No title

Archives of Sexual Behavior, Vol. 26, No. 4, 1997

A Monoamine Hypothesis for the Pathophysiology of Paraphilic Disorders Martin P. Kafka, M.D.1,2

A monoamine pathophysiological hypothesis for paraphilias in males is based on the following data: (i) the monoamines norepinephrine, dopamine, and serotonin are involved in the appetitive dimension of male sexual behavior in laboratory animals; (ii) data gathered from studying the side effect profiles of antidepressant, psychostimulant, and neuroleptic drugs in humans suggest that alteration of central monoamine neurotransmission can have substantial effects on human sexual functioning, including sexual appetite; (iii) monoamine neurotransmitters appear to modulate dimensions of human and animal psychopathology including impulsivity, anxiety, depression, compulsivity, and pro/antisocial behavior, dimensions disturbed in many paraphiliacs; (iv) pharmacological agents that ameliorate psychiatric disorders characterized by the aforementioned characteristics, especially central serotonin enhancing drugs, can ameliorate paraphilic sexual arousal and behavior. KEY WORDS: paraphilias; monoamines; sexual behavior; serotonin; comorbidity: pharmacotherapy.

PARAPHILIC SEXUALITY: A DIATHESIS MODEL FOR SEXUAL

IMPULSIVITY

Paraphilias are socially deviant, repetitive, highly arousing sexual fantasies, urges, and activities that have a duration of at least 6 months. In addition, paraphilias must be accompanied by clinically significant distress or impairment in social, occupational, or other areas of interpersonal functioning (American Psychiatric Association [APA], 1994, pp. 522-532). 1Harvard

Medical School, Boston, Massachusetts. whom correspondence should be addressed at McLean Hospital, 115 Mill Street, Belmont, Massachusetts 02178.

2To

343

0004-0002/97/0800-0343$12.50/0 c 1997 Plenum Publishing Corporation

344

Kafka

Although more than eight distinct primary paraphilic subtypes are consistently identified by DSM-III, DSM-III-R, and DSM-IV nosology, there is substantial overlap among paraphilic categories as well as between paraphilic sexuality and socially nondeviant forms of sexual impulsivity, e.g., compulsive masturbation, protracted promiscuity, and pornography dependence (Carnes, 1983, 1989; Kafka, 1994a 1994b; Kafka and Prentky, 1992a). Large-sample reports of male paraphilic sex offenders have demonstrated that it is more common for male paraphiliacs to have multiple paraphilias, sometimes simultaneously and sometimes serially (Abel et al, 1988; Abel and Osborn, 1992; Bradford et al, 1992). Pedophiles, for example, commonly acknowledge repetitive acts of exhibitionism, voyeurism, and rape (Abel and Osborn, 1992; Bradford et al, 1992) and only a relatively low proportion of incarcerated pedophiles report that acts against children represent either their primary or exclusive source of deviant sexual arousal (Freund et al, 1991; Gebhard et al, 1965; Langevin, 1983; Mohr et al, 1964). Thus, contemporary data suggest that men with paraphilias have a diathesis for multiple sexual impulsivity disorders (Abel et al, 1988; Abel and Osborn, 1992; Bradford et al, 1992; Kafka, 1995, 1997) characterized by heightened sexual arousal, socially deviant sexual preference, volitional impairment, and periods of temporal stability. In addition, Kafka (1991a, 1991b, 1995, 1997) reported that men with paraphilias frequently report persistently increased sexual appetite as measured by their weekly total sexual outlet, an attribute characterized by other theoreticians as "sexual addiction" (Carnes, 1983, 1989, 1990, 1991), "sexual compulsivity" (Coleman, 1986, 1987; Quadland, 1985), "sexual impulsivity" (APA, 1994; Barth and Kinder, 1987), or "male hypersexuality" (Brotherton, 1974; Davies, 1974; Kafka, 1997; Oxford, 1978). Persistently elevated total sexual outlet is correlated with the presence of multiple paraphilias (Kafka, 1997, 1995; Malamuth et al, 1995), sexual promiscuity (Malamuth et al, 1991), and measures of expressed aggression, sadism, and pornography use (Malamuth et al, 1995). PARAPHILIAS AND PSYCHIATRIC COMORBIDITY

A pathophysiological framework that can encompass the diversity, persistence, and socially deviant proclivities of paraphilic sexuality should help to explain those nonsexual comorbid characteristics that appear to be associated with paraphilic disorders. Many of the extant studies examining comorbid psychiatric disorders in paraphilic sex offenders report symptom clusters derived from personality inventories (e.g., MMPI) and do not identify syndromes as defined by Axis I of DSM-III, DSM-III-R, and DSM-IV

Hypothesis for the Pathophysiology of Paraphilias

345

(e.g., Anderson et al, 1979; Grossman and Cavanaugh, 1990; Kalichman, 1990; Langevin et al, 1990a, 1990b; McCreary, 1975). Although paraphilic sex offenders are a heterogeneous group, the cooccurrence of substance abuse, especially alcohol abuse (Langevin and Langs, 1990; Mio et al, 1986; Rada, 1975), conduct and attention deficit disorder (Hunter and Goodwin, 1992; Kavoussi et al, 1988) and antisocial impulsivity (Prentky and Knight, 1986, 1991; Rosenberg et al., 1988), mixed anxiety-depression (Becker et al, 1991; Fagan et al, 1991; Grossman and Cavanaugh, 1990; Kavoussi et al, 1988; Wise et al, 1991) and social anxiety with concomitant social skill deficits (Baxter et al, 1984; Cohen et al, 1969; Pagan and Wexler, 1986; Knight and Prentky, 1990; Levin and Stava, 1987; Marshall, 1989) have been documented repetitively in adolescent and adult sex offender populations. Pearson (1990) and Kafka and Coleman (1991) first suggested that paraphilias are disorders whose pathophysiology might include a disturbance of central serotonin neurotransmission. In this manuscript, a monoaminergic hypothesis for the pathophysiology of paraphilias is proposed based on the following data: (i) the monoamines norepinephrine, dopamine, and serotonin are involved in both the appetitive and performance related dimensions of male sexual behavior in laboratory mammals; (ii) data gathered from studying the side-effect profiles of antidepressant, psychostimulant, and neuroleptic drugs in humans also suggest that alteration of central monoamine neurotransmission can have substantial effects on human sexual functioning, including sexual appetite; (iii) monoamine neurotransmitters appear to modulate dimensions of human psychopathology and animal behavior including impulsivity, anxiety, depression, compulsivity, and pro/antisocial behavior, dimensions that are disturbed in many paraphiliacs; and (iv) certain pharmacological agents that ameliorate psychiatric disorders characterized by the aforementioned characteristics may ameliorate paraphilic sexual arousal and behavior as well.

MONOAMINES AND SEXUAL BEHAVIOR: ANIMAL DATA The central monoamines are synthesized and released by a relatively small number of neurons whose cell bodies lie in the brain stem but whose projections are extensive throughout the brain. Prior to a discussion of the functional role of monoamines in sexual as well as other behavioral dimensions, it must be emphasized that monoaminergic neuronal pathways interact both with each other as well as with other neurotransmitter and neuromodulatory systems (Hsiao et al, 1987; Linnoila et al, 1988; Meltzer and Lowy, 1987; Siever, 1987; Sulser, 1987; van Praag et al, 1991; Wald-

346

Kafka

meier and Delini-Stula, 1979). While it is beyond the scope of this paper to rigorously review the current concepts of the functional role of the monoamines, neuronal pathways transmitting monoaminergic signals have been conceptualized as neuromodulators mediating attention, learning, physiological functions, affective states, goal-motivated and motor behavior as well as appetitive states such as sleep, sex, thirst, and appetite (Heninger, 1995; Le Moal, 1995; Robbins and Everitt, 1995). Although many studies of psychiatric pathologies report alterations of monoamines in the context of either enhanced or diminished neurotransmission, these are relatively simplistic approaches to the complexities of integrated brain functions. Alternative theoretical models might include constructs such as neurotransmitter dysregulation, alterations in the relative influence of one neurotransmitter in homeostatic balance with others or, as yet undetermined downstream effects of monoamines on nonmonoamine neuromodulators. Thus, a hypothesized role of monoamines for sexual impulsivity disorders does not necessarily imply that monoaminergie dysfunction or dysregulation is synonymous with a simplistic etiological explanation for these complex behavioral disorders. Monoamines can mediate similar as well as distinct sexual behavioral response sets in males and females both within and across species (Gorzalka at al., 1990; Meyerson, 1984; Pfaus and Everitt, 1995). In addition, monoaminergie neuronal pathways can exhibit sexually dimorphic patterns in several brain areas (Siddiqui and Gilmore, 1988; Watts and Stanley, 1984). Inasmuch as paraphilic disorders are almost exclusively disorders affecting human males (APA, 1994), this review emphasizes mammalian studies of masculine sexual behavior. Although it is beyond the scope of this review to speculate as to why paraphilias are almost exclusively human male disorders, it is known that androgenic hormones, especially testosterone and its metabolically active metabolites dihydrotestosterone and estradiol, mediate sexually dimorphic behavior including sexual appetitive and copulatory response repertoire differences between males and females (Everitt and Bancroft, 1991; Feder, 1978; Meyerson, 1984). The efficacy of antiandrogens, cyproterone acetate and medroxyprogesterone acetate to mitigate paraphilic arousal and reduce sexual appetite in males (Bradford, 1995a, 1995b) also lends credibility to the assumption that androgens contribute to the sex differences in the prevalence of paraphilic sexuality. Although there are no animal models for "sexual impulsivity disorders," it has been suggested that the dimension of rodent sexuality most likely warranting comparison with the humans is the dimension of sexual appetite rather than the copulatory response repertoire (Everitt and Bancroft, 1991). Inasmuch as I suggest that paraphilias can be conceptualized

Hypothesis for the Pathophysiology of Paraphilias

347

as disorders that include increased sexual appetite, a comparison with male animal models of sexual appetite is germane. Taken together, the animal data suggest that decreased central (i.e., brain) serotonin (5-hydroxytryptamine [5-HT]) may disinhibit or increase appetitive sexual behavior and, conversely, enhancing central serotonin activity may inhibit or reduce sexual appetitive behavior in some male and female mammalian species (Everitt and Bancroft, 1991; Mas, 1995; Tucker and File, 1983). Increased noradrenergic postsynaptic activity enhances sexual behavior while certain noradrenergic antagonists can have an inhibitory effect on sexual arousal (Clark et al, 1985). Last, decreased central dopaminergic neurotransmission decreases motivated behavior, including male sexual appetite behavior (Everitt, 1983; Pfaus and Everitt, 1995; Segraves, 1989). Conversely, pharmacological enhancement of dopaminergic neurotransmission can augment male sexual behavior, including sexual appetite (Everitt and Bancroft, 1991; Gessa and Tagliamonte, 1975; Mas, 1995; Pfaus and Everitt, 1995; Segraves, 1988). Laboratory studies of rats have demonstrated that the effect of monoamines on sexual behavior can be as profound as the effect of the depletion of "sex" hormones. For example, the depletion of central serotonin in rats, through the administration of parachlorophenylalanine (PCPA), a selective 5-HT biosynthesis inhibitor, appears to increase measures of sexual excitement and mounting behavior and provided a model of "compulsive" sexual behavior in rats (Sheard, 1969; Tagliamonte et al, 1969) and cats (Ferguson et al, 1970). In addition, two serotonin blocking agents, cyproheptadine and methysergide, both facilitate sexual activity in male rats (Menendez-Abraham et al, 1988). Sexual behavior in castrated male rats can be restored with a very low dose of testosterone in combination with a pharmacological agent that reduces the availability of central serotonin while a low dose of testosterone alone failed to restore sexual activity (Haensel et al, 1993; Sodersten et al, 1976). In contrast, treatment of female rhesus monkeys with clomipramine, a potent enhancer of serotonin neurotransmission, markedly decreased sexual receptivity in the presence of male monkeys. These changes in sexual behavior were correlated with increased cerebrospinal fluid levels of the metabolites of serotonin (Everitt, 1980). Yohimbine and idozoxan, a2-receptor antagonists that enhance noradrenergic neurotransmission, increase mounting and intromission and decrease ejaculation latency in sexually naive male rats (Clark et al, 1984, 1985) while receptor agonists have an opposing effect (Clark et al, 1985). Central dopamine D2-receptor blockade, utilizing traditional neuroleptic drugs, can abolish all male rat sexual behavior, including sexual appetite, mounting, intromission, and ejaculation (Baum and Starr, 1980;

348

Kafka

Malmnas, 1973). Conversely, dopaminergic agonists such as apomorphine and L-DOPA enhance male rat copulatory behavior in gonadally intact (Paglietti et al., 1978) as well as castrated male rats maintained on low dose testosterone (Malmnas, 1973) and d-amphetamine released into the nucleus accumbents (ventral striatum) increased male rat instrumental responding to an estrus female (Everitt, 1990). The sex hormones estradiol, testosterone, and progesterone, can induce alterations in the binding of monoamine neurotransmitters in the limbic system, suggesting that one of the cellular mechanisms of hormone action may be at the level of monoamine receptors (Everitt, 1983). In male rats, for example, testosterone may affect changes in central serotonin concentrations that accompany sexual maturity (Kendall and Tonge, 1976) and induce alterations in the binding of norepinephrine, dopamine, and serotonin in the hypothalamus (Everitt et al, 1983). Testosterone effects on male rat sexual response are also mediated by its primary active metabolites, estradiol and 5-a-dihydrotestosterone. In some reports, these metabolites augment mesolimbic dopaminergic and inhibit hypothalamic serotonergic neurotransmission (Alderson and Baum, 1981; Baum and Starr, 1980; Bitran and Hull, 1987; Sodersten et al, 1976), effects that synergistically enhance male sexual appetite and copulatory response. It is most likely that hormones and monoamine neurotransmitters interact in a dynamic fashion that determine the form and intensity of drive behaviors, including sexual behavior (Sicuteri, 1974). The exact nature of the interrelationship of these hormone-monoamine systems in humans may hold clues to further elucidating the biological basis for sexual behaviors, including human sexual impulsivity disorders.

MONOAMINES AND SEXUAL BEHAVIOR: HUMAN DATA The neuromodulation of sexual desire in men and women is poorly understood in comparison to laboratory mammals. The elucidation of the biological mechanisms that determine human sexuality are hampered by limitations in the selectivity of biological probes, undesirable side effects of selective biological agents, as well as the absence of a noninvasive methodology to identify, localize, and selectively affect the brain areas most responsible for the substrates of sexual appetite and copulatory responses (Everitt and Bancroft, 1991). For example, there is less direct evidence that diminished central serotonin may increase sexual desire and performance in the human species (Everitt and Bancroft, 1991; Segraves, 1989). The use of PCPA in human males as a probe for sexual behavior mediated by central serotonin has produced negative results in comparison with experi-

Hypothesis for the Pathophysiology of Paraphilias

349

ments with laboratory mammals (Benkert et al, 1976; Cremata and Koe, 1966). In addition, I could not find studies that specifically measure monoaminergic metabolites or utilize biological probes to provoke monaminergic neurotransmission in male paraphiliacs. Such studies are clearly needed to support or revoke the hypothesis proposed in this manuscript. Despite these limitations, data supporting this hypothesis are presented below. The serotonin-reuptake inhibitors, clomipramine and fluoxetine, increase postsynaptic serotonergic effects and have been reported to produce a high frequency of human sexual dysfunction side effects including the cluster of loss of sexual desire and impaired copulatory response in males (ejaculatory delay, erectile dysfunction, and anorgasmia) (Jacobsen, 1992). These reports support the observation that enhanced central serotonin neurotransmission reduces or inhibits sexual desire and associated sexual performance behaviors. Yohimbine, a a2-receptor antagonist has been shown to enhance erectile function in heterogeneous samples of males with erectile disorders (Morales et al., 1987; Sonda et al, 1990). It is not clear, however, that yohimbine has a direct effect on sexual appetite. In humans, D2-receptor blockade utilizing traditional neuroleptic drugs diminishes sexual appetite and has been used to reduce paraphilic arousal (Tennent et al., 1974; Zbytovsky, 1993). Conversely, increased sexual desire, as measured by self-report of fantasies, erections, and activities, has been reported in men treated with dopamine agonists such as L-DOPA (Bowers et al, 1971; O'Brien et al, 1971) and amphetamine (Angrist and Gershon, 1976; Bell and Trethowan, 1961).

MONOAMINES AND COMORBID SYNDROMES ACCOMPANYING PARAPHILIAS IN HUMANS The neurobiological substrates for impulsivity, compulsivity, anxiety, and depression are better elucidated in humans in comparison with sexual behavior. Inasmuch as some paraphilias can include acts of sexual aggression and are, by current nosological definition, socially deviant disorders of impulse control, it is noteworthy that human studies have consistently pointed to an association between reduced central serotonin neurotransmission and the release of suppressed behaviors (Soubrie, 1986) including impulsive behaviors and aggression (Brown and Linnoila, 1990; Coccaro, 1989; Insei et al, 1990; Kavoussi and Coccaro, 1993; Linnoila et al, 1983; Stein et al, 1993). Central serotonergic disturbance may accompany some depressive disorders (Meltzer, 1990; Risch and Nemeroff, 1992), anxiety

350

Kafka

disorders (Kahn et al, 1987; van Praag et al, 1987), and consummatory behaviors (Amit et al, 1991), including alcoholism (Amit et al, 1991). The functional role of dopamine and norepinephrine for psychopathological disorders associated with paraphilic sexuality is less clear. Research literature exploring a relationship between impulsivity and these latter catecholamines is scant in comparison with the burgeoning data regarding serotonin. Dopamine appears to be associated with the mobilization, facilitation, and sustenance of goal- or incentive-related behavior (Blackburn et al, 1992; Crow, 1973; van Praag et al., 1990) and impairment in dopamine neurotransmission may mediate impulsivity and inattentiveness characteristic of attention-deficit disorder (Levy, 1991). Van Praag has hypothesized that central noradrenergic neurotransmission modulates the experience of pleasure and sensation. For example, alterations in plasma and CSF metabolites of norepinephrine have been reported in high sensation-seeking individuals (Ballanger et al., 1983; Roy et al, 1989) and may contribute to the etiology and symptomatology of depressive conditions (Siever and Davis, 1985).

PSYCHOPHARMACOLOGY, MONOAMNES, AND PARAPHILIC DISORDERS If monoamine neuorotransmitters contribute to the pathophysiology of sexual impulse disorders and these same neuromodulatory agents are implicated in the Axis I disorders and personality characteristics that are comorbid with sex offender paraphiliacs, then pharmacological agents that affect monoaminergic neurotransmission should affect paraphilic arousal as well. The three classes of drugs most commonly used to affect the neuroregulation of monoamines are the neuroleptics, psychostimulants, and antidepressants. A limited role for neuroleptics in the treatment of paraphilias has been discussed above. There are no published reports regarding the use of psychostimulants for the control of deviant sexuality, despite the diagnosis of learning disabilities and attention deficit hyperactivity disorder in some male paraphiliacs. It is possible that these agents have been underutilized because of the difficulty in assessing residual symptoms of attention deficit hyperactivity disorder in adults (Shaffer, 1994) as well as some case reports suggesting that psychostimulant abuse may disinhibit sexual behavior (Boffum et al, 1988). There have been case reports suggesting the efficacy of tricyclic antidepressants (Goldner, 1961; Kafka, 1991a; Snaith, 1981) and two case series with lithium carbonate (Bartova and Nahanek, 1979; Bartova et al., 1978) reporting a mitigation of

Hypothesis for the Pathophysiology of Paraphilias

351

paraphilic behaviors. A small but double-blind placebo-controlled crossover study comparing clomipramine, a predominantly serotonin reuptake inhibitor with desipramine, a predominantly norepinephrine reuptake inhibitor (Kruesi et al, 1992), reported that both were effective pharmacological agents to reduce paraphilic behavior. There are multiple case reports utilizing fluoxetine (e.g., Bianchi, 1990; Emmanuel et al, 1991; Jorgensen, 1991; Kafka, 1991a, 1991b; Lorefice, 1991; Perilstein et al, 1991), and open trials in inked groups of paraphiliacs utilizing fluoxetine in 16 males (Kafka and Prentky, 1992b), measuring outcome at 12 weeks, and either sertraline or fluoxetine in 24 males, measuring outcome at last visit, up to 1 year after the initiation of pharmacotherapy (Kafka, 1994b). In the aforementioned open-trial reports, there were reductions in total time consumed by deviant fantasies as well as reductions in paraphilic-related masturbation and overt sexual behaviors, regardless of primary paraphilic diagnosis. In general, socially conventional sexual behavior was not adversely affected by pharmacotherapy. Although many of the males were diagnosed with depressive conditions, a positive treatment response was independent of baseline depression rating. Bradford (1995a) reported a 12-week open trial of sertraline for 18 predominantly personality disordered heterosexual pedophiles. Statistically significant last-visit effects were reported including diminished pedophilic fantasies, decreased masturbation to pedophilic fantasies, and decreased penile tumescence to sexually arousing descriptions of pedophilia in comparison to baseline measures. In all the aforementioned reports utilizing serotonin reuptake inhibitors, statistically significant effects on target sexual behaviors are usually apparent after 4 weeks on active drug. Inasmuch as published manuscripts emphasize "positive" findings, it is worth mentioning that there are no published reports that contradict the aforementioned ameliorative effects of serotonergic agents on paraphilias. In conclusion, the review of the literature presented here affords suggestive evidence that paraphilias are disorders characterized by pathological dimensions including altered sexual preference, relative temporal stability, volitional impairment, and, possibly, increased sexual drive-related behaviors. Multiple paraphilic disorders appear to occur more commonly in affected persons implying that there is a paraphilic diathesis that predisposes to a variety of socially anomalous sexual outlets. In addition, paraphilic disorders can be accompanied by low self-esteem, social anxiety, social skills impairment, low grade anxious and depressive symptoms, and additional expressions of socially deviant impaired impulse control. These symptoms and syndromes may be, in part, mediated by perturbations in central monoamine neurotransmitters and can be ameliorated by pharmacological agents that affect serotonin, norepinephrine, and dopamine. Although

352

Kafka

much more research utilizing antidepressants and perhaps other pharmacologic monoaminergic neuromodulators (e.g., psychostimulants) remains to be investigated and reported, the most recent data are suggestive that serotonergic agents may represent a contemporary advance in the treatment of deviant sexuality.

REFERENCES Abel, G. G., Becker, J. V., Cunningham-Rathner, J., Mittelman, M., and Rouleau, J. L. (1988). Multiple paraphilic diagnoses among sex offenders. Bull. Am. Acad. Psychiat. Law 16: 153-168. Abel, G. G., and Osbom, C. (1992). The paraphilias: the extent and nature of sexually deviant and criminal behavior. In Bradford, J. M. W. (ed.), Psychiatric Clinics of North America: Clinical Forensic Psychiatry, W. B. Saunders, Philadelphia, pp. 675-687. Alderson, L. M., and Baum, M. J. (1981). Differential effects of gonadal steroids on dopamine metabolism in mesolimbic and nigro-striatal pathways of male rat brain. Brain Res. 218: 189-206. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th ed., APA, Washington, DC. Amit, Z., Smith, B. R., and Gill, K. (1991). Serotonin uptake inhibitors: effects on motivated consummatory behaviors. J. Clin. Psychiat. 52(12, Suppl.): 55-60. Anderson, W. P., Kunce, J. T. and Rich, B. (1979). Sex offenders: Three personality types. J. Clin. Psychol. 35: 671-6. Angrist, B. M., and Gershon, S. (1976). Clinical effects of amphetamine and L-dopa of sexuality and aggression. Comprehen. Psychiat. 17: 715-722. Ballenger, J. C, Post, R. M., and Jimerson, D. C. (1983). Biochemical correlates of personality in normals. Pers. Indiv. Diff. 4: 615-625. Barth, R. J., and Kinder, B. N. (1987). The mislabeling of sexual impulsivity. J. Sex Marital Ther. 13: 15-23. Bartova, D., and Nahanek, D. (1979). A comparative study of prophylactic lithium and diethylstilbestrol in sexual deviants. Activ. Nerv. Sup. (Praha) 21: 163-164. Bartova, D., Nahunek, K., and Svestka, J. (1978). Pharmacological treatment of deviant sexual behavior. Activ. Nerv. Sup. (Praha) 20: 72-74. Baum, M. J., and Starr, M. S. (1980). Inhibition of sexual behavior by dopamine antagonists or serotonin agonist drugs in castrated male rats given estradiol or dihydroteslosterone. Pharmacol. Biochem. Behav. 13: 47-67. Baxter, D. J., Marshall, W. L., and Barbaree, H. E. (1984). Deviant sexual behavior: Differentiating sex offenders by criminal and personal history, psychometric measures and sexual response. Grim. Justice Behav. 11: 477-501. Becker, J. V., Kaplan, M. S., Tenke, C. E., and Tartaglini, A. (1991). The incidence of depressive symptomatology in juvenile sex offenders with a history of abuse. Child Abuse Neglect 15: 531-6. Bell, D. S., and Trethowan, W. H. (1961). Amphetamine addiction and disturbed sexuality. Arch. Gen. Psychiat. 4: 74-78. Benkert, O., Bender, M., Bidlingmaier, P., and Butenandt, O. (1976). Effect of p-chlorophenylalanine (PCPA) on pituitary hormones and testosterone in the human. Arzneim. Forsck. 26: 1369-1371. Bianchi, M. (1990). Fluoxetine treatment of exhibitionism (letter). Am. J. Psychiat. 147: 1089-1090. Bitran, D., and Hull, E. M. (1987). Pharmacological analysis of rat male sexual behavior. Neurosci. Biobehav. Rev. 11: 365-389.

Hypothesis for the Pathophysiology of Paraphilias

353

Blackburn, J. R., Pfaus, J. G., and Phillips, A. G. (1992). Dopamine functions in appetitive and defensive behaviors. Prog. Neurobiol. 39: 247-279. Boffum, J., Moser, C, and Smith, D. (1988). Street drugs and sexual function. In Sitsen, J. M. A. (ed.), Handbook of Sexology, Elsevier, New York, pp. 462-477. Bowers, M. B., Woert, M. V., and Davis, L. (1971). Sexual behavior during L-DOPA treatment for parkinsonism. Am. J. Psychiat. 127: 1691-1693. Bradford, J. M. W. (1995a). An open pilot study of sertraline in the treatment of outpatients with pedophilia. Presented at the 148th Annual Meeting of the American Psychiatric Association, Miami FL. Bradford, J. M. W. (1995b). Pharmacological treatment of the paraphilias. In Oldham, J. M., and Riba, M. B. (eds.), Review of Psychiatry, American Psychiatric Press, Washington, DC, pp. 755-778. Bradford, J. M. W., Boulet, J., and Pawlak, A. (1992). The paraphilias: A multiplicity of deviant behaviors. Can. J. Psychiat. 37: 104-108. Brotherton, J. (1974). Effect of oral cyproterone acetate on urinary and serum FSH and LH levels in adult males being treated for hypersexuality. J. Reprod. Fertil. 36: 177-187. Brown, G. L., and Linnoila, M. I. (1990). CSF serotonin metabolite (5-HIAA) studies in depression, impuisivity, and violence. J Clin. Psychiat. 51(4, Suppl.): 31-41. Carnes, P. (1983). Out of the Shadows: Understanding Sexual Addiction, CompCare, Minneapolis, MN. Carnes, P. (1989). Contrary to Lave: Helping the Sexual Addict, CompCare, Minneapolis, MN. Carnes, P. (1990). Sexual addiction. In Morton, A., Johnson, B. L., and Roundy, L. M. (eds.), The Incest Perpetrator: A Family Member No One Wants to Treat, Sage, Newbury Park, CA. Carnes, P. (1991). Don't Call It Love: Recovery from Sexual Addiction, Bantam Books, New York. Clark,, J. T., Smith, E. R., and Davidson, J. M. (1984). Enhancement of sexual motivation in male rats. Science 222, 847-849. Clark, J. T., Smith, E. R., and Davidson, J. M. (1985). Evidence for the modulation of sexual behavior by alpha adenoreceptors in male rats. Neuroendocrinology 41: 36-43. Coccaro, E. F. (1989). Central serotonin and impulsive aggression. Br. J. Psychiat. 155 (Suppl. 8): 52-62. Cohen, M., Seghom, T., and Calmas, W. (1969). Sociometric study of the sex offender. J. Abn. Psychol. 74: 249-255. Coleman, E. (1986). Sexual compulsion vs. sexual addiction: The debate continues. SIECUS Rep. July, pp. 7-11. Coleman, E. (1987). Sexual compulsivity: Definition, etiology, and treatment considerations. /. Chem. Depend, Treat. 1: 189-204. Cremata, V. Y. J., and Koe, B. K. (1966). Clinical-pharmacological evaluation of p-chlorophenylalanine: A new serotonin-depleting agent. Clin. Pharmacol. Ther, 7: 768-776. Crow, T. J. (1973). Catecholamine-containing neurones and electrical self-stimulation: A theoretical interpretation and some psychiatric implications. Psychol. Med. 3: 66-73. Davies, T. S. (1974). Cyproterone acetate for male hypersexuality. J. Int. Med. Res. 2:159-163. Emmanuel, N. P., Lydiard, R. B., and Ballenger, J. C. (1991). Fluoxetine treatment of voyeurism (letter). Am. J. Psychiat. 148: 950. Everitt, B. J. (1980). Alterations in the sexual behavior and 5-hydroxyindolacetic acid in the cerebrospinal fluid of female rhesus monkeys treated with clomipramine. Postgrad. Med. J. 56(Suppl. 1): 53-57. Everitt, B. J. (1983). Monoamines and the control of sexual behavior. Psychol. Med. 13: 715-720. Everitt, B. J. (1990). Sexual motivation: A neural and behavioral analysis of the mechanisms underlying appetitive and copulatory responses of the male rat. Neurosei. Biobehav. Rev. 14: 217-232.

