1972). Status characteristics and social interaction. Am. Sociol. Rev. 37(3), 241–255. Bernhardt, P. C. (1997). Influences of serotonin and testosterone in aggression and dominance: Convergence with social psychology. Curr. Dir. Psychol. Sci. 6(2), 44–48. Boehm, C. (1999). ‘‘Hierarchy in the Forest: The Evolution of Egalitarian Behavior.’’ Harvard University Press, Cambridge, MA. Bolger, N., and Schilling, E. A. (1991). Personality and the problems of everyday life: The role of neuroticism in exposure and reactivity to daily stressors. J. Pers. 59(3), 355–386. Borkenau, P., and Liebler, A. (1995). Observable attributes as manifestations and cues of personality and intelligence. J. Pers. 63(1), 1–25. Bourdieu, P. (1985). The social space and the genesis of groups. Theory Soc. 14(6), 723–744. Brauer, M., and Bourhis, R. Y. (2006). Social power. Eur. J. Soc. Psychol. 36(4), 601–616. Brewer, M. B. (2004). Taking the social origins of human nature seriously: Toward a more imperialist social psychology. Pers. Soc. Psychol. Rev. 8(2), 107–113. Brown, P., and Levinson, S. C. (1987). ‘‘Politeness: Some Universals in Language Usage.’’ Cambridge University Press, New York, NY, US. Bugental, D. M. (2000). Acquisition of the algorithms of social life: A domain based approach. Psychol. Bull. 126(2), 187–219. Buss, D. M. (1996). Social adaptation and five major factors of personality. In ‘‘The Five Factor Model of Personality’’ ( J. S. Wiggins, ed.), pp. 180–208. Guilford Press, New York, NY, US. Buss, D. M., and Craik, K. H. (1981). The act frequency analysis of interpersonal dispositions: Aloofness, gregariousness, dominance and submissiveness. J. Pers. 49, 175–192. Buss, D. M., Gomes, S., Higgins, D. S., and Lauterbach, K. (1987). Tactics of manipulation. J. Pers. Soc. Psychol. 52(6), 1219–1229. Caporael, L. R. (1997). The evolution of truly social cognition: The core configurations model. Pers. Soc. Psychol. Rev. 1(4), 276–298. Caporael, L. R., and Brewer, M. B. (1995). Hierarchical evolutionary theory: There is an alternative, and it’s not creationism. Psychol. Inq. 6(1), 31–34. Chance, M. R. A. (1967). Attention structure as the basis of primate rank order. Man 2, 503–518. Chartrand, T. L., and Bargh, J. A. (1999). The chameleon effect: The perception-behavior link and social interaction. J. Pers. Soc. Psychol. 76(6), 893–910. Chen, S., and Andersen, S. M. (1999). Relationships from the past in the present: Significantother representations and transference in interpersonal life. In ‘‘Advances in Experimental Social Psychology’’ (M. P. Zanna, ed.), Vol. 31, pp. 123–190. Academic Press, San Diego, CA. Chen, S., Lee-Chai, A. Y., and Bargh, J. A. (2001). Relationship orientation as a moderator of the effects of social power. J. Pers. Soc. Psychol. 80, 173–187. Chen, S., Ybarra, O., and Kiefer, A. K. (2004). Power and impression formation: The effects of power on the desire for morality and competence information. Soc. Cognition 22(4), 391–421. Chiao, J. Y. (2006). Building blocks to human social hierarchy: Psychological and neural investigations of social dominance perception. Dissertation Abstr. Int.: Section B: Sci. Eng. 67(5-B), 2851. Copeland, J. T. (1994). Prophecies of power: Motivational implications of social power for behavioral confirmation. J. Pers. Soc. Psychol. 67(2), 264–277. Craik, K. (2008). ‘‘Reputation.’’ Manuscript in preparation. Curhan, J. R., and Pentland, A. (2007). Thin slices of negotiation: Predicting outcomes from conversational dynamics within the first 5 minutes. J. Appl. Psychol. 92(3), 802–811.
188
Dacher Keltner et al.
Dabbs, J. M. (1997). Testosterone, smiling, and facial appearance. J. Nonverbal Behav. 21(1), 45–55. Dawson, E. C., and Gilovich, T. (2004). ‘‘Perceptions of Status in the Self and Others.’’ Working paper, Yale School of Management. De Dreu, C. K. W., and Van Kleef, G. A. (2004). The influence of power on the information search, impression formation, and demands in negotiation. J. Exp. Soc. Psychol. 40, 303–319. de Waal, F. B. M. (1989). ‘‘Peacemaking Among Primates.’’ Harvard University Press, Cambridge, MA. De´pret, E., and Fiske, S. T. (1999). Perceiving the powerful: Intriguing individuals versus threatening groups. J. Exp. Soc. Psychol. 35(5), 461–480. Domhoff, G. W. (1998). ‘‘Who Rules America.’’ Mayfield Publishing, Mountain View, CA. Dovidio, J. F., and Ellyson, S. L. (1982). Decoding visual dominance: Attributions of power based on relative percentages of looking while speaking and looking while listening. Soc. Psychol. Quart. 45(2), 106–113. Dunbar, N. E. (2004). Dyadic power theory: Constructing a communication-based theory of relational power. J. Fam. Commun. 4(3 and 4), 235–248. Dunning, D., Leuenberger, A., and Sherman, D. A. (1995). A new look at motivated inference: Are self-serving theories of success a product of motivational forces? J. Pers. Soc. Psychol. 69(1), 58–68. Eagly, A. H., and Johnson, B. T. (1990). Gender and leadership style: A meta-analysis. Psychol. Bull. 108(2), 233–256. Elias, N. (1978). ‘‘The History of Manners.’’ Pantheon Books, New York, NY. Emerson, R. M. (1962). Power dependence relations. Am. Sociol. Rev. 27(1), 30–41. Emler, N. (1994). Gossip, reputation, and social adaptation. In ‘‘Good Gossip’’ (R. F. Goodman and A. Ben-Ze’ev, eds.), pp. 117–138. University Press of Kansas, Wichita, KA. Erber, R., and Fiske, S. T. (1984). Outcome dependency and attention to inconsistent information. J. Pers. Soc. Psychol. 47(4), 709–726. Fiske, A. P. (1991). ‘‘Structures of Social Life: The Four Elementary Forms of Human Relations: Communal Sharing, Authority Ranking, Equality Matching, Market Pricing.’’ Free Press, New York, NY, US. Fiske, S. T. (1993). Controlling other people: The impact of power on stereotyping. Am. Psychol. 48(6), 621–628. Fiske, S. T., and De´pret, E. (1996). Control, interdependence, and power: Understanding social cognition in its social context. In ‘‘European Review of Social Psychology’’ (W. Stroebe and M. Hewstone, eds.), Vol. 7, pp. 31–61. Wiley, Chichester, UK. Foucault, M. (1977). ‘‘Discipline and Punishment.’’ Vintage Books, New York, NY. Frank, R. H. (1988). ‘‘Passions Within Reason: The Strategic Role of the Emotions.’’ W.W. Norton, New York. French, J. R. P., Jr., and Raven, B. (1959). ‘‘The Bases of Social Power.’’ University of Michigan, Oxford, England. Friedman, R., Anderson, C., Brett, J., Olekalns, M., Goates, N., and Lisco, C. C. (2004). The positive and negative effects of anger on dispute resolution: Evidence from electronically mediated disputes. J. Appl. Psychol. 89, 369–376. Funder, D. C. (1999). ‘‘Personality Judgment: A Realistic Approach to Person Perception.’’ Academic Press, San Diego, CA. Galinsky, A. D., Gruenfeld, D. H., and Magee, J. C. (2003). From power to action. J. Pers. Soc. Psychol. 85(3), 453–466. Gladue, B. A., Boechler, M., and McCaul, K. D. (1989). Hormonal response to competition in human males. Aggress. Behav. 15(6), 409–422.
Reciprocal Influence and Power
189
Goffman, E. (1957). On face-work: An analysis of ritual elements in social interaction. Psychiatry 18, 213–231. Goffman, E. (1967). ‘‘Interaction Ritual: Essays on Face-To-Face Behavior.’’ Anchor BooksGarden City, NY. Goodwin, S. A., Gubin, A., Fiske, S. T., and Yzerbyt, V. Y. (2000). Power can bias impression processes: Stereotyping subordinates by default and by design. Group Processes Intergroup Relations 3(3), 227–256. Guinote, A. (2007). Power and goal pursuit. Pers. Soc. Psychol. Bull. 30(8), 1076–1087. Guinote, A., Judd, C. M., and Brauer, M. (2002). Effects of power on perceived and objective group variability: Evidence that more powerful groups are more variable. J. Pers. Soc. Psychol. 82(5), 708–721. Guinote, A., Brown, M., and Fiske, S. T. (2006). Minority status decreases sense of control and increases interpretive processing. Soc. Cognition 24(2), 169–186. Haidt, J. (2001). The emotional dog and its rational tail: A social intuitionist approach to moral judgment. Psychol. Rev. 108(4), 814–834. Haidt, J., Koller, S. H., and Dias, M. G. (1993). Affect, culture, and morality, or is it wrong to eat your dog? J. Pers. Soc. Psychol. 65(4), 613–628. Hall, J. A., Coats, E. J., and LeBeau, L. S. (2005). Nonverbal behavior and the vertical dimension of social relations: A meta-analysis. Psychol. Bull. 131(6), 898–924. Hampson, S. E., Goldberg, L. R., and John, O. P. (1987). Category-breadth and socialdesirability values for 573 personality terms. Eur. J. Pers. 1(4), 241–258. Hart, D., and Edelstein, W. (1992). The relationship of self-understanding in childhood to social class, community type, and teacher-rated intellectual and social competence. J. Cross Cult. Psychol. 23(3), 353–365. Hecht, M. A., and LaFrance, M. (1998). License or obligation to smile: The effect of power and sex on amount and type of smiling. Pers. Soc. Psychol. Bull. 24(12), 1332–1342. Henley, N. M. (1973). Power, sex, and nonverbal communication. Berkeley J. Sociol. 18, 1–26. Henley, N., and LaFrance, M. (1984). Gender as culture: Difference and dominance in nonverbal behavior. In ‘‘Nonverbal Behavior: Perspectives, Applications, Intercultural Insights’’ (A. Wolfgang, ed.), pp. 351–371. C.J. Hogrefe, Inc., Lewiston, NY. Hrdy, S. (1999). ‘‘Mother Nature: A History of Mothers, Infants and Natural Selection.’’ Pantheon Books, New York. Insel, T. R., and Fernald, R. D. (2004). How the brain processes social information: Searching for the social brain. Annu. Rev. Neurosci. 27, 697–722. John, O. P., and Srivastava, S. (1999). The big five trait taxonomy: History, measurement, and theoretical perspectives. In ‘‘Handbook of personality: Theory and research’’ (L. Pervin and O. P. John, eds.), 2nd edn., pp. 102–139. Guilford Press, New York, NY. Jost, J. T. (2006). The end of the end of ideology. Am. Psychol. 61(7), 651–670. Jost, J. T., and Banaji, M. R. (1994). The role of stereotyping in system-justification and the production of false consciousness. Br. J. Soc. Psychol. Special Issue: Stereotypes: Struct. Funct. Process 33(1), 1–27. Jost, J. T., Glaser, J., Kruglanski, A. W., and Sulloway, F. J. (2003). Political conservatism as motivated social cognition. Psychol. Bull. 129(3), 339–375. Keltner, D. (1996). Facial expressions of emotion and personality. In ‘‘Handbook of Emotion, Aging, and the Lifecourse’’ (C. Malatesta-Magai and S. H. McFadden, eds.), pp. 385–402. Academic Press, New York. Keltner, D., and Anderson, C. (2000). Saving face for Darwin: The functions and uses of embarrassment. Curr. Dir. Psychol. Sci. 9(6), 187–192. Keltner, D., and Haidt, J. (2001). Social functions of emotions. In ‘‘Emotions: Current Issues and Future Directions’’ (T. Mayne and G. Bonanno, eds.), pp. 192–213. Guilford Press, New York.
190
Dacher Keltner et al.
Keltner, D., Ellsworth, P. C., and Edwards, K. (1993). Beyond simple pessimism: Effects of sadness and anger on social perception. J. Pers. Soc. Psychol. 64(5), 740–752. Keltner, D., Young, R. C., Heerey, E. A., Oemig, C., and Monarch, N. D. (1998). Teasing in hierarchical and intimate relations. J. Pers. Soc. Psychol. 75(5), 1231–1247. Keltner, D., Capps, L., Kring, A. M., Young, R. C., and Heerey, E. A. (2001). Just teasing: A conceptual analysis and empirical review. Psychol. Bull. 127(2), 229–248. Keltner, D., Gruenfeld, D. H., and Anderson, C. (2003). Power, approach, and inhibition. Psychol. Rev. 110(2), 265–284. Kipnis, D. (1972). Does power corrupt? J. Pers. Soc. Psychol. 24(1), 33–41. Kipnis, D. (1976). ‘‘The Powerholders.’’ University of Chicago Press, Chicago, IL. Kraus, M. W., and Keltner, D. (2008). ‘‘Socioeconomic Status, the Sense of Control, and Explanations of Life Outcomes.’’ Manuscript in preparation. Krebs, J. R., Davies, N. B., and Parr, J. (1993). ‘‘An Introduction to Behavioural Ecology’’, 3rd edn. Blackwell Scientific Publications, Cambridge, MA, US. Kruger, J., and Dunning, D. (1999). Unskilled and unaware of it: How difficulties in recognizing one’s own incompetence lead to inflated self-assessments. J. Pers. Soc. Psychol. 77(6), 1121–1134. Kyl-Heku, L. M., and Buss, D. M. (1996). Tactics as units of analysis in personality psychology: An illustration using tactics of hierarchy negotiation. Pers. Individ. Dif. 21(4), 497–517. Langner, C. A., and Keltner, D. (2008). Power and emotion. J. Exp. Soc. Psychol. Manuscript in preparation. Leary, M. R., Cottrell, C. A., and Phillips, M. (2001). Deconfounding the effects of dominance and social acceptance on self-esteem. J. Pers. Soc. Psychol. 81(5), 898–909. Lewin, K. (1951). In ‘‘Field Theory in Social Science: Selected Theoretical Papers’’ (D. Cartwright, ed.). Harpers, Oxford, England. Logli, M. A., Keltner, D., Campos, B., and Oveis, C. (2008). Gossip and reputation within social hierarchies. Manuscript in preparation. Magee, J. C., Galinsky, A. D., and Gruenfeld, D. H. (2007). Power, propensity to negotiate, and moving first in competitive interactions. Pers. Soc. Psychol. Bull. 33(2), 200–212. Mazur, A., and Booth, A. (1998). Testosterone and dominance in men. Behav. Brain Sci. 21(3), 353–397. McArthur, L. Z., and Apatow, K. (1983). Impressions of baby-faced adults. Soc. Cognition 2(4), 315–342. McArthur, L. Z., and Berry, D. S. (1987). Cross-cultural agreement in the perception of babyfaced adults. J. Cross Cult. Psychol. 18(2), 165–192. McClintock, M. K. (1971). Menstrual synchrony and suppression. Nature 229(5282), 244–245. Messick, D. M., Bloom, S., Boldizar, J. P., and Samuelson, C. D. (1985). Why we are fairer than others. J. Exp. Soc. Psychol. 21(5), 480–500. Milgram, S. (1963). Behavioral study of obedience. J. Abnorm. Soc. Psychol. 67(4), 371–378. Moskowitz, D. S. (1994). Cross-situational generality and the interpersonal circumplex. J. Pers. Soc. Psychol. 66(5), 921–933. Nemeth, C. J. (1986). Differential contributions of majority and minority influence. Psychol. Rev. 93(1), 23–32. Ng, S. H. (1980). ‘‘The Social Psychology of Power.’’ Academic Press, San Diego, CA. Overbeck, J. R., and Park, B. (2001). When power does not corrupt: Superior individuation processes among powerful perceivers. J. Pers. Soc. Psychol. 81(4), 549–565. Overbeck, J. R., and Park, B. (2006). Powerful perceivers, powerless objects: Flexibility of powerholders’ social attention. Organ. Behav. Hum. Decis. Process. 99(2), 227–243.
Reciprocal Influence and Power
191
Overbeck, J. R., Tiedens, L. Z., and Brion, S. (2006). The powerful want to, the powerless have to: Perceived constraint moderates causal attributions. Eur. J. Soc. Psychol. 36(4), 479–496. Owens, D. A., and Sutton, R. I. (2001). Status contests in meetings: Negotiating the informal order. In ‘‘Groups at Work: Theory and Research’’ (M. Turner, ed.), pp. 299–316. Erlbaum, Mahwah, NJ. Parker, G. A., and Rubenstein, D. I. (1981). Role assessment, reserve strategy, and acquisition of information in asymmetric animal conflicts. Anim. Behav. 29(1), 221–240. Preston, S. D., and de Waal, F. D. M. (2002). Empathy: Its ultimate and proximate bases. Behav. Brain Sci. 25, 1–20. Raven, B. H. (1999). Influence, power, religion, and the mechanisms of social control. J. Social Issues Special Issue: Social Influence and Social Power: Using Theory for Understanding Social Issues 55(1), 161–186. Ray, J. C., and Sapolsky, R. M. (1992). Styles of male social behavior and their endocrine correlates among high-ranking wild baboons. Am. J. Primatol. 28(4), 231–250. Reid, S. A., and Ng, S. H. (1999). Language, power, and intergroup relations. J. Social Issues Special Issue: Social Influence and Social Power: Using Theory for Understanding Social Issues 55(1), 119–139. Riggio, R. E., and Friedman, H. S. (1986). Impression formation: The role of expressive behavior. J. Pers. Soc. Psychol. 50(2), 421–427. Rusbult, C. E., Verette, J., Whitney, G. A., and Slovik, L. F. (1991). Accommodation processes in close relationships: Theory and preliminary empirical evidence. J. Pers. Soc. Psychol. 60(1), 53–78. Russell, B. (1938). ‘‘Power: A New Social Analysis.’’ Allen and Unwin, London. Sapolsky, R. M., and Ray, J. C. (1989). Styles of dominance and their endocrine correlates among wild olive baboons (Papio anubis). Am. J. Primatol. 18(1), 1–13. Savin-Williams, R. C. (1977). Dominance in a human adolescent group. Anim. Behav. 25(2), 400–406. Sidanius, J., and Pratto, F. (2001). Social dominance: An intergroup theory of social hierarchy and oppression. Am. J. Sociol. 106(5), 1436–1440. Sinaceur, M., and Tiedens, L. Z. (2006). Get mad and get more than even: When and why anger expression is effective in negotiations. J. Exp. Soc. Psychol. 42(3), 314–322. Snibbe, A. C., and Markus, H. R. (2005). You can’t always get what you want: Educational attainment, agency, and choice. J. Pers. Soc. Psychol. 88(4), 703–720. Solberg, M. E., Olweus, D., and Endresen, I. M. (2007). Bullies and victims at school: Are they the same pupils? Br. J. Educ. Psychol. 77(2), 441–464. Sulloway, F. J. (1996). ‘‘Born to Rebel: Birth Order, Family Dynamics, and Creative Lives.’’ Pantheon Books, New York, NY. Taylor, S. E., and Brown, J. D. (1988). Illusion and well-being: A social psychological perspective on mental health. Psychol. Bull. 103(2), 193–210. Tetlock, P. E. (1992). The impact of accountability on judgment and choice: Toward a social contingency model. In ‘‘Advances in Experimental Social Psychology’’ (M. P. Zanna, ed.), Vol. 25, pp. 331–376. Academic Press, New York. Thibaut, J. W., and Kelley, H. H. (1959). ‘‘The Social Psychology of Groups.’’ Wiley, Oxford, England. Tiedens, L. Z. (2001). Anger and advancement versus sadness and subjugation: The effect of negative emotion expressions on social status conferral. J. Pers. Soc. Psychol. 80(1), 86–94. Tiedens, L. Z., and Fragale, A. R. (2003). Power moves: Complementarity in dominant and submissive nonverbal behavior. J. Pers. Soc. Psychol. 84(3), 558–568. Tiedens, L. Z., Ellsworth, P. C., and Mesquita, B. (2000). Sentimental stereotypes: Emotional expectations for high- and low-status group members. Pers. Soc. Psychol. Bull. 26(7), 560–575.
192
Dacher Keltner et al.
Van Kleef, G. A., and Coˆte´, S. (2007). Expressing anger in conflict: When it helps and when it hurts. J. Appl. Psychol. 92(6), 1557–1569. Van Kleef, G. A., De Dreu, C. K. W., and Manstead, A. S. R. (2004). The interpersonal effects of emotions in negotiations: A motivated information processing approach. J. Pers. Soc. Psychol. 87, 510–528. Van Kleef, G. A., De Dreu, C. K. W., Pietroni, D., and Manstead, A. S. R. (2006). Power and emotion in negotiation: Power moderates the interpersonal effects of anger and happiness on concession making. Eur. J. Soc. Psychol. Special Issue Social Power 36, 557–581. Van Kleef, G. A., Oveis, C., Van der Lo¨we, I., and Keltner, D. (2008). ‘‘Power and Compassion: How Social Power Shapes Emotional Responses to the Suffering of Others.’’ Manuscript in preparation. Vescio, T. K., Snyder, M., and Butz, D. A. (2003). Power in stereotypically masculine domains: A social influence strategy X stereotype match model. J. Pers. Soc. Psychol. 85(6), 1062–1078. Vescio, T. K., Gervais, S. J., Snyder, M., and Hoover, A. (2005). Power and the creation of patronizing environments: The stereotype-based behaviors of the powerful and their effects on female performance in masculine domains. J. Pers. Soc. Psychol. 88(4), 658–672. Weber, M. (1947). In ‘‘The Theory of Social and Economic Organization’’ (A. M. Henderson and T. Parsons, trans.) Oxford University Press, New York. Whitmeyer, J. M. (2000). Effects of positive reputation systems. Soc. Sci. Res. 29(2), 188–207.
C H A P T E R
F O U R
Psychological Aspects of Retributive Justice Kevin M. Carlsmith* and John M. Darley† Contents 1. Retributive Justice, in Relation to Other Kinds of Justice 1.1. Other systems for harm-based justice 1.2. Retributive principles, criminal justice system practices, and policy issues 2. What Motivates the Desire to Punish? 2.1. Methodological issues 2.2. Verbal reports 2.3. Behavioral measures 2.4. Retribution 2.5. Utility 2.6. Specific deterrence 2.7. Incapacitation 2.8. General deterrence 2.9. Carlsmith (2006) 2.10. Darley et al. (2000) 2.11. Carlsmith et al. (2002) 2.12. Individual differences 2.13. Connections with previous research 3. The Impulse to Punish as an Intuition 3.1. The role of the reasoning system in moral judgments 4. Policy Implications 4.1. Evidence for discrepancies between legal codes and community intuitions 4.2. When do steps toward a crime cross the line to become a crime? 4.3. The ex-anti function: Do people know the criminal law? 5. Conclusion References
* {
194 195 196 197 197 198 198 199 200 200 200 201 203 204 205 206 209 211 213 218 219 221 223 233 235
Department of Psychology, Colgate University, Hamilton, NY 13346 Department of Psychology, Princeton University, Princeton, NJ 08540
Advances in Experimental Social Psychology, Volume 40 ISSN 0065-2601, DOI: 10.1016/S0065-2601(07)00004-4
#
2008 Elsevier Inc. All rights reserved.
193
194
Kevin M. Carlsmith and John M. Darley
Abstract Retributive justice is a system by which offenders are punished in proportion to the moral magnitude of their intentionally committed harms. This chapter lays out the emerging psychological principles that underlie citizens’ intuitions regarding punishment. We rely on experimental methods and conclude that intuitions of justice are broadly consistent with the principles of retributive justice, and therefore systematically deviate from principles of deterrence and other utilitarian-based systems of punishing wrongs. We examine the recent contributions of social-neuroscience to the topic and conclude that retributive punishment judgments normally stem from the more general intuitive-based judgment system. Particular circumstances can trigger the reasoning-based system, however, thus indicating that this is a dual process mechanism. Importantly, though, evidence suggests that both the intuitive and reasoning systems adhere to the principles of retribution. The empirical results of this research have clear policy implications. Converging evidence suggests that the formal U.S. justice system is becoming increasingly utilitarian in nature, but that citizen intuitions about justice continue to track retributive principles. The resulting divide leads people to lose respect for the law, which means that they do not rely on the law’s guidance in ambiguous situations where the morally correct behavior is unclear. These are the dangers to society from having justice policies based jointly on the contradictory principles of retribution and utility, and we lay out an argument for enacting public policies more exclusively based on retributive principles of justice. The field of psychological research on retributive justice, as compared to other kinds of justice, is of more recent empirical investigation and is correspondingly less well researched. Recently, though, it has attracted enough psychological interest so that once can glimpse the beginnings of a conceptual account on how people think about this sort of justice. Some psychological principles begin to emerge. In this chapter, we will summarize some of this research, with the customary emphasis on research conducted in our own laboratories, and drawing in work of others where relevant.
1. Retributive Justice, in Relation to Other Kinds of Justice Retributive justice is roughly the question of how people who have intentionally committed known, morally wrong actions that either directly or indirectly harm others, should be punished for their misdeeds. Retributive justice is sometimes regarded as part of distributive justice, because distributive justice is in turn concerned with what people deserve for their actions. To some extent, of course, this is a question of nomenclature, but the great bulk of distributive justice research has concerned itself with what one positively deserves, in terms of a share-out of the rewards produced by
Psychological Aspects of Retributive Justice
195
a group product. Distributive justice theories tend to assume positive outcomes and are concerned with whether those positive outcomes should be shared out equally, equitably in proportion to individual’s contributions to the group product, or on the basis of needs. In keeping with the standard conventions in this area, we will reserve the term ‘‘distributive justice’’ to discuss reward allocations, and use ‘‘retributive justice’’ to refer to punishments that people deserve for their wrongdoing. As we will illustrate, the principles that distribute punishments are quite different from those used to distribute rewards.
1.1. Other systems for harm-based justice There actually are several other systems for dealing with the perpetrators of harmful actions, and we need to position them in our account. We begin by discussing the Tort Law system. The defining characteristic of the harms that these other systems deal with is that the perpetrator does not inflict the harms in question intentionally on others. The steam railroads of the 19th century were notorious inflictors of harm: burning bits of coal billowing out of locomotive smokestacks set farmers’ fields on fire, and quite a large number of livestock were slaughtered by speeding trains at unfenced rail crossings (Friedman, 1985). Cattle ranchers also made their own contributions to the infliction of damages on nearby farmers as their cattle broke through fences into farmer’s fields and consumed or trampled remarkably large amounts of farm product, or bulls impregnated carefully bred dairy cows. Mills and factories inflicted injuries on workers, and at least occasionally it was realized that the workers were due compensation for the medical costs of these harms. But it was not the case that the railroad owners, cattle ranchers, or mill owners intended to inflict harms on those injured. The legal system that grew up to deal with these issues in the American context is the Tort Law system. Within that system, the United States’ treatment of these sorts of cases heavily relies on the concept of negligence. The thought here is that we all owe our neighbors a duty of care such that our activities do not inflict damages on them. If I have not acted up to a standard of care that a reasonable, prudent person would exhibit, and I am the proximate cause of the damage to you and your property, and the damage would not have occurred but for my imprudent actions, then I owe you compensation for the damages inflicted. The essential notion here is compensation, ‘‘making you whole’’ for the damages you have suffered. Similar issues arise in other situations such as medical malpractice, and liability for manufacturers for the harms caused by dangerous products they manufacture. That notion is to some extent similar to the notion of punishment, a topic to which we will return after we present
196
Kevin M. Carlsmith and John M. Darley
some of the psychological discoveries concerning what it is that people seek to do when they impose punishment.
1.2. Retributive principles, criminal justice system practices, and policy issues One task of the psychologist is to discover how ordinary people conceptualize retributive justice principles. This is a task that is interesting in its own right, but it is also a genuinely important contribution to public policy issues because citizens have some right to have their opinions registered about what behaviors will be targeted by the criminal justice system. At a minimum, these opinions should be available to criminal code drafters; some would make the claim that they have some right to be represented in the criminal codes. We will return to this question after we examine some of the discoveries that have come from the psychological research. In our research, the contours of the decision rules that citizens use in thinking about issues of culpability are often contrasted to the decision rules embedded in legal codes. One reason for this, of course, is the old but useful academic habit of having a ‘‘straw man’’ to spar with. However, as legal scholars have pointed out, what is called ‘‘common law’’ is likely to have a good many citizens’ views contained within it. This is because the common law, which was the base for much of the criminal law of the United States up to the first half of the 20th century, is based on judgments made by judges and jurists (stare decisis) as opposed to statutes passed by legislative bodies. As a result, the criminal law, the part of the law that deals with retributive justice, has been to a considerable extent made by citizens in their role of jurists. Looking back over those juridical decisions, one is struck by how often the judges explicitly appealed to common sense notions of justice thought to be held by all citizens, or at least all righteous citizens. And as we know, the common law tradition is embedded in the legal systems of the United Kingdom, and thus in many former British colonies, including the United States. Therefore, many elements of our foundational criminal doctrines incorporate strong elements of ‘‘common sense,’’ if common sense is construed as the way that citizens of a society think about justice issues. Therefore often when one examines criminal legal doctrines, one is doing it not to erect straw men to oppose, but to discern how people who had to think very hard about questions of culpability and responsibility organized those thoughts. There is also a case to be made for bilateral causality here. Our legal system commands respect from large groups of citizens, and thus is a credible source of what citizens’ moral principles should be (Tyler, 1990). Over the course of the many generations that people have been interacting with the legal system, the system has had the chance to influence citizens’ moral thinking through the vehicle of trials, films, courtroom dramas, and all the other ways that legal practice gets transferred to the popular
Psychological Aspects of Retributive Justice
197
imagination. Such claims as ‘‘I have a right to a trial by a jury of my peers’’ or ‘‘a man is innocent unless proven guilty,’’ which initially seem products of moral intuitions, are more likely to be products of socialization into our particular culture. Regardless of the origin, there are clear intuitions among citizens about what constitutes a just response to intentional violations of norms and laws. We turn next to these intuitions.
2. What Motivates the Desire to Punish? Even a casual observer of human behavior comes away with the indisputable conclusion that punishment is ubiquitous within human communities. It is not clear, however, whether punishment is an emergent property of moral development, or rather is an extension of the human proclivity for tools. That is, does punishment stem from violations of moral strictures, or is punishment a technique for social control? It is unlikely, of course, that a behavior as common and as important as punishment would depend solely upon a single motive. Reality almost certainly contains multiple motivational elements. Nevertheless, it is useful to identify the motives that dominate behavior. With regard to societal punishment, the rich traditions of moral philosophy, criminology, and legal theory provide a concrete starting point. Retributive justice, of course, plays a critical role in each of these areas, but so too does deterrence theory and incapacitation. We open this paper, then, seeking the origin of the desire to punish among individuals. When a person seeks to punish, what is he or she trying to achieve? Moral philosophers have long struggled with the justifications that permit a society to punish its members. The debate is often simplified into two mutually incompatible schools of thought. Kant (1952/1790) argued that perpetrators deserve to be punished in proportion to their ‘‘internal wickedness,’’ and that the imperative to punish derived not from the future consequences of the punishment, but rather from a universal goal of giving people what they deserve. By contrast, Bentham (1962/1843), representing a utilitarian approach to ethics, argued that punishment, and indeed all action, ought to be assessed by the potential harm or benefit to society. Any decision to punish individuals, then, must weigh the potential harms to the individual (and, thus, to society) against the benefits to society, and punishment is moral only if society stands to benefit.
2.1. Methodological issues Although the basic question of why citizens punish is straightforward, the route to an answer is not. There exist important methodological considerations, and variations in how the question is addressed have led to starkly
198
Kevin M. Carlsmith and John M. Darley
divergent conclusions. We will attempt to show that some of the confusion within the field stems from these different methodological approaches. There are two fundamental approaches to this question: one can ask people why they punish, or one can observe people as they punish and infer the underlying motive. We will refer to the first as verbal reports and the second as behavioral measures.
2.2. Verbal reports The advantage of verbal reports is that they are simple to measure, and they contain a relatively high degree of face validity. The problem with verbal reports is that they are frequently incorrect. Nisbett and Wilson (1977) first demonstrated this point in a series of experiments demonstrating that people had only limited insight into the motivations for their behavior. When asked why they made a particular choice, participants appeared to formulate an answer spontaneously and after-the-fact rather than introspecting and discovering the actual reason. Nisbett and Wilson concluded that although people do, indeed, have discrete motivations that follow clear rules, they have only limited insight into those motivations and are frequently unaware of them entirely, or are unable to articulate them accurately. Thus, verbal reports reveal little about the true internal states of the person, and are no more insightful about motivation than a stranger might be after observing the target. This general point holds true for punishment motivations as well, and is discussed in Carlsmith (2008) as well as later in this chapter.
2.3. Behavioral measures A second approach, and one that we tend to favor, is to largely ignore the reasons that people furnish and to focus instead on their behavioral responses to various situations. Our general strategy has been to ask citizens to recommend punishments for a wide variety of carefully calibrated harms, and to draw inferences about their underlying motives for punishment. This approach allows us to create situations that cry out for severe punishment from one philosophical perspective, yet little or no punishment from other perspectives. The citizens’ responses (e.g., recommended sentence) clarify which perspective guides their intuition. This approach, frequently described as a policy-capturing approach (Cooksey, 1996), is useful because it can help formulate public policies that will most closely align with people’s intuitions of justice. We will say more about the limits of this approach toward the end of the chapter. Our prototypical experimental design pits two perspectives against each other in a 2 2 arrangement. We create four versions of a vignette that describes an offense, and ask participants to recommend an appropriate punishment for one of those cases. For each perspective, perhaps
Psychological Aspects of Retributive Justice
199
Retribution and Deterrence, we create a moderate and severe case. That is, we modify the vignette so that it is either relevant or irrelevant to each perspective. For example, the theory of retribution is concerned with the moral culpability of the perpetrator, and thus the vignette is adjusted such that the moral culpability is either high or low. A retributivist will adjust his or her punishment accordingly. By contrast, a committed deterrence theorist would be relatively unmoved by this variable, and far more sensitive to a deterrence-related variable such as the frequency of the particular type of crime. Across subjects and vignettes, one can see patterns of response, and from those infer which philosophical perspective is most relevant to punishment. This information can then be used to shape public policy so that it better matches public intuitions of justice and fairness. The logic of these studies depends critically upon how these variables are instantiated, and work under the assumption that what is relevant for one perspective (e.g., moral culpability for ‘‘just deserts’’), is irrelevant for another (e.g., deterrence). This is not to say that the perspectives never overlap in the factors that they consider relevant, but rather that the experimenters must choose those factors that are unique to each perspective. Several of our earlier papers (Carlsmith, 2006; Carlsmith et al., 2002) devoted substantial effort to defining and justifying our choice of manipulations, and we will summarize those efforts here because they are essential to understanding the results.
2.4. Retribution Kant (1790/1952) best articulates the retributive justice stance by arguing that perpetrators ought to be punished in proportion to the moral offensiveness of their action. The goal of retributive justice is to give people their ‘‘just deserts,’’ and any future consequences of the punishment (e.g., utilitarian outcomes) are irrelevant to the assignment of punishment. For Kant, punishment needs no justification beyond the deservingness of the perpetrator. The critical variable for retribution, then, is the moral outrage generated by the crime. More serious crimes deserve more serious punishments. At the same time, though, factors that mitigate or exacerbate the morality of the action also come into play. Thus, holding the severity of the crime constant—say, theft of $100—one would punish less severely if the money was needed to feed a starving child, and more severely if the money was needed to create the world’s largest margarita. Similarly, any circumstance that mitigates responsibility for the action would also mitigate the moral severity of the offense. Thus, negligent actions ought to be punished less severely than intentional actions, and unforeseeable harms ought to be punished less severely than foreseeable harms. In short, a theory of retributive justice will be concerned with those factors that increase or decrease the
200
Kevin M. Carlsmith and John M. Darley
moral outrage that the action generates. The theory is focused solely upon the action and seeks to balance moral magnitude of the offense with the punishment. Those who cry, ‘‘let the punishment fit the crime’’ are probably operating from a retributive stance. Similarly, the biblical decree of ‘‘an eye for an eye’’ reflects retributive imperative that crimes and punishments ought to be commensurate. Modern views of retributive justice seek proportionality rather than absolute parity. Thus, more serious crimes receive more serious punishments, but the punishments, on average, could be either more or less severe than the offense.
2.5. Utility There are other justifications for punishment independent of retribution. Most notably, there is utility. One punishes an offender to reduce the likelihood of offenses in the future. This approach stems from the more general utilitarian movement in philosophy, in which all of one’s actions should be aimed towards improving the world. In brief, the morality of punishment rests solely upon whether it is likely to increase or decrease the total happiness in the world. Although punishment will almost certainly make the recipient less happy, it also has the potential to make society at large happier to the extent that it reduces future crime. If the ‘‘expected utility’’ of the proposed punishment is positive, then the punishment should take place. Within this general framework, there exist several versions of utility.
2.6. Specific deterrence Here the assignment of punishment is for the purpose of deterring the specific offender from reoffending when he has completed the sentence just imposed. One could imagine a complex individualized calculation scheme that would attempt to determine the sentence that would be sufficient to deter exactly this individual in the future, but it is usually assumed that the standard sentence for the crime will deter this criminal from repeating the offense.
2.7. Incapacitation Incapacitation, like specific deterrence, is designed to affect the behavior of the specific individual that has committed the crime, but does not do so through threat of punishment for future crimes. Instead, it renders any future offenses impossible, at least for a relatively long time period. This is accomplished most typically through incarceration in prison for a long duration sentence, although other means are also available. Sentences that
Psychological Aspects of Retributive Justice
201
include capital punishment (for murderers), castration (for rapists), severing a hand (for thieves), disbarment (for lawyers), branding a scarlet ‘‘A’’ (for adulterers), and deportation (for nearly all classes of criminals), often reflect the goals of incapacitation. The goal of an incapacitationist is to prevent known criminals from reoffending, and thus the single most important criteria in deciding whether to sentence is the likelihood of recidivism. This construct is frequently evaluated by examining the criminal’s prior record, his or her impulsiveness, and/or other actions that suggest imminent criminal behavior. When a person is deemed to represent a threat to the future lawfulness of society, that person is incapacitated. The inevitable tension generated by this approach is that society punishes not for the action committed, but for the probability that an action will be performed. Thus, when the theory operates perfectly, no crimes are committed, but not one person in jail has actually committed a crime.
2.8. General deterrence This approach punishes one person in order to deter others from committing a similar crime. When the punishment is levied against an appropriately guilty perpetrator, and the punishment is proportionate to the offense, then the penalty can serve the utilitarian function of general deterrence, but also of specific deterrence and retribution. In its pure form, however, general deterrence need not serve these other functions. Indeed, it often makes sense to punish a perpetrator with excessive and disproportionate severity so as to make a clearer point to potential thieves. Although this violates the retributive tenets of fairness, it is clearly in society’s interest to punish one person severely if it will prevent dozens, hundreds, or thousands of crimes from occurring in the future. Extending this logic, one can see that society might do well to punish celebrities more severely than noncelebrities. Likewise, according to this theory, if the punishment will receive no publicity (e.g., a private punishment, or a punishment that nobody notices) then the punishment is immoral because it creates unhappiness with no offsetting deterrent benefit. By contrast, if the recipient is famous and will receive extensive press coverage, then the punishment ought to be as severe as possible because of the tremendous potential deterrent effect. Taken to its logical extreme, philosophers have argued that the key features of retribution—severity of the crime, responsibility, mitigating factors, etc.—are entirely irrelevant to a theory of deterrence. In fact, even guilt is irrelevant. One can imagine circumstances in which punishing an innocent would have such tremendous societal benefit (e.g., forestalling an imminent riot that would otherwise kill hundreds of innocents), that utility would demand punishment of the innocent for the greater good. Realizing the perceived injustice of this, most who advocate a deterrent approach to sentencing hold
202
Kevin M. Carlsmith and John M. Darley
that it can be used only on those guilty of crimes, and they also generally limit the duration of the sentence that can be imposed to one that does not exceed the bounds of what is justly inflicted for the crime the offender committed. We have stated that the severity of the harm is relevant to retribution, yet irrelevant to utility. Readers may wonder whether severity of harm is truly orthogonal to utilitarian justifications of punishment. Intuitively it may seem that society has more need to deter serious crimes than petty crimes, and thus that utilitarians ought to carefully consider the danger posed by a particular crime. There are two responses to this. First, the effect of a given crime extends far beyond the initial victim, and includes all those who subsequently live in fear of future crime and those who become more likely to commit crimes as a result of witnessing an unpunished crime. Thus, the ‘‘cost’’ of punishment (borne by the offender) is almost certainly outweighed by the benefit to society, regardless of the punishment’s severity. Phrased differently, the cost to society for even minor crimes is so high, that a proportional response would always be severe. A utilitarian, however, would always limit the punishment to be just sufficient to deter future instances of the crime, and thus deservingness never enters the equation. Second, the logic of deterrence is that it will prevent future crimes merely by the threat of severe sanction. Indeed, if one believes in the efficacy of deterrence, then one need not be concerned about ‘‘overpunishing’’ minor crimes, because the threat of punishment alone will ensure that the punishment is never, in fact, carried out. We have argued, as have numerous philosophers, that retribution and utility cannot be effectively integrated and that logic dictates an ‘‘either/or’’ approach. This strikes many readers as overly simplistic because it is easy to think ‘‘I want both: I want to stop future offenses and I want the perpetrator to receive a fair punishment.’’ There is certainly nothing wrong with this sentiment, and holding these desires simultaneously is not illogical. Indeed, it is frequently the case that a punishment can serve both functions effectively. Nonetheless, the two justifications are not isomorphic and frequently diverge on appropriate sentences. It is in these situations that one must make a choice, and that one motive will trump the other. We pause here to note that there are other justifications derived from utilitarian theory that we have not studied and do not discuss in great detail. We omit them in large part because their primary supporters appear to us to be within academic circles rather than the citizenry. And because our primary focus in this paper and our research has been citizen intuitions of justice, they have been less important. These derivations include: rehabilitation, restoration, and restitution (but see pp. 208–209 for a discussion of restorative justice). Armed with the methodological and philosophical background, we turn now to some of our empirical findings.
Psychological Aspects of Retributive Justice
203
2.9. Carlsmith (2006) A logical place to begin the search for motives is to examine the cognitive processes that lead to a particular behavioral outcome. In this case, we1 looked at what sort of information participants recruited when they were required to make punishment decision. Borrowing from the method of ‘‘behavioral process tracing’’ ( Jacoby et al., 1987), Carlsmith created a paradigm that revealed whether participants were seeking information relevant to deterrence, incapacitation, or retribution when they were asked to assign criminal punishment. Participants began the experiment knowing only that a crime had been committed, and that they were responsible for recommending a sentence. The experimenter presented them with different categories of information about the crime, each of which was uniquely relevant to deterrence, incapacitation, or retribution. Thus, for example, participants could select ‘‘frequency of crime in society’’ (a deterrence-related category), ‘‘likelihood of repetition’’ (an incapacitation category), and seven other possible bits of information that could be readily classified into the three categories. Once selected, they learned a particular detail of the case from that category, such as ‘‘crimes of this sort are extremely rare,’’ or ‘‘experts report that the perpetrator is very likely to repeat this crime given the opportunity.’’ The dependent measure was a count of frequency with which each category was chosen, and the order in which they were chosen. In short, people could choose to learn different types of information, and their choices could inform us of the underlying theory that guided their behavior. Participants overwhelmingly chose the retributive items. On the first of five repetitions, 97% of participants selected a retributive piece of information. On the second and third repetitions, strong majorities (64% and 57%) also selected retributive items. As the retributive category ran out of content, people shifted their requests to the incapacitative category. Only when no other category was available did people seek out deterrence-related information. In a follow-up study, participants were randomly assigned information about the crime from each of these categories, and asked to assign a punishment and to report their confidence in the correctness of those assignments. We found, not surprisingly, that people became more confident as they obtained more information. Importantly, though, those gains were also dependent upon the category from which they came. Each piece of retribution-related information provided a significantly larger confidence boost than did information from either the deterrence or incapacitation categories. 1
The pronoun "we" is used because although it is a single authored paper, it derived from the Carlsmith’s doctoral dissertation and Darley was deeply involved in the design and execution of the work.
204
Kevin M. Carlsmith and John M. Darley
This study reveals that when people are placed in the position of assigning punishment, they seek out information relevant to a retributive perspective and not to one of the utilitarian perspectives. Moreover, when they are forced to use utilitarian information, they feel less confident in the accuracy of their assignments than when they receive information pertaining to the perpetrator’s deservingness of punishment.
2.10. Darley et al. (2000) The previous study showed which information people sought when punishing. This next study reveals people’s sensitivity to variations in the different types of information. Together, these studies reveal whether people seek to give perpetrators ‘‘what they deserve,’’ or whether they seek to punish solely for the purpose of preventing future crimes. Participants in this study read 10 vignettes that described various crimes ranging from quite minor (e.g., stealing music CDs from a store) to quite severe (e.g., political assassination). The vignettes also varied on the prior history of the perpetrator: he was either described as a first time offender with no history of misbehavior, or as a repeat offender who had committed similar crimes in the past. In this way, we manipulated the moral severity of the crime (a retributive factor) and the likelihood of future offenses (an incapacitative factor). The dependent variable was recommended sentence. The design of this study allowed us to examine which factors elicited strong punitive responses, and thus which factors mattered to the punishers. In this way, we inferred which theory of punishment best captured people’s actual behavior, and it was not necessary to have people articulate or endorse a particular theory. The results were quite clear. When it came to sentencing, people were highly sensitive to the severity of the offense, and largely ignored the likelihood that the person would offend again. Manipulation checks revealed that people clearly understood whether the perpetrators would be likely to reoffend, but that knowledge did not translate into sentence severity. Even for the most serious crimes that involved physical injury, the threat of future harm did not evoke incapacitative responses. Rather, people punished exclusively in proportion to the harm done and not for the harm that might be done in the future. To further clarify our results, we asked the participants to review the vignettes a second and a third time, and to do so from a retributive perspective and an incapacitative perspective. We explained to them in some detail how these theories differed from each other, the goals of the theories, and how they operated in general. We then asked them to assign punishment again from these perspectives. Our purpose was twofold: first, we could reassure ourselves that people understood each perspective well enough to behave as an incapacitationist if they so desired (and, thus, that
Psychological Aspects of Retributive Justice
205
our primary results were not merely the result of trying unsuccessfully to express their incapacitative desires), and second, to compare their ‘‘default’’ assignments to each of the ‘‘pure’’ perspectives, and thus more clearly reveal their goals of punishment. As we predicted, people had no difficulty with the retributive perspective. Their sentences tracked the moral severity of the offense closely, and they completely ignored the incapacitative factors. When they adopted the utilitarian perspective, however, the results were more complex. They became highly sensitive to the future risk of the offenders as we had predicted, but they did not ignore the moral severity of the offense as the theory dictates. Even when participants were instructed to ignore the retributive factors, the moral severity of the crime intruded on their sentencing and remained a significant predictor of the sentence. This failure reveals, we think, the importance of the just deserts rationale to the ordinary person. Finally, a comparison of the three adopted perspectives (default, retributive, and incapacitative) revealed that the first two were nearly indistinguishable from each other, and that both yielded responses quite different from the third. This finding, in conjunction with the earlier results, led to the conclusion that people spontaneously punish in a manner that is highly consistent with a theory of retributive justice and not in a manner consistent with the utilitarian goals of incapacitation.
