Immigrant Integration: A Cross-National Study (The New Americans)

The New Americans Recent Immigration and American Society

Edited by Steven J. Gold and Rubén G. Rumbaut

A Series from LFB Scholarly

This page intentionally left blank

Immigrant Integration A Cross-National Study

Frank van Tubergen

LFB Scholarly Publishing LLC New York 2006

Copyright © 2006 by LFB Scholarly Publishing LLC All rights reserved. Library of Congress Cataloging-in-Publication Data Van Tubergen, Frank, 1976Immigrant integration : a cross-national study / Frank van Tubergen. p. cm. -- (The new Americans) Includes bibliographical references and index. ISBN 1-59332-153-8 (alk. paper) 1. Immigrants--Cultural assimilation--Cross-cultural studies. I. Title. II. Series: New Americans (LFB Scholarly Publishing LLC) JV6342.V36 2006 305.9'06912--dc22 2006014610

ISBN 1-59332-153-8 Printed on acid-free 250-year-life paper. Manufactured in the United States of America.

Contents

1. Introduction

1

2. Theory

7

3. Data and Methods

43

4. Employment

73

5. Occupational Status

101

6. Self-Employment

117

7. Language

139

8. Religion

167

9. Conclusions

193

Appendix A: Micro-Level Data

213

Appendix B: Macro-Level Data

223

References

235

Index

257

v

This page intentionally left blank

CHAPTER 1

Introduction

In many Western countries, there is a deep concern among policy makers and social scientists about the integration of the foreign-born population. Mexican immigrants in the United States are a case in point. The 2000 Census of Population shows, for instance, that 50 percent of the Mexican immigrants do not speak English well. Furthermore, 9.4 percent of the foreign-born Mexicans are unemployed, as against 5.6 percent among natives. The average income for adult Mexican immigrants is about half that of natives. Why are Mexican immigrants so poorly integrated in the United States? At the other side of the Atlantic Ocean, similar questions are raised in Europe about other immigrant groups. In the Netherlands, for instance, the integration of Moroccans attracts major attention. About 44 percent of Moroccans do not speak the Dutch language well (Dagevos, Gijsberts and Van Praag 2003). Furthermore, 13 percent of the foreign-born Moroccans were unemployed in 2000, compared to 3 percent of natives (SCP, WODC, CBS 2005). And the income of Moroccans is about 65 percent that of natives (SCP et al. 2005). Why are Moroccans not well incorporated in the Netherlands? This book tries to answer these questions. The main thesis is that in order to understand the integration of an immigrant group, we need to compare the experience of that group with other groups, both within and across nations. A better understanding of the poor integration of Mexicans is obtained by contrasting their situation to that of other immigrant groups in the United States, some of which speak better English and fare better in the labor market. Examples are immigrants of Chinese, Cuban, Japanese, Indian, Italian and Philippine origin. Moreover, this book maintains that more can be learned about the 1

2

Immigrant Integration

Mexican experience in the United States by comparing it with the situation of Mexicans and other immigrant groups abroad. How well are Mexican immigrants integrated in Australia and Canada? How does the problematic experience of Mexican immigrants in the United States relate to the problematic situation of Moroccans in the Netherlands? And what can we learn from comparing the integration of other immigrant groups cross-nationally? At first sight, it would seem impossible to answer these questions. How could one compare the experience of Mexicans in the United States with that of Chinese, Cubans and Japanese –all of whom migrated to the United States for different reasons, in different periods, and with different skills? Even more problematic is the question of comparing Mexicans in the United States with other immigrant groups in other countries. Indeed, many studies have focused on a single immigrant group in a single country to learn more about the causes of immigrant integration. This book takes a different standpoint. Rather than conceiving group differences within and between nations as problematic, it maintains that examining these differences provides an important way to understand the experience of a single immigrant group in a single country. The main insight is that groups and countries can be fruitfully compared by looking systematically at their characteristics. This study argues that the specific situation of an immigrant group, such as the Mexicans in the United States, is an outcome of three kinds of contextual characteristics. First, the integration of immigrants is affected by characteristics of their country of origin. Immigrants are socialized in their country of origin, which can have an enduring effect on their integration in the destination country. Economic and political conditions in the country of origin also affect the migration motive and the intentions to migrate permanently. For those reasons, the integration of immigrants strongly differs between origin groups. These kinds of characteristics are labeled “origin effects” in this book, and they reflect the role of the country of origin irrespective of the country of destination. If they play an important role in immigrant integration one should observe, for instance, that Mexicans are not only poorly integrated in the United States but equally so in other countries. It can also be that differences across immigrant groups do not travel across receiving countries. Although Mexican immigrants speak the

Introduction

3

host language less well than other groups in the United States, they possibly speak the language considerably better than other groups in Australia and Canada. These group differences that do not “travel” across countries would result from characteristics that are specific for an immigrant group in a certain destination, such as the size of the community. These are called “community” effects and they reflect the role of the immigrant community. Third, receiving countries affect the integration of immigrants as well. For example, host countries differ in their immigration policies, leading to different selection of their immigrant populations. Countries also differ in the degree and direction to which they aim to integrate immigrants. Some countries are more in favor of an assimilation policy, whereas other nations adopt a multiculturalism model of integration. In this study the influence of these and other characteristics of host countries are referred to as “destination effects.” This book examines the importance of the country of origin, the country of destination and the immigrant community for the integration of immigrants. It introduces a novel approach to answer these questions. It presents a “double comparative” research design, which studies the integration of multiple immigrant groups in multiple destinations at the same time. In this way, it studies immigrants from a variety of origins, both Western and non-Western, Christian and nonChristian, refugees and labor migrants, rich and poor. The integration of multiple immigrant groups is studied in 19 Western countries, including “classic” immigration countries (Australia, Canada, and the United States) as well as “new” immigration countries in Europe (Austria, Belgium, Denmark, Finland, France, Germany, Great Britain, Greece, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, and Sweden). DEMOGRAPHIC OPPORTUNITIES Two demographic changes make the double comparative approach to immigrant integration especially attractive: many more countries are affected by immigration, and immigrant populations have become more diverse. Since the 1960s, migration flows to Western countries have increased substantially and have affected more countries than ever before (OECD 2001). In the United States, for instance, the number of

4

Immigrant Integration

immigrants granted legal permanent resident status rose from about 450,000 per year in the period of 1971-1980, to more than 900,000 per year in the period of 1991-2000 (Castles and Miller 2003). Likewise, in Canada, the number of entries increased from 89,000 in 1983 to nearly 190,000 in 1999. The 1996 Canadian census counted about 5 million immigrants, which amounts to 17 percent of the total population. While the classic immigration countries again experienced largescale immigration, the immigrant population has increased in European countries as well. In Germany, for example, the total number of foreigners increased from 686,000 in 1960 to more than 5 million in 1990 (Castles and Miller 2003, OECD 1994). In addition, countries in the south of Europe that were once major sources of migration, like Greece and Italy, have been transformed into migrant-receiving nations (Massey et al. 1993). Given that many more countries are lately being affected by immigration it becomes especially attractive to examine the role of characteristics of the receiving country in immigrant integration. There is a second demographic change that provides an opportunity for double comparative research. International migration has not only grown in volume and scope but it has also changed in character. Migration flows to Western countries have become more diverse. That is, immigrants today originate from a variety of countries (Castles and Miller 2003). Whereas in the period of the first great migration (i.e., 1880-1924) the United States imported predominantly immigrants from Europe, nowadays it also attracts a sizable number of immigrants from Latin America (e.g., Cuba, Mexico) and Asia (e.g., China). This provides a unique opportunity to compare the integration of a variety of immigrant groups in the United States, which differ in characteristics of their country of origin and their immigrant community. Increasing diversity among the immigrant population is also observed in European countries. For example, in the 1950s, the Netherlands predominantly received immigrants from its former colonies (e.g., Indonesia). In the 1960s, the Netherlands imported labor immigrants from Mediterranean countries (e.g., Italy, Morocco, Spain, Turkey). Since the 1980s it has attracted political immigrants from a number of countries (e.g., Afghanistan, Iran, Iraq), and after the collapse of the Soviet Union, new immigrants from Eastern Europe arrived (e.g., from Poland). Nowadays, like many other countries in Europe, the Netherlands is a heterogeneous society consisting of immigrants from multiple origin countries.

Introduction

5

Immigration is not a temporary issue. On the contrary. In view of family reunification, globalization, growing inequalities between the North and the South, political and ethnic conflicts, and the creation of free trade areas, international population movements are expected to intensify in the future (Castles and Miller 2003). Moreover, many more countries will be affected by immigration in the future, and the immigrant population will become even more heterogeneous. Given the recent demographic trends, as well as the expected changes in the migration pattern in the future, it is imperative to look at immigrant integration from a comparative perspective. OUTLINE OF THE STUDY The structure of this study is as follows. Chapter 2 reviews the theories and findings on immigrant integration. It argues that research on immigrant integration needs to consider immigrants’ country of origin, country of destination, and the immigrant community. It shows that an “assimilation” or “micro-level” approach to immigrant integration is inadequate. For example, research findings in the United States show that Mexican immigrants speak the language less well and have a poor economic position, even when individual characteristics are taken into account. These empirical difficulties for the assimilation theory call for theories that also incorporate immigrants’ country of origin, country of destination, and the immigrant community. The chapter outlines how to understand such “macro” differences in immigrant integration. Chapter 3 provides the methodological basis of the book. It explains the double comparative design in more detail, and discusses how it deviates from comparative designs adopted in earlier research. Furthermore, Chapter 3 lists problems addressed in cross-national research. These are problems of data collecting, comparability, and statistical analysis. It will be explained how these problems are tackled in this research. In subsequent chapters empirical analysis is conducted in order to show the importance of the country of origin, the country of destination, and the immigrant community. Different aspects of immigrant integration are examined. Chapter 4 focuses on labor-force activity and unemployment. It is first examined whether immigrants are actively participating in the labor market and trying to find employment, as opposed to being

6

Immigrant Integration

engaged in housekeeping, obtaining additional schooling, suffering from a disability, or taking (early) retirement. Subsequently, it is studied whether the immigrants seeking employment are successful in finding jobs or whether they remain unemployed. The two subsequent chapters examine the working population in more detail. Chapter 5 investigates immigrants’ occupational status by examining if immigrants who have a job succeed in finding a job with either higher or lower prestige. Chapter 6 studies immigrants’ self-employment. Researchers have argued that, because mobility in the labor market is restricted for immigrants, self-employment is used as an alternative avenue for immigrants’ economic mobility (Raijman and Tienda 1999b; Sanders and Nee 1996). Next to considering these ‘economic’ aspects of integration, two later chapters deal with the ‘cultural’ dimension of integration. Chapter 7 focuses on immigrants’ language proficiency. Language is a crucial part of culture, and learning the language of the host country is therefore an important aspect of cultural integration. Furthermore, language proficiency is associated with better interethnic relations in a country (Espenshade and Calhoun 1993; Gordon 1964). It is important to study language acquisition because many immigrants do not speak the official language of the host society well (Portes and Rumbaut 1996). Chapter 8 studies immigrants’ religious affiliation (i.e., whether immigrants think of themselves as members of a religious community, denomination, or religion) and immigrants’ religious participation (i.e., the frequency with which immigrants attend religious meetings). Religion is a core element of culture, and plays a pivotal role in many societal issues, in people’s opinions on family-life and politics, and in people’s life courses. Chapter 9 summarizes the main findings of this book and gives suggestions for future comparative research on immigrant integration.