354

Kafka

Everitt, B. J., and Bancroft, J. (1991). Of rats and men: The comparative approach to male sexuality. In Bancroft, J., Davis, C. M, and Ruppel Jr., H. J. (eds.), Annual Review of Sex Research, Society for the Scientific Study of Sex, Mt. Vernon, IA, pp. 77-118. Everitt, B. J., Herbert, J., and Keverne, E. B. (1983). Neuroendocrine anatomy of the limbic system. In Navaratnam, V. (ed.), Progress in Anatomy, Cambridge University Press, Cambridge, U.K. Fagan, J., and Wexler, S. (1986). Explanations of sexual assault among violent delinquents. /. Adolescent Res. 3: 363-385. Fagan, P. J., Wise, T. N., and Schmidt, C. W. (1991). A comparison of five-factor personality dimensions in males with sexual dysfunction and males with paraphilia. J. Pers. Assess. 57: 434-448. Feder, H. M. (1978). Specificity of steroid hormone activation of sexual behavior in rodents. In Hutchison, J. B. (ed.), Biological Determinants of Sexual Behavior, Wiley, Chichester, UK., pp. 395-424. Ferguson, J., Henriksen, S., Cohen, H., Mitchell, G., Barchas, J., and Dement, W. (1970). "Hypersexuality" and behavioral changes in cats caused by administration of p-chlorophenylalanine. Science 168: 499-501. Freund, K., Watson, R., and Dickey, R. (1991). Sex offenses against female children perpetrated by men who are not pedophiles. /. Sex Res. 28: 105-112. Gebhard, P. H., Gagnon, J., Pomeroy, W., and Christenson, C. (1965). Sex Offenders: An Analysis of the Types, Harper & Row, New York. Gessa, G. L., and Tagliamonte, A. (1975). Role of brain serotonin and dopamine in male sexual behavior. In Sandier, M., and Gessa, G. L. (eds.), Sexual Behavior: Pharmacology and Biochemistry, Raven, New York. Goldner, R. D. (1961). Control of minor criminal sexual compulsions with imipramine and amine oxidase inhibitors. Presented at the Third World Congress of Psychiatry, Montreal, Canada (June 4-10, 1961). Gorzalka, B. B., Mendelson, S. D., and Watson, N. V. (1990). Serotonin receptor subtypes and sexual behavior. In Whitaker-Azmitia, P. M., and Peroutka, S. J. (eds.), Annals of the New York Academy of Sciences: The Neuropharmacology of Serotonin, New York Academy of Sciences, pp. 435-446. Grossman, L. S., and Cavanaugh, J. L. (1990). Psychopathology and denial in alleged sex offenders. J. Nerv. Ment. Dis. 178: 739-744. Haensel, S. M., Mos, J., VanderSchoot, P., and Slobak, L. (1993). Testosterone is required for the stimulatory effects of 8-OH-DPAT on sexual behavior in castrated male rats. Eur. J. Pharmacol. 233: 187-192. Heninger. G. R. (1995). Indolamines: The role of serotonin in clinical disorders. In Bloom, F. E., and Kupfer, D. J. (eds.), Psychopharmacology: The Fourth Generation of Progress, Raven, New York, pp. 471-482. Hsiao, J. K., Agren, H., Bartko, J. J., Rudorfer, M. V., Linnoila, M., and Potter, W. Z. (1987). Monoamine neurotransmitter interactions and the prediction of antidepressant response. Arch. Gen. Psychiat. 44: 1078-1083. Hunter, J. A., and Goodwin, D. W. (1992). The clinical utility of satiation therapy with juvenile sex offenders: variations and efficacy. Ann. Sex Res. 5: 71-80. Inset, T. R., Zohar, J., Benkelfat, C., and Murphy, D. L. (1990). Serotonin in obsessions, compulsion, and the control of aggressive impulses. In Whitaker-Azmitia, P. M., and Peroutka, S. (eds.), The Neuropharmacology of Serotonin, The New York Academy of Sciences, New York, pp. 574-586. Jacobsen, F. M. (1992). Fluoxetine-induced sexual dysfunction and an open trial of yohimbine. /. Clin. Psychiat. 53: 119-122. Jorgensen, V. I. (1991). Cross-dressing successfully treated with fluoxetine (letter). N.Y. State J. Med. 90: 566-567. Kafka, M. P. (1991a). Successful antidepressant treatment of nonparaphilic sexual addictions and paraphilias in men. 7. Clin. Psychiat. 52: 60-65. Kafka, M. P. (1991b). Successful treatment of paraphilic coercive disorder (a rapist) with fluoxetine hydrochloride. Br. J. Psychiat. 158: 844-847.

Hypothesis for the Pathophysiology of Paraphilias

355

Kafka, M. P. (1994a). Paraphilia-related disorders: common, neglected, and misunderstood. Harvard Rev. Psychiat. 2: 39-40. Kafka, M. P. (1994b). Sertraline pharmacotherapy for paraphilias and paraphilia-related disorders: An open trial. Ann. Clin. Psychiat. 6: 189-195. Kafka, M. P. (1995). Sexual impulsivity. In Hollander, E., and Stein, D. J. (eds.), Impulsivity and Aggression, Publisher, Chichester, U.K., pp. 201-228. Kafka, M. P. (1997). Hypersexual desire in males: An operational definition and clinical implications for men with paraphilias and paraphilia-related disorders. Arch. Sex. Behav. (in press). Kafka, M. P., and Coleman, E. (1991). Serotonin and paraphilias: The convergence of mood, impulse and compulsive disorders (letter). J. Clin. Psychopharmacol. 11: 223-224. Kafka, M. P., and Prentky, R. (1992a). A comparative study of nonparaphilic sexual addictions and paraphilias in men. J. Clin. Psychiat. 53: 345-350. Kafka, M. P., and Prentky, R. (1992b). Fluoxetine treatment of nonparaphilic sexual addictions and paraphilias in men. J. Clin. Psychiat. 52: 351-358. Kahn, R. S., Westenberg, H. G. M., and Verhoven, W. M. A. (1987). Effect of serotonin precursor uptake inhibitors in anxiety disorders: A double blind comparison of 5-hydroxytryptophan, clomipramine and placebo. Int. Clin. Psychopharm. 2: 33-45. Kalichman, S. C. (1990). Affective and personality characteristics of MMPI profile subgroups of incarcerated rapists. Arch. Sex. Behav. 19: 443-459. Kavoussi, R. J., and Coccaro, E. F. (1993). Impulsive personality disorders and disorders of impulse control. In Hollander, E. (ed.), Obsessive-Compulsive-Related Disorders, American Psychiatric Press, Washington DC, pp. 179-202. Kavoussi, R. J., Kaplan, M., and Becker, J. V. (1988). Psychiatric diagnoses in adolescent sex offenders. J. Am. Acad. Child Adolescent Psychiat. 27: 241-243. Kendall, D., and Tonge, S. (1976). Monoamine concentrations in eight areas of the brain of mature and immature male rats and the effects of injected hormones. J. Pharm. Pharmacol. 28(Suppl.): 37. Knight, R. A., and Prentky, R. A. (1990). Classifying sex offenders: The development and corroboration of taxonomic models. In Marshall, W. L., Laws, D. R., and Barbaree, H. E. (eds.), Handbook of Sexual Assault: Issues, Theories, and the Treatment of Offenders, Plenum Press, New York, pp. 23-52. Kruesi, M. J. P., Fine, S., Valladares, L., Phillips, R. A., and Rapoport, J. L. (1992). Paraphilias: A double-blind crossover comparison of clomipramine versus desipramine. Arch. Sex. Behav. 21: 587-593. Langevin, R. (1983). Sexual Strands: Understanding and Treating Sexual Anomalies in Men, Erlbaum, Hillsdale, NJ. Langevin, R., and Langs, R. A. (1990). Substance abuse among sex offenders. Ann. Sex Res. 3: 397-424. Langevin, R., Wright, P., and Handy, L. (1990a). Use of the MMPI and its derived scales with sex offenders: 1. Reliability and validity studies. Ann. Sex Res. 3: 245-291. Langevin, R., Wright, P., and Handy, L. (1990b). Use of the MMPI and its derived scales with sex offenders: 2. Reliability and criterion validity. Ann. Sex Res. 3: 453-486. Le Moal, M. (1995). Mesocorticolimbic dopaminergic neurons. In Bloom, F. E., and Kupfer, D. J. (eds.), Psychopharmacology: The Fourth Generation of Progress, Raven, New York, pp. 283-293. Levin, S. M., and Stava, L. (1987). Personality characteristics of sex offenders: A review. Arch. Sex. Behav. 16: 57-79. Levy, F. (1991). The dopamine theory of attention deficit hyperactivity disorder (ADHD). Aust. N.Z. J. Psychiat. 25: 277-283. Linnoila, M., Oliver, J., Adinoff, B., and Potter, W. (1988). High correlations of norepinephrine, dopamine and epinephrine and their major metabolic excretion rates. Arch. Gen Psychiat. 45: 710-714. Linnoila, M., Virkhunen, M., Scheinin, A., Nuutila, A., Rimon, R., and Goodwin, F. K. (1983). Low cerebrospinal spinal fluid 5-hydroxyindoleacetic acid concentration differentiates impulsive from nonimpulsive violent behavior. Life Sci. 33: 2609-2614.

356

Kafka

Lorefice, L. S. (1991). Fluoxetine treatment or a fetish (letter). J. Clin. Psychiat. 52: 41. Malamuth, N., Knight, R., and Prentky, R. (1995). A unified developmental theory of sexual aggression: Models in the making. Presented at the Fourteenth Annual Research and Treatment Conference of the Association for the Treatment of Sexual Abusers (ATSA), October 11-14, New Orleans, LA. Malamuth, N. M., Sockloskie, R. J., Koss, M. P., and Tanaka, J. S. (1991). Characteristics of aggressors against women: Testing a model using a national sample of college students. /. Consult. Clin. Psychol. 59: 670-681. Malmnas, C. O. (1973). Monoaminergic influence on testosterone activated copulatory behavior in castrated male rats. Acta Physiol. Scand. 395(Suppl.): 1-128. Marshall, W. L. (1989). Intimacy, loneliness, and sexual offenders. Behav. Ther. Res. 27: 491-503. Mas, M. (1995). Neurobiological correlates of masculine sexual behavior. Neurosci. Biobehav. Rev. 19: 261-277. McCreary, C. P. (1975). Personality profiles of persons convicted of indecent exposure. J. Clin. Psychol. 31: 260-2. Meltzer, H. (1990). Role of serotonin in depression. In Whitaker-Azmitia, P. M., and Peroutka, S. J. (eds.), The Neuropharmacology of Serotonin, New York Academy of Sciences, New York, pp. 486-500. Meltzer, H. Y., and Lowy, M. T. (1987). The serotonin hypothesis in depression. In Meltzer, H. Y. (ed.), Psychopharmacology: The Third Generation of Progress, Raven, New York, pp. 513-526. Menendez-Abraham, E., Moran-Viesca, P., and Velasco-Plaza, A. (1988). Modifications of the sexual activity in male rats following the administration of antiserotoninergic drugs. Behav. Brain Res. 30: 251-258. Meyerson, B. J. (1984). Hormone-dependent socio-sexual behaviors and neurotransmitters. In DeVries, G. J., DeBruin, J. P. C, Uylings, H. B. M., and Corner, M. A. (eds.), Sex Differences in the Brain, Elsevier, Amsterdam, pp. 271-280. Mio, J. S., Nanjundappa, G., Verleur, D. E., and de Rios, M. D. (1986). Drug abuse and the adolescent sex offender: A preliminary analysis. J. Psychoactive. Drugs 18: 65-72. Mohr, J. W., Turner, R. E., and Jerry, M. B. (1964). Pedophilia and exhibitionism. University of Toronto Press, Toronto, ONT. Morales, A., Condra, M., Owen, J. A., Surridge, D. H., Fenemore, J., and Harris, C. (1987). Is yohimbine effective in the treatment of organic impotence? Results of a controlled trial. J. Urol 137: 1168-1172. O'Brien, C. P., DiGiacommo, J. R, Fahn, S., and Schwartz, G. A. (1971). Mental effects of high dose levodopa. Arch. Gen. Psychiat. 24: 61-64. Orford, J. (1978). Hypersexuality: Implications for a theory of dependence. Br. J. Addict. 73: 299-310. Paglietti, E., Pelligrini-Quarantotti, B., Merev, G., and Gessa, G. L. (1978). Apomorphine and L-DOPA lower ejaculation threshold in the male rat. Physiol. Behav. 20: 559-562. Pearson, H. J. (1990). Paraphilias, impulse control, and serotonin [letter]. J. Clin. Psychopharmacol. 10: 233. Perilstein, R. D., Upper, S., and Friedman, L. J. (1991). Three cases of paraphilias responsive to fluoxetine treatment. J. Clin. Psychiat. 52: 169-170. Pfaus, J. G., and Everitt, B. J. (1995). The psychopharmacology of sexual behavior. In Bloom, F. E., and Kupfer, D. J. (eds.), Psychopharmacology: The Fourth Generation of Progress, Raven, New York, pp. 743-758. Prentky, R. A., and Knight, R. A. (1986). Impulsivity in the lifestyle and criminal behavior of sexual offenders. Crim. Justice Behav. 13: 141-164. Prentky, R. A, and Knight, R. A. (1991). Identifying critical dimensions for discriminating among rapists. J. Consult. Clin. Psychol. 59: 643-661. Quadland, M. C. (1985). Compulsive sexual behavior: Definition of a problem and an approach to treatment J. Sex Marital Ther. 11: 121-132. Rada, R. T. (1975). Alcoholism and forcible rape. Am. J. Psychiat. 132: 444-6.

Hypothesis for the Pathophysiology of Paraphilias

357

Risch, S. C, and Nemeroff, C. B. (1992). Neurochemical alterations of serotoninergic neuronal systems in depression. J. Clin. Psychiat. 53(Suppl.): 3-7. Robbins. T. W., and Everitt, B. J. (1995). Central norepinephrine neurons and behavior. In Bloom, F. E., and Kupfer, D. J. (eds.), Psychopharmacology: The Fourth Generation of Progress, Raven, New York, pp. 363-372. Rosenberg, R., Knight, R. A., and Prentky, R. A. (1988). Validating the components of a taxonomic system for rapists: A path analytic approach. Bull. Am. Acad. Psychiat. Law 16: 169-185. Roy, A., De Jong, J., and Linnoila, M. (1989). Extraversion in pathological gamblers: Correlates with indexes of noradrenergic function. Arch. Gen. Psychiat. 46: 679-681. Segraves, R. T. (1988). Drugs and desire. In Lieblum, S. R., and Rosen, R. C. (eds.), Sexual Desire Disorders, Guilford, New York, pp. 313-340. Segraves, R. T. (1989). Effects of psychotropic drugs on human erection and ejaculation. Arch. Gen. Psychiat. 46: 275-284. Shaffer, D. (1994). Attention deficit hyperactivity disorder fa adults. Am. J. Psychiat. 151: 633-638. Sheard, M. H. (1969). The effects of p-chlorophenylalanine on behavior in rats: Relation to brain serotonin and 5-hydroxyindole acetic acid. Brain Res. 15: 524-528. Sicuteri, F. (1974). Serotonin and sex in man. Pharm. Res. Commun. 4: 403-411. Siddiqui, A., and Gilmore, D. P. (1988). Regional differences in the catecholamine content of the rat brain: Effects of neonatal castration and androgenization. Acta Endocrinol. 118: 483-494. Siever, L. J. (1987). Role of noradrenergic mechanisms in the etiology of affective disorders. In Meltzer, H. (ed.), Psychopharmacology: The Third Generation of Progress, Raven, New York, pp. 493-504. Siever, L. J., and Davis, K. L. (1985). Overview: Toward a dysregulation hypothesis of depression. Ann. J. Psychiat. 142: 1017-1031. Snaith, R. P. (1981). Five exhibitionists and a method of treatment. Br. J. Psychiat. 138: 126-130. Sodersten, P., Larsson, K., Ahlenius, S., and Engel, J. (1976). Sexual behavior in male rats treated with monoamine synthesis inhibitors and testosterone. Pharmacol. Biochem. Behav. 5: 319-327. Sonda, L. P., Mazo, R., and Chancellor, M. B. (1990). The role of yohimbine in the treatment of erectile impotence. J. Sex Marital Ther. 16: 15-21. Soubrie, P. (1986). Reconciling the role of central serotonin neurons in human and animal behavior. Behav. Brain Res. 9: 319-364. Stein, D. J., Hollander, E., and Liebowitz, M. R. (1993). Neurobiology of impulsivity and the impulse control disorders. J. Neuropsychiat. 5: 9-17. Sulser, F. (1987). Serotonin-norepinephrine receptor interactions in the brain: Implications for the pharmacology and pathophysiology of affective disorders. J. Clin. Psychiat. 48(3, Suppl.): 12-18. Tagliamonte, A., Tagliamonte, P., and Gessa, G. L. (1969). Compulsive sexual activity induced by p-chlorophenylalanine in normal and pinealectomized rats. Science 166: 1433-1435. Tennent, G., Bancrott, J., and Cass, J. (1974). The control of deviant sexual behavior by drugs. A double-blind controlled study of benperidol, chlorpromazine and placebo. Arch. Sex. Behav. 3: 261-271. Tucker, J. C, and File, S. E. (1983). Serotonin and sexual behavior. In Wheatley, D. (ed.), Psychophamacology and Sexual Disorders, Oxford University Press, Oxford, U.K., pp. 22-49. van Praag, H. M., Asms, G. M., Kahn, R. S., Brown, S. L., Korn, M., Harkavy-Friedman, J. M., and Weltzer, S. (1990). Monoamines and abnormal behavior: A multiaminergic perspective. Br. J. Psychiat. 157: 723-734. van Praag, H. M., Brown, S. L., Asnis, G. M., Kahn, R. S., Kom, M. L., Harkavy-Friedman, J. M., and Weltzer, S. (1991). Beyond serotonin: A multiaminergic perspective on abnormal behavior. In van Praag, H. M., and Brown. S. L. (eds.). The Role of Serotonin in Psychiatric Disorders, Bruner/Mazel, New York, pp. 302-332.

358

Kafka

van Praag, H. M, Kahn, R. S., and Asnis, G. M. (1987). Denosologization of biological psychiatry or the specificity of 5-HT disturbances in psychiatric disorders. J. Affect. Disord. 13: 1-8. Waldmeier, P. C., and Delini-Stula, A. A. (1979). Serotonin-dopamine interactions in the nigro-striatal system. Eur J. Pharmacol. 55: 363-373. Watts, A. G., and Stanley, H. F. (1984). Indolamines in the hypothalamus and the area of the midbrain raphe nuclei of male and female rats throughout postnatal development. Neuroendocrinology 38: 461-466. Wise, T. N., Pagan, P. J., and Schmidt, C. W. (1991). Personality and sexual functioning of transvestic fetishists and other paraphiliacs. J. Nerv. Ment. Dis. 179: 694-698. Zbytovsky, J. (1993). Haloperidol decanoate (Janssen) in the treatment of sexual deviations. Cesk. Psychiat. 89: 15-17.

Archives of Sexual Behavior, Vol. 26, No. 4, 1997

Psychoendocrinological Assessment of the Menstrual Cycle: The Relationship Between Hormones, Sexuality, and Mood Stephanie H. M. Van Goozen, Ph.D.,1.6 Victor M. Wiegant, Ph.D.,2.3 Erik Endert, Ph.D.,4 Frans A. Helmond, Ph.D.,3 and Nanne E. Van de Poll, Ph.D.5

The role of sex hormones in sexuality and mood across the menstrual cycle was investigated. Twenty-one normal healthy women were followed for one menstrual cycle. Blood samples were taken frequently, and analyzed for estradiol, progesterone, testosterone, androstenedione, dehydroepiandrosterone sulfate, cortisol, and sex hormone-binding globulin. A diary concerning sexual interest and behavior, and different moods, was completed daily. Although the sample was not large, a clear effect of menstrual cycle phase on levels of testosterone and the free testosterone index was demonstrated. In a preliminary screening interview, 11 of the 21 women had reported that they suffered from premenstrual complaints (PC), the other 10 had reported no complaints in the premenstrual phase (NPC). Significant differences between the two groups were established in estradiol and the estradiol-progesterone ratio, with the NPC group having higher levels of both endocrine parameters across different menstrual samples. Psychologically, a cycle effect on tension and sexual interest was demonstrated. The NPC group reported a peak in sexual interest in the The study was supported by grants from the "De Drie Lichten" Foundation the "Dr. Saal Van Zwanenberg" Foundation and Organon International BV. 1Department of Child and Adolescent Psychiatry, University Hospital Utrecht B01-201, P.O. Box 85500, 3508 GA Utrecht, The Netherlands. 2Departmentof Medical Pharmacology, Rudolf Magnus Institute of Neurosciences, Utrecht University, The Netherlands. 3Department of Human and Animal Physiology, Agricultural University, Wageningen, The Netherlands. 4Laboratory for Endocrinology and Radiochemistry, Academic Medical Center, Amsterdam, The Netherlands. 5Department of Psychology, University of Amsterdam, The Netherlands. 6To whom correspondence should be sent.

359 0004-0002/97/0800-0359$12.50/0 © 1997 Plenum Publishing Corporation

360

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

premenstrual phase, whereas the PC group reported a peak in the ovulatory phase. There was a difference between the two groups in feelings of fatigue but not in other moods across the menstrual cycle. The study provides further evidence of the importance of androgen levels in women's sexuality and shows again that the relationship between menstrual cycle phase and sexuality is much clearer than between phase and mood. KEY WORDS: sexuality; mood; androgens; premenstrual complaints; sex hormones.

INTRODUCTION Background During the last decades the menstrual cycle has captured the interest of scientists as a naturalistic model for investigating the influence of gonadal steroids on brain and behavior (Hamilton et al., 1988). Activating effects of sex hormones on various mental and behavioral functions in women have been demonstrated by studying the covariation of levels of sex steroids and self-reports of these psychological factors at different phases of the menstrual cycle. Much attention has focused on sexuality. Earlier studies on a variety of female mammals indicated that the ovulatory period, when estrogen levels are high, is the prime phase of the cycle in which the female is responsive to males and sexual behavior is shown (Beach, 1976). In line with these findings, a midcycle peak in sexual activity has been reported in human females (Adams et al., 1978; Matteo and Rissman, 1984). However, there have also been frequent failures to find an ovulatory peak in sexual behavior (Sanders and Bancroft, 1982); some studies found evidence of a premenstrual peak (Schreiner-Engel et al., 1981; Bancroft, 1984), whereas others found evidence of both an ovulatory and pre- or postmenstrual peak (Bancroft et al., 1983). The inconsistency of results concerning changes in sexuality throughout the cycle could not lead to a consistent idea about which hormones are primarily involved in these behavioral functions. Studies showing a midcycle ovulatory peak in sexuality have led researchers to infer that estrogens are the most important hormones for sexual activity in women. However, despite the large differences in estradiol and progesterone levels resulting from normal variations in steroid levels during the menstrual cycle, Abplanalp et al. (1979) found that measures of sexuality were stable throughout the cycle; they also did not find a relationship between estradiol levels and enjoyment of heterosexual activity or number of heterosexual activities. Persky, Charney, et al. (1978) failed to demonstrate a direct relationship between plasma estradiol levels and sexual arousal, intercourse frequency, and

Hormones, Sexuality, and Moods in Women

361

sexual gratification. The occurrence of a premenstrual peak in sexuality has been ascribed to an inhibiting effect of progesterone (Bancroft, 1984). Schreiner-Engel et al., (1981), on the other hand, found a positive correlation between subjectively experienced sexual arousal and progesterone levels during the luteal phase. Animal studies have indicated that androgens contribute significantly to the expression of sexual behavior in female mammals (Van de Poll and Van Goozen, 1992). A distinction has been drawn between receptivity, which is the passive acceptance of a male animal by the female, which is primarily estrogen-dependent, and proceptivity, which is the active searching for a male and the initiation of copulation by the female (Beach, 1976). Studies have provided evidence of a relation between androgens and proceptivity in female rats (De Jonge and Van de Poll, 1984; De Jonge et al., 1986). In women, both the adrenal cortex and the ovary contain the biosynthetic pathways necessary for androgen synthesis and secretion, with the ovary producing 5-25% of plasma-testosterone, 45-60% of androstenedione, and approximately 20% of dehydroepiandrosterone (DHEA), whereas the adrenal cortex produces 5-25% of circulating testosterone, 3040% of androstenedione, 80% of DHEA, and more than 95% of dehydroepiandrosterone sulfate (DHEAS). The remainder of circulating androgens in the female results from peripheral conversion which presumably accounts for a production rate of 50-70% of testosterone and 10% of androstenedione (Greenspan, 1991; Sherwin, 1988). On the basis of animal research and a growing number of studies on women, the current balance of opinion is leaning toward the view that androgens play a more important role than estradiol. Higher levels of testosterone have been associated with more intense sexual interest and arousal (Alexander and Sherwin, 1993; Bancroft et al., 1980; Dabbs and Mohammed, 1992; Schreiner-Engel et al., 1981). In two studies (Persky, Lief et al., 1978; Morris et al., 1987) a relation was demonstrated between midcycle peak values of testosterone in women and intercourse frequency of couples. Moreover, research on the effects of androgen replacement therapies in postmenopausal women (Sherwin et al., 1985; Sherwin and Gelfand, 1987) and studies on the effects of testosterone administration to female-to-male transsexuals and of androgen-deprivation in male-to-female transsexuals (Van Goozen et al., 1995) demonstrated that androgenic hormones are an important factor in libido of both men and women. It is unclear which androgens are responsible for the behavioral effects of androgens. For example, not only is androstenedione—as a prehormone—peripherally converted to estrogen and other androgens, but it is unclear whether androstenedione has much psychological action in its

362

Van Goozen, Wiegant, Endert, Heimond, and Van de Poll

own right. Ribeiro et al. (1974) tested the hypothesis that serum levels of androstenedione, as a precursor of estradiol, had a similar secretion pattern. In six normal menstrual cycles no clear pattern of changing levels of androstenedione was found equivalent to the one observed for estradiol and progesterone. Although the daily fluctuations of androstenedione were great, the mean levels of androstenedione generally increased from menstruation to midcycle, but did not parallel the patterns of estradiol, and declined again from midcycle to menstruation. There were no clearcut differences between the levels of androstenedione in the follicular and luteal phases. In a sample of six women, Judd and Yen (1973) found that the mean androstenedione level during the middle third of the menstrual cycle was significantly higher than during the earlier or later portions. The same applied to the mean testosterone level, but the rise in testosterone was smaller than that of androstenedione. And finally, it is unclear whether there are cycle-related fluctuations in baseline levels of the androgenic hormones with a strong adrenocortical component (DHEA and DHEAS) and of cortisol (Abplanalp et al., 1977; Beck et al., 1990; Saxena et al., 1974), and if so, whether these are related to changes in psychological attributes. Research on the relation between hormonal changes during the menstrual cycle and changes in the experience of emotions has focused mainly on whether the period preceding the onset of menstruation is accompanied by an increase in negative mood. The results of different studies are conflicting: sometimes an increase in negative mood during the premenstrual phase has been reported (Backstrom et al., 1983; Sanders et al, 1983), and sometimes not (Abplanalp et al., 1979). Parlee (1982) even found significantly lower negative mood scores in the premenstrual than in the periovulatory phase. Studies on the effects of hormonal fluctuations during the menstrual cycle on sexuality and emotional well-being suffer from a number of methodological problems: (i) different aspects of sexuality have been studied, ranging from frequency of intercourse and masturbation to intensity of feeling passionate and affectionate; (ii) different groups of women have been involved: normal women, normal women who suffer from emotional and physical complaints in the premenstrual phase (i.e., premenstrual symptoms), and women who attend a gynecological clinic because of their severe premenstrual complaints; and (iii) different methodologies have been used to infer menstrual phases (e.g., body temperature measurements, subjective reports, hormone measurements). The fact that studies vary along these lines is probably to a large extent responsible for the mixed and inconsistent findings in this domain.

Hormones, Sexuality, and Moods in Women

363

Purposes of the Study For the duration of one menstrual cycle a sample of normal and regularly menstruating women was followed. Blood samples were gathered frequently to assess a number of different ovarian and adrenocortical hormones, including some for which there is hardly any information concerning their secretion pattern during the menstrual cycle [cortisol, DHEAS, sex hormone binding globulin (SHBG), and the free testosterone index (FTI), i.e., the proportion of unbound, biologically active, testosterone]. Daily reports of sexual motivation and behavior, and emotional wellbeing were made. All endocrinological and psychological variables were analyzed to establish whether or not there was a cyclic variation throughout the cycle. As yet, it is unclear whether women suffering from premenstrual complaints have an aberrant menstrual pattern of changing ovarian hormones: It is unclear whether sex hormones are involved, and if so, which ones they are and whether a deviant pattern is restricted to the premenstrual phase or also extends to other phases of the menstrual cycle. For example, it has been claimed that premenstrual syndrome (PMS) is related to the withdrawal of estradiol or of progesterone, or to a disturbance in the estrogen-progesterone ratio (Clare, 1985; Janowsy and Rausch, 1985). However, Backstrom et al. (1983) did not find evidence of a hormonal difference between women with and without PMS complaints on progesterone, estradiol, testosterone, and androstenedione. As a subsidiary goal of the present study we investigated whether there were endocrinological and/or psychological differences between women who reported suffering from premenstrual complaints and women who reported being free of premenstrual problems. To that end, the group was composed of women who differed in the perception of the severity of their premenstrual complaints.