2.11. Carlsmith et al. (2002) In a subsequent companion study, we tested the retributive motive against the general deterrence motive. In the first study, we asked over 300 university students to read a vignette of criminal wrongdoing and to provide a recommended sentence for the perpetrator. Across participants, we manipulated the moral severity of the harm and the need to generally deter the particular crime. To further generalize our findings, we instantiated each of the manipulations in several ways. For example, one vignette manipulated the retributive factor by increasing or decreasing the harm committed (e.g., embezzling modest amounts of money from an employer versus illegally dumping toxic chemicals near a town’s water table to avoid expensive disposal fees). In another vignette, the harm was kept the same but the morality of the action was varied (e.g., stolen funds were used to benefit underpaid factory workers at the company’s overseas plant, vs. using the stolen funds to finance a lavish lifestyle and extensive gambling debts). Again, the results were quite clear and mirrored our previous results. Regardless of how we instantiated the manipulations, people were highly sensitive to changes in the retributive factors, and ignored the deterrence factors. Figure 4.1 shows the path analysis conducted in Carlsmith et al. (2002),
206
Kevin M. Carlsmith and John M. Darley
Probability of non-detection Deterrence factors
0.82 0.68 Amount of publicity
Absence of mitigating circumstance Just deserts factors
0.68
0.31 0.50
0.25 0.41
0.55 Seriousness of offense
Moral outrage
0.30
Severity of punishment
0.27
Figure 4.1 Path model of predictors of punishment severity, N ¼ 329 (Carlsmith et al., 2002, Study 2).
and one can see quite clearly that severity of punishment is determined by the just deserts manipulation and mediated by moral outrage.
2.12. Individual differences In this study, we also measured individual differences to see if there are different types of people: those who punish for retribution and those who punish for deterrence. We explained the basic outlines of each theory to our participants, and asked them to choose the one that best described their personal perspective on punishment. We found that people varied widely on this question, and that it was quite easy to split people into three groups: retributivists, deterrists, and mixed perspectives. However, their actual behavior did not differ much across the three groups. Yet, those who selfidentified as deterrists were slightly more sensitive to the deterrence manipulation (and gave more severe sentences for crimes that needed more general deterrence), but the effect was extremely small, 2 ¼ .02. Thus, although individual differences mattered, they were easily dwarfed by the more universal sensitivity to the retributive factors. The third experiment within this paper further explored the disjunction between people’s stated goals of punishment and their actual behavior. As before, we asked people to self-identify themselves as retributivists or deterrists, and then had them evaluate one of our vignettes. People were again highly sensitive to retribution and ignored the deterrence manipulation, and individual differences only marginally affected their punitive responses. This time, however, we asked them to also engage in a resource
Psychological Aspects of Retributive Justice
207
allocation task. They were given the option to allocate city resources toward catching the perpetrators of the crime (an action consistent with the retributive goals of punishment), or toward preventing future crimes of this sort (an action consistent with deterrence). We found that people expressed strong support for preventing future occurrences of the crime, but that this desire was in no way related to the details of the crime. That is, people did not recommend more ‘‘prevention allocation’’ when the deterrence manipulation was high. Thus, what we find is that although people like and strongly endorse the outcome goals for utilitarian theories (e.g., reduced crime), they do not endorse the punishments that, according to the theories, would be required to achieve those goals. In other words, although people want punishment to reduce crime, they are not willing to incapacitate an individual merely because he or she is likely to commit a crime. Instead, they want to give a person what they deserve, and a person only deserves punishment after the fact. Likewise, people are unwilling to increase punishment when a sentence will be highly publicized, even though such an action would yield a greater deterrence benefit. Rather, people want the punishment to match the moral severity of the crime. In short, people want punishment to incapacitate and to deter, but their sense of justice requires sentences proportional to the moral severity of the crime (see also Carlsmith et al., 2007). The preceding discussion implies a fundamental conflict within many individuals: they want and expect punishment to serve utilitarian goals, but when faced with a particular case they consistently choose punishments that serve retributive goals. This, in turn, suggests that people do not have a good sense of their own motives for punishment, and as a consequence may endorse public policies that will violate their own intuitions of justice. This point was highlighted by Carlsmith (2008). The first experiment examined whether people could correctly articulate the motive that drove their decision to punish. Participants, who were broadly representative of the United States and other English speaking countries, evaluated a series of vignettes and assigned punishments in the usual manner. Each participant saw vignettes from each level of the retributive manipulation, and from each level of the utilitarian manipulation. By examining their pattern of responses to these vignettes, each participant’s sensitivity to the retributive and utilitarian factors was quantitatively assessed. Participants also self-identified their motive for punishing, both through single-item endorsements (e.g., ‘‘Which perspective is more important to you, deterrence or retribution’’), and through multi-item scales designed by other researchers for the same purpose (Clements et al., 1998; McKee and Feather, 2006). If people know why they punish and are internally consistent, then these two measures ought to be highly related. That is, the behavior and the attitude ought to correlate. But the results showed that these two constructs had no relation to each other. Indeed, the reasons that people generated for their
208
Kevin M. Carlsmith and John M. Darley
just-completed punishment bore no relationship to the pattern of their responses across the four vignettes and across multiple measurement constructs (all r s < .08, ns). The second study examined the consequences of this disjunction between what people say and what people do when it comes to punishment. We hypothesized, and found, that people will endorse utilitarian policies in the abstract not realizing that they will reject those same policies in practice because the policies violate retributive tenets of punishment. The inspiration for these studies came from the numerous observations of citizens enacting utilitarian legislation in one year, only to repeal that legislation shortly afterwards. California’s 3-strikes law is a case in point. Citizens overwhelmingly enacted this incapacitative law with a 70% majority in a statewide referendum. The law stated, simply, that after a third criminal conviction perpetrators go to jail for a very, very long time. What the citizens did not realize at the time, though, is that this law would deviate from their retributive intuitions. As a result of this law, offenders guilty of quite minor crimes (e.g., theft of a children’s video tape, or a Snickers Bar) faced sentences of 25 years to life. Once citizens witnessed this law in action, public support plummeted and surveys showed support to be under the 50% mark just a few years later. The second study of Carlsmith (2008) mimicked this phenomenon. It first described the problem that schools have had with student drug use, and asked people (again, broadly representative) to evaluate several different approaches to the problem. Although the majority preferred proportional responses to offenses that would be called for by a retributive approach, a substantial minority (30%) supported a zero-tolerance approach based on utilitarian theory. When these same people discovered that these zerotolerance rules violated proportionality, and thus their intuitive sense of fairness, support for the rules dropped precipitously to 6%. The paper concluded that people have only limited insight into the reasons for their punishment, and that this ignorance can lead them to endorse policies that, ironically, they will soon reject as unfair and unjust. One more research finding is relevant to the present argument. ‘‘Restorative justice’’ is a phrase that refers to a set of justice practices that many who are dissatisfied with the more punitive aspects of our criminal justice system find attractive. These practices focus on justice procedures that bring about repair of the harms done to crime victims, restoration of the offenders to a law-abiding life style, and involve the community in the process. The debate within the restorative justice movement concerns whether retributive punishments are also permissible within the system; many, following Braithwaite’s lead, hold that they are not. We find a great deal to admire in the ideals of the restorative justice community, but given what we have reported about the intensity of the just deserts punitive impulse, we doubt that citizens will be willing to allow a purely restorative,
Psychological Aspects of Retributive Justice
209
punishment-absent, treatment of at least reasonably serious crimes. In a scenario experiment (Gromet and Darley, 2006) testing this intuition, participants played the role of a judge assigning criminal cases to various court systems. One system was ‘‘purely restorative,’’ and it was made clear that the usual punitive options were not available. A second system, modeled on a procedure suggested by Barton (1999), allowed a mix of both restorative and punitive procedures. The third system was similar to the traditional court system in that prison sentences were the only corrective recourse. The findings were quite clear. For crimes of very low moral magnitudes (e.g., Halloween mischief ), about 80% of the respondents were willing to send the offender to the court that did not have punitive procedures at its disposal. However, for crimes of more moderate seriousness, just over 40% were sent to pure restorative courts, with a bit over 40% going to the mixed restorative/punitive procedure. For the serious crimes, such as attempted murder or rape only 10% sent to the pure restorative court, a bit over 65% sent to the mixed court, and about 25% sent to the court that had only punishments at its disposal. This was a small sample study, but if its results replicate on larger, more representative segments of the population, then the policy conclusion is that a majority of people will not tolerate a purely restorative justice system that does not allow for retributive punishments for moderately serious and more serious offenses. But if it is granted that there are actual gains for the victim and perhaps for the offender in restorative justice practices even if they take place in a context in which punitive sanctions are also likely to be imposed, then the ‘‘mixed’’ branch of the restorative justice system gains empirical support from these results. At the level of psychological theory, the conclusion might be that there is indeed a requirement that, when serious crimes are committed, punishment ensues; but most people would also allow for restorative practices as well. In a later experiment in the article, in which the respondents were told that the offender had successfully completed the victim-helping commitments he had made as part of the restorative agreement, across all the various offenses, they suggested imposing reduced prison sentences on the offender.
2.13. Connections with previous research Our work on punishment motives is derived from the philosophical debate articulated by Kant, Bentham, and others. However, that particular debate can be usefully traced back to Plato (cf. Republic) and Aristotle (cf. Nichomanchean Ethics), who discussed underlying principles for justice and fairness. Briefly, Aristotle believed that justice is achieved through a bottom-up process in which a series of fair decisions at the individual level inevitably and ultimately lead to a fair outcome for society. By contrast, Plato argued that fairness was best achieved by a top-down process
210
Kevin M. Carlsmith and John M. Darley
in which the fair outcomes for society were defined first, and individual cases were decided so as to achieve the broad outlines of justice. This distinction has been usefully updated and articulated by Brickman et al. (1981) in their discussion of micro- and macrojustice orientations. They suggest that people use different criteria for justice depending on which orientation they ascribe to, and that the criteria for each orientation inevitably conflict with each other. According to these authors, microjustice is concerned with the qualifying attributes of individuals. That is, it takes a myopic view of the individual without regard for the larger needs of society and examines the extent to which the particular individual deserves a particular outcome. By contrast, macrojustice is concerned with the appropriate order and well being of society. The deservingness of the individual takes a secondary role to that of society. Thus, microjustice is individuating whereas macrojustice is deindividuating. Brickman et al. (1981) discuss these two orientations with regard to reward allocation rather than punishment, but we think that their points are broadly applicable and can be usefully mapped onto our discussion of utility and deservingness, a mapping that Miller and Vidmar (1981) foreshadowed in their chapter on the Psychology of Punishment Reactions. Miller and Vidmar presented a 2 2 table in which the basic motives for punishment (behavior control and retribution) were crossed with the target of the punishment (offender vs. other in the social environment). A utilitarian mindset is triggered when one is oriented toward macrojustice, and a deservingness or retributive mindset is triggered when one is oriented towards microjustice. Brickman et al. (1981) suggest that the orientations are mutable, but that microjustice is more psychologically salient across situations. A basic social psychological question is raised by both Miller and Vidmar (1981) and Brickman et al. (1981) in their discussion of justice orientations. Namely, under what conditions might a person adopt one or the other orientation? We would suggest that a macrojustice orientation is produced in an individual when he or she is asked about whether punishments should be increased for a particular crime when the rate of that crime is perceived to be rising and thus threatens their sense of safety. The micro-orientation is triggered when a person is deciding what penalty is just for a person who has been convicted of a specific crime. Recent work on Temporal Construal Theory by Liberman and Trope (Liberman and Trope, 1998; Trope and Liberman, 2000) also provides some guidance on this question. Briefly, these researchers show that temporally distant situations elicit abstract outcome-based orientations whereas temporally proximate situations elicit the opposite. We suspect, but do not yet have data to support, that temporally proximate decisions guide individuals to microjustice concerns and thus to applying principles of
Psychological Aspects of Retributive Justice
211
deservingness and retribution. Temporally distant decisions guide individuals to macrojustice concerns focused on achieving a good outcome for society at large, and thus to applying principles of utility. These two approaches are consistent with our findings that people are intuitive retributivists. We have found scant evidence for utilitarian orientations in behavior, but have found substantial support for it in self-reported goals of punishment (Carlsmith, 2008; Carlsmith et al., 2002). It seems probable that when an individual is faced with a specific sentencing task, it is temporally proximate and this triggers a microjustice orientation. Accordingly, the individual is highly sensitive to the extent to which the perpetrator merits punishment and assigns a correspondingly proportional sentence. By contrast, when the punisher is asked to report their general goals for punishment or to recommend sentencing policies and guidelines, he or she is cued to think at higher, more abstract, levels and thus adopt a macrojustice orientation. This in turn leads the punisher to rely on principles and sentencing strategies that will achieve societal goals even at the expense of proportional outcomes for individual offenders.
3. The Impulse to Punish as an Intuition Increasingly, the evidence suggests that the decision to punish an individual is normally an intuitive rather than a completely reasoned decision. What does asserting that these judgments are intuitive mean here? In the last decades, judgment and decision-making researchers have demonstrated that people frequently make decisions via heuristic processes, in contrast to the conscious, deliberative processes that we think of as ‘‘reasoning processes.’’ Kahneman (2003) and others have suggested that it is important for psychologists to differentiate between judgments and decisions that are made by heuristic, intuitive methods and those made by conscious reasoning. The processes of the intuitive system are identical to those of the perception system, in that they are rapid, can occur in parallel with other thought processes, and are automatic. However, their outputs are similar to the outputs of the reasoning process, in that they are likely to be conceptual representations, judgments, or decisions. As is well-known, people experience their perceptions as simple correct representations of ‘‘what is out there’’; that is, people experience the perceptual world in the mode of a naı¨ve realist. In turn, intuitions, like perceptions, are often taken by ‘‘the intuitor’’ to be as unproblematically correct as perceivers take their perceptions to be. Thus, there are a set of intuition-produced decisions, choices, and problem solutions that are experienced as summaries of the ways
212
Kevin M. Carlsmith and John M. Darley
‘‘the world is,’’ because the person having the intuitions is unaware of the complex, potentially incorrect, cognitive processes that produced them.2 As we mentioned previously, when respondents receive a scenario in which some person commits a known morally wrong action, respondents experience a reaction of moral outrage that is a substantial predictor of the relative punishments that will be assigned to the perpetrator of the immoral action. We would suggest that this ‘‘feeling’’ of moral outrage is the conscious registration of the intuitive reaction to instances of moral wrongdoing. Kahneman (2003) comments that the assessment of the degree of ‘‘badness’’ represented in a stimulus is an intuitive system judgment: ‘‘Some attributes, which Tversky and Kahneman (1983) called natural assessments, are routinely and automatically registered by the perceptual system or by (the intuitive system) without intention or effort . . .. The evaluation of stimuli as good or bad is a particularly natural assessment. The evidence, both behavioral and neurophysiological, is consistent with the idea that the assessment of whether objects are good (and should be approached) or bad (and should be avoided) is carried out quickly and efficiently by specialized neural circuitry.’’ Several psychological researchers have also concluded that the sorts of punishment decisions that we have been considering are products of intuitive reasoning. They demonstrate that people report quick judgments that certain acts are morally wrong, or deserving of a certain level of punishment, but are unable to report the reasoning that led them to that decision. Haidt’s (2001) well-known work on moral dumbfounding demonstrates that people quickly report that, for instance, an act of brother–sister sexual intercourse is wrong. But when Haidt’s experimenter pushes subjects to supply reasons why it is wrong, they often cite reasons, for instance the psychic damage that could be done to one of the participants, or that a genetically defective baby could result. However, those reasons cannot be true, as the experimenter points out, because elements in the story told to the participants rule out those possibilities. Subjects often end, having run out of reasons why the act is wrong, asserting, ‘‘it is just wrong,’’ thus falling back on their intuitions. Other researchers, who have worked with what are called Trolley Car problems, report a similar effect. The core of the trolley car scenario, invented by Phillippa Foot, describes a runaway trolley car that, if it continues, will kill five people who are on the track farther down the incline.3 But this outcome can be mitigated if you, the respondent, throw
2
3
This section draws on work that JMD drafted for a forthcoming paper by Paul Robinson and John Darley (in preparation). In Foot’s version of the problem, the five persons "have been tied to the track by a mad philosopher." Foot was a philosopher, and allowed herself this in-joke. More recent researchers alter the story, and simply say that the five are, for instance, workers who will not see the run-away trolley soon enough to escape it.
Psychological Aspects of Retributive Justice
213
a switch to divert the trolley to another track, in which case it will kill only one person. In an alternative scenario, the respondent must take direct action against an innocent to save the imperiled workers: ‘‘As before, a trolley is hurtling down a track towards five people. You are on a bridge under which it will pass, and you can stop it by dropping a heavy weight in front of it. As it happens, there is a very fat man next to you—your only way to stop the trolley is to push him over the bridge and onto the track, killing him to save five. Should you proceed?’’ The utilitarian considerations are identical here in the two cases: killing one person to save many. However, most people report that throwing the switch would be the moral action and they would do it, yet most people report that pushing the person onto the tracks would be immoral and they would not do it. When one set of researchers (Hauser et al., 2007), who had administered both scenarios to the same respondents, (a within subjects design), confronted their respondents with their discrepant decisions about what seemed a morally identical case, about 70% of the respondents could not give an explanation of why they differed on the two cases, echoing Haidt’s moral dumbfounding work. Both sets of researchers assert that these judgments are intuitive ones: Haidt (2001, p. 814) concludes that moral judgments derive from ‘‘quick automatic evaluations (intuitions),’’ and Hauser et al. (2007, p. 125) suggest, ‘‘much of our knowledge of morality is . . . intuitive, based on unconscious and inaccessible principles.’’
3.1. The role of the reasoning system in moral judgments Provisionally, let us accept that it is often, perhaps always, the case that a person’s first reactions to a moral violation are intuitive in nature, and largely driven by just deserts concerns. The next task is to examine the role of the reasoning system in the final decisions. Kahneman sets the question: ‘‘In the context of an analysis of accessibility, the question of when intuitive judgments will be corrected is naturally rephrased: When will corrective thoughts be sufficiently accessible to intervene in the judgment?’’ (2003, p. 711). As this suggests, it is possible for a person to learn to apply reasoning processes that can correct and override the conclusions presented by the intuitive system. As an example, many statistically trained psychologists have done so in the area of probability updating. Specifically they have learned that, relying on intuitions, they do a bad job of probability updating, and so know that they need to process the information through Bayes equations to come to the correct answer. However, statistically sophisticated researchers do not always draw on their knowledge when they should. In one of their earliest studies, Tversky and Kahneman (1971) demonstrated that
214
Kevin M. Carlsmith and John M. Darley
statistically trained social scientists often failed to do this. As Kahneman comments (2003, p. 697), ‘‘remarkably, the intuitive judgments of these experts did not conform to statistical principles with which they were thoroughly familiar . . . We were impressed by the persistence of discrepancies between statistical intuition and statistical knowledge, which we observed both in ourselves and in our colleagues. We were also impressed by the fact that significant research decisions, such as the choice of sample size for an experiment, are routinely guided by the flawed intuitions of people who know better.’’ He goes on to conclude: ‘‘People are not accustomed to thinking hard and are often content to trust a plausible judgment that quickly comes to mind’’ (Kahneman, 2003, p. 699). There are reasons for this. Intuitive judgments, like perceptual judgments, do quickly come to mind, and the fact that they are products of interpretive processes is not apparent to the person making the judgments. Because they are rapid, automatic judgments, they leave no reminders that they are interpretations, and thus provide no cues to the person that they might be incorrect. The intuitive system processes are like those used in our visual perception system, and psychologists have recognized that people are ‘‘naı¨ve realists’’ with respect to what they perceive. They give percepts an ‘‘it couldn’t be otherwise’’ character. We are suggesting that the products of the intuitive system have a similar character. Still, the reasoning system can be drawn on to override the conclusions of the intuitive system. As this suggests, the intuitive and reasoning system distinction creates yet another dual process theory in psychology in the area of human moral judgments. The question for those attempting a description of moral reasoning then becomes the question that faces all dual process theorists, which is to give some specification of the conditions that toggles the processor between experiencing intuitions and acting on them, as opposed to bringing reasoning processes to bear and following the reasoned conclusions, overriding the intuitions. For this article, a great deal rides on the answer to this question. If it is true that a person’s initial response to an incident of wrong doing is an intuitive, just desertsdriven one, yet this initial response is always or often overridden by a more reason-based decision, then we should quit analyzing the intuitive moral system and begin understanding the principles of the reason-based account. We will argue that the kinds of moral reasoning that leads to the punishment judgments that comprise what we are attempting to model is almost always just deserts based. The first part of the argument is that moral reasoning often stops with the intuitive system result. This intuitive reaction is generally rapidly, automatically, and nonoptionally produced. And for the reasons we just mentioned, these intuitive reactions have a ‘‘truthy’’ character that makes them likely to be acted on without scrutiny.
Psychological Aspects of Retributive Justice
215
There is a second reason why intuition often prevails. Moral judgments are often made by persons who are under what we now call ‘‘cognitive load.’’ Kahneman comments that: As in several other dual-process models, one of the functions of System 2 (the reasoning system) is to monitor the quality of both mental operations and overt behavior (citations omitted). As expected for an effortful operation, the self-monitoring function is susceptible to dual-task interference. People who are occupied by a demanding mental activity (e.g., attempting to hold in mind several digits) are more likely to respond to another task by blurting out whatever comes to mind. (citation omitted)
To sum, punishment reactions are the product of a dual-process system in which the retributive desire is automatic, and the reasoning process that might override it is only selectively brought online. But even if the reasoning process were triggered into action, would reasoning always override intuition? Previously we suggested that the intuitions that were driving reactions of moral outrage and relative severity of sentence assigned to crime vignettes were based on just deserts considerations. Is it the case that these just desert sentencing intuitions would be overridden if more deliberative reasoning processes were applied to the cases? In other words, is it the case that people will display more complex thinking about punishments for morally wrong actions if they were caused to reason about the cases? Two studies we reviewed previously suggest this is not the case, and that reasoning about punishments is often done from a just deserts perspective. First, in Carlsmith (2006) respondents were led to engage a reasoning process; they acquired an item of information that they selected, thought through how that affected their assessment of the appropriate punishment, and then chose another item to further refine their judgment. Results showed that, overwhelmingly, respondents acquired items giving themselves retributive information to inform their judgments. Thus, their conscious reasoning system sought retributive information as the relevant decision inputs. This is quite important. It suggests that the study respondents are not motivated to deploy a reasoning process that deviated from a just deserts-based approach when they had the chance to do so. Second, recall that in Darley et al. (2000), participants were instructed to take either a retributive or utilitarian stance before assigning punishments. Participants had no trouble taking a retributive, just deserts stance, and the punishments they assigned while taking that stance did not differ from the punishments they assigned previously, when they were free to assign whatever punishments they thought appropriate. However, even when asked to punish according to a utilitarian stance, the participants were not able completely to do so. Their punishment patterns showed considerable reliance on just deserts, retributive considerations. From these two studies, we conclude that although it is undoubtedly the case that people are capable of reasoning about appropriate punishment
216
Kevin M. Carlsmith and John M. Darley
assignments, these reasoning processes do not routinely contradict the retributive, desert-based judgments that are supplied by the intuitive system. In fact, reasoning processes often also seem to be driven by just desert considerations. We next want to present brain-imaging research, but want to be quite clear on what it adds to the case we are presenting. First, and most impressively, it supports the claim that moral reasoning can involve dual processes, and adds the possibility that these dual processes take place in somewhat different brain locations. Second, it supports the idea that the first process is rapid, one that is generally computed when the moral case involves an individual committing direct harm to another. We interpret this to mean that it is an intuitive system product. There is also a slower process, involving brain areas associated with higher order reasoning. This slower process can sometimes override the first process decision. These dual process results come from two neural imaging studies conducted by Joshua Greene and his collaborators (Greene et al., 2001, 2004). In these studies, the respondent was told a story, and the story ended with an action that the protagonist could take. The task of the respondent was to decide whether the action was inappropriate or appropriate. Concretely, the respondent might be told the trolley problem in which the protagonist threw the switch to save five and kill one. In the first study, using trolley car and similar other problems, they found that a set of problems they call ‘‘personal violations’’ activated brain regions that previous research has associated with both emotion and social cognition activities. Other, ‘‘impersonal moral dilemmas’’ activated increased activity in areas associated with abstract reasoning and problem solving. The trolley car problem that posed the choice of letting the trolley continue on its tracks and killing five persons versus throwing a switch to shunt the trolley to a track on which it would kill only one, was an impersonal moral dilemma. It did not engage moral reasoning areas. Our previous ‘‘footbridge’’ example, in which the possible trolley-stopping action was pushing a specific other individual in the path of the oncoming trolley car, is a personal violation case: the action required is a direct personal one in which ‘‘I act directly to harm a specific other person.’’ These personal violation cases generally drew quick reaction time decisions that the action in question was an inappropriate one to do and involved heightened brain activity in the emotion and social cognition areas (specifically, the medial prefrontal cortex, posterior cingulate/precuneus, and superior temporal sulcus/temperoparietal junction). Most of the personal moral dilemmas were of a simple character, for instance, should a teenage mother kill her unwanted newborn infant. These were the cases decided rapidly, and overwhelmingly, the proposed response was judged ‘‘inappropriate.’’
Psychological Aspects of Retributive Justice
217
But a few cases elicited quite a different pattern of results. Here is one such case. Enemy soldiers have taken over your village. They have orders to kill all remaining civilians. You and some of your townspeople have sought refuge in the cellar of a large house. Outside you hear the voices of who have come to search the house for valuables. Your baby begins to cry loudly. You cover his mouth to block the sound. If you remove your hand from his mouth, his crying will summon the attention of the soldiers who will kill you, your child, and the others hiding out in the cellar. To save yourself and the others, you must smother your child to death. Is it appropriate for you to smother your child in order to save yourself and the other townspeople?
Some respondents reported that in this case the personal violation action in question was appropriate rather than inappropriate. Generally those respondents realized that smothering the baby, although a horrible thing to do, was required to save the group and decided the action was appropriate. Our guess is that it was the final sentence of the scenario, ‘‘smother . . . in order to save . . .’’ that triggered the reasoning system into action. Other subjects thought for a long time, but decided the action was still inappropriate. (Both groups could be matched on reaction time.) Researchers interpreted the associated brain activation patterns as evidence that when participants responded in a utilitarian manner ( judging personal moral violations to be acceptable when they serve a greater good) as reflecting both the involvement of abstract reasoning centers (regions in the dorsolateral prefrontal cortex) and also the engagement of cognitive control (specifically the anterior cingulate cortex, a brain region associated with cognitive conflict) in order to overcome the social-emotional responses elicited by the thought of smothering the baby. On this account, dual processes make moral judgments: one process that is relatively rapidly produced is the product of social cognitive and emotional responses, and takes place nonoptionally. This is the intuitive system we discussed earlier. The second process involves abstract reasoning areas of the brain, ones that developed evolutionarily later than did the social cognitive and emotional brain areas, and is not always triggered into action. This, we suggest, is the reasoning system. Further, when this latter system is activated, its results are sometimes in conflict with the intuitions of the other system. The conflict is perhaps resolved by some assessment of the competing strengths of the two sets of signals. So the reasoning system that here is giving us the utilitarian result and overriding the impulse against killing a single other individual, is acting to monitor or limit the intuitive system result. It is this ‘‘override’’ response that is possible, but that we suggest is rare for the punitive judgments we are researching.
218
Kevin M. Carlsmith and John M. Darley
4. Policy Implications The previously reviewed results seem to us to have some policy implications for various aspects of our justice system, and we now turn to what those might be. We argue that when people seek to follow the law, their impressions of what the law is, what the law allows or prohibits, is drawn less from their knowledge of actual laws than it is from their intuitions of what the law must be. And those intuitions are read off from their moral intuitions. That is, people tend to think that the legal codes are in accordance with their own moral intuitions. One reason that this is so is that these intuitions that give people their take on what the law holds come to mind automatically, and this gives the intuitions a character of certainty that resembles the certainty that people give to the products of their perceptual system. This in turn means that, if legal codes differ markedly from citizens’ moral intuitions, then the legal system needs to make the considerable efforts necessary to inform citizens about what the law should be, and persuade them of the moral correctness of the court-held laws. Otherwise citizens will not agree with the criminal codes, and as we will demonstrate later, this will have negative consequences for their willingness to voluntarily obey the law. This persuasion can be done, but it is difficult. It entails either engaging the reasoning system to override the intuitive system, or altering the intuitive processes to yield the desired answer. We suspect that both of these routes would be difficult, but little has been written on the topic because scholars have only recently conceptualized intuition and reasoning as dual process systems (cf. Oswald and Stucki, 2008). We will make a few preliminary observations. For a reasoning system to override an intuitive judgment or decision, the individual must recognize that the intuition comes from a class of judgments that the individual has learned should be overridden. The question, then, is what percept triggers this override and brings the specific reasoning principles into operation. Sometimes a cue to engage the reasoning system comes from the form of the problem itself. One of us has (finally) learned to recognize when the form of a statistical question involves updating prior probabilities based on new information. This realization cues us that Bayesian updating calculations solve the problem, and further, that his intuitions about these calculations are generally wildly wrong. Reasoning ensues; intuition is overridden. For this process to occur, however, the individual must recognize the fallibility of intuition, and this requirement may be particularly difficult when it involves the appropriate punishment for some specific case of wrongdoing. Those intuitions come with powerful associated emotions, and those emotions exert influence on both the reasoning and intuitive systems of judgment. Moreover, as we will show later in this paper, both
Psychological Aspects of Retributive Justice
219
intuitions about punishment and reasoning about punishment appear to draw on principles of just desert. Thus, it is not clear that a person is likely to switch systems, or that switching systems are likely to alter the outcome. The alternate process would involve changing the intuitive reaction. The nearest process to this that psychology is familiar with is emotional conditioning. The task here is to condition emotional reactions to certain stimuli. One common suggestion in the moral judgment literature is that emotions, such as anger, disgust, and contempt (that are the ones linked to moral judgments of the sort we are discussing), can be conditioned to respond—or not to respond—to a given stimulus. An experiment by Wheatley and Haidt (2005, experiment 2) has a condition that is wonderfully on point for our purposes. They showed highly hypnotizable respondents stories with no morally transgressive actions and that elicited no strong reactions from control groups. For subjects who were hypnotized to experience disgust at trigger words in the story, however, the story reliably elicited ratings that the actions were morally wrong! If conditioning generally produces the emotional components of reactions to moral transgressions, then an insight about moral learning follows. As we know, emotional reactions to situations are often learned with few trials. This could explain how it is that children in a particular culture rapidly learn what counts as ‘‘wrongdoing’’ in their cultures. They do so because they experience the emotional reactions of adults as they witness the wrongdoing actions. Sometimes, of course, the task of the person is to override or cancel the intuition that a particular action, which currently triggers the moral emotional response, is a wrong action. The emotional conditioning literature suggests, however, that eliminating the emotional component of the response is not an easy task. Habituation, such as happens for medical students dissecting corpses, may eventually come about, but via the dual process avenue we have suggested. The habituation may not be the disappearance of the emotional component of the response, but rather its weakening. This weakening allows it to be overridden by processes of reasoning that provide justification for the hitherto transgressive action.
4.1. Evidence for discrepancies between legal codes and community intuitions In 1995, Paul Robinson4 and John Darley published Justice, Liability and Blame: Community Views and the Criminal Law. They report the results of 18 studies that searched for discrepancies between legal codes and 4
Paul Robinson is the Colin S. Diver Professor of Law at the University of Pennsylvania School of Law. His area of specialty is in criminal law, and particularly the empirical justifications for the doctrines from which criminal law is generated.
220
Kevin M. Carlsmith and John M. Darley
community views on what those codes should be. The method of the studies was standardized. Scenarios were created describing various actions that were possibly criminal in nature. It was the task of the respondent to assign whatever penalty he or she thought appropriate to the action. Respondents could decide that the action generated no liability—in other words the actor did not do anything that was criminal, or that the action generated liability and the condemnation associated with that, but did not require a prison sentence, or could assign a prison sentence that ranged in ascending steps from one day in prison through life in prison to the death sentence. The designs were usually within-subject designs, in that each respondent assigned penalties to a number of different scenarios. The within subject designs meant that subjects were made aware that they were being asked if they thought that the presented cases should be treated differently. (Although they were told that they would sometimes receive cases, to which it would be perfectly appropriate to assign the same penalties.) This we thought was the correct methodology, because we were interested in whether subjects’ judgments about cases differed under conditions of conscious comparison. The scenarios were chosen to fit the definitions of crimes that were generated from the relevant laws (Robinson and Darley, 1995, Study 5, scenarios 2, 4.). The issue investigated in this study was the correspondence between the legal rules about what constituted appropriate use of violence in self-defense and the citizens’ moral intuitions about this. The first scenario describes what the criminal code regards as a legitimate act of self-defense because the person cannot retreat from the threat. However, in the second scenario, shooting the other is not legitimate, because the threatened person is aware he could retreat and the codes generally demand that the person retreat when it is possible to do so. More precisely, according to the law, the first shooting is legitimate self-defense and receives no penalty but the second case is murder and should receive a high penalty.5 In fact the average liability assigned by our respondents to the second scenario was between 6 months and a year in prison and this was all that was assigned even though the respondents agreed with the statement that the person could have safely retreated. Further, 23% of the respondents assigned no liability and an additional 17% thought that the shooter should receive the censure of conviction, but serve no prison time. ‘‘Americans stereotypically ‘stand their ground’ and our subjects seem to want them to even when legal codes say they should not’’ (Robinson and Darley, 1995, p. 60). There has been a tendency for codes to shift toward approving this; several states have weakened the requirement to retreat when faced with a threat of violence.
5
This is a fair characterization of the law as it was stated at that time. Since then some states have changed their laws in the direction of not requiring retreat.
Psychological Aspects of Retributive Justice
221
4.2. When do steps toward a crime cross the line to become a crime? Legal codes have needed to consider at what point a person’s actions toward committing a crime become the crime of attempt and what the sentence should be for attempted crimes. Interestingly, the criminal codes have changed their stance on this question, and done so in a way that seemed to modern code drafters to make psychological sense. Some brief history of the criminal law of the United States is needed here. The phrase, ‘‘common law,’’ used in the context of the law of the United States, refers to a patchwork of laws originally inherited from England in the period when the United States was a colony of England, and then modified by rulings made by judges or legislatively imposed by state congresses over the years. The criminal codes in effect in the United States until the middle of the 20th century were common law products. In its treatment of attempt, the common law focused on how close the actor had come to completing the offense—his proximity to the offense, as it was called. The usual test for this was ‘‘dangerous proximity,’’ a test that would be fulfilled, for instance, if a burglar were on the property of a house he intended to rob. In the 1960s, it was increasingly recognized that the common law criminal codes sometimes contained provisions that seemed odd, archaic, unjust, or contradictory. Further, criminal codes differed significantly by state. All of this seemed irrational and unjust to the prestigious American Legal Institute who spent some years designing the ‘‘Model Penal Code.’’ The code rationalized and systematized the chaotic patchwork of common law criminal codes. The orienting assumption that the code drafters took on was to create the code that a reasoning person in the modern era would judge that the penal code should be. Their efforts were successful: The model code they drafted has been adopted in whole or in part by 37 states. The model penal code took a very different stance on defining criminal attempts than did the older common law codes. For example, the common law generated some counterintuitive results in its treatment of ‘‘impossible attempts.’’ It allowed a defense if an actor bought goods believing them to be stolen goods but it later turned out to be the case that they were not actually stolen. This counter-intuitive result came about because if the goods were not stolen, then it was impossible for the purchaser to buy stolen goods. The model penal code drafters found that absurd. What matters is that the actor believed that they were stolen, and thus what he did was buy what he thought were stolen goods. The model penal code drafters held that the actor’s liability was determined from the point of view of the actor, with the circumstances as the actor believed them to be. So he was liable for the offense of attempting to buy stolen goods.
222
Kevin M. Carlsmith and John M. Darley
Reasoning from this, the model penal code drafters thought through the question of what conduct counted as an attempted crime. They reasoned that the essence, the gravamen, of the offense, was that the actor had formed the settled intent to commit the crime. More specifically, they looked at how far the actor had moved toward the offensive conduct, and held that the commission of a substantial step toward the commission of the criminal offense was the key test. It showed that the actor had a willingness to commit the offense. This is often called the subjectivist stance of criminalization, because it focuses on actors’ subjective culpability in attempting to bring about the criminal action. Coming close to bringing about the harm or evil of the offense was not required. All that was required was that the actor had progressed far enough to demonstrate a fully formed intention to commit the crime, and that was what taking a substantial step toward commission of the crime showed. This subjectivist view gives little weight to whether the crime was committed or not, because that does not alter the actor’s culpability. Therefore, the Model Penal Code had an easy and consistent answer to what the penalty should be for the crime of attempted X; it should be equivalent to the penalty for the crime of committing X. The older common law, focused as it was on the actual occurrence of the harm as the offense, habitually gave some sentence discount to actors even when they were discovered in dangerous proximity to the crime, because the harm had not occurred. (Different offenses had different specified discounts for attempts, further depending on jurisdictions.) Thinking about these formulations, we thought that the Model Penal Code drafters might have gone too far in the subjectivist direction from the point of view of the citizens. So in a study of citizens’ perceptions of the appropriate penalty for attempt offenses (Robinson and Darley, 1998), various aspects of the difference between the subjective and the objective perspectives were tested. In the first test, it was found that for the case of impossible attempts, respondents primarily agreed with the subjectivist perspective. The actor who smuggled talcum powder back from Columbia because he had been tricked into thinking it was cocaine received a penalty only slightly less that the actor that smuggled real cocaine, rather than the zero penalty that the objectivist stance would assign. But in a second set of tests, the pattern of results matched the pattern that would be given by an old-fashioned common law objectivist. The completed crime scenario was the murder of a prosecutor who had prosecuted a family member of the murderer. For this, the murderer received a life sentence, with a few respondents giving the death sentence. The second scenario involved the same actor, now a would-be murderer, who shot at the prosecutor from close range but missed. This is known in criminal law circles as a completed attempt. He received a considerable sentence
Psychological Aspects of Retributive Justice
223
reduction to the successful murder—just above 15 years in prison versus the life sentence given to the murderer. In yet other scenarios built around the same core, the actor fully intended to kill the prosecutor, but was halted even farther from the completion of the act. Sentences were correspondingly reduced. Further, as accomplices were introduced into the scenario, the actor’s sentences were again reduced. It seemed to us that the respondents were assigning punishment graded in accordance with their notions of the actor’s psychological distance from the murder act. (For an attempt to model this distance in terms of additive discounts from the prototype offense, see Robinson and Darley, 1998, pp. 436–440.) What this tells us is that the respondents held an objectivist view of how offenses and attempted offenses should be graded—and they feel that graded offenses should have calibrated punishments assigned to them. But our respondents were staunchly subjectivist in punishing those who completed attempts at, for instance, smuggling when they had been tricked into smuggling talcum powder rather than cocaine. As the authors (Robinson and Darley, 1998, pp. 429–430) comment: When we consider what our respondents took to be the minimum requirements for having committed a crime, the subjectivist perspective fares better. Even when the actions taken ended far away from the successful commission of the crime, and the actor assisted in the steps toward the crime rather than was the intended principal, substantial sentences were assigned. The respondents criminalized and punished many actions than a strict objectivist would if the objectivist held to the rule of no liability is to be assigned short of actions being committed that were ‘‘in dangerous proximity to the offense.’’ But, and again, they did not punish these offenses as if they were equivalent to the completed offense.
It is possible to speculate that, in our modern society, in which dangers can be inflicted at long range, and lethal weapons enable preparations to turn quickly to completions, the point at which people’s thinking about the ‘‘lines’’ people should not cross has moved further away from the completion of a dangerous action, toward not starting the action. If so, we would expect criteria for when actions can be taken in self-defense to move comparably. We did notice that the respondents who defended themselves and killed when they could have retreated did not receive the murder sentence that the law calls for.
4.3. The ex-anti function: Do people know the criminal law? One further study was done on the subjective objective issue and the crime of attempt. First, it gives us a more fine grain picture of the way that the punishment for the attempted crime scales up as the attempt converges with the completed crime. Second, it gives us our first information on the important question of whether and to what extent citizens are aware of the contours of the criminal law.
224
Kevin M. Carlsmith and John M. Darley
The researchers (Darley et al., 1996) also presented attempt scenarios to respondents, describing increasing steps toward the commission of a crime, ending with the completed crime of murder in one instance and robbery in another. Again, as in the previous experiment, the dependent variable, the duration of the prison sentence assigned to the various scenarios, increased as the nearness of the actions to the complete commission of the crime increased. Taking an initial step toward the commission of the crime received a relatively low sentence; reaching the dangerous proximity stage just prior to the planned commission of the act received a sentence of about 60–70% of the duration of the sentence for the completed robbery or murder. Again, respondents are grading objectivists in that they assign punishments that are linked to the nearness of the action chain to its completion. In this experiment, the researchers also asked the respondents to report how they thought the legal code of the state they lived in would treat the actions described in the scenarios. More specifically, they were asked to report the duration of the penalties that they thought were assigned to the various scenarios by the laws of the state in which they resided. Subjects resided in New Jersey, which is a Model Penal Code state, so the actual laws criminalized attempt at the substantial step level and, it will be remembered, assigned actors who took the substantial step toward the crime the full duration of the sentence they would receive for committing the crime. This result can be seen in Fig. 4.2. The line representing the sentence duration that the respondents would assign to the various scenarios is far below the high and flat line representing the sentences that the legal code would actually assign. (This is the line that, starting at the substantial step scenario, assigns punishments equivalent to the punishment assigned for the attempt.) However, in this study, we remembered to ask the respondents what sentence they thought the state of New Jersey actually assigned. All answered. Their answers are shown in the third line on the figure, which is essentially identical to their answers as to what they thought the morally appropriate punishment should be. What is suggested by this high degree of overlap between the duration that the subject assigns to each scenario and the duration they think that the law assigns is that they are generating ‘‘what they think the law is’’ from their own moral intuitions about what would be the morally right sentence for each scenario. It strikes us that this tendency to assume correspondence between what the laws ‘‘must say’’ and one’s own moral intuitions may be generalizable; a point to which we will return. On public policy terms this discrepancy between what the respondents think the law holds and what the law actually holds means that the respondents, many of whom were long-term residents of the state of New Jersey, are unaware of some very important doctrinal elements in the law of the state in which they live, are in this case blissfully unaware that actions that
225
Psychological Aspects of Retributive Justice
7 6
Liability judgments
5 4 3 2 1
Subjects' sentences Subjects' perceived code Actual code
0 −1 Core
Thought only
Substantial Unequivstep ocality
Dangerous Completed proximity offense
Attempt scenarios
Figure 4.2 Subjects’ sentences, subjects’ perceived code sentences, and actual code sentences for robbery (Darley et al., 1996).
they think would attract only trivial criminal sanctions would in fact draw sentences equal in severity to those visited on offenders who complete the actual crime. Is this an atypical finding, or are citizens often ignorant of important elements of the criminal law? The importance of the question is this. The legal code in a complex society is designed to have several functions. First, it is designed to announce beforehand the rules by which citizens must conduct themselves, on pain of criminal punishment. This is called the ex ante function, and is one of the central constraints imposed on societies that wish to claim that they are fulfilling the ‘‘rule of law’’ requirements. A law that is hidden from the society that it is designed to govern is either no law, or if it is hauled out of hiding and imposed on some unlucky citizen, it is an unjust law. The criminal justice system relies on people knowing the law and knowing where the boundaries for their conduct lie. Ignorance does not excuse unlawful conduct; a fact summarized in the phrase ‘‘ignorance of the law is no excuse.’’ In an empirical study, we (Darley et al., 2001) investigated the generality of this finding of ignorance. Are citizens aware of the lines drawn by the criminal codes of their states? Do people know what the law says? Before answering this, we need to specify what sets of laws we are talking about. It would be possible to rig our test of the question by ferreting out obscure laws about trivial matters. On the other hand, it would be possible to rig our test in the other direction by examining, for instance, whether people knew
226
Kevin M. Carlsmith and John M. Darley
that murder was against the law. People know murder is against the law not because they have scrutinized legal codes but because they cannot conceive that a criminal code would fail to criminalize murder. For our research, we chose to see whether people are aware of the laws about such issues as being required to assist a stranger in distress, report a known felon, or retreat before using deadly force in defense of self or property. These laws struck us as being about important conduct, about situations in which many of us may find ourselves and in which, therefore, we look to the guidance of the legal code. The state-by-state adoption of the Model Penal Code provided a natural experiment, because different states slightly modified different aspects of the code as they adopted it. Clearly, these state legislatures were scrutinizing the code, and deciding that certain of the proposed model code formulations were inappropriate and needed to be changed. So they changed what they enacted into law. But were the citizens of that state knowledgeable about what the law held in their state? Testing this question on these laws seemed to us to set up a fair test of whether people are aware of the contours of various important laws that were intended to govern their behavior. We selected four states, each with a somewhat deviant law about what counts as criminal conduct. We asked selected residents of these states to read a series of scenarios. One scenario described an offense that is criminal in most states, but not in the deviant state, or an action that is criminal in the deviant state but not in the other states. Each state served as an experimental group with the deviant law for one scenario, and as one part of three-part control group for the other three laws.6 Our question is whether the citizens of a state with a deviant law knew the content of that state’s law; therefore, the residents were asked for each of the four ‘‘crimes’’ to report what liability and punishment the law of the state in which they lived would assign. In this study, we were also able to test our prediction that a respondent would infer that the law of the state was what the respondent thought it ought to be, and in turn the respondent’s general punitive attitudes would generate what the respondents thought the law ought to be. Here, on more psychological terms, is the chain of reasoning that we suggest generates a respondent’s answer to the question of whether the state has made an action unlawful. First, general moral attitudes determine whether a person thinks a particular action is morally acceptable or unacceptable. Second, when people are asked their personal view on the criminality of the action, their own moral attitudes determine whether they perceive the action as criminal. Finally, when people are asked whether the state in which they reside criminalizes that action, they answer yes 6
In one case, two states held the deviant version of the law and two did not.