CHAPTER 2

Theory

The consequences of the growing share and diversity of immigrants in Western societies have been much discussed in public and scientific circles. A number of questions have been raised on immigrant integration, questions that are directly linked to the classical topics of social cohesion and inequality studied in the social science literature. With respect to social cohesion or “socio-cultural integration” researchers have looked at ethnic intermarriage, religiosity, language proficiency, residential segregation, and ethnic identity (Alba and Nee 2003). Socio-cultural integration refers to two closely related aspects of immigrant integration: social integration (the extent to which immigrants interact socially with natives), and cultural integration (the degree to which cultural values and patterns are shared among immigrants and natives). Immigrants are said to be more socially and culturally integrated in the host country when, for example, interethnic contacts, friendships, and marriages are common, and when immigrants speak the destination language well. In addition, researchers have studied ethnic inequalities or “economic integration” in terms of unemployment, occupational status, self-employment and income (Alba and Nee 2003). Economic integration indicates the degree of economic equality between immigrants and natives. It is assumed that the economic integration of immigrants is stronger, when they have higher participation rates, lower unemployment levels, better jobs and a higher income. In this chapter, I review theories and findings on immigrant integration. I do so by looking at the literature on immigrants’ sociocultural and economic integration simultaneously. Although these dimensions of immigrant integration differ from one another, each having their specific causes and research literature, I show that research 7

8

Immigrant Integration

traditions in both fields have actually followed a similar path of development. Briefly stated, I will argue that initially researchers theorized that immigrants would gradually assimilate both socioculturally and economically. Subsequently, I show that for both dimensions of integration, the assimilation theory failed to tell the whole story. More specifically, it turned out that immigrants’ integration also depends on macro-level factors: the role of the immigrant group and the influence of the receiving context. These research findings stress the importance of the role of the country of origin, the country of destination and the immigrant community, which are the main interest of this book. Finally, I discuss how several researchers have attempted to incorporate the role of macro-level factors by developing a specific immigrant framework, whereas others have relied on general theories. I discuss how the general theories will be used in this book. ASSIMILATION THEORY Empirical Success Research on integration of immigrants developed shortly after the first great migration to the United States (i.e., 1880-1920). It was there that sociologist Park lay down the now famous idea of assimilation: “a process of interpretation and fusion in which persons and groups acquire the memories, sentiments, and attitudes of other persons and groups and, by sharing their experience and history, are incorporated with them in a common cultural life” (Park and Burgess [1921] 1969:735). Park suggested that over their life-course, immigrants would gradually assimilate to the mainstream culture of the host society. At the end of World War II, Warner and Srole (1945) elaborated on Park’s idea. They argued that, next to immigrants’ life-course, another major motor of assimilation is the succession of generations. With each successive immigrant generation, they argued, people would increasingly adopt the cultural elements of the host country. At the beginning of what is called the second great migration to the United States (i.e., from the 1960s onwards), sociologist Gordon (1964) further extended the idea of assimilation and gave it concrete forms.

Theory

9

Whereas Park used assimilation to refer to cultural processes in a rather vague manner, Gordon argued that this is only one dimension of assimilation. He suggested to make a distinction between seven dimensions instead (i.e., cultural, social-structural, marital, identificational, attitudinal, behavior receptional, and civic) and proposed concrete operationalizations. In this way, Gordon considerably increased the theoretical scope and usefulness of the assimilation idea. Remarkably, however, Gordon did not consider economic assimilation, although subsequent researchers have done so (Alba and Nee 1999, 2003). The central question of the researchers working within an assimilation framework was: “how does the integration of immigrants change over their life-course and between generations?”. From the 1970s onwards, researchers have used the assimilation theory as an answer to this question. Although they have not always expressed their ideas quite clearly, three rather concrete hypotheses were subsumed under the assimilation idea. It was hypothesized that integration will be higher among (1) immigrants who arrived at a younger age, (2) immigrants who have resided longer in the host country, and (3) successive immigrant generations.1 The assimilation theory has received ample empirical support in studies on immigrants’ socio-cultural and economic integration, in a wide number of countries, over a large time span, and relying on different sources. With regard to immigrants’ economic integration, studies conducted in the United States found out that during the first few years in the host country, immigrants were more often unemployed than natives, and that this gap diminished with length of stay in the host country (Chiswick 1982; Chiswick, Cohen, and Zach 1997; Chiswick and Hurst 2000). In studies on earnings, it was found that over their life-course (Chiswick 1978, 1979; Chiswick and Miller 2002) and

1

It should be remarked that several other ideas are also subsumed under the assimilation framework, most notably those that deal with successive stages of integration and the relationship between forms of integration. One idea is that immigrants who are integrated in terms of one dimension will be more likely to integrate in terms of other dimensions. For example, Logan and Alba (1993) argued that the higher the income of immigrants (“economic integration”), the more likely they are to settle in non-ethnically segregated neighborhoods (“spatial integration”). I do not treat these versions of assimilation theory here.

10

Immigrant Integration

across generations (Kalmijn 1996; Neidert and Farley 1985) immigrants gradually catch up economically with the native population. Researchers has shown that immigrants’ proficiency in the destination-language decreases with their age at the time of migration (Carliner 2000; Chiswick and Miller 1995, 1998; Dustmann 1994, 1997; Stevens 1999), increases with the length of residence in the destination country (Chiswick and Miller 1994, 1995, 1998, 2001; Evans 1986), and increases with successive generations (Veltman 1983). Confirmations for the assimilation idea were also found with regard to other measures of socio-cultural integration, such as ethnic intermarriage. In studies conducted in the United States (Lieberson and Waters 1988) and Australia (Jones and Luijkx 1996), it was found that immigrant groups that are more recent in the host society have lower intermarriage rates. In addition, Rosenfeld (2002) showed that intermarriage rates among first generation immigrants in the United States are lower than those of their children. A growing cultural assimilation with successive generations has also been documented in terms of ethnic identification with the host society (Alba 1990). Empirical Problems Although numerous studies confirmed the predictions of the assimilation theory, many were confronted with findings that the theory could not explain. Most of these findings do not contradict the assimilation idea, however. That is to say, researchers rarely found that over the life-course or with successive generations, the socio-cultural and economic integration of immigrants declined. The basic problem for the assimilation theory was that it turned out not to tell the whole story about the integration of immigrants. What happened was that researchers discovered important macro differences in integration: between immigrant groups, and between immigrants in different destinations (i.e., cities, regions, countries). Although these macro differences could be partially attributed to different composition in terms of assimilation factors (i.e., age at the time of migration, length of stay, generation) and demographic characteristics (e.g., gender, marital status), the theory could not explain that, after considering the influence of these assimilation factors, macro differences persist. These macro differences have been observed in many countries, among a variety of

Theory

11

groups, and with respect to a number of indicators of both sociocultural and economic integration. Research on immigrants’ language proficiency in the United States found that Mexicans have lower English proficiency than other origin groups, even after individual characteristics were taken into account (Carliner 2000; Veltman 1983; Portes and Rumbaut 2001). Likewise, in Germany, Dustmann (1994, 1997) observed that Spanish and Turkish immigrants had lower proficiency levels than Yugoslavian immigrants. Similar group differences in destination-language acquisition have also been observed in Australia (Chiswick and Miller 1996; Evans 1986), Belgium (Lesthaeghe 2000), Canada (Lieberson 1970), France (Tribalat 1995), Israel (Beenstock 1996), the Netherlands (Tesser, Merens and Van Praag 1999), Norway (Hayfron 2001), and the United Kingdom (Modood et al. 1997). Group differences were also found with respect to another core aspect of socio-cultural integration: ethnic intermarriage. In two classical studies of intermarriage in New Haven, Connecticut (the United States), Kennedy (1944, 1952) showed that in-marriage rates differed between the seven groups considered. For instance, averaged over the period 1870 to 1950, out-marriage rates of men were 16% among Italians, and 61% among Scandinavians. In an analysis of nine groups on a national level, Pagnini and Morgan (1990) showed that the percent outmarried in the United States in 1910 was 19 among Italians and 32 among Scandinavians. Similar group differences were reported in other studies conducted in the United States (Kalmijn 1993; Lieberson and Waters 1988), as well as in studies done on ethnic intermarriage in Australia (Jones and Luijkx 1996), Belgium (Lievens 1998), and Sweden (Cretser 1999). Group differences were also observed with regard to immigrants’ economic integration. Studies of the early waves of migration to the United States at the turn of the 20th century already showed that some groups, such as the Russian Jews, performed especially well economically, whereas other groups were less successful in the labor market (Perlmann 1988; Thomas and Znaniecki 1958). More recent studies of immigrants in the United States have again stressed the influence of the group on immigrants’ economic integration (Borjas 1999; Jasso and Rosenzweig 1990a). Similar country studies, reporting differences in the economic integration among (different) origin groups, have appeared in other countries as well (Basavarajappa and Jones 1999; Bell 1997; Kelley and McAllister 1984; Lesthaeghe 2000;

12

Immigrant Integration

Longva and Raaum 2003; Raijman and Semyonov 1995; Tribalat 1995). Researchers have found that the socio-cultural and economic integration of immigrants also differs across destinations, these being either regions or countries. Probably the first study in which such differences were found, was conducted in Canada by Lieberson (1970). He compared the ability to speak French and English among ethnic groups with French mother tongue in nine Provinces in Canada in 1931 and 1961. His results showed that in Quebec less than 1 percent of this group is able to speak English, whereas this percentage was more than 44 in British Colombia. A more recent study on language is that of Solé (1990), who used regional data for the United States. Solé compared the usage of English as a home language of four immigrant groups in eight states in the US and showed that 13 percent of Mexican immigrants in New York use English at home, whereas this is true for less than three percent of this group in Texas. Similar regional differences in immigrants’ language have also been observed in Australia. A relevant study is that of Clyne (1991), who compared percentages of persons using the official language of the country of origin by eight regions in Australia in 1976 and 1986. He found evidence to suggest that minority languages are more often used by immigrants in Victoria than in Queensland. All in all, these studies suggested that the socio-cultural integration of immigrants, at least measured in terms of language usage and proficiency, differs between regions within the same country. The economic integration of immigrants was also found to depend on the receiving context. Sanders and Nee (1987) found that (among paid employees) the earnings of Cuban immigrants in Miami were lower than those of Cuban immigrants in other parts of Florida. In addition, they found that the earnings of Chinese immigrants in San Francisco were lower than those of Chinese immigrants in other parts of California. Zhou and Logan (1989) reported that self-employed Chinese immigrants in New York City had lower earnings than in other parts of the state New York. In addition to these regional differences, researchers have also documented different economic outcomes across countries (Reitz 2003). Model (1997) and Model and Lapido (1996) showed that (nonwhite) immigrants in a major city in the United States –New York– had a higher occupational status than (non-white) immigrants in a major