METHOD Subjects An advertisement was placed in local newspapers, asking for subjects willing to participate in research on the relation between hormonal changes during the menstrual cycle and moods and sexuality. Four hundred fortyfour women responded by telephone. During this telephone call questions were asked about the length and regularity of the menstrual cycle, use of contraception, marital status, and occupation, and the procedure of the study was explained. On the basis of the information gathered 93 women

364

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

were selected for an extensive structured interview held at our laboratory. The other 348 women were not invited because they had no regular menstrual cycles, they were using contraceptives, or they were not willing to take part in a 1-month study involving, among other things, coming to the laboratory every other day to give blood samples. The focus of the selection interview was broad, ranging from questions about sociodemographic status, personal history and development, education, employment status, and sexuality, to current and past menstrual cycle functioning. The goal of the selection procedure was to compose a group of women satisfying the following criteria: report of a regular menstrual cycle of 26-30 days, no use of oral contraceptives, leading a sexually active life, and making a psychologically stable impression. Fifty-eight women fulfilled these criteria and took part in a study on the effects of the premenstrual phase on emotionality during aversive events (Van Goozen et al., 1996). Of this group, 21 women were willing to take part despite the rather arduous procedure of this particular study in which blood samples were gathered frequently. Of these, 11 women had reported diverse premenstrual problems returning every cycle. The most frequently mentioned complaints were irritability, mood swings, headaches, depression, nausea, and complaints related to water retention. None of the women was under treatment by a physician for these complaints.7 Furthermore, it was established that they had these complaints only in the premenstrual phase. This was the premenstrual complaints or PC group. No attempt was made to verify whether these women actually experienced the symptoms they complained of because this was not the focus of the study. The other 10 women reported not suffering from any complaints across all phases of the cycle (NPC group). Of the NPC group 7 women used condoms, 1 a coil, and 2 were lesbian; of the PC group 7 women used condoms and 4 a coil. The age of the women was between 24 and 40 years (mean age of the PC group = 29.9; mean age of the NPC = 27.9). All subjects were involved in a stable, single-partner sexual relationship, but not all of them were married or lived together. Fifteen subjects had at some time in the past taken oral contraceptives, but all had ceased taking the pill at least 6 months prior to the time of the study. Medical histories based on interview information indicated that all subjects were in good health and were not taking any medication that might interfere with endocrine function or emotional state; none of the women were overweight. All had reported a history of normal and regular 7We

emphasize that the group of women who reported suffering from premenstrual complaints (PC) cannot be compared to women suffering from PMS or late luteal phase dysphoric disorder. For that, the complaints should meet certain criteria. Our distinction between NPC and PC women is solely based on the subjective perception of women about their premenstrual experience.

Hormones, Sexuality, and Moods in Women

365

menstrual cycles (28 ± 2 days) and the sociodemographic data of the two groups were similar.

Procedure

Each subject was studied for one complete menstrual cycle. For each subject, depending on the length of the menstrual cycle, 12-14 blood samples were collected: two 10-ml blood samples were taken by venipuncture. Since it is known that cortisol displays a marked diurnal variation, blood samples were drawn between 7.30 and 8.30 AM, three times per week at 2-day intervals. Subjects were not allowed to smoke, drink coffee, or have a high cholesterol breakfast before blood was collected. Throughout the study subjects were asked to refrain from drinking alcohol and from going to bed late. Subjects were asked to choose a period during which they expected no special life events (negatively or positively stressful occasions, such as parties, birthdays, vacations, exams) to occur. The total sample was divided into three groups of 7 women; all subjects within one group started on the same day, irrespective of the day of her menstrual cycle. Moods, sexuality, physical symptoms, and experience of stressors were reported every evening: subjects filled out a shortened (32 item) version of the Profile of Mood States (POMS; McNair et al., 1971), consisting of five subscales: depression, anger, fatigue, tension, and vigor. Each subscale contains a number of different emotions and moods (for example: lonely, sad, depressed) and the subject had to indicate on a 5-point intensity rating scale how strongly she experienced that particular feeling that day. A mean mood score was calculated for each subscale on each day. Also, a self-constructed questionnaire containing items about interest in own appearance, interest in sex with and without a partner, and need for physical contact (each item rated on a 7-point intensity scale) was administered daily. Descriptive data were collected concerning sexual behavior: On a daily basis subjects indicated whether they had had sexual activity with a partner, who had taken the initiative (self, partner, jointly; percentage score), whether they had masturbated (either when being alone or in the company of a partner), and whether they had had an orgasm (and if so, how many). Further questionnaires about physical symptoms and daily stressors are not described here since the data will be presented elsewhere. Complete confidentiality of all data was assured. Subjects received Hfl. 500 (approx. $250) after participation in the study.

366

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

Hormonal Assays Cortisol, estradiol, and progesterone were assayed in plasma without prior extraction using coated-tube radioimmunoassay (RIA) test kits (EuroDiagnostica B.V., Apeldoorn, The Netherlands). All RIAs were performed in duplicate. Characteristics of the RIA test kits, as provided by the manufacturer, were as follows: Cortisol RIA. Sensitivity, 6 nmol/L; intra-assay coefficient of variation (CV), 2.5-7.9%, interassay CV, 2.6-8.9%; crossreactiviry of cortisol, 100%; prednisolone, 33%; corticosterone, 9%; cortisone, 2.2%; dexamethasone, 0.4%; 11-deoxycorticosteroids (DOC), 3.8%; 17-OH-progesterone, 1.0%; testosterone, 0.14%. Estradiol RIA. Sensitivity, 0.2 nmol/L; intra-assay CV, 6.7-9.7%, interassay CV, 9.0, 9.4%; crossreactivity of estradiol, 100%; estradiol, 0.57%; cortisol, 0.003%; testosterone, 0.001%; 17-p-estradiol, 0.43%. Progesterone RIA. Sensitivity, 0.3 nmol/L; intra-assay CV, 5.6-7.5%, interassay CV, 4.2-5.3%; crossreactivity of progesterone, 100%; corticosterone, 2.46%; estriol, 0.24%; cortisol, 0.13%; 11-DOC, <0.1%; DHEAS, <0.1%; estradiol, <0.1%; testosterone, <0.1%; cortisone, <0.1%; 17-8OH-progesterone, <0.1%. DHEAS and androstenedione were measured by commercial RIA (COAT-A-COUNT, Diagnostic Products Corporation, Los Angeles, CA); SHBG by an immunoradiometric assay (Farmos Diagnostica, Turku, Finland). Total plasma testosterone concentrations were determined by an inhouse radioimmunoassay, without extraction and chromatography and with tritiated testosterone as label (Pratt et al., 1975). The FTI is calculated as the level of testosterone divided by the level of SHBG times 100 (Sherwin, 1988). The estradiol-progesterone (EP) ratio is calculated as the level of estradiol divided by the level of progesterone times 100.

Statistical Analysis of Endocrine and Psychological Data Since women were not studied in one complete menstrual cycle but across two cycles starting at different times in the cycle, hormonal data for each subject were transformed into a standardized menstrual cycle of 29 days with the first day of menstruation being Day 0. For all subjects 13 samples were used, which means that for those subjects with a shorter cycle one sample was used twice, whereas for those with a longer cycle one sample was omitted. This was done by taking note of all the individual estradiol and progesterone values of the subject and the report in the diary of the onset of menstrual bleeding.

Hormones, Sexuality, and Moods in Women

367

Although menstrual cycle data are sometimes analyzed using timeseries analysis, Cook and Campbell (1979) argue that this is an inappropriate technique under such circumstances.8 Accordingly, the present data were analyzed using repeated measures analysis of variance. Two kinds of statistical analyses were performed: One repeated measures ANOVA was done over 13 samples; for the second analysis hormonal data were used to divide each individual cycle into four endocrinologically defined phases: preovulatory, ovulatory, midluteal, and premenstrual. For each woman this phase division was based on her individual hormonal values over 13 samples. The preovulatory phase encompassed all samples with low estradiol and progesterone values from the onset of menstrual bleeding onwards (4-5 samples). The ovulatory phase contained the one or two samples centered around a sharp midcycle estradiol peak while progesterone values remained low. The midluteal phase included a second estradiol peak (often a moreor-less sustained elevation) and a rise in progesterone (3-4 samples). Finally, the premenstrual phase was defined by the remaining 3-4 samples before the first day of menses. Psychological data were matched to the hormonal data, such that diary data were averaged across the days covering one blood sample (2 or 3 days). A second analysis was conducted by dividing the menstrual cycle data of each woman into the described hormonally defined four phases. This enabled intracycle comparisons of behavioral data as a function of cycle phase. The descriptive data concerning sexual behavior were analyzed in a similar manner using a scale with four phases, which were determined for each subject based on her estradiol and progesterone values: (i) preovulatory phase, (ii) ovulatory phase, (iii) midluteal phase, and (iv) premenstrual phase.

RESULTS Descriptive Data Concerning Sexual Behavior Figure 1 shows the frequency of different sexual activities (intercourse and masturbation) for the whole sample of 19 women. (The data of two 8Cook

and Campbell (1979) say in their chapter on time-series analysis: "Specifically, it is difficult to use ARIMA modeling procedures with fewer than SO to 100 observations. Therefore, if the number of time-series observations is small and the errors are independent, repeated measures analysis of variance may be most suitable for evaluating the significance of an intervention effect" (p. 235).

368

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

Fig. 1. Total number of occasions of intercourse and masturbation (n = 19 women). Four phases: preovulatory (Day 1-9), ovulatory (Day 10-14), midluteal (Day 15-24), and premenstrual (Day 25-29). The frequencies of the two longer phases were divided to match the length of the shorter phases.

subjects were excluded from analysis because of an insufficiently clear lateluteal phase. For further details, see section on endocrine data.) There is clearly less sexual activity (intercourse and masturbation) in the first third of the menstrual cycle, after which there is a steep increase around the ovulatory period (Day 10-14). The frequency of these sexual activities remains at a higher level until the first day of menstrual bleeding. Overall, there seems to be a higher frequency of masturbation than of intercourse, with the exception of the preovulatory period, however this difference in frequencies was not significant, x2(3) = 0.90, ns. Figure 2 shows the relative sexual initiative score (in percentages) across the menstrual cycle. Apparently, after menstrual bleeding, in the ovulatory period, there is a tendency for the woman to take more initiative in sexual activity, whereas in the second half of the cycle (between ovulation and menstruation) either the partner takes the initiative or it is taken jointly.

Hormones, Sexuality, and Moods in Women

369

Fig. 2. Sexual initiative across menstrual cycle. Foumphases: preovulatory (Day 1-9), ovulatory (Day 10-14), midluteal (Day 15-24) and premenstrual (Day 25-29).

Endocrine Data Hormonal data of two subjects were excluded from analyses because of an insufficiently clear mid- to late-luteal phase, involving very low progesterone (<19.0 nmol/L) values. The low progesterone levels of the two women did not seem to have affected them because the diaries of the two women and other personally reported information did not indicate any unusual or aberrant data. All of the remaining 19 cycles (8 NPC and 11 PC) were normal. For each subject 13 samples were used. Variations in mean endocrine values across the menstrual cycle were analyzed with repeated measures ANOVAs. There were highly significant differences over the cycle in estradiol, F(12, 204) = 6.04, p < 0.01; progesterone, F(12, 204) = 48.14, p < 0.01; the EP ratio, F(12, 204) = 8.69, p < 0.01; testosterone, F(12, 204) = 6.37, p < 0.01; SHBG, F(12, 204) = 2.45, p < 0.01; and the FTI, F(12, 204) = 5.08, p < 0.01. Naturally, there were elevated, estradiol levels around the time of ovulation, and higher estradiol and progesterone levels in the period between ovulation and menstruation. The established differences in testosterone and FTI across the menstrual cycle, with peak values half-way through the cycle and correspondingly lower values in SHBG around that time, indicate that the period around ovulation is the time of increased

370

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

testosterone and unbound, biologically active, testosterone. A marginally significant sample effect for androstenedione was noted, F(12, 204) = 1.71, p < 0.10. There were significant group differences between the NPC and PC groups in the levels of estradiol, F(l, 17) = 13.45, p < 0.01, and in the EP ratio, F(l, 17) = 8. 77, p < 0.01, with the NPC group showing evidence of higher estradiol levels in practically all samples. In all 13 samples the EP ratio was lower in the PC group. No group differences between the NPC and PC groups were noted in any of the other assayed hormones. There were significant interactions between group and sample for cortisol, F(12, 204) = 1.77, p < 0.05, and SHBG, F(12, 204) = 1.92, p < 0.05, indicating that the NPC and PC groups had a different pattern at different samples of the cycle. For cortisol the samples were quite similar throughout the cycle. However, when comparing the two groups, two phases in the cycle stood out: (i) There was a large difference just before the onset of menstruation; and (ii) there were smaller differences on samples 3, 4, and 6, which could be related to the period of ovulation. The interaction involving SHBG reflects the fact that although SHBG levels tended to be higher in the PC group in all samples, this difference was noticeably larger between shortly before and during menstruation. No changes in DHEAS (umol/L) throughout the cycle were found in this analysis, all levels remaining remarkably constant. The results of a classification of the 13 samples into four hormonally defined menstrual phases (preovulatory, ovulatory, midluteal, and premenstrual) are presented in Fig. 3. The results of these analyses underline the already established differences between the NPC and PC groups on the one hand: estradiol (F = 21.3, p < 0.01; EP ratio (F = 5.73, p < 0.05) and the hormonal changes in the different phases on the other: estradiol, (F = 34.6, p < 0.01); progesterone (F = 40.0, p < 0.01); EP ratio (F = 29.9, p < 0.01); testosterone (F = 11.4, p < 0.01); SHBG (F = 10.3, p < 0.01); FTI (F = 9.8, p < 0.01). However, the significant interactions between group and phase of the cycle for estradiol (F = 13.4, p < 0.01) and the EP ratio (F = 8.12, p < 0.01) indicate that the NPC and PC groups had different patterns at different cycle phases, with the NPC group showing a noticeable ovulatory peak pattern in estradiol, whereas the PC group had a lower, flatter profile. When t tests for unpaired groups were used to test the significance of the differences between the NPC and PC at different phases of the cycle (twotailed tests), it was found that the NPC group had higher estradiol values than the PC group in all phases (preovulatory phase: p = 0.11, ovulatory phase: p < 0.0001, midluteal phase: p = 0.10, and premenstrual phase: p = 0.02). In addition, a higher EP ratio was established for the NPC group

Hormones, Sexuality, and Moods in Women

371

Fig. 3. Hormonal data of the 13 samples were used to divide each individual cycle into 4 endocrinologically defined phases: preovulatory, ovulatory, midluteal, and premenstrual (see Method section).

372

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

than the PC group in the ovulatory phase (p = 0.005) and premenstrual phase (p = 0.07). For testosterone, androstenedione, cortisol, DHEAS, SHBG, and FT1, there were no differences between the two groups in the levels of these hormones across different phases of the menstrual cycle. Psychological Data The diary data of the two subjects omitted from the analyses of the hormonal data were also omitted from the present analyses. For the remaining 19 subjects mean diary data corresponding to the blood sampling day (i.e., mean diary values calculated over 2 or 3 days) were entered, thus providing 13 data points per individual cycle. Variations in mean values across the menstrual cycle were analyzed by means of repeated measures ANOVAs. There were significant sample differences on "tension," F(12, 20) = 2.48, p < 0.01; and "interest in own appearance," F(12, 204) = 2.44, p < 0.01; with higher values on both variables in the period preceding the onset of menstrual bleeding. There were significant group differences in "fatigue," F(l, 17) = 7.75, p < 0.01; "the need for physical contact," F(l, 17) = 6.51, p < 0.05; and "interest in own appearance," F(l, 17) = 3.14, p < 0.10, with the PC group scoring higher on all three variables. No differences were found on the other variables and nor were there any significant interactions. Next, the 13 samples were classified into four hormonally defined menstrual phases: preovulatory, ovulatory, midluteal, and premenstrual. The results of analyses using this classification are shown in Fig. 4. With respect to emotional well-being, there was a significant phase effect for tension only (F = 3.25, p < 0.05). Although the PC group reported stronger feelings of tension during premenstrual phase and NPC group reported more tension in the ovulatory phase, neither of these differences attained statistical significance when analyzed by t tests. There were significant differences between the two groups in fatigue (the PC group reporting significantly more fatigue in all four phases; F = 7.99, p < 0.01), need for physical contact (the PC group having a stronger need for physical contact in all four phases; F = 12.56, p < 0.01), and interest in own appearance (the PC group reporting a significantly stronger interest in own appearance in the ovulatory and midluteal phases; F = 4.64, p < 0.05). The absence of interaction effects indicates that the differences between the two groups were constant across the menstrual cycle. Significant phase and interaction effects were established for "sexual interest" (phase: F = 3.22, p < 0.05; phase by group: F = 2.36, p < 0.10) and "interest in sex with partner" (phase: F = 4.65, p < 0.01; phase by group: F = 2.77, p < 0.05). The trend in both items is for PC women to report stronger

Hormones, Sexuality, and Moods in Women

373

Fig. 4. Diary data were matched to the hormonal data and mean diary data belonging to 4 endocrinologically defined phases were entered: preovulatory, ovulatory, midluteal, and premenstrual (see Method section).

374

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

sexual interest in the period around ovulation and for NPC women to report a peak in the premenstrual phase of the cycle. When t tests were conducted it was found that on both items the PC group had a higher score than the NPC group in the ovulatory phase (p < 0.05), whereas the difference in the premenstrual phase was not significant. Correlations between androgen levels and parameters of sexual behavior (intercourse, masturbation, and orgasm frequency, and the average "sexual interest" score over the menstrual cycle) were calculated. For androstenedione, testosterone, DHEAS, and FTI two indicators were used: hormone level at ovulation, and average hormone level across the cycle. The resulting correlations are shown in Table I. Whether one takes the hormone level at ovulation or the average hormone level across the cycle makes little difference to the correlations. In both cases, parameters of sexual behavior correlated with androgenic hormones: Intercourse frequency was correlated significantly with FTI, masturbation frequency correlated significantly with the other three androgens: androstenedione (A), testosterone (T) and DHEAS. Orgasm frequency was not significantly related to the levels of any of these androgens, although the direction of the relationship was consistently positive. The different androgens had a more-or-less equally strong relationship with the tonic level of sexual interest, but only the correlation with FTI was statistically significant. Correlations between these parameters of sexual behavior and mean estradiol and progesterone values were all small and nonsignificant. We have no explanation why there are different correlations between FTI, T, and A, just as the present state of the art means that it is unclear why some parameters of sexuality relate to androgens in different ways. Finally, since the NPC and PC groups differed in sexual interest over the cycle, correlations between mean T and FTI levels per sample, and mean sexual interest per sample were calculated separately for the NPC and PC groups. There was a strong correspondence within the PC group between androgen levels and sexual interest (r = .50 between T and mean sexual interest, r = 0.69 between FTI and mean sexual interest). In the NPC group, on the other hand, there were negative correlations between androgens and sexual interest, a result related to the fact that their sexual interest tended to peak in the premenstrual phase whereas their peak androgen levels were around the time of ovulation (r = -.35 between T and sexual interest; r = -.49 between FTI and sexual interest). DISCUSSION The aim of the study was to investigate the levels of ovarian and adrenocortical hormones throughout the menstrual cycle, and to relate

Hormones, Sexuality, and Moods in Women

375

376

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

changes in hormone levels to various measures of mood and emotionality, including those of sexuality. To summarize the most clear-cut findings, the study confirms the important role of androgens in women's sexuality, particularly sexual interest. The evidence of mood-related changes across the cycle was less clear, despite the fact that a subgroup had been composed consisting of women who had reported to suffer from diverse premenstrual complaints returning every cycle. Hormonal Changes and the Menstrual Cycle There were large differences in estradiol and progesterone levels across the menstrual cycle, following the normal cyclic pattern. Since circulating testosterone mainly arises from the peripheral conversion of androstenedione (about 60%) it has been claimed that the timing of the testosterone rise is similar to that of androstenedione but that the rise is smaller (Judd and Yen, 1973; Ribeiro et al., 1974). In contrast to these results, we established a significant phase effect for testosterone—with a clear peak in testosterone around the time of ovulation—but not for androstenedione. Moreover, the fact that correlation coefficients between androstenedione and various sexual parameters did not stand out, compared to those of the other androgens, suggests that this hormone does not have much psychological action in its own right. The part of testosterone that is unbound is able to exert its androgenic effects on body and behavior. We found not only a peak in testosterone in the ovulatory phase but also one in FTI, as well as a corresponding decrease in SHBG, thus providing evidence for the claim that the bioavailability of testosterone is increased during the time of ovulation. As yet, not much is known about the levels of adrenocortical hormones across the menstrual cycle. In the present study no evidence of significant variations in cortisol or DHEAS was found across different phases of the menstrual cycle for the group as a whole. Mood and the Menstrual Cycle Evidence for the existence of menstrual cycle-related changes in mood has been inconsistent. Despite the fact that we found a clear temporal pattern in the levels of estrogen, progesterone, and androgenic hormones, unlike others (Backstrom et al., 1983; Sanders et al., 1983) we did not find a corresponding temporal pattern in emotional changes within our sample, apart from a cyclic effect on tension. When calculating an overall adverse moods score, based on the sumscore of the subscales depression, tension,

Hormones, Sexuality, and Moods in Women

377

anger, and fatigue, again no cycle-related changes were found (F = 1.75, p = 0.17), whereas there was a marginally significant group effect (F = 3.11, p = 0.10) with the PC group having higher adverse moods scores across all four phases. These results are in line with other prospective studies in which a low incidence of premenstrual complaints has usually been reported (McFarland et al., 1989). Our study therefore provides no support for the hypothesis that in normal healthy women there is a direct link between absolute estradiol and/or progesterone levels, or menstrual cycle phase on the one hand, and emotional states on the other. Sexuality and the Menstrual Cycle From the present study and many others comes evidence of a cyclic pattern in sexuality. This pattern is not similar to the typical estrous pattern of other mammals, nor is it easy to explain in hormonal terms. The presence of an ovulatory peak in intercourse and masturbation might imply that the effect is mediated by fluctuations in hormone levels associated with ovulation (estradiol), but it might just as well be related to the established peak in testosterone and FTI at midcycle. The peak in frequency of selfinitiated intercourse in the late-menstrual phase, before ovulation, argues against an estrogen explanation. According to Sanders and Bancroft (1982) the amount of estradiol required for normal female sexuality is likely to be low and fluctuations in estradiol above that level are probably irrelevant. In their view, the role of estradiol in women is comparable to that of testosterone in men. On the basis of our data and those of others (Abplanalp et al., 1979; Persky et al., 1978), we conclude that estradiol levels cannot account for the changing patterns in female sexuality found in different studies. As to the role of progesterone, our results and those of others (Schreiner-Engel et al., 1981) indicate that it is even more ambiguous. Our study provides clear support for the role of androgens in various aspects of sexuality. According to Sherwin (1988) androgens have a specific influence on motivational or "preceptive" aspects of female sexual behavior, similar to that found in animals. Her conclusion was based on the fact that testosterone (peak and/or average levels) was associated with the mean frequency of sexual desire, sexual interest, and autoerotic behaviors throughout the cycle, whereas it was not related to the frequency of either coitus or orgasmic response. Our results differ superficially from those of Sherwin, in that the FTI correlated most strongly with sexual interest, intercourse frequency, and orgasm frequency, whereas testosterone, androstenedione, and DHEAS correlated more strongly with masturbation frequency. However, the correlation between intercourse frequency and

378

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

sexual interest (.47) was significant, whereas that between masturbation frequency and sexual interest (0.31) was not. We therefore agree with Sherwin that the level of biologically active testosterone and different, motivationally influenced, parameters of sexual behavior are distinctly related. Specifically, the mean frequency of sexual interest and sexual behavior (intercourse and masturbation frequency) throughout the cycle are associated with the midcycle androgen peak and/or average level of androgens throughout the cycle. At present, the timing of the relationship between hormone levels and aspects of sexual interest and behavior is unclear. According to Sherwin (1988) there is no reason to believe that an increase in the level of a hormone (like the midcycle testosterone peak) causes an immediate behavioral response. It is very likely that the increase in sexual interest and/or behavior observed shortly before the onset of menstruation is due to the midcycle testosterone peak and that it takes several days to a week for behavioral changes to become noticeable. Future studies on different groups of endocrinological patients who are treated with hormones (e.g., hypogonadal men, transsexuals) should aim to combine frequent hormonal sampling and behavioral measurements to elucidate this temporal relationship. In summary, the present study demonstrates the importance of different androgens to sexual interest and behavior in women, a role clearly more important than the one played by either estrogens or progesterone. Differences Between the NPC and PC Groups Our sample of subjects consisted of normal women differing in their perception of the intensity of emotional complaints during the premenstrual phase, whereas other researchers have often used women suffering from premenstrual complaints or premenstrual syndrome who meet certain specific criteria. At present, it is unclear whether the results from clinical samples can be generalized to normal populations of women, or, stated differently, whether the difference between these populations is a quantitative or a qualitative one (Bancroft et al., 1993; Brooks-Gunn, 1986). The hormonal status of the premenstrual syndrome is unclear: PMS has been related to estradiol, progesterone, and/or an aberrant EP ratio. The picture is complicated further by the fact that in some studies (e.g., Reame et al., 1992) no differences in progesterone and estradiol were found between severe PMS patients and normal controls. We found clear differences between the NPC and PC groups with respect to estradiol, but these were opposite to the endocrinological pattern that has been suggested in PMS, namely, an excess of estradiol (Hammerback et al., 1989). Moreover, the interindividual differences between our two groups were consistent and

Hormones, Sexuality, and Moods in Women

379

not limited to the premenstrual phase. The picture for progesterone was that of the NPC group having generally higher levels, although these differences only reached significance in the preovulatory and ovulatory phases (p < 0.05). Some studies (see Clare, 1985) have posited a role for the EP ratio in the etiology of PMS, with an excess of estrogen and a relative lack of progesterone being the important factors. In the present study the EP ratio was generally higher in the NPC group, with the exception of the midluteal phase; only for the ovulatory phase was the difference significant. The significant group and phase effects, as well as the highly significant interaction, make it difficult to interpret these results. It is clear that future studies, using larger groups of subjects differing in the perception of the intensity of their menstrual complaints, should focus not only on levels of steroid hormones, but, if possible, also on differences in steroid receptors. It could well be that differences in steroid receptors are more responsible for menstrual cycle-related problems than peripheral steroid levels. By taking the role of the biological substrate into account, one might start to resolve the intricate relationships between abnormal steroid levels and premenstrual tension. No meaningful differences were found between the NPC and PC groups in androstenedione, testosterone, FTI, and DHEAS. Eriksson et al. (1992) demonstrated significantly higher free testosterone levels across all phases of the cycle for a group of PMS patients as compared to a normal control group, whereas there were no differences in the serum levels of progesterone, total testosterone, androstenedione, DHEAS, SHBG, and the total testosterone-SHBG ratio. They concluded that androgens are possibly involved in the pathophysiology of premenstrual irritability and dysphoria. Our finding of no differences between the NPC and PC groups in androgen levels is probably due to the fact that our sample did not contain women suffering from severe premenstrual irritability and that we did not find cycle-related changes in irritable mood. Although the levels of cortisol were consistent across the menstrual phases for the sample as a whole, an interesting interaction effect was established, F(12, 204) = 1.77. p = 0.05, with the PC group having higher cortisol levels shortly before the onset of menstrual bleeding (in the premenstrual phase), as well as slightly higher levels around the period of ovulation. As yet, it is uncertain how these results should be interpreted. Clearly, different levels of cortisol over the cycle may reflect phasic group differences in neuroendocrinological as well as psychological phenomena. More data are needed before PMS symptomatology can be linked to cycle-related changes in baseline cortisol levels. The data concerning emotions and moods clearly indicate that the conclusion that the menstrual cycle has no strong effect on positive and negative

380

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

moods as reported on a day-to-day basis also applies to those women who before the study reported suffering from an increase in negative mood during the premenstrual phase. Other studies (for example, Trunnell et al., 1988) have found that women with PMS had significantly higher levels of depression than women without PMS during the luteal phase, whereas there were no differences between the groups during the follicular phase. Again, the most likely explanation for the different outcomes is that our PC women suffer less severely from premenstrual symptoms or complaints than the clinical PMS women who have been involved in other studies (Sanders et al., 1983). With respect to the day-to-day report of sexual interest a significant interaction effect between phase and group was found, with PC women reporting a peak in the ovulatory phase and NPC women reporting a peak in the premenstrual phase. It may be that in PC women the awareness of the premenstrual phase and the approaching menstruation had a negative, inhibiting effect on their sexual interest, whereas in women without complaints there was no such effect. It is unclear why there was a difference in sexual interest between the two groups at the time of ovulation. The fact that no differences were established between the two groups in positive and/or negative mood across the cycle implies that these findings are specifically sexual, and not caused by changing emotion levels. The two groups differed with respect to the phase of the cycle at which they reported peak sexual interest. This result is somewhat similar to a study reported by Bancroft et al. (1983), in which a peak in sexual activity and sexual feelings for the PMS group was found in the midfollicular phase, whereas the no-PMS group showed stronger sexual feelings in the late luteal phase. Moreover, there was a clear relationship between levels of androgens (testosterone and biologically active testosterone) and sexual interest within the PC group, whereas these variables tended to be inversely related in the NPC group. The present study provides evidence of a difference between women with and without reports of premenstrual emotional complaints. However, these differences were not found on the day-to-day measures of mood or emotion; rather, they were found on the endocrinological measures and the measures of sexuality. This raises the possibility that the "emotional complaints" reported by women in the PC group summarize subjective perceptions of endocrinological activity rather than actual differences in emotion. ACKNOWLEDGMENTS

The authors thank Merel Kindt for her assistance during the course of this 6 months study and Martine Janzen for her contribution to hormonal analysis.