Psychological Aspects of Retributive Justice
227
or no not because they know what the code says but because they assume that the state, in its moral wisdom, shares their personal moral views. Of course, some steps in this chain of reasoning may occur out of the awareness of the respondent. The inference is more likely to be spontaneous and automatic. To test the knowledge of state law hypothesis, we asked respondents whether or not their state criminalized the four actions, and the severity of sentences that should be assigned to each action. To test for support for this ‘‘inferencing’’ hypothesis, we measured a few of the respondents’ punishment relevant attitudes and also whether they themselves thought the action in question should be criminal or not. We then compared the answer patterns across the four states. The study was conducted by enlisting staff 7 of universities in the selected states to be respondents. The laws in question were these: most states impose no duty to assist a person in trouble, even if it could be done without much inconvenience. Wisconsin imposes such a duty. North Dakota and Wisconsin require that a person with force must retreat if he can instead of using deadly force on the attacker. Texas and South Dakota do not require retreat. South Dakota requires that one report to the authorities the location of a person known to be a felon, no other state does. Finally, no state allows the use of deadly force on an offender who is taking only property, except Texas. The results of the study are complex, but gave little support to the notion that the citizens of the state with the variant law accurately predicted that law. Respondents were first asked whether the action in question was criminalized by their state or not. For three of the four comparisons, the citizens of the deviant states had no reliable differences from the control states in their predictions of what was or was not the law. For instance, 36% of the Wisconsin respondents believed their state would punish a person who failed to assist a person in need—which of course was the right answer. However, 39% of the respondents from the three other states believed their state would punish although in fact their states would not punish. On whether there was a duty to retreat when attacked, similar results occurred. North Dakota and Wisconsin hold that a person cannot use force against another if the person can retreat, and 79% of their citizens said the duty to retreat existed. But 71% of the citizens in the states in which no such duty exists thought it did. Finally, South Dakota has a duty to report the location of a felon and 80% of its citizens thought so, but 76% of the citizens in states with no such duty thought their states did have the duty to report. To conclude, all this looks like citizens are guessing rather than knowing the laws of their state.
7
Staff rather than students because we wanted respondents who were long-term residents of the state and likely to have been so during their adult years.
228
Kevin M. Carlsmith and John M. Darley
The exception occurred on the question of whether it was lawful to use deadly force in defense of one’s property—specifically to shoot somebody who was running away with a computer monitor taken from the car in his garage. Seventy-seven percent of the Texas respondents thought that was legal, and they were right. In Texas, it was legal. Only 49% of the citizens of the other states thought so. Does this mean that the Texans knew the law of their state? Hold that question until we present more of the results. What did predict whether respondents favored a deviant or standard law, regardless of their state of residence, were their background attitudes (Darley et al., 2001, p. 189 for these questions). Using structural equation modeling on each case, we determined that for the three cases the background attitudes were a strong determinant of the respondent’s personal judgment about what the appropriate sentence for the action should be, and that personal judgment in turn predicted what the respondent reported was the sentence that the state law would give. The actual law of the state did not predict what we would now call the respondent’s guesses about what the state law was. With respect to the analysis of the Texas data, we said this: ‘‘As in the other models, people’s general attitudes toward the use of deadly force to protect property predicted the respondents’ personal sentence (b ¼ 0.58), and their personal sentence predicted their estimates of the state’s sentence (b ¼ 0.52). The presence of a state law that permits the use of deadly force was predictive of reduced sentencing (b ¼ 0.24), and it was somewhat predictive of attitudes (b ¼ 0.15). It is important to note, however, that there was no direct or mediated relationship between state law and personal sentence. The fit of this model was excellent (RMSEA < 0.001)’’ (p. 181). Our conclusion was that respondents generally did not actually know what the law of their state held. Instead they guessed that their state had gotten the law right, which meant that they guessed that the state law agreed with what they thought was the morally appropriate answer. In the case of Texas, we thought that the ‘‘lenient opinions of the Texans were there first and influenced the passage of the correspondingly lenient laws’’ (p. 182). For the moment, accept the conclusion that in many cases, the citizens of a state are not particularly aware of the exact content of the laws of their states, that is, are not aware of the laws of the jurisdictions that govern them. We should qualify this. They do get it right that prototypical criminal acts, such as actors of unjustified violence, or theft, are criminal. Common law codes were made by judges who held the moral intuitions of their communities, and those intuitions held that these acts were immoral and thus criminal penalties were assigned to them. Later, as codes became products of legislative enactments, legislatures criminalized these actions for the same reasons. On these cases, it is not the case that the citizens consciously recognize that they are predicting what the codes hold. Instead, they simply hold that
Psychological Aspects of Retributive Justice
229
of course the codes criminalize those morally wrong actions. Indeed, we suspect that the legitimacy that citizens grant to the law (Tyler, 1990) stems largely from the presumed (and occasionally erroneous) similarity between their personal moral codes and their State’s legal codes. However, if people, in a blanket way assume that they know the exact content of the criminal codes, based on the code’s exact congruence with their own moral codes, the research we have reviewed in this section, which has demonstrated that citizens intuitions about codes often get it importantly wrong, then they are living in a fool’s paradise. One last fact is worth noting. We searched the newspapers of the deviant states via Lexis-Nexis for stories about why it was that the legislatures were considering adopting variant versions of standard laws. We thought that if information about contemplated laws was ever going to attract public interest, it was the debates about why the legislatures should adopt these laws that were different from the normal laws that would do so. We did not find any such articles, using all the key words we could think of. If we take newspapers to be a proxy for coverage in the public media, then apparently the public media is not a medium to count on for transmission of legal codes to the general public. This may be unfortunate. Among legal philosophers, there is considerable agreement that the Rule of Law is a precondition for the just society. One central tenet of the rule of law is that laws must be promulgated. The notion here is that they must be available to the public so that the public can know the laws and use them to guide their behavior. To serve their ex ante function, that is, the law codes must be promulgated. What, specifically, does this mean? Various dictionaries signal two possible threads of the definition of promulgate. Merriam-Webster’s online dictionary says, ‘‘to make (as a doctrine) known by open declaration.’’ The American HeritageÒ Dictionary of the English Language (Fourth Edition, 2000) says ‘‘To make known (a decree, for example) by public declaration; announce officially.’’ Both of these call to mind images of a public pronouncement of the new law or code, more specifically, attempts to communicate the substance of the law to the public. However, MerriamWebster’s also gives us this: ‘‘to put (a law) into action or force.’’ What might that mean? Apparently, concerning the content of the country’s laws, it is this second thread that governs. What it means with respect to laws is that they be published in some ‘‘official register,’’ generally State publications on the criminal, or administrative laws of the state. The official register no doubt meets the legal definition of ‘‘promulgated’’ but from an intelligent social policy perspective, it certainly fails. That is, if our account of how people think it is that they know the laws of their state—as read off their moral intuitions, the ‘‘register as promulgation’’ notion certainly fails any sensible test. As our examples of laws, the legislatures have chosen to modify makes clear, it is exactly the parts of the law
230
Kevin M. Carlsmith and John M. Darley
that are morally unclear and under debate that are the ones on which legislative modifications are likely. No one is debating the law against murder, or against theft. What is debated, inevitably, are the specifications of the margins of what counts as theft, or duty, or exactly when violence in self-defense is allowable. So it seems truistic that it is exactly this aspect of the laws, on which controversy exists, that large segments of the populace will get wrong. For the legislative community to fail to grasp the chance to explain why it is they draw code lines where they do is a terrible failure of responsibility. Recent shifts in grounds for making actions ‘‘against the law’’ make the situation worse. The existence of contradictions between code and community views has been increased by the recent tendency to blur the line between criminal offenses and violations of regulatory and other civil codes. The criminal justice system deviates from the community’s intuitions about appropriate criminal laws in two ways: by failing to ‘‘criminalize’’ actions that the community thinks are morally wrong, and by criminalizing actions that the community regards as morally acceptable. Depending on which of these intuitions is violated, the community may respond with different specific actions. If the law ‘‘allows’’ actions that the community regards as condemnable, then the community may seek to mobilize informal sanctions on these activities, including inflicting vigilante penalties as severe in magnitude as death. If, on the other hand, the law criminalizes actions that the community, or a segment of the community, thinks are morally permissible, then it is likely that those actions will continue to be practiced by people but will ‘‘go underground.’’ This in turn generates the rise of entrepreneurs who profit by facilitating people’s indulgence in those activities, the establishment of venues such as ‘‘speakeasies’’ in which those activities are practiced, and bribes to legal authorities willing to overlook the existence of these actions and venues. But in addition to these undesirable reactions that attempt to engage or escape control mechanisms for these actions, we suggest that when the criminal justice system is seen as out of tune with community sentiments a less obvious but more common and troublesome reaction occurs—the loss of moral credibility that the justice system standardly possesses in a reasonable society. The justice system loses its relevance as a guide to good conduct. Initially, the reactions may be limited to conclusions about the idiocy of the specific legal rule that offends the community’s morality, but as the apparatus of the social control agents of the government, such as police and the courts are mobilized to enforce the senseless laws, or as those forces seem to stand by passively as moral offenses are committed, there develops a generalized contempt for the system in all its aspects, and a generalized suspicion of all of its rules.
Psychological Aspects of Retributive Justice
231
Others have recognized this as well. Deviations from what is perceived as deserved punishment may have consequences for the respect in which the criminal justice system is held. As Robinson and Darley (manuscript in preparation) state: Just as the institution of the criminal law may be brought into disrepute by the too easy attribution of criminality in situations where the label criminal is generally thought inappropriate, so also may the institution be undercut if it releases as non-criminal those society believes should be punished. This does not mean that the criminal law may not be a means of educating the public as to the conditions under which moral condemnation and punishment is inappropriate. It does mean that the results may not depart too markedly from society’s notions of justice without risking impairment of the acceptability and utility of the institution.
Vigilante justice is an extreme version of a possible reaction. A host of other less dramatic—but more common—forms of resistance and subversion also can be committed when citizens perceive injustice. Jurors may disregard the jury instructions and convict or acquit according to their personal moral judgments rather than the legal rules. Police officers may not arrest or prosecutors prosecute if they think that the law in question criminalizes an innocent action or minor transgression. Or alternatively, police may informally punish actions that they think the legal codes fail to criminalize, or trump up charges against the offenders. Witnesses may lose any incentive to offer their information and testimony. Tom Tyler’s (1990) groundbreaking research on legitimacy supports this. He has found that Americans are likely to obey the law when they view it as a legitimate moral authority. In turn, they are likely to regard the law as a legitimate moral authority when they regard the law as being in accord with their own moral codes. As Tyler concludes, ‘‘the most important normative influence on compliance with the law is the person’s assessment that following the law accords with his or her sense of right and wrong’’ (p. 64). Two recent experiments (Greene, 2003; Nadler, 2005) provide evidence for the beginnings of this rejection process in individuals who discover contradictions between the legal codes in force and their own moral intuitions of justice. In these studies, participants read about a case in which there was a mismatch between the laws in existence and the moral intuitions of the participants. The mismatch was one that many participants found shocking. It was derived from a real world instance in which one young man dragged a young girl into a semi-private space and raped and killed her. A friend, aware of the series of actions, took no steps to intervene or report the action to authorities that might have intervened. The watcher and the friend spent the next two days gambling
232
Kevin M. Carlsmith and John M. Darley
in casinos, later, the watcher bragged to friends about the crime. Because there was no law against the watcher’s actions in the state in question, no legal action was brought against him.8 Participants made aware of this and other unjust cases rated themselves significantly less likely to cooperate with police and less likely to use the law to guide their behavior after reading these intuition-violating cases. More specifically, participants who had read cases in which the legal system behaved in ways counter to their moral intuitions rated themselves more likely to take steps aimed at changing the law (including replacing legislators and prosecutors and breaking the law while taking part in demonstrations), less likely to cooperate with police, more likely to join a vigilante or watch group, and less likely to use the law to guide behavior. Overall, participants appeared less likely to give the law the benefit of any doubt after reading cases where the law was at odds with their intuitions. In the Nadler study, learning about a case in which a similar mismatch occurred on one law ‘‘caused participants to report a willingness to flout unrelated laws commonly encountered in everyday life as well as willingness of mock jurors to engage in juror nullification’’ (Nadler, 2005, p. 1399). John Coffee (1992, pp. 1881–1882) makes the point that the actual frequency of such intuition-violating laws is irrelevant. Actual use of the criminal sanction might remain rare, but it is the threat of its use that must be chiefly considered in evaluating the degree of freedom within a society. To be sure, some may justify pervasive use of the criminal sanction based on simple cost-benefit reasoning: the loss to those imprisoned is less than the harm thereby averted through specific and general deterrence. Yet this analysis depends on a myopic social cost accounting. Even if the deterrent effect gained under such a system of enforcement exceeded the penalties actually imposed, additional costs need to be considered, including the fear and anxiety imposed on risk-averse individuals forced to live under the constant threat of draconian penalties. These citizens would bear not only the risks of false accusation and erroneous conviction, but also the constant fear that they might commit an unintentional violation. Ultimately, if we measure the success of the criminal law exclusively in terms of the number of crimes prevented, we could wind up, in Herbert Packer’s memorable phrase, ‘‘creating an environment in which all are safe but none is free.’’
8
For a narrative describing the case facts and a discussion of the issues arising in it, see Paul H. Robinson, Criminal Law: Case Studies and Controversies, Section 15, "The Act Requirement and Liability for an Omission" (2005). Many respondents found the absence of a law criminalizing this conduct unbelievable. In their study of community intuitions of justice, Robinson and Darley (1995, Study 4) found that their respondents generally punished actors who failed in a duty to rescue a person in distress, if the intervention could be achieved without serious inconvenience or danger to the potential rescuer, as it could have been in the case described here.
Psychological Aspects of Retributive Justice
233
5. Conclusion Penal sentencing practices are much contested in our society, and one message from this chapter is that it is time for empirical psychology to speak-up. In the same way that psychology has made contributions to so many other aspects of justice policy, so too must it make research contributions to the punishment policy ‘‘contest.’’ What do we suggest those contributions might be? In our society at least, crime descriptions mobilize a reaction that we have characterized as ‘‘moral outrage.’’ This is an intuitive rather than a reasoned reaction, one that brainimaging studies suggest contains both emotional and cognitive elements. When presented with specific scenarios describing concretely specified crimes, a person’s intuitive reactions are driven primarily by just desert concerns. Utilitarian considerations, by contrast, are much less influential. As is often the case with intuitive judgments, it is possible that the judgment maker may mobilize reasoning processes to correct and override the intuitive judgment. May mobilize, but does not always or even often mobilize: And even when reasoning processes are mobilized, they frequently rely on just deserts considerations. In some of the studies we reported, the respondents used a reasoning process to arrive at their decisions, and these reasoning processes were also driven by just deserts considerations. More research, with larger and more representative samples of citizens, needs to be done before we can claim these generalizations are descriptive of the responses to crimes by citizens of our society. Fascinating research will test whether the findings are consistent across societies—whether just deserts reactions are ‘‘human nature’’ or the results of cultural socialization. But for a moment, let us assume that they are general at least within our society, and comment on the policy relevant implications. To begin, we need to say more about what the just deserts reaction is and is not. It is best characterized as a ‘‘proportional just deserts’’ stance. Von Hirsch has been the most consistent advocate of punishment according to a just deserts rule, and Wasik and Hirsch (1990) define proportional just deserts as the system in which punishment ‘‘severity should be allocated according to the blameworthiness of the criminal conduct’’ (p. 509). This is elaborated as proportional ‘‘to the conduct’s harmfulness or potential harmfulness and to the offender’s degree of culpability in committing the conduct.’’ Several comments are important to make here. First, there is no requirement that the punishment be ‘‘equal to the crime,’’ only that it be proportionate to it. Different societies have evolved toward having wider or narrower ranges of prison sentence durations, which are the normal mechanism for varying punishment severity. The United States, as we documented, has a wide range of durations including life without parole and the death
234
Kevin M. Carlsmith and John M. Darley
penalty, and assigns high-severity punishments far more than do other Western societies. The Scandinavian countries generally use a more restricted range of punishments. But all of them can employ proportionate punishment systems. What is required is that the most severe punishments the society gives are assigned to what the society regards as the most blameworthy offenses. This can be stated quite precisely on psychometric terms. The rank ordering of the moral wrongness of crimes should correlate with the rank ordering of punishments used by the society. To make what we are saying quite clear, a person or a society can decide never to give a death penalty, even to murderers, and still be doing proportional just deserts if it assigns the murderer the most severe prison sentence the person or the society allows. Proportional justice is not an ‘‘eye for an eye,’’ lex talionis, system. Looking back on our discussion of when it is that the citizens of a society move toward disrespecting their justice system, it too can be stated on psychometric terms. Contempt will develop when the sentencing practices of the society are importantly out of synchrony with the citizens’ rank orderings of the blameworthiness of crimes. We suggested that this was what was developing in the United States. We reviewed cases in which the criminal justice system failed to criminalize actions that the citizens found morally appalling, and did criminalize actions that the citizens found undeserving of the criminal sanction. We reviewed some preliminary evidence that when citizens discovered these discrepancies, they began to lose respect for the justice system, began to lose confidence in the legal system as a guide to moral behavior. We suggested that this was not a necessary result of community-code discrepancies. It should be possible to convince citizens of the reasoned conclusion that the criminal code ‘‘had gotten it right,’’ but that the criminal justice system was failing in its duty to persuade citizens that its criminalization patterns were the correct ones. Many critics of a proportional just deserts system assert that the necessary consensus on blameworthiness cannot be found within any society, or at least within our society. They are simply empirically incorrect. We earlier discussed the correlations that have been found in samples in our society, and found them remarkably high.9 There is a simple formulation of what we wish to say. For a number of reasons, the forces acting to shape the criminal laws of our society are moving toward creating laws that are more and more distant from the moral judgments of its citizens about which behaviors are immoral and ought to be criminalized, and which are moral and should not be subject to criminal sanctions. As we, as well as others have documented, the psychological costs 9
We freely acknowledge that researchers such as Haidt (Haidt and Hersh, 2001) and Jost ( Jost et al., 2003) have identified important topics in which delineated groups of people disagree passionately about the moral trespass of certain actions (e.g., abortion). But these strike us as notable exceptions to the otherwise consensual view within society of what constitutes a crime and harm doing.
Psychological Aspects of Retributive Justice
235
of this is that the law loses credibility as an honest guide to moral behavior. Worse, it draws contempt on itself when, for instance, it fails to punish callous disregard for victims in distress, or in other cases demands punishment for actions that seem innocent to citizens. The ‘‘rule of law’’ is threatened when the rules of law violate citizen intuition.
REFERENCES Barton, T. D. (1999). Therapeutic jurisprudence, preventive law, and creative problem solving: An essay on harnessing emotion and human connection. Psychol. Public Policy Law 5, 921–943. Bentham, J. (1962). Principles of penal law. In ‘‘The Works of Jeremy Bentham’’ ( J. Bowring, ed.), p. 396. W. Tait, Edinburgh. (Original work published 1843). Brickman, P., Folger, R., Goode, E., and Schul, Y. (1981). Microjustice and macrojustice. In ‘‘The Justice Motive in Social Behavior’’ (M. Lerner, ed.), pp. 173–204. Plenum, New York. Carlsmith, K. M. (2006). The roles of retribution and utility in determining punishment. J. Exp. Soc. Psychol. 42, 437–451. Carlsmith, K. M. (2008). On justifying punishment: The discrepancy between words and actions. Soc. Justice Res. doi: 10.1007/s11211-008-0068-x. Carlsmith, K. M., Darley, J. M., and Robinson, P. H. (2002). Why do we punish? Deterrence and just deserts as motives for punishment. J. Pers. Soc. Psychol. 83, 284–299. Carlsmith, K. M., Monahan, J., and Evans, A. (2007). The function of punishment in the ‘‘civil’’ commitment of sexually violent predators. Behav. Sci. Law 25, 437–448. Clements, C., Wasieleski, D. T., Chaplin, W. F., Kruh, I. P., and Brown, K. P. (1998). The sentencing goals inventory: Development and validation. Poster session presented at the biennial meeting of the American Psychology-Law Society, Redondo Beach, CA. Coffee, J. C., Jr. (1992). Paradigms lost: The blurring of the criminal and civil law models. And what can be done about it. Yale Law J. 101(8), 1875–1893. Cooksey, R. W. (1996). ‘‘Judgment Analysis: Theory, Methods, and Applications.’’ Academic Press, San Diego, CA. Darley, J. M., Sanderson, C. A., and LaMantia, P. S. (1996). Community standards for defining attempt: Inconsistencies with the model penal code. Am. Behav. Sci. 39, 405–420. Darley, J. M., Carlsmith, K. M., and Robinson, P. H. (2000). Incapacitation and just deserts as motives for punishment. Law Hum. Behav. 24, 659–683. Darley, J. M., Carlsmith, K. M., and Robinson, P. H. (2001). The ex ante function of the criminal law. Law Soc. Rev. 35, 701–726. Friedman, L. M. (1985). ‘‘A History of American Law,’’ (2nd edn). Simon and Schuster, New York. Greene, E. J. (2003). Effects of disagreements between legal codes and lay institutions on respect for the law. Dissertation Abstracts International: Section B: The Sciences and Engineering 64(2-B), 1001. Greene, J. D., Sommerville, R. B., Nystrom, L. E., Darley, J. M., and Cohen, J. D. (2001). An f MRI investigation of emotional engagement in moral judgment. Science 293, 2105–2108. Greene, J. D., Nystrom, L. E., Engell, A. D., Darley, J. M., and Cohen, J. D. (2004). The neural bases of cognitive conflict and control in moral judgment. Neuron 44, 389–400. Gromet, D. M., and Darley, J. M. (2006). Restoration and retribution: How including retributive components affects the acceptability of restorative justice procedures. Soc. Justice Res. 19, 395–432.
236
Kevin M. Carlsmith and John M. Darley
Haidt, J. (2001). The emotional dog and its rational tail: A social intuitionist approach to moral judgment. Psychol. Rev. 108, 814–834. Haidt, J., and Hersh, M. A. (2001). Sexual morality: The cultures and emotions of conservatives and liberals. J. Appl. Soc. Psychol. 31, 191–221. Hauser, M. D., Young, L., and Cushman, F. A. (2007). Reviving Rawls’ linguistic analogy. In ‘‘Moral Psychology: Vol. 2. The Cognitive Science of Morality’’ (W. SinnottArmstrong, ed.). Oxford University Press, Oxford, NY. Jacoby, J., Jaccard, J. J., Kuss, A., Troutman, T., and Mazursky, D. (1987). New directions in behavioral process research: Implications for social psychology. J. Exp. Soc. Psychol. 23, 146–175. Jost, J. T., Glaser, J., Kruglanski, A. W., and Sulloway, F. J. (2003). Political conservatism as motivated social cognition. Psychol. Bull. 129, 339–375. Kahneman, D. (2003). A perspective on judgment and choice: Mapping bounded rationality. Am. Psychol. 58, 697–720. Kant, I. (1952). The science of right (W. Hastie, trans.). In ‘‘Great Books of the Western World’’ (R. Hutchins, ed.), pp. 397–446. T., and T. Clark, Edinburgh. (Original work published 1790). Liberman, N., and Trope, Y. (1998). The role of feasibility and desirability considerations in near and distant future decisions: A test of temporal construal theory. J. Pers. Soc. Psychol. 75, 5–18. McKee, I., and Feather, N. T. (2006). Sentencing goals, human values and punitive personality types: The role of revenge in attitudes towards punishment and rehabilitation. Symposium on ‘‘Legal Justice’’. In ‘‘11th Biennial Meeting of the International Society for Justice Research,’’ Berlin, Germany. Miller, D. T., and Vidmar, N. (1981). The social psychology of punishment reactions. In ‘‘The Justice Motive in Social Behavior’’ (M. Lerner, and S. Lerner, eds.), pp. 145–172. Academic Press, New York. Nadler, J. (2005). Flouting the law. Tex. Law Rev. 83(5), 1399–1441. Nisbett, R. E., and Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychol. Rev. 84, 231–259. Oswald, M. E., and Stucki, I. (2008). A two-process model of punishment. In ‘‘Social Psychology of Punishment of Crime’’ (M. E. Oswald, S. Bieneck, and J. HupfeldHeinemann, eds.). Wiley, Chichester. Robinson, P. H., and Darley, J. M. (1995). ‘‘Justice, Liability and Blame: Community Views and the Criminal Law.’’ Westview Press, Boulder, CO. Robinson, P. H., and Darley, J. M. (1998). Objectivist vs. subjectivist views of criminality: A study in the role of social science in criminal law theory. Oxf. J. Leg. Stud. 18, 409–447. Robinson, P. H., and Darley, J. M. (manuscript in preparation). Intuitions of justice: Implications for criminal law and justice policy (March 22, 2007). University of Penn Law School, Public Law Research Paper No. 07–12 Available at SSRN:http://ssrn.com/ abstract¼976026. Trope, Y., and Liberman, N. (2000). Temporal construal and time dependent changes in preference. J. Pers. Soc. Psychol. 79, 876–889. Tversky, A., and Kahneman, D. (1971). Belief in the law of small numbers. Psychol. Bull. 76, 105–110. Tversky, A., and Kahneman, D. (1983). Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment. Psychol. Rev. 90, 293–315. Tyler, T. R. (1990). ‘‘Why People Obey the Law.’’ Yale University Press, New Haven, CT. Wasik, M., and Hirsch, A. V. (1990). Statutory sentencing principles: The 1990 white paper. Mod. Law Rev. 53(4), 508–517. Wheatley, T., and Haidt, J. (2005). Hypnotic disgust makes moral judgments more severe. Psychol. Sci. 16, 780–784.
Index
A Acceptance, 7, 9–10, 12, 16, 18, 20, 35 Achievement behaviors, 132 Affect, in theories of emotion, 67 Affection, 5 Agitation, 160 Agreeableness, 160, 165–167 Alarm contingency, 5–6 for newlyweds, 21–23 Ambivalent Sexism Inventory (ASI), 76, 123 Ambivalent sexism theory, 122 Amygdala’s vigilance system, 135 Antipathy, 67, 75 Antisocial behavior, 130, 132–133, 160 Anti-Tutsi propaganda, 129 Anxieties, 6, 23, 160 Appraisal theories of emotion, 67 Approach-related behavior, 153 Argument quality effect, in message processing, 253, 255–256, 260 Attitude change assessment of, 293–294 to majority and minority, 252 in majority and minority influence, 252 and nonsystematic processing, 271–272 and systematic processing, 271 Attitude theories, 67 Attributional bias, 132 B Behavior from intergroup affect and stereotypes (BIAS), 63 Behaviors hypotheses, 108–119. See also BIAS map active behavioral responses, 110–111 behavioral tendencies correlations of, 115 standardized means, 115 bias clusters, 113 correlational and experimental support, 114–119 dimensions of behaviors, 108–109 active harm, 109 passive facilitation and passive harm, 110 passive behavioral responses, 111 pattern of behaviors, 112–113 Benevolent sexism (BS), 76–77, 122–124
BIAS map, 66 basic tenets of, 70–71 active and passive behaviors, 70–71 specific behavior patterns, 71 Black–White intergroup relations, 63 C California’s 3-strikes law, 208 Career women, 123–124 Carlsmith work’s, on retributive justice, 203–206 Cognitive appraisals, 67 Collectivism, 66 Commitment-insurance model, 5 Commitment-insurance system, 3–4 alarm and repair contingencies dependence of partners, 26–29 and exchange concerns, 23–25 moderating role of self-esteem, 30–32 for newlyweds, 21–23 partner commitment, 29–30 connectedness in, 9–11, 33–38 empirical support for, 11–12 perceived commitment, comparisons shift, 14–21 perceiving match or equivalence, 13–14 setting aspirations, 12–13 future research on, 54–55 historical perspectives of, 38–40 If–then, contingencies of exchange, 4–9 interdependence theory, motivational perspectives, 40–46 judging equivalence, 46–48 sociometer model of self-esteem, 11 Committing a crime, 221, 224 Common law, 196, 221 Common sense, 196 Communal traits, 76, 91, 101 Communication, 178 Communion, 66 Compensatory contingency, functions of, 6 Competence dimension, 63 Competence-related traits, 120 Competence scales, 65 Competition, 96, 101 Competition drives perceptions of warmth, 95 Competitive-frame participants, 98 Conflict elaboration theory, 246–247. See also Conversion theory
327
328
Index
Confucianism, 168 Conscientiousness, 161, 165 Contempt, 219, 230, 234–235 Content of thoughts, 251 Context/categorization theory, 247–248 Convergent-divergent theory, 243–244 Convergent thinking, 243 Converging theory and evidence, 101–102 Conversion theory, 240–243 Cooperative-frame participants, 98 Counter-attitudinal majority, 250 Counter-attitudinal message, 257–258 Counter-attitudinal minority, 243 Counter-attitudinal position, 309 Counter-persuasive communication, resistance of, 287–292 Crime of attempt, 221 Criminal attitude, 221–223 Criminal justice system, 196–197, 225, 230 Criminal law awareness, 196, 222 community’s intuitions and, 230 importance of, 223 intuition-violating law and, 232 legal code and, 224–226 model penal code and, 226 standard law adoption, 229 state law hypothesis, 227–228 vigilante justice and, 231 Cultural default groups, 95 D Darley works’s, on retributive justice, 204–205 Dating couples, 12–13, 15 Democracy and minorities, 313–314 Deservingness, 13 Deterrence, 199–203, 206, 232 Discursive reputation, 164–165 Distress, 160 Distributive justice, 195 Distributive reputation, 164–165 Divergent thinking, 243 Divorce, 7 Drug addicts, 80, 132–133, 135–136 Dual-process models, 304–306 Dual role model, 248–249. See also Convergentdivergent theory; Objective consensus approach Dual source paradigms, 301–302 Dyadic competition, 98 Dynamism, 66 E East Asian collectivistic cultures, 121 Elaboration continuum, 267–269 Elaboration Likelihood Model, 264, 268, 271, 300, 303–305
Elaborative processing, stages, 266. See also Source-context-elaboration model (SCEM) Electromyogram (EMG), 300–301 ELM. See Elaboration Likelihood Model EMG. See Electromyogram Emotional convergence process, power in, 174–175 Emotions, 67, 153 Emotions hypotheses admiration, 105–106 contempt, 104–105 envy, 104 pity, 103–104 Enemy image theory, 67 Envious prejudice, 76, 104, 122 Asians and Asian-Americans, 125–127 Jews, 127–129 Ethnic Chinese, migration, 62 Ex-anti function, 223, 225 Exchange or even trade, relationships, 23 Exchange script priming and partner dependency, 27–28 Explicit contingency rules, for self-esteem, 8, 51, 53 Extraversion, 159–161, 165 Extraverts, 159–160 F Femininity, 66 Frans de Waal work’s, on social power, 155 Freedom and obligation, ideologies of, 184 G Gender stereotypes, 65 Glick’s ideological theory, 129 Goal-directed social cognition, 153 Gossip definition of, 165 discursive reputation transmission, 170 frequent targets of, 167 hypothesis testing of, 166 role in group members, 168, 178 Group-based representations, 178 Group members, role in social power, 171 H Heiderian analysis, warmth and competence, 134 Heuristic processing, 264 Heuristic Systematic Model, 263–264, 304–305 High partner-dependence condition, 29 High partner standard condition, 23 High-power individuals, study of, 176 High status, competitive groups, 95 Hostile sexism (HS), 76, 122–124 HSM. See Heuristic Systematic Model
329
Index
Human behavior meta analysis of, 161–162 perceived and actual relations, 162 and punishment incapacitation, 200–201 measures of, 198–199 methodological issues, 197–198 retribution, 199–200 specific deterrence, 200 verbal reports, 198 Human hierarchies, 154–156, 170, 185 Human sociality, basic elements of, 155 Human social relationship, complexity of, 155–156 Human, ultrasocial species, 155 I Ideal partner, 5, 12–13 Ideological model of scapegoating, 128 If-then, contingencies of exchange, 4–9, 18, 32–33 Image theory, 67, 101 Immoral action, 212 Impersonal moral dilemmas, 216 Implicit contingency rules, for self-esteem, 8, 51, 53 Incapacitation, 197, 200–201, 203, 205 Incompatible goals, 96 Incompetence stereotypes, 120 Individual differences, 306 Individualism, 66 In-group membership, 246–248 Intellectual incompetence, 120 Interdependence analyses, of partners, 3–4, 7 duality of, 48–54 Intergroup attributions, 130–135 Intergroup emotions theory (IET), 67 Intergroup perception ambivalent stereotypes, 75–78 competent but cold, 77–78, 125 univalent, 78 warm but incompetent, 76–77, 120 centrality of warmth and competence, 74–75 Princeton stereotyping series, 75 specific social groups, stereotypes of, 74–75 cross-cultural SCM studies, 82–89 BIAS map, models of stereotypes, 88 East Asia, 86–87 Europe, 84–86 favoritism ratings, reference-group, 89 Israel, 87–88 Latin America, 88–89 questionnaires and samples, 82–83 stereotype content model, 85, 87 US SCM studies, 79–82 cluster solutions for SCM samples, 81 participants rated groups, 79–80
Interpersonal acceptance, 35–36 Interpersonal level of analysis, 97 Interpersonal perception centrality of warmth and competence, 71–74 Interpersonal relationships, 3 Intuition-violating law, 232 Intuitive judgments, 214 Intuitive system, process of, 211 Issue-relevant thinking, 244 J Jewish conspiracy, 129 L Legal codes and community intuitions, discrepancies between, 219–220 Leniency contract, 248 Lexis-Nexis, 229 Low partner-dependence condition, 29 Low partner standard condition, 23 Low-power individuals, study of, 155, 174–177 M Machiavellianism, 166–167 Macrojustice orientation, 211 Majority default option, 266–267 Majority-minority status, 308–310 Majority vs. minority, social influence of contingency models, 249–250 conflict elaboration theory, 246–247 context/categorization theory, 247–248 dual role model, 248–249 source-position congruity, 245–246 experimental studies message processing conditions, 279–287 message processing source, status and types, 277–279 nature of attitudes, 287–296 main effects models for, 249 convergent-divergent theory, 243–244 conversion theory, 240–243 objective consensus approach, 244 methodological issues, 297–304 assessment elaboration, 299–301 control conditions, 302–304 manipulation of source, 297–299 Single vs. dual source paradigms, 301–302 procedure and methodology, 272–277 source-context-elaboration model contexts of, 267–271 nature of attitudes, 271–272 processes in, 263–267 theoretical and future issues, 304–312 Medial prefrontal cortex (mPFC), 107–108, 135–136
330
Index
Medical disorders, genetic screening for, 290 Message-congruent thought index, 254 Message elaboration, consequences of, 272 Message processing, 251–254 conditions for, 279–287 majority and minority in different situations, 256–260 majority and minority source, 255–256 majority source in, 245, 247–248, 254 source status and types, 277–279 Message-relevant processing, 244 Meta-self-perceptions, 137 Microjustice orientation, 210 Minority behavioral style threshold, 267 Minority influence, consequences of, 242–243 Minority messages, advantages of, 314–315 Model of social power, 152 Model Penal Code, 221–222, 224, 226 Modern globalization, 62 Modesty, 168–169 Morality, 66, 181, 213 Moral judgments, role of reasoning system, 213 brain-imaging research, 216 dual process, 217 intuitive judgments, 214–215 Moral traits, 65 Moscovici’s majorities and minorities, 242 Motivational assessments of partners, 5 N National stereotypes, 64 Natural assessments, 212 Need for Closure scale, 307 Need for cognition, 306 Negative outcome bias, 280–283 Negative personal outcome, 260 Neural signatures for SCM space, 135 Neuroticism, 160, 165 NfC. See Need for cognition NFCL. See Need for Closure scale Nonelaborative processing, 266 Nonresponsiveness, 5 Nonverbal behaviors and power, relation of, 161 Non-violent persuasion, 311 O Objective consensus approach, 244 Obligation, 180–181, 184 Older people, 74, 77, 120–122 Optimal distinctiveness theory, 307–308 Optimism, 14–15, 182 Ostracism, 9 Outcome of influence, 252 Out-group membership, 247–248
P Pan-cultural ageism, 121 Partners, committment, 5–7, 29, 33, 40, 42–44, 46, 48 comparisons in, 14–21 self-esteem in, 7–11 Paternalism, 76–78, 104, 120, 122–124 Perception, 63 intergroup, 74–89 interpersonal, 71–74, 93 Persist over time, 293 Personality psychology, 65 Personal Need for Structure scale, 306–307 Personal violations, 216 Pitying prejudice, 120 Pitying stereotypes, 75 Plaintiveness, 160 PNS. See Personal Need for Structure scale Policy implications, 218–219 criminal attitude, 221–223 criminal law awareness, 223–232 discrepancies between legal codes and community intuitions, 219–220 Positive intentions, 96 Positive stereotype, 96 Power, 152–154 and amplification of preexisting inclinations, 173–174 displays, 161–163 experience of, 178 power and ideologies of agency and obligation, 180–185 reciprocal influence dynamics, 178–180 as interaction heuristic, 169–170 and interpersonal responsiveness, 174–177 a social affordance, 170–172 and social constraint, 163–164 modesty, 168–169 reputation and gossip, 164–167 social psychological study of, 153 Predict behaviors, 293–296 Prejudiced emotions, 102 correlational evidence, 106–107 emotions hypotheses, 102–106 experimental evidence, 107–108 Primacy of warmth, 89 accessibility of warmth information, 91 diagnosticity of warmth and competence information, 92 moderation of primacy of warmth, 90–91 profitable traits, importance of, 90 Rapidity of warmth judgments, 90 Pro-attitudinal message, 257 Pro-attitudinal minority, 250 Proportional justice, 234 Psychological components of bias, 67
331
Index
Punish, impulse to, 211–217 intuitive system judgment, 212 moral judgments, 213–217 process of intuitive system, 211 Punishment severity, path model of, 206 Punitive judgments, 217. See also Criminal justice system R Reasoning processes, 211 Reasoning system, 217 Reciprocal influence dynamics, 178–180 model, of social power, 156–158 Repair contingency, 5–6 for newlyweds, 21–23 Reputation content theme of, 166 definition of, 164–165 Research program. See also Majority vs. minority, social influence of; Message processing experimental studies, 277–296 message development, 273–275 statistical analysis, 277 stimulus materials, 273 topic of influence and dependent measures, 276 Resilience, 10, 38 Responsiveness, 3 Restorative justice system, 209 Retribution theory, 199–200 Retributive justice, 194 Aristotle and Plato’s work’s on, 209–210 Brickman work’s on, 210 Carlsmith works’s on, 203–206 Darley works’s, 204–205 deterrence, 201–202 goal of, 199–200 harm-based justice, 195–196 incapacitation of, 200–201 individual differences, 206–209 measures of, 198–199 methodological issues, 197–198 principles of, 196–197 specific deterrence, 200 utility of, 200 verbal reports, 198 Risk regulation system, 49 S SCM structural hypotheses cross-cultural tests of, 98 US tests of, 96 Self-assessments, 157, 178 Self-esteem, 106, 136–137 and commitment-insurance, 5, 7–9 moderating role of, 10–11, 30–32, 34–38
sociometer model, 11 Self-profitable traits, 65 Self-protection goals, 53–54 Self-uncertain and self-certain condition, 23 Sense of humor, 27, 45–46 SES. See Socio-economic status Sex differences, 65 Social affordance, 159, 170–172 Social categories in society, 62 Social cognition, 135 Social dynamic laws, 152 Social-emotional responses, 217 Social engagement, 163 hypothesis, 159–160 signals of, 161–163 Social motives, 65 Social movements and minority influence, 310–312 Social power acquisition of, 158–159 displays of power, 161–163 disposition, 159–161 definition of, 154–155 empirical study of, 153–154 gossipers role in, 167 judgments of, 171 perceptions of, 172 reciprocal influence model of, 156–158 Social status accuracy in perceiving, 172 correlations of dimensions, 160 Socio-economic status, 181, 184–185 Sociometer model of self-esteem, 11 Sociometric ratings and personality measures, relationship of, 166–167 Source-context-elaboration model (SCEM). See also Conversion theory; Dual role model contexts of, 267–271 nature of attitudes, 271–272 processes in, 263–267 Source-position congruity, 245, 280 Stare decisis, 196 Status–competence effects, 98 Stereotype-Confirming Attribution Bias (SCAB), 131, 133–134 Stereotype content model (SCM), 63–64, 66 basic tenets of, 68–70 ambivalent stereotypes, 68–69 signature emotions, 69–70 social structural origins, 69 warmth and competence, centrality of, 68 cross-cultural tests, structural hypotheses, 98–99 competitiveness and warmth, relationship, 99 social structure–stereotype correlations, 99–100 image theory and social role theory, 101–102
332
Index
Stereotype content model (SCM) (cont) neural signatures for SCM space, 135–136 US tests, structural hypotheses, 96 correlational evidence, 96–97 experimental evidence, 97–98 Stereotypes, changes in, 62 Sympathy, 76, 103–104, 106–107, 120 Systematic processing, 264 T Taoism, 168 Task performance, majority and minority influence on, 253 Team game, cooperation, 97–98 Temporal Construal Theory, 210–211 Thought-listing technique, 299–301 Tort Law system, 195 Traditional women, 122–124 Trait adjectives, 72 Trivial criminal sanctions, 225 Trolley Car problem, 212–213, 216 U Ultimate attribution error (UAE), 130–131, 133 Ultrasociality, 154–155, 170 V Validation process, 242 Verbal reports, advantage of, 198 Vigilante justice, 231 Volatile ambivalence, 130 Voluntary student unions, 295–296 VSU. See Voluntary student unions
W Warm–incompetent cluster, 107, 120 See also Older people Warmth and competence centrality of, 64 fundamental dimensions of social perception, 71 Warmth and competence judgments (See also Stereotype content model (SCM)) ambivalent combinations on, 119 envious prejudice, 125–130 pitying prejudice, 120–124 emotional and behavioral consequences assessment, 64 locating origins of, 64 social structural roots of, 92–94 competition and warmth judgments, 95–96 social status and competence judgments, 94–95 in self perception, 136–137 Warmth dimension, 63 Warmth-trustworthiness dimension, 65 Wiggins’s interpersonal circumplex of behaviors, 64 Willingness to risk interdependence, 52 Winner takes all, competition, 98 Women’s influx, paid work force, 122 Working mother, 124 Y Yoked judges, 98
C H A P T E R
F I V E
Majority Versus Minority Influence, Message Processing and Attitude Change: The Source-ContextElaboration Model Robin Martin* and Miles Hewstone† Contents 1. Introduction 2. Theoretical Models 2.1. Main effects models 2.2. Contingency models 2.3. Summary of approaches 3. Research Examining Message Processing 3.1. Message processing for majority source only 3.2. Message processing for both majority and minority source 3.3. Message processing for either majority or minority in different situations 3.4. Integration of findings 4. The Source-Context-Elaboration Model of Majority and Minority Influence 4.1. Processes 4.2. Contexts affecting when processes occur 4.3. Nature of attitudes following majority and minority influence 5. Research Program 5.1. General methodology and procedure 5.2. Experimental studies 5.3. Summary 6. Methodological and Theoretical Issues in Current and Future Research 6.1. Methodological issues in our research program 6.2. Theoretical issues and future research
* {
238 240 240 244 249 250 254 255 256 260 263 263 267 271 272 272 277 296 297 297 304
Work and Organisational Psychology Group, Aston Business School, Aston University, Birmingham, B4 7ET, United Kingdom Department of Experimental Psychology, Oxford University, Oxford, OX1 3UD, United Kingdom
Advances in Experimental Social Psychology, Volume 40 ISSN 0065-2601, DOI: 10.1016/S0065-2601(07)00005-6
#
2008 Elsevier Inc. All rights reserved.
237
238
Robin Martin and Miles Hewstone
7. Conclusions 7.1. Majority versus minority influence: When, not whether 7.2. Minorities, democracy, and fairness 7.3. Epilogue Acknowledgments References
312 312 313 314 315 315
Abstract This chapter examines the contexts in which people will process more deeply, and therefore be more influenced by, a position that is supported by either a numerical majority or minority. The chapter reviews the major theories of majority and minority influence with reference to which source condition is associated with most message processing (and where relevant, the contexts under which this occurs) and experimental research examining these predictions. The chapter then presents a new theoretical model (the source-context-elaboration model, SCEM) that aims to integrate the disparate research findings. The model specifies the processes underlying majority and minority influence, the contexts under which these processes occur and the consequences for attitudes changed by majority and minority influence. The chapter then describes a series of experiments that address each of the aspects of the theoretical model. Finally, a range of research-related issues are discussed and future issues for the research area as a whole are considered. Never doubt that a small group of thoughtful, committed citizens can change the world. Indeed, it is the only thing that ever has. (Margaret Mead) . . . if either of the two opinions has a better claim than the other, not merely to be tolerated, but to be encouraged and countenanced, it is the one which happens at the particular time and place to be in a minority. ( John Stuart Mill, On Liberty, 1859/1986)
1. Introduction We sat down to write this chapter in a year that saw the two hundredth anniversary of the abolition of the slave trade in the British Empire. Like many other instances in which prevailing attitudes were overturned and social movements galvanized, the seeds of the abolition protest were sown by a minority. Every social psychologist knows Sidney Lumet’s film Twelve Angry Men as an example of minority, and majority, influence. Twelve men meeting in a printing shop above a London pub are also credited with beginning the process whose bicentenary is celebrated this year, ‘‘what these citizens began rippled across the world and we feel its aftereffects still’’ (Hochschild, 2005, p. 1).