Theory

13

city in the United Kingdom –London. Other studies found that immigrant earnings relative to those of natives were lower in the United States than in Australia and Canada (Borjas 1988; Reitz 1998). Kogan (2003) examined the integration of Yugoslav immigrants in the labor markets in two destinations: Austria and Sweden. Her findings showed that Yugoslavs fared better in Austria than in Sweden in terms of laborforce participation and employment. All in all, these studies suggested that the economic opportunities of immigrants depend on their destinations, being either regions or countries. ANSWERS TO MACRO QUESTIONS The assimilation theory was thus confronted with findings it was incapable to explain. The general conclusion reached was that both the socio-cultural and economic integration of immigrants depend on the immigrant group and on its destination, even after assimilation and other individual factors were taken into account. Establishing these macro differences is one thing, explaining them is quite another. The latter issue is, thus, the central focus of the subsequent research: “Why do the immigrant group and the receiving context play a role in the socio-cultural and economic integration of immigrants?”. Contexts of Reception One answer to this question is the idea of “contexts of reception” (Portes and Rumbaut 1996, 2001). According to this notion, the economic fate of immigrants depends on their “contexts of reception” or, put differently, “modes of incorporation.” There are three such contexts of receptions: government policy, labor market conditions, and ethnic community characteristics. In a later study, Model and Lin (2002) elaborated on this idea. With regard to the role of the government, it is argued that policies concerning immigration laws differ between immigrant groups. Some groups are welcomed and are supported actively by the government. Other groups, however, are passively accepted, whereas some groups are even excluded. It was therefore suggested by Portes and Rumbaut (1996, 2001) that the more favorable the government policies towards certain groups are, the better the economic integration of these groups.

14

Immigrant Integration

Model and Lin (2002) argued that government policy is not only directed to immigration law, but also to discrimination law. They suggested that the more strongly countries forbid discrimination of immigrants, the better will be the integration of immigrants in the labor market. The second contextual factor outlined is the labor market. It was argued that economic factors such as “the stage in the business cycle, demand for specific kinds of labor, and regional wage differentials” (Portes and Rumbaut 1996:85) play a role in shaping the context of reception. Furthermore, some immigrant groups are “negatively typified,” whereas other groups are more “positively typified” by the native population. More favorable attitudes depend on the similarity between immigrant groups and the native population in terms of “physical appearance, class background, language, and religion to society’s mainstream” (Portes and Rumbaut 2001:47). Because social distance towards less similar groups is larger, these groups will experience more discrimination and hence fewer prospects in the labor market. The ethnic community is the third dimension of the context of reception. It is argued that for some immigrants no cohesive community exists, but that a cohesive, well-integrated community is more common. It is assumed that all co-ethnics are willing to help their newly arrived immigrants, but that they do so within the limits of the information and resources available to them. Hence it was suggested that immigrants who belong to more advanced communities have better economic opportunities than those belonging to ethnic groups with fewer human skills and limited information and resources. The theory about “contexts of reception” is important because it may contribute to explaining macro differences in immigrant integration. The real strength of the theory is acknowledged when the general insights on which it is grounded are made more explicit. That is to say, the idea of “contexts of reception” is a theory specifically designed to explain the situation of immigrants, but the basic assumptions are obtained from general theories. For instance, the theory points to the importance of social distance, and this notion is part of a more general theory of prejudice and discrimination. To give another example, the theory on context of reception refers to the role of the ethnic community, and this idea rests on insights from social capital theory. By making the underlying assumptions more explicit, the

Theory

15

empirical scope of the theory on the contexts of reception can be expanded. Thus, whereas the theory is focused on the economic incorporation of immigrants, explicating the general theories on which it rests makes it possible to derive hypotheses on social integration (e.g., ethnic intermarriage) and cultural integration (e.g., language, religion) as well. I distinguish four such general theories, and I will discuss them now. Four General Theories The four general theories discussed here are: human capital theory, structural opportunity theory, prejudice theory, and social capital theory. The paradoxical status of these general theories is that they have been capable of explaining the influence of macro effects on immigrants’ socio-cultural and economic integration, whereas, at least initially, these theories were not developed for that purpose. I will reconstruct the development of these theories and empirical applications, and show, first, that each of the theories has its specific history, and, second, that they were subsequently introduced in the migration literature, although reference to these more general theories has sometimes remained implicit. Human Capital The first general theory was developed by economists in the 1950s and 1960s. Mincer (1958) and Becker (1964; Becker and Chiswick 1966), in particular, advanced the human capital theory, which rests on two general assumptions: (1) people’s life-chances depend on their humancapital, and (2) people are aware of this relationship between individual skills and their life-chances and therefore rationally invest in their own human capital. Thus, it was argued, people would invest in themselves as they expect that future material and financial benefits outweigh the temporary costs of investing. Initially, the theory has been used with considerable success to explain patterns of labor market participation (Becker 1964). Later on, the theory was also applied to explain noneconomic phenomena, such as marriage patterns, division of household tasks and fertility rates (Becker 1976, 1981). In the migration literature, the human capital theory was used first and foremost in studies of immigrants’ economic integration. Of crucial importance was the work of Chiswick (1978, 1979). First of all,

16

Immigrant Integration

Chiswick provided a human capital explanation for the assimilation theory, the idea and observation that immigrants have a disadvantaged economic position upon arrival, but improve their economic position over time. According to Chiswick, immigrants have a weaker economic position at arrival than natives, because they have less human capital. To begin with, most immigrants originate from non-Western countries and for that reason have less “general” human capital than natives, who received higher education on average. Second, immigrants lack “hostcountry specific” human capital. Educational qualifications obtained in the country of origin are not easily transferable and equally valued in the host country (Friedberg 2000). Furthermore, many immigrants do not speak the language of the host country well upon arrival, which limits their job opportunities. Finally, the labor market experience immigrants obtained in the origin country is of little value in the host country, which has a different structure and different rules. Because Chiswick assumed that immigrants invest in post-school training, gradually learn the host language and acquire knowledge of the host country labor market, he could explain that with length of stay, the economic integration of immigrants would progress. Furthermore, Chiswick hypothesized that after a certain number of years immigrants would eventually outperform natives, because immigrants are assumed to have more talents and other unobserved human capital than natives. These hypotheses were supported with data from the 1970 U.S. census of population. In this way, the human capital theory had provided a plausible explanation of why economic assimilation occurs. Subsequent studies elaborated on Chiswick’s ideas. It was reasoned that, if Chiswick was right in equating the assimilation effect with postmigration investments in human capital, such investments should have a positive effect on immigrants’ economic integration. Several studies supported this theoretical extension. Using data on the United States, Chiswick (1979) himself found a positive effect of educational qualifications and occupational experience on immigrants’ labor market outcomes. In more recent studies, it was shown that education obtained in the host country has a particularly pronounced effect on economic incorporation (Bratsberg and Ragan 2002; Friedberg 2000). Furthermore, Kossoudji (1988) showed that immigrants’ economic attainment directly varies with proficiency in the destination language.

Theory

17

This finding was replicated in a number of studies thereafter (Chiswick and Miller 1995; Shields and Price 2002). An important contribution to the human capital approach to immigrants’ economic incorporation was a study of another economist, Borjas (1985). First, pooling multiple censuses of the United States, and distinguishing the effect of length of stay from the immigrant cohort, Borjas showed that the economic progress of immigrants over the life-course was much smaller than maintained by Chiswick (1978, 1979), and that immigrants do not surpass natives economically. Second, Borjas found that the more recent cohorts were less successful economically than the older cohorts. These observations induced new theoretical developments in the human capital theory, in which the role of macro-level factors in the economic integration of immigrants was incorporated (Borjas 1985, 1987, 1988). In addition to the already established idea that human capital affects the economic position of immigrants, it was maintained that macro-level factors influence the selection of immigrants. Immigrants could be either favorably selected (i.e., with much human capital) or unfavorably selected (i.e., with little human capital). In this way, Borjas (1985) argued and indeed found that the declining economic success of immigrant cohorts in the United States could be partly ascribed to migration policies that have become less selective over time. This change in immigration policy shifted the skill composition of immigrant groups in the United States from mainly highly skilled European immigrants to less skilled immigrants from Third World countries. In a similar way, researchers have used the human capital theory to hypothesize about the influence of the immigrant group and the receiving context on their economic integration. The presumed causal relationship is depicted in Figure 2.1. This figure is related to the models advanced by Coleman (1990). The model contains two levels: a macro level (at the top) and a micro level (below). The causal process runs from the left to the right, and the signs express either a positive (+) or negative (–) effect. The nice feature of these models is that one gets a clear understanding of how a relationship at the macro level can be explained in terms of macro-micro, micro-micro, and micro-macro processes.