Hormones, Sexuality, and Moods in Women

381

REFERENCES Abplanalp, J. M, Livingston, L., Rose, R. M., and Sandwisch, D. (1977). Cortisol and growth hormone responses to psychological stress during the menstrual cycle. Psychosom. Med. 39: 158-177. Abplanalp, J. M., Rose, R. M., Donnelly, A.F., and Livingston-Vaughan, L. (1979). Psychoendocrinology of the menstrual cycle: II. The relationship between enjoyment of activities, moods, and reproductive hormones. Psychosom. Med. 41: 605-615. Adams, D. M., Gold, A. R., and Burt, A. D. (1978). Rise in female-initiated sexual activity at ovulation and its suppression by oral contraceptives. New Eng. J. Med. 299: 1145-1150. Alexander, G. M., and Sherwin, B. B. (1993). Sex steroids, sexual behavior, and selection attention for erotic stimuli in women using oral contraceptives. Psychoneuroendocrinology 18: 91-102. Backstrom, T., Sanders, D., Leask, R., Davidson, D., Warner, P., and Bancroft, J. (1983). Mood, sexuality, hormones, and the menstrual cycle. II. Hormone levels and their relationship to the premenstrual syndrome. Psychosom. Med. 45: 503-507. Bancroft, J. (1984). Hormones and human sexual behavior. J. Sex and Marital Ther. 10: 3-21. Bancroft, J., Davidson, D. W., Warner, P., and Tyrer, G. (1980). Androgens and sexual behavior in women using oral contraceptives. Clin. Endocrinol. 12: 327-340. Bancroft, J., Sanders, D., Davidson. D., and Warner, P. (1983). Mood, sexuality, hormones, and the menstrual cycle. III. Sexuality and the role of androgens. Psychosom. Med. 45: 509-516. Bancroft, J., Williamson, L., Warner P, Rennie, D., and Smith, S. K. (1993). Perimenstrual complaints in women complaining of PMS, menorrhagia, and dysmenorrhea: Toward a dismantling of the premenstrual syndrome. Psychosom. Med. 55: 133-145. Beach, F. A. (1976). Sexual attractivity, proceptivity, and receptivity in female mammals. Horm. Behav. 70: 105-138. Beck, L. E., Gevirtz, R., and Mortola, J. F. (1990). The predictive role of psychosocial stress on symptom severity in premenstrual syndrome. Psychosom. Med. 52: 536-543. Brooks-Gunn, J. (1986). Differentiating premenstrual symptoms and syndromes. Psychosom. Med. 48: 385-387. Clare, A. W. (1985). Hormones, behaviour and the menstrual cycle. J. Psychosom. Res. 29: 225-233. Cook, T. D., and Campbell, D. T. (1979). Quasi-Experimentation: Design & Analysis Issues for Field Settings, Rand McNally, Chicago. Dabbs, J. M., and Mohammed, S. (1992). Male and female salivary testosterone concentrations before and after sexual activity. Physiol. Behav. 52: 195-197. de Jonge, F. H., Eerland, E. M. J., and Van de Poll, N. E. (1986). The influence of estrogen, testosterone and progesterone on partner preference, receptivity and proceptivity. Physiol. Behav. 37: 885-891. de Jonge, F. H., and Van de Poll, N. E. (1984). Relationships between sexual and aggressive behavior in male and female rats. In de Vries, G. J., de Bruin, J. P. C, Uylings, H. B. M., and Corner, M. A. (eds.), Sex Differences in the Brain—Progress in Brain Research, Vol. 61, Elsevier, Amsterdam, pp. 283-302. Eriksson, E., Sundblad, C, Lisjo, P., Modigh, K., and Andersch, B. (1992). Serum levels of androgens are higher in women with premenstrual irritability and dysphoria than in controls. Psychoneuroendocrinology 17: 195-204. Greenspan, F. S. (1991). Basic Clinical Endocrinology, 3rd ed., Prentice-Hall, Englewood Cliffs, NJ. Hamilton, J. A., Parry, B. L., and Blumenthal, S. J. (1988). The menstrual cycle in context I: Affective syndromes associated with reproductive hormonal changes. J. Clin. Psychiat. 49: 474-480. Hammerback, S., Damber, J. E., and Backstrom, T. (1989). Relationship between symptom severity and hormone changes in women with premenstrual syndrome. J. Clin. Endocrinol. Metab. 68: 125-130. Janowsky, D. S., and Rausch, J. (1985). Biochemical hypotheses of premenstrual tension syndrome. Psychol. Med. 15: 3-8.

382

Van Goozen, Wiegant, Endert, Helmond, and Van de Poll

Judd, H. L, and Yen, S. S. C. (1973). Serum androstenedione and testosterone levels during the menstrual cycle. J. Clin. Endocrinol. Metab. 36: 475-481. Matteo, S., and Rissman, E. F. (1984). Increased sexual activity during the midcycle portion of the human menstrual cycle. Horm. Behav. 18: 249-255. McFarland, C, Ross, M., and DeCourville, N. (1989). Women's theories of menstruation and biases in recall of menstrual symptoms. J. Personal. Social Psychol. 57: 522-531. McNair, D. M., Lorr, M., and Droppleman, L. F. (1971). Manual for the Profile of Mood States, Educational and Industrial Testing Service, San Diego, CA. Morris, N. M., Udry, J. R., Khan-Dawood, F., and Dawood, M. V. (1987). Marital sex frequency and midcycle female testosterone. Arch. Sex. Behav. 16: 27-37. Parlee, M. B. (1982). Changes in moods and activation levels during the menstrual cycle in experimentally naive subjects. Psychol. Women Quart. 7: 119-131. Persky, H., Charney, N., Lief, H. I., O'Brien, C. P., Miller, W. R., and Strauss, D. (1978). The relationship of plasma estradiol level to sexual behavior in young women. Psychosom. Med. 40: 523-535. Persky, H., Lief, H. I., Strauss, D., Miller, W. R., O'Brien, C. P., and Miller, W. R. (1978). Plasma testosterone level and sexual behavior of couples. Arch. Sex. Behav. 70: 157-173. Pratt, J. J., Wiegman, T., Lapphohn, R. E., and Wolding, M. G. (1975). Estimation of plasma testosterone without extraction and chromatography. Clin. Chem. Acta. 597: 337-346. Reame, N. E., Marshall, J. C, and Kelch, R. P. (1992). Pulsatile LH secretion in women with premenstrual syndrome (PMS): Evidence for normal neuroregulation of the menstrual cycle. Psychoneuroendocrinology 17: 205-213. Ribeiro, W. O., Mishell, D. R., and Thorneycroft, I. H. (1974). Comparison of the patterns of androstenedione, progesterone, and estradiol during the human menstrual cycle. Am. J. Obstet. Gynecol. 119: 1026-1032. Sanders, D., and Bancroft, J. (1982). Hormones and the sexuality of women—The menstrual cycle. Clin. Endocrinol Metab. 11: 639-659. Sanders, D., Warner, P., Backstrom, T., and Bancroft, J. (1983). Mood, sexuality, hormones, and the menstrual cycle. I. Changes in mood and physical state: Description of subjects and method. Psychosom. Med. 45: 487-501. Saxena, B. N., Dusitsin, N., and Lazarus, L. (1974). Human growth hormone (HGH), thyroid stimulating hormone (TSH) and cortisol levels in the serum of menstruating Thai women. J. Obstet. Gynaecol. 81: 563-567. Schreiner-Engel, P., Schiavi, R. C, Smith, H., and White, D. (1981). Sexual arousability and the menstrual cycle. Psychosom, Med. 43: 199-214. Sherwin, B. B. (1988). A comparative analysis of the role of androgen in human male and female sexual behavior: Behavioral specificity, critical thresholds, and sensitivity. Psychobiology 16: 416-425. Sherwin, B. B., Gelfand, M., and Brender, W. (1985). Androgen enhances sexual motivation in females: A prospective cross-over study of sex steroid administration in the surgical menopause. Psychosom. Med. 47: 339-351. Sherwin, B. B., and Gelfand, M. M. (1987). The role of androgen in the maintenance of sexual functioning in oophorectomized women. Psychosom. Med. 49: 397-409. Trunnell, E. P., Turner, C. W., and Keye, W. R. (1988). A comparison of the psychological and hormonal factors in women with and without premenstrual syndrome. J. Abn. Psychol. 97: 429-436. Van de Poll, N. E., and Van Goozen, S. H. M. (1992). Hypothalamic involvement in sexuality and hostility: Comparative psychological aspects. In Swaab, D. F., Hofman, M. A., Mirmiran, M., Ravid, R, and Van Leeuwen. F. W. (eds.), The Human Hypothalamus in Health and Disease—Progress in Brain Research, Vol. 93, Elsevier, Amsterdam, pp. 343-361. Van Goozen, S. H. M., Cohen-Kettenis, P. T., Gooren, L. J. G., Frijda, N. H., and Van de Poll, N. E. (1995). Gender differences in behaviour: Activating effects of cross-sex hormones. Psychoneuroendocrinology 20: 343-363. Van Goozen, S. H. M., Frijda, N. H., Wiegant, V. M., Endert, E., and Van de Poll, N. E. (1996). The premenstrual phase and reactions to aversive events: A study of hormonal influences on emotionality. Psychoneuroendocrinology 21: 479-497.

Archives of Sexual Behavior, Vol. 26, No. 4, 1997

Comparing Gay and Bisexual Men on Sexual Behavior, Condom Use, and Psychosocial Variables Related to HIV/AIDS Joseph P. Stokes, Ph.D.,1.2 Peter Vanable, M.A.,1 and David J. McKirnan, Ph.D.1

Interviews were conducted with 750 men, recruited from a variety of sources in Chicago, who reported sex with men in the past 3 years. Behavioral criteria were used to establish groups of gay and bisexual men. We predicted that gay men, compared to bisexual men, would report more male sexual partners, more experience with receptive sex, and more tolerant attitudes toward homosexuality. The only reliable difference between the gay and bisexual men with respect to number of partners was that gay men were more likely to have had a steady male partner or lover. Gay men were more likely than bisexual men to have engaged in receptive sex, including unprotected receptive anal sex. Bisexual men were more self-homophobic and saw other people as less accepting of same-sex activity. There were no differences between gay and bisexual men in other psychosocial variables. Interventions designed to reduce the transmission of HIV/AIDS need to consider differences in gay and bisexual men's sexual behavior and attitudes toward homosexual behavior. KEY WORDS: bisexual; gay; HIV; AIDS; sexual behavior; condom use.

INTRODUCTION To understand and prevent the transmission of HIV, researchers in the past decade have studied the sexual behavior of men who have sex Support for this research was provided through Cooperative Agreement Number U64/CCU506809-02 with the Centers for Disease Control and Prevention. 1Department of Psychology (mail code 285), University of Illinois at Chicago, 1007 West Harrison Street, Chicago, Illinois 60607-7137. 2To whom correspondence should be sent.

383 0004-0002/97/0800-0383$12.50/0 C 1997 Plenum Publishing Corporation

384

Stokes, Vanable, and McKirnan

with men. Little information, however, is available about the general sexual behavior and attitudes of subgroups within this diverse population. Differences between gay and bisexual men, the focus of the current study, are seldom examined. Gay and bisexual men are often considered as a single group, both in research and in the reporting of epidemiological data about HIV/AIDS. Men who have sex with other men do not fall neatly into gay and bisexual groups, and definitions distinguishing these two groups are, to some degree, arbitrary. Nonetheless, there is a valid distinction to be made using behavioral criteria, since some men have both male and female partners over long periods, whereas others have only male partners. Understanding differences in these groups of men will help interventionists design programs that are optimally effective for reducing the spread of HIV and other sexually transmitted diseases (STDs). Existing studies that contrast gay and bisexual men have relied on sampling procedures that limit the generalizability of results. Typically respondents are from HIV testing sites (e.g., Chetwynd et al., 1992) or STD clinics (e.g., Hernandez et al., 1992), or data come from people with HIV or AIDS (e.g., Diaz et al. 1993). Also, bisexual men are often subsamples of gay cohorts, defined after data are collected; few researchers sample explicitly, as we did, for bisexual behavior. Moreover, the few existing studies were conducted in several different countries; cultural differences increase the difficulty of drawing conclusions from this literature. Studies have suggested some differences in sexual behavior between gay and bisexual men (see Stokes, Taywaditep, et al., 1996, for a more thorough review of this literature). Most studies have shown that bisexual men, compared with homosexual men, have fewer male sexual partners (Diaz et al., 1993; Doll et al., 1992; Hernandez et al., 1992; Lever et al., 1992; Ross et al., 1992; Wood et al., 1993). Exceptions to this pattern exist, usually when the type of partner is restricted to casual partners (Kippax et al., 1994) or when the time frame is limited (Doll et al., 1992). Studies have also suggested that bisexual men are less likely than homosexual men to have engaged in anal intercourse with other men (Hernandez et al., 1992; Lever et al., 1992; Lewis and Watters, 1994; Morse et al., 1991), although two Australian studies do not support this conclusion (Hood et al., 1994; Kippax et al., 1994). One could also speculate that bisexual men, compared to homosexual men, are more likely to take the insertive role (Carrier, 1985), and data from Hernandez et al. (1992) supported this speculation. Previous research has provided few data about differential rates of condom use among bisexual and homosexual men. Some experts have argued that since bisexual men are less likely than homosexual men to have

Comparing Gay and Bisexual Men

385

been reached with HIV prevention information, condom use among bisexual men might be lower. On the other hand, a few studies show trends for bisexual men to have lower rates of unprotected anal sex with men than do homosexual men (Doll et al., 1992; Hood et al., 1994; Ross et al., 1992). A review of 17 studies that reported HIV prevalence rates for homosexual and bisexual men (Stokes, Taywaditep, et al., 1996) showed that all but one of these studies reported higher prevalence among homosexual than bisexual men. One factor contributing to this result may be that homosexual men were less likely than bisexual men to use condoms during anal intercourse with male partners. In summary, the existing literature provides some basis for hypothesizing differences in sexual behavior and condom use in gay and bisexual men. We predicted that gay men, compared to behaviorally bisexual men, would report more male partners, more experience with anal sex, and more experience with the receptive role in anal and oral sex with men—all factors that increase risk for HIV. We had no hypothesis about how rates of condom use would differ between gay and bisexual men. Previous research has not examined differences in gay and bisexual men on psychosocial variables relevant to HIV/AIDS, although such differences might be important to inform interventions aimed at reducing the spread of HIV (Doll and Becker, 1996). We examined differences between gay and bisexual men on self-homophobia, perceived acceptance of samesex sexual behavior by others, and self-disclosure of homosexual activity to others. These variables are important in reaching men for HIV prevention messages; men who are closeted, self-homophobic, and who see others as not accepting of same-sex sexual behavior will not be reached well by messages that emphasize homosexuality. Part of our rationale for studying bisexual men was the prospect that they are less accepting of homosexual activity than are gay men. We predicted that the bisexual men would be more self-homophobic, would perceive others as less accepting of same-sex behavior, and would not have disclosed their homosexual activity as widely as the gay men. Because the African American community has been described as relatively unaccepting of same-sex activity (Mays and Cochran, 1987; Peterson and Marin, 1988; Wright, 1983), we thought African American bisexual men might be especially prone to self-homophobia, perceived lack of acceptance, and low levels of disclosure. Additional psychosocial measures included normative beliefs concerning same-sex behavior, normative beliefs about condom use, perceived vulnerability to AIDS, and attitudes toward condom use. We did not have a priori hypotheses about how gay and bisexual men might differ on these variables.

386

Stokes, Vanable, and McKirnan

Sampling problems are inherent in studying gay and bisexual men, in part because the undertying population is not identifiable. Although we do not claim that our sample is representative of the underlying population—it is a sample of convenience—we did recruit from a variety of sources in the community at large, and our sample may be more representative than other samples in this type of literature.

METHOD Respondents The initial pool of respondents comprised 750 18 to 30-year-old African American and white men from Chicago who had engaged in sex with other men. From this group we selected two groups according to behavioral criteria. The bisexual subsample was restricted to men who reported having both male and female sexual partners in the past 6 months. The behavioral criteria for the homosexual subsample were sex with a man in the past 6 months and no sex with a woman in the past 3 years. Although we recognize that there is some fluidity between these two groups and that there is danger in using categorical labels, for convenience we refer to these as the bisexual and gay subsamples. These behavioral designations were largely consistent with respondents' self labels (McKirnan et al., 1995). Procedure To minimize the inherent difficulties in obtaining a representative sample from a hidden, hard-to-find population, we used multiple recruitment sources and strategies. Recruitment sources were: (i) print ads in newspapers widely distributed throughout-Chicago (25.6% of the sample), including one with a predominantly African American readership; (ii) outreach by recruiters into community settings and sexual "pick-up" areas, including coffee shops and cafes, street fairs, and public parks and streets where men look for casual sex partners (31.2%); (iii) outreach by recruiters into gay and "mixed" (gay and heterosexual) bars (23.5%) (The bars tended to be relatively segregated, with African American respondents being recruited from bars whose clientele are predominantly people of color and whites from bars with predominantly white clientele.), and (iv) snowball sampling, referrals from men who had been interviewed (19.8%). The print ad invited men to call the university for further information about a study being conducted at the university with the sponsorship of

Comparing Gay and Bisexual Men

387

the Centers for Disease Control and Prevention. The recruiters in the field described the study as a confidential, anonymous interview, and gave potential respondents a card with information about contacting the university to schedule an appointment. Determination of eligibility for the study was made when potential respondents called our research office at the university. Callers who met the eligibility requirements were given an interview appointment. Trained interviewers conducted individual, face-to-face, structured interviews. The mean interview length was 1 hr, and respondents were paid $25. Respondents were given the option of remaining anonymous, and all interviews were confidential. Measures Demographics. Measures included age, highest education level, income, and employment status. Income for the previous year was measured on a scale ranging from 1 (less than $10,000) to 6 (over $60,000). Employment and educational status were open-ended items that were coded for standard categories. Number of Partners. Respondents estimated the number of male sexual partners they had had in their lives, as well as the number of male partners in the past 6 months in each of three categories: (i) steady partner or lover (steady); (ii) someone they knew well who was not a steady partner (friend); and (iii) casual acquaintance or anonymous partner (casual). Sexual Behavior and Condom Use. Respondents were asked whether they had ever in their lives engaged in the following sexual behaviors with male partners: insertive and receptive oral sex, insertive and receptive anal sex without a condom, and insertive and receptive anal sex with a condom. Respondents were also asked if they had engaged in these behaviors in the past 6 months with the three types of male partners described in the preceding paragraph. Self-Homophobia and Self-Acceptance of Homosexual Behavior. Selfhomophobia, defined as the degree to which individuals hold negative attitudes toward themselves due to their homosexual feelings and behaviors, was measured with a three item scale (a = .87; e.g., "I feel stress or conflict within myself over being attracted to men"). The response scale ranged from 1 (disagree) to 5 (agree). Respondents also indicated their self-acceptance of having sex with men on a single-item scale ranging from 1 (not at all accepting) to 7 (very accepting). Perceived Acceptance of Respondent's Homosexual Behavior by Others. Respondents listed the initials of the six nonrelative friends or associates

388

Stokes, Vanable, and McKirnan

to whom they were closest. For each of these network members they rated how accepting she or he was about "your having sex with a man, or how accepting they would be if they were to find out," using the scale described in the preceding paragraph. Respondents also provided global acceptance ratings for their immediate friends overall, for their family overall, and for the people in the neighborhood where they lived. Thus, four variables tapped perceived acceptance of homosexual behavior by others: the mean network rating, and single-item ratings for friends, family, and neighbors. Disclosure of Homosexual Activity. A respondent's disclosure of homosexual behavior was the mean rating of seven items (a = .82): For each of seven people or groups (e.g., father, mother, siblings, close male friends), the respondent indicated whether he had explicitly disclosed his homosexual behavior or feelings with the person or group (rated 3), whether he thought the person/group suspected in the absence of explicit disclosure (2), or that he thought the person/group was not aware of his homosexual behavior or feelings (1). Normative Beliefs About Same-Sex Sexual Activity and About Condom Use, Respondents were asked to think about "100 men who are about your age and ethnic group" and to estimate how many of that 100 "are bisexually active; that is, at least occasionally have sex with both men and with women." Respondents also estimated of 100 such men "who are having vaginal or anal sex with women, how many always use a condom?" and of 100 men "who are having anal sex with a man, how many always use a condom?" Perceived Vulnerability to HIV/AIDS. Respondents estimated the likelihood that they were "carrying the AIDS virus now" and the likelihood that they would "get the AIDS virus any time in your life." These two items were not asked of the 27 men who reported that they had tested positive for HIV. Attitudes Toward Condom Use. Respondents' attitudes toward using condoms with male sexual partners was measured with a three-item scale (a = .65; e.g. "I insist on using a condom for anal sex with any male partner"), with responses ranging from 1 (disagree) to 5 (agree).

RESULTS Overview of Analyses Because the distributions of number of sexual partners were skewed, with some men reporting a very large number of partners, these variables were transformed to their common logarithm. Group differences in sexual

Comparing Gay and Bisexual Men

389

behaviors and condom use were tested using logistic regression, which allows for regressing a dichotomous criterion variable on either continuous or dichotomous predictor variables. The significance test for the contribution of gay vs. bisexual group membership in the logistic regressions was the Wald statistic, which has a chi-square distribution with 1 degree of freedom (Norusis, 1990). Group differences in criterion variables that were continuous were examined using multiple regression. Both types of regression analyses were conducted hierarchically, with demographic measures (age, ethnicity, income, education, and dummy-coded variables representing employment status) and recruitment source (dummy-coded) entered as the first step. Next, a dichotomous predictor variable for subsample (behaviorally bisexual or homosexual) was entered. Finally, a predictor that carried the interaction of ethnicity and subsample was entered. With two exceptions mentioned below, the interaction term was not a significant predictor of the criterion variable, and was dropped from the model. Respondents The sample included 205 men who met our definition of gay and 310 who met the criteria for bisexual. The mean age for the combined sample was 25.3. The median annual income was about $15,000, with 31% of the sample earning less than $10,000, and 12% more than $30,000. Twentyeight percent had a high school diploma or less education, 46% had some college (including 18% of the sample who were students at the time of the interview), and 26% had a bachelor's degree or more education. Forty-six percent were employed full-time at interview, 17% were employed parttime, 18% were students, and 20% unemployed. There were no differences between the gay and bisexual subsamples on age, income, or employment status. Gay men reported somewhat higher levels of education, x2(3) = 8.70, p < 0.05, and were less likely than bisexual men to have been recruited from snowball techniques and print ads, x2(3) = 30.89, p < 0.001. The bisexual group contained a higher proportion of African American men than the gay group (62.3 vs. 47.4%), x2(l) = 11.36, p < 0.001. Number of Partners Table I shows the means for various measures of number of male sexual partners. Gay men reported more male sexual partners in their lives than bisexual men. Because gay men had been sexually active with male partners for a slightly longer time than bisexual men (Xs = 8.56 vs. 7.94 years), we added tune since first sexual experience with men to the variables

390

Stokes, Vanable, and McKirnan

Table I. Mean Number of Male Sexual Partners by Partner Type and Time Period for Gay and Bisexual Mena No. male sexual partners Total lifetime Total past 6 months Steady partners, past 6 months Friend partners, past 6 months Casual partners, past 6 months

Gay (« = 205)

Bisexual (n = 310)

37.5 4.76

29.5 5.61

2.62c

.89

.60

3.82d

1.18 2.69

1.39 3.79

2.70c

tb ns ns

aReported

means were computed after truncating the variables as follows: total lifetime at 100, total 6 months at SO, steady at 4, friend at 15, and casual at SO. bTests of subsample (gay vs. bisexual) as a predictor of means of log-transformed data, after controlling for age, ethnicity, income, education, employment status, and recruitment source. cp < 0.01. dp < 0.001.

that were controlled. With this variable controlled, gay vs. bisexual group membership was not a significant predictor of the number of lifetime male sexual partners. There was no significant difference between the gay and bisexual men in total number of partners in the past 6 months, although differences did emerge by partner type. Bisexual men reported more casual/anonymous partners, and gay men more steady partners or lovers. Perhaps the gay men have fewer casual/anonymous partners than the bisexual men because the gay men are more likely to be in a steady relationship with a man (39.5% of the gay men and 25.2% of the bisexual men reported being in a relationship with a man), x2(l) = 11.91, p < 0.001. When men in a steady relationship with another man were excluded from analysis, group membership (gay or bisexual) was no longer a statistically significant predictor of number of casual/anonymous partners. Group membership was also eliminated as a significant predictor of casual/anonymous partners when the sample excluded men who had been paid for sex by other men in the past 6 months (8.3% of the gay subsample and 23.8% of the bisexual subsample). Sexual Behavior

Table II shows sexual behavior data for the two subsamples. Gay men were more likely than bisexual men to have engaged in receptive sex, both anal and oral, for both time periods. Rates of unprotected sex are shown

Comparing Gay and Bisexual Men

391

Table II. Percentages of Gay and Bisexual Men Who Have Engaged in Various Sexual Behaviors for Lifetime and Past Six Monthsa Gay Bisexual (n = 205) (n = 310) OR (CI) Wald x2 Lifetime Insertive anal Receptive anal Insertive oral Receptive oral

94.7 86.6 98.6 94.3

Past 6 months Insertive anal Receptive anal Insertive oral Receptive oral

89.4 62.9 98.4 79.9

ns 29.66d ns 12.36d

71.8

77.7

67.0 92.3 87.1

46.5 94.3 75.2

ns 17.11d ns 4.20b

3.92 (2.40--6.40) 3.42 (1.72--6.75)

2.26 (1.54--3.33) 1.73 (1.02-2.91)

aTests

of significance reflect differences in subsamples after controlling for age, ethnicity, income, education, employment status, and recruiting source. bp < 0.05. cp < 0.01. dp < 0.001.

Table III. Percentages of Gay and Bisexual Men Who Have Engaged in Unprotected Sexual Behaviors for Lifetime and Past Six Monthsa Gay Bisexual Wald x2 OR (CI) (n = 205) (n = 310) Lifetime Insertive anal Receptive anal

75.5 68.4

66.0 41.5

5.51b 31.77d

1.69 (1.09-2.61) 3.21 (2.14-4.81)

Past 6 months Insertive anal Receptive anal

26.8 27.3

31.2 18.2

ns 7.45C

1.89 (1.20-3.00)

aTests

of significance reflect differences in subsamples after controlling for age, ethnicity, income, education, employment status, and recruiting source. bp < 0.05. cp < 0.01. dp < 0.001.

in Table III. More men in the gay subsample than in the bisexual subsample reported unprotected receptive anal sex for both time periods and unprotected insertive anal sex for the lifetime time frame. These differences are a function of more men in the gay subsample having engaged in any type of anal sex, protected or unprotected. The differences by subsample were

392

Stokes, Vanable, and McKirnan

eliminated if the sample was restricted to men who had engaged in any type of anal sex during the respective time periods.

Self-Homophobia, Perceived Acceptance, and Disclosure As shown in Table IV, the gay men were less self-homophobic and more self-accepting of their same-sex sexual behavior. Compared to the bisexual men, the gay men perceived the six people named in their social network list, their friends overall, their family, and their neighbors to be more accepting of their homosexual activity. Men in the gay subsample were also more likely to have disclosed their homosexual behavior to other people. The self-homophobia and the disclosure measures were the only variables predicted by the interaction of ethnicity (African American and white) and subsample (gay and bisexual), t(498) = -2.47, p < 0.02 for self-homophobia, and t(498) = 3.70, p < 0.001 Table IV. Gay Versus Bisexual Men on Psychosocial Variablesa Variable

Gay Bisexual (n = 205) (n = 310)

df

t

Self-homophobiab

2.10

2.74

499

4.67g

Acceptance of homosexual behavior byc Respondent Mean of six network members Friends overall Family Neighbors

6.55 6.34 6.07 4.07 4.00

5.69 5.69 4.73 3.30 3.43

499 495 499 499 499

-6.11g -9.13* -7.90g -4.09g -2.09e

Disclosure of homosexual behavior

2.42

1.86

499

-11.05g

33.58 30.17 56.41 13.34 24.04 4.37

42.64 32.90 50.09 14.93 27.25 4.38

499

2.81f

Estimated % men who Are behaviorally bisexual Always use condoms w/women Always use condoms w/men Likelihood of having HIV now (%) a Likelihood of getting HIV in future (%)e Attitudes toward using condoms aTests

ns ns ns ns ns

of significance reflect differences in subsamples after controlling for age, ethnicity, income, education, employment status, and recruiting source. bHigher score is more homophobic. cHigher score is more accepting. dns = 193 for gay subgroup and 294 for bisexual subgroup. ep < 0.05. fp < 0.01. Up < 0.001.

Comparing Gay and Bisexual Men

393

for disclosure. For self-homophobia, the white gay men scored lower than the other three groups; for disclosure, gay men were more disclosing than bisexual men for both ethnic groups, but the difference in disclosure between gay men and bisexual men was greater for the white men than for the African American men. Normative Beliefs, Perceived Vulnerability to HIV/AIDS, and Attitudes Toward Condom Use Bisexual men gave higher estimates of bisexual activity than did gay men (42.6 vs 33.6%), although both groups thought bisexual activity among men their age was prevalent (see Table IV). The two groups did not differ in their estimates of the percentage of men who use condoms consistently with either male or female sexual partners. Neither did the two groups differ on either measure of perceived vulnerability to HIV/AIDS or in their attitudes toward using condoms.