Majority Versus Minority Influence
239
There are, of course, many attempts to influence our opinions every day that might not be as dramatic as the ones noted above. These attempts can come from a variety of directions—reading a newspaper, listening to the television/radio, hearing a debate, and face-to-face interactions are all situations where one person or group is trying to change the attitudes and opinions of another person or group. On occasions, the numerical support of the position advanced by the source of influence is salient by referring to the position as being supported by the ‘‘majority’’ or ‘‘minority’’ of the population or by revealing the percentage of people supporting the position. It is this aspect of attitude change that is the subject matter of this chapter and it addresses the basic research question—what are the contexts in which people will process more deeply, and therefore be more influenced by, a position that is supported by either a numerical majority or minority. The potential influence of minorities was only belatedly introduced into social psychology by the pioneering theorizing and research of Serge Moscovici (1976, 1980; Moscovici and Faucheux, 1972), following our discipline’s initial focus exclusively on majorities. Yet the history of ideas, not to mention the rise and fall of fashions, musical styles, and so on, is replete with examples of minority influence. Galileo was humiliated by the Roman Catholic Church (for promulgating the view that the earth revolved around the sun; see White, 2007).1 In nineteenth century London, English physician Dr. John Snow attempted to persuade majority opinion that the epidemic disease it faced was carried in the capital’s water supply, and not due to atmospheric ‘‘miasmas’’ (see Johnson, 2006); but he was only vindicated after his death. Contemporary examples abound too, such as Peter Roberts, who founded Compassion in World Farming in 1967, and effected a transformation in more humane practices of animal welfare, which was ‘‘achieved by force of argument, not by violence or threats’’ (The Economist, December 2, 2006). The remainder of the chapter is organized into six sections where we (1) briefly review the major theories of majority and minority influence with specific reference to which source condition is associated with most message processing; (2) review and evaluate research examining which source leads to most message processing; (3) outline a new theoretical model that specifies the processes underlying majority and minority influence, the contexts under which they occur and the consequences for attitudes changed by majority and minority influence; (4) describe a series of experiments that address each of the aspects of our theoretical model; (5) critically consider methodological and theoretical issues in current and future research; and (6) draw some conclusions for the research area as a whole.
1
In 1633, Galileo was forced by the Italian Inquisition to retract his heretical teaching. In 1980, the Pontifical Academy of Science reviewed his case, taking 12 years to concede that he was right.
240
Robin Martin and Miles Hewstone
2. Theoretical Models You can’t turn a thing upside down if there’s no theory about it being the right way up. –G. K. Chesterton (attributed)
We begin, as a necessary prologue to our review of the literature and then introduction to our own model, by reviewing some of the main models of majority and minority influence. We focus on the predictions each model makes concerning the processes associated with each source and the consequences of these processes for attitude change (for reviews of the area see De Dreu, 2007; De Dreu and De Vries, 2001; Martin and Hewstone, 2003a; Martin et al., 2008b). Of course, it would be ideal if one theory could account for all the data; but as Quiamzade et al. (in press) have written of this area, ‘‘straightforward theories are elegant and stimulating, but they do not account for the complexity of social influence dynamics’’ (p. 260). For clarity of exposition, we have organized the theoretical models into those that specify main effects (i.e., predict specific, and different, processes and outcomes for each source) versus contingency models (i.e., predict processes and outcomes for each source dependent upon one or more contingency factor). This classification also reflects a chronological order in theory development with the main effects models being developed before the contingency models (Martin and Hewstone, 2003a). It should be noted that we only focus on theories that explicitly consider the issue of message processing and do not consider those that do not address this issue, such as mathematical models (Latane´ and Wolf, 1981; Tanford and Penrod, 1984) and self-categorization theory (David and Turner, 1996, 1999, 2001). Also, we only focus on those aspects of the models that specify which source is associated with most message processing and, in doing this, we do not provide exhaustive research evidence for each model (for relevant research on each model see De Dreu, 2007; Levine and Russo, 1987; Martin and Hewstone, 2003a, in press, Martin et al., 2008b).
2.1. Main effects models In this section, we describe three models that propose specific, and different, processes for each source condition: conversion theory, convergent-divergent theory, and the objective consensus approach. The main predictions of these models are shown in Table 5.1. 2.1.1. Conversion theory The dominant perspective in this area has been Moscovici’s (1980, 1985) (see also Moscovici, 1976 and Moscovici and Faucheux, 1972 for developments of the model) conversion theory which argues that all forms of
Table 5.1
Key comparisons concerning main effects models Conversion theory (Moscovici, 1980) Majority
Convergent-divergent theory (Nemeth, 1986)
Objective consensus approach (Mackie, 1987)
Minority
Majority
Minority
Majority
Validation
Convergence
Divergence
Systematic
Characteristics of majority
None
High
Low
High
Low
Process Cognitive activity Focus of attention
Comparison
Characteristics of majority
Content of arguments
Content of arguments
Direct/public attitude change Indirect/private attitude change
High
Low
High
Wider issues related to topic Low
Low
High
Low/High
High
<
Minority
>
Nonsystematic
242
Robin Martin and Miles Hewstone
influence, whether from a majority or minority, result in conflict and that individuals are motivated to reduce that conflict. The resolution of conflict, however, varies depending on the nature of the source. In the case of majority influence, Moscovici proposes that individuals engage in a comparison process where they concentrate attention on ‘‘. . . what others say, so as to fit in with their opinions or judgments’’ (Moscovici, 1980, p. 214). In this situation, people wish to belong to the majority group, as identification with a majority is desirable and, through the social comparison processes, conform to the majority position. Because the goal is to attain majority group membership, and to avoid being categorized as a minority, people accept the majority position without the need for a detailed appraisal of the majority’s arguments. Because acceptance of the majority position has resulted from public compliance to this position without considering in details its arguments, the attitude change tends to be located only on a public/direct level and underlying attitudes are not affected. In the case of minority influence, favorable social comparison is unlikely as minority membership is typically associated with undesirable characteristics. They are seen as deviant, rebarbative, and even sinister. This is why minority influence tends to occur later; as Gandhi wrote ‘‘First they ignore you, then they laugh at you, then they fight you, then you win.’’ In order to influence despite these difficulties, Moscovici (1976) argued that a persuasive minority would have to adopt a behavioral style of ‘‘consistency’’; by ‘‘standing-up’’ to the majority, the minority shows that it is certain, confident, and committed to its position, and will not be easily swayed. William Wilberforce’s 20-year effort to introduce a Bill into parliament to abolish the slave trade is an example of such consistency and commitment. But consistency can also be thought of, and operationalized as, a more shortterm variable, holding to a line in an argument, making the same point in several ways in a pamphlet, and so on. Through this behavioral style, majority members begin to see that the minority has a valid alternative that deserves, perhaps demands, to be considered. Moscovici proposes that the minority can encourage a validation process, leading the recipient of minority influence to ‘‘. . . examine one’s own responses, one’s own judgments, in order to confirm and validate them . . . one’s main preoccupation [is] to see what the minority saw, to understand what it understood’’ (1980, p. 215). By being consistent, the minority is ‘‘visible’’ within the group and attracts, or even demands, attention (Schachter, 1951). Whereas minority influence may not lead to public agreement, due to fear of being categorized as a minority member (Mugny, 1982), the close examination of the minority’s position may bring about attitude conversion on an indirect, latent, or private level (i.e., a more unconscious level). According to Moscovici, majorities and minorities cause people to focus their attention on different aspects of the situation, ‘‘It would be an overstatement but not a mistake to say that in the face of a discrepant majority all
Majority Versus Minority Influence
243
attention is focused on others, whereas in the face of a discrepant minority, all attention is focused on reality; that, in the first case, the conflict is primarily a conflict of responses, and in the second case, it is a conflict of perceptions’’ (Moscovici, 1980, p. 215). In his conversion theory, Moscovici explicitly states that exposure to a counter-attitudinal minority will result in greater cognitive activity (in terms of analyzing the message) than exposure to a counter-attitudinal majority (where there may be only a superficial analysis of the message content). 2.1.2. Convergent-divergent theory Nemeth’s (1986, 1995) convergent-divergent theory focuses on the different types of thinking styles that can be instigated by majority and minority sources. Nemeth argues that people are motivated to assume that the majority is correct, and therefore when they are exposed to a counterattitudinal majority this is stressful because it suggests that their position is wrong and also they fear disapproval from the majority group. In general, stress is thought to reduce a person’s focus of attention (Easterbrook, 1959) and lead to what is termed convergent thinking which is characterized by a ‘‘. . . convergence of attention, thought, and the number of alternatives considered’’ (Nemeth, 1986, p. 25). In the case of minority influence, the experience of stress is less than is the case with a majority as people believe that the minority is incorrect. As a result of the lower stress, and the minority being consistent and confident in their position, people are stimulated to understand a range of alternative views leading to divergent thinking which involves ‘‘. . . a greater consideration of other alternatives, ones that were not proposed but would not have been considered without the influence of the minority’’ (Nemeth, 1986, p. 25). What is radical about Nemeth’s perspective is that she suggests that the cognitive activity associated with minority influence results in a wider consideration of alternatives than would have been otherwise considered. This is why dissenting voices are so important for our democracies, whether in the legislature or the judiciary; Supreme Court justice Antonin Scalia is so well known for his dissenting opinions that he has earned the nickname the ‘‘Lone Ranger’’ (Talbot, 2005). Research has shown that this wider consideration of alternatives can, in some, but not all, settings result in improved judgments and performance. Indeed, in tasks that benefit from divergent thinking, minority influence has been shown to lead to better performance (e.g., Nemeth and Kwan, 1987; Nemeth and Wachtler, 1983), whereas on tasks that benefit from convergent thinking, a majority source was found to lead to better performance (e.g., Nemeth et al., 1992; Peterson and Nemeth, 1996). In developing her theory, Nemeth has avoided the question of which source condition is associated with more message processing (Nemeth, 2003). The theory is widely interpreted as either showing no difference in the level of cognitive activity between majority and minority influence or as not
244
Robin Martin and Miles Hewstone
specifying the relationship (see De Dreu and De Vries, 1996). However, Nemeth does give a clear indication of the direction of thinking by predicting ‘‘. . . message-specific cognitive activity in response to a majority and issuespecific activity in response to a minority’’ (Nemeth, 1995, p. 284) and stating furthermore that ‘‘. . . cognitive activity induced by the minority is not about the message but, rather, about the issue’’ (Nemeth, 1995, p. 284). The theory argues that both majority and minority influence can lead to cognitive activity but that the direction, or focus, varies; majority influence leads to message processing whereas minority influence leads to issue processing. However, a potential concern here is the distinction between message- and issue-processing and how one can think of issues related to the message without first processing the content of that message. It seems more likely that message-relevant processing is a necessary precursor for issue-relevant thinking. 2.1.3. Objective consensus approach The third main effects model is based on the objective consensus approach (Mackie, 1987) which draws upon concepts developed in the persuasion literature. In direct contradiction to conversion theory, this approach argues that it is a majority source that leads to message elaboration rather than a minority one. There are two reasons for this. First, people assume that the majority view reflects reality in the sense that ‘‘several pairs of eyes are better than one’’ and the majority position ‘‘. . . informs recipients about the probable validity of the arguments presented, directs attention to them, and results in the majority messages receiving considerable processing’’ (Mackie, 1987, p. 50). Second, based on the ‘‘false consensus effect’’ (Ross et al., 1977), people believe that they share similar attitudes to members of the majority group and hold different attitudes from those in the minority group (similar to convergent-divergent theory). As a consequence of these expectations, individuals expect to agree with a majority and disagree with a minority. When faced with a counter-attitudinal majority, the consensus expectation is broken, which is surprising, and this motivates people to analyze the majority arguments in an attempt to understand the difference in opinion. As Mackie proposes, ‘‘. . . disagreement with the majority is likely to induce considerable issue-relevant cognitive activity, thus producing internalization’’ (1987, p. 42). By contrast, exposure to a counter-attitudinal minority is consistent with the consensus heuristic, is therefore not surprising, and consequently one is less likely to process the minority’s message.
2.2. Contingency models Next, we describe four models that propose that majority and minority influence can lead to different processes depending on the presence of one of more contingency variables. These models predict that both majority and minority can lead to message processing, but under different circumstances.
Majority Versus Minority Influence
245
2.2.1. Source-position congruity Based upon the objective consensus approach, Baker and Petty (1994) have proposed that the degree of message processing is determined by the relationship between the source and whether it breaks one’s expectation concerning consensus (i.e., the assumption that our attitudes and opinions are similar to the majority of people and different from the minority of people). When the source/position is expected or ‘‘balanced’’ (pro-attitudinal majority or counter-attitudinal minority), this situation is unsurprising and therefore it is unlikely to lead to deeper message processing. However, when the source/ position is unexpected or ‘‘imbalanced’’ (counter-attitudinal majority or pro-attitudinal minority), this is surprising and it motivates individuals to process the message in order to understand the incongruity. Other reasons why a majority might lead to greater message processing have been outlined by Baker and Petty (1994). People may assume that a majority-endorsed position is more likely to become adopted (or accepted by the population) than a minority-endorsed position, and therefore people believe it would be more important to process the majority’s arguments. Another reason might be that individuals prefer to identify with a majority group rather than a minority group and therefore process the majority message as a guide to what their own attitudes should be. Whereas these factors might lead to elaboration of the majority’s message, Baker and Petty recognize the multiple roles that source status can play and that in some situations (e.g., when elaboration was constrained to be low), majority status might act as peripheral cue leading to agreement without elaboration. Whereas Baker and Petty (1994) propose that the source-position incongruence leads to message processing, Maheswaran and Chaiken (1991) argue that it is source-content incongruence that is most important. They propose that individuals expect the majority to have strong arguments and the minority to have weak arguments (though one could equally argue that people would expect a minority to have strong arguments as it has to be sure of its position in the face of the majority, whereas the power associated with a majority position is such that it need not have convincing arguments). Thus, message processing may occur when the source-content relationship is unexpected (i.e., majority/ weak message or minority/strong message) and not when the relationship is expected (i.e., majority/strong message or minority/weak message). Baker and Petty (1994) (see also Baker, 1992) suggest a third approach involving a two-stage process. The first stage concerns the source-position relationship discussed above. If participants are motivated to process the message (in the imbalanced situation), then the second stage of source-content occurs. Thus, individuals will be more likely to process a pro-attitudinal message from a majority with weak arguments than one from the same source with strong arguments. In balanced situations, the motivation is low but may be sufficient for individuals to examine the nature of the message. If this occurs, then the second stage is reached. The two-stage process predicts
246
Robin Martin and Miles Hewstone
message processing in all conditions except those where the source/position and source/content pairings are expected (i.e., counter-attitudinal minority with weak arguments, and pro-attitudinal majority with strong arguments). 2.2.2. Conflict elaboration theory The second contingency model for analyzing majority and minority influence is conflict elaboration theory proposed by Mugny, Pe´rez, and their colleagues (Mugny et al., 1995; Mugny and Butera, 1995; Pe´rez and Mugny, 1996; see also Quiamzade et al., in press). The model is similar to conversion theory, in proposing that divergence from a source of influence causes conflict. The resulting psychological processes arising from that conflict depend on whether it originates from a majority or minority source. However, unlike conversion theory, which focuses on conflict resolution, this approach considers conflict elaboration, a process that ‘‘. . . refers to the way people give meaning to this divergence’’ (Mugny et al., 1995; p. 161). Unlike conversion theory, conflict elaboration theory specifies a contingency approach specifying that different source characteristics are relevant in different tasks. The nature of the conflict elaboration, and the type of influence, depends on (a) the nature of the task and (b) the nature of the source introducing the divergence. Two contingency variables are associated with the nature of the task. The first concerns the relevance of making an error. If the task is objective with a clearly correct response (with all other responses being wrong), then the cost to the individual of an error is high, whereas if the task is one where objectively correct responses cannot be determined, then the cost of making an error is low. The second dimension concerns whether the responses are socially anchoring. If the response defines the individual in terms of a particular group membership, then it is socially anchoring, whereas if the response does not define an individual in terms of a particular social category, then the task is not socially anchoring. These two dimensions yield four quadrants, each of which is associated with different psychological processes involved in conflict elaboration. Conflict elaboration theory attempts to explain many social influence phenomena, across many social situations, within a single theoretical framework. One of these quadrants is particularly relevant to the present discussion, as it is synonymous with the typical majority and minority influence research scenario that involves subjective tasks (concerning attitudes/opinions) that are socially anchoring.2 The conflict associated with a source in this situation is determined by the meaning attached to it in terms of in-group or out-group membership. When the source is the in-group, as one may assume a majority may be, then normative influence would be 2
Research in other quadrants has, however, proliferated, such as when the task has a clear objective answer and the source introduces a ‘‘conflict of competences’’ (see Butera and Mugny, 1995; Butera et al., 1996; Legrenzi et al., 1991; Quiamzade et al., 2006).
Majority Versus Minority Influence
247
increased and conformity to the majority position will occur with little need to consider the content of the majority message. This is similar to Moscovici’s (1980) description of the comparison process for majority influence. However, it is possible that a majority might lead to a latent influence in a specific situation where there is ‘‘. . . both an intense process of self-categorization and an identification conflict’’ (Pe´rez and Mugny, 1996, p. 204). In this situation, a fear of loss of majority membership or reducing group cohesion can lead to message evaluation. When the source is an out-group such as a minority, and thereby associated with negative connotations, then agreement would be threatening to self-image, and an ‘‘identification conflict’’ arises. Private and indirect influence can occur through a process of dissociation between social comparison and validation, whereby targets of influence resolve the intergroup conflict because ‘‘Only then can subjects focus their attention on the content of the minority position’’ (Mugny et al., 1995, p. 166; see also Falomir et al., 2000; Quiamzade et al., 2003). The Conflict Elaboration Theory, in this specific situation, proposes hypotheses similar to those of Moscovici and suggests that a majority source leads to minimal processing of its message, whereas a minority source (especially when it has out-group status) can lead to detailed consideration of its message especially when social comparison processes are weak. 2.2.3. The context/categorization theory The next theory we will consider in this section is the context/categorization theory developed by Crano and colleagues (Alvaro and Crano, 1997; Crano, 2001, in press; Crano and Alvaro, 1998a,b; Crano and Chen, 1998; Crano and Hannula-Bral, 1994). This model identifies two processes that are important in determining whether there is direct or indirect influence: message elaboration and source derogation. The model proposes a number of contingency factors (including, in-group vs out-group status, subjective vs objective task, involving vs uninvolving issue). The model also makes an important distinction between situations that involve attitude formation (where no prior attitude existed) and attitude change (Crano, in press). The model specifies some situations when a majority or minority source can lead to message elaboration. For example, if the message concerns weak or unvested attitudes, an in-group minority can be persuasive because it is perceived by majority members as being distinctive, and this leads to message elaboration (in a manner similar to that proposed by conversion theory). If the minority is part of the in-group, it is unlikely to be derogated by the majority because the attitude dimension has little implication for in-group membership. Majorities, however, are unlikely to lead to much influence because the majority is not distinctive and therefore does not trigger message elaboration. If the message concerns central or vested attitudes, targets of in-group minority influence are wary of being identified with the minority position, yet there is
248
Robin Martin and Miles Hewstone
a reluctance to derogate other in-group members. This leads to what Crano and Alvaro (1998a) term the ‘‘leniency contract,’’ which allows the target to elaborate on the in-group minority’s message without source derogation, ‘‘. . . open-mindedly, with little defensiveness or hostility’’ (Crano and Alvaro, 1998a, p. 180), and this can lead to indirect attitude change. In the case of out-group minorities, however, the source is derogated, resulting in little direct or indirect influence. If the message relates to important grouprelevant dimensions, there can be compliance without message elaboration leading to influence on both a public or direct level. Majorities can lead to message elaboration when two situations occur. First, when the majority is ‘‘self relevant’’ (i.e., it is important to people’s social identity) and second, when the majority is a ‘‘legitimate’’ source (i.e., the majority is seen as a credible in relation to the message issue). 2.2.4. Dual role model The final contingency model we consider is the Dual Role Model proposed by De Dreu, De Vries, and colleagues (De Dreu, 2007; De Vries et al., 1996). The approach relies upon similar assumptions as the objective consensus approach (Mackie, 1987) and the convergent-divergent theory (Nemeth, 1986) approaches to majority and minority influence described earlier. In the case of majority influence, people find identification with the majority group attractive and therefore, it is both important and relevant that they process the majority’s message to understand their position (and why they hold a position different from the majority). In this case, the majority leads to convergent thinking (i.e., focusing on the specific issues contained in the arguments) resulting in more attitude change on a direct/ public than an indirect/private level. In the case of minority influence, people find identification with the minority group unattractive, and consequently it is neither important nor relevant to understand the minority position. In most cases, the minority position is rejected unless there is some aspect of the context or of the minority itself that triggers systematic processing. Whether a minority source leads to systematic processing relies upon a range of contextual factors (e.g., topic relevance, De Dreu and De Vries, 1996; comparison focus, De Dreu and De Vries, 1993; historical growth of source size, Gordijn et al., 2002) or on minority-specific characteristics (e.g., displaying a positive behavioral style in the way described by Moscovici, 1976) that ‘‘. . . make it difficult or impossible to further neglect the minority’s discrepant message’’ (De Vries et al., 1996, p. 158). If any of the above factors are present (see De Vries et al., 1996 for additional examples), then this encourages people to engage in divergent thinking (i.e., focusing on a wide range of issues other than those presented in the arguments). Because people do not wish to identify with the minority, the results of message processing
Majority Versus Minority Influence
249
will be greater attitude change on an indirect/private than a direct/public level (see also De Dreu et al., 1999).
2.3. Summary of approaches In comparing these theoretical perspectives, one needs to distinguish between the amount and type of cognitive activity proposed for majority and minority influence. We consider predictions for each of these dimensions separately for main effects and contingency models. 2.3.1. Main effects models For the main effects models (see Table 5.1), conversion theory clearly predicts that a minority source will result in a greater amount of cognitive activity than a majority source, whereas the objective consensus approach predicts the opposite. The convergent-divergent theory makes no specific predictions concerning the different amounts of cognitive activity, but suggests that each source leads to considerable cognitive activity (but with a different focus). With regard to the type of cognitive processes, the theories make different predictions according to the source of influence. All three perspectives would agree that a majority can cause influence, but for conversion theory, this would be due to the desired relationship with the source without considering, in depth, the majority arguments. In contrast, both the objective consensus approach and the convergent-divergent theory argue that majority influence is determined by detailed processing of the majority’s message, but the objective consensus approach makes no distinction between message- and issue-relevant thinking, while the convergent-divergent theory predicts that majority influence is determined only by message-relevant thinking. The biggest differences between the theories emerge when we consider minority influence. According to conversion theory, when exposed to a minority, individuals process its arguments and think of pro- and counter-arguments to assess its validity. By engaging in these thought processes, individuals begin to see the logic of the minority’s position and can be influenced by it (assuming that its position has some logic, and its arguments are compelling). The objective consensus approach, by contrast, proposes that the motivation to process the minority position is low and, as a consequence, the minority arguments are not analyzed in detail. Finally, convergent-divergent theory argues that a counter-attitudinal minority induces issue-relevant rather than message-relevant thinking, a systematic consideration of issues associated with the minority position but not stated by it explicitly. 2.3.2. Contingency models In terms of the amount of message processing, the source-position congruity approach predicts that both a majority and a minority can lead to message processing of its arguments in different situations. Both sources will lead
250
Robin Martin and Miles Hewstone
to more message processing when the source-position is imbalanced (i.e., counter-attitudinal majority/pro-attitudinal minority) than when the source-position is balanced (i.e., pro-attitudinal majority/counter-attitudinal minority). The types of processes underlying these effects are similar to those employed in the objective consensus approach—when the situation is imbalanced this breaks the consensus heuristic, which is surprising and counter-intuitive, and this leads to an evaluation of the source’s arguments to understand the imbalance. In the conflict elaboration theory, in the particular situation we identified (subjective tasks where responses define group membership), the prediction is similar to conversion theory: message processing only for the minority source. The context/categorization theory predicts message processing for both a majority (when self-relevant to group membership and it is seen as a legitimate source) and minority (when it has in-group status to the recipients of influence and attitudes are unvested). Finally, the dual role model predicts message processing for both a majority source (when majority group membership is attractive) and the minority (when contextual factors draw attention to its arguments). However, the focus of the message processing differs with majority and minority influence leading to convergent versus divergent thinking respectively resulting in differences in attitude change on different levels of influence. 2.3.3. Summary To summarize, we can categorize the predictions made by models reviewed above in terms of which source they predict will lead people to attend to, and scrutinize, the source’s arguments
only a majority source (objective consensus approach, convergentdivergent theory) only a minority source (conversion theory, conflict elaboration theory when in subjective tasks/socially anchoring quadrant) both majority and minority sources dependent upon contingency factor(s) (source-position congruity model when counter-attitudinal majority and pro-attitudinal minority; context/categorization theory when majority is self-relevant and legitimate, and minority is part of in-group and topic concerns uninvested attitudes; dual role model when majority group membership is highly attractive and minority occurs in context that focuses attention on message and it cannot be ignored).
3. Research Examining Message Processing I dislike arguments of any kind. They are always vulgar, and often convincing. (Oscar Wilde, The importance of being Earnest and other plays. Harmondsworth, UK: Penguin, 1988, p. 310).
Majority Versus Minority Influence
251
A large literature has emerged comparing a wide range of factors related to majority and minority influence. While much of this literature does not directly examine the level of message processing directly, it does give important insights into which source is associated with detailed scrutiny of the arguments. We briefly describe some of the key studies in this literature in relation to: focus of attention, types of thoughts, content of thoughts, recall of source arguments, and outcome of influence (attitude and task performance). A few studies have directly examined differences in the focus of attention following majority and minority influence. Some of these studies give limited support to the notion that perceivers pay more attention to minority than majority sources, also reporting some evidence for greater attention to the majority than the minority source (Campbell et al., 1986; Guillon and Personnaz, 1983; Tesser et al., 1983). A number of studies have examined the types of thoughts people generate following exposure to a majority or a minority (by asking participants to list thoughts they had while reading the source’s arguments). There is some evidence suggesting differences in the quantity and quality of thinking following majority and minority influence, but the results have been inconsistent (e.g., Alvaro and Crano, 1997; De Dreu and De Vries, 1993, 1996; Maass and Clark, 1983; Mackie, 1987; Martin, 1996; Mucchi-Faina et al., 1991). A more successful technique has been to examine the content of thoughts, with some evidence showing that minority influence led to issue-relevant thinking, whereas majority influence led to message-relevant thinking (e.g., De Dreu and De Vries, 1993; De Dreu et al., 1999; Trost et al., 1992). Finally, other experiments have examined the strategies people employ following majority and minority influence. For example, there is evidence that minority influence leads to the use of multiple strategies in solving problems whereas majority influence leads individuals to focus on the majority-endorsed strategy (e.g., Brodbeck et al., 2002; Butera et al., 1996; Legrenzi et al., 1991; Martin and Hewstone, 1999; Mugny and Pe´rez, 1991; Nemeth and Kwan, 1985, 1987; Peterson and Nemeth, 1996). Another way to examine the types of processes involved in majority and minority influence is to examine the recall of the source’s arguments. It might be reasoned that the more a person processes the source’s arguments, then the greater should be the recall of those arguments. Whereas some studies showed greater recall of a minority message (e.g., Moscovici et al., 1981; Nemeth et al., 1990), other studies revealed greater recall for a majority message (e.g., Mackie, 1987; Maass and Clark, 1983; Trost et al., 1992), and some studies showed no difference in recall between a majority and minority (e.g., Alvaro and Crano, 1997). One of the problems in this research is that the recall measures were often taken at the end of the experiment, with many dependent variables intervening between message exposure and the recall, thus making it difficult to determine accurately the effects of source
252
Robin Martin and Miles Hewstone
on recall. Furthermore, the assumption that message recall is an indicator of the extent of message processing has been questioned by some attitude researchers (e.g., Eagly and Chaiken, 1993). Another method to assess the processes involved in majority and minority influence has been to examine the outcome of influence, that is, the extent to which a majority or minority leads to a change in participants’ attitudes or improved performance on a task. One might assume that the consequence of engaging in detailed message processing is greater agreement with the source’s position or improved performance on a task. A large number of experiments have examined the effects of majority and minority influence on attitudes and attitude change. The potential problem with this assumption, however, is that attitude change can be due to public compliance without detailed appraisal of the source’s arguments. Researchers have therefore developed methodologies to examine the impact of majorities and minorities beyond the public/direct level to examine influence on a private/indirect dimension which might be a better reflection of people’s true attitudes. In a review, Maass et al. (1987) identified four dimensions on which attitude change to a majority and minority have been examined: a. time—influence measured immediately following exposure to the source versus influence measured latter in time (e.g., Crano and Chen, 1998; Moscovici et al., 1981) b. specificity—influence specific to the message versus influence that goes beyond the message and considers a wider set of issues. This dimension is commonly referred to as ‘‘direct’’ and ‘‘indirect’’ influence, respectively (e.g., Alvaro and Crano, 1997; Falomir et al., 2000; Mackie, 1987; Moscovici et al., 1981; Mugny and Pe´rez, 1991) c. privacy—responses which are made in public versus those that are made anonymously and in private (e.g., Maass and Clark, 1983; Martin, 1988a,b) and d. awareness—participants are aware of the connection between source message and influence dimension versus not being aware of this connection (e.g., Brandsta¨tter et al., 1991; Martin, 1998; Moscovici and Personnaz, 1980, 1991; Mugny, 1984). Overall, research tends to support conversion theory’s prediction of greater public/direct influence for a majority and greater private/indirect influence for a minority (Martin et al., 2008b). A meta-analysis of 97 studies by Wood et al. (1994) also supported this conclusion, concluding that ‘‘Minority impact was most marked on measures of influence that were private from the source and indirectly related to the content of the appeal and less evident on direct private influence measures and on public measures’’ (p. 323). Furthermore, minorities had as much, if not more, impact on indirect measures of influence as did majorities.
Majority Versus Minority Influence
253
Another set of studies has examined the effects of majority and minority influence on task performance. Most of these experiments test Nemeth’s (1986) prediction that minority influence leads to divergent thinking, which entails a consideration of a wide range of issues related to the topic of the source’s message, and this can lead to more original and creative thinking. The results of many experiments tend to support Nemeth’s theory. In experiments where task performance benefits from divergent thinking, minority influence leads to better performance than majority influence (e.g., Martin and Hewstone, 1999; Nemeth and Kwan, 1987; Nemeth and Wachtler, 1983), whereas on tasks where performance benefits from convergent thinking, majority influence leads to better performance than minority influence (e.g., Nemeth et al., 1992; Peterson and Nemeth, 1996). Finally, minority influence has been shown to lead to more creative and novel judgments compared to majority influence (e.g., Mucchi-Faina et al., 1991; Nemeth and Kwan, 1985; Nemeth and Wachtler, 1983; Volpato et al., 1990; see also Martin, 1996). To summarize, a range of different methodologies have been employed to examine majority and minority influence with the aim of determining the underlying processes. However, most of these studies did not manipulate variables to examine directly the amount of message processing. A more useful methodology to identify whether message processing has occurred, developed in the persuasion area, is to vary the quality of the arguments in the message (see Petty and Cacioppo, 1986). Manipulating argument quality to determine message processing was first employed by Petty et al. (1976). If individuals process a message in detail, that is they pay close attention to the content of the arguments, they should be more influenced by strong and persuasive arguments than by weak and nonpersuasive arguments—we refer to this as the argument quality effect. By contrast, if individuals do not process the message, and therefore do not attend to the content of the arguments, then argument quality should have a small (or no) impact on attitudes. By crossing source status with argument quality, one should be able to detect whether either, both, or neither source leads to systematic processing of the message. A reliable interaction between source status and argument quality can indicate different processes underlying majority and minority influence processing (see Kruglanski and Mackie, 1990). We now present a detailed review of studies that have manipulated both source status and argument quality to determine which source is associated with message processing. Our selection criteria for the review were that the experiment (a) was concerned with a social attitude or opinion, (b) manipulated both source conditions (majority vs minority), and (c) exposed participants to a message containing either strong or weak arguments as a within- or between-subjects design. Some studies that did not fit these criteria, principally because they did not manipulate both source conditions (e.g., Baron and Bellman, 2007; Gordijn et al., 2002), or they exposed
254
Robin Martin and Miles Hewstone
participants to both majority- and minority-endorsed arguments, making it difficult to attribute attitude change to a particular source (e.g., Garlick and Mongeau, 1993), are not discussed. In presenting the review, we focus mainly on the findings concerning argument quality. The reader should be forewarned that the review contains many conflicting findings and that we try to make sense of these apparently inconsistent findings when we present our model in the next section. In presenting the review, we have sorted the studies into those showing first, message processing only for a majority source; then, message processing for both a majority and minority; and finally, message processing for a majority and minority under different situations.
3.1. Message processing for majority source only A number of experiments have found evidence to suggest that only the majority leads to detailed message processing. Baker and Petty (1994, Experiment 1) had university students read either a strong or weak counterattitudinal message concerning a ‘‘. . . mandatory service program in which university students must commit 2 years of community service in exchange for maintaining current tuition rates. If students declined to participate, their tuition rates would automatically increase threefold’’ (p. 8). Participants were told that either 78% (majority) or 12% (minority) of students in universities in their state supported the proposal. After indicating their attitude towards the message, participants completed a thought-listing task whereby they listed all the thoughts they had while reading the message (after this they indicated whether each thought was positive, negative, or neutral towards the topic). There was a reliable interaction between source status and argument quality on posttest attitude scores. While there was no difference between strong and weak messages in the minority condition, there was a tendency for the strong message to have greater influence than the weak message in the majority condition. The thought-listing data were used to calculate a messagecongruent thought index (ratio of positive to positive and negative thoughts, such that higher scores show a greater proportion of thoughts in the direction of the message); however, this index did not show a reliable interaction, although the pattern was similar to the attitude data. Gordijn (1998, Experiment 1) gave Dutch University students a message in favor of courses in the university being taught in the English language. The message was attributed to either a numerical majority or minority of fellow students, and participants read either strong or weak arguments. Gordijn found a reliable interaction between source status and argument quality for posttest attitudes. The interaction showed a reliable difference between the strong and weak messages (argument quality effect) in the majority condition, but not the minority condition, suggesting participants had processed only the majority message.
Majority Versus Minority Influence
255
3.2. Message processing for both majority and minority source Several experiments have found message processing for both a majority and a minority source. For example, Bohner et al. (1998) had participants read a report of a meeting where a communicator gave either strong or weak arguments that were anti-animal experimentation (which was known to be counter-attitudinal to the participants) and they were told that either 80% (majority) or 20% (minority) of people held the same view as the person whose arguments they read. The researchers also manipulated the ‘‘distinctiveness’’ of the source’s minority (majority) position by informing recipients either that the source usually held minority (majority) positions with respect to most topics (low distinctiveness), or that the source’s minority (majority) position was unique to the current topic (high distinctiveness). On the basis of Kelley’s (1967) attribution theory, Bohner et al. proposed that high distinctiveness might ‘‘. . . increase the likelihood of attributions to a position’s external validity, and . . . lead to more positive source-related attributions and judgments’’ (p. 856) and presumably increase message processing. The results showed that attitudes were affected by both argument quality and distinctiveness (high > low) for both a majority and minority source. Crano and Chen (1998, Experiment 3) had participants read a message in favor of a university service program, which contained either strong or weak arguments (a topic and messages employed extensively in the persuasion literature; see Petty and Cacioppo, 1986). Outcome relevance was manipulated by telling participants that the new program would be introduced either the following year (high relevance) or in 8 years time and therefore after participants had completed their course (low relevance). Attitudes towards the issue of a university service program were measured at three times: before message exposure, immediately after message exposure, and 1 week after message exposure. Unexpectedly, outcome relevance did not affect the attitude scores. There was, however, a reliable argument quality effect for both the majority and minority conditions, suggesting that participants had systematically processed the message for each source. In addition, there was a reliable source effect, showing that the majority had greater impact than the minority (the interaction between source status and argument quality was not reliable). When examining the attitude scores taken 1 week after the experiment, there was no effect of argument quality, but the source effect had persisted. The final two experiments, by Erb et al. (1998), differed from others in the literature in two respects: first, they manipulated argument quality as a within-subjects factor; second, the researchers deliberately used issues on which participants would have no prior attitude (see Erb and Bohner, 2007); thus, the studies can best be classified as dealing with attitude
256
Robin Martin and Miles Hewstone
formation rather than change. In the first experiment, German students read a message that was in favor of building a tunnel under Rotterdam harbor in the Netherlands. The position was presented as being either high (85%) or low (15%) in consensus. The message contained six arguments; from pretesting one item was considered to be strong, because it was highly persuasive (concerning resident issues) whereas another was considered to be weak and unpersuasive (concerning economy issues). Attitudes were assessed towards the building of the tunnel and also the specific items for which strong and weak arguments were presented in the message. Overall, the majority led to more positive attitudes towards the building of the tunnel than did the minority. Further analyses showed a reliable argument quality effect (in this case a difference between attitudes towards the issue raised in the strong and weak arguments) for the majority, minority, and control (where no consensus information was given) conditions. The lack of a reliable interaction between consensus condition and argument quality showed that the difference between strong and weak arguments was the same in all three experimental conditions. A measure of thought valence obtained from the thought-listing task (number of pro-message thoughts minus number of anti-message thoughts) mediated the relationship between consensus level and general attitudes. A second experiment (Erb et al., 1998, Experiment 2) concerning the development of a new holiday resort area in Brazil, also manipulated the order of presentation of whether source information preceded or followed the message, and replicated the finding (irrespective of whether the source information had been given before or after reading the message). As was the case in the first experiment, thought valence mediated attitudes when consensus was given before the message.
3.3. Message processing for either majority or minority in different situations In this final section, we review a number of experiments showing message processing for both a majority and minority source, but under different situations. De Dreu and De Vries (1993, Experiment 2) were the first researchers to cross source status and argument quality. They also manipulated the level of evaluation comparison of the message. Dutch college students were exposed to a counter-attitudinal message that was in favor of the introduction of admission exams for university. The participants were told that the message was supported by either 82% (majority) or 18% (minority) of the population and they read a set of strong or weak arguments concerning the introduction of admissions exams for university. The information processing manipulation concerned how participants evaluated the source’s message. Participants judged how ‘‘correct’’ they thought their own attitude and that espoused in the message were in one of two different ways. In one condition, they divided a score of 100 ‘‘points
Majority Versus Minority Influence
257
of correctness’’ between their own and the message position (comparative evaluation condition) which the authors reasoned should increase comparison between self and source, necessitating the processing of the source message. In the other condition, participants allocated a score out of 100 ‘‘points of correctness’’ for both self and source, independently (noncomparative evaluation condition) which should not encourage comparison, and therefore participants should be able to avoid systematically processing the source’s message. Attitudes towards the topic were measured before and after message exposure and the results are reported as change scores. The reliable three-way interaction between the independent variables showed a differential pattern of attitude change between the evaluation conditions. In the comparative evaluation condition, there was a reliable argument quality effect in the minority (strong > weak) but not the majority condition. In the noncomparative evaluation condition, the opposite pattern was observed: a reliable argument quality effect for the majority but not the minority source. Notwithstanding the important contribution of this experiment, we raise a methodological issue concerning the impact of the strong/weak arguments. The pattern of attitude change scores showed that three of the eight experimental cells showed a negative change (indicating they were less in favor of admissions exams than they were at pretest) whereas two further cells showed no change. This leads us to wonder whether the strong and weak arguments did, in fact, have a differential impact on attitudes. Furthermore, the findings in the critical conditions of comparative/minority and noncomparative/majority showed a different pattern between strong and weak messages (due to a large negative change for the weak message in the former, and a large positive change for the strong message in the latter); these results suggest that the underlying processes might not have been the same. Baker and Petty (1994, Experiment 2) reported an experiment that crossed source status with argument quality and also manipulated the source-position congruity. As described earlier, Baker and Petty argue that participants are more likely to process a message systematically when the source-position is incongruent or ‘‘imbalanced’’ (i.e., counterattitudinal majority/pro-attitudinal minority) than when source-position is congruent or ‘‘balanced’’ (pro-attitudinal majority/counter-attitudinal minority). To address the issue of balanced versus imbalanced situations, Baker and Petty (1994, Experiment 2) conducted an experiment that crossed counter- and pro-attitudinal messages with majority and minority status. The topic and message were similar to those used in the first experiment described in an earlier section. The pro-attitudinal message argued that performance of university services would lead to a full tuition break (message heading, ‘‘Majority [or Minority] Favors Tuition Break,’’ p. 12) whereas the counter-attitudinal message proposed an increase in tuition for those that did not participate (message heading, ‘‘Majority [or Minority]
258
Robin Martin and Miles Hewstone
Favors Tuition Increase,’’ p. 12). Given that participants in a pilot study thought that fewer than 55% of people would support the pro-message position, and that the mean pretest agreement was only marginally above the scale midpoint (M ¼ 5.44 on a 9-point scale), assumptions of majority/ minority-held positions were far from universal (a point supported in the source manipulation check for the main study where perceived support for the pro-message was only 54.5%, although 78% had, in fact, in the experimental manipulation supported this position). The three-way interaction between the independent variables was reliable. The results for post-influence attitudes supported Baker and Petty’s predictions. There was a reliable argument quality effect in the imbalanced conditions for both majority and minority sources, and no effect of argument quality in the balanced conditions. The results for the proportion of positive thoughts supported the attitude data. In the balanced conditions, there was no effect of argument quality in the proportion of positive thoughts. In the imbalanced conditions, as expected, argument quality predicted the proportion of positive thoughts, which in turn predicted attitudes. This finding further supports the idea that when the situation is imbalanced, for either a majority or minority, this leads to message processing which affects attitudes. Shuper and Sorrentino (2004) conducted an almost exact replication, using the same messages in a sample of Canadian students. However, they found no reliable effects of the independent variables on attitude scores. Like the experiment by Crano and Chen (1998, Experiment 3) reported earlier, Kerr (2002) examined outcome relevance (low vs high), and manipulated a second factor termed source advocacy (active vs passive). The topic of the study was the familiar one of senior comprehensive exams for university students (cf. Petty and Cacioppo, 1986). Outcome relevance was manipulated by informing participants that these exams would be implemented before they graduated and therefore would apply to them (high relevance), or in 4–5 years time and therefore after they graduated (low relevance). Source status was manipulated in a novel way—participants were told either that the message was supported by a large majority (87%) of students at the university (majority) or that it was opposed by a large majority (87%) of students at the university (minority). Normally the minority position is defined by having a low percentage of support, and it is unusual to define the minority with reference to having the opposite position to the majority. For participants in the low relevance condition, where one would not expect the topic to induce message processing, there was no effect of argument quality for the majority but there was for the minority—especially when the minority was ‘‘active’’ in orientation (i.e., ‘‘are aware of the relative popularity of their positions, are interdependent with others in the group, and do expect to interact with others in the group in a context
Majority Versus Minority Influence
259
requiring them to reveal their opinion’’ p. 472). In the high outcome relevance condition, where participants should be motivated to process the message, there was an argument quality effect for both source conditions, indicating that message processing had occurred, and a trend towards a source effect, the majority having more influence than the minority (replicating Crano and Chen, 1998, Experiment 3). Erb et al. (2002) reported two studies showing that under high conflict, minority messages were processed more extensively than majority messages, whereas under low conflict, the opposite was true. Experiment 1 examined strong and weak arguments as a within-subject variable, as they did in their previous research (Erb et al., 1998), and also examined participants’ prior attitude. German students were exposed to a message that advocated the building of a road tunnel under the river Rhine, the construction of which was planned for the following year. Initially, participants read a summary of the topic to induce either an opposing attitude or a moderate attitude towards the topic. The source information was manipulated by a report of agreement with the tunnel from a local meeting on the issues—either 87% (majority) or 13% (minority). The message contained five arguments; the second argument (concerning advantages for residents) was strong whereas the fourth argument (concerning economic benefits) was weak. There was a reliable interaction between the independent variables (prior attitude, source status, argument quality). The pattern of means suggests that argument quality had a bigger impact for the minority than the majority when participants held opposing attitudes to the source and for the majority than the minority when they held moderate attitudes to the source. The final experiment was also conducted by Erb et al. (2002, Experiment 2). German participants were exposed to a message advocating the fluoridation of drinking water. Pilot work had shown that participants were either strongly against or moderate towards this issue which provided a naturally occurring situation in which to examine the moderating role of opposing versus moderate attitude on majority and minority influence. The participants were informed about a discussion on the topic by representatives of public health insurance plans where either 88% (majority) or 12% (minority) favored the fluoridation of drinking water, and then read a set of strong or weak arguments. The predicted three-way interaction between the independent variables (prior attitude, source status, argument quality) was reliable. The results supported the authors’ prediction. When participants held an opposing attitude, there was an argument quality effect for the minority, but not the majority source. However, when the participants held a moderate attitude, the opposite pattern was observed; there was an argument quality effect for the majority, but not the minority source. A similar three-way interaction was also found for a measure of thought valence, and the comparison between strong and weak messages was similar to the attitudes data.
260
Robin Martin and Miles Hewstone
In conditions where message processing was predicted (opposing attitude/ minority and moderate attitude/majority), the level of thought valence mediated attitudes. However, no mediation was found in conditions where message processing was not predicted (opposing attitude/majority and moderate attitude/minority). This is compelling evidence that the argument quality effect was due to the sources differentially affecting participants’ elaboration of the message which, in turn, affected their attitudes.
3.4. Integration of findings The above review has described several experiments that have crossed source status and argument quality to determine if either, neither, or both source conditions lead to systematic processing of its message. A variety of findings have shown message processing, only for a majority, for both a majority and minority, and for each source under different conditions. This pattern of results does not give clear support for any of the models described earlier. We have proposed that many of these seemingly inconsistent findings might be explained with reference to the types of topics and experimental manipulations employed (Martin and Hewstone, 2003b). Interestingly, those experiments that found an argument quality effect only for a majority source, suggesting that message processing had occurred (supporting the objective consensus approach), employed messages about a topic that had a very high impact and would be seen as being undesirable both to the source and the message recipients: delivering courses to Dutch University students in English (Gordijn, 1998, Experiment 1), a mandatory community service program to maintain tuition rates for undergraduate student participants (Baker and Petty, 1994, Experiment 1). As Martin and Hewstone (2003b) note, such topics might introduce a negative personal outcome. Under the atypical situation when a negative personal outcome is proposed, it is not surprising that a counter-attitudinal majority arguing for a negative outcome invoked greater message scrutiny than did a minority. When a majority endorses such a position, it is likely to raise suspicion or curiosity in the participants (‘‘why are the majority arguing against my, and their, interest?’’) and this may result in closer examination of the message. When a minority argues against its self-interest for a negative outcome, however, its message may not be scrutinized for a number of reasons: such as, the minority’s deviance might be attributed to an underlying bias; because minority-supported positions are unlikely to be adopted; or to protect one from identifying with a (deviant) minority group. For those experiments finding an argument quality effect for both a majority and minority, suggesting that both sources led to systematic message processing, most employed experimental instructions that potentially increased participants’ motivation to process the source’s arguments.