Figure 2.1 Effects of macro selection on economic integration

Characteristics of groups and destinations

+

Favorable macro selection

Economic integration

(4) (1)

+

+ Human capital immigrants

+ (2)

Economic attainment

(3)

Theory

19

Figure 2.1 shows that, within the human capital framework, the variation across immigrant groups and destinations in the economic integration should be interpreted in terms of macro selection. That is to say, for example, that in groups with certain characteristics (macro level) more favorably selects immigrants (macro level), which, in turn, positively affects the economic integration of that group (macro level). In order to understand the relationship between the group’s favorable macro selection and the group’s economic integration (i.e., relationship 4 in Figure 2.1), researchers have proposed macro-micro propositions (i.e., relationship 1), micro-micro propositions (i.e., relationship 2), and micro-macro propositions (i.e., relationship 3). More specifically, it has been postulated that: (1) members of favorably selected groups have on average more human capital than members of unfavorably selected groups, (2) human capital has a positive effect on immigrants’ economic integration, and, by aggregation, (3) better economic opportunities lead to more economic integration. Note that (1) is not a contextual effect, but rather a composition effect: favorably selected groups are composed of members with high human capital, but this macro property does not have a direct effect on immigrants’ economic outcomes. In order to explain macro differences in the economic integration of immigrants, researchers using a human capital framework have proposed various characteristics of groups and destinations that affect skill selection. It was hypothesized and empirically found that crossnational differences in the economic performance of immigrants are partly an outcome of immigration selection policies (Borjas 1988; Reitz 1998). Furthermore, evidence was found that group differences are the result of selection processes associated with the income inequality in the origin and destination country (Borjas 1987, 1988; Cobb-Clark 1993), economic development in the source and host country (Borjas 1987, 1988; Chiswick 1978, 1979; Jasso and Rosenzweig 1990a), travel distance ( Blau and Duncan 1967; Borjas 1987; Cobb-Clark 1993; Jasso and Rosenzweig 1990a), pre-migration exposure to the host-country language (Borjas 1987; Chiswick and Miller 2002; Jasso and Rosenzweig 1990a; Kossoudji 1988; Shields and Price 2002), political conditions in the country of origin and the reasons of migration (Chiswick 1978, 1979, 1999). Thus far, the contributions of human capital theory can be summarized as follows. The theory was, first of all, used to explain the

20

Immigrant Integration

assimilation theory, the idea and observation that immigrants become more economically integrated with length of stay in the host country. In addition, researchers used this theory to explain macro differences in immigrants’ economic incorporation. Human capital theory can explain why immigrants in one country perform better economically than in another country, and why some groups are more incorporated economically than other groups. In more recent times, economists and sociologists have also introduced the human capital theory in fields other than immigrants’ economic integration. It was again Chiswick who pioneered, and who applied the theory to a core element of immigrants’ socio-cultural integration: language proficiency. Chiswick (1991) argued that language skills are a form of hostcountry specific human capital that may improve one’s economic integration, an assumption that has been supported in earlier studies on the impact of immigrants’ language fluency on earnings (Kossoudji 1988). Chiswick thus relied on assumption (2) of the human capital theory, mentioned above, which states that people would invest in themselves as they expect that future material and financial benefits outweigh the temporary costs of investing. As all forms of human capital, he reasoned, language skills are embodied in a person and immigrants invest deliberately in learning the second language after arriving in the host country. In this way, the human capital theory was able to explain another finding that supported the assimilation theory, that is, the observation that immigrants’ language proficiency increases with length of stay in the host society. Chiswick (1991) further used the human capital theory to explain macro differences in immigrants’ language proficiency, but only as they pertain to differences between groups, not between destinations. Figure 2.2 presents the presumed causal model. It shows that group differences in language proficiency are explained in terms of both macro selection and economic context. When groups are more favorably selected, immigrants from these groups have more abilities to learn a new language. That is to say, they are more efficient in language acquisition, which means that they more quickly improve their language proficiency given a certain amount of exposure. Group differences in language proficiency are also assumed to be an outcome of the incentives to invest in language learning. Chiswick argued that economic incentives of investing in a second language can be higher or lower for immigrants in the same group, depending on the expected

Figure 2.2 Effects of macro selection and economic context on language proficiency

Characteristics of groups and destinations

+

Favorable macro selection and economic context

Language proficiency

+

+ Efficiency and expected benefits from investments in language l i

+

Language acquisition

22

Immigrant Integration

returns in terms of, for instance, employment or higher wages. Because these economic incentives are similar for all immigrants in a certain group, they can explain group differences. Thus, by postulating several characteristics of groups that affect macro selection or economic context, researchers have been able to explain group differences in language proficiency. Several of these characteristics were also highlighted in explanations of immigrants’ economic incorporation, such as political conditions in the source country (Chiswick and Miller 2001), economic development (Chiswick and Miller 2001, 2002; Espenshade and Fu 1997), and geographic distance (Borjas 1987; Chiswick 1999). Chiswick and Miller 2001; Espenshade and Fu 1997; Jasso and Rosenzweig 1990a). Other macro-level characteristics were examined specifically with regard to language, most notably group size (Clyne 1991; Dustmann 1994; Evans 1986, 1989; Lopez 1996; Portes and Bach 1985; Stevens 1992; Veltman 1983) and “linguistic distance” –the resemblance between the languages spoken in the origin and destination country (Beenstock, Chiswick and Repetto 2001; Carliner 2000; Chiswick and Miller 2001; Espenshade and Fu 1997). Structural Opportunities The second general theory used to explain macro differences in immigrants’ socio-cultural and economic integration is the structural opportunity theory advanced by sociologist Blau (1977, 1994; Blau and Schwartz 1984). Blau’s theory rests on two assumptions: (1) people have a preference to interact with other people of the same social standing, and (2) structural characteristics of people’s environment restrict their contact opportunities. Whereas the first assumption remains empirically untested in this research tradition, the second idea is worked out theoretically and empirically in the literature. Blau outlined four structural conditions that foster or hinder contact opportunities: group size, segregation, heterogeneity/inequality, and intersection/consolidation. He argued that members of a numerically larger group or a group that is more geographically segregated, have more opportunities to meet members of their own group and less chances to meet members of other groups. With heterogeneity –or horizontal differentiation– Blau refers to the distribution of a population across groups in terms of a nominal parameter (e.g., sex, race). Inequality –or vertical differentiation– refers to the distribution of a population across groups in terms of a graduated parameter (e.g.,

Theory

23

education, income). Heterogeneity and inequality increase, according to Blau, as people belong to groups of more equal size, and decrease as people belong to fewer (larger) groups. Higher levels of heterogeneity or inequality are therefore hypothesized to increase contact opportunities with members of other groups, and decrease chances to meet members of the own group. Intersection and consolidation, finally, point towards the degree to which various parameters are correlated. For instance, if differences in race, occupation and education are substantially associated, group barriers are consolidated, resulting in few contact opportunities between groups. By contrast, if these dimensions intersect (i.e., poorly associate), contacts between groups will be more frequent. Blau’s structural opportunity theory has been used to explain patterns of intermarriage and, less often, patterns of crime and violence. It was predicted and indeed found that the larger and more geographically concentrated the group to which people belong, the lower their chance to marry outside their own group (Blau, Blum and Schwartz 1982; Labov and Jacobs 1986) and the lower their chance to become a victim of crime or violence committed by members of other groups (Blau and Blau 1982; Sampson 1984; South and Messner 1986). Blau’s ideas were also introduced in the migration literature, but only with respect to immigrants’ socio-cultural integration. Figure 2.3 depicts the assumed causal model. It shows that immigrants’ sociocultural integration depends on the structural opportunities for immigrants to meet natives. These opportunities, in turn, are assumed to be an outcome of characteristics of groups and destinations, such as group size, segregation, inequality, and intersection. In studies on immigrants’ socio-cultural integration, Blau’s ideas were more often applied to the social dimension than to the cultural dimension. One measure of social integration, ethnic intermarriage, was particularly frequently examined. It was argued that when there is ample opportunity for immigrants in a certain group (or in a certain destination) to meet natives, this would result in frequent social interaction with natives –exemplified by high intermarriage rates. Studies have indeed found support for this line of thought. For example, in a study of six Asian American groups in multiple regions in the United States in 1980, Hwang, Saenz, and Aguirre (1997) found that relative group size has a negative effect on intermarriage rates.

Figure 2.3 Effects of structural opportunities on socio-cultural integration

Characteristics of groups and destinations

+

Structural opportunities to meet natives

Socio-cultural integration

+

+ Immigrants’ social interaction with natives

+

Immigrants’ cultural adaptation

Theory

25

They also found evidence to suggest that intermarriage rates increase with a shortage of the opposing sex of the same group in the same area. Furthermore, they showed that economic inequality between groups in a certain area decreases the probability of intermarriage. Lievens (1998) examined intermarriage rates of two groups (Turks and Moroccans) in multiple regions in Belgium in 1991. Lievens showed that ethnic intermarriage decreases with the relative size of a group in a region and with the degree of consolidation between ethnicity and socio-economic status. Blau’s theory was also introduced into studies on cultural integration. However, this application required an additional assumption, namely that frequent social interaction with natives results in stronger cultural integration. Two sociologists made precisely this assumption: Evans (1986) and Stevens (1992). They suggested that structural characteristics of groups do not only influence the chance to have positive (marriage) or negative (crime, violence) relations with people, but also affect the language spoken with others. They argued, furthermore, that language is learned and used through exposure to the out-group and language learning and usage is to some extent inhibited by exposure to the in-group. Demographic characteristics of the “language market” therefore affect immigrants’ language proficiency and usage. Thus, Stevens (1992:172), in her study of immigrants’ language use in the United States, argued that “geographical segregation and the numerical size of minority language groups underlie the probabilities that non-English language Americans are able to use their non-English language with other people.” In accordance with Blau’s theory, a negative relationship was found between group size and spatial concentration on the one hand and destination-language proficiency on the other (Chiswick 1991; Clyne 1991; Dustmann 1994; Evans 1986; Jasso and Rosenzweig 1990a; Veltman 1983). Prejudice The third general theory that is used to explain the role of macro factors in the socio-cultural and economic integration of immigrants is about “prejudice” among natives. I outlined that the structural opportunity theory assumes that people prefer someone of the own group rather than someone of another group, without further questioning this assumption. Ideas on prejudice, however, specifically theorize about this point. Although there is no overarching prejudice theory it is

26

Immigrant Integration

generally assumed in the sociological (e.g., Coenders 2001; Lubbers 2001) and socio-psychological (e.g., Brown 1995; Duckitt 1992) literature that people have (1) a positive attitude towards their own group, and (2) a negative attitude towards out-groups. To these two assumptions, a third one is often added: (3) negative attitudes towards an out-group lead to negative action towards the members of that group. These core ideas on prejudice have been worked out in a number of directions. The first two assumptions are, first of all, grounded in sociopsychological insights (Brown 1995; Duckitt 1992). There, it is argued that people strive for a positive self-concept, part of which people derive from identification with a social group (“social identity”). Because people mainly perceive positively valued characteristics of the own group and compare this to the mainly negatively valued characteristics of other groups, they have a positive attitude towards the in-group, and a negative attitude towards out-groups. Furthermore, researchers have maintained that people identify themselves with a number of groups –not just one or two. Allport (1954), for example, mentions the following group memberships: family, neighborhood, city, state, nation, racial stock, sex, age group, occupational, social (club and friendship), religious, ethnic, and ideological. Accordingly, there is not only prejudice along, say, racial grounds, but there exists “sexism” and “age-discrimination” as well. It is important to emphasize that, in consequence, prejudice theory has been applied outside the migration literature and can therefore be labeled as a “general theory.” Nevertheless, a substantial body of research has been involved in attitudes towards immigrants and “ethnic minorities,” which is discussed below. In addition to arguing that people generally have (1) a positive attitude towards the own group, and (2) negative attitudes towards other groups, the degree of positive and negative attitudes has been made “conditional” or “variable.” That is to say, sometimes in-group preferences are stronger than in other times, and some out-groups are more disliked than others. What causes these varying degrees of positive in-group attitudes and negative out-group attitudes is quite complex, and, at present, there is still a good deal of scientific discussion on this issue. In the present context, however, it is important to trace the developments with regard to negative out-group attitudes, and it seems safe to say that social psychologists and sociologists agree