DISCUSSION These data do not support previous reports that gay men have more male partners than bisexual men. In fact, bisexual men had more casual partners in the previous 6 months, although this difference was eliminated when either men in a steady relationship with another man or men who have engaged in sex for pay were eliminated from the analysis. These data suggest the need for interventions to decrease risk for HIV among bisexual men. The one reliable finding with respect to number of partners is that gay men are more likely than bisexual men to be in a steady relationship with a man. This result suggests that interventions to prevent the spread of HIV might place special emphasis on safer sex in the context of a relationship when the target audience is gay men. The lack of differences in number of nonsteady male sexual partners between gay and bisexual men supports the need to intervene with bisexual men about HIV/AIDS. The gay men were more likely than the bisexual men to have engaged in receptive sex, a behavior that carries high risk for HIV transmission. This finding might reflect a "natural" preference that is different between gay men and bisexual men. Perhaps receptive homosexual activity, especially receptive anal sex, tends to occur developmentally later than other homosexual activity. If some of the bisexual men are in transition to a homosexual identity, perhaps their frequency of receptive sex will increase

394

Stokes, Vanable, and McKirnan

over time. If so, preventive interventions for bisexual men should educate them to the risks of receptive sex as soon as possible, ideally before they initiate a pattern of unprotected receptive anal sex. Such early intervention might increase the likelihood that condom use would become habitual. Additional analyses of our data provide some support for the idea that receptive anal sex occurs developmentally later than other sexual behaviors. The correlation of the number of years since first sexual experience with a man and having engaged in receptive anal sex in the past 6 months was small but statistically significant for the bisexual subsample (r = .13, p < 0.02) and for the combined gay and bisexual subsample (r = .13, p < 0.01). The comparable correlations for insertive anal sex were not significant and were significantly different from the correlations for receptive anal sex (rs = -.01). Gay men were more likely to have had unprotected receptive sex. Therefore, the general category of gay men are at a higher absolute risk for exposure to HIV or other STDs. However, this difference was a function of differences in the base rate of the behavior; restricting the samples to men who have experienced receptive anal sex in the relevant time period eliminated the differential rates of unprotected sex between the gay and bisexual men. In other words, we found no evidence that rates of condom use for anal sex are different for the two groups. The gay and bisexual men did not differ in their perceived norms of condom use with either male or female partners, in their attitudes towards using condoms, or in their perceived vulnerability to HIV/AIDS. Thus, attitudes toward core HIV-prevention behaviors were similar in the two groups. There were differences in how normative the two groups estimated bisexual behavior to be, with the bisexual men seeing bisexual behavior as more prevalent among men their age than the gay men did, but both groups saw bisexual activity as very common. Both groups saw condom use with male partners in general as more common than condom use with female partners in general, a perception that data from bisexual men suggest is accurate (McKirnan et al., 1995). The only consistent and significant differences in psychosocial variables were with self-homophobia and perceived acceptance of same-sex behavior by others. This finding suggests that it will be difficult to recruit bisexual men into interventions that target gay men. Perhaps their self-homophobia and fear of others' reaction influences some bisexual men to maintain relationships with women, as opposed to having only male partners. Perhaps the gay men, being more comfortable with their homosexual behavior, have disclosed to more people and have received largely positive reactions, leading them to revise their estimates of others' reactions to their same-sex behavior. Whatever its origin, this group difference may be im-

Comparing Gay and Bisexual Men

395

portant in designing HIV prevention programs that effectively reach bisexual men. Differences between the gay and bisexual subsamples interacted with ethnicity (African American or white) in predicting two variables, self-homophobia and disclosure to others. In each case the difference between the gay and bisexual subsamples were greater for the white men than for the African American men. Put another way, there were minima! ethnic differences in self-homophobia and disclosure for bisexual men, but white gay men were less self-homophobic and more likely to have disclosed their same-sex behavior than were African American gay men. This phenomenon may occur because white gay men are more likely than black gay men to be part of a gay community, which is supportive of their same-sex behavior (Peterson, 1992). The existence of a supportive gay community may facilitate the development of positive, accepting attitudes about one's homosexual behavior. There are, of course, great variability and individual differences among groups of gay men and groups of bisexual men. Various subgroups of behaviorally bisexual men, in particular, may exist (Stokes and Damon, 1995) and might be reached and influenced by different intervention strategies and messages. Some bisexual men may be in transition to a homosexual identity. For these men, interventions need to encourage a positive gay identification and exposure to role models who emphasize avoiding sexual activity, avoiding anal sex, and/or using condoms consistently. Other bisexual men will never identify as gay, and may not even identify as bisexual. These men likely have the relatively high levels of self-homophobia and low levels of perceived acceptance of same-sex activity that characterized the bisexual men in the current study. For them, prevention messages and materials that are not gay-oriented will be most successful (Doll and Beeker, 1996). Ideally, preventive interventions would address perceived and actual lack of acceptance of homosexual behavior and lack of disclosure of samesex activity, including lack of disclosure to female partners (Stokes, McKirnan, et al., 1996). To encourage men to be thoughtful about their sexual behavior, we need to create a more tolerant climate for the diverse desires and experiences of sexual behavior. Increased awareness for the prevalence of bisexual behavior is a first step, which might make it easier for concerned friends and relatives to raise issues of sexuality and to encourage condom use. Increased tolerance and understanding of diverse expressions of sexuality would also facilitate bisexual men's disclosure of their homosexual activity to others, including potential female partners. In summary, these data indicate that gay men, compared to bisexual men, are more likely to have engaged in receptive sex, including unpro-

396

Stokes, Vanable, and McKirnan

tected receptive anal sex. Nonetheless, substantial numbers of bisexual men have engaged in unprotected receptive anal sex, and interventions targeting bisexual men are needed to reduce HIV transmission. Our finding that bisexual men have high levels of self-homophobia and low levels of perceived acceptance of same-sex sexual behavior suggests that gay-oriented interventions or interventions that require men to disclose homosexual activity are unlikely to be successful with bisexual men (Doll and Beeker, 1996). Community level interventions, including media campaigns, are needed to increase awareness and acceptance of bisexual activity, and to encourage men who have sex with both men and women to use condoms consistently.

ACKNOWLEDGMENT

We acknowledge the contributions of Rebecca Burzette, Marshall Davis, and Borris Powell in conducting this research and of Lynda Doll for providing helpful feedback on drafts of this manuscript.

REFERENCES Carrier, J. (1985). Mexican male bisexuality. J. Homosex. 11: 75-87. Chetwynd, J., Chambers, A., and Hughes, A. J. (1992). Condom use in anal intercourse amongst people who identify as homosexual, heterosexual, or bisexual. New Zeal. Med. J. 105: 262-264. Diaz, T., Chu, S. Y., Frederick, M., Hermann, P., Levy, A., Mokotoff, E., Whyte, B., Conti, L., Herr, M., Checko, P. J., Rietmeijer, C. A., Sorvillo, F., and Mukhtar, Q. (1993). Sociodemographics and HIV risk behaviors of bisexual men with AIDS: Results from a multistate interview project. AIDS 7: 1227-1232. Doll, L. S., and Beeker, C (1996). Male bisexual behavior and HIV risk in the United States: Synthesis of research with implications for behavioral interventions. AIDS Educ. Prevent. 8: 205-225. Doll, L. S., Petersen, L. R., White, C. R., Johnson, E., Ward, J. W., and the Blood Donor Study Group. (1992). Homosexually and nonhomosexually identified men who have sex with men: A behavioral comparison. /. Sex Res. 29: 1-14. Hernandez, M., Uribe, P., Gortmaker, C, Avila, L., De Caso, L. E., Mueller, N., and Sepulveda, J. (1992). Sexual behavior and status for Human Immunodeficiency Virus Type 1 among homosexual and bisexual males in Mexico City. Am. J. Epidemol. 135: 883-894. Hood, D., Prestage, G., Crawford, J., Sorrell, T., and O'Reilly, C. (1994). Report on the B.A.N.G.A.R. Project: Bisexual Activity/Non-Gay Attachment Research, National Centre in HIV Epidemiology & Clinical Research, Darlinghurst, Australia. Kippax, S. Crawford, J., Rodden, P., and Benton, K. (1994). Report on Project Male-Call: National Telephone Survey of Men Who Have Sex with Men, Australian Government Publishing Service, Canberra, Australia. Lever, J., Kanouse, D., Rogers, W., Carson, S., and Hertz, R. (1992). Behavior patterns and sexual identity of bisexual males. J. Sex Res. 29: 141-167.

Comparing Gay and Bisexual Men

397

Lewis, D. K., and Watters, J. K. (1994). Sexual behavior and sexual identity in male injection drug users. J. AIDS 7: 190-198. Mays, M., and Cochran, S. D. (1987). Acquired immunodeficiency syndrome and Black Americans: Special psychosocial issues. Public Health Rep. 102: 224-231. McKirnan, D. J., Stokes, J. P., Doll, L., and Burzette, R. G. (1995). Bisexually active men: Social characteristics and sexual behavior. J. Sex Res. 32: 64-75. Morse, E. V., Simon, P. M., Osofsky, H. J., Balson, P. M., and Gaumer, H. R. (1991). The male street prostitute: A vector for transmission of HIV infection into the heterosexual world. Soc. Sci. Med. 32: 535-539. Norusis, M. J. (1990). SPSS/PC+ Advanced Statistics 4.0, SPSS, Chicago. Peterson, J. (1992). Black men and their same-sex desires and behaviors. In Herdt, G. (ed.), The Culture of Gay Men, Sage, Thousand Oaks, CA, pp. 147-164. Peterson, J. L., and Marin, G. (1988). Issues in the prevention of AIDS among black and Hispanic men. Am. Psychol 43: 871-877. Ross, M. W., Wodak, A., and Gold, J. (1992). Sexual behaviour in injecting drug users. J. Psychol. Hum. Sex. 5: 89-104. Stokes, J. P., and Damon, W. (1995). Counseling and psychotherapy for bisexual men. Dir. Clin. Psychol. 5: 1-14. Stokes, J. P., McKirnan, D. J., Doll, L., and Burzette, R. G. (1996). Female partners of bisexual men: What they don't know might hurt them. Psychol. Women Quart. 20: 267-284. Stokes, J., Taywaditep, K., Vanable, P., and McKirnan. D. (1996). Bisexual men, sexual behavior, and HIV/AIDS. In Firestein, B. (ed.), Bisexuality: The Psychology and Politics of an Invisible Minority, Sage, Newbury Park, CA. Wood, R. W., Krueger, L. E., Pearlman, T. C, and Goldbaum, G. (1993). HIV transmission: Women's risk from bisexual men. Am. J. Public Health. 83: 1757-1759. Wright, J. W. (1983). African American male sexual behavior and the risk for HIV infection. Hum. Organ. 52: 421-431.

Archives of Sexual Behavior, Vol. 26, No. 4, 1997

Correlates of Increased Sexual Satisfaction Elina Haavio-Mannila, Ph.D.,1,3 and Osmo Kontula, Ph.D.2

Comparisons of nationally representative survey data of the population ages 18-54 years in 1971 (N = 2252) and 1992 (N = 1718) from Finland show that sexual satisfaction has greatly increased particularly among women. Some predictors of sexual satisfaction of men and women are examined on the basis of the 1992 survey data on people ages 18-74 years (N = 2250). Correlations between social background factors, sexual ideas and assertiveness, optional relationships, sexual practices, orgasm, and satisfaction with sexual intercourse were calculated. To control the simultaneous effect of the variables explaining satisfaction, path analyses were conducted. Results show that young age, a sexually unreserved and a nonreligious childhood home, early start of sexual life, high education, sexual assertiveness, considering sexuality important in life, reciprocal feeling of love, use of sex materials, frequent intercourse, many-sided (versatile) sexual techniques, and frequent orgasm correlate with finding sexual intercourse pleasurable. There were some gender differences in the connections between the independent factors and satisfaction with coitus. The importance of sexuality in life, love, and the use of sexual materials were connected directly to physical sexual satisfaction among men but only indirectly among women. For women, but not for men, young age and early start of sexual life correlated with enjoyment of intercourse. The greater sexual dissatisfaction of women compared to men, which still prevails, may be due to their late start of sexual life, conservative sexual attitudes, unimportance of sexuality in life, lack of sexual assertiveness, and use of restricted sexual techniques. The emancipation of women may change these ideas and practices of women. This might lessen the gender gap in physical sexual satisfaction. KEY WORDS: sexual satisfaction; orgasm; sexual attitudes; love; gender differences.

1Department of 2Departmentof 3To

Sociology, University of Helsinki, Box 18, FIN-00014, Finland. Public Health, University of Helsinki, Finland whom correspondence should be addressed.

399 000«KWOT/080(MB99Jli50/0 C 1997 Plenum Publishing Corporation

400

Haavio-Mannila and Kontula

INTRODUCTION According to Davidson et al. (1995), "a sense of enjoyment or satisfaction with one's sexual life is a highly personal sentiment greatly related to an individual's past sexual experiences, current expectations, and future aspirations" (p. 237). Sexual satisfaction has been shown to be related to the characteristics and behavior of the partner, emotions, sexual behaviors as well as to social background factors. In a review of studies conducted in the U.S. (Sprecher and McKinney, 1993), sexual satisfaction was found to be associated with young age and middle class background. Sexual behaviors connected to sexual satisfaction included frequent intercourse, oralgenital sex, experimental lovemaking, and orgasm. Laumann et al. (1994) found there was more emotional satisfaction and physical pleasure in a monogamous relationship than in sexual intercourse with a primary partner while also having a sexual relation with one or more other partners within the past 12 months. This may be due to having learned what excites and pleases that partner. In a sample of 868 nurses located in 15 states women with many partners expressed the least psychological (not physiological) sexual satisfaction. One explanation may be that women with one partner indicate that their partners often delay orgasm until after their own first orgasm. (Davidson and Darling, 1988). Analyses of the behavioral and attitudinal variables have indicated that college-age women (N = 275) in committed relationships, who use effective contraceptive methods, who are more consistently orgasmic, and who report higher frequencies of sexual intercourse tend to report higher levels of sexual satisfaction (Pinney et al., 1987). However, Greeley (1991) found that frequency of sex did not affect overall relationship quality once sexual satisfaction was controlled. Effective communication has been found to be important to the sexual relationship. Communication can enhance sexual arousal, is necessary for the initiation and refusal of sex, and is related to sexual satisfaction (Sprecher and McKinney, 1993, p. 124). Couples who maintain a high quality of communication about sex are more likely to have a satisfying sexual relationship. Feeling incapable of communicating sexual desires has been found to be a common attribute related to anorgasmia (Huberle, 1991). The belief that such feelings may be caused by a lack of self-esteem were supported by Huberle's findings that showed that sexually assertive women reported higher frequencies of sexual activity and orgasm, rated themselves as having greater subjective sexual desire, and reported greater marital and sexual satisfaction. Orgasm is only one facet of the total sexual experience, and many factors influence both orgastic capacity and sexual satisfaction (Morokoff,

Correlates of Increased Sexual Satisfaction

401

1978). Orgasm nevertheless remains the most easily quantifiable index of sexual satisfaction. American studies indicate that 4-10% of adult women have never experienced orgasm. Inhibited female orgasm is typically cited as the most common sexual difficulty presented to practitioners in clinic settings. Women have been found to be orgasmic only 40-80% of the time, regardless of the stimulation method, (cf. Alzate 1985; Darling and Davidson, 1986; Darling et al., 1991; Spector and Carey, 1990.) In this study, emotional sexual satisfaction is operationalized as the "happiness of the steady relationship." Physical sexual satisfaction was in our questionnaire defined as "pleasurableness of sexual intercourse." Finding sexual intercourse pleasurable was chosen as the main dependent variable because we had found that it varies by course of time and gender (Kontula and Haavio-Mannila, 1995a). Physical sexual satisfaction is also influenced by other social factors and is a relevant object for a sociological study. On the basis of earlier studies we assume that pleasurableness of sexual intercourse is influenced by social background, sexual ideas, emotional relations between the partners, sexual practices, and orgasm. We measure the assumed determinants of sexual satisfaction by using the following predictors (see Appendix for details): Social Background. Age of respondents, sexual and religious atmosphere in childhood homes, educational resources, and the timing of first sexual experiences. Sexual Ideas. Importance of sexual life among different spheres of life and sexual assertiveness, i.e., the subjective perception of being sexually skillful, active, and attractive. Emotional Ties Between the Partners. That there is a person who really loves the respondent and that he or she really loves some woman or man. Sexual Techniques and Practices. Use of sexual materials, frequency of intercourse and the application of many-sided techniques in sexual encounters. The last-mentioned variable includes having used several positions or woman on the top position in the latest intercourse, having engaged in oral and anal sex, and stimulating partner's genitals by hand. Orgasm. How often the intercourse has led to the ending of sexual tension and orgasm. Changes in satisfaction with sexual intercourse over more than 20 years are analyzed by comparing results of two national representative surveys of 18- to 54-year-old Finnish people conducted in 1971 and 1992. On the basis of the latter study (Kontula, and Haavio-Mannila, 1995a), predictors of physical sexual satisfaction will be examined separately for men and women ages 18 to 74 years.

402

Haavio-Mannila and Kontula

METHOD

In 1992, adult sexual behavior was studied in Finland after an interval of 20 years. As one of the most important objectives of the survey was to make comparisons with the 1971 survey (Sievers et al., 1974), the method used was necessarily as close as possible to the one used in 1971. Thus, a two-stage face-to-face interview/self-administered questionnaire survey, mainly implemented in the home of the respondent, was chosen. Two questionnaires were used, one filled out by predominantly female interviewers and the other, by respondents. The interviewer did not see the answers. Questionnaires were enclosed in envelopes for mailing to the research team. One third of questions were the same as asked in the 1971 study. A large number of new questions on sexual life and new topics never before studied in Finland were added. There were 207 questions and 404 variables (Kontula and Haavio-Mannila, 1995a). At the end of 1991 and the beginning of 1992, survey data were collected by face-to-face interviews among a population ages 18 to 74 years. The sample was drawn at random from the central population register and was nationally representative of the population in the age bracket 18-74. The sample was limited in the same manner as that of the 1971 study. The upper age limit of 74 was chosen because those in the age cohort 54 or younger were now 74 or younger. Data collection was conducted by 164 interviewers of Statistics Finland. In 1971, interviews were done by the personnel of the health services (public health nurses and midwives). The average time spent on the interview and on filling out the questionnaire was 78 min, slightly less than in 1971. In 1992, there were 2250 responses to the survey, and the response rate was 76%. A total of 1146 women and 1104 men participated. Response rates were 78% for women and 74% for men. The most responsive group were those less than 25 years (83%) and the least responsive (72%) those ages 35-44 years. Young and elderly people responded better than expected (of those more than 65 years 77% participated in the study). In 1971, 2188 persons ages 18 to 54 years responded to the survey, and the response rate was as high as 91%. Geographically, the best response was obtained in the provinces of Pohjois-Karjala (88%) and Mikkeli (84%) in Eastern Finland, which are relatively rural, nonindustrialized areas. Uusimaa, a highly developed area in Southern Finland, was the worst (70%). On the basis of marital status, widows and widowers were the most responsive (83%), while divorcees (72%) were the least. Unmarried people were slightly more responsive.

Correlates of Increased Sexual Satisfaction

403

The decrease in response rate (i.e., the increase in the share of refusals) from the 1971 study to the 1992 study was due to many factors: people's increased reluctance to respond to surveys; the opportunity of the health care personnel, in 1971, to use their authority (also uniform) in persuading the people; the people's possibility to refuse meeting in the telephone contact in 1992 (in 1971 the people were met without any previous telephone contact); and the open announcement of the research subject (sexual issues), in 1992, in the first letter of contact (contrary to the 1971 survey). The response rate in 1992 did not deviate from the surveys by Statistics Finland of the same time concerning the spending of leisure time and elections. The decline of the response rate did not cause any systematic selfselection of respondents. Responses concerning first sexual experiences were in agreement in the different age cohorts in 1992 and 1971. For example, the 50-year-olds recalled and reported their first sexual experiences in 1992 in the same way as the 30-year-olds in 1971. This could be explained by similar biases in these surveys. However, this finding may be due to other factors. Data were analyzed by calculating percentages and means and testing the statistical significance of differences between groups. To control the simultaneous effect of the independent variables used in predicting sexual satisfaction, path analyses based on stepwise regression analyses were conducted. As many weak links between the variables are not included in the path models, original correlation matrices are presented from which the reader can evaluate results and interpretations.

RESULTS Increase in Sexual Satisfaction in the Last Twenty Years

Perhaps the most positive result of the recent Finnish sex survey is the observation (Table I) that people ages 18 to 54 years (in order to compare with the 1971 data, the older people interviewed in 1992 are omitted from Tables I and II) nowadays report greater satisfaction was reported with their sexual life than 20 years ago. In 1992, sexual intercourse was considered more pleasurable, steady relationship was experienced as happier, and sexual life as a whole was estimated to be more satisfying than was reported in 1971. These results hold true even when the influence of age, education, marital status, type of community, and the importance of religion to one's life were controlled.

404

Haavio-Mannila and Kontula

Correlates of Increased Sexual Satisfaction

405

406

Haavio-Mannila and Kontula

From the point of view of gender equality, in 1992 women considered sexual intercourse more pleasurable than in 1971 and nearly as pleasurable as men did (Table II, Fig. 1). Men still said they enjoyed intercourse in general more than women did but the evaluation of the most recent intercourse was the same for both genders. The gender gap in experiencing sexual intercourse as pleasurable has thus decreased significantly. Emotional satisfaction, i.e., the happiness of the present steady relationship, has also increased in the last 20 years (Table I, Fig. 2). Especially the younger (18-34 years old) women now report more happiness. The gender difference in the happiness of a present steady relationship is not statistically significant: Men and women are as happy. Of subjects ages 18-74 years, of men 29% and of women 33% were very happy and only 2% were unhappy in 1992, the rest being inbetween. The following results are based on the 1992 sample of people 18-74 years old: 42% of the men and 32% of the women thought that intercourse was mostly very pleasurable, and 43 and 39%, respectively, reported that the latest sexual intercourse was a very pleasurable experience. Only 2-3% reported that intercourse is unpleasant. For the rest, intercourse experiences were seen as quite pleasurable or neither pleasurable nor unpleasant. The gender differences in physical sexual satisfaction are statistically significant (p < 0.001). Although women found their latest intercourse pleasurable as often as men did, a difference still exists in the experience of first intercourse: 36% of the men and 10% of the women considered their first intercourse

Fig. 1. Experience intercourse mostly as very pleasant (1971-1992).

Correlates of Increased Sexual Satisfaction

407

fig. 2. Are very happy in their present steady relationship (1971-1992).

very pleasurable and 5 and 33%, respectively, unpleasant. The gender difference is significant (p < 0.001). General sexual satisfaction was the same for both sexes: 26% of men and 29% of women found their sexual life as a whole very satisfying, 58 and 52%, respectively, quite satisfying, 11 and 13% neither satisfying nor unsatisfying, and 4 and 6% unsatisfying. Emotional and physical sexual satisfaction are related. The correlation coefficient (r) between considering one's relationship to one's present steady partner happy and finding sexual intercourse pleasurable is for men .25 (p < 0.001) and for women .29 (p < 0.001). Happiness in steady relationship and satisfaction with sexual intercourse influence the evaluation of one's sexual life as a whole to the same extent. The correlation coefficient between overall sexual satisfaction and satisfaction with sexual intercourse was for men .40 (p < 0.001) and for women .47 (p < 0.001) and that between overall sexual satisfaction and happiness of the steady relationship was, respectively, for men .36 (p < 0.000) and for women .44 (p < 0.001). Emotional and physical sexual satisfaction are thus equally important predictors of overall sexual satisfaction. Even though we concentrate on changes and predictors of physical sexual satisfaction, we do not devalue the emotional aspects of sexuality. Gender Differences in the Assumed Predictors of Sexual Satisfaction Before the causal analysis, we compared men and women with regard to the factors we assumed to be connected with sexual satisfaction. We examined gender differences in social background, sexual ideas, emotional

408

Haavio-Mannila and Kontula

Table III. Means and Statistical Significance of Gender Differences in Assumed Determinants of Sexual Satisfaction in 1992 Men Women Assumed determinants of sexual satisfaction X X Social background Age, years 41.3 43.1. Sexually unreserved home (range 1-5) 2.4 2.4 Nonreligious home (range 1-5) 3.8* 3.5 16.9 Age at starting of sexual life, years 18.0b 16.9 17.4 High (at least 15 years of) education (%) Sexual ideas Liberal sex attitudes (6-item scale, range 6-30) Sexually assertive (3 item scale, range 3-15) Considers sex important in life (range 1-5) Emotional relations between partners Feels and gets love (2-item scale, range 2-6) Sexual techiques and practices Use of sex materials (5-item scale, range 5-20) Frequent sexual intercourse (2-item scale, range 2-16) Many-sided sexual techniques (5-item scale, range 5-23) Orgasm and satisfaction Orgasm (2-item scale, range 2-9) Sexual satisfaction (2-item scale, range 2-10) No. of respondents ap bp

21.2* 10.0b 4.1*

19.7 9.2 3.6

5.1

5.0

8.4* 9.3* 14.7*

5.6 8.2 13.3

8.3* 8.8*

6.5 8.3

1103

1144

< 0.01. < 0.001.

relations, sexual practices, and orgasm. In this way we attain a better comprehension of the gender-specific aspects of sexual satisfaction. If women were found to rank low on some predictors of sexual satisfaction, the result would give us some clues for understanding why they do not find sexual intercourse as pleasurable as men do. All gender differences cited in the following paragraphs are statistically significant. Table III reveals that women in our sample on average are older (43 years) than men (41 years). There is no significant gender difference in the educational level of the interviewees. Neither did the sexual atmosphere in the childhood home differ by gender. Nevertheless, the men recall their home as less religious than the women do. The sexual histories of the women differ from those of the men. The women started their sexual life (kissing, dating, sexual intercourse, and achieving orgasm in intercourse) at a later age than did the men. Thus the social background of the women is, according to their subjective evaluation, somewhat more

Correlates of Increased Sexual Satisfaction

409

traditional and their sexual initiation had taken place at a later age than is the case among the men. Sexual ideas, i.e., attitudes and sexual assertiveness, differ by gender. Women are more conservative in their sexual attitudes, rate sexual life less important among the 13 different life spheres mentioned in the questionnaire, and are sexually less assertive. Thus sexuality seems to have a less important role in the lives of women. Sexual techniques and practices reported by men and women vary considerably. Men use more sexual materials (movies, videos, TV programs, magazines, books and calendars), engage more often in sexual intercourse, use more techniques, for example, nontraditional positions in intercourse, anal sex, oral sex, and stimulation by hand. Finnish women report relatively high rates of orgasm during intercourse, but they still have significantly lower rates than men (Darling et al. 1996). Fifty-three percent of men but only 6% of women reported "always" having had an orgasm during sexual intercourse, and 44 vs. 48%, respectively, have achieved it "almost always or usually." Of the Finnish women, 4% had never experienced an orgasm during sexual intercourse (of the men, 0.3%). During the most recent intercourse as many as 92% of the men but only 56% of the women reported that they had achieved an orgasm. Women had their first orgasm at a considerably later age (on the average 21 years) than men (18 years) even though both genders had started sexual intercourse at the same age (18 years). Pathways to Physical Sexual Satisfaction To study the factors related to physical sexual satisfaction (finding intercourse pleasurable) we conducted path analyses using stepwise regression analysis. In the path models (Figs. 3 and 4) explaining men's and women's physical sexual satisfaction, 13 independent variables are included. They are defined in the Appendix. Correlations between the variables are shown in Table IV so that one can see the connections on which the multitvariate analysis is based. Physical sexual satisfaction of women is more predicted by social factors examined in our study: The path models explain only 21% of men's but as much as 46% of women's satisfaction with sexual intercourse. In most cases, men's and women's sexual satisfaction is explained by the same factors. Satisfaction with sexual intercourse is directly connected to young age (women only), considering sexuality important in life (men only), sexual assertiveness, reciprocal love (men only), using sex materials (men only), frequent sexual intercourse, many techniques in sexual encounters, and or-

410

Haavio-Mannila and Kontula

Correlates of Increased Sexual Satisfaction

411

412

Haavio-Mannila and Kontula

Correlates of Increased Sexual Satisfaction

413

gasm. Several social factors included in the path models have an indirect influence on sexual satisfaction when the impact of the other variables is controlled. These include nonreligious and sexually unreserved home in childhood, early start of sex life, and liberal sex attitudes. Nevertheless, these social factors are important to sexual satisfaction because they are related to variables that directly increase it. The indirect influence of some variables on physical sexual satisfaction can be explained as follows. For men, age has no direct connection to sexual satisfaction. Young age, nevertheless, indirectly predicts men's sexual satisfaction as young men are sexually assertive and use many sexual techniques that increase sexual satisfaction. For women, considering sex important in life does not directly predict sexual satisfaction. However, women reporting that sex is important frequently engage in intercourse and use many techniques in sexual encounters, factors that contribute to physical sexual satisfaction. And even though love does not have a direct impact on women's sexual satisfaction, women who love and are loved engage frequently in sexual intercourse and often achieve orgasm in intercourse. These factors increase sexual satisfaction. Women's use of sex materials contributes to their sexual satisfaction through many sexual techniques which predict finding sexual intercourse pleasurable. An unexpected result is that, for women, high education is associated with lack of orgasm. In the highest educational category (15+ years of education) only 33%, in the middle group (10-14 years) as many as 75%, and in the lowest category (at most 9 years of school) 62% of women reported orgasm in intercourse at least most of the time. Even though there is no statistically significant correlation between education and achieving orgasm (Table IV), the regression coefficient is -.07 (p < 0.05) when one uses the dichotomy 15+ vs. fewer years of education (Fig. 4). In the Introduction we discussed the role of effective communication in increasing sexual satisfaction. Finding it easy to discuss sexual matters with one's partner correlates with sexual satisfaction among men (r = .28, p < 0.001). For women the correlation is even stronger (r = .40, p < 0.001). Nevertheless, the impact of this variable, which correlates with many other predictors of sexual satisfaction, disappeared when we controlled for other factors. Thus, this variable was excluded from path analyses. We also discussed the sexual satisfaction of people with one or more sexual partners. No statistically significant results were found between finding sexual intercourse pleasurable and having one or more sexual partners during the last year (Table V). Neither did living together with a steady sexual partner (being married or cohabiting) predict sexual satisfaction. However, there was a significant (p < 0.025) interaction in predicting sexual

Haavio-Mancila and Kontula

414

Table V. Physical Sexual Dissatisfaction According to Type of Partner Relationship and Number of Sexual Partners During Last Year and Gendera Sexual dissatisfactionb No. of sexual Men Women partners during Type of partner relationship X last year n X n Not married nor cohabiting Married or cohabiting

22

1

3.39 3.21

118 115

3.55 3.32

121 88

1 >2

3.16 3.37

561 119

3.52 3.50

622 46

aScale means (range 2-10). The higher the mean, the lower the satisfaction. bStatistical significance of the gender difference: p < 0.0001, and of the interaction

between

type of relationship and number of partners: p < 0.025.

satisfaction between being married or cohabiting and having had more than one sexual partner during the last year. The sexual satisfaction of nonmarried noncohabiting people with one partner was lower than that of people with several partners. Among the married or cohabiting men there was a tendency for men with one partner to enjoy intercourse more than men with several partners. The Finnish results thus support the findings by Laumann et al. (1994) and Davidson and Darling (1988): Monogamous men enjoy sexual intercourse more than men with several partners.