Majority Versus Minority Influence
261
Crano and Chen (1998, Experiment 3) manipulated outcome relevance by telling participants that a university service program would be introduced either the following year or in several years time. Unexpectedly, outcome relevance did not affect the attitude scores. The authors reasoned that ‘‘. . . the manipulation was superfluous. By virtue of the topic’s relevance for their social identities, participants were already involved in the university service issue well in advance of the manipulation . . . Natural involvement was already high, thereby rendering the manipulation inconsequential.’’ (p. 1448). This conclusion is supported by the results of the outcome relevance manipulation. Although there was a reliable difference between the low versus high relevance conditions, the mean for the low relevance condition was very near the scale midpoint—participants in both outcome relevance conditions believed the issue was personally relevant to them. The Erb et al. (1998, Experiments 1 and 2) studies included task instructions that explicitly encouraged message processing. In the first experiment, the participants were told ‘‘. . . to read the text attentively because later they would be asked some questions about it’’ (p. 623), whereas in the second experiment they were told ‘‘. . . to evaluate two texts as part of a study on text comprehension and were asked to read the texts attentively to ensure systematic processing’’ (p. 627). By directing participants to carefully attend to the arguments, the experimenters encouraged systematic processing of the message for both source conditions. We now turn to those experiments finding message processing for a majority and minority source in different situations. De Dreu and De Vries (1993, Experiment 2) employed a message that could be seen to propose a negative personal outcome as described earlier (in favor of the introduction of admission exams for university). On the basis of the above analysis, we would predict that when the source argues for a negative personal outcome, there should be systematic processing only for the majority source. The experiment also included a manipulation that did or did not encourage participants to compare their own and the source’s position (comparative and noncomparative conditions). In the noncomparative condition, De Dreu and De Vries found an argument quality effect only for the majority condition, which is consistent with other experiments employing a topic arguing for a negative personal outcome (Baker and Petty, 1994, Experiment 1; Gordijn, 1998, Experiment 1). However, in the comparative condition, where participants’ awareness of their own position, and how it differs from the source position, would be salient, one might expect participants to try to resist influence from the source (indeed there was less attitude change in the comparative than the noncomparative conditions). In the comparative condition, the only experimental condition that differed from the others was the minority/ weak message combination which showed a large change away from the minority’s position.
262
Robin Martin and Miles Hewstone
Baker and Petty (1994, Experiment 2) manipulated source/position congruency by creating balanced versus imbalanced settings. This was achieved by varying the direction of the message (pro- vs counterattitudinal), by reversing the position advocated in the message (concerning students performing university services). However, the experiment not only varied the direction of message (pro- vs counter-attitudinal), but also the desirability of outcome. The authors state that the pro-attitudinal position ‘‘. . . seemed as if it would only have benefits for students’’ (p. 12) and because the counter-attitudinal message took the opposite position, it would be against the participants’ interests (a negative personal outcome). With the pro-attitudinal message (in favor of participants’ interest), there was message processing only for the minority source. However, with the counter-attitudinal message, that might have induced the negative personal outcome, the results were consistent with the experiments discussed above, showing message processing only for the majority source. The final experiments by Erb et al. (2002, Experiments 1 and 2) show that the initial attitude, in relation to the source, moderated when message processing would occur. When participants initially had opposing attitudes there was message processing only for the minority source but when they held moderate attitudes there was message processing only for the majority source. However, in these experiments, it is not clear what factor actually moderated message processing. There are many differences between people who hold opposing attitudes compared to those who hold moderate attitudes, other than their prior attitude position (such as, attitude strength, certainty, centrality—to name a few) that might affect information processing strategies. Indeed, as noted by Stroebe (in press), there is no theoretically compelling reason why people with a moderate attitude should process a majority message more than a minority one. To summarize, the above analysis shows that some of the inconsistencies in the literature can be potentially explained by taking into account the topic of influence and the experimental procedures employed. In experiments that employed a topic that might have proposed a negative personal outcome, there was message processing only for a majority source (Baker and Petty, Experiment 1; Baker and Petty, Experiment 2, counterattitudinal message; Gordijn, 1998, Experiment 1). In experiments that did not induce a negative personal outcome, there was message processing only for the minority source (Baker and Petty, 1994, Experiment 2, proattitudinal message). In experiments finding message processing for both a majority and minority, the topics used were high in personal relevance (e.g., increase in tuition fees, Crano and Chen, 1998, Experiment 3; Kerr, 2002), or the studies deliberately used instructions that explicitly encouraged systematic processing of the source’s arguments (e.g., Erb et al., 1998, Experiments 1 and 2). Interestingly, in all these experiments, the predicted argument quality effect was found for both majority and
Majority Versus Minority Influence
263
minority source conditions, and there was also a main effect for source status (majority > minority). We recognize that this is a post hoc interpretation of these findings and that not all experiments fit in with the analysis (e.g., Bohner et al., 1998). However, there does seem some support for the view that when the situation does not introduce any processing biases (such as, negative personal outcome, high personal relevance) there is message processing only for a minority (supporting conversion theory). Majorities can also lead to message processing, and therefore attitude change, but only when the situation introduces a factor that motivates people to pay closer attention to the content of the majority’s arguments.
4. The Source-Context-Elaboration Model of Majority and Minority Influence You see, it is a very dangerous thing to listen. If one listens one may be convinced . . . (Oscar Wilde, An ideal husband; reprinted in Oscar Wilde, The importance of being Earnest and other plays. Harmondsworth, UK: Penguin, 1988, p. 163).
Our review of the relevant literature suggests that both a majority and minority can lead to message processing under different contexts. However, it appears that for majorities to lead to message processing, there must be some aspect of the context that motivates people to pay close attention to the content of their arguments. It is the interplay between source status and the contextual factor(s), that might encourage or constrain message processing, that forms the central theme of our theoretical model. We refer to our theoretical framework as the source-context-elaboration model (SCEM). The model predicts that a majority and minority can lead to message elaboration in specific contexts. The model proposes a contingency framework where message elaboration is contingent upon source status (majority vs minority) and the ‘‘context’’ in which influence occurs, which includes both cognitive and motivational factors. Our theoretical framework attempts to specify (a) the types of processes involved in majority and minority influence, (b) the contexts under which these processes occur, and (c) the consequences of these processes for people’s attitudes. Each of these aspects of the framework is described below.
4.1. Processes Our model is based on an integration of aspects of Moscovici’s (1980) conversion theory and dual-process models of persuasion. We turn to dual process models of persuasion (Chaiken et al., 1989, Heuristic
264
Robin Martin and Miles Hewstone
Systematic Model, HSM, and Petty and Cacioppo’s, 1986, Elaboration Likelihood Model, ELM) as potentially providing a detailed analysis of the underlying processes involved in majority and minority influence (Maass and Clark, 1983). Although the ELM and the HSM make some different predictions (see Eagly and Chaiken, 1993; Visser and Cooper, 2003), they share some common assumptions. Both the ELM and HSM distinguish two strategies of information processing in persuasion settings. ‘‘Central-route persuasion’’ (ELM) or ‘‘systematic processing’’ (HSM) entails thinking carefully about persuasive arguments and other issue-related information. Systematic processing refers to a ‘‘. . . comprehensive, analytic orientation in which perceivers access and scrutinize all information input for its relevance and importance to their judgmental task’’ (Chaiken et al., 1989, p. 212); it can be biased or unbiased, depending upon motivational factors (processing goals). In order to engage in systematic processing, people have to be both motivated and able to process the source’s message. In situations where people are unmotivated and/or unable to process the source’s message, attitudes may be changed by ‘‘peripheral-route persuasion’’ (ELM) or ‘‘heuristic processing’’ (HSM), whereby systematic processing is minimal, and persuasion occurs due to some cue(s) in the persuasion environment (e.g., status of source) or use of simple heuristics (e.g., ‘‘the majority is always right’’). Heuristic processing refers to a ‘‘. . . more limited processing mode that demands much less cognitive effort and capacity than systematic processing . . . [it involves a] focus on that subset of available information that enables [respondents] to use simple, inferential rules, schemata, or cognitive heuristics to formulate their judgements and decisions’’ (Chaiken et al., 1989, p. 213). Maass and Clark (1983) were the first to draw a parallel between Moscovici’s concepts of comparison and validation and those of nonsystematic and systematic processing respectively. Table 5.2 gives some of the original definitions of key processes proposed in conversion theory, the ELM and the HSM, although it should be acknowledged that there has been (understandably) some variation in these definitions over time. We have organized the definitions into those that predict elaboration of the source’s arguments and those that do not. With regard to processes that describe message elaboration (validation/central/systematic processing), there are common themes concerning (a) evaluation of the merits of the source’s arguments and (b) assimilation of the results of the evaluation to preexisting attitudes. In this situation, attitude change is a direct consequence of engaging in message-relevant thinking—the greater the message-relevant thinking, the greater is the corresponding attitude change. With regard to processes that do not involve message elaboration (comparison/peripheral/ heuristic processing), there is somewhat less agreement amongst the theories but there are some common themes concerning (a) attention being drawn away from the content of the source’s arguments and (b) inferences are
Table 5.2
Definition of key processes Attention to message content Low
High
Conversion theory (Moscovici, 1980)
Comparison process: [the target of influence will] ‘‘. . . concentrate all his attention on what others say, so as to fit in with their opinions or judgments’’ (p. 214)
Elaboration Likelihood Model (ELM: Petty and Cacioppo, 1986)
Peripheral processing: based on ‘‘. . . superficial analyses of the veracity of the recommendation. Acceptance or rejection of the appeal is not based on the careful consideration of issue-relevant information . . . but rather it is based on the issue or object being associated with positive or negative cues . . . or with the individual’s drawing a simple inference based on various cues in the persuasion context’’ (p. 13) Heuristic processing: involves a ‘‘. . . focus on that subset of available information that enables [respondents] to use simple, inferential rules, schemata, or cognitive heuristics to formulate their judgments and decisions’’ (p. 213)
Validation process: ‘‘. . . an examination of the relation between its responses and the object of reality . . . One will at the same time examine one’s own responses, ones own judgments, in order to confirm and validate them’’ (p. 215) Central processing: ‘‘. . . directed at evaluating the merits of the arguments for a recommendation (p. 12) . . . typically result in new arguments, or one’s personal translations of them, being integrated into the underlying belief structure (schema) for the attitude object’’ (p. 12)
Heuristic Systematic Model (HSM: Chaiken et al., 1989)
Systematic processing: ‘‘. . . a comprehensive, analytic orientation in which perceivers access and scrutinize all informational input for its relevance and importance to their judgment task, and integrate all useful information in forming their judgments’’ (p. 212)
266
Robin Martin and Miles Hewstone
made concerning the validity of the arguments. In this situation, attitudes are changed via linking the source position to a positive external cue, but because no message scrutiny has taken place, the attitude change is superficial and there is no change in underlying attitudes. We do not claim there is a perfect parallel between these processes but suggest that there is sufficient similarity to make some integration, in the context of majority and minority influence, worthwhile. On the basis of our integration of the above theories, and as applied to our model of majority and minority influence, we envisage two major forms of message processing: elaborative and nonelaborative processing. Elaborative processing involves five main stages: (a) attending to the content of the source’s arguments and trying to understand the underlying reasons why the source believes these arguments, (b) generating pro- and counter-arguments, (c) evaluating the merits of these arguments in the light of preexisting attitudes, (d) assimilating these new arguments into the attitude object, and (e) being aware of the consequences of these ‘‘new’’ attitudes to source group membership (and, when necessary, redefining group identification). By contrast, nonelaborative processing involves none of the above stages and when it occurs preexisting attitudes remain intact. Sometimes nonelaborative processing can guide attitudes by a heuristic cue. Whether source status will act as heuristic cue depends upon the extent to which source group membership is psychologically important; that is, when majority group membership is desirable (e.g., it forms part of the person’s in-group) and/or minority membership is undesirable (e.g., it has deviant status). If majority/minority group membership is not psychologically meaningful, then we do not expect people to engage in elaborative processing, and existing attitudes are not changed. Because information concerning source status typically precedes message exposure, then majority or minority endorsement will be a major determinant of the type of information processing people will engage in. Moreover, source status is likely to affect many aspects of the elaboration processes. For example, it could affect the focus of elaborative processing (on attitudes central to the source’s message and/or of attitudes related to the source’s message even when not mentioned in it); stimulate a range of processing goals (such as, accuracy, defence, and impression); lead to selective attention to different aspects of information (Frey, 1986; Nemeth and Rogers, 1996; Schulz-Hardt et al., 2000); and affect the recipients’ level of confidence in their attitudinal judgements (what the HSM refers to as the ‘‘sufficiency principle,’’ see Maheswaran and Chaiken, 1991). There are two guiding principles concerning when people are influenced by a majority or minority. First, in many situations where there is a choice between agreeing with the majority or with a minority, most people will choose the former—which we refer to as the majority default option. This is because it is psychologically easier to comply with a majority-endorsed
Majority Versus Minority Influence
267
position than it is to agree with a minority one. Most people wish to belong to majority groups, both for social acceptance and for consensual validation of attitudes and opinions, and also to avoid being categorized as part of a deviant minority. Agreeing with a minority intuitively entails rejecting majority-group status, expending considerable cognitive effort to assimilate the minority’s arguments into one’s attitude structure, and also becoming prepared to defend the minority position against the majority. In most situations, majority status can act as a heuristic cue (e.g., through assumptions that ‘‘several pairs of eyes are better than one,’’ ‘‘majority more likely to be correct,’’ ‘‘majority views prevail,’’ and so on) and this will lead people to comply with the majority, without changing their underlying attitudes, to satisfy their desire to be part of the majority group and for consensual validation of their opinions. While we believe that in most situations majority status might act as a heuristic cue, we can also envisage some, albeit rare, situations when minority heuristics might occur. Examples would be elite or high status minorities, minorities with lofty positions and/or special knowledge concerning the topic of influence, or when prior majority-endorsed decisions have proved unsuccessful. The second guiding principle states that minorities will only lead to elaboration of their arguments, and therefore have influence, if they are ‘‘distinctive’’ in the eyes of the majority, and they are consistent and committed to their position—the minority behavioral style threshold. Numerous studies show the importance of the minority’s behavioral style, especially consistency, in determining the ability of the minority to influence the majority (see section on conversion theory above, Moscovici, 1976; Moscovici and Lage, 1976; Moscovici and Nemeth, 1974; Mugny, 1982 and also meta-analysis by Wood et al., 1994). In the absence of external factors increasing elaboration, the minority will not influence the majority unless this threshold has been exceeded.
4.2. Contexts affecting when processes occur The second part of our theoretical framework specifies the ‘‘contexts’’ under which majority and minority influence will lead to either elaborative or nonelaborative processing. In common with dual-process models of persuasion, we propose that whether a source will lead to elaborative versus nonelaborative processing will depend upon where the context falls on an ‘‘elaboration continuum’’ (Petty and Cacioppo, 1986). The position on the elaboration continuum is determined by a range of dispositional (e.g., need for cognition, Cacioppo and Petty, 1982) and situational (e.g., relevance of topic, Johnson and Eagly, 1989) factors that affect people’s ability and/or motivation to engage in elaborative processing. In short, the elaboration continuum goes from ‘‘. . . no thought about the issue-relevant information presented, to complete elaboration of every argument, and complete
268
Robin Martin and Miles Hewstone
integration of these elaborations into the person’s attitude schema’’ (Petty and Cacioppo, 1986, p. 8). But how can one determine where a person is on the elaboration continuum? According to Petty et al. (2005), ‘‘. . . the location of the target of influence along the continuum is determined by a person’s overall ability and motivation to think about the issue, object, or person under consideration.’’ Although the elaboration continuum is conceived to be a continuous variable, most research on the ELM has examined situations at the extremes of the elaboration continuum, from virtually no thinking about the message (low elaboration) to very detailed consideration of the content of the message (high elaboration) (Petty and Cacioppo, 1986). Relatively little consideration has been given to the situation between low and high levels of the continuum (intermediate elaboration) which is interesting precisely because it is the situation in which most persuasion situations occur (cf. Petty et al., 1999). Clearly, the definition of ‘‘intermediate’’ elaboration is based upon the target’s ability/motivation to think about the message and it is a relational one—it is between low and high elaboration situations. The ELM provides some insights into the potential effects of source status at different levels of the elaboration continuum (see Petty and Cacioppo, 1984). Petty and colleagues referred to the ‘‘multiple roles’’ of majority/minority source status (Baker and Petty, 1994; Petty and Wegener, 1998). More specifically, they suggest that the effects of majority versus minority source status will vary along the elaboration continuum. For example, Petty and Wegener (1998) state that ‘‘. . . when the elaboration likelihood is low, majority/minority source status is most likely to serve as a simple cue . . . when people are unsure whether they should carefully scrutinize the message or not, majority/minority status can determine the amount of message scrutiny . . . when motivation and ability to process an incoming message are high, majority/minority status should impact persuasion primarily by influencing the nature of the thoughts that come to mind’’ (pp. 347–348). Thus, majority/minority source could influence attitudes as a peripheral cue (at low levels of motivation and/or ability), as a factor that influences the amount of scrutiny of the message (at more intermediate levels of motivation and/or ability—although Petty and Wegener do not specify which source would have this affect), and as an argument, or through biased processing (at high levels of motivation and/or ability). Drawing together these theoretical ideas, we make the following predictions concerning the effects of source status (majority vs minority) at three levels of the elaboration continuum (low vs intermediate vs high, see Table 5.3). When the elaboration context is low (e.g., the topic is low in personal relevance, presence of a distractor task), the motivation and/or ability to process the message is correspondingly low, leading to nonelaborative processing. In this context, attention focuses on the characteristics of the source and two possible outcomes can occur. If majority/minority
Table 5.3
Key comparisons concerning source-context elaboration model Elaboration situation Low
Intermediate
High
Source status
Process
Majority
Minority
Majority
Minority
Nonelaborative/ Heuristic
Nonelaborative
Nonelaborative/ Heuristic
Elaborative
High
Minority
High
Low
Elaborative Elaborative (sometimes plus Heuristic) < (High) Characteristics Content of Characteristics of Content of arguments of majority arguments majority/ content of arguments High Low High High
Low
Low
Low
Message processing (None/Low) Focus of attention Characteristics None of majority
Direct/public attitude change Indirect/private attitude change
Majority
High
High
270
Robin Martin and Miles Hewstone
group memberships are not important to the recipients of influence, and therefore holding a different attitude from the source does not affect selfidentity, then there will be no attitude change for either the majority and/or minority and original attitudes will remain intact. If majority/minority group membership has social meaning, either because majority group membership is desirable (e.g., it forms part of the person’s in-group) or minority membership is undesirable (e.g., it has deviant status), then majority status can act as a heuristic cue. In this context, the majority default option typically applies. As outlined above, in some situations, a minority might also act as a heuristic cue. When heuristic processing occurs, then there is public compliance to the source’s position without affecting preexisting attitudes. When the elaboration context is at an intermediate level, we make specific predictions concerning the impact of majority and minority influence. When people have the motivation and ability to elaborate the message, there is some choice as to which strategy to follow. The transition from low to intermediate elaboration contexts occurs when a threshold is crossed that allows (or encourages) the message recipient to engage in sufficient elaborative processing of the message to potentially allow for attitude change to occur. In this situation of moderate levels of ability and/or motivation, Petty et al. (1999) state that ‘‘. . . when thinking is not constrained to be high or low by other variables . . . source variables can determine the extent of thinking’’ (p. 20). What determines this choice, and the type of processing strategy, is the status of the source. If the elaboration context is not too high, then the ‘‘majority default option’’ applies. Even though people could systematically process the message, they typically will not do so, and take the option to comply with the majority position (to attain desirable majority group status, and so on) without considering the majority’s arguments in detail. Therefore, people’s underlying attitudes remain unchanged. However, when the source is a minority and it meets the behavioral style threshold (Moscovici, 1976), then this can lead to elaborative processing of the minority’s arguments to see if it ‘‘. . . may contain some truth . . . as a result of trying to see or understand what the minority saw or understood, the majority begins to see and understand as the minority would’’ (Moscovici and Personnaz, 1980, pp. 271–272). These predictions follow those outlined by conversion theory and are also supported by research in the persuasion literature showing that, under intermediate processing demands, people process messages from a stigmatized source more systematically than from a nonstigmatized source (see Fleming et al., 2005; Petty et al., 1999; White and Harkins, 1994). When the elaboration context is high (e.g., the topic is high in personal relevance, secondary tasks focus attention on the arguments), the motivation and/or ability to process the message is correspondingly high, leading
Majority Versus Minority Influence
271
to elaborative processing for both the majority and the minority. Because people are highly motivated and able to process the message, the majority default option is overridden and attention focuses on the content of the message. Thus, both the majority and minority source can lead to elaborative processing and attitude change is primarily determined by the amount of message-congruent thinking. In addition, it is possible that both elaborative and heuristic processing might occur, especially when the results of each type of processing complement each other, for example, the majority uses convincing arguments (the ‘‘co-occurrence hypothesis,’’ Bohner et al., 1995; see Erb and Bohner, 2007) or majority status acts as a message argument (Petty and Wegener, 1998). In this situation, there may be greater attitude change to a majority than a minority source. To summarize, in a low elaboration context attention focuses on source status and attitudes can be guided by heuristic cues resulting in public compliance without affecting preexisting attitudes. In an intermediate elaboration context, source status determines the type of processing. The majority default option applies and there is compliance to the majority without message elaboration. For a minority that adopts a positive behavioral style, elaborative processing can occur leading to attitude change. In a high elaboration context, attention focuses primarily on the content of the source’s arguments and elaborative processing can occur for both a majority and minority. Attitude change is related to the amount and direction of message-relevant thinking. In some situations, elaborative and heuristic processing can co-occur leading to majorities having more influence than minorities. Overall, majorities lead to message processing in a high elaboration context (e.g., high outcome relevance) whereas minorities lead to message processing in intermediate/high elaboration contexts when there is sufficient opportunity for their positive behavioral style to impact upon the majority.
4.3. Nature of attitudes following majority and minority influence The third part of our theoretical framework considers the nature of attitudes that are formed following majority and minority influence. According to the ELM, attitudes changed via systematic processing result from a detailed cognitive elaboration of the source’s message and these are referred to as ‘‘strong’’ (Krosnick et al., 1993; Petty, 1995). According to Krosnick and Petty (1995), there are four key characteristics of strong attitudes: they are resistant to counter-attitudinal appeal (Haugtvedt and Petty, 1992; Petty et al., 1995), they persist over time (Visser and Krosnick, 1998), they are more likely to guide behavior (Holland et al., 2002; Leippe and Elkin, 1987), and, finally, they guide information-processing (Houston and Fazio, 1989; Lord et al., 1979). By contrast, attitudes formed via
272
Robin Martin and Miles Hewstone
nonsystematic processing tend to be relatively ‘‘weak,’’ in that they have not been based upon message elaboration, and this means that they are unable to resist counter-persuasion, do not persist over time, do not predict behavior, and they do not guide information-processing. Petty et al. (1995) describe three reasons why message elaboration should result in these consequences. First, message elaboration can increase the structural consistency for an attitude because the process of elaboration might resolve preexisting inconsistencies amongst attitude components. Second, message elaboration can increase the association between various aspects of people’s attitudes and therefore increase their accessibility in memory (which is known to increase the attitude-behavior link, Fazio, 1995). Third, message elaboration may increase people’s belief that they have expanded considerable cognitive effort leading to increased confidence in their attitudes. Applying this reasoning to our framework leads to the prediction that attitudes changed via elaborative processing (whether from a majority or minority source) should lead to attitudes that resist counter-persuasion, persist over time and guide behaviors compared to attitudes changed via nonelaborative or heuristic processing.
5. Research Program It is a capital mistake to theorize before one has data. (Sir Arthur Conan Doyle, Scandal in Bohemia)
We now review a series of key experiments from our own research program that have tested various aspects of the theoretical framework presented earlier. We have organized the experiments around the three aspects of our framework concerning (a) source status and types of message processing, (b) the contexts under which these processes occur, and (c) the nature and consequences of attitudes formed following majority and minority influence. Before we present these experiments, we briefly review the methodology and procedure employed as these were common to all the experiments (precise details of experimental procedure and results are, of course, given in the primary publications cited throughout).
5.1. General methodology and procedure While there were some variations between the experiments described in this chapter, they followed a similar methodology and procedure which are described below. For further information, the reader is directed towards the individual papers.
Majority Versus Minority Influence
273
5.1.1. Methodology 5.1.1.1. Stimulus materials We employed a wide range of attitudinal issues in our studies, all pretested to be topical for participants. These included the legalization of voluntary euthanasia (the right of someone with a terminal illness to end their life), the use of animal experimentation for scientific research, the introduction of oral exams in universities, the introduction of the Euro currency in the UK, and the introduction of voluntary student unionism in Australia. 5.1.1.2. Message development We undertook extensive pretesting to generate attitude materials (following standard procedures described by Eagly and Chaiken, 1993, p. 311; and Petty and Cacioppo, 1986, p. 133). The aim was to establish two messages that contained either strong or weak arguments concerning the social issue and that only differed in terms of their level of persuasiveness. Message development involved the following main stages. The first stage involved argument generation. We collected arguments concerning each topic from a number of sources such as newspaper articles, campaign materials, and student discussion groups. We drew up a list of about 24 arguments including roughly equal numbers of what seemed, intuitively, to be weak versus strong arguments. We ensured that each argument made one point with reference to the topic and that all the arguments were similar in word length (about two sentences each). The second stage involved argument assessment and this involved judges (separately and independently), rating each argument on two dimensions: how ‘‘persuasive’’ and ‘‘believable’’ they were. The third stage concerned argument selection. This involved the selection of an equal number of arguments (typically six) that were rated as being the most (strong) or least (weak) persuasive. We were careful to ensure that we did not include arguments that were low on believability (we checked that the mean believability rating for the strong and weak arguments was similar). We also ensured that the word count for all the strong and weak messages was similar. The fourth stage concerned message piloting. We constructed separate messages containing either the strong or weak arguments and then asked a group of participants (from the same target population as the main experiments) to read the message and then: indicate their attitude towards the topic, list the thoughts they had had when reading the arguments, and then rate the arguments on several dimensions (such as, believability; comprehensibility; complexity; familiarity). The messages were deemed suitable if the strong versus weak message led to more agreement with the argument position, a greater proportion of message-congruent thoughts (strong messages should elicit more favorable than unfavorable thoughts, and the reverse for weak messages; Petty and Cacioppo, 1986), and the two messages were rated as being similar in terms of believability, comprehension, complexity, and familiarity. Example arguments, developed for the strong and weak versions of the pro- and anti-voluntary euthanasia messages, are shown in Table 5.4, and
Table 5.4
Examples of strong and weak arguments employed in the pro- and anti-voluntary euthanasia messages
Pro-message
Anti-message
Strong arguments
Weak arguments
We live in a society where freedom of choice is highly valued. This freedom of choice should be extended to include the individual’s choice to have their life ended under certain circumstances. Voluntary euthanasia is practiced in secret by some doctors anyway. Legalization would provide standardized guidelines and, therefore the system would be less open to abuse. Doctors should not be able to decide what is right and wrong for someone else. Voluntary euthanasia gives ultimate control of their life back to the patient.
There is nothing new in euthanasia. In the days when most people died at home, a suffering patient’s death was often hastened by the relatives or G.P.
Euthanasia is an extremely complex issue, covering a wide range of situations. Legalizing any form of euthanasia could eventually lead to decisions being made to terminate someone’s life in situations that are morally and ethically unacceptable. Euthanasia leaves the way open for bribery and financial gain to influence doctors’ decision-making. This could lead to an abuse of the system, for example, relatives could see it as a short-cut to their inheritance. Practicing doctors must know of several cases in which death seemed inevitable, yet the patient made a good recovery. Doctors are not always right in their diagnosis; people get better against the odds. Euthanasia does not allow for this possibility.
Medical technology is used to save, transform, and prolong lives. Why then, should it not be used to end lives, if that is what the individual wants? Voluntary euthanasia is a humane and economic alternative to the suffering and problems associated with old age. People could choose to die when they wanted to and not have to accept the decline into old age. We are put on this earth to suffer and endure life to the end. Therefore, no one should abuse this natural law. The act of voluntary euthanasia is therefore wrong and against the natural cycle of life. Society should ensure that people are treated with dignity. This means that for the good of society as a whole, life should be preserved regardless of what the individual who is terminally ill requests. One should not consider ending one’s life, even if there is no hope of recovery. Everyone needs to live their life to the end, no matter what is their quality of life. Therefore opposition to euthanasia ensures respect for life.
275
Majority Versus Minority Influence
Table 5.5 Pilot study mean results concerning development of strong and weak messages that are pro- and anti-voluntary euthanasia Argument direction Pro-voluntary euthanasia
Anti-voluntary euthanasia
Argument quality
Positive thoughts number (%) Negative thoughts number (%) Attitude Message ratings: Believability Comprehensiveness Complexity Familiarity
Strong
Weak
Strong
Weak
2.23 (74)
1.27 (37)
1.46 (57)
0.64 (30)
0.77 (26)
2.20 (63)
1.08 (43)
1.50 (70)
7.69
5.73
4.69
7.21
5.69 5.92 3.46 5.69
4.73 5.27 3.07 5.20
4.69 5.08 3.46 4.62
4.14 5.21 3.36 4.93
pilot data for these arguments are shown in Table 5.5 (note, positive and negative thoughts refer to message-congruent and message-incongruent thoughts respectively). We asked a sample of participants to read one of the messages (strong vs weak pro- vs anti-voluntary euthanasia) and then complete a thought-listing task, a 9-point scale indicating their attitude to voluntary euthanasia and ratings of the message on the dimensions noted above. The thought-listing data showed a significant interaction between argument quality and message direction. Strong arguments yielded more positive than negative thoughts (66% vs 34%, respectively) whereas weak arguments resulted in more negative than positive thoughts (67% vs 33%, respectively). Importantly, the three-way interaction was nonsignificant, showing the above pattern was found for both pro- and anti-voluntary euthanasia messages. For the attitude scores, there was a reliable argument quality by message direction interaction. The pattern of means perfectly corresponds to our expectations; for the pro-voluntary euthanasia arguments, the strong version led to a more favorable attitude (M ¼ 7.69) than did the weak version (M ¼ 5.73), whereas for the anti-voluntary euthanasia message, the strong version led to a less favorable attitude (M ¼ 4.69) than did the weak version (M ¼ 7.21). Clearly, the strong and weak versions of the messages were leading to differential cognitive elaborations which are consistent with the resulting attitude change. There were no differences in ratings of believability, comprehensibility, complexity, and familiarity.
276
Robin Martin and Miles Hewstone
5.1.2. Procedure Most experiments consisted of three main stages, with a separate booklet for each: pretest, message exposure, and posttest. In the pretest, participants were asked to indicate their attitude to a range of social issues, one of which concerned the topic of influence. As well as providing a pretest score, this measure also served as a screening item to ensure that only participants were selected who met sample selection criteria. For example, in some experiments, participants were only included if the source message was counter-attitudinal with respect to their initial attitude. The second booklet reported the results of a recent survey of students at their college concerning the topic of influence. This (fictitious) survey formed the basis for the source status manipulation, and it should be noted that we made explicit that the survey was conducted amongst their peers in order to locate the source status variable within the in-group population. Participants were informed about the proportion of the population that supported the message (which was stated as being ‘‘in favor’’ of or ‘‘against’’ the topic). We manipulated source status in one of two ways: (a) via numerical support such as, 82% (majority) or 18% (minority) support3; or (b) via results of a panel discussion showing 9 out of 11 (majority) or 2 out of 11 (minority) support. Source status information was printed in bold in order to make this salient. Participants were then informed that the main arguments given by this section of the population were reported on the following page. The message contained either five or six separate arguments and these were either listed separately or contained within an article within a student magazine. Participants read either the strong or weak version of the message. The third booklet contained the dependent measures which typically included (1) manipulation checks on the efficacy of the independent variables (e.g., percent of students agreeing with the message); (2) a measure of attitude towards the topic using either a single item 9-point item or a reliable multi-item scale (consisting of four or six 9-point semantic differential scales); (3) a thought-listing task in which participants reported their thoughts to the message by writing their thoughts about the message (one thought into each of several ‘‘idea’’ boxes, later rated by the participant as in favor, against or neutral towards the topic). Some experiments had additional measures and stages specific to their design (e.g., age, gender, religion, indirect attitudes, behavioral intentions, an additional booklet containing a countermessage, and so on). At the end of the experiment, the participants were thanked and debriefed as to the aims of the research.
3
Gross and Miller (1997) have argued that the ‘‘Golden Section’’ (61.8% majority; 32.8% minority) may reflect the point at which, subjectively, majority size is recognized. Our standard manipulation (82:18) clearly exceeds this criterion, although some of the studies reported below where we varied the size of majority and minority do not.
Majority Versus Minority Influence
277
5.1.3. Statistical analysis Our main analyses focused on the effects of the two independent variables (i.e., source status and argument quality) on the dependent measures (e.g., posttest attitude or attitude change scores) and the thought-listing responses. In line with previous research, we also computed a ratio of message-congruent thoughts (number of thoughts in favor of the position advocated by the source/total number of thoughts in favor and against the source’s position). This ratio indicates the proportion of message-congruent thoughts in the direction of the source’s position. For each dependent variable, we conducted factorial analyses of variance (ANOVA) and, where appropriate, tested hypotheses using post hoc tests or planned comparisons. To explore the potential mediating role of thought elaborations in determining attitudes, we computed regression procedures outlined by Baron and Kenny (1986). It should be noted that in many experiments, preliminary analyses examined the effects of gender and age on the dependent variables. No reliable findings were found for either of these factors and so they are not reported in subsequent analyses.
5.2. Experimental studies In this section, we review our experimental studies in relation to the three aspects of our theoretical model we described earlier (1) source status and types of message processing, (2) conditions affecting when message processing occurs, and (3) the nature of attitudes following majority and minority influence. 5.2.1. Source status and types of message processing In these studies, we examined the processes underlying majority and minority influence. In one experiment, we exposed participants to a message in favor of the use of animals in scientific research (Martin and Hewstone, 2003b, Experiment 1). They were told that either 82% (majority) or 18% (minority) supported the pro-animal research position. They then read a set of either strong or weak arguments (a third independent variable concerned the initial attitude position, pro- vs anti-animal experimentation, but this did not affect influence). The results for the posttest attitude score showed a reliable interaction between source status and argument quality (see Fig. 5.1). Additional analyses showed a reliable argument quality effect for the minority (strong > weak) but not for the majority source. The data from the thought-listing task yielded a reliable difference in the minority condition, showing that strong arguments led to a greater proportion of message-congruent thinking than did weak arguments. Furthermore, the thoughts ratio mediated the relationship between argument quality and attitude scores. The attitude and thought-listing data support the view that the minority source led to elaborative processing and this affected attitudes. By contrast, the lack of an argument quality effect for the
278
Robin Martin and Miles Hewstone
7
Strong
Weak
Agreement with source
6 5 4 3 2 1 Majority
Minority
Figure 5.1 Mean attitude scores as a function of source status and argument quality (Martin and Hewstone, 2003b, Experiment 1).
majority on both attitude and thoughts ratio suggests that participants had engaged in nonelaborative processing. The results of the attitude and thought-listing data support conversion theory’s prediction that it is the minority source that leads to detailed message processing. This first experiment showed that people processed the minority’s arguments, which affected their attitudes, but it does not indicate why this might occur. We have argued that for minorities to have influence, they need to have a positive behavioral style and be seen as ‘‘distinctive’’ from the majority. One aspect of distinctiveness concerns the numerical size of the minority. It could be assumed that being a numerically small group might make a minority more distinctive in the eyes of the majority, and that this would lead to, possibly even be necessary for, message elaboration. We tested this idea in an experiment that manipulated the consensus level of the majority and minority as being either a large difference (82 vs 18%), as operationalized in the previous experiment, or relatively small (52 vs 48%) (Martin et al., 2002, Experiment 3). If minority distinctiveness is important in attracting attention to the minority’s message, we reasoned that a numerically small (18%) minority would be more likely to lead to systematic processing than would a ‘‘large’’ (48%) minority. The message used in this study concerned the legalization of voluntary euthanasia and employed either a strong or weak message that was antivoluntary euthanasia. The predicted three-way interaction on posttest attitudes between consensus differences, source status, and message quality was significant (see Fig. 5.2: note, scores are reverse coded to show agreement with source). Confirming our predictions, there was no difference between the strong and weak messages for the majority, irrespective of its consensus level (82% or 52%). However, consensus difference did have an effect for minority influence: there was a reliable argument quality effect only for the numerically
279
Majority Versus Minority Influence
7
Strong
Weak
Agreement with source
6 5 4 3 2 1 82% majority 18% minority 52% majority 48% minority
Figure 5.2 Mean attitude scores as a function of numerical size, source status, and argument quality (Martin et al., 2002, Experiment 3).
small (18%) minority. Furthermore, the proportion of message-congruent thinking mediated the relationship between this interaction and attitudes. This shows that participants had processed the message when attributed to an 18% minority. An additional finding of interest concerns when the consensus level between the majority and minority was small (52 vs 48%). Here, we found no effect of argument quality in either source condition, but a reliable source effect showing more influence from the 52% majority than the 48% minority (for an alternative effect of the numerical size of the majority and minority on attitude change, see Erb et al., 2006). Another further experiment dealing with the size of the source confirmed that being perceived as ‘‘small’’ was important in leading people to elaborate the minority’s arguments (Gardikiotis et al., 2005, Experiment 2). In this experiment, participants read either strong or weak arguments against voluntary euthanasia. Two independent variables were manipulated in relation to the supporter of the arguments: source status (majority vs minority) and consensus adjective (large vs small); thus the majority or minority was referred to as being either ‘‘large’’ or ‘‘small.’’ There was a reliable argument quality effect only for the ‘‘small minority.’’ In summary, these first three experiments all point to the elaboration of messages from the minority, but not the majority, and the minority source received elaboration of its message only when it was seen as distinctive (whether by its size in percentage terms, or by being labeled ‘‘small’’ in size). 5.2.2. Conditions affecting when message processing occurs We turn now to some of the potential contingency variables that might determine when people will engage in elaborative processing of the majority or minority’s arguments. We deal, in turn, with source-position
280
Robin Martin and Miles Hewstone
congruency, negative outcome bias, and studies that manipulated ability or motivation to elaborate the message. 5.2.2.1. Source-position congruity In this section, we explore the source-position congruity predictions proposed by Baker and Petty (1994). To recap, they propose that people will engage in systematic processing when their consensus expectations are broken (i.e., when the source-position is ‘‘imbalanced’’: counter-attitudinal majority or proattitudinal minority) and will not engage in systematic processing when their consensus expectations are confirmed (i.e., when the situation is ‘‘balanced’’; pro-attitudinal majority or counter-attitudinal minority). In our review of the Baker and Petty experiments, we noted that their basis for determining source-position congruity had not been satisfactorily established. Therefore, it was the aim of this experiment to employ a methodology to ensure that balanced and imbalanced conditions had been created (Martin and Hewstone, 2003b, Experiment 1). We have described this experiment in the last section; here, we present additional information concerning how we ensured that balanced and imbalanced conditions had been met. We did this by asking participants, prior to exposure to the message, to indicate the percentage of students they believed were ‘‘in favor,’’ ‘‘against’’ or ‘‘don’t know’’ regarding the issue of animal research. Analyses of these scores indicated, as expected, that participants tended to believe that more students shared their opinion than opposed it, and this was true for those initially against and in favor of animal research. To ensure that participants met our selection criteria, they were only included in the experiment if their pretest attitude score placed them unambiguously in a counter- or pro-attitudinal condition with respect to the message, and their percentage estimates showed they believed more students agreed with their attitude than opposed it. As indicated above, the experiment found a reliable argument quality effect only for the minority condition, and this occurred in both the balanced and imbalanced conditions. Contrary to Baker and Petty (1994), a study that used a topic with a negative outcome, in this experiment which carefully established balanced and imbalanced conditions, and did not employ a topic that might have induced a negative personal outcome bias, we found message processing only in the minority condition (see also Gardikiotis, 2005). 5.2.2.2. Negative outcome bias In our review of the research, we proposed that the message topic might moderate when a majority or minority induces message processing. In particular, we noted that experiments showing only a majority leading to systematic processing had employed a message that advocated a counter-attitudinal position and was against the participants’ self-interest—a position we have termed as ‘‘negative personal
Majority Versus Minority Influence
281
outcome.’’ When a majority endorses such a position, it is likely that this leads to curiosity (‘‘why are the majority arguing against my, and their, interest?’’). Because the majority position often prevails, due to the weight of numerical support, this provides a dilemma for the individual—‘‘most people are arguing for something that is likely to occur, and that is against my self-interest.’’ This situation might lead people to process the majority’s arguments to understand the proposal and its implications to them personally. However, when the minority argues for a negative personal outcome, it may be ignored, either because weakly supported positions are unlikely to be adopted, or to protect one’s self-interest (‘‘a few people are arguing for something that is unlikely to be adopted and therefore is not threatening’’). By contrast, when the topic does not lead to a negative personal outcome, we predict a similar pattern of results to our earlier experiments; namely, message processing only for the minority. Thus far, however, topics varying in negative outcome bias have been used in different experiments. In our next experiment, we therefore manipulated whether the topic induced a negative personal outcome, or not (Martin and Hewstone, 2003b, Experiment 2). We did this by using different topics which varied in the extent to which they argued in favor of a negative personal outcome (low vs high). Message topics were selected from a pilot study that asked participants to rate a number of social issues with respect to (a) their agreement with the issue, (b) whether the issue affected their self-interest, and (c) the potential outcome of the proposed change (negative vs positive). On the basis of these ratings, two topics were chosen which differed reliably on these dimensions. The first topic, ‘‘the legalization of voluntary euthanasia,’’ was rated as one on which the respondents were in favor of the issue, it was moderate in self-interest, and the proposed change had a positive outcome. By contrast, the second topic, concerning ‘‘the introduction of a single currency in Europe (the ‘Euro’),’’ was rated by our British participants as one on which they were against the issue, it was high in self-interest, and the proposed change was perceived as leading to a negative outcome. We then developed strong and weak counter-attitudinal arguments for the anti-voluntary euthanasia and pro-Euro issues which represented low and high negative personal outcomes, respectively. As expected, this experiment yielded a reliable three-way interaction between message topic, source status, and argument quality (see Fig. 5.3). When the arguments did not induce a negative personal outcome (voluntary euthanasia), there was an argument quality effect only for the minority condition which replicates our earlier experiments. However, when the topic was associated with a negative personal outcome (Euro), consistent with our predictions, the pattern of results was reversed (replicating Baker and Petty, 1994, Experiment 1). In this case, there was a reliable argument quality effect for the majority source but not for the minority source. Furthermore, in the conditions showing an argument quality effect, the relationship between
282
Robin Martin and Miles Hewstone
7
Strong
Weak
Agreement with source
6 5 4 3 2 1 Majority
Minority Majority Minority Low High Negative personal outcome
Figure 5.3 Mean attitude scores as a function of negative personal outcome, source status, and argument quality (Martin and Hewstone, 2003b, Experiment 2).
argument quality and attitudes was mediated by the proportion of messagecongruent thinking—suggesting that participants had systematically processed the arguments. Therefore, a topic that induces a negative personal outcome— that is, a position that argues against the self-interest of the source and recipients of influence and proposes a negative outcome—moderates message processing of both the majority and minority source. A potential problem with the last experiment is that it manipulated negative outcome bias by using different topics (and arguments) and it is possible that topic differences, rather than outcome bias differences, determined the results. To overcome this potential problem, we conducted an experiment that employed the same topic for all participants (pro-animal experimentation) and we manipulated personal outcome bias via feedback about the source’s belief concerning their self-interest in the topic (Martin et al., under review, Experiment 2). After the source status manipulation, participants were informed either that ‘‘These students thought their position was in favor of their own and other peoples’ self-interest’’ (positive personal outcome) or that ‘‘These students thought their position was against their own and other peoples’ self-interest’’ (negative personal outcome). Manipulation checks subsequently established that this manipulation had successfully led to different beliefs about the source’s self-interest in the topic. Attitudes towards animal experimentation were taken before and after exposure to the arguments. Mean attitude change scores showed a reliable three-way interaction between source status, argument quality, and
283
Majority Versus Minority Influence
Attitude change
3
Strong
Weak
2
1
0 Majority Minority Positive
Majority Minority Negative
Stated personal outcome
Figure 5.4 Mean attitude change scores as a function of stated personal outcome, source status, and argument quality (Martin et al., under review, Experiment 2).
self-interest (see Fig. 5.4). The pattern of results replicated those found for the last experiment. When the source argued for a positive personal outcome, there was a reliable argument quality effect only for the minority but not for the majority—a consistent finding in our experiments. However, when the source argued for a negative personal outcome, the pattern was reversed; there was a reliable argument quality effect for the majority but not for the minority. The data for the proportion of message-congruent thinking mirrored the pattern found for attitude change and further analyses showed that message-congruent thinking mediated the relationship between the independent variables and attitude change. Taken together, the results of these two experiments (and also Martin et al., under review, Experiment 1, that examined self-reported self-interest) provide strong evidence for the role of outcome bias, and help us to provide theoretical integration of a previously disparate set of results from different research labs. On the basis of the attitude and thought-listing data, in an intermediate processing situation when there is no negative outcome bias, people engage in elaborative processing of a minority message. However, when the topic induces a negative outcome bias, then there is only elaborative processing for the majority source. 5.2.2.3. Manipulating ability/motivation for message elaboration Thus far, our experiments have demonstrated that in intermediate processing situations, where there are no situational factors that either increase or
284
Robin Martin and Miles Hewstone
decrease elaboration of the message, message recipients will engage in elaborative processing of the minority’s arguments more than the majority’s arguments, provided that the personal outcome is not negative. However, our theoretical framework makes specific predictions concerning how people will process a majority and minority message at different levels of elaboration (low, intermediate, and high). To briefly recap our predictions: under low elaboration, attitudes will not be guided by argument quality but by heuristic cues (such as, ‘‘majority more likely to be correct than minority’’); under intermediate elaboration (a situation akin to many conditions described above), attitudes will be guided by source status (elaborative processing for a minority); and under high elaboration, attitudes will be guided by argument quality for both majority and minority and, in specific situations, by both elaborative and nonelaborative processing. We now review two experiments that manipulated the elaboration context and investigated the effects of source status on attitudes (see also Martin and Martin, 2006, who increased message elaboration via ingestion of caffeine). In the first experiment, we manipulated elaboration context using a method inspired by Craik and Lockhart’s (1972) levels-of-processing approach to memory. In a highly influential paper, Craik and Lockhart suggested that incoming stimuli can be processed at different levels and that this has consequences for their retention in memory. Their main postulate was that the more deeply information is processed, the more it will be remembered (the ‘‘depth effect’’). They proposed that there is a continuum from phonological (based on sounds) to semantic (based on meaning of words) processing, which corresponds to a shallow/deep level of processing and that memory is a by-product of the depth of processing. Evidence for the depth effect comes from studies that have used the incidental learning paradigm. In these studies, participants are presented with a list of words and are asked to make various judgments about them (a so-called orientating task). Following this, and unknown to the participant at the time of presenting the test stimuli, they are given a memory test. Recall is taken as an index of level of processing. By using a variety of orientating tasks, it is possible to manipulate the level of processing participants are required to undertake. For example, asking participants to indicate whether a stimulus word is in capitals would require only shallow processing; thinking of a word that rhymes with the stimulus word would need medium level of processing; while evaluating whether the stimulus word fits into a sentence would need deeper processing. Many studies have since found that the greater the depth of processing, the greater the recall (e.g., Craik and Tulving, 1975). In this experiment, we presented participants with a set of strong or weak anti-voluntary euthanasia arguments (Martin et al., 2007a, Experiment 2). Portions of the arguments were repeated alongside the original message
285
Majority Versus Minority Influence
(the selected text). The level of message elaboration was manipulated by asking participants to conduct one of three orientating tasks on the selected text (manipulation checks established that these tasks had been performed correctly). In a lexical-processing condition (low elaboration condition), participants were asked to check whether the selected text was written in the same font type and size as the original message. This task should result in only shallow message processing and therefore there should be no effect of argument quality. However, we predicted that the majority source, operating via a consensus heuristic, would have greater impact on attitudes than the minority source. In a ‘‘surface’’ semantic-processing condition (intermediate elaboration condition), participants were asked to check the spelling of words in the selected text. This task requires semantic processing of single words and we would expect intermediate processing to occur; we predicted message processing only for the minority source, due to its distinctiveness leading to message validation (cf., Moscovici, 1980). In a ‘‘deep’’ semantic-processing condition (high elaboration condition), participants were asked to rewrite the selected texts in different words but to convey the same meaning. This task requires semantic processing of sentences and we predicted that strong arguments (regardless of majority/ minority support) would have greater impact on the reported attitudes than weak arguments. We tested our hypotheses, within each processing condition, with a series of planned contrasts. These confirmed our hypotheses (see Fig. 5.5). 6
Strong
Weak
Agreement with source
5
4
3
2
1 Majority
Minority Low
Majority Minority Intermediate Message processing task
Majority
Minority High
Figure 5.5 Mean attitude change scores as a function of message processing task, source status, and argument quality (Martin et al., 2007a, Experiment 2).