Theory

27

that the “group-level approach” to prejudice is empirically quite successful and valuable for explaining differences of prejudice across time, regions, and groups (Coenders 2001; Lubbers 2001; Quillian 1995). This approach encompasses the contributions of several closely associated ideas: realistic group conflict theory (Bobo 1983; Le Vine and Campbell 1972), group-threat theory (Blumer 1958, Quillian 1995), and ethnic competition theory (Coenders 2001; Scheepers, Gijsberts and Coenders 2002). To simplify the matters, the group-level approach argues that prejudice directly varies with perceived threat (e.g., cultural, economic) from an out-group. In turn, perceived threat is assumed to increase with the number of out-group members, the visibility of out-group members, and the scarcity of goods that are at stake (Blalock 1967). Social psychologists and sociologists have also elaborated on the third assumption, the idea that negative attitudes towards a group result in negative actions towards the members of that group. Here it is useful to mention several degrees of negative action, listed in the classic study of Allport (1954:14–15). The first stage is antilocution, referring to people who express their prejudice with like-minded friends, and sometimes strangers. Then comes avoidance, meaning that people avoid members of the disliked group. In the third phase people discriminate the out-group, by means of excluding members of that group from certain types of employment, residential housing, political rights, and so on. Acts of violence and physical attack towards the outgroup constitute the fourth level of negative actions. Finally, Allport mentions the most extreme expression of negative attitudes, extermination, indicated by lynching, pogroms, and massacres. In the following, I will use the label “prejudice theory” when referring to the combined ideas on the three assumptions. These ideas have been introduced in the migration literature to explain macro differences in immigrants’ economic and socio-cultural integration. Figure 2.4 shows the presumed causal model with respect to immigrants’ economic integration. In order to apply prejudice theory to the integration of immigrants, researchers have filled in the three assumptions mentioned above. The first assumption made is that natives identify themselves with their country (in-group), and have a positive attitude towards their country fellows. Another assumption is that natives consider immigrants or “ethnic minorities” as the out-group, towards whom they have negative attitudes. Moreover, in order to explain differences between immigrant

Figure 2.4 Effects of anti-immigrant attitudes on economic integration

Characteristics of groups and destinations

Anti-immigrant attitudes

Economic integration

–

+

+ Discrimination of immigrants in the labor market

–

Immigrants’ economic opportunities

Theory

29

groups, immigrants are not considered as a homogenous out-group; instead, anti-immigrant attitudes vary between groups. This assumption was supported in the migration literature, where it was shown that negative attitudes or “social distance” (Bogardus 1959) of natives towards immigrants varied by immigrant group. In studies of natives in Canada (Pineo 1977) and the United States (Owen, Eisner, and McFaul 1981), northern Europeans ranked at the top of the social hierarchy, followed by eastern and southern Europeans and, at the lowest level of prestige, nonwhite immigrants. Researchers in the migration literature have also used the third assumption of prejudice theory, the idea that negative attitudes result in negative actions. As Figure 2.4 shows, the degree of anti-immigrant attitudes is considered an important determinant for negative actions towards immigrants. That is to say, negative attitudes of natives towards an immigrant group result in discrimination of the members of that group in the labor market, which hampers their economic opportunities. In the words of Evans and Kelley (1991:722): The underlying theory is that employers’ ethnic prejudice leads them to avoid hiring immigrants and sometimes their descendants. A parallel argument holds that native workers are prejudiced and so resist working with immigrants. Such discrimination could push immigrants into a secondary labor market where rewards are meager and promotion opportunities limited. By aggregation (i.e., from micro to macro), this implies that immigrant groups that are most disliked by natives are also the groups with the lowest economic integration. In addition, anti-immigrant attitudes could also vary between regions, explaining why in some destinations immigrants are better incorporated economically than in other regions. The crucial point for researchers was therefore to come up with characteristics of immigrant groups and receiving contexts that are associated with anti-immigrant sentiments. Why is it that some immigrant groups are at the top of the “social hierarchy,” and why is prejudice towards immigrants stronger in one region than in another? In line with prejudice theory, several authors have suggested that social distance towards groups and regional variation in prejudice are determined by a distinction between immigrants and natives in terms of

30

Immigrant Integration

culture and physical appearance, a low socioeconomic background of immigrants, and the size of the immigrant group (Hodge 1973; Lieberson 1980; Massey 1981; Portes and Rumbaut 2001; Van den Berghe 1967). These factors are all associated with perceived (cultural or economic) threat, which is assumed to result in stronger antiimmigrant attitudes among the native stock. Two studies have examined the role of the visibility of immigrant groups due to race or skin color. Evans and Kelley (1991) argued that natives’ negative attitudes towards an immigrant group increase with the distinctiveness in terms of race and color. They hypothesized that in Western countries, immigrants from “Third World” countries (e.g., Vietnam, the Indian subcontinent, Malaysia, Indonesia) would have a lower economic status than immigrants from other, racially less distant, countries. Using 1981 census data for Australia, however, they did not find that Third World immigrants had lower occupational and income returns to their education qualifications. In contrast, in Canada, Basavarajappa and Jones (1999), using the 1991 census, found that, after taking into account individual-level factors, non-white immigrant groups had a lower income than white groups, a finding assumed to concur with the idea of prejudice on racial grounds. Authors have also suggested that prejudice, and, in turn, discrimination, increases with the relative size of the immigrant group (Blalock 1967; Quillian 1995). They argued that sizable groups are more visible and are perceived as a potential threat to the native population in terms of political and economic power. As a consequence, they expected a negative relationship between the relative size of an immigrant group in a host society and the economic performance of its members in that society. Studies have indeed found that the relative size of the black population in an area is positively related to prejudice and discrimination towards blacks in that area (Fossett and Kiecolt 1989; Quillian 1995) and negatively related to their economic attainment (Frisbie and Neidert 1977). However, with regard to immigrant groups there have been mixed results with respect to the effect of group size on prejudice and economic attainment (Taylor 1998; Tienda and Lii 1987; Tolnay 2001). In another application of prejudice theory, researchers have relied on direct measures of prejudice towards immigrant groups. One such study was conducted by Wanner (1998), who relied on the results of a

Figure 2.5 Effects of anti-immigrant attitudes on socio-cultural integration

Characteristics of groups and destinations

Anti-immigrant attitudes

Socio-cultural integration

–

–

+ Immigrants’ social interaction with natives

+

Immigrants’ cultural adaptation

32

Immigrant Integration

survey in which native Canadians were asked how comfortable people would feel being around individuals from certain immigrant groups. He used the scores of these attitudes towards 12 groups to predict group differences in the economic integration of the members of these groups. However, using the 1996 Canadian census, and applying multilevel techniques, he failed to find a significant effect of negative attitudes towards the group on the occupational status and earnings of immigrants. Researchers have also applied prejudice theory to immigrants’ socio-cultural integration. Figure 2.5 depicts the causal argument. Using assumption (3), researchers have assumed that negative attitudes do not only result in discrimination (lowering their economic integration), but also in avoidance of immigrants (lowering their sociocultural integration). The ideas on prejudice were used in studies on immigrants’ socio-cultural integration, in particular with regard to ethnic intermarriage. In accordance with the idea on prejudice along racial lines, it was found that outmarriage rates among Black ancestry groups are much lower than those of other groups in the United States (Lieberson and Waters 1988). Note that Figure 2.5 also shows that prejudice theory could be applied to immigrants’ cultural integration, assuming that social interaction with natives increases immigrants’ cultural adaptation. Although this assumption has been made by researchers using structural opportunity theory (Evans 1986; Stevens 1992), at present, the prejudice theory has not yet been extended in this direction. Social Capital The fourth general theory is that of social capital, developed most notably by the sociologists Bourdieu (1980) and Coleman (1988, 1990). Although there is discussion in the literature about definitions of social capital, its sources and its forms (Portes and Sensenbrenner 1993; Portes 1998), most researchers agree that social capital indicates “the importance of the resources that are available to a person through his or her social relations with others” (Flap, Kumcu and Bulder 2000:147). It seems fruitful to outline four core assumptions. Basically, the argument is that the amount of social capital people have, increases with (1) the willingness of other people to offer help (“solidarity”), (2) opportunities for other people to offer help (“structural opportunities”), and (3) the resources available to people (“human capital”). To these another

Theory

33

assumption is added: (4) the more social capital people have, the better their life-chances. The theory has been used to explain a variety of social outcomes, including “school attrition and academic performance, children’s intellectual development, sources of employment and occupational attainment, juvenile delinquency and its prevention” (Portes 1998:9). Researchers have also introduced social capital theory in the field of migration, but only in so far as it explains differences between immigrant groups in their economic integration. Figure 2.6 presents the causal relationships. Figure 2.6 shows that the economic integration of immigrants is an outcome of the “social” or “ethnic” capital of their group. It is assumed that immigrants are willing to help co-ethnics, for example, by offering jobs, buying goods and lending money. In groups with more ethnic capital, immigrants have more resources available to them, which, in turn, promotes their economic opportunities. By aggregation, this means that these groups are also better integrated economically. The idea of ethnic capital has been worked out in two specific directions. In one line of thought, researchers have emphasized the importance of structural opportunities (assumption 2) for immigrants to share ethnic resources. It has been suggested that structural characteristics of immigrant groups affect the opportunity to form an “ethnic enclave” or “ethnic labor market” (Wilson and Portes 1980, Portes and Bach 1985). More sizable and geographically concentrated immigrant groups, it is argued, have an opportunity to develop an independent, mono-ethnic labor market in which immigrants can obtain positions otherwise held by natives (Semyonov 1988). However, there have been mixed results with respect to the effect of group size and ethnic segregation on the economic integration of groups (Sanders and Nee 1987; Taylor 1998; Tienda and Lii 1987; Tolnay 2001; Zhou and Logan 1989).2 It was also predicted that immigrants in larger groups are more often selfemployed, a finding supported in Yuengert (1995), but not in other studies (Fairlie and Meyer 1996; Razin and Langlois 1996).

2

Note that Portes and his colleagues (Portes and Bach 1985; Wilson and Portes 1980; Portes and Jensen 1987) do not equate the ethnic enclave idea with the geographic concentration of groups. Other researchers, however, such as Sanders and Nee (1987), have done so, and the literature cited here concurs with this conception of the ethnic enclave idea.