CONCLUSION The results of the two Finnish nationally representative sex surveys show an improvement in satisfaction with sexual life between 1971 and 1992. The growth of sexual satisfaction applies to women in particular. Women of today find their steady relationships as happy and are just as satisfied with their sexual life as a whole as men. However, women still find sexual intercourse less pleasurable than men do. Few Finns are dissatisfied with their sexual life as a whole. One wonders if the low proportion of people reporting emotional and physical sexual dissatisfaction indicates the existence of an unhappiness barrier (Roos, 1988, calls it "happiness barrier") beyond which it is not appropriate to go when one participates in a survey, even when the answers are written by the respondent and not shown to the interviewer. It is not easy to reveal unhappiness and dissatisfaction in sexual life, to admit that one has failed in this central area. In the 161 sexual autobiographies collected for the

Correlates of Increased Sexual Satisfaction

415

Finnish sex research project FINSEX, there are many stories of sexual failures and frustrations (Kontula and Haavio-Mannila, 1995b; HaavioMannila and Roos, 1995). This may be partly due to the self-selection of people who write autobiographies even though they demographically closely resemble the general population. But it is also possible that it is easier to express negative feelings and experiences when writing one's sexual history than when answering emotionally neutral survey questions. An autobiography can be compared to a diary, into which one pours the misery of the dark moments in order to get relief. Sexual life is now less connected with reproduction and the institution of marriage than earlier; it is more plastic and based on "pure" relationships (Giddens, 1991,1992). Our analysis of the Finnish survey data shows that overall sexual satisfaction is associated to the same extent with both physical and emotional sexual satisfaction. Satisfaction with "sublime love" and "bestial sex" (cf. Luhmann, 1984) belong together even though the acceptance of sex without love has increased in the last 20 years (Kontula and Haavio-Mannila, 1995a, pp. 65-66). According to the results of our survey, many social factors are connected to sexual satisfaction. Young age, sexually unreserved and nonreligious childhood home, early start of sexual life, high education, liberal sexual attitudes, sexual assertiveness, high importance given to sexuality in life, feeling of love, use of sex materials, frequent sexual intercourse, versatile sexual techniques, and frequent orgasm are associated with satisfaction of sexual life. Importance of sexuality in life, love, and use of sexual materials are directly connected to sexual satisfaction among men but only indirectly among women. Young women who are sexually assertive, use many sexual techniques, frequently engage in sexual intercourse, and often achieve orgasm in intercourse are sexually as satisfied as men with similar characteristics. Women used to be viewed as the traditional guardians of religion and morality in the family and community. The public has become more secular and straightforward in their treatment of women, sexuality, and other bodily pleasures (Davidson et al., 1995; Kontula and Kosonen, 1996). This enhances the sexual life of women. Our examination of gender differences in the predictors of sexual satisfaction makes us conclude that the sexual dissatisfaction of women is, at least to some extent, due to their late start of sexual life, conservative sexual attitudes, low importance of sexuality in life, lack of sexual assertiveness, and not using versatile sexual techniques. The emancipation of women probably will make them sexually less inhibited and change their ideas and practices. This will lessen the gender gap in physical sexual satisfaction.

416

Haavio-Mannila and Kontula APPENDIX Definitions of Variables in the Path Analyses (Figs. 3 and 4)

The questionnaire has been published in Sexual Pleasures—Enhancement of Sexual Life in Finland 1972-1992 (Kontula and Haavio-Mannila 1995a, pp. 239-287). A detailed version of the definition of the variables is available upon request from the authors. It includes response alternatives. Ranges of the sum scales are shown in Table III. Sexually Unreserved Home. What was your childhood home like? Sexual matters were kept secret—Unreserved about sexual matters. Nonreligious Home. What was your childhood home like? ReligiousNonreligious. Early Start of Sex Life. How old were you when you: Kissed for the first time? Started going steady for the first time? Had sexual intercourse for the first time? The first time had an orgasm during sexual intercourse? The reliability (Cronbach's alpha) of the sum scale is .76. Sexual Assertiveness. What is your opinion of the following statements concerning your sexual life and your sexual capacity? I have rather great sexual skills. I am sexually active. I am sexually attractive. The reliability of the sum scale is .80. High education. What is the total amount of years you have gone to school or college or some other educational institution (including both basic education and full-time study in addition to basic education)? People having gone to school for 15 or more years were classified as having high education. Love. Right now, is there some man or woman that you really love? Is there some man or woman who really loves you? The correlation coefficient is .74. Liberal Sex Attitudes. Sexual attitudes were measured on the basis of six items. The following statements were chosen from 33 different statements referring to sexuality on the basis of a factor analysis: It is good that there are reform-minded groups of people in society, who defend more free moral trends.

Correlates of Increased Sexual Satisfaction

417

Women have every right to take the initiative when they want sexual contact with men. Homosexual behavior among adults is the private affair of the people concerned, with which officials and the law should in no way interfere. Completely temporary sexual relations can be happy and satisfying to both parties. People over 16 years of age should be able to freely buy pornographic magazines from specially licensed stores. One must be able to accept a wife's temporary infidelity. The reliability of the sum scale is .62. Sex Important. People value differently various parts of their life. We ask you to evaluate an importance of each of the following areas in your life. Sexual life was one of the 13 areas listed in the interview form. The others were: Work; family; your own health; enjoying life; social work; selftraining, studying; close friendly relations; living standard, material well-being; religion, spiritual matters; love; high status in society; neat appearance. Sex Materials. In the last 12 months, have you read or watched following materials, which in your opinion are sexually interesting or arousing (including so-called pornography)? Sex movies or other sex programs on television Sex videos Sex magazines Wall calendars with naked pictures In the last year, have you read or glanced through a magazine or a book that you think could be called pornographic? The reliability of the sum scale is .71. Frequent Intercourse. Sexual intercourse was defined as "sexual interaction in vaginal, oral or anal intercourse or stimulation by hand." Its frequency was measured by two questions: When was the last time that you had sexual intercourse? How often have you had sexual intercourse during the last month (30 days)? The correlation coefficient is .81. Manysided Techniques. In what position did your last sexual intercourse take place? Have you ever had anal intercourse (a man's penis entering a partner's anus)?

418

Haavio-Mannila and Kontula

When have you last stimulated your partner's genitals and given (for women: him, for men: her) satisfaction without sexual intercourse? (e.g., fondling and stimulating by hand)? In the last 5 years, how often have you had oral sex in your sexual relations, that is, fondling a man's penis or a woman's genitals by mouth? The reliability of the sum scale is .74. Orgasm. An ending of sexual tension into coming and an intense feeling of pleasure is called orgasm. Have you had an orgasm during sexual intercourse? Did you have an orgasm during your last sexual intercourse? The correlation coefficient is .82. Satisfaction with Intercourse. What kind of an experience was your last sexual intercourse? Have your intercourse mostly been very unpleasant or very pleasurable? The correlation coefficient is .54. REFERENCES Alzate, H. (1985). Vaginal eroticism and female orgasm: A current appraisal. J. Sex Marital Ther. 11: 271-284. Dariing, C. A., and Davidson, J. K. (1986). Enhancing relationships: Understanding the feminine mystique of pretending orgasm. J. Sex Marital Ther. 12: 182-196. Darling, C A., Davidson, J. K., and Cox, R. P. (1991). Female sexual response and the timing of partner orgasm. /. Sex Marital Ther. 17: 3-21. Darling, C A., Haavio-Mannila, E., and Kontula, O. (1996) Orgasmic frequency in Finland: Do age and gender make a difference? Paper presented at the annual meeting of The Society for the Scientific Study of Sexuality, Houston, Texas. Davidson, J. K., and Darling, C. A. (1988). The sexually experienced woman: Multiple sex partners and sexual satisfaction. J. Sex Res. 24: 141-154. Davidson, J. K., Darling, C. A, and Norton, L. (1995). Religiosity and the sexuality of women: Sexual behavior and sexual satisfaction revisited. J. Sex Res. 32: 235-243. Giddens, A. (1991). Modernity and Self-Identity, Stanford University Press, Stanford, CA. Giddens, A. (1992). The Transformation of Intimacy, Stanford University Press, Stanford, CA Greeley, A. (1991). Faithful Attraction: Discovering Intimacy, Love and Fidelity in American Marriage, Tor, New York. Haavio-Mannila, E., and Roos, J. P. (1995). Love, Generation and Gender. Paper presented at the Second European Conference for Sociology, Budapest, August 30-September 2. Huberle, D. F. (1991). The Role of assertiveness in female sexuality: A comparative study between sexually assertive and sexually nonassertive women. J. Sex Marital Thar. 17: 183-190. Kontula, O., and Haavio-Mannila, E. (1995a). Sexual Pleasures—Enhancement of Sexual Life in Finland, 1971-1992, Dartmouth, Aldershot.

Correlates of Increased Sexual Satisfaction

419

Kontula, O., and Haavio-Mannila, E. (19955). Matkalla intohimoon [On the way to passion], WSOY, Helsinki. Kontula, O., and Kosonen, K. (1996). Sexuality changing from privacy to the open—A study of the Finnish press over the years from 1961 to 1991. Nord. Sexol 14: 34-47. Laumann, E. O., Gagnon, J. H., Michael, R. T., and Michaels, S. (1994). The Social Organization of Sexuality. Sexual Practices in the United States. Chicago and London, University of Chicago Press, Chicago. Luhmann, N. (1984). Liebe als Passion, Surkamp, Frankfurt am Main. Morokoff, P. (1978). Determinants of female orgasm. In LoPiccolo, J., and LoPiccolo, L. (eds.), Handbook of Sex Therapy, Plenum Press, New York, pp. 147-165. Pinney, E. M., Gerrard, M., and Denney, N. W. (1987). The Pinney Sexual Satisfaction Inventory. J. Sex Res. 23: 233-251. Roos, J. P. (1988). Beyond the happiness barrier. Soc. Indicators Res. 20: 141-163. Sievers, K., Koskelainen, O., and Leppo, K. (1974). Suomalaisten sukupuolielama [The Sexual Life of the Finns], WSOY, Helsinki. Spector, I. P., and Carey, M. P. (1990). Incidence and prevalence of the sexual dysfunctions: A critical review of the empirical literature. Arch. Sex. Behav. 19: 389-408. Sprecher, S., and McKinney, K. (1993). Sexuality, Sage, Newbury Park, CA.

Archives of Sexual Behavior, Vol. 26, No. 4, 1997

Self-Reported Childhood Abuse Among Sex and Violent Offenders Jaana Haapasalo, Ph.D.,1,2 and Marjo Kankkonen, M.A.1

Few prior studies have compared the childhood abuse experiences of different types of offenders. In this pilot study, the childhood abuse experiences of sex offenders (n = 16) and violent offenders (n = 16) were examined with retrospective structured interviews. The sex offenders reported significantly more psychological abuse, especially verbal abuse, and also tended to have experienced slightly more physical and sexual abuse than the violent offenders. In general, the sex offenders remembered their parents in a more negative light than the comparison group. The offender groups did not differ from each other in the total amount of self-reported early conduct disorder symptoms. The sex offenders appeared to come from more abusive childhood family environments, but the conclusions remain tentative due to the small sample and retrospective data collection. KEY WORDS: child abuse; psychological abuse; sex offender; violent behavior. INTRODUCTION Several studies have indicated a relationship between childhood abuse and later aggressive and criminal behavior (e.g., Luntz and Widom, 1994; Widom, 1989, 1991). Rivera and Widom (1990) reported that traumatic abusive incidents in childhood are associated with an increased risk for violent offending later in life. Stein and Lewis (1992) found that abused chil-

The study was supported in pan by Grant SA 8482 from the Academy of Finland. 1Department of Psychology, University of Jyvaskyla, P.O. Box 35, FIN-40351 Jyvaskyla, Finland. 2To whom correspondence should be addressed. 421 0004-0002/97/D800-0421$12.50A>« 1997 Plenum Publishing Corporation

422

Haapasalo and Kankkonen

dren were more aggressive than their nonabused peers. Also, sexual offending has been linked to childhood sexual and physical abuse (e.g., Becker et al., 1991; Benoit and Kennedy, 1992; Prentky et al., 1989; Widom and Ames, 1994). According to existing studies both sex and violent offenders have grown up in violent childhood family environments. There are, however, relatively few studies focusing on the differences between the childhood histories of sex and violent offenders. We still know more about the adult personality of sex offenders (e.g., Levin and Stava, 1987) than the developmental aspects of sex offending. However, some data have accumulated on the development of sex offenders. According to Knight and Prentky (1993), sex offenders who started their criminal career in adolescence reported having experienced more serious and earlier sexual abuse than those who started later. Sexual deviation and abuse within the childhood family have been found to be related to the severity of sexual aggression, whereas physical abuse and neglect were associated with severity of nonsexual aggression (Prentky et al., 1989). Among sex offenders, a greater proportion of pedophiles than rapists have reported childhood sexual abuse, 57 vs. 23% in one study and 56 vs. 5% in another (Watkins and Bentovim, 1992). Early-starting child molesters have also been found to have experienced more physical abuse than those who started their offenses later (Knight and Prentky, 1993). Lewis et al. (1979) noted that young sex offenders and violent offenders had experienced childhood abuse approximately to the same extent. On the other hand, a follow-up study indicated that juvenile sex offenders committed a greater number of and more serious violent offenses in adulthood as compared to juvenile violent offenders (Rubinstein et al., 1993). This finding gives reason to assume that sex offenders originally were more disturbed than violent offenders. A greater degree of psychological disturbance, in turn, may reflect the more dysfunctional childhood family environment among sex offenders compared to violent offenders. There could be further differences in childhood experiences between various types of sex offenders, such as those whose victims are adults and those who victimize children. The main objective of the present study was to compare adult sex offenders (rapists) with violent offenders in terms of their childhood abuse recollections. The study served as a pilot study for a larger research project targeting the relationship between childhood abuse and neglect experiences among offenders.

Childhood Abuse Among Offenders

423

METHOD Subjects The study and comparison groups were randomly sampled from inmates currently serving a sentence in prison. The study group consisted of 16 adult male sex offenders whose victims were women over the age of 18 years. The mean age of the offenders was 35.1 years (SD = 7.68), the age range was 24-50. They had been convicted for rape (n = 9), assault and rape (n = 4), aggravated assault and rape (n = 1), and rape and manslaughter (n = 2). The sample comprised the majority of the convicted rapists in prison at the time of study. The comparison group comprised 16 adult male violent offenders who had no convictions for sex offenses. The violent offenders' mean age was 33.9 years (SD = 7.73); age ranged from 23-50. The mean ages of the offender groups did not differ significantly from each other. The study and comparison groups were matched according to a childhood family problem index. The index varied from 0 to 5 and was based on the following variables: placed in institutional/foster care in childhood, experienced parental divorce in childhood, witnessed spousal abuse in childhood, parents/custodians abused alcohol or drugs, and parents/custodians were mentally disordered. The offender groups did not differ from each other in terms of the family problem index. Measures

Structured Interview and Files Each subject was interviewed using a structured form which included three parts: the Family-of-Origin Scale (Brassard et al., 1993), a questionnaire based on the Maltreatment Classification System developed by Barnett et al. (1993), and a Self-Evaluation of Difficulties Questionnaire (Meek, 1990). Also, the prison files containing information about the offenders' criminal history and childhood family environment were reviewed. Family-of-Origin Scale (FOO). The FOO provides information about the composition of the subject's childhood family environment, such as criminality, alcohol abuse and mental problems in family members, and behavior towards the subject by parents and peers in childhood. The original FOO Scale has been found to be both a reliable and valid measure of childhood family environment (Capps et al. 1993; Manley et al., 1990).

424

Haapasalo and Kankkonen

Maltreatment Classification System. The system is based on abuse literature, information from child abuse experts, and collaborative work with clinicians (Barnett et al, 1993). It lists various types of physical, psychological, and sexual abuse, and neglect. Also, the severity of abuse/neglect is expressed. Barnett et al. emphasized that empirical studies are needed to gather validity and reliability data concerning the system. According to them, the system can be modified after more information about its ability to target child maltreatment becomes available. In the Maltreatment Classification System-based part of the interview, the subject was asked to answer Likert-type statements concerning different types of physical, psychological, and sexual abuse with five options: never, rarely, sometimes, often, always. There were 21 physical abuse statements (e.g., "As a child, I was hit with a belt by my parents.") and 12 statements pertaining to sexual abuse (e.g., "As a child, I was forced to witness sexual activities by my parents.")- In addition, there were 21 psychological abuse statements which were divided into two subgroups: verbal abuse (e.g., yelling, threatening, ridiculing) and rejecting/isolating (e.g., rejecting: "As a child, I experienced an openly hostile and negative attitude on the part of my parents"; isolating: "As a child, I was locked up in a closet by my parents."). Moreover, 5 statements inquired whether the offender had been subjected to antisocial influences in his childhood family environment (e.g., "As a child, I saw my parents commit crimes."). Antisocial influences may include letting the child witness criminal and antisocial acts or instilling immoral or antisocial values in the child. These influences were treated as a separate form of maltreatment, although they are similar to the moral corruption that is often incorporated in psychological abuse (see Hart and Brassard, 1991). They differ from intentional psychological abuse in that the parents may not be aware of the child being influenced by them and that the parents do not deliberately cause psychological harm to the child. Statements were weighted according to the severity of abuse as follows: 1 = mild abuse (e.g., physical abuse that left no visible marks, such as hair-pulling), 2 = moderate abuse (e.g., physical abuse that left bruises or other visible marks, such as hitting with an object), and 3 = severe abuse (e.g., physical abuse that required medical attention or caused permanent injuries or even life endangerment, such as strangling). The detailed interview is available from the authors. Self-Evaluation of Difficulties Questionnaire (SEDQ). The SEDQ is a self-report instrument in which various maladjustment symptoms before the age of 15 years are charted (Meek, 1990). It was added to the interview to look for conduct disorder symptoms, such as truancy, expulsion from school, arrests, running away, alcohol and drug abuse, stealing, and violent behavior. For both offender groups, a total sum of conduct disorder symp-

Childhood Abuse Among Offenders

425

toms was computed, although the groups were compared with each other also in individual symptoms.

Procedure The study was conducted in prisons, where the potential subjects first signed an informed consent. Confidentiality regarding all information was emphasized. After having signed the written consent, subjects participated in individual interviews that took from about 1 to 2 hr. Next, the prison files were read.

Data Analyses In case of skewed distributions of variables, nonparametric MannWhitney tests and Spearman correlations were computed. If the distribution of Likert-type variables was approximately normal, t tests were used. All analyses were carried out using the SPSS for Windows 6.1 program.

RESULTS Childhood Abuse Experiences Table I presents the differences between sex and violent offenders in their abuse experiences. In addition to psychological, physical, and sexual abuse, having been subjected to antisocial influences such as parental criminality or alcoholism was taken into account as a separate type of maltreatment. The sex offenders tended to have experienced more abuse than the violent offenders, but the differences between the groups reached significance only for verbal abuse. The sex offenders reported more verbal abuse, for example having been threatened and ridiculed more than the violent offenders. In Fig. 1, the scales of abuse types were modified by giving each scale the same maximum so as to make them comparable with each other. Figure 1 indicates that the reports of rejecting and isolating were more common than the reports of physical abuse in the sex offender group, but in the violent offender group physical abuse was more common than rejecting and isolating. In both groups, sexual abuse was reported much less than other forms of abuse.

426

Haapasalo and Kankkonen

Table I. Differences Between Sex and Violent Offenders in Childhood Abuse (Mann-Whitney) n Type of abuse and offender group X z P Psychological (verbal") Sex Violent Psychological (rejectingb/isolatingc) Sex Violent Physical Sex Violent Antisocial influencesd Sex Violent

16 16

21.44 11.56

2.98

<0.01

16 16

19.47 13.53

1.86

<0.10

16 16

17.59 15.41

0.66

ns

15 16

17.50 14.59

0.93

ns

Sexual 18.16 1.15 16 ns Sex 14.84 16 Violent aYelling, threatening, ridiculing. bShowing open hostility and negativity towards the child, ignoring, favoring siblings. Locking up the child in a closed environment, isolating him from other people. dLetting the child witness criminal and antisocial behavior, such as parental substance abuse and criminality, or inculcating immoral or antisocial values (moral corrupting).

In the sex offender group, physical abuse correlated with rejecting/isolating (r = .92, p < 0.001), as did verbal abuse (r - .71, p < 0.01). In the violent offender group, rejecting/isolating correlated with sexual abuse (r = .71, p < 0.01) and antisocial influences (r = .68, p < 0.01). The differences between the two offender groups were examined also in a number of interview variables reflecting the quality of attachment with parents. The sex offenders recalled the relationship with their parents as significantly worse than did the violent offenders, as shown in the significant differences between the groups in the following statements: "Mother did not want me near her." (t = 2.65, p < 0.05) "Father did not want me near him." (t = 3.47, p < 0.01) "Mother threatened to hurt me." (t = 3.24, p < 0.01) "Father threatened to hurt me." (t = 3.52, p < 0.001) "Mother belittled me." (t = 2.30, p < 0.05) "Father belittled me." (t = 3.80, p < 0.001) "Father made me do shameful things." (t = 3.41, p < 0.05)

Childhood Abuse Among Offenders

427

Fig. 1. The proportions of abuse types in the offender groups.

Conduct Disorder Symptoms The sex offenders did not differ significantly from the violent offenders in the total sum of conduct disorder symptoms before the age of 15 years. School maladjustment was common in both groups: 88% of the sex offenders reported truancy (vs. 69% of the violent offenders) and 44% had been expelled from school (vs. 31%). The groups differed most in initiating fights. In the violent offender group, 10 (63%) subjects reported having started rights often in childhood, whereas only 2 (13%) sex offenders reported this. In the sex offender group, the total sum of conduct disorder symptoms correlated with psychological abuse (verbal abuse: r - .53, p < 0.05; rejecting/isolating: r ~ .54, p < 0.05). Also, having been subjected to antisocial influences correlated with the symptoms (r = .54, p < 0.05). In the comparison group, the conduct disorder symptoms were related to rejecting/isolating (r = .54, p < 0.05) and antisocial influences (r = .50, p < 0.05).

428

Haapasalo and Kankkonen

DISCUSSION The main aim of the present study was to compare the self-reported childhood abuse experiences among sex and violent offenders. The sex offenders reported more psychological abuse than did the violent offenders. In particular, the sex offenders recalled verbal abuse, such as yelling and threatening, more than the violent offenders. On the other hand, the groups failed to differ from each other in the reported physical or sexual abuse experiences and antisocial influences in the childhood family environment. The sex offenders tended, however, to come from slightly more abusive family environments than the violent offenders. In the total amount of conduct disorder symptoms, the groups were not different from each other. Although this study can be best considered as a pilot study due to the small sample, the findings lend support to previous investigations concerning the relationship between childhood abuse and later offending. A great number of adult offenders have been abused in childhood according to retrospective reports (e.g., Lewis et al., 1979; Widom, 1989), but physical abuse has been more related to general than sexual aggression (Prentky et al., 1989). Sexual abuse has been found in the family background of sex offenders (e.g., Prentky et al., 1989; Rubinstein et al., 1993; Widom and Ames, 1994), although more commonly among pedophiles than rapists (Watkins and Bentovim, 1992). The prevalence of psychological abuse among sex offenders has rarely been studied. This may relate to the fact that the definition of psychological abuse is elusive (Hart and Brassard, 1991). In this study, forms of psychological abuse were not examined separately but were split into two groups, one of which represented verbal abuse (yelling, threatening, ridiculing) and the other nonverbal abuse (rejecting, isolating). This dichotomization seemed appropriate in light of the fact that it revealed the high degree of verbal abuse dominating the retrospective recollections of sex offenders. The interpretation of the high prevalence of reported verbal abuse experiences in the sex offender group could be linked to the negative qualities of the parent-child relationship. The sex offenders recalled their parents as having been more rejecting, threatening, and belittling than did the violent offenders. This could, of course, result from a biased negative retrospective perception of parental behavior, not the actual incidents of verbal abuse or rejection. Alternatively, it could be assumed that in the childhood family environment of the violent offenders there was more positive interaction between the parents and the child compared to that of the sex offenders. Physically abusive parents, such as the parents of the violent offenders, have been found to be expressive both with regard to positive

Childhood Abuse Among Offenders

429

and negative affect (Widom, 1989). In such families, abusive incidents may alternate with warm and positive affective interaction. Regardless of the higher degree of verbal abuse in the childhood family environment of the sex offenders as compared to the violent offenders, the sex offenders evidenced no more conduct disorder symptoms in childhood. It could be assumed that the developmental pathway of both groups is relatively similar in terms of the early conduct disorder symptoms. The violent offenders reported, however, having started fights in then- childhood more often than the sex offenders. In contrast, Rubinstein et al. (1993) found that sex offenders early in their lives were as violent as violent offenders. There is evidence that violent family environment and witnessing parental violence may be linked to later aggressive behavior among sex offenders (see Marshall and Barbaree, 1990). Aggression is one of the dimensions that differentiates between various types of sex offenders (Prentky and Knight, 1991). It should be noted that adverse childhood experiences may not be causally related to aggressive offending, but genetically mediated biochemical characteristics and antisocial personality traits could also account for the sexually aggressive behavior (Nagayama Hall and Hirschman, 1991). For example, offenders scoring high on the Hare Psychopathy Checklist have been found to commit more serious violent offenses than other offenders (Hare and Hart, 1993). Two limitations of the present study should be noted. The sample can best be considered as representative of those rapists who were in prison at the time of the study. The rapists who did not get caught or who committed less serious sex offenses as first-time offenders, and were thus given conditional sentences, remained outside the scope of the study. Second, the retrospective nature of data collection may have led to distorted reports of abuse and conduct disorder symptoms. To measure the prevalence of abuse and conduct disorder symptoms in a more reliable manner, it would be useful to rely more on early documents and files than on retrospective interview reports. It has been reported that salient experiences and objective facts are usually easier to recall than generalized parental behavior, such as parental conflict in the home (Brewin et al., 1993; Henry et al., 1994). Some offenders may have a tendency to deny or rationalize harsh parental treatment (Delia Femina et al., 1990), and this could explain why physical and sexual abuse were not abundantly reported in this sample. The fact that the more the offenders reported conduct disorder symptoms, the more they also reported childhood abuse, is, however, a consistent finding and increases the reliability of the retrospective reports.