286
Robin Martin and Miles Hewstone
In the low elaboration condition, the only reliable effect was for source status, with the majority having more influence than the minority. The lack of an argument quality effect suggests the source status main effect was due to nonelaborative processing. In the intermediate elaboration condition, there was a reliable interaction between source status and argument quality, showing a difference between strong and weak messages for the minority but not majority source. Finally, in the high elaboration condition, there was an argument quality effect for both majority and minority sources, showing participants had engaged in elaborative processing for both sources. The pattern of results for message-congruent thoughts complemented the attitude data. In the low elaboration condition, there were no reliable effects, as one would expect; here we predicted attitudes would be guided by heuristic cues and not by elaboration of the source’s arguments. In the intermediate elaboration condition, there was a reliable interaction between source status and argument quality that mirrored the pattern found for attitudes. Finally, in the high elaboration condition, the only reliable finding was for argument quality. Further evidence that the argument quality effect was due to elaborative processing comes from the finding that the proportion of message-congruent thinking in both the intermediate and high elaboration conditions mediated attitude scores. An additional experiment in the same paper also manipulated levels of message elaboration but this time using a motivational, rather than cognitive, manipulation. Specifically, we varied the outcome relevance of the topic (Martin et al., 2007a, Experiment 1). In this study, participants read either a strong or weak message that advocated the introduction of oral exams as part of their college examinations. They were told either that oral exams would be introduced into another college, and not their own, and so they would not be affected by them (low relevance), or no information was given about oral exams being introduced into a college (intermediate relevance), or they were told that oral exams would be introduced into their college during their time at college (high relevance). The manipulation of personal relevance is conceptually analogous to the three elaboration levels obtained by using orientating tasks in the previous experiment, and the pattern of results for attitudes was very similar across the two studies. The key effects were a source main effect (majority > minority) under low relevance, an argument quality effect only for the minority under intermediate relevance, and an argument quality effect for both the majority and minority under high relevance (in this latter condition there was also a source main effect; majority > minority). The results across these experiments were consistent and supported our predictions (see also Petty and Cacioppo, 1984). When the elaboration context was low, there was no effect of argument quality but a reliable source status effect. In both experiments, there was heuristic acceptance of the majority position without detailed processing of its arguments.
Majority Versus Minority Influence
287
When the elaboration context was intermediate, there was an interaction between source status and message quality. As we predicted, only the minority source triggered elaborative processing. This finding supports Moscovici’s (1980) conversion theory and shows one of the boundary conditions under which it might apply. Finally, when the elaboration context was high, there was a reliable effect of argument quality, showing elaborative processing for both a majority and minority source. As we noted above, when high elaboration was manipulated by having high personal relevance, there was, in addition to the argument quality effect, a source main effect (but this was not found when high elaboration was achieved via an orientating task). Both Crano and Chen (1998) and Kerr (2002) found a similar pattern in experiments that also manipulated message elaboration by means of personal relevance. It is possible that when message elaboration is manipulated via motivational means, a number of processes can operate together. Hence, encouraging people to read arguments that are personally relevant both triggers an elaborative appraisal of the arguments and leads to greater focus on the source of the message, resulting in heuristic acceptance (this is consistent with the co-occurrence hypothesis in the HSM, see Bohner et al., 1995). 5.2.3. Nature of attitudes following majority and minority influence In this final empirical section, we review some of our studies that have examined the consequences of engaging in elaborative processing following majority and minority influence. Specifically, these studies tested the prediction that minority-supported messages resulted in ‘‘strong’’ attitudes (Krosnick et al., 1993; Petty, 1995), which resist counter-attitudinal appeal, persist over time, and guide behavior (we have not conducted research on whether these minority-led attitudes also guide information-processing). It should be recalled that we predict that in most persuasion situations (what we refer to as ‘‘intermediate’’ processing contexts), there will be heuristic acceptance of a majority position (without detailed consideration of the majority’s arguments), whereas a minority should stimulate elaborative processing of the content of its arguments. If this is the case, then attitudes formed following minority influence should be relatively ‘‘strong,’’ compared to those formed following majority influence, and therefore message recipients should be better able to resist counter-persuasion, should persist over time, and should predict behavior. We now review some of our experiments on each of these issues. 5.2.3.1. Resist counter-persuasion To test further our model, we wanted to examine the strength of attitudes that had first been formed by a minority or majority, in order to see if these attitudes would be able to resist a counter-persuasive communication. To achieve this aim, we exposed participants, in turn, to two messages that argued different positions
288
Robin Martin and Miles Hewstone
on the same topic, with attitude scores taken after each message (for use of this paradigm, see, e.g., Haugtvedt and Petty, 1992; Haugtvedt and Wegener, 1994; Wu and Shaffer, 1987). In terms of cognitive models of persuasion, elaborative processing of the first (initial) message potentially provides individuals with arguments to resist the ‘‘attack’’ from the second (counter) message (in some respects the first message ‘‘inoculates’’ participants against attack from the second message, McGuire, 1964). However, if attitudes to the initial message are formed via nonelaborative processing, they should be relatively weak and susceptible to the influence of the second countermessage. To summarize our prediction, if minority influence leads to conversion, as predicted by Moscovici (1980) and by our theoretical framework, then, attitudes formed following exposure to a minority should be more resistant to counter-persuasion than attitudes formed following majority influence. We tested this hypothesis (Martin et al., 2003, Experiment 2) in an experiment that measured participants’ attitudes towards voluntary euthanasia at three time points: at the onset of the experiment (pretest), after exposure to a counter-attitudinal message (initial message, posttest I), and finally after exposure to a pro-attitudinal message (countermessage, posttest II). Some procedural issues need to be mentioned. First, we only indicated the status of the source (majority or minority) to the initial message. Second, we ensured that there was a delay between exposure to the two messages (5 min) to ensure that participants had sufficient time to engage in message elaboration, if they chose to, before exposure to the countermessage. Finally, because we wished to encourage elaborative processing of the initial message (and, potentially, resistance to the countermessage), we used only the strong version of each message. The results for the attitude data showed the expected two-way interaction between source status and measurement times (see Fig. 5.6). Note that because of the difference in the directions of the messages, influence to the source is shown by high scores on the initial message and by low scores on the countermessage. There was a reliable change across the measurement times for both the majority and minority conditions—but the pattern varied. When the source was a majority, participants were influenced by the initial message as shown by a reliable increase between pretest and initial message/posttest I; however, these attitudes were vulnerable to the second countermessage as shown by the reliable reduction between posttest I and posttest II. In fact, for the majority source, there was no difference between pretest and posttest II attitude scores, showing that attitudes following the countermessage returned to their pretest levels. This pattern suggests that participants had heuristically accepted the majority’s position to the initial message, without detailed processing of its arguments, and therefore, the attitudes that resulted were weak and yielded to the countermessage. In the minority condition, participants were also influenced by the initial message
289
Majority Versus Minority Influence
Attitude score
5
Majority
Minority
4
3
2 Pre-test
Initial message
Countermessage
Figure 5.6 Mean attitude scores as a function of measurement time and source status (Martin et al., 2003, Experiment 2). Note: greater agreement with the source is reflected by high scores on the initial message and low scores on the countermessage.
but, unlike in the majority condition, these attitudes did not change as a result of exposure to the countermessage. This pattern suggests that participants had processed the minority’s arguments in the initial message and, as a result of doing this, these attitudes were relatively strong and they were able to resist the countermessage. The data from the thought-listing task (completed after the initial message) support the above conclusion. Those in the minority condition engaged in more message-congruent thinking than did those in the majority condition, suggesting that minorities encouraged greater elaborative processing. Furthermore, the proportion of message-congruent thinking mediated the change in attitudes between the initial and countermessage. Further experiments reported by Martin et al. (2003) addressed some additional methodological and theoretical issues associated with this paradigm. For example, in one experiment, the order of the messages was reversed with the initial message being pro-attitudinal and the countermessage being counter-attitudinal (Martin et al., 2003, Experiment 1). Thus, the first message was consistent with preexisting attitudes, whereas the second message was inconsistent. Attitudes did move towards the countermessage position in both source conditions but the reduction was smaller, indicating greater elaborative processing, when the source of the initial message was a minority than a majority. A potential methodological problem in this paradigm is whether the participants resisted the countermessage or, in fact, were influenced by it. We tested this competing hypothesis in an experiment that replicated the majority and minority
290
Robin Martin and Miles Hewstone
conditions but also added conditions in which participants were exposed to only a majority- or minority-endorsed countermessage, in order to determine how persuasive it was without prior exposure to the initial message (Martin et al., 2003, Experiment 3). The results supported the view that participants in the majority and minority conditions had, indeed, yielded in the majority condition, and resisted the countermessage in the minority condition. The assumption in the above experiments is that the resistance to the countermessage was due to recipients processing the minority arguments. This is based on the cognitive-elaboration approach, which assumes that it is only when attitudes have been changed via systematic processing that these attitudes will be resistant to counter-persuasion. Resistance occurs through engaging in message-congruent elaborations to the initial message that render participants able to resist a countermessage. We tested this assumption further in an experiment that varied the quality of the arguments contained in the initial message (Martin et al., 2008a, Experiment 1). We reasoned that if attitude change following majority influence led to compliance (through nonsystematic processing), then the extent to which these ‘‘new’’ attitudes yielded to counter-persuasion should not be affected by the quality of the arguments in the initial message. However, if attitude change following minority influence was due to systematic message processing, then there should be greater attitude change following the initial message when it contained strong compared to weak arguments, and these attitudes should be better able to resist the countermessage. This experiment was similar in procedure to that of Martin et al. (2003, Experiment 2) described earlier, except that the initial message contained either strong or weak arguments. To further examine social influence processes, this experiment also included an indirect measure of attitude change that was not mentioned in the message; the indirect attitude concerned ‘‘Genetic screening for medical disorders (e.g., cystic fibrosis, sickle cell anemia),’’ whereas the direct attitude concerned voluntary euthanasia. A pilot study showed that there was a moderate, but reliable, positive correlation between the direct and indirect attitudes at pretest. The results for the direct attitude showed the expected reliable threeway interaction between source status, argument quality, and measurement time. For the majority source, the participants were influenced by the initial message but, as predicted, these attitudes yielded to the second countermessage, with attitudes returning to the pretest levels. This pattern of results was the same irrespective of the quality of the arguments in the initial message, showing that participants had heuristically accepted the majority position without considering its arguments in detail. For the minority source, participants were influenced by the initial message when it contained strong arguments and, as expected, these attitudes subsequently resisted the countermessage (supporting the findings of Martin et al., 2003).
Majority Versus Minority Influence
291
When the minority had weak arguments, however, a different pattern emerged in that the minority-endorsed initial message did not affect attitudes and these attitudes then yielded to the countermessage. This is what we expected: if the minority encourages elaborative processing, then elaboration of weak arguments should not lead to influence (which it did not), and because no attitude change occurs, message recipients’ attitudes are vulnerable to counter-persuasion. As for the previous experiments, the proportion of message-congruent thinking mediated attitude change between the initial and countermessage giving further support that resistance to counter-persuasion was due to elaborative processing of the initial message. We can now turn to the results for the indirect attitude—an issue related to the message topic but not mentioned within it. The results revealed one important finding and that concerned the minority condition containing the strong arguments. In this condition, participants were influenced by the initial message, as shown by the increase from pre- to posttest I, but these attitudes resisted counter-persuasion as there was no difference between posttest I and posttest II. These results indicate that in the minority condition with strong arguments the elaborative processing of these arguments led to conversion to the minority position that was detected on an issue related to, but not mentioned in, the message. So far we have found reliable evidence that majority and minority influence lead to nonelaborative and elaborative processing, respectively, and we have shown that this differentially affects the tendency of people to yield or to resist counter-persuasion. However, theoretically, we propose this relationship only when the elaboration context is at an intermediate level and we were careful to choose a pretested procedure and message we have shown to be at this level (Martin et al., 2007a). We predict that changing the elaboration context should have corresponding effects on the types of processes instigated by majority and minority sources and the nature of attitudes formed subsequently. In a further experiment (Martin et al., 2008a, Experiment 2), we examined this idea by using an orientating task to increase the elaboration context which, we hypothesized, should lead to elaborative processing of both the majority and minority source and therefore resistance to a countermessage (in the same way as the orientating task that led to high elaboration did in the experiment discussed in the last section, Martin et al., 2007a, Experiment 2). This new experiment employed the same procedures as the last experiment with two exceptions. First, we used only the strong arguments for the initial message, and second, in one condition, we told participants that they would later be required to recall the arguments in the message (this procedure should lead participants to pay closer attention to the message, and process the arguments in greater depth). When the participants were not told that they would have to recall the arguments in the initial message, the pattern of results was the same as in
292
Robin Martin and Miles Hewstone
previous experiments—attitude change to the initial message yielded or resisted the countermessage when the initial message was attributed to a majority or minority, respectively (see Fig. 5.7). In the condition where participants were told that they would later recall the arguments in the initial message, which should increase elaborative processing of the arguments, attitudes following the initial message resisted the countermessage for both the majority and the minority source. Thus, this experiment shows that people will elaboratively process majority-endorsed arguments, resulting in ‘‘strong’’ attitudes that resist counter-persuasion, in a high elaboration context. In summary, these studies on resistance to counter-persuasion show that attitudes formed following minority influence were relatively strong in that they were able to resist counter-persuasion. On the other hand, attitudes formed following majority influence were relatively weak and yielded to counter-persuasion, unless there was a secondary task that increased message elaboration.4
Attitude score
7
Majority
Minority
6
5
4 Initial message Countermessage
Control
Initial message Countermessage
Recall
Figure 5.7 Mean attitude scores as a function of measurement time, source status, and recall condition (Martin et al., 2008a, Experiment 2). Note: greater agreement with the source is reflected by high scores on the initial message and low scores on the countermessage. 4
Some resent research by Tormala et al. (2007) offers potentially conflicting results (see also Tormala et al., in press). They found that after initially resisting a minority message, attitudes became weaker and they were less resistant to the second message. The key difference between the Tormala et al. (2007) studies and our own studies is that in the Tormala et al. studies, participants resisted the initial message whereas in our studies, participants were persuaded by the initial message. It should be noted that theoretically we are interested in attitudes that have been formed following minority influence-a situation very different from the Tormala et al. (2007) studies. However, we recognize that whether participants resist or are persuaded by the initial message might moderate the effects of the countermessage.
Majority Versus Minority Influence
293
5.2.3.2. Persist over time According to our approach, attitudes changed by minorities should be relatively strong and should persist over time compared to attitudes changed by a majority (see also, Crano and Chen, 1998; Moscovici et al., 1981; Tafani et al., 2003). We tested this hypothesis in an experiment in which participants were exposed to strong arguments against voluntary euthanasia message (Martin and Hewstone, in preparation). We assessed attitudes on three occasions: prior to reading the message (pretest), immediately after reading the message (posttest: immediate), and 2 weeks following the experiment (posttest: delayed). On each occasion, we measured attitudes towards both voluntary euthanasia (direct attitude) and abortion (indirect attitude). Although the message concerning voluntary euthanasia did not mention the issue of abortion, pretesting had shown that people perceive a link between the two (they are both concerned with the control and sanctity of life), and there was a moderately positive correlation between the two.5 There was a reliable two-way interaction between source status and measurement time for both the direct and indirect attitudes. For the direct attitude, the majority led to an immediate change in attitudes towards its position (pretest vs posttest: immediate) but attitudes returned to their pretest level 2 weeks later. In the case of the minority, there was a small immediate effect on attitudes (between pretest and posttest: immediate), but there was a larger shift to the minority position 2 weeks after message exposure. For the indirect attitude, there was no change over time for the majority source. In the case of the minority, however, there was reliable change towards the indirect attitude after exposure to the message and this change persisted over time. Taken together, these results suggest that the majority led to compliance to its position, without detailed message processing, which resulted in an immediate change to direct attitudes that did not persist over time, and had no effect on the indirect attitude. The minority source led to elaborative processing that brought about a delayed change in direct attitudes, and an immediate change to the indirect attitude that persisted over time. This different pattern of effects, across time, for the minority on direct and indirect attitude measures has also been noted in other experiments (e.g., Personnaz, 1981). 5.2.3.3. Predict behaviors The final strand of our research program concerns the impact of majority and minority influence on intentions to behave in line with one’s attitudes (behavioral intentions) and on actual attituderelevant behaviors. It should be noted that nearly all research on majority and minority influence has focused on attitudes and beliefs, and very few 5
It is not only our participants who link issues in this way. Recently Poland blocked a European Union protest against the death penalty, arguing for parallel European condemnation of abortion and euthanasia (see The Guardian, September 12, 2007).
294
Robin Martin and Miles Hewstone
studies have investigated changes to people’s behaviors. Some related research has examined the role of expert versus nonexpert sources on behavioral intentions (e.g., Falomir-Pichastor et al., 2002; Joule et al., 1988), but these studies did not manipulate explicitly both majority and minority source status. As described earlier, one consequence of elaborative processing is that it should lead to greater prediction of attitude-consistent behaviors. Indeed, persuasion research shows that when people process a message systematically, then they are more likely to engage in behaviors that are consistent with their attitudes (behavioral intention) than when they have engaged in nonsystematic processing (e.g., Cacioppo et al., 1986a; Leippe and Elkin, 1987; Petty et al., 1983). Applied to the context of majority-minority influence, we hypothesize that if attitude change following majority influence is due to nonelaborative processing, whereas attitude change following minority influence is due to elaborative processing, then attitude-consistent behavioral intentions should be expressed more strongly following a minority- than a majority-endorsed message. We tested this hypothesis in an experiment in which participants were exposed to a counter-attitudinal message (anti-voluntary euthanasia) attributed to either a numerical majority or minority (Martin et al., 2007b, Experiment 1). After reading the message, participants indicated their intention to engage in an attitude-consistent behavior (in this case signing a ‘‘living will,’’ a form of advance directive that stipulates that, under specified circumstances, medical procedures should not be used to keep the signer alive; see Dworkin, 1993). If participants agree with the message (i.e., their attitude becomes less in favor of voluntary euthanasia), then they should be less likely to engage in an attitude-consistent behavior (i.e., less likely to sign a living will). In addition, we predicted that the above effects should occur only when participants had been influenced by the message. When participants were not influenced by the message, presumably because they had not elaborated on the source’s arguments, we did not expect any effect of source status on behavioral intentions. To test this idea, we split the participants into those who did, and did not, change their attitudes. As predicted, there was a reliable two-way interaction between source status and whether attitudes had been changed or not (see Fig. 5.8). When participants did not change their attitude there was no difference between the majority and minority conditions. We suspect that this reflects the fact that no elaborative processing had occurred for these respondents. However, when participants were influenced by the source (i.e., became less in favor of voluntary euthanasia), those in the minority condition reported higher attitude-consistent behavioral intentions (i.e., less likely to sign living will) than did those in the majority condition. Although the majority resulted in the same amount of attitude change to the message as the minority, the change in attitude was due to nonelaborative processing and
295
Majority Versus Minority Influence
7
Behavioral intentions
6
No attitude change Attitude change
5 4 3 2 1 Majority
Minority
Figure 5.8 Mean behavioral intention scores as a function of attitude change and source status (Martin et al., 2007b, Experiment 1).
thus, it was ‘‘weak’’ in nature and therefore did not affect attitude-consistent behavioral intentions. A potential methodological problem with the last experiment is that it used a self-report measure of behavioral intention, and that this measure might be affected by participants’ reported attitude scores. In our next experiment, therefore, we tried to overcome this potential problem by having a measure of ‘‘actual’’ behavior that was taken outside the context of the experimental situation and under conditions of participant anonymity (Martin et al., 2007b, Experiment 2). We believe that this is the first study of majority and minority influence to include a direct behavioral measure. The study concerned the introduction of voluntary student unions (VSU) into an Australian University which we knew, through piloting, that most students were in favor of. We therefore exposed participants to an anti-VSU message. At the end of the experiment, participants were given a card, addressed to the Australian Minister for Education, Science and Training, that stated ‘‘. . . opposition at any moves to introduce voluntary student unionism at Australian universities.’’ Participants were told that if they wished they could sign the card and place it in a sealed box (which was located away from the experimental room). Because the message was anti-VSU and the card voiced opposition to VSU, the act of signing and posting the card constituted an attitude-consistent behavior. It should be noted that the participants were under no obligation to sign the card and that they received their course credit before they could have posted the card. Participants believed that the experimenter would not know if they had signed the card but, by the use of invisible ink, we were able to identify which participants had posted the card. We predicted that when the message was low/medium personal relevance, as in the last experiment, engaging in an attitude-consistent behavior would be most likely for those in the minority condition who changed their
296
Robin Martin and Miles Hewstone
100
Performing behavior (%)
90
No attitude change Attitude change
80 70 60 50 40 30 20 10 0 Majority
Minority
Figure 5.9 Percentage of participants performing behavior as a function of attitude change and source status (Martin et al., 2007b, Experiment 2).
attitude. This is indeed what was found (see Fig. 5.9). When participants had not changed their attitude towards VSU, there was no effect of source status on the number of people engaging in the behavior. However, when participants had changed their attitude, they were more likely to sign and return the card when exposed to a minority compared to a majority source. 5.2.3.4. Summary For the first time, we explored the resistance, persistence, and prediction of behavior induced by majority versus minority messages. First, we showed that majority messages were vulnerable to a second, counterpersuasion message, whereas minority messages resisted the second message, due to elaborative processing of the first message that also affected an indirect measure of attitude. When, however, we experimentally manipulated high elaboration of the message, resultant attitudes were strong, resisting persuasion, in both source conditions. Further evidence of stronger attitudes in the minority condition was gleaned by use of a persistence paradigm. Whereas majority influence resulted only in immediate compliance, minority influence persisted in a 2-week follow-up. Finally, we showed a greater impact of the minority than the majority source on not only guiding behavioral intentions, but also behaving in a manner consistent with the position argued by the minority. Taken together, this is impressive cross-paradigm evidence that minority sources engender stronger resultant attitudes than do majority sources.
5.3. Summary In this section, we have reviewed evidence from our research program in support of our model. First, we explored message processing in intermediate processing conditions. We provided evidence for elaborative
Majority Versus Minority Influence
297
processing in the minority, but not the majority, condition, and we showed that this effect only occurred when the minority was distinctive. Next, we looked further at conditions affecting when message processing occurred. We showed elaborative processing of the minority message irrespective of the level of source-position congruency (balanced vs imbalanced), but only when the minority did not argue for a negative personal outcome; if the source argued for a negative personal outcome, there was elaborative processing of the majority, but not the minority, message. Third, we manipulated ability or motivation for message elaboration across three levels. Under low elaboration, majority influence was greater, but without elaborative message processing. Under intermediate elaboration, there was only elaborative processing of the minority message. Under high elaboration, messages from both sources were elaboratively processed. Finally, we explored the nature of attitude change induced by minority versus majority sources. The minority source was clearly associated with stronger attitudes that resisted counter-persuasion, persisted over time, and guided behavior.
6. Methodological and Theoretical Issues in Current and Future Research Read not to contradict and confute, nor to believe and take for granted, nor to find talk and discourse, but to weigh and consider. Francis Bacon (Of Studies)
6.1. Methodological issues in our research program Notwithstanding its achievements in terms of better understanding when, how, and why messages associated with minority or majority sources receive detailed scrutiny, we are fully aware of some of the limitations of our research program to date, and which directions future research might move in. We deal, in turn, with our manipulation of the key source variable, our assessment of elaboration via the thought-listing procedure, and our exclusive use of a single-source paradigm. Finally, we deal with the absence of control conditions in this area of research. 6.1.1. The manipulation of source In all our research to date, we have manipulated source status by means of consensus information, informing participants of the proportion of the relevant group that hold the view propounded in a message. This has long been used as a manipulation in this research domain (e.g., Moscovici and Personnaz, 1980).
298
Robin Martin and Miles Hewstone
We have likened this manipulation to the real-life experience of reading a newspaper article above which is a headline that conveys the percentage of people holding a view, normally referring to a recently conducted poll (e.g., ‘‘Exclusive poll reveals 68% of English voters want own parliament,’’ Daily Telegraph, November 26, 2006 and ‘‘Only 34% back being in the EU,’’ Daily Mail, December 12, 2006, see Gardikiotis et al., 2004, for an analysis of ‘‘majority’’ and ‘‘minority’’ terms in newspaper headlines). As we reported earlier, we have also varied the size of the majority-minority difference (Martin et al., 2002, Experiment 3), and the adjectives used to describe either group (Gardikiotis et al., 2005, Experiment 2). We acknowledge that such manipulations do not exhaust the range of possible operationalizations. The very first laboratory research on minority influence was actually carried out by Solomon Asch (1951), using his paradigm of face-to-face interacting members of a small group, some of whom were confederates. He investigated the impact of a lone minority member (a confederate) on naı¨ve participants, and reported no influence at all. Indeed, the majority ridiculed the minority. This is not to say, however, that face-to-face minorities have no impact. Moscovici et al. (1969) also opposed two minority members (confederates) to four majority members of a face-to-face group. Far from being ineffective, a consistent minority was able successfully to influence the majority. But as Smith and Tindale (in press) note, whereas Moscovici’s (1976) early theoretical work regarding minority impact conceived of social influence as an interactive phenomenon, involving reciprocal influence between minority and majority factions, most of his empirical work highlighted a much narrower and less dynamic aspect of the influence process. We do not think that either of these manipulations (percentages vs faceto-face groups using confederates) is preferable to the other; they reflect different operationalizations of the same conceptual variable, as should be the case in a representative research program designed to yield generalizable results. However, as Smith and Tindale (in press) warn, we should be alert to the possibility that some findings may be paradigm-specific (e.g., it might be primarily, or even exclusively, in freely interacting groups that minority influence is more likely to be indirect rather than direct, found in private than public, and delayed rather than immediate). There are also many research questions that can only be addressed with more dynamic paradigms (e.g., see Prislin’s studies on changing relations of size between majorities and minorities; for reviews see Prislin, in press; Prislin and Christensen, 2005). There are, of course, constraints on being able to conduct research using real groups: it takes much longer and, with the group as the unit of analysis, it would require huge numbers of participants to carry out the kinds of multifactorial studies we have reported. It remains the case, moreover, that such interacting groups are not suitable for investigating our main concern here, namely message processing of majority and minority sources, where it
Majority Versus Minority Influence
299
is essential to keep different messages separate and to obtain detailed measures of scrutiny that refer to discrete arguments. We should also not lose sight of the fact that minorities and majorities in the real world do not differ simply in terms of numbers. Positions of power are often occupied by majority, rather than minority, members, giving them direct control of important outcomes for minorities, and there are often differences of status as well as size (for a discussion see Vocci et al., in press). Moving in a less social and more cognitive direction, we believe that future research could usefully exploit priming paradigms (e.g., Bargh and Pietromonaco, 1982), using both supra- and subliminal manipulations, to further our understanding of the processes underlying majority and minority influence. Priming could be used to manipulate not only the size of the source, but also its connotations. It is well known, for example, that being in the minority, especially under conditions where group consensus is required, reduces liking and acceptance of the minority source of influence (Levine, 1980); and research by Mugny and Papastamou (1980) contrasted positive and negative views of the minority as, respectively, ‘‘flexible’’ versus ‘‘rigid.’’ Consideration of how we view minorities in society confirms immediately that they can have either positive (e.g., elites, experts) or negative (e.g., deviants, loners) connotations. When talking about minorities who are our heroes (whether Aung San Suu Kyi or Bob Geldof), we might be inclined to use the words ‘‘indefatigable,’’ ‘‘determined,’’ or ‘‘courageous,’’ but when talking about some minorities (whether animalrights protesters or anti-abortion groups), they become ‘‘intransigent,’’ ‘‘dogmatic,’’ and so on. Priming techniques could be used to investigate the consequences of characterizing what Hitchens (2001) calls ‘‘contrarians’’ as ‘‘dissidents,’’ ‘‘radicals,’’ ‘‘mavericks,’’ ‘‘loose cannons,’’ ‘‘rebels,’’ ‘‘fanatics,’’ ‘‘troublemakers,’’ and so on. 6.1.2. Assessing elaboration by means of thought listing All of the key studies in our research program use the thought-listing technique to assess the degree of message recipients’ elaboration of majority versus minority messages. The introduction of the thought-listing technique into this area (by Maass and Clark, 1983, Experiment 2) was a welcome addition to an area that assumed much, but measured little, about processing. However, persuasion researchers acknowledge that there are still problems with what remains a central dependent measure for much persuasion research. In particular, recipients may be differentially motivated to generate and/or report thoughts that justify the attitude expressed, as a function of movement towards a majority versus minority message (for reservations about thought-listing, see Miller and Colman, 1981). However, a series of elegant studies by Romero et al. (1996) do provide compelling support for the cognitive mediation hypothesis across a range of paradigms.
300
Robin Martin and Miles Hewstone
Miller and Colman (1981) note that thought listing is, essentially, a measure of introspection, which psychologists have long treated with suspicion. Research on both the ELM and HSM models, including our own, seeks to overcome any inherent weakness of this technique, to some extent, by relying heavily on the manipulation of strong versus weak arguments, in conjunction with measurement of thought listing, to demonstrate information-based systematic processing. Thus, it is the comparison of thought listing, whatever its demerits, between argument conditions that is crucial. When systematic processing occurs, people should be more persuaded by messages that contain strong rather than weak arguments (the ‘‘argument quality effect’’); when nonsystematic processing occurs, the nature of the arguments has no significant impact on persuasion. A further issue with the thought index (see Miller and Colman, 1981) is the order in which thoughts and attitudes should be measured. On the one hand, because thoughts are the putative mediator of attitudes, one can argue that the order thoughts-attitudes is the more logical, or appropriate, given the theoretical model explicit in the cognitive theories of persuasion that thoughts cause (and therefore must precede) attitudes (see Wood et al., 1996). The order adopted in our own research has therefore been for participants to give their thoughts in response to the persuasive message prior to filling in their attitudes. On the other hand, one could argue that asking participants to list thoughts might induce them to do something (e.g., think) that they might not normally do (perhaps in response to a particular source or argument that they would otherwise ignore). From the outset, we thought it unlikely that this procedural difference could account for the different results reported across studies, because Petty et al. (1976) compared the two orders of thoughts and attitudes in a standard attitude-change paradigm (i.e., one that did not manipulate majority vs minority source) and found no difference. However, it is conceivable that order of measures could interact with source status, and for this reason we conducted a study that varied the order of thoughts and attitudes. Using the topic of voluntary euthanasia, we presented participants with a counter-attitudinal message that was anti-voluntary euthanasia (Martin and Hewstone, 2001). Thus, the design of the study was a 2 (order of measurement: thoughtsattitudes vs attitudes-thoughts) 2 (source: majority vs minority) 2 (argument quality: strong vs weak) factorial. We found an argument quality effect only for the minority source, and this occurred regardless of the order of the measurement of the dependent variables (see Fig. 5.10). Although we believe that our research has dealt with most of the critical issues surrounding the use of the thought-listing technique, we should point out that, although relatively convenient, it is not the only possible means of assessing elaboration. Cacioppo et al. (1986b) showed that the electromyogram (EMG) could be used to measure both the valence and intensity of emotional reactions, and it would be interesting to introduce such measures
301
Majority Versus Minority Influence
6
Strong
Weak
Agreement with source
5
4
3
2
1 Majority
Minority
Thoughts-attitude
Majority
Minority
Attitude-thoughts
Measurement order
Figure 5.10 Attitude scores as a function of measurement order, source status, and argument quality (Martin and Hewstone, 2001).
into this area. The rise of social cognitive neuroscience (Ochsner and Lieberman, 2001) also offers exciting possibilities here, including use of techniques to measure (e.g., functional magnetic resonance imaging, fMRI, and event-related potentials, ERPs), and to disrupt (Transcranial Magnetic Stimulation, TMS) brain activity. If minority messages are processed more systematically, we might expect to see greater activity in the medial frontal cortex in recipients of minority messages studied while in a scanner (see Amodio and Frith, 2006), and also to detect greater activity in scalp sites over the frontal cortex (e.g., Ito and Urland, 2003). Similarly, if recipients of a minority message were subjected to TMS in the area found to be active during processing of such messages (e.g., Harmer et al., 2001), then processing should be interrupted, and previously reported differences between the source conditions should be attenuated. 6.1.3. Single versus dual source paradigms All the studies from our research program reported in this chapter have used a single-source paradigm: participants received either a majority or a majority message.6 However, there is an alternative paradigm that can be, and has 6
It is useful to note that in our studies that examined resistance to a countermessage we did, in fact, expose participants to two messages, each representing a different source position (Martin et al., 2003; Martin et al., 2008a). However, in these experiments, each source message was presented in a sequential order, delayed in time, with attitudes taken after exposure to each. In this section, we consider the issue of presenting both majority and minority messages simultaneously.
302
Robin Martin and Miles Hewstone
been, used in research on majority-minority influence: participants are exposed simultaneously to both majority and minority influence (see Maass and Clark, 1983; Mackie, 1987). However, these studies used different manipulations of source, and different topics, and reported contrasting results. Maass and Clark provided participants with a transcript of a group, whereas Mackie provided consensual support for attitudinal positions. Maass and Clark (1983) found that a minority source induced greater attitude change than a majority source on a private measure of attitude, whereas the majority had a stronger effect when respondents believed that their attitudes would be disclosed in public. Mackie (1987), in contrast, reported that the majority had greater impact than the minority on both public and private measures. We do not wish to engage in a discussion of which of these findings is ‘‘correct,’’ but rather to acknowledge the strengths as well as the shortcomings of the dual-source paradigm. In our view, the dual-source paradigm is not any ‘‘better,’’ more ‘‘realistic,’’ or more ‘‘accurate’’ a representation of majority-minority influence settings than our own single-message paradigm. It simply reflects one type of setting, where both groups’ opinions may be presented in juxtaposition. This happens sometimes in real life (e.g., committee meetings), but not always (e.g., newspaper headlines, leaflets dropped through doors and mailboxes attempting to persuade recipients to vote for a particular party or policy). We see both approaches as representative of the range of settings in which we should study majority-minority social influence. Furthermore, we admire Maass and Clark, and Mackie, for engaging in the complexities of the dual-message paradigm, which is more complex than the singlemessage paradigm. However, it is this very complexity which led us to eschew it, given our focus on message processing. In order to draw conclusions about the extent of argument scrutiny for majority versus minority sources, comparing strong and weak messages, we believe that the singlesource paradigm is preferable. As Chaiken and Stangor (1987) pointed out, a within-subjects design is inappropriate if one is interested in the amount of thinking instigated by a specific source. Nonetheless, we fully acknowledge the contribution made by dual-source studies, whose complexity is perhaps reflected in their relative scarcity, and argue that a full understanding of majority-minority influence will only be achieved via research using a wide range of paradigms. Ideally, these should range from single-source studies, through dual-source studies, to studies of face-to-face interaction in real groups. 6.1.4. The curious absence of control conditions The main focus in our research program (see Section 5, above) has been to compare thought-listing and attitude measures in recipients of majority and minority messages comprising either strong or weak arguments. This has also been the standard factorial design adopted in research by other scholars
Majority Versus Minority Influence
303
on this topic (see Section 3, above). The focus in the broader literature on majority-minority influence (i.e., dealing with judgments and physical stimuli, and not just attitudes) has also been on this comparison (although some studies have drawn conclusions, unwisely in our view, on the basis of just one source). However, whereas explicit comparison simply of majority and minority sources suffices to answer certain questions (which source leads to greater attitude change?, which source is associated with greater argument scrutiny?), multiple control conditions would be required to answer additional questions. Stroebe (in press) was the first to consider which additional control conditions would be needed to assess further hypotheses derived from the ELM and the HSM. For example, to test the hypothesis that message recipients might (under some circumstances) wish to distance themselves from a negatively viewed minority, one would need a message-only control condition, in which participants were exposed to the same message as other participants, but without being told that the source was either majority or minority. Stroebe explains that without this particular control condition it is not possible to ascertain whether any biasing effects are due to the majority source inducing a positive bias (i.e., more positive than the control) or the minority source inducing a negative bias (i.e., less positive than the control), or both. Stroebe notes that only one study appears to have run this control condition (Erb et al., 1998). Notwithstanding its utility for testing such hypotheses, the adoption of such a control condition is not without difficulty; the problem is that participants may infer the source of the message when none is given, and, given the false consensus effect (Ross et al., 1977), they are likely to infer a pro-attitudinal majority and a counter-attitudinal minority. This problem may explain, in part, the curious absence of this control condition. Stroebe argues, however, that this condition is essential for testing hypotheses about the impact of consensus information on the processing of persuasive arguments. We believe that it is the most important control condition for our research focus. The researchers’ task is made still harder by Stroebe’s (in press) arguments (strong ones, at that) in favor of two further control conditions. The first, consensus-information only, would specify majority or minority source, but present them without any persuasive arguments, allowing the researcher to assess the impact of the consensus information as a peripheral cue (only one study appears to have included this control condition; Mackie, 1987, Experiment 4). The third and final control condition, is a no information condition (where neither consensus information nor message is given) as a baseline for assessing the amount of majority and minority influence. But perhaps the absence of these control conditions is not so ‘‘curious’’ after all. As a cursory review of studies reported in this chapter reveals, most of the experiments conducted, especially those in our own research program, are complex, multifactorial experiments involving between-subjects
304
Robin Martin and Miles Hewstone
manipulation of source (majority vs minority), argument quality (strong vs weak), and then some additional factor that adds value to the literature (e.g., low vs high relevance; positive vs negative personal outcome; low vs intermediate vs high elaboration). Adding multiple control conditions would, however, significantly increase the complexity of the design. As a compromise, for now, we suggest that future research should begin systematically to include some of the control conditions, beginning with the key message-only condition. 6.1.5. Summary Although there is clearly room for more research using, and comparing, operationalizations of the key source variable, the manipulation we have used thus far can be defended on sound methodological grounds, as can the use of both the thought-listing procedure and the single-source paradigm. Future research should become both more and less social. More social, in that it would include some studies using interacting groups (both with, and without, the use of confederates) and simultaneous presentation of majority and minority messages. Less social, in that it would exploit priming paradigms and social neuroscience methods to explore in greater depth information processing in majority and minority influence.
6.2. Theoretical issues and future research Although we have worked on this research program for over a decade, we acknowledge that we have barely scraped the surface, in terms of what could still be achieved using this approach. We begin, remaining within the confines of our adopted paradigm, by considering how best to exploit the rich potential of the cognitive approach to persuasion. We first explore further hypotheses concerning how source status might impact on information processing and attitude change, and then consider the possible role of need states and individual differences. Next, we adopt a different approach to explore the neglected ‘‘phenomenology’’ of majority-minority status, by looking at what is ‘‘special’’ about minority status and how that impacts on the arguments minorities advance and defend. Finally, we move even further from the laboratory-based message-processing approach we have adopted thus far to speculate about the missing link in social change between minority influence and social movements. 6.2.1. Realizing the full potential of dual-process models of persuasion for the study of social influence We hope that this exposition of our research program has shown, at least, some of the gains of applying a cognitive-persuasion (message-processing) approach to the study of majority and minority influence. However, both the ELM (Petty and Wegener, 1998) and the HSM (Bohner et al., 1995) are
Majority Versus Minority Influence
305
complex, sophisticated models that include a wide range of variables that we have not touched on, and allow for types of processing we have not yet explored. Other research (see Bohner et al., 1995; De Dreu et al., 1999; Erb and Bohner, 2007, in press) has made greater use of the HSM in its study of majority-minority social influence than we have, and Stroebe (in press) has outlined the full range of predictions that could be derived from dualprocess theories concerning the impact of majority and minority influence on the processing of persuasive messages (from which we sample just a few). In particular, the HSM (Bohner et al., 1995) assumes that under high levels of motivation and ability both systematic and heuristic processing modes are likely to affect persuasion. Future research could, for example, explore evidence for the HSM’s ‘‘additivity hypothesis’’ (e.g., when both heuristic and systematic processing point to the same conclusion both heuristic cues and content information exert main effects on persuasion). Thus, one might manipulate both the expertise of a minority (low/high) and the quality of its arguments (strong/weak). As Stroebe (in press) suggests, however, if the expert source presents many strong arguments, then the HSM’s ‘‘attenuation hypothesis’’ may be confirmed, whereby the independent effect of the heuristic cue is weakened by the mass of content information. Stroebe (in press) also suggests investigation of the different types of motivation conceived by the HSM to underpin attitude change—accuracy, impression, and defense. As he points out, accuracy motivation encourages objective and unbiased information processing. In contrast, impression motivation (with its focus on the expression of socially acceptable attitudes) is likely to be aroused in influence settings in which the identities of significant audiences are salient, or people must communicate their attitudes to others who have the power to reward to punish them. He predicts that, given people’s concern to be liked by others (Deutsch and Gerard, 1955) and especially by the majority, impression motivation should favor majority over minority influence. We know of no work that has explicitly manipulated these types of motivation and investigated their impact in this area. Research has, at least, begun to explore processing ability that, along with motivation, is central to both ELM and HSM. As we reported earlier, we investigated majority versus minority influence by comparing message processing under low, intermediate, and high elaboration conditions (using both cognitive and motivational manipulations; Martin et al., 2007a). We found a heuristic effect for the majority in the low elaboration condition, an argument quality effect for the minority in the intermediate condition, and an argument quality effect for both majority and minority in the high-elaboration condition. We know of only one study, however, that has manipulated processing ability directly, as done for the ELM (e.g., using a distraction paradigm) and compared majority and minority influence (Schuurman et al., 1995;
306
Robin Martin and Miles Hewstone
see De Vries et al., 1996). In the distraction condition, contrary to the idea that numerical support might serve as a heuristic cue (e.g., ‘‘consensus implies correctness’’), there was no difference in the impact of majority and minority sources; however, in the no-distraction condition, majority support produced more favorable attitudes than did minority support. As De Vries et al. (1996) point out, this experiment leaves open whether distraction interferes with reception, rather than systematic processing. This could, of course, be shown in an experiment that manipulated distraction, source, and argument quality. Future research should therefore compare majority and minority influence in classic distraction paradigms used in attitude research (e.g., Petty et al., 1976), but also in single- versus dual-task paradigms (e.g., Macrae et al., 1994), which can reveal how much processing capacity is usurped by the type of processing engaged for each source (e.g., see Pendry and Macrae’s, 1994, Experiment 2, use of a probe reaction time measure, based on Bargh, 1982). If our theorizing is correct, minority influence both triggers and requires deeper, more time-consuming processing; it should therefore usurp more cognitive capacity, leaving less residual capacity for secondary tasks. 6.2.2. Investigation of need states and individual differences Thus far, our research program has neglected this topic; however, findings from the vast literature on attitude change suggest that some measures of individual difference might moderate effects of interest, and that manipulation of need states might also impact the effectiveness of influence. Need for cognition (NfC) is an obvious first measure to explore, because research has shown stronger argument quality effects on attitude change for individuals with high rather than low need for cognition (Cacioppo and Petty, 1982; Cacioppo et al., 1996). Need for cognition has also been shown to affect attitude strength. Haugtvedt and Petty (1992) reported that attitude change in respondents with a high need for cognition was more persistent and more resistant against counter-argumentation than in individuals with low need for cognition. It is not clear immediately, however, how NfC might interact with majority-minority status. One hypothesis is that those who frequently engage in and enjoy effortful cognitive activity might be more willing to consider different views, including that of the minority and not just the majority. Relatedly, Neuberg and Newsom (1993) proposed that individuals differ in the extent to which they are dispositionally motivated to simplify and structure their environment. The Personal Need for Structure scale (PNS, Thompson et al., 1993) measures people’s chronic need for a simple structure, and correlates positively with people’s likelihood of organizing both social and nonsocial information in less complex ways (Neuberg and Newsom, 1993, Experiment 3).