Figure 2.6 Effects of social capital on economic integration Characteristics of groups

+

Social capital immigrant group

Economic integration

+

+ Resources available to immigrants

+

Immigrants’ economic opportunities

Theory

35

Another research line, using assumption (3), more strongly focused on the role of resources available in the immigrant group. One such measure of ethnic capital was proposed by Borjas (1992): the group’s average education. This indicator was used in his study on the relationship between ethnic capital and intergenerational mobility in the United States. Borjas showed that, taking parental and neighborhood characteristics into account, the higher the average educational attainment among a certain immigrant group, the better the economic outcomes of the children in that group. A THEORETICAL ASSESSMENT In this chapter, I outlined the development of theories and research on the socio-cultural and economic integration of immigrants. Initially, researchers were primarily concerned with the question how the sociocultural and economic integration of immigrants changes over time. The answer they came up with was simply that immigrants will become like natives (“assimilation”), a process that takes place in people’s own life-course, as well as across generations. These researchers, however, were confronted with empirical findings for which the assimilation theory provided no answer. The theory turned out to be incapable of explaining why, after taking into account assimilation and other individual factors, differences persist between immigrant groups and between immigrants who moved to different destinations. These macro differences became the central focus of subsequent studies. One answer to this “problem” is the idea of contexts of reception, suggesting three contexts that could explain macro differences: government policy, the labor market, and the ethnic community. This is an important theory that possibly explains the macro differences in the incorporation of immigrants that were unaccounted for by the assimilation theory. However, the real strength of the theory is acknowledged when the general insights on which it is grounded, are made more explicit. By explicating these underlying assumptions, the empirical scope of the idea of contexts of reception can be expanded. Thus, whereas the original theory is focused on the economic incorporation of immigrants, explicating the theories on which it rests makes it possible to derive hypotheses on social

36

Immigrant Integration

integration (e.g., ethnic intermarriage) and cultural integration (e.g., language) as well. The first general theory (i.e., human capital theory) was developed by the economist Becker (1964). This theory initially found its way in the migration literature in studies of the economic integration of immigrants. Later on, however, the human capital theory was also used to explain immigrants’ destination-language proficiency. Structural opportunity theory, which was originally formulated by Blau (1977), was the second general theory introduced in the migration literature. Researchers have used this theory to explain socio-cultural integration in terms of ethnic intermarriage, language use and language proficiency. The third theory stresses the role of ethnic prejudice. This theory, which was developed by sociologists and social psychologists, was introduced in studies on both immigrants’ socio-cultural integration and economic integration. Finally, the fourth theory is that of social capital, originally developed by Bourdieu (1980) and Coleman (1988), which was used to explain group differences in immigrants’ economic incorporation. The review presented here seems to warrant the conclusion that these four general theories are theoretically and empirically quite successful and promising for future work. First, they explain, in terms of human capital formation, why assimilation occurs within and between immigrant generations, both economically and socioculturally. Second, they incorporate the idea of contexts of reception and give explanations for macro differences in immigrants’ sociocultural and economic integration. Third, hypotheses derived from these general theories have received ample empirical support in the literature, including studies done on different immigrant groups, in a variety of countries across a wide time span, and relying on different data sources. However, despite the empirical success of these general theories, there are some challenges to be mentioned as well. One important issue is that these theories, although they are of a highly abstract level, have not been applied to all aspects of socio-cultural and economic integration. Although the human capital theory has been used to explain several aspects of immigrants’ economic integration (i.e., labor force status, occupational status, income), with regard to socio-cultural integration it has been restricted to only one indicator (i.e., language proficiency). It remains to be seen, therefore, if this theory is general

Theory

37

enough to be extended to other aspects of immigrants’ socio-cultural integration, such as language use, ethnic identity, religion, and ethnic intermarriage. Likewise, structural opportunity theory has been applied to several indicators of socio-cultural integration (i.e., intermarriage, language proficiency, language usage), but not to other outcomes that seem less related to structural opportunities (e.g., ethnic identity, religiosity). In addition, the theory says little about economic integration (i.e., labor-force status, occupational status, income, selfemployment). Third, prejudice theory has been used to explain different aspects of immigrants’ economic integration, but with regard to sociocultural integration, it only has been used to explain ethnic intermarriage. Hence, further attention should be given to the question if prejudice theory is also capable of understanding the role of the immigrant group and the receiving context in issues on cultural integration, for instance ethnic identity, religiosity, language usage and proficiency. Fourth, social capital theory has not been used to explain the role of the receiving context, nor has it been suggested as an explanation for macro differences in immigrants’ socio-cultural integration. THEORETICAL EXTENSIONS The theoretical and empirical insights of earlier studies are incorporated in this book. This study uses hypotheses derived from the four general theories, at least in so far as they relate to economic integration and cultural integration (i.e., excluding social integration). Although previous work has made progress in our understanding of macro-level effects on immigrant integration, much remains unknown. One major issue is that previous researchers used the theories to understand the role of the receiving context and the immigrant group, not questioning their interplay. That is to say, previous research has not explicitly distinguished between immigrant group differences that travel across destinations (“origin effects’”) from immigrant group differences that are specific for a certain destination (“community effects”). As this study will show, theoretical extensions can be made to incorporate their differential effects as well. Another way this book contributes to the understanding of macrolevel effects on immigrant integration is to confront the four theories with one another. Earlier studies typically have focused on only one of

38

Immigrant Integration

the four general theories. In this study, however, hypotheses from different research traditions are used and tested simultaneously. A third way of contributing to the literature discussed above is to derive and test new hypotheses. This will be done in two different ways: by elaborating the “causes” (i.e., the driving forces) and by expanding the “outcomes” (i.e., that what is to be explained). A few examples will be used to demonstrate these theoretical extensions. First, I will expand the outcomes studied in the human capital framework, so that hypotheses on immigrants’ self-employment are included. A number of researchers assumed that immigrants with little human capital have fewer opportunities in the labor market. To this idea researchers have added that, when chances of finding work as a normal salaried worker are difficult, immigrants use self-employment as an alternative strategy for social mobility (Clark and Drinkwater 2000). With this assumption in mind, one can predict that when immigrant groups are unfavorably selected according to human capital theory, they are more likely to be self-employed instead of being paid workers. Because economists have listed a number of macro-level hypotheses that predict immigrants’ degree of economic integration, these ideas can be fruitfully used to explain macro differences in immigrants’ self-employment. To date, however, these applications of human capital theory have not been made. I do so in the present study, in Chapter 6. I also make new predictions for structural opportunity theory. As stated, researchers adopting this theory have linked structural opportunities to social integration (e.g., intermarriage), and, moreover, assumed that social integration fosters cultural integration. This latter assumption has been restricted to patterns of language usage and proficiency. To this, I would add that social interaction with natives not only increases immigrants’ usage of and proficiency in the official language, but it also promotes religious’ adaptation. This assumption stems from a large body of research, which mainly argues that social groups shape one’s religious environment and are therefore important for determining one’s religion (Berger 1967; Durkheim 1961 [1897]; Kelley and De Graaf 1997; Need and De Graaf 1996; Te Grotenhuis and Scheepers 2001), as well as other forms of norm-guided behavior, including suicide (Van Tubergen, Te Grotenhuis, and Ultee 2005). I therefore argue in Chapter 8 that if immigrant groups are large and cohesive, members of those groups have fewer opportunities to meet

Theory

39

natives, which results in less religious adaptation to society’s mainstream. I also elaborate on prejudice theory. I argued that the role of antiimmigrant attitudes has been examined only with respect to immigrants’ social integration, not cultural integration. In principal, however, prejudice theory could be applied to explain immigrants’ cultural integration, assuming, again, that social interaction with natives increases immigrants’ cultural adaptation. Although this assumption has been made by researchers using structural opportunity theory, at present, researchers have not extended the prejudice theory in this direction. I will do so in this study with respect to immigrants’ language proficiency (Chapter 7). There I argue that if natives have strong anti-immigrant sentiments, it is more difficult for immigrants from all origins to interact with members of the receiving society, which results in poor language skills. Another way to expand the prejudice theory is to come up with causes of anti-immigrant attitudes not examined before in the migration literature. One such cause I propose in this study is associated with the religious background of immigrants. The idea was informed by Allport (1954:224), who argued that “if two religions (or branches of a religion) are militantly disposed, each claiming the rival sect, a genuinely realistic conflict will ensue.” Immigrants with religions different from that of natives (e.g., Islam, Buddhism) are therefore assumed to increase the perceived cultural threat among natives, which, in turn, fosters negative feelings. Thus, I assume that anti-immigrant sentiments in the Westerns countries (all predominantly Christian) are strongest against immigrant groups with non-Christian backgrounds. In Chapters 4 and 5, I examine if “religious distance” results in lower economic integration (due to discrimination), and in Chapter 7 if it leads to fewer language skills (due to less social interaction with natives), as well. The final theory tested in the present study is that of social capital. I examine new hypotheses that pertain to the resources available in the immigrant community. I introduce three theoretically informed macrolevel variables in studies on immigrants’ economic incorporation. The first one is the mean educational level of the immigrant group. This measure was used in Borjas’ (1992) study on the relationship between ethnic capital and the socioeconomic outcomes of the second generation. I simply introduce this measure in research on the economic integration of the first generation, and argue that groups with higher

40

Immigrant Integration

levels of education have more ethnic capital, and individuals in these groups are therefore better integrated economically. The second one is the percentage of employed immigrants. This indicator of ethnic capital was proposed by Boyd (1991) in studies of black and Asian selfemployment in the U.S. I not only use this measure to predict immigrants’ self-employment (Chapter 6), but also to predict immigrants’ occupational status (Chapter 5). The third macro-level variable introduced in this study, is the educational heterogeneity of the immigrant group, which is assumed to be associated with the amount of information available in the group. This measure of the resources in a group was proposed and tested in the study of Renzulli, Aldrich and Moody (2000) on self-employment in the general population. I introduce the heterogeneity variable in the migration literature in Chapter 6 (self-employment). Despite the fact that the four theories are quite general in scope, this book will rely on additional –field specific– notions, which could not be subsumed under the general theories. Furthermore, I will use somewhat different labels in the empirical chapters, without direct reference to the four general theories. For both reasons, it is valuable to take a closer look at the theories that will be applied in each chapter. In order to explain macro-level differences in labor force participation and unemployment (Chapter 4), I will use human capital theory, prejudice theory and social capital theory. The same theories are used in Chapter 5, which is concerned with immigrants’ occupational status. A somewhat different picture arises for the study of immigrants’ self-employment (Chapter 6). Although the theories on human capital, prejudice, and social capital are again used to derive hypotheses, I also rely on additional notions. I draw on the existing literature on immigrants’ self-employment, which leads to additional hypotheses on macro-level effects. These hypotheses, such as on the role of labor market opportunities for entrepreneurship, are not derived from one of the four general theories, but they are important to consider as well. The book then continues with studying cultural integration, to begin with immigrants’ proficiency in the host-country language (Chapter 7). In that chapter I rely on three well-known mechanisms that are associated with language proficiency. These are: incentives to learn a new language, the efficiency with which people acquire a new language, and the amount of exposure to that language. The first two

Theory

41

mechanisms can be subsumed under the human capital theory, whereas the role of exposure to the host language is connected to the theory on structural opportunities. Thus, although somewhat different labels are used, the general theories are applied in that chapter as well. As we will see, the general theories are less capable of understanding the religious affiliation and attendance of immigrants. Apart from the hypothesized role of structural opportunities, as outlined, Chapter 8 mainly draws on theories from the sociology of religion. These are, for instance, theories on the (enduring) role of being socialized in a religious country, the possible influence of a more secular receiving context and the role of religious pluralism in the host country. In summary, despite the fact that the four theories are general in scope and that this book makes some theoretical extensions, they need to be supplemented with field specific notions. At the present time, there is no grand theory of immigrant integration.