430

Haapasalo and Kankkonen REFERENCES

Barnett, D., Manly, J. T., and Cicchetti, D. (1993). Defining child maltreatment: The interface between policy and research. In Cicchetti, D., and Toth, S. L. (eds.), Child Abuse, Child Development, and Social Policy. Advances in Applied Developmental Psychology, Vol. 8. Ablex, Norwood, NJ, pp. 7-74. Becker, J. V., Kaplan, M. S., Tenke, C. E., and Tartaglini, A. (1991). The incidence of depressive symptomatology in juvenile sex offenders with a history of abuse. Child Abuse Neglect 15: 531-536. Benoit, J. L, and Kennedy, W. A. (1992). The abuse history of male adolescent sex offenders. J. Interpers. Violence T. 543-548. Brassard, M. R., Hart, S. N., and Hardy, D. B. (1993). The psychological maltreatment rating scales. Child Abuse Neglect 17: 715-729. Brewin, C. R., Andrews, B., and Gotlib, I. H. (1993). Psychopathology and early experience: A reappraisal of retrospective reports. Psychol. Bull 113: 82-93. Capps, S. C, Searight, H. R., Russo, J. R., and Temple, L. E. (1993). The Family-of-Origin Scale: Discriminant validity with adult children of alcoholics. Am. J. Fam. Ther. 21: 274-277. Delia Femina, D., Yeager, C. A., and Lewis, D. O. (1990). Child abuse: Adolescent records vs. adult recall. Child Abuse Neglect 14: 227-231. Hare, R. D., and Hart, S. D. (1993). Psychopathy, mental disorder, and crime. In Hodgins, S. (ed.), Mental Disorder and Crime, Sage, Newbury Park, pp. 104-115. Hart, S. N., and Brassard, M. R. (1991). Psychological maltreatment: Progress achieved. Dev. Psychopathol. 3: 61-70. Henry, B., Moffitt, T. E, Caspi, A., Langley, J., and Silva, P. A. (1994). On the "remembrance of things past": A longitudinal evaluation of the retrospective method. Psychol. Assess. 6: 92-101. Knight, R. A., and Prentky, R. A. (1993). Exploring characteristics for classifying juvenile sex offenders. In Barbaree, H. E., Marshall, W. L and Hudson, S. M. (eds.), The Juvenile Sex Offender, Guilford, New York, pp. 45-83. Levin, S. M., and Stava, L. (1987). Personality characteristics of sex offenders: A review. Arch. Sex. Behav. 16: 57-69. Lewis, D. O., Shanok, S. S., Pincus, J. H., and Glaser, G. H. (1979). Juvenile male sexual assaulters. Am. J. Psychiat. 136: 1194-1196. Luntz, B. K., and Widom, C. S. (1994). Antisocial personality disorder in abused and neglected children grown up. Am. J. Psychiat. 151: 670-674. Manley, C. M., Searight, H. R., Skitka, L. J., and Russo, J. R. (1990). The reliability of the Family-of-Origin Scale for adolescents. Fam. Ther. 17: 273-280. Marshall, W. L., and Barbaree, H. E. (1990). An integrated theory of the etiology of sexual offending. In Marshall, W. L., Laws, D. R., and Barbaree, H. E. (eds.), Handbook of Sexual Assault, Plenum Press, New York, pp. 257-275. Meek, C. L. (1990). Post-traumatic stress disorder. Self-Evaluation of Difficulties Questionnaire. In Meek, C. L. (ed.), Post-Traumatic Stress Disorder Assessment, Differential Diagnosis, and Forensic Evaluation, Pullman, Sarasota, pp. 197-215. Nagayama Hall, G. C, and Hirschman, R. (1991). Toward a theory of sexual aggression: A quadripartite model. J. Consult. Clin. Psychol 59: 662-669. Prentky, R. A., and Knight, R. A. (1991). Identifying critical dimensions for discriminating among rapists. J. Consult. Clin. Psychol 59: 643-661. Prentky, R. A., Knight, R. A., Sims-Knight, J. E., Straus, H., Rokous, F., and Cerce, D. (1989). Developmental antecedents of sexual aggression. Dev. Psychopathol. 1: 153-169. Rivera, B., and Widom, C. S. (1990). Childhood victimization and violent offending. Violence Victims 5: 19-35. Rubinstein, M., Yeager, C. A., Goodstein, C, and Lewis, D. O. (1993). Sexually assaultive male juveniles: A follow-up. Am. J. Psychiat. 150: 262-265.

Childhood Abuse Among Offenders

431

Stein, A., and Lewis, D. O. (1992). Discovering physical abuse: Insights from a follow-up study of delinquents. Child Abuse Neglect 16: 523-531. Watkins, B., and Bentovim, A. (1992). Male children and adolescents as victims: A review of current knowledge. In Mezey, G. C., and King, M. B. (eds.), Male Victims of Sexual Assault, Oxford University Press, Oxford, pp. 27-66. Widom, C. S. (1989). Child abuse, neglect, and adult behavior: Research design and findings on criminality, violence, and child abuse. Am. J. Orthopsychiat. 59: 355-367. Widom, C. S. (1991). Avoidance of criminality in abused and neglected children. Psychiatry 54: 162-174. Widom, C. S., and Ames, M. A. (1994). Criminal consequences of childhood sexual victimization. Child Abuse Neglect 18: 303-318.

Archives of Sexual Behavior, Vol. 26, No. 4, 1997

Sexual Orientation and Professional Dance J. Michael Bailey, Ph.D.,1 and Michael Oberschneider, B.A.1

The stereotypical professional male dancer is a gay man. However, little if any systematic research has investigated the validity of this stereotype, much less the reasons why male sexual orientation would be associated with interest in dance. We interviewed 136 professional dancers about the prevalence of homosexuality among dancers, the dancers' own sexual development, and relationships between dancers of different sexual orientations. Dancers estimated that over half of male dancers are gay, but that only a small minority of female dancers are lesbian. Gay men recalled more intense early interest in dance compared to heterosexual men and women, and were more feminine as boys than were heterosexual men. Gay men's homosexual feelings typically preceded their dance experience, and only one gay man felt that his dance experiences may have influenced his sexual orientation. Heterosexual men voiced some mild complaints about gay male dancers, but these were balanced by positive sentiments. KEY WORDS: sexual orientation; dance; careers; gender identity.

INTRODUCTION Stereotypes of gay men and lesbians include occupational components. Gay men are thought to be disproportionately represented among the arts (especially dance), fashion careers (hairdressing and design), and decorating (interior decorating and florists). Lesbians are widely believed to frequent the military and women's professional athletics. Occupations and professions stereotypically associated with gay men are considered feminine and have a high proportion of women; analogously, "lesbian" occupations are masculine and (with the exception of womens' athletics) con1Department

of Psychology, Northwestern University, Evanston, Illinois 60208-2710. 433 0004-0002/07/0800-0433$12.50/0 c 1997 Plenum Publishing Corporation

434

Bailey and Oberschneider

tain a high proportion of men. Despite the pervasiveness of these stereotypes, empirical research on their validity and related questions has been scanty. At least three general questions might be asked concerning an occupation hypothesized to have a high proportion of gay men or lesbians. First, is there, in fact, a relation between sexual orientation and membership in the occupation in question? Second, if there is such an association, why does it occur? Third, what are the implications of a high percentage of gay or lesbian coworkers for heterosexual members of the occupation? It would seem a fairly straightforward matter to determine if a given occupation has a disproportionate number of gay men or lesbians, but there are in fact serious barriers. For example, it is often difficult to get a random sample of the members of an occupation. A more serious difficulty concerns the logistics of surveying members of an occupation about their sexual behavior, especially often stigmatized sexual behavior, such as homosexuality. People are likely to be generally cautious about answering such questions, perhaps especially so in the context of an occupational survey. For example, at the time of this writing homosexual people are still expelled from the military if they engage in homosexual acts. Thus, gay men and lesbians in the military have a disincentive to cooperating in such research. However, most employers are less concerned than the military about their employees' private lives, and thus, it is often more feasible to conduct such research than it would be in the military. One alternative to asking people directly about their own sexual orientation is to ask more general questions about the sexual orientation of others in the occupation. For example, one might ask "What percentage of your male coworkers are gay," or "What percentage of men in your occupation are gay?" How accurately respondents could estimate these values would depend on factors such as how open coworkers are with each other about their sexuality and how aware they are of such trends outside their immediate place of employment. If people in a given profession could accurately answer such questions, it might partly circumvent the need to obtain random or representative samples, and respondents may feel more open answering questions about unspecified others than about themselves. If an occupation or profession has a disproportionate number of gay men or lesbians, one must then ask "why?" Researchers have offered at least two general explanations of the congregation of homosexual people in certain occupations. We distinguish these general explanations as "psychological" and "sociological." Whitam (in press; Whitam and Dizon, 1979; Whitam and Mathy, 1986) has been the primary proponent of the former position, that important psychological differences between homosexual and

Sexual Orientation and Professional Dance

435

heterosexual people, especially regarding the preference for typical versus atypical gender roles, predispose homosexual people to prefer occupations typically preferred by opposite-sex heterosexual people. Both prospective (Green, 1987) and retrospective (Bailey and Zucker, 1995) studies have demonstrated an association between homosexuality and sex-atypical behavior during childhood. Whitam argues that the adult occupational differences are developmental consequences of the early differences. Supporting the psychological position, Chung and Harmon (1994) found moderate differences in the career interests and aspirations of gay and heterosexual men, with gay men's interests less realistic and investigative, but more artistic and social. In contrast, the sociological position (see Murray, 1991) argues that occupations more or less arbitrarily become identified as friendly to gay or lesbian employees because there is less antihomosexual discrimination in them. Homosexual people transmit information to each other about job opportunities, and eventually these occupations become disproportionately homosexual. The sociological position denies that the aforementioned differences in childhood sex-typicality play an important role in occupational differences. Those occupations that have a highly disproportionate number of gay men or lesbians allow investigation of a number of interesting questions. For example, much of the debate over whether openly homosexual people should serve in the military was necessarily speculative, but could have been informed by research on relationships between homosexual and heterosexual people in disproportionately homosexual occupations. Some theories concerning origins of sexual orientation can also be examined in the context of research on "homosexual" occupations. For example, one common belief is that environments tolerant of homosexuality will spawn more homosexuals compared to less tolerant environments (e.g., Patullo, 1992). Another is that homosexuality is often the result of early seduction by older same-sex adults (Newton, 1978). If these hypotheses are true, then gay men and lesbians from "homosexual" occupations should often claim to have become homosexual due to their occupational experiences, especially seduction by coworkers. Heterosexual coworkers in occupations tolerant of homosexual behavior should be more likely to have had homosexual experiences and to have struggled with their own sexual orientation. In this paper, we report the results of a study concerning sexual orientation in professional dance, including ballet, modern, and jazz/musical theater. We chose to study professional dance for several reasons. First, professional male dancers are widely assumed to be disproportionately gay. Supporting the stereotype, one of the most famous male dancers of this century, Rudolph Nureyev, was gay, and several prominent figures in the dance world are openly gay. However, no one has at-

436

Bailey and Oberschneider

tempted a systematic survey of this question. Second, many dancers begin dancing quite young, and so may provide some insight into the role of psychological versus sociological factors in their career choice. Third, members of a dance company often socialize with each other, and thus might be expected to know more about each other's sexuality. Fourth, we believed that although a substantial proportion of professional male dancers are gay, an appreciable proportion are heterosexual. Thus, professional dance should provide an opportunity to study the resulting social dynamics. Finally, one of us (M.O.) was a professional ballet dancer for several years, and thus had some useful knowledge and helpful contacts.

METHOD

Participants Participants included present and past professional dancers. Participants were recruited unsystematically, via personal contacts, a solicitation in a Chicago dance magazine and "snowball sampling." We attempted to interview approximately equal numbers of gay men, heterosexual men, heterosexual women, and lesbians, but it was not possible to interview a comparable number of lesbian dancers, due to their apparent rarity. Procedure

We interviewed dancers either in person or over the telephone. Interviews lasted approximately 90 min, on average. Participants were guaranteed anonymity. Interviews included questions concerning dancers' professional experiences including the development of their interests, sexual histories, experiences with dancers of other sexual orientations and attitudes about those experiences, and dancers' estimates about the prevalence of homosexuality among dancers generally, and in their own companies. Participants also completed a short questionnaire assessing sexual orientation via Kinsey scores, and, for men, (retrospective) childhood gender typicality. The latter included items regarding participants' memories of sex-typed feelings and behavior (e.g., "I was called a sissy."). Scales containing similar items generally yield large differences between heterosexual and homosexual men (Bailey and Zucker, 1995).

Sexual Orientation and Professional Dance

437

RESULTS Participants We interviewed 48 gay men, 42 heterosexual men, 45 heterosexual women, and 1 lesbian. Sexual orientation was determined on the basis of self-identification. Unfortunately, we subsequently lost (through theft) interview tapes of 9 gay men, 1 heterosexual man, and 6 heterosexual women, and those data are unavailable. Questionnaires were obtained from 39 gay men, 38 heterosexual men, 36 heterosexual women, and 1 lesbian; for 9 gay men and 1 heterosexual men these were the only available data. (In the analyses that follow, variation in the degrees of freedom reflect missing data.) Dancers' ages ranged from 17 to 50, with a mean of 28.9 (SD = 6.9). The sample's ethnic composition was 80% white, 6.2% African American, 8.0% Hispanic, and 5.8% other or unspecified ethnicity. Most of the dancers (84%) were employed in dance companies at the time of the interviews. Eleven percent were dancing independently, and 6% were retired. Dancers represented 36 companies, based in several large U.S. cities. The companies varied considerably in their size, prestige, and regional appeal. We interviewed multiple members from 16 companies; the number interviewed in these companies ranged from 2 to 13. Estimates of Homosexuality Among Dancers We asked all participants several questions concerning the prevalence of homosexuality in the dance world. On average, dancers estimated that 57.8% of male dancers are gay, though individual estimates ranged from 25-98%. Estimates differed slightly but significantly between gay men, heterosexual men, and heterosexual women, with mean estimates of 58.0,52.4, and 64.3%, respectively, F(2, 108) = 4.9, p < 0.01. (Because there was only one lesbian, group comparisons must ignore her.) In contrast, the estimated prevalence of homosexuality among female dancers averaged 3.1%, and did not significantly differ between groups. Several of the dancers believed that the prevalence of lesbianism among modern dancers was considerably higher than among other kinds of professional dancers. Participants also provided estimates of the percentages of gay men and lesbians in their present companies. These figures were important for two reasons. First, to the extent that dancers know about the sexual orientations of others in their companies, then dancers in the same company should provide similar estimates. If they did not, one could have little confidence in their estimates regarding the prevalence of homosexuality among

438

Bailey and Oberschneider

dancers in general. Second, estimates of homosexuality in companies should be consistent with estimates concerning dancers overall. Dancers reported that, on average, 53% of the men in their companies were gay. (This percentage changed little, to 54%, even if the mean was not weighted by company.) The intraclass correlation between dancers' estimates across companies was .82, indicating a high degree of similarity for the estimates of dancers from the same company. Dancers reported that to their knowledge, no female dancer in their companies was lesbian, with the exception of the single lesbian participant. Early Interests On average, women began dancing earlier than the men, ages 7.3 (SD = 3.4) and 12.2 (SD = 4.6), respectively, t(lll) = 5.5, p < 0.001. Gay men began dancing slightly later than heterosexual men, 13.1 (SD = 4.5) versus 11.3 (SD = 4.5), respectively, t(78) = 1.9, p = 0.07. When asked who first inspired their interests in dance, gay men gave markedly different answers from the other groups. Sixty percent of the heterosexual dancers said "Parents," compared to only 13% of the gay male dancers. In contrast, 50% of the gay men said they became interested by themselves, compared to only 19% of the heterosexual dancers. (Heterosexual men and women gave similar responses. The final category, "Friends or Other," did not distinguish the three groups.) The overall pattern of responses differed significantly between gay male and heterosexual dancers, x2(2) = 23.1, p < 0.001. Furthermore, and consistent with these differences, heterosexual dancers recalled their parents as more supportive of their dancing compared to gay men's parental recollections, t(91) = 3.5, p < 0.001, though even gay men recalled their parents as somewhat supportive. Gay men recalled being different from heterosexual men in other respects as well. On a scale of childhood gender typicality, gay men reported significantly more feminine behavior than heterosexual men, t(62) = 7.4, p < 0.001. Development of Male Sexual Orientation Because of the high rate of homosexuality among male dancers, we investigated developmental aspects of male sexual orientation, including the degree to which male dancers thought their sexuality had been influenced by their dance experiences. On average, gay men recalled that their first attractions to men occurred at the age of 10.4 years (SD = 3.7). They first considered the possibility that they were gay at 11.5 years (SD = 4.4). On average, their first homosexual encounter occurred at age 16.4 years (SD

Sexual Orientation and Professional Dance

439

= 4.8). Seventy-one percent of the gay male dancers also reported past sexual attraction to women, at an average age of 13.2 (SD = 5.1). The same percentage had had at least one heterosexual encounter, occurring at age 18.3 (SD = 3.7). Heterosexual men reported that on average, they were aware of their first heterosexual feelings at age 8.9 (SD = 3.6). They first had heterosexual relations at age 16.5 (SD = 2.0). Only 15% of the heterosexual male dancers said they had ever been attracted to a male, and only 7% had ever had sex with a male, with first occurrence at 20.5 years (SD = 4.9). We asked dancers several questions concerning their opinions about origins of their sexual orientation. Only one gay male dancer believed that his experiences in dance influenced his sexual orientation. He was the only gay male dancer who believed that he was not gay before beginning his dance career or who believed that sometimes dancers become gay due to the high prevalence of homosexuality within dance. However, all gay male dancers, including this one, believed that they would have eventually become homosexual even if they had never danced. Twenty-three percent of the heterosexual dancers agreed that "heterosexual male dancers struggle with their sexuality more than heterosexual men in other professions," and an additional 23% believed that this might be true. A high proportion of heterosexual male dancers, 71%, had sometimes felt that gay men had tried to influence their sexual orientation or "recruit" them into homosexuality. However, only 5% of the heterosexual male dancers believed that some male dancers become gay because of the prevalence of homosexuality in the dance world. Sexual Orientation and Social Relations Between Dancers We investigated the consequences of the high prevalence of gay male dancers for social relations between dancers. Not surprisingly, all gay male dancers agreed that the dance culture is especially accepting of homosexuality. When asked how their sexuality affected their relationships with heterosexual male dancers, 74% of the gay men said it had a positive or no effect, and 27% said it had a negative effect. Only 6% of gay men believed that their relationships with women suffered, a significantly lower percentage, x2(l) = 4.9, p < 0.05. Eighty percent of gay men believed that it is not uncommon for gay male dancers to have sexual relationships with other dancers. Only 21% of the heterosexual male dancers and 26% of the heterosexual female dancers agreed that the high percentage of gay male dancers created special problems for them. Although we did not code this response,

440

Bailey and Oberschneider

our impression is that women who mentioned special problems most often meant the tragedy of having many of their gay male friends die of AIDS. In contrast, heterosexual men's problems were less sympathetic. Eighty percent of heterosexual male dancers said that others often assumed that they were gay due to their profession, but only 26% of these said that it bothered them. Ninety percent of heterosexual male dancers reported that they had been sexually propositioned by a gay male dancer. Of these, 19% felt "upset," 47% felt "neutral," and 34% felt "flattered." When asked if they liked anything about the high prevalence of gay male dancers, 58% of the heterosexual men and 61% of the heterosexual women said "yes." Many of the heterosexual men, for example, mentioned that they liked gay men's sense of humor. Seventy-six percent of heterosexual women said that their friendships with gay male dancers were closer than their friendships with heterosexual male dancers. (Unfortunately, we did not ask this question of heterosexual men.) Seventy-nine percent of the heterosexual men believed that the high prevalence of homosexual male dancers made them more sought after (by women) as romantic partners; most (81%) heterosexual women agreed. Seventy-seven percent and 84% of heterosexual men and women, respectively, said that they had become more tolerant of homosexuality due to their dance experiences.

DISCUSSION Limitations Before discussing the implications of our results, we address two major limitations of our data. The first limitation concerns sampling. As already noted, we cannot claim to have either a random or representative sample of dancers. We recruited most of our participants because they were acquaintances of one of us (M. O.) or of other participants. Nevertheless, we doubt that our sample is biased in a way that seriously compromises our results. Most of our participants had danced in companies other than their current ones. Furthermore, it is our impression that substantial socializing and related exchange of information occurs between dancers in different companies. Thus, dancers' knowledge of homosexuality in the dance world was typically based on a fairly extensive database. A second potential concern is that much of our data are retrospective. It is possible, for example, that either gay men or heterosexual men, or both, systematically distort their childhood memories in ways that magnify real differences, or even create apparent differences when none exist. Elsewhere (see Bailey and Zucker, 1995), we have reviewed evidence concerning the

Sexual Orientation and Professional Dance

441

validity of memories of childhood sex-typed behavior and their association with sexual orientation, and concluded that available evidence largely supported their validity. Furthermore, our unsystematic impression is that the gay men in our study were not especially eager to support the stereotype that gay men are feminine. Although for these reasons we doubt that childhood recollections differed between gay and heterosexual male dancers solely due to retrospective biases, we cannot exclude that possibility. A related concern is that we asked dancers' opinions of why they or others became homosexual. People are not always aware of the major causes of their behavior, and thus, for example, a dancer's belief that he would have been gay regardless of his dance experiences cannot be conclusive. On the other hand, it is unlikely to be irrelevant either. Dancers may reasonably accurately recall, for example, whether their first homosexual feelings preceded their first awareness of homosexuality in dance. Prevalence of Gay Men in Dance Dancers estimated that well over 50% of professional male dancers are gay. Because of our study's methodological limitations, and because dancers' estimates varied widely, our average figure should not be taken seriously as a precise estimate. On the other hand, it does not seem plausible to argue with the basic conclusion that gay men are massively overrepresented among professional male dancers. Results of careful population surveys (e.g., Billy et al., 1993; Gebhard, 1972) suggest that perhaps 1-4% of American men have a homosexual orientation. Even the lowest estimate of gay men in dance provided by one of our participants, 25%, is much higher than even the relatively generous rate of approximately 10% applicable to men from Western urban areas (e.g., Johnson et al., 1992). Our results support the psychological rather than the sociological explanation of the increased prevalence of homosexuality among professional male dancers. (Of course, sociological and psychological factors could both contribute.) Although, consistent with a sociological explanation, participants believed that the dance world is especially tolerant of homosexuality, this has not led to any evident excess in the prevalence of homosexuality among female dancers. Gay men appear to be especially motivated to seek dance careers. Even as children, gay male dancers appear to have differed from heterosexual male dancers in important respects. First, they recalled more feminine behavior and interests. Indeed, the effect size associated with the difference between gay and heterosexual men in the present study, 1.9, was larger than the effect size in typical retrospective studies of gay and heterosexual men, 1.3 (Bailey and Zucker, 1995). Second, gay men

442

Bailey and Oberschneider

appear to have been more self-motivated to dance than heterosexual men and women, who were more likely to have been encouraged by their parents. Early parental intervention probably explains why heterosexual dancers began dancing slightly earlier than gay men. Assuming momentarily that gay men's increased presence among professional dancers is better explained at a psychological than a sociological level, what are the responsible psychological factors? Why are gay men so motivated to dance? One hypothesis is that gay men dance in order to be feminine. In other words, gay men dance because women do. An alternative hypothesis is that gay men and women share a common factor in their emotional make-up that makes dancing especially enjoyable. By this hypothesis, gay men's and women's desire to dance have similar causes. A third very general hypothesis is that gay men's increased motivation to dance has nothing to do with their desire to be feminine or their feminine traits. Results of this study cannot clearly distinguish these alternatives. There are at least two important related issues. First, to what extent is dance (more specifically, Western professional dance and its analogs) cross-culturally identified as a feminine activity? Several of our dancers suggested that in Russia and Latin America, ballet is a less stereotypically feminine pursuit and that as a result, professional male dancers from those regions are less likely to be gay. In contrast, Whitam (in press; see also Whitam and Dizon, 1971) has observed an association between male sexual orientation and dance in several diverse cultures. A complete understanding of the phenomena described herein requires systematic cross-cultural data on both heterosexual sex differences in dance and the association between dance and sexual orientation. Development of Male Sexual Orientation Results of our study generally support those of previous research (e.g., Bell et al., 1981) that male sexual orientation is most often determined early, by adolescence, and is not susceptible to influence by later experiences. If, as some (e.g., Patullo, 1992) have suggested, tolerance of and exposure to homosexuality increases the likelihood of its occurrence, then dancers should be especially likely to adopt a gay identity after entering the dance world. In contrast to this prediction, however, we found very few lesbian dancers, and our gay male dancers appeared to have experienced homosexual feelings long (on average 4 years) before they entered the dance world. The vast majority of the gay men believed that their experiences in dance had no effect on their sexual orientation. Nor did heterosexual male dancers believe that gay men successfully recruited

Sexual Orientation and Professional Dance

443

heterosexual men into homosexuality. As we have acknowledged, memories are not always accurate, and people are often unaware of causes that shape them. There was little indication from our study, however, that the high visibility and tolerance of male homosexuality in dance causes dancers who would otherwise be heterosexual to become gay.

Social Relations Among Dancers Professional dance is a useful laboratory to study interactions between gay and heterosexual dancers, as well as how dancers cope with any resulting conflict and tension. Our study suggests that few problems occur between gay male and heterosexual female dancers. Heterosexual men did occasionally complain that it was bothersome to them that others assumed that they were gay because of their profession. Furthermore, an appreciable minority minded the sexual advances they had received from gay dancers. Generally, however, heterosexual men appeared to be relatively unconcerned about such problems, and a majority believed there were benefits to the strong gay presence in dance. Thus, our results suggest that conflict between gay and heterosexual male dancers is neither serious nor intractable. It is conceivable, however, that heterosexual men most aggravated by such conflicts tend to abandon their careers and hence may be underrepresented.

Conclusions Results of this study strongly suggest that gay men are overrepresented among professional dancers. The most likely explanation for this is that gay men are especially interested in dance by the early age when future professionals often begin dancing, and are sufficiently interested to make the strenuous commitments necessary to succeed. Professional experiences did not appear to have influenced dancers' sexual orientations. Our participants did not report that the high percentage of gay male dancers caused them serious harm.

REFERENCES Bailey, J. M., and Zucker, K. J. (1995). Childhood sex-typed behavior and sexual orientation: A conceptual analysis and quantitative review. Dev. Psychol. 31: 43-55. Bell, A. P., Weinberg, M. S., and Hammersmith, S. K. (1981). Sexual Preference: Its Development in Men and Women, Indiana University Press, Bloomington.

444

Bailey and Oberschneider

Billy, J. O. G., Tanfer, K., Grady, W. R., and Klepinger, D. H. (1993). The sexual behavior of men in the United States. Fam. Plann. Perspect. 25: 52-60. Chung, Y. B., and Harmon, L. W. (1994). The career interests and aspirations of gay men: How sex-role orientation is related. 7. Vocat. Behav. 45: 223-239. Gebhard, P. (1972). Incidence of overt homosexuality in the United States and western Europe. In Livingood, J. M. (ed), National Institute of Mental Health Task Force on Homosexuality: Final Report and Background Papers, U.S. Government Printing Office, Washington DC. Green, R. (1987). The "Sissy Boy Syndrome" and the Development of Homosexuality, Yale

University Press, New Haven. Johnson, A. M., Wadsworth, J., Wellings, K., Bradshaw, S., and Field, J. (1992). Sexual lifestyles and HIV risk. Nature 360: 410-412. Murray, S. O. (1991). "Homosexual occupations" in Mesoamerica? J. Homosex. 21: 57-65. Newton, D. E. (1978). Homosexual behavior and child molestation: A review of the evidence. Adolescence 13: 29-43.

Patullo, E. L. (1992). Straight talk about gays. Commentary, 93(December): 21-24. Whitam, F. L. (in press). Culturally universal aspects of male homosexual transvestites and transsexuals. In Bullough, B., and Bullough, V. (eds.), Gender and Transgender Issues, Prometheus Books, Amherst, NY. Whitam, F. L., and Dizon, M. J. (1979). Occupational choice and sexual orientation in cross-cultural perspective. Int. Rev. Mod. Soc. 9: 137-149. Whitam, F. L., and Mathy, R. M. (1986). Male Homosexuality in Four Societies: Brazil, Guatemala, the Philippines, and the United States, Prager, New York.