Majority Versus Minority Influence
307
The need for a simple structure has recently been conceptualized as part of a broader construct, the need for cognitive closure (see Neuberg et al., 1997), which has also been proposed to differ chronically across individuals. It has been measured as a dispositional variable using the Need for Closure scale (NFCL, Webster and Kruglanski, 1994), and obtained results included greater resistance to persuasion when prior information was present, and less resistance when no prior information was given (Kruglanski et al., 1993). Once again, it is not obvious how these variables might interact with majority-minority status. However, it appears plausible that message recipients with higher PNS or NFCL scores might be more inclined to go along with the safe majority, than entertain the views of the minority. Consistent with this idea, De Dreu and Koole (1997) found that experimentally lowering participants’ need for closure reduced their tendency to use the ‘‘consensus implies correctness’’ heuristic. Finally, Shuper and Sorrentino (2004) explicitly examined the relationship between people’s uncertainty orientation and majority and minority influence. Two final variables that might be interesting to explore in this context are the needs for uniqueness (conceived as a trait; Maslach, 1974) and the need for distinctiveness (considered as a need state; Brewer, 1991). A number of authors have alluded to the link between minorities and uniqueness. Crano (in press), for example, has argued that the minority enjoys the advantage of relative uniqueness. Meanwhile, Erb and Bohner (in press) have linked minority influence and deindividuation. On the basis of research findings showing that individuals sometimes strive to avoid deindividuation (e.g., Snyder and Fromkin, 1980), they proposed, and reported support for the idea, that low-consensus positions allow deindividuated individuals to regain a feeling of uniqueness and social distinction (Imhoff and Erb, 2006; cited in Erb and Bohner, in press). This research manipulated deindividuation, and future research might also explore the individual difference measure of ‘‘need for uniqueness’’ as a moderator of minority influence. Given that certain types of cultures (i.e., individualist) emphasize the uniqueness of individuals more than others (i.e., collective; Triandis and Trafimov, 2001), there is also a niche for cross-cultural studies of minority influence (see Ng and Van Dyne, 2001). Relatedly, optimal distinctiveness theory (Brewer, 1991) argues that individuals seek identification with social groups in order to satisfy two basic, independent psychological needs (inclusion/assimilation and differentiation/distinctiveness). This is an important theoretical perspective on self and identity that, in our view, also has implications for the social influence literature. For example, classic theorizing on social influence assumes that majority influence is caused by group members’ dependence on the group (Festinger, 1950). Yet, research on optimal distinctiveness generally reveals the positive value of numerical minorities for reasons of greater group distinctiveness (see Leonardelli and Brewer, 2001). Although this is not an
308
Robin Martin and Miles Hewstone
individual-differences perspective, we believe that manipulation and measurement of optimal distinctiveness as a situational or individual difference variable has something to offer research on majority-minority influence. In fact, we would predict, generally, that individuals with a higher need for uniqueness or a need for distinctiveness would be more susceptible to minority influence than individuals with a lower need for uniqueness or distinctiveness; and that, correspondingly, individuals with a higher need for assimilation would be more susceptible to majority influence. 6.2.3. The ‘‘phenomenology’’ of majority-minority status Our research program to date has focused exclusively, as has all prior research on both majority and minority influence, on the target, rather than on the source, of influence. We have studied in detail the information processing involved, and its consequences, when message recipients are exposed to majority and minority sources. Research has, however, generally neglected the behavior of sources of influence (for exceptions, see Levine and Kaarbo, 2001; Gordijn et al., 2001; Zdaniuk and Levine, 1996). We have recently reversed this focus by examining the nature of the arguments generated by individuals who find themselves in either a numerical majority or minority (Kenworthy et al., in press). If arguments presented by minorities and majorities can produce differential scrutiny and outcomes (e.g., Baker and Petty, 1994; Martin and Hewstone, 2003b; Mucchi-Faina et al., 1991), then what effect does majority versus minority status have on argument generation? We began to think along these lines because we were surprised at, and disagreed with, a claim made in the literature. Maheswaran and Chaiken (1991) proposed that the combinations of majority source-weak message and minority source-strong message are ‘‘incongruent’’ and thus trigger deeper processing (see also Baker and Petty, 1994). Yet, from the perspective of Moscovici’s (1980) conversion theory, we would argue that there is nothing incongruent in the combination of minority source and ‘‘strong’’ message. Minorities are often deeply committed, and this is arguably their greatest attribute. As John Stuart Mill wrote: ‘‘One person with a belief, is a social power equal to ninety-nine who have only interests’’ (Mill, 1861/ 1924, p. 155). From this perspective, minorities might well be expected to have strong, not weak, messages (Gerard, 1985; Kerr, 2001). We then began to consider what we refer to as ‘‘the phenomenology of minority numerical status.’’ Being in the numerical minority is typically aversive. Research from various domains of social psychology has shown that when people are in the numerical minority, they feel uncomfortable, stressed, and displeased (see Asch, 1956; Dion, 2002; Guinote et al., 2006; Kenworthy and Miller, 2001, 2002; Lu¨cken and Simon, 2005; Nemeth, 1986; Nemeth and Wachtler, 1983), as compared to those
Majority Versus Minority Influence
309
in the numerical majority. With respect to attitudes and opinions specifically, the aversiveness and discomfort of minority numerical status, most likely due to a heightened need for validation of one’s beliefs and opinions (see Deutsch and Gerard, 1955; Kenworthy and Miller, 2001; Kruglanski and Mayseless, 1987), leads to a search for information that may support one’s position. A number of studies have demonstrated, consistent with this view, that numerical minorities (vs majorities) show increased cognitive activity of various types (e.g., Guinote et al., 2006; Levine and Russo, 1995). Relatedly, studies on power show that individuals in powerful positions tend to focus their information-processing strategies on a single target, and ignore peripheral information, whereas individuals in powerless positions consider multiple sources of information, and generate additional information (see Guinote, 2008, for a review). These results suggest that being in the numerical minority might produce more original, creative, argumentation, and are consistent with Zdaniuk and Levine’s (1996) findings that participants in a minority faction, anticipating a confrontation with a numerical majority, were less biased in favor of their own position during a thought-generation task than were numerical majorities (see also Levine et al., 1996). In addition, as a function of their being outnumbered, numerical minorities must necessarily take multiple perspectives—both their own as well as that of the numerical majority—in order to reduce the threat of invalidation. By contrast, numerical majorities will be less motivated to adopt others’ perspective and to think in novel and creative ways to defend their position, because theirs has the weight of consensus (see Levine and Russo, 1995). In three studies, we examined the effect of minority versus majority group membership on argument originality in the absence of any exposure to minority or majority messages (Kenworthy et al., in press). Our general hypothesis was that being in the numerical minority would lead to the generation of more original and creative arguments for one’s position, as compared to being in the numerical majority. The first two studies found reliable evidence that arguments generated by participants told they were in a numerical minority condition were indeed rated more original and convincing (by judges blind to experimental condition and hypotheses of the experiment) than those produced by participants in the numerical majority condition. Moreover, the effect was not influenced by a measure of either mood or need for cognition. The third study replicated this minority-originality effect even when all participants were merely asked to imagine themselves as belonging to one of the numerical status factions taking part in a discussion group, and to generate arguments for a counter-attitudinal position instead of the one they reported favoring initially. In this way, we examined whether minority numerical status could produce stronger and more creative arguments even when the position being advocated was not initially endorsed (see Gordijn et al., 2001).
310
Robin Martin and Miles Hewstone
This design also allowed us to rule out an explanation in terms of vestedinterests (see Crano, 1995), whereby participants’ own attitudes would likely be more elaborated (e.g., Petty and Krosnick, 1995). What is not yet clear is why numerical minority members produced more original arguments, regardless of whether or not they personally endorsed those attitudinal positions. Future research might explore whether creative arguments result from either a defensive motivation (e.g., being in a minority elicits a tendency to resist the weight of consensus held by the majority faction), or an offensive motivation (e.g., not wishing merely to demonstrate the validity of one’s position, but also to convert the majority faction to one’s point of view; see Levine and Kaarbo, 2001). Turning the spotlight from the target to the source of influence, as this new research does, also opens up the possibility of a new line of studies on minority influence. Thus, whereas prior research has found that minorities can be persuasive in converting majority members if they are consistent and confident (Moscovici, 1980, 1985; Moscovici et al., 1969), what remains to be seen is whether actual numerical minorities are consistent and confident across time in naturalistic settings. Research in this vein is in its infancy, but we hope that it will expand and add to a fuller understanding of both minority influence as well as the psychology of numerical status generally. 6.2.4. From minority influence to social movements From the suffragettes to the Montgomery bus boycotters, history is littered with characters who kick-start progress by being awkward. (Editorial in The Guardian newspaper; May 30, 2007) You can alter policy. The individual is not powerless in the face of either political indifference or monstrous human tragedy. Let me say it embarrassedly, cornily, almost guiltily. Let me try to say it without sounding like some pious twat. You can change the world. (Bob Geldof; The Guardian, December 28, 2005)
Individuals were, of course, crucial to the slave-trade abolition movement. Thomas Clarkson, founder of the Society for Effecting the Abolition of Slavery in 1787, was indefatigable. He rode on horseback round England to spread his message and collect signatures, and published 700 iconic posters showing a diagram of the slave-ship Brookes. Granville Sharpe was the writer of countless polemics of protest. Former slave Olaudo Equiano spoke out for those still in bondage and had huge influence through his autobiography. And, of course, William Wilberforce was ultimately successful in parliament. Yet, successful abolition required a social movement to achieve its goal, one that owes a huge debt to the networking skills and social organization of the Quakers. They were crucial in organizing the boycott of slave-grown sugar, joined at one time by more than 300,000 people. It is
Majority Versus Minority Influence
311
fitting, in the present context, that Quakers are referred to as religious ‘‘nonconformists.’’ Determined, persistent, and persuasive as they may be, individual minorities often fail to have an impact unless they start a social movement. The anti-slavery movement can be seen as the prototypical social movement, involving popular agitation, legal challenges, and parliamentary lobbying (other examples include the Suffragettes in the UK, and the Southern Christian Leadership Conference, in the Deep South of the United States). Klandermans (e.g., 1997), one of the leading scholars of social movements, includes as one of three key processes involved in whether movements take off; ‘‘persuasive communication during mobilization campaigns by movement organizations, their opponents and countermovement organizations’’ (p. 45), yet he stops short of an explicit analysis of minority influence. We believe that such an analysis would illuminate both areas, social influence and social movements. We think it highly likely that a persuasion-based approach to minority influence, such as our own, may be especially useful at the initial stage of a social movement (where people potentially engage with and listen to novel, cogent arguments), but that a categorization-based approach (e.g., David and Turner, 1996, 1999, 2001) may be more significant in catalyzing the social movement. Stu¨rmer and Simon (2004) propose that collective identity should foster social movement participation by four main processes: making group members’ selfesteem more dependent on the in-group’s status relative to other groups; increasing perceptions of collective strength and in-group cohesion as well as intergroup differentiation; fostering the perception of commonalities of interests, thereby promoting trust and cooperation; and, most important for our focus, promoting the acceptance of social influence from in-group members (whereas influence from out-group members is rejected). Highly identified group members should be more willing to accept social influence attempts from fellow in-group members, thus influence would progress in line with selfcategorization theory. Individuals will perceive the minority in a wider context and begin to see the minority as ‘‘part of ‘us’ rather than ‘them’, basically on our side, standing for basic values that ‘we’ all share’’ (Turner, 1991, p. 171). It is for this reason that in-group minorities tend to have more public influence than out-group minorities (David and Turner, 1996, 1999), a fact apparently known to the Quakers, whose (out-group) voice against slavery had been ignored, until it was taken up by the Anglican Thomas Clarkson who could influence public opinion as an in-group minority. We do not, however, wish to over-play the hand of minority influence. Although our research program to date has focused on persuasion, we acknowledge that bringing about social change involves much more than persuasion. For example, Garrow (1978), writing about Martin Luther King and the ‘‘Protest at Selma,’’ argues that King was forced to accept that an approach based on ‘‘non-violent persuasion’’ (p. 221, emphasis in the original) would not bring about reforms and reduce racial injustice. Instead,
312
Robin Martin and Miles Hewstone
he shifted to a strategy of ‘‘coercive nonviolence’’ (p. 221, emphasis in the original), based no longer on convincing his opponents, but on forcing progressive change in the American South. 6.2.5. Summary One measure of the heuristic value of a new model and its associated research program is the extent to which it can guide future research. We have shown in this section that there is, first, still a rich vein to be exploited in terms of exploring the great potential of the cognitive approach to persuasion for the area of majority-minority influence. We have also proposed future research on individual differences and need states, highlighting the needs for cognition, structure, closure, uncertainty orientation, uniqueness, and distinctiveness. We have also suggested that research shift its focus from targets of influence alone, to include also sources of influence, and we have reported promising early research showing that minority sources generate more novel arguments. Finally, and more speculatively, we have called on research to integrate the findings from the literatures on minority influence and social movements, to provide a fuller understanding of the role of active minorities in bringing about real social change.
7. Conclusions Dare to be a Daniel, Dare to stand alone, Dare to have a purpose firm, Dare to make it known. (‘Dare to be a Daniel’, Salvation Army hymn; cited by Benn, 2004, p. 11). Daring ideas are like chessmen moved forward; they may be beaten, but they may start a winning game. ( Johann Wolfgang Goethe)
7.1. Majority versus minority influence: When, not whether Much time and effort in earlier studies of majority-minority influence went into answering the question of which kind of influence was stronger, majority or minority. Thanks, first, to Moscovici’s (1980) conversion theory, and, second, to the success of the contingency approaches (reviewed in Section 2.2), we believe that this question is no longer worth asking. Prior to our research, studies had already shown that the answer depends on whether attitude change is assessed in public versus private settings, and on direct versus indirect measures (Wood et al., 1994). In this chapter, we have added to this prior work by showing, consistently, that it also depends on the ‘‘context.’’
Majority Versus Minority Influence
313
According to both our review of the pertinent literature and the extensive results of our own research program, neither source, majority or minority, leads exclusively to more processing. Rather, whether the majority, or the minority, or both instigate elaborative message processing depends on a range of factors, including the nature of the topic (e.g., threatens self-interest, or not); and both cognitive (e.g., level of elaboration) and motivational (e.g., personal involvement) factors affecting ability and effort. The results of our research program show, for the first time, that minority influence is likely to lead to message processing when the topic does not threaten self-interest, and elaboration level is intermediate or high, and that this elaborative processing results in stronger attitudes that resist counter-persuasion, persist over time, and guide behavior. These findings underline an important change in the questions we should ask about this domain of social influence. We should no longer ask whether the majority or minority can instigate elaborative processing, but when either source does so. It seems that neither the majority nor the minority can persuade all of the people all of the time. Our research, using a range of attitude topics and paradigms, suggests that elaborative processing may often be the default, faced with a minority source, but that either source can, in principle and in practice, trigger elaborative processing.
7.2. Minorities, democracy, and fairness Although we tend to think of ‘‘persuasion’’ as something studied in our laboratories, Dunn (2005) has made an eloquent case for persuasion (in contrast to coercion) as being central to the practice of democracy, and he claims that it has been ever since its Athenian dawn. But, as John Stuart Mill (1859/1986), the great advocate of minorities, emphasized, within a democracy, there should not be tyranny of the majority over the minority. Democracy as majority rule is not without limitations and problems. Minorities must be protected, and we should be aware that democracy can lead to the accumulation of power and its abuse by an electoral majority. This happens in many parts of the world and, ironically, nonelected governments in some countries are considerably more liberal than democratically elected majorities (see Zakaria, 2004, for a detailed discussion). Minorities should also be listened to and engaged with. An appropriate guiding text would be the sentiment attributed to Voltaire, ‘‘I disapprove of what you say, but I will defend to the death your right to say it.’’ But because Voltaire, in fact, never said it (Pearson, 2005, traces the phrase to Tallentyre, but as a paraphrase of Voltaire’s attitude), we propose as a guiding philosophy what he did say, ‘‘I am a tolerant man, and I consider it a very good thing if people think differently from me’’ (Voltaire, 1737; quoted in Pearson, 2005). Within any democracy that is worth its name, and if it is to be a participatory and not merely a representative democracy (Ginsborg, 2005),
314
Robin Martin and Miles Hewstone
it is crucial that minorities are recognized, respected, treated fairly, and encouraged to speak up. Noelle-Neumann (1993) has referred to a ‘‘spiral of silence’’ phenomenon in the media, such that people are less likely to speak out when they believe they hold a minority opinion, which should lead other members of the minority to underestimate the consensus of their view, and members of the majority to overestimate the consensus of their view (see Glynn et al., 1997, for a review; and Hornsey et al., 2003, for evidence that under some conditions knowledge that one is in a minority position can motivate people to act even more strongly in line with their attitudes).
7.3. Epilogue The philosopher Michael Oakshott argued for the unceasing exchange of ideas between people of different beliefs and opinions that he termed ‘‘the conversation of mankind’’ (see Franco, 2004). Whether in the Supreme Court, parliament, or our faculty committees, one would hope that we would all, ultimately, be swayed by the best arguments, sometimes even arguments we do not initially personally agree with, made by people with whom we do not normally agree (what legal experts call ‘‘deliberativeness’’), and expressed by those in a numerical minority. Indeed, while writing this piece, one of us, an aggressive nonsmoker, found himself sitting in a college meeting at which the authorities were just about to ban smoking completely from all premises. He found himself, first, intrigued by the courage of the lone spokesman for the tiny rump of collegiate smokers; then listening to the eloquent, cogent arguments; then changing his mind about banning smoking from the college; and finally voting against it! This is the supreme gift of minorities: they can make us listen, to a new voice, and sometimes they can persuade us; but above all they often provide a novel perspective. Going back to the halcyon days of Solomon Asch’s (1956) studies of majority influence, and the subsequent ‘‘dependency’’ perspective on group influence, we are struck by the parallels between Serge Moscovici’s own minority position arguing for the influence of minorities against the dominant perspective of majority influence, and the very phenomenon whose study he initiated. We believe that there is also a certain romance to minority influence, and we admit that we have succumbed to it. Inspired many years ago, as students, by the original, consistent, and exciting message of Moscovici’s (1976, 1980; Moscovici et al., 1969) writing on the topic, we have both been not merely influenced, but persuaded. As our own experiments have shown, minority messages seem to have an advantage when it comes to perlustration, inviting attention and exerting an influence that can be direct, but is often indirect, latent, and removed in time (Wood et al., 1994). They can, moreover, help us to see the world differently, adopt new, more critical perspectives and ideas, and be agents of social change (Nemeth, 1986; Moscovici, 1976). The value
Majority Versus Minority Influence
315
of minority influence, with its ability to shift not just what we think but how we think, reminds us of the sage advice of Francis Bacon (1625, Book I. i. 3), ‘‘If a man will begin with certainties, he shall end in doubts; but if he will be content to begin with doubts, he shall end in certainties.’’ Yet, after the initial impact of Moscovici’s ideas in the 1970s, taken up by several European scholars (notably Mugny, 1982; Mugny and Pe´rez, 1991), the area of majority-minority seemed to have slipped into a period of torpor, as if there were no more work to be done, or no new theories and methods to guide the work. It is almost as if the very topic, its theories, and its pioneers have been forgotten. Astonishingly, a recent volume on ‘‘social influence’’ (Pratkanis, 2007)—listing the key phenomena/theories of the field of social influence, and its pioneers—failed to list either minority influence, or Serge Moscovici. We believe that the marriage of European and North American ideas adopted in our theoretical model (splicing conversion theory and cognitive models of persuasion) has brought new understanding to this area, making sense of a heretofore disparate set of results; we hope that our new integration will help to re-invigorate the area, and guide much further research too.
ACKNOWLEDGMENTS The publication of this chapter, to which we have made equal contribution, gives us an opportunity to integrate over a decade of joint research. Many of the ideas were hammered out over all-day breakfasts and Indian meals, in various locations around the world, but we hope that the present product, notwithstanding its bulk, is not equally indigestible. We gratefully acknowledge various sources of funding over the years for this program of work, including (to both authors) the Economic and Social Research Council, and (to Robin Martin), the Australian Research Council. We also thank the graduate students, research staff, and other collaborators who have made major contributions to this endeavor; these are made explicit in authorship of primary studies, but we especially thank Antonis Gardikiotis, Jared Kenworthy, John Levine, Pearl Martin, and Joanne Smith. We are very grateful to the following people for comments on a previous draft of this chapter: Gerd Bohner, Bill Crano, Carsten de Dreu, Jared Kenworthy, Pearl Martin, Serge Moscovici, Gabriel Mugny, Charlan Nemeth, Rich Petty, Wolfgang Stroebe, Zak Tormala, and Mark Zanna.
REFERENCES Alvaro, E. M., and Crano, W. D. (1997). Indirect minority influence: Evidence for leniency in source evaluation and counterargumentation. J. Pers. Soc. Psychol. 72, 949–964. Amodio, D. M., and Frith, C. D. (2006). Meeting of minds: The medial frontal cortex and Social cognition. Nat. Rev. Neurosci. 7, 268–277. Asch, S. E. (1951). Effects of group pressure upon the modification and distortion of judgments. In ‘‘Groups, Leadership, and Men’’ (H. Guetzhow, ed.), pp. 177–190. Carnegie Press, Pittsburgh, PA. Asch, S. E. (1956). Studies of independence and conformity. A minority of one against a unanimous majority. Psychol. Monogr. Gen. Appl. Whole No. 416.
316
Robin Martin and Miles Hewstone
Bacon, F. (1625). Essays or counsels, civil and moral: On studies. In ‘‘The Major Works’’ (F. Bacon, ed.). Oxford University Press (1996 edn.), Oxford. Baker, S. M. (1992). ‘‘Majority and Minority Influence: A Message Processing Perspective.’’ Ohio State University, Unpublished doctoral dissertation, Columbus, Ohio. Baker, S. M., and Petty, R. E. (1994). Majority and minority influence: Source-position imbalance as a determinant of message scrutiny. J. Pers. Soc. Psychol. 67, 5–19. Bargh, J. A. (1982). Attention and automaticity in the processing of self-relevant information. J. Pers. Soc. Psychol. 43, 425–436. Bargh, J. A., and Pietromonaco, P. (1982). Automatic information processing and social perception: The influence of trait information presented outside of conscious awareness on impression formation. J. Pers. Soc. Psychol. 43, 437–449. Baron, R. S., and Bellman, S. B. (2007). No guts, no glory: Courage, harassment and minority influence. Eur. J. Soc. Psychol. 37, 101–124. Baron, R. M., and Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. J. Pers. Soc. Psychol. 51, 1173–1182. Benn, T. (2004). ‘‘Dare to be a Daniel: Then and now.’’ Arrow Books, London. Bohner, G., Moskowitz, G., and Chaiken, S. (1995). The interplay of heuristic and systematic processing of social information. In ‘‘European Review of Social Psychology’’ (W. Stroebe and M. Hewstone, eds.), Vol. 6, pp. 33–68. John Wiley, Chichester. Bohner, G., Frank, S., and Erb, H. P. (1998). Heuristic processing of distinctiveness information in minority and majority influence. Eur. J. Soc. Psychol. 28, 855–860. Brandsta¨tter, V., Ellemers, N., Gaviria, E., Giosue, F., Huguet, P., Kroon, M., Morchain, P., Pujal, M., Rubini, M., Mugny, G., and Pe´rez, J. A. (1991). Indirect majority and minority influence: An exploratory study. Eur. J. Soc. Psychol. 21, 199–211. Brewer, M. B. (1991). The social self: On being the same and different at the same time. Pers. Soc. Psychol. Bull. 17, 475–482. Brodbeck, F. C., Kerschreiter, R., Mojzisch, A., Frey, D., and Schulz-Hardt, S. (2002). The dissemination of critical, unshared information in decision making groups: The effects of pre-discussion dissent. Eur. J. Soc. Psychol. 32, 35–56. Butera, F., and Mugny, G. (1995). Conflict between incompetences and influence of a lowexpertise source in hypothesis testing. Eur. J. Soc. Psychol. 25, 457–462. Butera, F., Mugny, G., Legrenzi, P., and Pe´rez, J. A. (1996). Majority and minority influence, task representation and inductive reasoning. Br. J. Soc. Psychol. 35, 123–136. Cacioppo, J. T., and Petty, R. E. (1982). The need for cognition. J. Pers. Soc. Psychol. 42, 116–131. Cacioppo, J. T., Petty, R. E., Kao, C. F., and Rodriguez, R. (1986a). Central and peripheral routes to persuasion: An individual difference perspective. J. Pers. Soc. Psychol. 51, 1032–1043. Cacioppo, J. T., Petty, R. E., Losch, M. E., and Kim, H. S. (1986b). Electromyographic activity over facial muscle regions can differentiate the valence and intensity of affective reactions. J. Pers. Soc. Psychol. 50, 260–268. Cacioppo, J. T., Petty, R. E., Feinstein, J. A., and Jarvis, W. B. G. (1996). Dispositional differences in cognitive motivation: The life and times of individuals varying in need for cognition. Psychol. Bull. 119, 197–253. Campbell, J. D., Tesser, A., and Fairey, P. J. (1986). Conformity and attention to the stimulus: Some temporal and contextual dynamics. J. Pers. Soc. Psychol. 51, 315–324. Chaiken, S., and Stangor, C. (1987). Attitudes and attitude change. Annu. Rev. Psychol. 38, 575–630. Chaiken, S., Liberman, A., and Eagly, A. H. (1989). Heuristic and systematic information processing within and beyond the persuasion context. In ‘‘Unintended Thought: Limits
Majority Versus Minority Influence
317
of Awareness, Intention and Thought’’ ( J. S. Uleman and J. A. Bargh, eds.), pp. 212–252. Guilford, New York. Craik, F. I. M., and Lockhart, R. S. (1972). Levels of processing: A framework for memory research. J. Verb. Learn. Verb. Behav. 11, 671–684. Craik, F. I. M., and Tulving, E. (1975). Depth of processing: A framework for memory research. J. Verb. Learn. Verb. Behav. 11, 671–684. Crano, W. D. (1995). Attitude strength and vested interest. In ‘‘Attitude Strength: Antecedents and Consequences. The Ohio State University Series in Attitudes and Persuasion’’ (R. E. Petty and J. A. Krosnick, eds.), Vol. 4, pp. 131–157. Erlbaum, Hillsdale, NJ. Crano, W. D. (2001). Social influence, social identity, and ingroup leniency. In ‘‘Group Consensus and Innovation’’ (C. K. W. De Dreu and N. K. De Vries, eds.), pp. 122–143. Blackwell, Oxford. Crano, W. D. (in press). Majority and minority influence in attitude formation and attitude change: Context/categorization—leniency contract theory. In ‘‘Minority Influence and Innovation: Antecedents, Processes and Consequences’’ (R. Martin and M. Hewstone, eds.). Psychology Press, Hove, E. Sussex. Crano, W. D., and Alvaro, E. M. (1998a). The context/comparison model of social influence: Mechanisms, structure, and linkages that underlie indirect attitude change. In ‘‘European Review of Social Psychology’’ (W. Stroebe and M. Hewstone, eds.), Vol. 8, pp. 175–202. John Wiley, Chichester. Crano, W. D., and Alvaro, E. M. (1998b). Indirect minority influence: The leniency contract revisited. Group Processes and Intergroup Relations 1, 99–115. Crano, W. D., and Chen, X. (1998). The leniency contract and persistence of majority and minority influence. J. Pers. Soc. Psychol. 74, 1437–1450. Crano, W. D., and Hannula-Bral, K. A. (1994). Context/categorization model of social influence: Minority and majority influence in the formation of a novel response norm. J. Exp. Soc. Psychol. 30, 247–276. David, B., and Turner, J. C. (1996). Studies in self-categorization and minority conversion: Is being a member of the outgroup an advantage? Br. J. Soc. Psychol. 35, 179–199. David, B., and Turner, J. C. (1999). Studies in self-categorization and minority conversion: The ingroup minority in intragroup and intergroup contexts. Br. J. Soc. Psychol. 38, 115–134. David, B., and Turner, J. C. (2001). Majority and minority influence: A single process selfcategorization analysis. In ‘‘Group Consensus and Innovation’’ (C. K. W. De Dreu and N. K. De Vries, eds.), pp. 91–121. Blackwell, Oxford. De Dreu, C. K. W. (2007). Minority dissent, attitude change, and group performance. In ‘‘The Science of Social Influence: Advances and Future Progress’’ (A. R. Pratkanis, ed.), pp. 247–270. Psychology Press, New York. De Dreu, C. K. W., and De Vries, N. K. (1993). Numerical support, information processing and attitude change. Eur. J. Soc. Psychol. 23, 647–663. De Dreu, C. K. W., and De Vries, N. K. (1996). Differential processing and attitude change following majority and minority arguments. Br. J. Soc. Psychol. 35, 77–90. De Dreu, C. K. W., and De Vries, N. K. (Eds.) (2001).‘‘Group Consensus and Innovation.’’ Blackwell, Oxford. De Dreu, C. K. W., and Koole, S. L. (1997). Motivated use of heuristics in negotiation. Unpublished data. University of Amsterdam, Amsterdam, The Netherlands. De Dreu, C. K. W., De Vries, N. K., Gordijn, E., and Schuurman, M. (1999). Convergent and divergent processing of majority and minority arguments: Effects on focal and related attitudes. Eur. J. Soc. Psychol. 29, 329–348. De Vries, N. K., De Dreu, N. K., Gordijn, E., and Schuurman, M. (1996). Majority and minority influence: A dual interpretation. In ‘‘European Review of Social Psychology’’ (W. Stroebe and M. Hewstone, eds.), Vol. 7, pp. 145–172. John Wiley, Chichester.
318
Robin Martin and Miles Hewstone
Deutsch, M., and Gerard, H. G. (1955). A study of normative and informational social influence upon individual judgment. J. Abnorm. Soc. Psychol. 51, 629–636. Dion, K. (2002). The social psychology of perceived prejudice and discrimination. Can. Psychol. 43, 1–10. Dunn, J. (2005). ‘‘Setting the People Free: The Story of Democracy.’’ Atlantic Books, London. Dworkin, R. (1993). ‘‘Life’s Dominion: An Argument About Abortion and Euthanasia.’’ Harper Collins, London. Eagly, A. H., and Chaiken, S. (1993). ‘‘The Psychology of Attitudes.’’ Harcourt Brace Jovanovich, Fort Worth, TX. Easterbrook, J. A. (1959). The effect of emotion on the utilization and the organization of behavior. Psychol. Rev. 66, 183–201. Erb, H.-P., and Bohner, G. (2007). Social influence and persuasion: Recent theoretical developments and integrative attempts. In ‘‘Social Communication’’ (K. Fiedler, ed.), pp. 191–221. Psychology Press, New York. Erb, H.-P., and Bohner, G. (in press). Consensus as the key: Towards parsimony in explaining majority and minority influence. In ‘‘Minority Influence and Innovation: Antecedents, Processes and Consequences’’ (R. Martin and M. Hewstone, eds.). Psychology Press, Hove, E. Sussex. Erb, H.-P., Bohner, G., Schmaelzle, K., and Rank, S. (1998). Beyond conflict and discrepancy: Cognitive bias in minority and majority influence. Pers. Soc. Psychol. Bull. 24, 620–633. Erb, H.-P., Bohner, G., Rank, S., and Einwiller, S. (2002). Processing minority and majority communications: The role of conflict with prior attitudes. Pers. Soc. Psychol. Bull. 28, 1172–1182. Erb, H.-P., Bohner, G., Hewstone, M., Werth, L., and Reinhard, M.-A. (2006). Large minorities and small majorities: Interactive effects of inferred and explicit consensus on attitudes. Basic Appl. Soc. Psychol. 28, 221–231. Falomir, J. M., Mugny, G., and Pe´rez, J. A. (2000). Social influence and identity conflict. In ‘‘Attitudes, Behavior, and Social Context: The Role of Norms and Group Membership’’ (D. Terry and M. Hogg, eds.), pp. 245–246. Erlbaum, Mahwah, NJ. Falomir-Pichastor, J. M., Butera, F., and Mugny, G. (2002). Persuasive constraint and expert versus non-expert influence in intention to quit smoking. Eur. J. Soc. Psychol. 32, 209–222. Fazio, R. H. (1995). Attitudes as object-evaluation associations: Determinants, consequences, and correlates of attitude accessibility. In ‘‘Attitude Strength: Antecedents and Consequences’’ (R. E. Petty and J. A. Krosnick, eds.). Erlbaum, Hillsdale, NJ. Festinger, L. (1950). Informal social communication. Psychol. Rev. 57, 271–282. Fleming, M. A., Petty, R. E., and White, P. H. (2005). Stigmatized targets and evaluation: Prejudice as a determinant of attribute scrutiny and polarization. Pers. Soc. Psychol. Bull. 31, 496–507. Franco, P. (2004). ‘‘Michael Oakshott: An introduction.’’ Yale University Press, New Haven, CO & London. Frey, D. (1986). Recent research on selective exposure to information. In ‘‘Advances in Experimental Social Psychology’’ (L. Berkowitz, ed.), Vol. 19, pp. 41–80. Academic Press, New York. Gardikiotis, A. (2005). Majority and minority source and processing of the influential message. Psychology: The J. Hellenic Psychol. Soc. 12, 455–469. Gardikiotis, A., Martin, R., and Hewstone, M. (2004). The representation of majorities and minorities in the British press: A content analytic approach. Eur. J. Soc. Psychol. 34, 637–646. Gardikiotis, A., Martin, R., and Hewstone, M. (2005). Group consensus in social influence: Type of consensus information as a moderator of majority and minority influence. Pers. Soc. Psychol. Bull. 31, 1163–1174.
Majority Versus Minority Influence
319
Garlick, R., and Mongeau, P. A. (1993). Argument quality and group member status as determinants of attitudinal minority influence. Western J. Comm. 57, 289–308. Garrow, D. (1978). ‘‘Protest at Selma.’’ Yale University Press, London and New Haven, CO. Gerard, H. B. (1985). When and how the minority prevails. In ‘‘Perspectives on Minority Influence’’ (S. Moscovici, G. Mugny, and E. V. Avermaet, eds.), pp. 171–186. Cambridge University Press, Cambridge, England. Ginsborg, P. (2005). ‘‘The politics of everyday life: Making choices, changing lives.’’ Yale University Press, New Haven, CO & London. Glynn, C. J., Hayes, A. F., and Shanahan, J. (1997). Perceived support for one’s opinions and willingness to speak out: A meta-analysis of survey studies on the spiral of silence. Public Opin. Q. 61, 452–463. Gordijn, E. H. (1998). ‘‘Being Persuaded or Persuading Oneself: The Influence of Numerical Support on Attitudes and Information Processing.’’ Unpublished Doctoral dissertation, University of Amsterdam, The Netherlands. Gordijn, E. H., Postmes, T., and de Vries, N. (2001). Devil’s advocate or advocate of oneself: Effects of numerical support on pro- and counterattitudinal self-persuasion. Pers. Soc. Psychol. Bull. 27, 395–407. Gordijn, E. H., De Vries, N. K., and De Dreu, C. K. W. (2002). Minority influence on focal and related attitudes: Change in size, attributions and information processing. Pers. Soc. Psychol. Bull. 28, 1315–1326. Gross, S. R., and Miller, N. (1997). The ‘‘Golden Section’’ and bias in perceptions of social consensus. Pers. Soc. Psychol. Rev. 1, 241–271. Guillon, M., and Personnaz, B. (1983). Analyse de la dynamique des representations des conflits minoritaire et majoritaire. Cah. Psychol. Cogn. 3, 65–87. Guinote, A. (2008). Behavior variability and the situated focus theory of power. In ‘‘European Review of Social Psychology’’ (W. Stroebe and M. Hewstone, eds.), Vol. 18, 256–295. Psychology Press (Taylor & Francis), Hove, E. Sussex. Guinote, A., Brown, M., and Fiske, S. (2006). Minority status decreases sense of control and increases interpretive processing. Soc. Cogn. 24, 169–186. Harmer, C. J., Thilo, K. V., Rothwell, J. C., and Goodwin, G. M. (2001). Transcranial magnetic stimulation of medial-frontal cortex impairs the processing of angry facial expressions. Nat. Neurosci. 4, 17–18. Haugtvedt, C. P., and Petty, R. E. (1992). Personality and persuasion: Need for cognition moderates the persistence and resistance of attitude change. J. Pers. Soc. Psychol. 63, 308–319. Haugtvedt, C. P., and Wegener, D. T. (1994). Message order effects in persuasion: An attitude strength perspective. J. Consum. Res. 21, 205–218. Hitchens, C. (2001). ‘‘Letters to a Young Contrarian.’’ Basic Books, New York, NJ. Hochschild, A. (2005). ‘‘Bury the Chains: Prophets and Rebels in the Fight to Free an Empire’s Slaves.’’ Houghton Mifflin. New York, NY. Holland, R., Verplanken, B., and van Knippenberg, A. (2002). On the nature of attitudebehavior relations: The strong guide, the weak follow. Eur. J. Soc. Psychol. 32, 869–876. Hornsey, M. J., Majkut, L., Terry, D. J., and McKimmie, B. M. (2003). On being loud and proud: Non-conformity and counter-conformity to group norms. Br. J. Soc. Psychol. 42, 319–335. Houston, D. A., and Fazio, R. H. (1989). Biased processing as a function of attitude accessibility: Making objective judgments subjectively. Soc. Cogn. 7, 51–66. Imhoff, R., and Erb, H.-P. (2006). cited in Erb, H.-P., and Bohner, G. (in press). Consensus as the key: Towards parsimony in explaining majority and minority influence. In ‘‘Minority Influence and Innovation: Antecedents, Processes and Consequences’’ (R. Martin and M. Hewstone, eds.). Psychology Press, Hove, E. Sussex.
320
Robin Martin and Miles Hewstone
Ito, T. A., and Urland, G. R. (2003). Race and gender on the brain: Electrocortical measures of attention to race and gender of multiply categorizable individuals. J. Pers. Soc. Psychol. 85, 616–626. Johnson, S. (2006). ‘‘The Ghost Map: The Story of London’s Most Terrifying Epidemic – and how it Changed Science, Cities, and the Modern World?’’ Riverhead books, New York. Johnson, B. T., and Eagly, A. H. (1989). Effects of involvement on persuasion: A metaanalysis. Psychol. Bull. 106, 290–314. Joule, R. V., Mugny, G., and Pe´rez, J. A. (1988). When a compliance without pressure strategy fails due to a minority dissenter. A case of ‘‘behavioral conversion’’ Eur. J. Soc. Psychol. 18, 531–535. Kelley, H. H. (1967). Attribution theory in social psychology. In ‘‘Nebraska Symposium on Motivation’’ (D. Levine, ed.), pp. 192–241. University of Nebraska Press, Lincoln. Kenworthy, J. B., and Miller, N. (2001). Perceptual asymmetry in consensus estimates of majority and minority members. J. Pers. Soc. Psychol. 80, 597–612. Kenworthy, J. B., and Miller, N. (2002). Attributional biases about the origins of attitudes: Externality, emotionality, and rationality. J. Pers. Soc. Psychol. 82, 693–707. Kenworthy, J. B., Hewstone, M., Levine, J. M., Martin, R., and Willis, H. (in press). ‘‘The Phenomenology of Minority-Majority Status: Effects on Innovation in Argument Generation.’’ Eur. J. Soc. Psychol. Kerr, N. L. (2001). Is it what one says or how one says it? Style vs. substance from an SDS perspective. In ‘‘Group Consensus and Minority Influence: Implications for Innovation’’ (C. K. W. De Dreu and N. K. De Vries, eds.), pp. 201–228. Blackwell, Oxford, UK. Kerr, N. (2002). When is a minority a minority? Active vs. passive minority advocacy and social influence. Eur. J. Soc. Psychol. 32, 471–484. Klandermans, B. (1997). ‘‘The Social Psychology of Social Protest.’’ MA: Blackwell, Oxford, UK & Cambridge. Krosnick, J. A., and Petty, R. E. (1995). Attitude strength: An overview. In ‘‘Attitude Strength: Antecedents and Consequences’’ (R. E. Petty and J. A. Krosnick, eds.), pp. 1–24. Erlbaum, Hillsdale, NJ. Krosnick, J. A., Boninger, D. S., Chuang, Y. C., Berent, M. K., and Carnot, C. (1993). Attitude strength: One construct or many related constructs?. J. Pers. Soc. Psychol. 65, 1132–1151. Kruglanski, A. W., and Mackie, D. M. (1990). Majority and minority influence: A judgmental process analysis. In ‘‘European Review of Social Psychology’’ (W. Stroebe and M. Hewstone, eds.), Vol. 1, pp. 229–261. John Wiley, Chichester. Kruglanski, A. W., and Mayseless, O. (1987). Motivational effects in the social comparison of opinions. J. Pers. Soc. Psychol. 53, 834–842. Kruglanski, A. W., Webster, D. M., and Klem, A. (1993). Motivated resistance and openness to persuasion in the presence or absence of prior information. J. Pers. Soc. Psychol. 65, 861–876. Latane´, B., and Wolf, S. (1981). The social impact of majorities and minorities. Psychol. Rev. 88, 438–453. Legrenzi, P., Butera, F., Mugny, G., and Pe´rez, J. A. (1991). Majority and minority influence in inductive reasoning: A preliminary study. Eur. J. Soc. Psychol. 21, 359–363. Leippe, M. R., and Elkin, R. A. (1987). When motives clash: Issue involvement and response involvement as determinants of persuasion. J. Pers. Soc. Psychol. 52, 269–278. Leonardelli, G., and Brewer, M. B. (2001). Minority and majority discrimination: When and why. J. Exp. Soc. Psychol. 37, 468–485. Levine, J. M. (1980). Reaction to opinion deviance in small groups. In ‘‘Psychology of Group Influence’’ (P. B. Paulus, ed.), pp. 375–429. Lawrence Erlbaum Associates, Hillsdale, NJ.
Majority Versus Minority Influence
321
Levine, J. M., and Kaarbo, J. (2001). Minority influence in political decision-making groups. In ‘‘Group Consensus and Minority Influence: Implications for Innovation’’ (C. K. W. De Dreu and N. K. De Vries, eds.), pp. 229–257. Blackwell, Oxford, UK. Levine, J. M., and Russo, E. M. (1987). Majority and minority influence. In ‘‘Group processes: Review of Personality and Social Psychology’’ (C. Hendrick, ed.), Vol. 8, pp. 13–54. Sage, Newbury Park, CA. Levine, J., and Russo, E. (1995). Impact of anticipated interaction on information acquisition. Soc. Cogn. 13, 293–317. Levine, J., Bogart, L., and Zdaniuk, B. (1996). Impact of anticipated group membership on Cognition. In ‘‘Handbook of Motivation and Cognition, Vol. 3: The Interpersonal Context’’ (R. M. Sorrentino and E. T. Higgins, eds.), pp. 531–569. Guilford Press, New York. Lord, C. G., Ross, L., and Lepper, M. R. (1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. J. Pers. Soc. Psychol. 37, 2098–2109. Lu¨cken, M., and Simon, B. (2005). Cognitive and affective experiences of minority and majority members: The role of group size, status, and power. J. Exp. Soc. Psychol. 41, 396–413. Maass, A., and Clark, R. D., III (1983). Internalization versus compliance: Differential processes underlying minority influence and conformity. Eur. J. Soc. Psychol. 13, 197–215. Maass, A., West, S., and Cialdini, R. B. (1987). Minority influence and conversion. In ‘‘Group Processes: Review of Personality and Social Psychology’’ (C. Hendrick, ed.), Vol. 8, pp. 55–79. Sage, Newbury Park, CA. Mackie, D. M. (1987). Systematic and nonsystematic processing of majority and minority persuasive communications. J. Pers. Soc. Psychol. 53, 41–52. Macrae, C. N., Milne, A. B., and Bodenhausen, G. V. (1994). Stereotypes as energy-saving devices: A peek inside the cognitive toolbox. J. Pers. Soc. Psychol. 66, 37–47. Maheswaran, D., and Chaiken, S. (1991). Promoting systematic processing in lowmotivation settings: Effects of incongruent information on processing and judgment. J. Pers. Soc. Psychol. 61, 13–25. Martin, P. Y., and Martin, R. (2006). The effects of caffeine consumption on direct and indirect majority and minority influence. J. Appl. Soc. Psychol. 36, 1961–1979. Martin, R. (1988a). Ingroup and outgroup minorities: Differential impact upon public and private responses. Eur. J. Soc. Psychol. 18, 39–52. Martin, R. (1988b). Minority influence and social categorization: A replication. Eur. J. Soc. Psychol. 18, 369–373. Martin, R. (1996). Minority influence and argument generation. Br. J. Soc. Psychol. 35, 91–103. Martin, R. (1998). Majority and minority influence using the afterimage paradigm: A series of attempted replications. J. Exp. Soc. Psychol. 34, 1–26. Martin, R., and Hewstone, M. (1999). Minority influence and optimal problem-solving. Eur. J. Soc. Psychol. 29, 825–832. Martin, R., and Hewstone, M. (2001). Determinants and consequences of cognitive processes in majority and minority influence. In ‘‘Social Influence: Direct and Indirect Processes’’ ( J. Forgas and K. Williams, eds.), pp. 315–330. Psychology Press, Philadelphia. Martin, R., and Hewstone, M. (2003a). Social influence processes of control and change: Conformity, obedience to authority, and innovation. In ‘‘Sage Handbook of Social Psychology’’ (M. A. Hogg and J. Cooper, eds.), pp. 347–366. Sage, London.