This page intentionally left blank

CHAPTER 3

Data and Methods

This chapter provides the methodological basis of the book. I explain the double comparative design in more detail, and discuss how it deviates from comparative designs adopted in earlier research. I also list a series of challenges to conducting cross-national research. Then I show how these problems are solved in this book. COMPARATIVE MACRO DESIGNS Studies on the role of macro-level factors in the socio-cultural and economic integration of immigrants have relied on four different types of research designs. Table 3.1 provides an overview of these designs. Table 3.1 Overview of macro-designs Destination Origin Single Single Case studies Multiple

Comparative origin studies

Multiple Comparative destination studies Double comparative studies

Case Studies Perhaps the classic design to study the integration of immigrants is the “case study,” in which a single origin group is examined in a single destination (e.g., city, region, country). This research design, often 43

44

Immigrant Integration

small-scale, using methods of participant observation and letters of immigrants, was developed and applied frequently after the first major migration wave to the United States at the turn of the 20th century. The Polish Peasant, by Thomas and Znaniecki (1958 [1918-20]), was probably the first study of importance using this design. The early sociological and anthropological studies of Italians in the United States used the case study design as well. Prominent examples are Ware’s (1935) Greenwhich Village 1920-1930, which gave a detailed account of first-generation Italians in a little city, Child’s (1943) Italian or American?, which described the conflicts experienced by secondgeneration Italians in the United States, Whyte’s (1943) Street Corner Society, and Gans’ (1962) Urban Villagers, both dealing with the Italian community in Boston. Many more case studies have appeared afterwards on other immigrant groups in the United States, as well as on groups in other countries. There exists a second version of case studies on immigrants. In this version all origin groups are lumped together into the heterogeneous group of “immigrants.” Thus, while immigrants from different countries of origin are included in the data, in the analysis this detail is lost, and no difference is made with regard to origin. This is basically the method applied in the tradition of Park (Park and Burgess 1969 [1921]) and Gordon (1964). In these studies, which are conducted in a single destination, the comparison is typically drawn between the native population and the immigrant population. It is, perhaps, relevant to emphasize that this methodological design perfectly suited the goals of researchers working within an assimilation framework. As outlined in Chapter 2, their central question was how immigrants behaved compared to natives, hypothesizing that over time immigrants would become like natives. The case-study approach has been a frequently used method to examine immigrants’ economic integration and, in particular, sociocultural integration. A number of detailed case studies have been conducted on immigrants’ language usage and acquisition. Such studies have been done, for instance, on the Dutch in New Zealand (Hulsen 2000) and the Chinese in the United States (Luo and Wiseman 2000). A similar design has been used to study immigrants’ religion, such as the research on Greek immigrants in New York (Veglery 1988), and that on Koreans in the United States (Hurh and Kim 1990). Case studies

Data and Methods

45

have also been done on immigrants’ ethnic identity, an example is the work of Alba (1985) on Italians in the United States. Few case studies have been done on immigrants’ economic integration. Moreover, these studies have mostly lumped together immigrants from all countries of origin into one single group. For example, Chiswick (1978) studied the earnings of all immigrants (i.e., irrespective of their origins) in a single destination, the United States. This approach, in which the economic standing of the entire immigrant population is contrasted with that of the entire population of natives in a single destination, has been replicated in other studies (Baker and Benjamin 1994; Duleep and Regets 1992). Comparative Origin Studies In the “comparative origin” design, one compares the integration of immigrants from a number of origins in a single destination. Without doubt this is a very popular design, leaving only the case study being used more frequently. One of the first studies that adopted this method was Warner and Srole’s (1945) account of immigrant groups in New Haven. They studied the integration of Irish, French Canadians, Armenians, Italians, Poles, Greeks, and Russians in the American society. Another classic study is Beyond the Melting Pot, by Glazer and Moynihan (1970), in which immigrants from Puerto Rico, Russia, Italy, and Ireland were compared in a single destination, New York. The comparative design was adopted by researchers interested in documenting and explaining differences between immigrant groups. This research design was widely used in studies on immigrants’ cultural integration, such as ethnic intermarriage (e.g., Jones and Luijkx 1996; Kalmijn 1993; Kennedy 1944; Lieberson and Waters 1988; Lievens 1998; Pagnini and Morgan 1990; Qian and Lichter 2001) and language proficiency (e.g., Carliner 2000; Chiswick and Miller 1996, 1998; Dustmann 1994, 1997; Espenshade and Fu 1997; Espinosa and Massey 1997; Stevens 1999). It was also frequently used in studies on immigrants’ economic integration (e.g., Fairlie and Meyer 1996; Jasso and Rosenzweig 1990a; Kelley and McAllister 1984; Longva and Raaum 2003; Perlmann 1988; Tribalat 1995).

46

Immigrant Integration

Comparative Destination Studies Researchers have also performed “comparative destination” studies. The use of this design is quite recent, however, and applications are rarer compared to case studies or comparative origin studies. In the comparative destination design, the integration of immigrants from a single country of origin is studied in multiple destinations. Or, in a second version –like in the case study method–, the comparative destination design encompasses “all immigrants” as a single group studied in multiple destinations. In the words of Model, Fisher, and Silberman (1999:187), the question is: “How does the choice of destination affect the [socioeconomic] outcomes of culturally similar migrants?” Researchers have used the comparative destination design to investigate and explain the role of the receiving context. With respect to immigrants’ socio-cultural integration, applications of this design have been restricted to within nations comparisons on language proficiency (Clyne 1991; Lieberson 1970; Solé 1990). The comparative destination design has been more frequently used in the literature on immigrants’ economic integration, where it was used to examine cross-regional (Sanders and Nee 1987; Zhou and Logan 1989) and cross-national differences (Cheng 1994; Kogan 2003; Martin 1991; Model, Fisher, and Silberman 1999; Roodenburg, Euwals, and Ter Rele 2003). Double Comparative Studies Beginnings have been made in the literature to study the cultural and economic integration of immigrants in a double comparative perspective. However, these studies have been primarily concerned with describing macro differences, no study has used this design to test macro-level hypotheses on the role of immigrants’ country of origin, country of destination, and the immigrant community. A “double comparative” research design examines the integration of multiple origin groups in multiple destinations simultaneously. With regard to immigrants’ socio-cultural integration, such studies have been done on immigrants’ language usage and proficiency. Solé (1990) compared the usage of English as a home language across four immigrant groups in eight states of the US. Veltman (1983) studied

Data and Methods

47

patterns of language use of immigrants with French mother tongue in two US regions, and, likewise, language use of immigrants with Spanish mother tongue in eight regions. Clyne (1991) presented figures on the percentage language shift in the first generation by 14 countries of origin in eight regions in Australia. Double comparative studies have been done on language at a national level as well, although comparisons have been made indirectly. Evans (1986), for example, compared her findings on destinationlanguage proficiency among several origin groups in Australia, to published findings from other studies in Germany and the United States. Extra and Verhoeven (1993) edited a volume, which contained separate descriptions of immigrants’ language acquisition in several European nations. Similarly, Chiswick and Miller (1995) conducted separate, but similar, analyses of language fluency across multiple origin groups in Australia, Canada, Israel, and the United States. The literature on immigrants’ economic integration has used the double comparative design somewhat more often. Model (1997) and Model and Lapido (1996) examined the occupational status of six groups of non-white immigrants in London and New York. Borjas (1988) and Reitz (1998) analyzed the economic attainment of multiple immigrant groups in three classic immigrant countries: Australia, Canada, and the United States. Langlois and Razin (1995) compared rates of self-employment among immigrants of French mother tongue in five regions in Canada in 1986, and, in the same regions, the selfemployment rates of immigrants from non-French mother tongue. Razin and Langlois (1996) compared self-employment rates among 65 immigrant groups in 25 metropolitan areas in Canada in 1991. Furthermore, several studies on self-employment have been conducted, in which separate country reports on multiple immigrant groups have been combined and compared (Kloosterman and Rath 2003; Rath 2002; Waldinger, Aldrich and Ward 1990). From a comparative macro-level perspective, the double comparative design is promising and is to be preferred above the other research designs. My general argument is that the advantages of this design have not been fully explored. The first advantage is the opportunity to study both the impact of immigrants’ origin and destination. In contrast, case studies provide no macro-level information at all, for immigrants from a single origin group in a single destination are examined. In the case-study method only within-group factors (i.e., individual-level effects) can be studied,

48

Immigrant Integration

and, therefore, the case study design is of no direct significance for understanding macro differences. Furthermore, whereas the comparative origin design yields information on the role of the country of origin, it holds destinations constant by design and it is therefore unable to examine destination differences. Similarly, comparative destination studies provide information on the importance of destinations on the integration of immigrants, but because origin differences are held constant by design, no information is obtained on the role of the country of origin. The second asset of the double comparative design is that it provides a more representative view of origin and destination effects. After all, case studies do not compare origins and destinations at all, comparative origin studies rely on a single destination to examine the impact of the country of origin, and comparative destination studies use one origin group to examine the influence of the country of destination. Conclusions about origin differences are therefore not generalizable to other destinations, and conclusions about destination differences may not be applicable to other origin groups. Consider, for example, that a researcher who uses a comparative origin design finds that in the United States, Mexicans fare less well in the labor market than Irish immigrants. On grounds of a single case, one cannot conclude that Mexican immigrants do less well economically, irrespective of their destination. A similar deficiency relates to the comparative destination design. For example, it could be observed that Moroccans in France perform better economically than they do in Germany and the Netherlands, but it would be impossible to say that France is the more favorable context for all immigrants. The third advantage of the double comparative design is related to the previous one. In the comparative origin design and the comparative destination design, the researcher (often implicitly) assumes that the pattern found for a single case applies equally to other cases. However, this generalization neglects the possibility that origin groups perform differently, depending on their destination. It could be, for example, that the disadvantage of Mexican immigrants compared to Irish immigrants in the United States is totally different in another destination country (e.g., the Netherlands). Similarly, with regard to destination effects, Moroccan immigrants in France could have an edge over Moroccan immigrants in the Netherlands and Germany. Would