Archives of Sexual Behavior, Vol. 26, No. 4, 1997

BOOK REVIEWS The Psychology of Gender. Edited by Anne E. Beall and Robert Stemberg. Guilford Press, New York, 1994, 278 pp., $28.95. Reviewed by M. Beth Casey, Ph.D.1

The Psychology of Gender is featured as addressing four critical questions in the field of gender and gender differences. One of those four is the question, How do heredity and environment interact to create gender differences? In the introductory chapter, Sternberg makes the further argument that the highest level question in relation to gender differences is, "To what extent are gender differences due to biology, to what to environment, and to what extent interaction between the two, as a function of time and place?" (p. 5). The reader is left with the initial impression that this volume is actually going to grapple with this complex question in interesting ways. Unfortunately, the subject of genetic-environmental interactions is then basically dropped except for four pages in one chapter on evolutionary pressures on gender differences and seven pages on childhood socialization. In general, this volume is essentially on social and cultural factors affecting gender differences. In this review, I summarize some points made in what is essentially a much less ambitious book, and then conclude by proposing a way that the topic of heredity-environmental interrelationships might be addressed. Contrasting with the lack of coverage of heredity-environment interactions, the concept of gender schemas is extensively woven throughout the book, and, in fact, is discussed in depth in 6 of 10 chapters. According to this cognitive-developmental perspective, gender is a socially constructed concept which evolves out of the tendency of humans to categorize and to stereotype. In a chapter on the social-psychological view of gender, Geis argues that in spite of conscious beliefs in gender equality, gender-based schemas unconsciously "interpret and guide our perceptions, inferences, 1Department

of Counseling, Development Psychology, and Research Methods, Boston College, Chestnut Hill, Massachusetts 02167. 445 0004-0002/97/0800-0445$12.50/0 C 1997 Plenum Publishing Corporation

446

Book Reviews

memories, and treatment of men and women" (p. 37). These biases invade what we notice, infer, and remember about one another and produce selffulfilling prophecies based on gender stereotypes. Geis cites the literature that shows, for example, that when the gender of applicants is randomly switched on resume summaries, there are systematic biases based on gender stereotypes. Geis considers the possibility of some biologically based gender differences, but concludes that most biological differences go across gender and are swamped by a cognitive system attuned to gender-based traits. Lott and Malusco also give scant credence to the importance of biological factors in their chapter on social learning, and imply that it is surprising that "Explanations for gender-related behaviors that call on innate predispositions continue to appear in the psychological literature" (p. 112). They discuss Lytton and Romney's (1991) meta-analytic review of the socialization literature only in passing, primarily citing it with regard to the one finding that is supportive of socialization theory. They never inform the reader of the nonsignificant findings that are directly contradictory to many of the underlying assumptions of social learning. Only in a later chapter on childhood socialization by Jacklin and Reynolds is it pointed out that Lytton and Romney examined 1250 studies, 172 of which they could use in their meta-analysis. They divided these studies into 19 areas of socialization (e.g., achievement encouragement, verbal interaction, restrictiveness, encouragement of dependency). In 18 of these areas, they found no evidence of differential socialization. In studies on populations in North America, in only one area (direct encouragement of gender-typed activities), did they find significant differences, (p. 199)

Cross and Markus, in their chapter on cognitive approaches to gender, stress the importance of examining individual variations in the extent to which gender schemas dominate behaviors and attitudes. Some situations are less likely to elicit gender schemas, namely, situations in which other categories are made more salient or when our cognitive resources are taxed. In addition, some individuals have strong gender schemas, where much of their perception and expectation of the world is based on this dichotomy, while for others this way of viewing the world is not so dominating. They argue that "People with self-schemas for masculinity or femininity will be sensitive to different information and events than people without these selfschemas, or than people who have both self-schemas as features of their self-concept." (p. 76) They note that Bem (1981a) originally proposed that individual differences in the robustness of gender schemas vary as a result of the degree to which this gender dichotomy has been emphasized during the socialization process. What is ignored in this conceptualization is how

Book Reviews

447

biologically based individual differences may interact with this socialization process to affect gender self-schemas. Now I will discuss the topic I had hoped would be addressed in this book—a proposal for ways in which heredity and environment factors interrelate to affect both within- and between-group gender differences. I suggest a mechanism for examining individual differences in gender selfschemas in terms of the interaction of biological and environmental factors and provide data based on this approach. Consistent with Sternberg's initial proposal in the introductory chapter, there is a recent theoretical movement to consider the complex ways that environmental factors and biological factors interrelate (Plomin, 1990; Wachs, 1992). Wachs called the research based on this conceptualization "Phase III" research, and argued that it is the wave of the future, since this type of research has moved to the point of addressing the interrelationship of biological and socialization factors through interactions, rather than only considering main effects. However, little systematic work has been done either to formulate theoretical models for this type of interrelationship or to provide specific research methodologies for investigating them. Casey & al. (1995a) adopted an interactive approach by investigating how socialization factors differentially affect the gender role identification of women with two different handedness patterns. We examined how biological predispositions based on the Geschwind, Behan, and Galaburda neuropsychological theory (assessed through handedness patterns) and socialization (assessed through subjects' perceptions of parental influence) interact to determine women's gender role identity. Geschwind and Behan (1984) and Geschwind and Galaburda (1987) posit that testosterone levels within the fetal brain are affected both by the maternal-fetal environment and genetic factors affecting both hormone levels and the permeability of the brain to these hormones. The level of testosterone that actually enters the fetal brain affects handedness patterns and laterality of language functions as well as masculinization and feminization. According to Geschwind and Galaburda (1987), the standard dominance pattern "involves strong left hemisphere dominance for language and handedness, and strong right hemisphere dominance for other functions," while "the term anomalous dominance refers to those in whom the pattern differs from the standard form" (p. 70). They state further that anomalous dominance is found most frequently among nonright-handers (left-handed and ambidextrous individuals), those with childhood learning disorders, and those right-handers with first-degree relatives in either of these categories. Geschwind and Galaburda estimate that over a third of the population is anomalous dominant. Thus, anomalous brain dominance is not "abnormal"

448

Book Reviews

or even unusual, but rather represents a less predictable pattern of brain organization than standard dominance. Geschwind and Galaburda also propose that anomalous dominant women are more likely to have increased masculinization and/or defeminization as a result of this exposure to elevated testosterone levels. Thus, they are suggesting a mechanism by which biological factors may have an impact on the formulation of gender role identity and gender self-schemas. They distinguish between masculinization and defeminization, stating that masculinizing effects lead to male-typical behavior in the female, while defeminizing effects lead to loss of female-typical traits. In our research, handedness and familial handedness patterns were used as markers for anomalous and standard brain dominance, representing biological influences on gender role identity. It was predicted that anomalous dominant women (nonright-handed females and right-handed females with nonright-handed relatives) would have a greater predisposition toward more masculine or less feminine gender role identity than would standard dominant women (right-handers with all right-handed relatives). Following our prior conceptualization of biological-environmental interactions using the bent twig approach (Casey and Brabeck, 1990), it was predicted that this effect on anomalous dominant girls would be moderated by socialization factors and would be expressed only among those anomalous dominant girls who perceived themselves as receiving parental permission to develop along this "bent." Our findings support the view that biological predisposition interacts with perceived parental influences to affect cross-gender role identity. Anomalous dominant women who perceived their parents as definitely giving them permission for cross-gender behavior demonstrated high masculine/low feminine characteristics on the Bem Sex Role Inventory (Bern, 1981b). The cross-gender role identity in this group differed significantly from the traditional gender role identity found in all other groups of females (including anomalous dominant women whose parents were perceived as less permissive and the women with standard dominance reporting both styles of parenting). An overview of this interactive research approach has been reviewed elsewhere (Casey, 1996a, 1996b; see also Halpern, 1996). I proposed a variety of ways that biological and environmental factors may interrelate to influence gender-related behaviors. I also suggested ways that our methodological approach can be used to examine the influence of biologicalenvironmental interactions on other types of behaviors. In addition to gender role development, we have used a similar approach to examine the interaction of environmental and biological factors among women on variations in one type of spatial skill, the ability to mentally rotate objects.

Book Reviews

449

This type of skill shows the largest cognitive gender difference, with males as a group excelling over females. The practical importance of individual differences in this type of spatial ability among females is presented elsewhere (Casey et al., 1995b). We showed that mental rotation ability predicts success on the math SAT among college-based samples, and that gender differences on the math SAT are eliminated when this type of spatial ability is statistically controlled. It is clear that variations based on gender have a major impact on the development of the individual and on the choices we make and successes we have in life. Thus, in my view it is important to move forward to disentangle the ways in which biological and environmental factors intertwine to influence gender-based behaviors—this is the research agenda that Sternberg proposed in his introductory chapter. It was frustrating that The Psychology of Gender did not build upon this initial promise, since Sternberg has the vision to break through the impasse in the field of gender differences to consider new methodologies for approaching gender issues from a heredity-environment interaction perspective.

REFERENCES Bem, S. L. (1981a). Gender schema theory: A cognitive account of sex-typing. Psychol. Rev. 88: 354-364. Bem, S. L. (1981b). Bem Sex-Role Inventory: Professional Manual, Consulting Psychologists Press, Palo Alto, CA. Casey, M. B. (1996a). Understanding individual differences in spatial ability within females: A nature/nurture interactionist framework. Dev. Rev. 16: 241-260. Casey, M. B. (1996b). A reply to Halpem's commentary: Theory driven methods for classifying groups can reveal individual differences in spatial ability within females. Dev. Rev. 16: 271-283. Casey, M. B., and Brabeck, M. M. (1990). Women who excel on a spatial task: Proposed genetic and environmental factors. Brain Cognit. 12: 73-84. Casey, M. B., Brabeck, M. M., and Nuttall, R. (1995a). As the twig is bent: The biology and socialization of gender roles in women. Brain Cognit. 27: 237-246. Casey, M. B., Nuttall, R., Pezaris, E., and Benbow, C P. (1995b). The influence of spatial ability on gender differences in math college test scores across diverse samples. Dev. Psychol. 31: 697-705. Geschwind, N., and Behan, P. (1984). Laterality, hormones, and immunity. In Geschwind, N., and Galaburda, A. M. (eds.), Cerebral Dominance: The Biological Foundations, Harvard University Press, Cambridge, MA, pp. 211-214. Geschwind, N., and Galaburda, A. M. (1987). Cerebral Lateralization, MIT Press, Cambridge, MA.

Halpern, D. F. (1996). Sex, brains, hands, and spatial cognition. Dev. Rev. 16: 261-270. Lytton, H., and Romney, D. M. (1991). Parents' differential socialization of boys and girls: A meta-analysis. Psychol. Bull. 109: 267-296. Plomin, R. (1990). Nature and Nurture: An Introduction to Human Behavioral Genetics, Brooks/Cole, Pacific Grove, CA. Wachs, T. D. (1992). The Nature of Nurture, Sage, Newbury Park, CA.

450

Book Reviews

Neither Man Nor Woman: The Hijras of India. By Serena Nanda. Wadsworth Publishing Company, Belmont, California, 1990, 170 pp., $23.95. Reviewed by Froukje M. E. Slijper, Ph.D.2

The hijras are a religious community of men who dress and act like women and whose culture centers on the worship of Bahuchara Mata, one of the many versions of the Mother Goddess worshipped throughout India. In connection with the worship of this goddess, the hijras undergo an operation in which their genitals are removed. The hijra emasculation operation is a ritual and consists of the surgical removal of the penis and testicles. The emasculation ritual is performed by a dai ma (midwife) who has no medical training. The dai ma is a hijra, often a guru (leader), who believes she operates with the power of the Mata so that the result is not in her hands. The nirvan (the one who is operated on) is in a state of trance during the emasculation operation so that he does not experience any pain. The operation is life-threatening and forbidden by Indian law. Because the operation is illegal and is always done in secret, there is no way to know how many hijras die as a result. Urinary tract infections are one of the most serious physical complications of the operation. The author was not allowed to witness an operation but has seen many results of the operation. The material in Nanda's book is based on tape-recorded narratives and conversations with at least 20 hijras, mainly from the three largest households in Bastipore. She also spoke to hijras who lived by themselves or with men they called their husbands. The data were collected in 19811982 during a stay of 8 months and return visits to India in 1985 and 1986. Because the hijras are often hostile to outsiders, from whom they frequently receive abuse, ridicule, or prurient curiosity, it was not easy for Nanda to get into contact with them and win their confidence. She started with hijras who were willing to talk with her and through them made contact with other hijras. Although the data collection could not be systematic, in view of the difficult circumstances, we should be grateful to the author for providing us with such a rich amount of material. Narratives and conversations with four hijras are presented in detail. Hijras join their community of their own free will, mostly after the onset of puberty, when they become aware of their sexuality. Often they first live for several years in the community before they decide to undergo the emasculation operation. The most important traditional role for the 2Department

of Child and Adolescent Psychiatry, Sophia Children's Hospital, Dr. Molewaterplein 60, 3015 GJ Rotterdam, The Netherlands.

Book Reviews

451

hijras is to perform at marriages or at homes where a male child has been born. The hijras bless the baby boy and his family and they provide entertainment (music, dancing, singing) for friends, relatives, and neighbors. This ritual is a source of income but often not enough to earn a living. As a consequence, many hijras live from prostitution. Although Nanda describes the attitude of Indian society toward the performances of the hijras as ambivalent, my impression is that the attitude is overtly rejecting—there is no respect for the hijras, only fear and contempt. Because many hijras are homosexuals and sometimes sexually provocative, they are teased and the victims of public mockery. The hijras force their services upon the family celebrating a marriage or birth because they need the money. The public fears the verbal and physical aggression of the hijras if they object to their services, and most of all they fear the threat of the hijras exhibiting their mutilated genitals. When interpreting the hijra culture, I look at it from my knowledge and experience as a psychoanalyst who treats children and adults with gender identity problems in a Western society, which differs from Nanda's perspective as an anthropologist. From this point of view, the performance of the hijras at marriages and births, which consists of the aggressive ridicule of women and worshipping of men, appears as an acting out of their anger and sorrow about having to live as a woman, a castrated man. In this performance, they mourn the loss of their male identity which is so much more valued in Indian society than female identity. At the same time, it appears to be an act of revenge on the society in which they are outcasts as a consequence of their hijra identity. For hijras, there is no way to have a normal professional life, they can only earn a living by performing and prostitution and often they are rejected by their family, so their only social life is with hijras. The author only mentions the aspect of revenge on society in the performances of the hijras, not the acting out of the psychological meaning of their castration. Nanda raises the problem of defining the gender identity of the hijras, since some told her they were neither men nor women, while others told her they felt and behaved like women. In Nanda's opinion, transsexuals in Western culture are a permanent gender category because they feel they are living in the wrong gender from birth onwards. The hijras, however, are a gender category that does not fit in the dichotomous and permanent gender categories of Western culture. I do not agree with this point of view. In analyzing the gender identity and sexual orientation from the narratives of the four hijras described in detail, from my Western point of view I can say the following: 1. Salima, who felt she was a "real" hijra because she had a physical intersex condition, the origin of which is not described by the author, could

452

Book Reviews

have had a normal life if she had been assigned at birth to the female sex and reared in that sex (Money et al., 1986; Slijper et al., 1996). From the literature, we know that individuals like Salima, who have to live as neither a woman nor a man, develop severe psychopathology because they cannot identify with a gender category (Money and Lobato, 1988). In contrast to other hijras, who are men until the moment they decide to become hijras, from birth on there was no identification possibility for Salima. She was neglected by her family and looked for comfort with the hijras, who treated her well but ostracized her when she made a mistake. 2. Meera, the guru, is the only hijra of the four who takes hormones (an over-the-counter preparation used by women to lessen menstrual cramps) to feel more feminine. For her, the motive behind the operation was to feel more like a woman. Meera had had several relations with men before she met her husband, with whom she developed a stable relationship. She also adopted a baby and tried to breastfeed him. Although she was able to lactate, breastfeeding failed. From the age of 4, Meera felt and behaved like a girl. Her cross-gender behavior was accepted by her parents until puberty. From the point of view of Western culture, Meera is a transsexual. 3. Kamladevi and Shushila, both 35 years of age at interview, were both feminine boys from the age of 5 onwards. Their cross-gender behavior was also tolerated until puberty. Shushila was not operated on and Kamladiva, who was operated at 33, was still ambivalent about it. Her motive for the operation was to obtain more social and economic security (only emasculated hijras are allowed to perform); the operation was not essential for her gender feelings. Kamladevi and Shushila were both feminine boys who developed into effeminate homosexuals. In conclusion, I can say that individuals with different gender identities and sexual orientations can join the hijras. The gradual process of becoming a hijra is determined more by social and economic factors than by gender identity factors. As we have seen, cross-gender behavior and cross-gender identification were already observed in the hijras from the age of 4 or 5, which means as long as their memory can reach. In Indian society, there is a strong boundary between the world of men and women. Men are not allowed to feel and behave like women. At puberty, they must prove their heterosexuality and being impotent in a heterosexual relationship makes them hijras. This means that for effeminate homosexual men and transsexuals there is no place in Indian society. The hijra culture gives them an opportunity to express their gender identity and offers them social and economic shelter. But I wonder how many of them really identify with the hijra culture, the essence of which is sexual asceticism and identification with the Mother Goddess. I also wonder how many would have had the

Book Reviews

453

emasculation operation if they could live as either an effeminate homosexual man or as a surgically corrected transsexual. I agree with Money, who states in the foreword: "On the criterion of surgery, hijras and transsexuals may gradually have more in common if ever the hijra communities incorporate modern plastic surgery for vaginoplasty as well as castration into their customs, and at the same time add feminization of the body by hormones to demasculinization effects of castration." (p. xiii) So I think that the majority of the hijras are either men (effeminate men) or women (transsexuals). From my work with the treatment of effeminate boys, I know that some of them, who are convinced at the start of therapy they will live in adult life as women, develop into homosexual men. Within these individuals, there appears to be variability of gender identity—sometimes they feel more masculine, sometimes more feminine—a fluctuation over which they can have control in their fantasy. Doom et al. (1994) make the assumption of two gender identity subsystems (a masculine and a feminine) that are present in any human being, and can have any relative strength. This results in two continua: one ranging from a strong feminine gender identity subsystem that is unconditionally expressed to weak unexpressed femininity, and one ranging from a strong and unconditionally expressed masculinity to a weak masculinity. We also know from experience with individuals with the intersex condition 5-alpha-reductase deficiency (Imperato-McGinley et al., 1979), who develop from females into males at puberty, that gender identity can be flexible. So I agree with Nanda that gender identity is not always fixed, only the concept of a third gender, neither man nor woman, which Nanda is proposing, underlines in my opinion the dichotomy. Nanda's book is challenging. It is written with respect and love for the hijras and reads like a novel.

REFERENCES Doom, C. D., Poortinga, J., and Verschoor, A. M. (1994). Cross-gender identity in transvestites and male transsexuals. Arch. Sex. Behav. 23: 185-201. Imperato-McGinley, J., Peterson, R. E., Gautier, T., and Sturla, E. (1979). Androgens and the evolution of male-gender identity among male pseudohermaphrodites with 5-alpha-reductase deficiency. New Engl. J. Med. 300:1233-1237. Money, J., Devore, H. M., and Norman, B. P. (1986). Gender identity and gender transposition: Longitudinal outcome study of 32 male hermaphrodites assigned as girls. J. Sex Marital Thar. 12:165-181. Money, J., and Lobato, C. (1988). Matched pair of siblings concordant for 46,XY hermaphroditism with female sex assignment, and discordant for erotosexual outcome. Psychiatry 51: 65-79.

454

Book Reviews

Slijper, F. M. E., Drop, S. L. S., Molenaar, J. C., and de Muinck Keizer-Schrama, S. M. P. F. (1996). Long-term psychological evaluation of children with genital anomalies. Manuscript submitted for publication.

Love, Sex, and Intimacy: Their Psychology, Biology, and History. By Elaine Hatfield and Richard L. Rapson. HarperCollins Publishers, New York, 1993, 520 pp. Reviewed by Pamela C. Regan, PH.D.3

Until recently, such "personal" experiences as love, sex, and intimacy were not viewed as legitimate topics of scientific inquiry. We tolerated songs, odes, and treatises on these experiences by poets, philosophers, and other starry-eyed sentimentalists, but we expected our rational, hard-headed scientists to know better than to waste valuable research time and money on such frivolous and irrational pursuits. It was such an attitude that led Senator William Proxmire to conclude that "right at the top of the list of the things we don't want to know is why a man falls in love with a woman and vice versa" (Hatfield and Walster, 1978, p. viii). But, millions of Americans want nothing more than an answer to that very question, as indicated by the proliferation of magazine "lust" quizzes that purport to measure our sexual desirability, talk shows devoted to discussions of what to wear, say, and do to attract and keep a romantic partner, and self-help books that promise to teach us in 10 easy steps how to rekindle the sexual ashes of our fading romances. Although this interest—both personal and professional—in interpersonal relationships is unlikely to disappear in the near future, there are notably few textbooks available that can serve as sources of general introduction to theory and research in this area (see, however, Brehm, 1992; Hendrick and Hendrick, 1992). Focusing specifically on romantic (i.e., dating, marital) relationships, Hatfield and Rapson's volume thus represents a particularly valuable addition to the existing literature. In keeping with the interdisciplinary nature of the field of interpersonal relationships, the authors draw upon theory and research from social psychology, developmental psychology, sociology, communication studies, marriage and family therapy, and clinical psychology. The judicious use of cartoons and quotations gleaned from writers, poets, biographers, historians, philosophers, and entertainers prevents the text from becoming a mere catalog of theoretical models and empirical findings, and seems likely to capture and maintain the interest of students as well as members of the 3Department

of Psychology, California State University, Los Angeles, California 90032.

Book Reviews

455

general public. Similarly, the authors' discussion of their own relationship histories in the Preface, coupled with their friendly, conversational style, renders this book well suited to classroom use and public consumption. Indeed, Hatfield and Rapson seem to have carefully considered these two audiences in constructing their book, for they appear as invested in teaching their readers how to understand and apply what is known about love, sex, and intimacy to their own lives and relationships as they do in reviewing the literature on those topics. For example, chapter 1 ("Beginnings") includes not only a discussion of the variables that influence interpersonal attraction (e.g., proximity, similarity, physical attractiveness) but also a section entitled, "Effective Strategies for Meeting Dates and Mates," which outlines some of the various approaches that might be used to successfully attract potential romantic partners. Similarly, chapter 10 ("Dealing With Problems: Communication") reviews theory and research on verbal and nonverbal communication and also provides concrete examples of ways in which to effectively transform vague gripes about the partner or the relationship into specific complaints and to then turn these more specific complaints into positive suggestions for change. In short, the authors have achieved an interesting and ultimately successful blend of literature review and self-help book. Although this may render the work less interesting to researchers than to other, nonprofessional audiences, it is refreshing to find well-known and influential psychologists who do not shy away from discussing the potential applicability of relationships theory and research to the lives and loves of the man or woman on the street. This book is also nicely organized. Each chapter includes an introductory overview, definitions of the phenomena to be discussed (e.g., passionate love, intimacy, power), reviews of relevant research on the causes, manifestations, consequences, and correlates of those phenomena, and major conclusions and implications of this research for the individual and his or her romantic relationships. The book itself follows a similarly orderly progression. Chapter 1 focuses on the initial stages of a romantic relationship and introduces readers to theory and research on the factors that attract us to one another. Chapters 2 through 6 explore particular topics in the relationship literature, including passionate love, sexuality, companionate love, intimacy and commitment, and power. For example, in chapter 6, Hatfield and Rapson discuss how researchers have defined and operationalized power, review the ways in which individuals express power and resolve conflict in interpersonal situations, and examine the different power tactics and strategies employed by men and women in their sexual encounters and romantic relationships. The next three chapters deal with the many variables that conspire to create tension and problems between the members of a couple. Specifically, chapter 7 ("Emotional Problems") discusses

456

Book Reviews

personal or individual factors that may adversely affect a relationship, including depression, anxiety, and anger; chapter 8 ("Relationship Problems") examines the influence of such interpersonal factors as inequity, conflict, and alcoholism; and chapter 9 ("The Larger World") reviews the impact of social factors on a couple's relationship, including parental interference, the presence of children and romantic or sexual rivals, and career demands. The next three chapters take a more optimistic turn, focusing not on the antecedents and consequences of interpersonal problems but rather on the management and repair of love relationships gone awry. Chapter 10 presents different methods of understanding and dealing with emotional problems, chapter 11 overviews various strategies for coping with conflict and jealousy, and chapter 12 discusses ways in which to more effectively communicate with one's partner. Replete with interesting and germane quotations and anecdotes, the final two chapters focus on the process of relationship dissolution (chapter 13) and on the personal, economic, and interpersonal consequences of breakup, divorce, and partner death (chapter 14). I thoroughly enjoyed this book and do not hesitate to recommend it to researchers, students, and nonprofessionals. Of particular note is the chapter on passionate love, which is one of the few treatments of the topic to discuss this experience as it occurs in children. I did have a few minor quibbles. In particular, chapter 3 ("Sexuality") seemed less "interpersonal" in orientation than the other chapters. The lion's share of this chapter is devoted to sex differences between heterosexual men and women (e.g., differences between men and women in the developmental timing of various sexual behaviors, the types of sex fantasies utilized during masturbation, the experience of orgasm). Although sex differences in the area of sexuality and romantic relationships undoubtedly exist and have important interpersonal consequences, little attention was given to delineating these consequences and to exploring sexuality as it is experienced within the confines of close relationships; that is, as it is experienced not by men and women in general but by relationship partners. It should be noted that one such topic, sexual coercion in relationships, is discussed in some depth in chapter 6 ("Power"). An additional note, again involving chapter 3, concerns the discussion of pornography. Although the authors state that most of the pornographic material written during the 1950s and 1960s was designed for a male audience and was consequently disliked by women, they only include four very explicit passages from a novel by Henry Miller that are indeed "threaded through with strands of male power, sexuality, and hatred of women" (p. 93). No complementary pornographic material written for a female audience is included, although the authors do discuss gender differences in re-

Book Reviews

4S7

sponses to different types of pornography. Readers may be offended by these explicit passages (indeed, the book is interspersed with sexually explicit passages, although these are for the most part consensual, egalitarian, and nonviolent in tone). A final point concerns the lack of discussion given to homosexual relationships. In the Preface, Hatfield and Rapson state that their text will discuss research and theory on the interpersonal experiences of both heterosexuals and homosexuals. However, other than a brief discussion of the sexuality of homosexual men and women, the book is noticeably reticent on the homosexual experience in close relationships. This may be due in part to the fact that the authors are reviewing areas of theory and research that have focused on the heterosexual experience (for discussion, see Herek et al., 1991); at the very least, this deserves mention. In sum, these issues aside, Hatfield and Rapson have again made a significant contribution to the interpersonal relationships literature. The book is well written, entertaining, empirically and theoretically rich, and will prove a valuable resource for researchers, students, and nonprofessionals.

REFERENCES Brehm, S. (1992). Intimate Relationships, 2nd ed, McGraw-Hill, New York. Hatfield, E., and Walster, G. W. (1978). A New Look at Love, Addison-Wesley, Reading, MA. Hendrick, S. S., and Hendrick, C. (1992). Liking, Loving, & Relating, 2nd ed., Brooks/Cole, Pacific Grove, CA. Herek, G. M., Kimmel, D. C, Amaro, H., and Melton, G. B. (1991). Avoiding heterosexist bias in psychological research. Am. Psychol. 46: 957-963.

Archives of Sexual Behavior, Vol. 26, No. 4, 1997

Letter to the Editor

HOMOSEXUALITY AND DANCE: RELATION TO AIDS AND HAWAI I

To the Editor: The study by Bailey and Oberschneider (1997, this issue) looks at relationships between sexual orientation and dance. The hypothesis under investigation essentially asks: "Are there known relationships between homosexuality and dance as a profession?" This prompts me to write of an association I became aware of several years ago. An acquaintance of mine, Thomas J. Aguilar, died in Honolulu in May 1993 from AIDS complications. In H a w a i i Aguilar was open about his homosexuality and his AIDS condition. He was active in educating about the disease and fighting against homophobia. Our professional association was primarily related to my capacity as Co-Founder and Director of the Hawai i AIDS Task Group. Aguilar was a showman with talents in acting, singing, and dance who also choreographed and directed. He came to prominence for his performances as a lead character in the London and Broadway productions of "A Chorus Line." The gay newspaper, New York Native, in an obituary (26 May 1993) reported on Aguilar's accomplishments and noted that to one of his interviews he talked of his HIV infection. In that interview he mentioned that of the 17 original men cast in "A Chorus Line," only he and one other still were alive. Now he too is gone. The musical opened on Broadway at the Public Theater on May 21, 1975. This was some half-decade before the AIDS plague was noted in the United States. In the 20 years since the play's opening, however, 16 of the 17 men fa the cast have been reported to have died of AIDS. Nothing is known about the drug taking, blood transfusion, or surgery histories of these 17 men and I am not privy to the sexual orientation of the male cast members. One can, however, speculate on the high correlation between AIDS-related death, homosexuality, and dance. A second separate item is related to the hypothesis. In traditional Hawai ian dance, men and women danced, but not together. As with many other matters of life, dance was associated with religion and to such things the sexes were often kept separate, e.g., food preparation and eating was 459 0004-0002/97/0800-0459$12.50/0 C 1997 Plenum Publishing Corporation

460

Letter to the Editor

also associated with the kapu system and men and women were not allowed to eat together although they could have almost any type of sex together (Diamond, 1990). Sexual expression of all types was accepted in male and female dancing and the medium was used to tell erotic stories, for sexual invitation, celebrating religious and social events, to perpetuate oral history, and more. Essentially, thus in traditional pre-Captain Cook Hawai i, all males and females were expected to know something about dance. In many of these dances the male displays were often, without doubt, what we in present day society would consider very masculine. They might depict aggression, battles, and such. The female dances often were about love, sacrifice, relationships. Dance (hula; hulahula) itself was sacred and protected under the goddess Laka. According to tradition, Laka would take mild possession of the dancer—become one with the dancer—and dance through him or her. During hula training the dancer was dedicated to Laka and "he or she had to stay absolutely away from the opposite sex until after graduation" (Pukui et al., 1972; emphasis mine). Nothing is recorded about same-sex relations but it could be assumed that that too was prohibited. After graduation any type of sexual activity could be resumed. During the Makahiki festival, the celebration of the New Year according to the Hawai ian calendar, dance was a major factor. The dances often had an erotic character and the transition from idea to deed was often expected to follow. "The dancers, whatever their sex, cannot refuse the sexual advances of the spectators they have aroused." (Valeri, 1985: emphasis mine). In contemporary Hawai i society, the situation is quite different. Now the typical traditional type dance group (hula halau) is filled by females.1 Many male troupes do, nevertheless, exist both separately and as part of a coed halau. But even in these halau, the men and women dance separately; only for tourist shows might they dance together; never for a Hawai ian audience. These groups dance for professional and recreational purposes. The general impression among dancers and the general population is that a high proportion of the male dancers are gay and that among the male leaders and teachers of the groups, the kumu hula, almost all are gay. This is said matter-of-fact without any stigma or condescension. It might be added that homosexual and ambisexual activities, as indeed almost any sexual activity, were accepted without stigma in traditional Hawai i (Diamond, 1990). 1The

term hula halau originally referred to the structure where the hula was taught. Usually the majority of females learned to dance and the majority of males learned chanting and musical accompaniment (Handy, 1965).

Letter to the Editor

461

Milton Diamond, Ph.D. University of Hawai i-Manoa John A. Bums School of Medicine

Department of Anatomy & Reproductive Biology Pacific Center for Sex & Society 1951 East-West Road, Honolulu, Hawai i 96822

REFERENCES Bailey, J. M., and Oberschneider, M. (1997). Sexual orientation and professional dance. Arch. Sex. Behav. 26: 427-438. Diamond, M. (1990). Selected cross-generational sexual behavior in traditional Hawai i: A sexological ethnography. In Feierman, J. (Ed.), Pedophilia: Biosocial Dimensions, Springer-Verlag, New York, pp. 422-443. Handy, E. S. C (1965). Religion and education. In: Handy, E. S. C, Emory, K. P., Bryan, E. H., Buck, P. H., Wise, J. H., et al., Ancient Hawaiian Civilization: A Series of Lectures Delivered at the Kamehameha Schools, Rev. ed., Charles E. Tuttle, Rutland, VT, pp. 47-59. New York Native (1993). Obituary, 26 May, Aguilar, Thomas J. Pukui, M. K., Haertig, E. W., and Lee, C. A. (1972). Nana I Ke Kumu (Look to the Source], Vols. 1 and 2, Hui Hanai: Queen Lili uokalani Children's Center, Honolulu. Valeri, V. (1985). Kingship and Sacrifice: Ritual and Society in Ancient Hawaii (P. Wissing, trans.), The University of Chicago Press, Chicago.