322
Robin Martin and Miles Hewstone
Martin, R., and Hewstone, M. (2003b). Majority versus minority influence: When, not whether, source status instigates heuristic or systematic processing. Eur. J. Soc. Psychol. 33, 313–330. Martin, R., and Hewstone, M. (Eds.) (in press). ‘‘Minority influence and innovation: Antecedents, processes and consequences.’’ Psychology Press, Hove, E. Sussex. Martin, R., and Hewstone, M. (in preparation). ‘‘Attitude Persistence to Persuasive Messages as a Function of Majority and Minority Source Status.’’ Manuscript in preparation. Martin, R., Gardikiotis, A., and Hewstone, M. (2002). Levels of consensus and majority and minority influence. Eur. J. Soc. Psychol. 32, 645–665. Martin, R., Hewstone, M., and Martin, P. Y. (2003). Resistance to persuasive messages as a function of majority and minority source status. J. Exp. Soc. Psychol. 39, 585–593. Martin, R., Hewstone, M., and Martin, P. Y. (2007a). Systematic and heuristic processing of majority- and minority-endorsed messages: The effects of varying outcome relevance and levels of orientation on attitude and message processing. Pers. Soc. Psychol. Bull. 33, 43–56. Martin, R., Martin, P. Y., Smith, J., and Hewstone, M. (2007b). Majority and minority influence and prediction of behavioral intentions and behavior. J. Exp. Soc. Psychol. 43, 763–771. Martin, R., Hewstone, M., and Martin, P. Y. (2008a). Majority versus minority influence: The role of message processing in determining resistance to counter-persuasion. Eur. J. Soc. Psychol. 38, 16–34. Martin, R., Hewstone, M., Martin, P. Y., and Gardikiotis, A. (2008b). Persuasion from majority and minority groups. In ‘‘Attitudes and Attitude Change’’ (W. Crano and R. Prislin, eds.). pp. 361–384. Psychology Press, New York. Martin, R., Hewstone, M., Martin, P. Y. (under review). ‘‘The Role of Message Processing and Attitude Change in Majority and Minority Influence: The Moderating Effect of Positive vs. Negative Personal Outcome.’’ Manuscript under review. Maslach, C. (1974). Social and personal bases of individuation. J. Pers. Soc. Psychol. 29, 411–425. McGuire, W. J. (1964). Inducing resistance to persuasion: Some contemporary approaches. In ‘‘Advances in Experimental Social Psychology’’ (L. Berkowitz, ed.), Vol. 1, pp. 191–229. Academic Press, San Diego, CA. Mill, J. S. (1861/1924). ‘‘Considerations on Representative Government.’’ Oxford University Press, London. Mill, J. S. (1859/1986). ‘‘On Liberty.’’ Prometheus Books, New York, NY. Miller, N., and Colman, D. E. (1981). Methodological issues in analyzing cognitive mediation in persuasion. In ‘‘Cognitive Responses in Persuasion’’ (R. E. Petty, T. M. Ostrom, and T. C. Brock, eds.), pp. 105–125. Erlbaum, Hillsdale, NJ. Moscovici, S. (1976). ‘‘Social Influence and Social Change.’’ Academic Press, London. Moscovici, S. (1980). Toward a theory of conversion behavior. In ‘‘Advances in Experimental Social Psychology’’ (L. Berkowitz, ed.), Vol. 13, pp. 209–239. Academic Press, New York. Moscovici, S. (1985). Social influence and conformity. In ‘‘The Handbook of Social Psychology’’ (G. Lindsey and E. Aronson, eds.), 3rd edn., Vol. 2, pp. 347–412. Random House, New York. Moscovici, S., and Faucheux, C. (1972). Social influence, conformity bias and the study of active minorities. In ‘‘Advances in Experimental Social Psychology’’ (L. Berkowitz, ed.), Vol. 6, pp. 149–202. Academic Press, New York. Moscovici, S., and Lage, E. (1976). Studies in social influence III: Majority versus minority influence in a group. Eur. J. Soc. Psychol. 6, 149–174.
Majority Versus Minority Influence
323
Moscovici, S., and Nemeth, C. (1974). Social influence II: Minority influence. In ‘‘Social Psychology: Classic and Contemporary Integrations’’ (C. Nemeth, ed.), pp. 217–249. Rand McNally, Chicago. Moscovici, S., and Personnaz, B. (1980). Studies in social influence. V. Minority influence and conversion behavior in a perceptual task. J. Exp. Soc. Psychol. 16, 270–282. Moscovici, S., and Personnaz, B. (1991). Studies in social influence VI: Is Lenin orange or red? Imagery and social influence. Eur. J. Soc. Psychol. 21, 101–118. Moscovici, S., Lage, E., and Naffrechoux, M. (1969). Influence of a consistent minority on the responses of a majority in a color perception task. Sociometry 32, 365–380. Moscovici, S., Mugny, G., and Papastamou, S. (1981). ‘‘Sleeper effect’’ et/ou effet minoritaire? Etude the´orique et expe´rimentale de l’influence sociale a` retardement. Cah. Psychol. Cogn. 1, 199–221. Mucchi-Faina, A., Maass, A., and Volpato, C. (1991). Social influence: The role of originality. Eur. J. Soc. Psychol. 21, 183–197. Mugny, G. (1982). ‘‘The Power of Minorities.’’ Academic Press, London. Mugny, G. (1984). Compliance, conversion and the Asch paradigm. Eur. J. Soc. Psychol. 14, 353–368. Mugny, G., and Butera, F. (1995). Influence majoritaire et minoritaire: Vers une inte´gration. Psychol. Franc¸aise 40, 339–346. Mugny, G., and Papastamou, S. (1980). When rigidity does not fail: Individualization and psychologization as resistances to the diffusion of minority innovations. Eur. J. Soc. Psychol. 10, 43–61. Mugny, G., and Pe´rez, J. (1991). ‘‘The Social Psychology of Minority Influence.’’ Cambridge University Press, Cambridge. Mugny, G., Butera, F., Sanchez-Mazas, M., and Pe´rez, J. A. (1995). Judgements in conflict: The conflict elaboration theory of social influence. In ‘‘Perception-Evaluation-Interpretation’’ (B. Boothe, R. Hirsig, A. Helminger, B. Meier, and R. Volkart, eds.), pp. 160–168. Hogrefe and Huber, Go¨ttingen. Nemeth, C. (1986). Differential contributions of majority and minority influence. Psychol. Rev. 93, 23–32. Nemeth, C. (1995). Dissent as driving cognition, attitudes and judgements. Soc. Cogn. 13, 273–291. Nemeth, C. (2003). Minority dissent and its ‘‘Hidden’’ benefits. New Rev. Soc. Psychol. 2, 21–28. Nemeth, C. J., and Kwan, J. (1985). Originality of word associations as a function of majority and minority influence. Soc. Psychol. Q. 48, 277–282. Nemeth, C. J., and Kwan, J. (1987). Minority influence, divergent thinking and detection of correct solutions. J. Appl. Soc. Psychol. 17, 788–799. Nemeth, C., and Rogers, J. (1996). Dissent and the search for information. Br. J. Soc. Psychol. 35, 67–76. Nemeth, C. J., and Wachtler, J. (1983). Creative problem solving as a result of majority vs. minority influence. Eur. J. Soc. Psychol. 13, 45–55. Nemeth, C., Mayseless, O., Sherman, J., and Brown, Y. (1990). Exposure to dissent and recall of information. J. Pers. Soc. Psychol. 58, 429–437. Nemeth, C., Mosier, K., and Chiles, C. (1992). When convergent thought improves performance: Majority vs. minority influence. Pers. Soc. Psychol. Bull. 18, 139–144. Neuberg, S. L., and Newsom, J. T. (1993). Personal need for structure: Individual differences in the desire for simple structure. J. Pers. Soc. Psychol. 65, 113–131. Neuberg, S. L., Judice, T. N., and West, S. G. (1997). What the Need for Closure Scale measures and what it does not: Toward differentiating among related epistemic motives. J. Pers. Soc. Psychol. 72, 1396–1412.
324
Robin Martin and Miles Hewstone
Ng, R. Y., and Van Dyne, L. (2001). Individualism-collectivism as a boundary condition for effectiveness of minority influence in decision making. Organ. Behav. Hum. Decis. Process. 84, 198–225. Noelle-Neumann, E. (1993). ‘‘Spiral of Silence: Public Opinion – our Social Skin.’’ 2nd edn. University of Chicago Press, Chicago. Ochsner, K. N., and Lieberman, M. D. (2001). The emergence of social cognitive neuroscience. Am. Psychol. 56, 717–734. Pearson, R. (2005). ‘‘Voltaire Almighty.’’ Bloomsbury, London. Pendry, L. F., and Macrae, C. N. (1994). Stereotypes and mental life: The case of the motivated but thwarted tactician. J. Exp. Soc. Psychol. 30, 303–325. Pe´rez, J. A., and Mugny, G. (1996). The conflict elaboration theory of social influence. In ‘‘Understanding Group Behavior: Small Group Processes and Interpersonal Relations’’ (E. H. Witte and J. H. Davis, eds.), Vol. 2, pp. 191–210. Lawrence Erlbaum Associates, New Jersey. Personnaz, B. (1981). Study in social influence using the spectrometer method: Dynamics of the phenomena of conversion and covertness in perceptual responses. Eur. J. Soc. Psychol. 11, 431–438. Peterson, R., and Nemeth, C. J. (1996). Focus versus flexibility: Majority and minority influence can both improve performance. Pers. Soc. Psychol. Bull. 22, 14–23. Petty, R. E. (1995). Attitude change. In ‘‘Advanced Social Psychology’’ (A. Tesser, ed.), pp. 195–255. McGraw-Hill, New York. Petty, R. E., and Cacioppo, J. T. (1984). Source factors and the Elaboration Likelihood Model of persuasion. In ‘‘Advances in Consumer Research’’ (T. C. Kinnear, ed.), Vol. 11, pp. 668–672. Association for Consumer Research Provo, UT. Petty, R. E., and Cacioppo, J. T. (1986). ‘‘Communication and Persuasion: Central and Peripheral Routes to Attitude Change.’’ Springer-Verlag, New York. Petty, R. E., and Krosnick, J. A. (Eds.) (1995). ‘‘Attitude Strength: Antecedents and Consequences.’’ Erlbaum, Hillsdale, NJ. Petty, R. E., and Wegener, D. T. (1998). Attitude change: Multiple roles for persuasion variables. In ‘‘The Handbook of Social Psychology’’ (D. T. Gilbert, S. T. Fiske, and G. Lindzey, eds.), 4th edn., pp. 323–390. McGraw-Hill, Boston. Petty, R. E., Wells, G. L., and Brock, T. C. (1976). Distraction can enhance or reduce yielding to propaganda: Thought disruption versus effort justification. J. Pers. Soc. Psychol. 34, 874–884. Petty, R. E., Cacioppo, J. T., and Schumann, D. W. (1983). Central and peripheral routes to advertising effectiveness: The moderating role of involvement. J. Consum. Res. 10, 135–146. Petty, R. E., Haugtvedt, C., and Smith, S. M. (1995). Elaboration as a determinant of attitude strength: Creating attitudes that are persistent, resistant, and predictive of behavior. In ‘‘Attitude Strength: Antecedents and Consequences’’ (R. E. Petty and J. A. Krosnick, eds.), pp. 93–130. Erlbaum, Hillsdale, NJ. Petty, R. E., Fleming, M. A., and White, P. H. (1999). Stigmatized sources and persuasion: Prejudice as a determinant of argument scrutiny. J. Pers. Soc. Psychol. 76, 19–34. Petty, R. E., Cacioppo, J. T., Strathman, A., and Priester, J. R. (2005). To think or not to think? Exploring two routes to persuasion. In ‘‘Persuasion: Psychological Insights and Perspectives’’ (T. C. Brock and M. C. Green, eds.), 2nd edn., pp. 81–116. Sage Publications, Thousand Oaks, CA. Pratkanis, A. R. (2007). An invitation to social influence research. In ‘‘The Science of Social Influence: Advances and Future Progress’’ (A. R. Pratkanis, ed.), pp. 1–15. Psychology Press, New York.
Majority Versus Minority Influence
325
Prislin, R. (in press). Dynamics of change: Minority influence makes the world go around. In ‘‘Minority Influence and Innovation: Antecedents, Processes and Consequences’’ (R. Martin and M. Hewstone, eds.). Psychology Press, Hove, E. Sussex. Prislin, R., and Christensen, P. N. (2005). Social change in the aftermath of successful minority influence. In ‘‘European Review of Social Psychology’’ (W. Stroebe and M. Hewstone, eds.), Vol. 16, pp. 43–73. Psychology Press (Taylor & Francis), Hove, E. Sussex. Quiamzade, A., Pe´rez, J. A., Mugny, G., and Alonso, R. (2003). Le patron de la conversion minoritaire, vers une articulation des the´ories de la dissociation et de l’auto-cate´gorisation: Les multiples cate´gorisation simultane´es. Int. Rev. Soc. Psychol. 16, 87–123. Quiamzade, A., Mugny, G., Falomir, J. M., and Chatard, A. (2006). De la psychologie sociale de´veloppementale a` l’influence sociale dans les taˆches d’aptitudes: Retour sur l’ontoge´ne`se de la notion de conflit sociocognitif dans l’e´cole genevoise. In ‘‘Bilans et Perspectives en Psychologie Sociale’’ (R. V. Joule and P. Huguet, eds.), Vol. 1, pp. 171–205. Presses universitaires de Grenoble, Grenoble. Quiamzade, A., Mugny, G., Falomir-Pichastor, J. M., and Butera, F. (in press). The complexity of majority and minority influence processes. In ‘‘Minority Influence and Innovation: Antecedents, Processes and Consequences’’ (R. Martin and M. Hewstone, eds.). Psychology Press, Hove, E. Sussex. Romero, A. A., Agnew, C. R., and Insko, C. A. (1996). The cognitive mediation hypothesis revisited: An empirical response to methodological and theoretical criticism. Pers. Soc. Psychol. Bull. 22, 651–665. Ross, L., Greene, D., and House, P. (1977). The ‘‘false consensus effect’’: An egocentric bias in social perception and attribution processes. J. Exp. Soc. Psychol. 13, 279–301. Schachter, S. (1951). Deviation, rejection, and communication. J. Abnorm. Soc. Psychol. 46, 190–207. Schulz-Hardt, S., Frey, D., Lu¨thgens, C., and Moscovici, S. (2000). Biased information search in group decision making. J. Pers. Soc. Psychol. 78, 655–669. Schuurman, M. K., Siero, F. W., De Dreu, C. K. W., and Buunk, A. P. (1995). Differentie¨le verwerking van numerieke steun. [Differential processing of majority and minority arguments]. In ‘‘Fundamentele Sociale Psychologie’’ (N. K. De Vries, N. E. Ellemers, R. Vonk, and C. K. W. De Dreu, eds.), Vol. 9, pp. 1–9. Tilburg University Press, Tilburg. Shuper, P. A., and Sorrentino, R. M. (2004). Minority versus majority influence and uncertainty orientation: Processing persuasive messages on the basis of situational expectancies. J. Soc. Psychol. 144, 127–147. Smith C., and Tindale S. (in press). Direct and indirect minority influence in groups. In ‘‘Minority Influence and Innovation: Antecedents, Processes and Consequences’’ (R. Martin and M. Hewstone, eds.). Psychology Press, Hove, E. Sussex. Snyder, C. R., and Fromkin, H. L. (1980). ‘‘Uniqueness, the Human Pursuit of Difference.’’ Plenum Press, New York. Stroebe W. (in press). Minority influence and information processing: A theoretical and methodological analysis. In ‘‘Minority Influence and Innovation: Antecedents, Processes and Consequences’’ (R. Martin and M. Hewstone, eds.). Psychology Press, Hove, E. Sussex. Stu¨rmer, S., and Simon, B. (2004). Collective action: Towards a dual-pathway model. In ‘‘European Review of Social Psychology’’ (W. Stroebe and M. Hewstone, eds.), Vol. 15, pp. 59–99. Psychology Press (Taylor & Francis), Hove, E. Sussex. Tafani, E., Souchet, L., Codaccioni, C., and Mugny, G. (2003). Influences majoritaire et minoritaire sur la repre´sentation sociale de la drogue. Nouvelle Revue de Psychologie Sociale 2(3), 343–354. Talbot, M. (2005). Supreme confidence: The jurisprudence of Justice Antonin Scalia. The New Yorker pp. 38–55. March 28, 2005.
326
Robin Martin and Miles Hewstone
Tanford, S., and Penrod, S. (1984). Social influence model: A formal integration of research on majority and minority influence processes. Psychol. Bull. 95, 189–225. Tesser, A., Campbell, J., and Mickler, S. (1983). The role of social pressure, attention to the stimulus, and self-doubt in conformity. Eur. J. Soc. Psychol. 13, 217–233. Thompson, M., Naccarato, M., Parker, K., and Moskowitz, G. (1993). The development and validation of the Personal Need for Structure (PNS) and the Personal Fear of Invalidity (PFI) measures. Unpublished manuscript, University of Waterloo. Tormala, Z. L., De Sensi, V. L., and Petty, R. E. (2007). Resisting persuasion by illegitimate means: A meta-cognitive perspective on minority influence. Pers. Soc. Psychol. Bull. 33, 354–367. Tormala, Z. L., Petty, R. E., and De Sensi, V. L. (in press). Multiple roles for minority sources in persuasion and resistance. In ‘‘Minority Influence and Innovation: Antecedents, Processes and Consequences’’ (R. Martin and M. Hewstone, eds.). Psychology Press, Hove, E. Sussex. Triandis, H. C., and Trafimov, D. (2001). Culture and its implications for intergroup behavior. In ‘‘Blackwell Handbook of Social Psychology: Intergroup Processes’’ (R. Brown and S. L. Gaertner, eds.), pp. 367–385. Blackwell, Oxford. Trost, M. R., Maass, A., and Kenrick, D. T. (1992). Minority influence: Personal relevance biases cognitive processes and reverses private acceptance. J. Exp. Soc. Psychol. 28, 234–254. Turner, J. C. (1991). ‘‘Social Influence.’’ Open University Press, Milton Keynes. Visser, P. S., and Cooper, J. (2003). Attitude change. In ‘‘Sage Handbook of Social Psychology’’ (M. A. Hogg and J. Cooper, eds.), pp. 211–231. Sage, London. Visser, P. S., and Krosnick, J. A. (1998). Development of attitude strength over the life cycle: Surge and decline. J. Pers. Soc. Psychol. 75, 1389–1410. Vocci, A., Hewstone, M., Crisp, R. J., and Rubin, M. (in press). ‘‘Majority, minority, and parity: Effects of gender and group size on perceived group variability.’’ Soc. Psychol. Q. Volpato, C., Maass, A., Mucchi-Faina, A., and Vitti, E. (1990). Minority influence and social categorization. Eur. J. Soc. Psychol. 20, 119–132. Webster, D. M., and Kruglanski, A. W. (1994). Individual differences in need for cognitive closure. J. Pers. Soc. Psychol. 67, 1049–1062. White, M. (2007). ‘‘Galileo antichrist: A biography.’’ Orion Books, London. White, P. H., and Harkins, S. G. (1994). Race of source effects in the Elaboration Likelihood Model. J. Pers. Soc. Psychol. 67, 790–807. Wood, W., Lundgren, S., Ouellette, J. A., Busceme, S., and Blackstone, T. (1994). Minority influence: A meta-analytic review of social influence processes. Psychol. Bull. 115, 323–345. Wood, W., Pool, G. J., Leck, K., and Purvis, D. (1996). Self-definition, defensive processing, and influence: The normative impact of majority and minority groups. J. Pers. Soc. Psychol. 71, 1181–1193. Wu, C., and Shaffer, D. R. (1987). Susceptibility to persuasive appeals as a function of source credibility and prior experience with the attitude object. J. Pers. Soc. Psychol. 52, 677–688. Zakaria, F. (2004). ‘‘The Future of Freedom: Illiberal Democracy at Home and Abroad.’’ W. W. Norton & Company, New York, NY. Zdaniuk, B., and Levine, J. M. (1996). Anticipated interaction and thought generation: The role of faction size. Br. J. Soc. Psychol. 35, 201–218.
CONTENTS OF OTHER VOLUMES
Volume 1
From Acts to Dispositions: The Attribution Process in Person Perception Edward E. Jones and Keith E. Davis Inequality in Social Exchange J. Stacy Adams The Concept of Aggressive Drive: Some Additional Considerations Leonard Berkowitz Author Index—Subject Index
Cultural Influences upon Cognitive Processes Harry C. Triendis The Interaction of Cognitive and Physiological Determinants of Emotional State Stanley Schachter Experimental Studies of Coalirion Formation William A. Gamson Communication Networks Marvin E. Shaw A Contingency Model of Leadership Effectiveness Fred E. Fiedler Inducing Resistance to Persuasion: Some Contemporary Approaches William J. McGuire Social Motivation, Dependency, and Susceptibility to Social Influence Richard H. Walters and Ross D. Purke Sociability and Social Organization in Monkeys and Apes William A. Mason Author Index—Subject Index
Volume 3 Mathematical Models in Social Psychology Robert P. Abelson The Experimental Analysis of Social Performance Michael Argyle and Adam Kendon A Structural Balance Approach to the Analysis of Communication Effects N. T. Feather Effects of Fear Arousal on Attitude Change: Recent Developments in Theory and Experimental Research Irving L. Janis Communication Processes and the Properties of Language Serge Moscovici The Congruity Principle Revisited: Studies in the Reduction, Induction, and Generalization of Persuasion Percy H. Tannenbaum Author Index—Subject Index
Volume 2 Vicarious Processes: A Case of No-Trial Learning Albert Bandura Selective Exposure Jonathan L. Freedman and David O. Sears Group Problem Solving L. Richard Hoffman Situational Factors in Conformity Vernon L. Allen Social Power John Schopler
Volume 4 The Theory of Cognitive Dissonance: A Current Perspective Elliot Aronson
333
334
CONTENTS OF OTHER VOLUMES
Attitudes and Attraction Donn Byrne Sociolinguistics Susan M. Ervin-Tripp Recognition of Emotion Nico H. Frijda Studies of Status Congruence Edward E. Sampson Exploratory Investigations of Empathy Ezra Stotland The Personal Reference Scale: An Approach to Social Judgment Harry S. Upshaw Author Index—Subject Index Volume 5 Media Violence and Aggressive Behavior: A Review of Experimental Research Richard E. Goranson Studies in Leader Legitimacy, Influence, and Innovation Edwin P. Hollander and James W. Julian Experimental Studies of Negro-White Relationships Irwin Katz Findings and Theory in the Study of Fear Communications Howard Leventhal Perceived Freedom Ivan D. Steiner Experimental Studies of Families Nancy E. Waxler and Elliot G. Mishler Why Do Groups Make Riskier Decisions than Individuals? Kenneth L. Dion, Robert S. Baron, and Norman Miller Author Index—Subject Index Volume 6 Self-Perception Theory Daryl J. Bem Social Norms, Feelings, and Other Factors Affecting Helping and Altruism Leonard Berkowitz The Power of Liking: Consequence of Interpersonal Attitudes Derived from a Liberalized View of Secondary Reinforcement Albert J. Lott and Bernice E. Lott
Social Influence, Conformity Bias, and the Study of Active Minorities Serge Moscovici and Claude Faucheux A Critical Analysis of Research Utilizing the Prisoner’s Dilemma Paradigm for the Study of Bargaining Charlan Nemeth Structural Representations of Implicit Personality Theory Seymour Rosenberg and Andrea Sedlak Author Index—Subject Index Volume 7 Cognitive Algebra: Integration Theory Applied to Social Attribution Norman A. Anderson On Conflicts and Bargaining Erika Apfelbaum Physical Attractiveness Ellen Bersheid and Elaine Walster Compliance, Justification, and Cognitive Change Harold B. Gerard, Edward S. Connolley, and Roland A. Wilhelmy Processes in Delay of Gratification Walter Mischel Helping a Distressed Person: Social, Personality, and Stimulus Determinants Ervin Staub Author Index—Subject Index Volume 8 Social Support for Nonconformity Vernon L. Allen Group Tasks, Group Interaction Process, and Group Performance Effectiveness: A Review and Proposed Integration J. Richard Hackman and Charles G. Morris The Human Subject in the Psychology Experiment: Fact and Artifact Arie W. Kruglanski Emotional Arousal in the Facilitation of Aggression Through Communication Percy H. Tannenbaum and Dolf Zillman The Reluctance to Transmit Bad News Abraham Tesser and Sidney Rosen Objective Self-Awareness Robert A. Wicklund
CONTENTS OF OTHER VOLUMES
Responses to Uncontrollable Outcomes: An Integration of Reactance Theory and the Learned Helplessness Model Camille B. Wortman and Jack W. Brehm Subject Index Volume 9 New Directions in Equity Research Elaine Walster, Ellen Berscheid, and G. William Walster Equity Theory Revisited: Comments and Annotated Bibliography J. Stacy Adams and Sara Freedman The Distribution of Rewards and Resources in Groups and Organizations Gerald S. Leventhal Deserving and the Emergence of Forms of Justice Melvin J. Lerner, Dale T. Miller, and John G. Holmes Equity and the Law: The Effect of a Harmdoer’s ‘‘Suffering in the Act’’ on Liking and Assigned Punishment William Austin, Elaine Walster, and Mary Kristine Utne Incremental Exchange Theory: A Formal Model for Progression in Dyadic Social Interaction L. Lowell Huesmann and George Levinger Commentary George C. Homans Subject Index Volume 10 The Catharsis of Aggression: An Evaluation of a Hypothesis Russell G. Geen and Michael B. Quanty Mere Exposure Albert A. Harrison Moral Internalization: Current Theory and Research Martin L. Hoffman Some Effects of Violent and Nonviolent Movies on the Behavior of Juvenile Delinquents Ross D. Parke, Leonard Berkowitz, Jacques P. Leyens, Stephen G. West, and Richard Sebastian
335
The Intuitive Psychologist and His Shortcomings: Distortions in the Attribution Process Less Ross Normative Influences on Altruism Shalom H. Schwartz A Discussion of the Domain and Methods of Social Psychology: Two Papers by Ron Harre and Barry R. Schlenker Leonard Berkowitz The Ethogenic Approach: Theory and Practice R. Harre On the Ethogenic Approach: Etiquette and Revolution Barry R. Schlenker Automatisms and Autonomies: In Reply to Professor Schlenker R. Harre Subject Index Volume 11 The Persistence of Experimentally Induced Attitude Change Thomas D. Cook and Brian F. Flay The Contingency Model and the Dynamics of the Leadership Process Fred E. Fiedler An Attributional Theory of Choice Andy Kukla Group-Induced Polarization of Attitudes and Behavior Helmut Lamm and David G. Myers Crowding: Determinants and Effects Janet E. Stockdale Salience: Attention, and Attribution: Top of the Head Phenomena Shelley E. Taylor and Susan T. Fiske Self-Generated Attitude Change Abraham Tesser Subject Index Volume 12 Part I. Studies in Social Cognition Prototypes in Person Perception Nancy Cantor and Walter Mischel A Cognitive-Attributional Analysis of Stereotyping David L. Hamilton
336
CONTENTS OF OTHER VOLUMES
Self-Monitoring Processes Mark Snyder Part II. Social Influences and Social Interaction Architectural Mediation of Residential Density and Control: Crowding and the Regulation of Social Contact Andrew Baum and Stuart Valins A Cultural Ecology of Social Behavior J. W. Berry Experiments on Deviance with Special Reference to Dishonesty David P. Farrington From the Early Window to the Late Night Show: International Trends in the Study of Television’s Impact on Children and Adults John P. Murray and Susan Kippax Effects of Prosocial Television and Film Material on the Behavior of Viewers J. Phillipe Rushton Subject Index
Cognitive, Social, and Personality Processes in the Physiological Detection of Deception William M. Waid and Martin T. Orne Dialectic Conceptions in Social Psychology: An Application to Social Penetration and Privacy Regulation Irwin Altman, Anne Vinsel, and Barbara B. Brown Direct Experience and Attitude–Behavior Consistency Russell H. Fazio and Mark P. Zanna Predictability and Human Stress: Toward a Clarification of Evidence and Theory Suzanne M. Miller Perceptual and Judgmental Processes in Social Contexts Arnold Upmeyer Jury Trials: Psychology and Law Charlan Jeanne Nemeth Index Volume 15
Volume 13 People’s Analyses of the Causes of AbilityLinked Performances John M. Darley and George R. Goethals The Empirical Exploration of Intrinsic Motivational Processes Edward I. Deci and Richard M. Ryan Attribution of Responsibility: From Man the Scientist to Man as Lawyer Frank D. Fincham and Joseph M. Jaspars Toward a Comprehensive Theory of Emotion Howard Leventhal Toward a Theory of Conversion Behavior Serge Moscovici The Role of Information Retrieval and Conditional Inference Processes in Belief Formation and Change Robert S. Wyer, Jr. and Jon Hartwick Index
Balance, Agreement, and Positivity in the Cognition of Small Social Structures Walter H. Crockett Episode Cognition: Internal Representations of Interaction Routines Joseph P. Forgas The Effects of Aggressive-Pornographic Mass Media Stimuli Neil M. Malamuth and Ed Donnerstein Socialization in Small Groups: Temporal Changes in Individual–Group Relations Richard L. Moreland and John M. Levine Translating Actions into Attitudes: An Identity-Analytic Approach to the Explanation of Social Conduct Barry R. Schlenker Aversive Conditions as Stimuli to Aggression Leonard Berkowitz Index
Volume 14
Volume 16
Verbal and Nonverbal Communication of Deception Miron Zuckerman, Bella M. DePaulo, and Robert Rosenthal
A Contextualist Theory of Knowledge: Its Implications for Innovation and Reform in Psychological Research William J. McGuire
CONTENTS OF OTHER VOLUMES
Social Cognition: Some Historical and Theoretical Perspectives Janet Landman and Melvin Manis Paradigmatic Behaviorism: Unified Theory for Social-Personality Psychology Arthur W. Staats SocialPsychologyfromthe Standpointofa StructuralSymbolicInteractionism:Toward anInterdisciplinarySocial Psychology Sheldon Stryker Toward an Interdisciplinary Social Psychology Carl W. Backman Index Volume 17 Mental Representations of Self John F. Kihlstrom and Nancy Cantor Theory of the Self: Impasse and Evolution Kenneth J. Gergen A Perceptual-Motor Theory of Emotion Howard Leventhal Equity and Social Change in Human Relationships Charles G. McClintock, Roderick M. Kramer, and Linda J. Keil A New Look at Dissonance Theory Joel Cooper and Russell H. Fazio Cognitive Theories of Persuasion Alice H. Eagly and Shelly Chaiken Helping Behavior and Altruism: An Empirical and Conceptual Overview John F. Dovidio Index Volume 18 A Typological Approach to Marital Interaction: Recent Theory and Research Mary Anne Fitzpatrick Groups in Exotic Environments Albert A. Harrison and Mary M. Connors Balance Theory, the Jordan Paradigm, and the Wiest Tetrahedon Chester A. Insko The Social Relations Model David A. Kenny and Lawrence La Voie Coalition Bargaining S. S. Komorita
337
When Belief Creates Reality Mark Snyder Index Volume 19 Distraction–Conflict Theory: Progress and Problems Robert S. Baron Recent Research on Selective Exposure to Information Dieter Frey The Role of Threat to Self-Esteem and Perceived Control in Recipient Reaction to Help: Theory Development and Empirical Validation Arie Nadler and Jeffrey D. Fisher The Elaboration Likelihood Model of Persuasion Richard E. Petty and John T. Cacioppo Natural Experiments on the Effects of Mass Media Violence on Fatal Aggression: Strengths and Weaknesses of a New Approach David P. Phillips Paradigms and Groups Ivan D. Steiner Social Categorization: Implications for Creation and Reduction of Intergroup Bias David A. Wilder Index Volume 20 Attitudes, Traits, and Actions: Dispositional Prediction of Behavior in Personality and Social Psychology Icek Ajzen Prosocial Motivation: Is It Ever Truly Altruistic? C. Daniel Batson Dimensions of Group Process: Amount and Structure of Vocal Interaction James M. Dabbs, Jr. and R. Barry Ruback The Dynamics of Opinion Formation Harold B. Gerard and Ruben Orive Positive Affect, Cognitive Processes, and Social Behavior Alice M. Isen
338
CONTENTS OF OTHER VOLUMES
Between Hope and Fear: The Psychology of Risk Lola L. Lopes Toward an Integration of Cognitive and Motivational Perspectives on Social Inference: A Biased Hypothesis-Testing Model Tom Pyszczynski and Jeff Greenberg Index Volume 21 Introduction Leonard Berkowitz Part I. The Self as Known Narrative and the Self as Relationship Kenneth J. Gergen and Mary M. Gergen Self and Others: Studies in Social Personality and Autobiography Seymour Rosenberg Content and Process in the Experience of Self William J. McGuire and Claire V. McGuire Information Processing and the Study of the Self John F. Kihlstrom, Nancy Cantor, Jeanne Sumi Albright, Beverly R. Chew, Stanley B. Klein, and Paula M. Niedenthal Part II. Self-Motives Toward a Self-Evaluation Maintenance Model of Social Behavior Abraham Tesser The Self: A Dialectical Approach Carl W. Backman The Psychology of Self-Affirmation: Sustaining the Integrity of the Self Claude M. Steele A Model of Behavioral Self-Regulation: Translating Intention into Action Michael F. Scheier and Charles S. Carver Index Volume 22 On the Construction of the Anger Experience: Aversive Events and Negative Priming in the Formation of Feelings Leonard Berkowitz and Karen Heimer Social Psychophysiology: A New Look John T. Cacioppo, Richard E. Petty, and Louis G. Tassinary
Self-Discrepancy Theory: What Patterns of Self-Beliefs Cause People to Suffer? E. Tory Higgins Minding Matters: The Consequences of Mindlessness-Mindfulness Ellen J. Langer The Tradeoffs of Social Control and Innovation in Groups and Organizations Charlan Jeanne Nemeth and Barry M. Staw Confession, Inhibition, and Disease James W. Pennebaker A Sociocognitive Model of Attitude Structure and Function Anthony R. Pratkanis and Anthony G. Greenwald Introspection, Attitude Change, and Attitude–Behavior Consistency: The Disruptive Effects of Explaining Why We Feel the Way We Do Timothy D. Wilson, Dana S. Dunn, Dolores Kraft, and Douglas J. Lisle Index Volume 23 A Continuum of Impression Formation, from Category-Based to Individuating Processes: Influences of Information and Motivation on Attention and Interpretation Susan T. Fiske and Steven L. Neuberg Multiple Processes by Which Attitudes Guide Behavior: The MODE Model as an Integrative Framework Russell H. Fazio PEAT: An Integrative Model of Attribution Processes John W. Medcof Reading People’s Minds: A Transformation Rule Model for Predicting Others’ Thoughts and Feelings Rachel Karniol Self-Attention and Behavior: A Review and Theoretical Update Frederick X. Gibbons Counterfactual Thinking and Social Perception: Thinking about What Might Have Been Dale T. Miller, William Turnbull, and Cathy McFarland Index
CONTENTS OF OTHER VOLUMES
Volume 24 The Role of Self-Interest in Social and Political Attitudes David O. Sears and Carolyn L. Funk A Terror Management Theory of Social Behavior: The Psychological Functions of Self-Esteem and Cultural Worldviews Sheldon Solomon, Jeff Greenberg, and Tom Pyszczynski Mood and Persuasion: Affective States Influence the Processing of Persuasive Communications Norbert Schwarz, Herbert Bless, and Gerd Bohner A Focus Theory of Normative Conduct: A Theoretical Refinement and Reevaluation of the Role of Norms in Human Behavior Robert B. Cialdini, Carl A. Kallgren, and Raymond R. Reno The Effects of Interaction Goals on Person Perception James L. Hilton and John M. Darley Studying Social Interaction with the Rochester Interaction Record Harry T. Reis and Ladd Wheeler Subjective Construal, Social Inference, and Human Misunderstanding Dale W. Griffin and Lee Ross Index Volume 25 Universals in the Content and Structure of Values: Theoretical Advances and Empirical Tests in 20 Countries Shalom H. Schwartz Motivational Foundations of Behavioral Confirmation Mark Snyder A Relational Model of Authority in Groups Tom R. Tyler and E. Allan Lind You Can’t Always Think What You Want: Problems in the Suppression of Unwanted Thoughts Daniel M. Wegner Affect in Social Judgments and Decisions: A Multiprocess Model Joseph Paul Forgas The Social Psychology of Stanley Milgram Thomas Blass
339
The Impact of Accountability on Judgment and Choice: Toward a Social Contingency Model Philip E. Tetlock Index Volume 26 Attitudes Toward High Achievers and Reactions to Their Fall: Theory and Research Concerning Tall Poppies N. T. Feather Evolutionary Social Psychology: From Sexual Selection to Social Cognition Douglas T. Kenrick Judgment in a Social Context: Biases, Short-comings, and the Logic of Conversation Norbert Schwarz A Phase Model of Transitions: Cognitive and Motivational Consequences Diane N. Ruble Multiple-Audience Problems, Tactical Communication, and Social Interaction: A Relational-Regulation Perspective John H. Fleming From Social Inequality to Personal Entitlement: The Role of Social Comparisons, Legitimacy Appraisals, and Group Membership Brenda Major Mental Representations of Social Groups: Advances in Understanding Stereotypes and Stereotyping Charles Stangor and James E. Lange Index Volume 27 Inferences of Responsibility and Social Motivation Bernard Weiner Information Processing in Social Contexts: Implications for Social Memory and Judgment Robert S. Wyer, Jr. and Deborah H. Gruenfeld The Interactive Roles of Stability and Level of Self-Esteem: Research and Theory Michael H. Kernis and Stephanie B. Waschull
340
CONTENTS OF OTHER VOLUMES
Gender Differences in Perceiving Internal State: Toward a His-and-Hers Model of Perceptual Cue Use Tomi-Ann Roberts and James W. Pennebaker On the Role of Encoding Processes in Stereotype Maintenance William von Hippel, Denise Sekaquaptewa, and Patrick Vargas Psychological Barriers to Dispute Resolution Lee Ross and Andrew Ward Index Volume 28 The Biopsychosocial Model of Arousal Regulation Jim Blascovich and Joe Tomaka Outcome Biases in Social Perception: Implications for Dispositional Inference, Attitude Change, Stereotyping, and Social Behavior Scott T. Allison, Diane M. Mackie, and David M. Messick Principles of Judging Valence: What Makes Events Positive or Negative? C. Miguel Brendl and E. Tory Higgins Pluralistic Ignorance and the Perpetuation of Social Norms by Unwitting Actors Deborah A. Prentice and Dale T. Miller People as Flexible Interpreters: Evidence and Issues from Spontaneous Trait Inference James S. Uleman, Leonard S. Newman, and Gordon B. Moskowitz Social Perception, Social Stereotypes, and Teacher Expectations: Accuracy and the Quest for the Powerful Self-Fulfilling Prophecy Lee Jussim Jacquelynne Eccles, and Stephanie Madon Nonverbal Behavior and Nonverbal Communication: What do Conversational Hand Gestures Tell Us? Robert M. Krauss, Yihsiu Chen, and Purnima Chawla Index Volume 29 Counterfactual Thinking: The Intersection of Affect and Function Neal J. Roese and James M. Olson
Terror Management Theory of Self-Esteem and Cultural Worldviews: Empirical Assessments and Conceptual Refinements Jeff Greenberg, Sheldon Solomon, and Tom Pyszczynski The Flexible Correction Model: The Role of Naı¨ve Theories of Bias in Bias Correction Duane T. Wegener and Richard E. Petty Self-Evaluation: To Thine Own Self Be Good, to Thine Own Self Be Sure, to Thine Own Self Be True, and to Thine Own Self Be Better Constantine Sedikides and Michael J. Strube Toward a Hierarchical Model of Intrinsic and Extrinsic Motivation Robert J. Vallerand Volume 30 Promotion and Prevention: Regulatory Focus as a Motivational Principle E. Tory Higgins The Other ‘‘Authoritarian Personality’’ Bob Altemeyer Person Preception Comes of Age: The Salience and Significance of Age in Social Adjustments Joann M. Montepare and Leslie A. Zebrowitz On the Perception of Social Consensus Joachim Krueger Prejudice and Stereotyping in Everyday Communication Janet B. Ruscher Situated Optimism: Specific Outcome Expectancies and Self-Regulation David A. Armor and Shelley E. Taylor Volume 31 Affect and Information Processing Robert S. Wyer, Jr., Gerald L. Clore, and Linda M. Isbell Linguistic Intergroup Bias: Stereotype Perpetuation through Language Anne Maass Relationships from the Past in the Present: Significant-Other Representations and Transference in Interpersonal Life Serena Chen and Susan M. Anderson
CONTENTS OF OTHER VOLUMES
341
The Puzzle of Continuing Group Inequality: Piecing Together Psychological, Social, and Cultural Forces in Social Dominance Theory Felicia Pratto Attitude Representation Theory Charles G. Lord and Mark R. Lepper Discontinuity Theory: Cognitive and Social Searches for Rationality and Normality— May Lead to Madness Philip G. Zimbardo
G. Daniel Lassiter, Andrew L. Geers, Patrick J. Munhall, Ian M. Handley, and Melissa J. Beers Effort Determination of Cardiovascular Response: An Integrative Analysis with Applications in Social Psychology Rex A. Wright and Leslie D. Kirby Index
Volume 32
Uncertainty Management by Means of Fairness Judgments Kees van den Bos and E. Allan Lind Cognition in Persuasion: An Analysis of Information Processing in Response to Persuasive Communications Dolores Albarracin Narrative-Based Representations of Social Knowledge: Their Construction and Use in Comprehension, Memory, and Judgment Robert S. Wyer, Jr., Rashmi Adaval and Stanley J. Colcombe Reflexion and Reflection: A Social Cognitive Neuroscience Approach to Attributional Inference Matthew D. Lieberman, Ruth Gaunt, Daniel T. Gilbert, and Yaacov Trope Antecedents and Consequences of Attributions to Discrimination: Theoretical and Empirical Advances Brenda Major, Wendy J. Quinton, and Shannon K. McCoy A Theory of Goal Systems Arie W. Kruglanski, James Y. Shah, Ayelet Fishbach, Ron Friedman, Woo Young Chun, and David Sleeth-Keppler Contending with Group Image: The Psychology of Stereotype and Social Identity Threat Claude M. Steele, Steven J. Spencer, and Joshua Aronson Index
The Nature and Function of Self-Esteem: Sociometer Theory Mark R. Leary and Roy F. Baumeister Temperature and Aggression Craig A. Anderson, Kathryn B. Anderson, Nancy Dorr, Kristina M. DeNeve, and Mindy Flanagan The Importance of Being Selective: Weighing the Role of Attribute Importance in Attitudinal Judgment Joop van der Pligt, Nanne K. de Vries, Antony S. R. Manstead, and Frank van Harreveld Toward a History of Social Behavior: Judgmental Accuracy from Thin Slices of the Behavioral Stream Nalini Amabady, Frank J. Bernieri, and Jennifer A. Richeson Attractiveness, Attraction, and Sexual Selection: Evolutionary Perspectives on the Form and Function of Physical Attractiveness Dianne S. Berry Index Volume 33 The Perception–Behavior Expressway: Automatic Effects of Social Perception on Social Behavior Ap Dijksterhuis and John A. Bargh A Dual-Process Cognitive-Motivational Theory of Ideology and Prejudice John Duckitt Ambivalent Sexism Peter Glick and Susan T. Fiske Videotaped Confessions: Is Guilt in the Eye of the Camera?
Volume 34
Volume 35 Social Identity and Leadership Processes in Groups Michael A. Hogg and Daan van Knippenberg
342
CONTENTS OF OTHER VOLUMES
The Attachment Behavioral System in Adulthood: Activation, Psychodynamics, and Interpersonal Processes Mario Mikulincer and Phillip R. Shaver Stereotypes and Behavioral Confirmation: From Interpersonal to Intergroup Perspectives Olivier Klein and Mark Snyder Motivational Bases of Information Processing and Strategy in Conflict and Negotiation Carsten K. W. De Dreu and Peter J. Carnevale Regulatory Mode: Locomotion and Assessment as Distinct Orientations E. Tory Higgins, Arie W. Kruglanski, and Antonio Pierro Affective Forecasting Timothy D. Wilson and Daniel T. Gilbert Index
Over Thirty Years Later: A Contemporary Look at Symbolic Racism David O. Sears and P. J. Henry Managing Group Behavior: The Interplay Between Procedural Justice, Sense of Self, and Cooperation David De Cremer and Tom R. Tyler So Right it’s Wrong: Groupthink and the Ubiquitous Nature of Polarized Group Decision Making Robert S. Baron An Integrative Theory of Intergroup Contact Rupert Brown and Miles Hewstone Says Who?: Epistemic Authority Effects in Social Judgment Arie W. Kruglanski, Amiram Raviv, Daniel Bar-Tal, Alona Raviv, Keren Sharvit, Shmuel Ellis, Ruth Bar, Antonio Pierro, and Lucia Mannetti Index
Volume 36
Volume 38
Aversive Racism John F. Dovidio and Samuel L. Gaertner Socially Situated Cognition: Cognition in its Social Context Eliot R. Smith and Gu¨n R. Semin Social Axioms: A Model for Social Beliefs in Multicultural Perspective Kwok Leung and Michael Harris Bond Violent Video Games: Specific Effects of Violent Content on Aggressive Thoughts and Behavior Craig A. Anderson, Nicholas L. Carnagey, Mindy Flanagan, Arlin J. Benjamin, Jr., Janie Eubanks, and Jeffery C. Valentine Survival and Change in Judgments: A Model of Activation and Comparison Dolores Albarracı´n, Harry M. Wallace, and Laura R. Glasman The Implicit Volition Model: On the Preconscious Regulation of Temporarily Adopted Goals Gordon B. Moskowitz, Peizhong Li, and Elizabeth R. Kirk Index
Exploring the Latent Structure of Strength-Related Attitude Attributes Penny S. Visser, George Y. Bizer, and Jon A. Krosnick Implementation Intentions and Goal Achievement: A Meta-Analysis of Effects and Processes Peter M. Gollwitzer and Paschal Sheeran Interracial Interactions: A Relational Approach J. Nicole Shelton and Jennifer A. Richeson The Psychology of Self-Defense: Self-Affirmation Theory David K. Sherman and Geoffrey L. Cohen Intergroup Beliefs: Investigations from the Social Side Charles Stangor and Scott P. Leary A Multicomponent Conceptualization of Authenticity: Theory and Research Michael H. Kernis and Brian M. Goldman Index
Volume 37 Accuracy in Social Perception: Criticisms, Controversies, Criteria, Components, and Cognitive Processes Lee Jussim
Volume 39 Culture and the Structure of Personal Experience: Insider and Outsider Phenomenologies of the Self and Social World Dov Cohen, Etsuko Hoshino-Browne, and Angela K.-y. Leung
CONTENTS OF OTHER VOLUMES
Uncertainty–Identity Theory Michael A. Hogg Metacognitive Experiences and the Intricacies of Setting People Straight: Implications for Debiasing and Public Information Campaigns Norbert Schwarz, Lawrence J. Sanna, Ian Skurnik, and Carolyn Yoon Multiple Social Categorization Richard J. Crisp and Miles Hewstone On the Parameters of Human Judgment Arie W. Kruglanski, Antonio Pierro, Lucia Mannetti, Hans-Peter Erb, and Woo Young Chun
343
Panglossian Ideology in the Service of System Justification: How Complementary Stereotypes Help Us to Rationalize Inequality Aaron C. Kay, John T. Jost, Anesu N. Mandisodza, Steven J. Sherman, John V. Petrocelli, and Amy L. Johnson Feeling the Anguish of Others: A Theory of Vicarious Dissonance Joel Cooper and Michael A. Hogg Index