Data and Methods

49

one consider Surinamese immigrants in these three countries, conclusions on the role of receiving nations may be totally different. This implies that, using the double comparative design, one could go beyond establishing and explaining the role of the immigrant group and the receiving context. Although earlier researchers have used this design, the possibility of examining three macro-level effects has gone unnoticed. First, it is possible to study the general impact of characteristics of the country immigrants come from, irrespective of their destination. An example of such a factor is the political condition in the origin country. I label these and other factors that relate to the country of origin as “origin” effects. Second, one could study the general role of characteristics of destinations, notwithstanding immigrants’ origins. For instance, countries have certain immigration and integration policies that may affect the integration of all immigrants. These I call “destination” effects. Third, there is another macro effect, which relates to the combination between immigrants’ origin and destination. A good example is the relative size of an origin group in a certain destination country. I refer to these and similar factors as “community” effects. Whereas the multiple origin-multiple destination design can disentangle these three effects, the other designs cannot. The results of comparative origin studies indicate either origin effects or community effects, being unable to distinguish them. Likewise, comparative destination studies cannot distinguish community effects from destination effects. Case studies, finally, are the most ambiguous. Because these studies focus on a single origin group in a single destination, such as the Italians in the United States, their findings could mean three things. It could say something about Italian immigrants (i.e., their origin), about immigrants in the US (i.e., their destination), or about the Italian community in the US (i.e., their community). For that reason, double comparative research designs are to be preferred in studies on the socio-cultural and economic integration of immigrants. Because studies that adopted such a design have done so implicitly, the advantages have gone unnoticed. COMPARATIVE MICRO DATA Although the double comparative design is especially attractive to study the role of the country of origin, the country of destination and

50

Immigrant Integration

the immigrant community in immigrant integration, it also poses data problems. Two issues, in particular, deserve attention. The first is that several authors have argued that cross-national studies in general are hampered by problems of comparability. The second is that the small number of cases at the macro level could flaw comparative studies. It is important to emphasize that these two issues have been frequently addressed by proponents of the case study design, and are used as arguments against comparative designs –in particular designs in which nations are compared. In a review of this critique, Goldthorpe (1997) showed that case studies are equally vulnerable to this problem. I will not go into this debate here. Instead, I discuss the two problems first and subsequently show how I overcome them. Comparability Problems An important issue in cross-national research is comparability. Some authors have argued that quantitative cross-national analyses are impossible to conduct (Ragin 1987). Basically, they have two main points. Theoretically, they argue that macro-level cases differ so much from one another that comparisons are deemed to fail from the beginning. Societies are multifaceted, each having a different culture and a unique historical background. Methodologically, it is argued that survey differences in, for example, sampling, mode of data collection, questioning, and coding, obscure true differences. And, even if surveys are conducted in a similar way cross-nationally, they argue that the meaning and interpretation of words and phrasing differ too much to make meaningful comparisons. Thus, from a theoretical and methodological view they argue that comparing different macro cases is like comparing “apples and oranges,” that is, that countries can only be studied as entities in themselves. Hence they advocate an ideographic way of doing science, or a case-study approach, in which a single (or just a few) case(s) should be studied in detail, taking into account the “complex and unique sociohistorical configuration” (Skocpol and Somers 1980:178).3 According to Ragin (1987) case 3

Out of the critique on quantitative cross-national studies, Ragin (1987) developed an alternative method of comparing nations –“the comparative method.” This method, however, is of little use to researchers interested in

Data and Methods

51

studies allow nations to be considered as meaningful wholes rather than serving simply the basis on which to place boundaries around the measurement of variables. This critique, however, has met strong and convincing counterarguments (Goldthorpe 1997; Lieberson 1991, 1994). For one thing, it is argued, the slogan “everything and everybody is different and unique” is at the same time true and false. It is trivially true, because societies (like individuals) naturally differ from one another. However, it is false to draw the conclusion that two societies (or individuals) that are different cannot be compared. On the contrary, differences observed between societies are “social variation” and such differences are the beginning of any theoretically driven research. If there would be no variation between societies, there won’t be comparative macro research at all. And the more variation, the better. The aim of science is precisely to generalize and simplify and yet describe the more complex reality at the same time (Popper 1972). In this nomothetic strategy of doing science, or “variable-oriented” style, cross-national research can be done perfectly well. The starting point of this line of research was a classic work of Przeworski and Teune (1970). They argued that countries can be compared by looking at their characteristics. If the proper names of countries are replaced by the values of variables, comparisons can be made perfectly well. For example, to understand that water boils at different temperatures in New York and Denver, one need not attribute this to “New York-ness” or “Denver-ness”. Instead, the researcher should theorize about the causes of the differences between the two observations, and subsequently examine this idea by attaching values of the theoretically informed macro-level variable (i.e., atmospheric pressure). Properties of countries could then explain the observed similarities and differences between countries. To give another, classic example: apples and

individual-level outcomes (all outcomes have to be macro-aggregates, and need to be dichotomized), and, moreover, technically extremely dubious, for it assumes a deterministic pattern of social phenomena (ignoring the probabilistic and multivariate revolution in the social sciences –see Goldthorpe 2000). Detailed critique on this method is provided by Lieberson (1991, 1994) and Goldthorpe (1997).

52

Immigrant Integration

oranges could be compared with respect to sweetness, size, color, growth-rate, and so on.4 The variable-oriented approach overcomes the theoretical critique of case-study proponents against cross-national research. The argument that societies are too complex or too multifaceted to be compared with one another can be easily refuted. Quantitative social research can deal with this complexity by theorizing about these important constructs, measuring these concepts, and applying multivariate regression techniques to assess their importance. In a similar vain the idea that unique historical contexts hinder cross-national (or cross-group) comparisons can be met. The variable-oriented approach can deal with this objection, for change can be measured and likewise incorporated in theoretical and empirical models. With regard to the methodological critique against cross-national research, one solution is to collect data that are comparable across nations. That is to say, one needs to assemble data sets that use similar coding of variables, similar survey designs, and so on. Another solution is to measure survey differences that can bias cross-national comparisons. Just like cross-national differences may be explained by theoretically-informed variables on characteristics of countries, so too could cross-national differences be “explained” by differences in surveys. Including these survey characteristics in the analysis is a way to control for survey differences in a statistical way. Below, I will show how the present study deals with these issues and I will give other strategies to solve problems of comparability. Small N Problems Another well-known problem in macro-comparative research is the small number of macro-cases generally included in the analysis 4

It is interesting to reflect on the objections of the case-study approach in light of quantitative studies at the individual level. If, as argued by proponents of the case-study style of research, countries could not be compared because they are situated in a unique sociohistorical context, so too could individuals not be compared –since they are equally the product of unique historical and temporal circumstances. Strange enough, no one would argue so with respect to the individual level, but such arguments are made with regard to the macro level.

Data and Methods

53

(Goldthorpe 1997; Lieberson 1991, 1994). Lieberson (1991) phrased the problem as “Small-N, Big Conclusions.” More precisely stated, the number of macro-cases is generally too small relative to the number of explanatory variables. This makes testing (competitive) theories difficult, and conclusions based on a few cases may be drawn incorrectly. The problem pertains to earlier studies on macro effects on the socio-cultural and economic integration. Previous research has typically used a few groups to examine group differences or a few destinations to examine destination differences. Warner and Srole (1945), for instance, studied immigrants from seven different countries of origin to examine group effects. In a similar vain, Borjas (1988) and Reitz (1998) analyzed the economic attainment of immigrants in only three countries to interpret the role of host societies. The small N problem is therefore a serious one that needs to be dealt with carefully. Below, I will show how the present study aims to do so. The International File of Immigration Surveys In view of the above discussion, it is important to have micro data on the cultural and economic integration of immigrants that are both comparable across nations and provide enough cases at the macro level to perform detailed analysis. It is important for theory testing that this data set contains as many origin groups in as many countries of destinations as possible, while at the same time preserving comparability. I collected existing surveys containing individual-level information on immigrants in highly industrialized, Western countries in the period 1970–2002. The surveys were combined into one file that I call the International File of Immigration Surveys (Van Tubergen 2004a, see Appendix A). To obtain high quality and cross-national comparable data, surveys included in the meta-file fulfill several criteria. First, surveys should contain a sufficiently large number of immigrants to provide detailed analyses and the survey sample should be (approximately) nationally representative. Second, surveys had to contain standard questionnaires with fixed response categories. Third, surveys should contain cross-national comparable independent and dependent variables. Regarding the analysis of language proficiency,

54

Immigrant Integration

surveys also had to be translated in the mother tongue of immigrants and/or bilingual interviewers must be used in the field. I was able to find about 150 surveys that fulfill these criteria for a total of 19 Western countries: three classic immigrant societies (Australia, Canada and the United States) and 16 new immigrant societies (Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, and the United Kingdom).5 Appendix A presents detailed information on the measurement of the independent variables, on the surveys used in each chapter and it gives detailed references to all surveys. Three sorts of surveys were collected: census data for the three classic immigrant countries, labor force surveys for 15 EU countries, and additional specific immigrant surveys. Table 3.2 presents the pros and cons of these surveys, which I will discuss briefly. The first “characteristic” of surveys is the number of societies in which they are conducted. Regarding the classic immigrant countries (Australia, Canada, and the United States) the census has been a reliable and often used data source for the study of immigrants. By contrast, census data on the European continent mostly do not provide information on country of birth, and are therefore not appropriate for this study. For Europe, two other options are open to study immigrants. First, cross-national comparable labor force surveys are available for the 15 countries of the European Union. Second, specific immigrant surveys are conducted, but for a limited number of countries. On these grounds, I collected census data for the three classic immigration countries, labor force surveys for the 15 EU countries, and additional specific immigrant surveys in both old and new immigration countries. Another aspect of the surveys in IFIS is the number of respondents and the number of origin groups. The public use files of the census data in the traditional immigrant countries contain large numbers of immigrants to perform meaningful analyses. Also, census data for the classic immigrant countries contain many origin groups. Labor force data provide information on multiple origin groups, but the categories in the surveys do not always include each origin country. 5

I used various strategies to obtain these surveys. First, I directly contacted researchers who conducted national representative surveys on the integration of immigrants. Second, I requested several data sets from institutes, which provided general population surveys. Third, I contacted national data archives.

Table 3.2 Pros and cons of different survey data on immigrants Census data in Labor force surveys in EU classic immigrant countries countries Number of nations – + Number of origin groups + + Number of respondents + +/– Translated/bilingual interviewers + / – – Background information +/– – Information on language +/– – Information on religion +/– – Information on economic + + integration Note: + advantage of survey; – drawback of survey

Specific immigrant surveys – – +/– + + + + +

56

Immigrant Integration

Moreover, a problem in the study of immigrants in Europe is that the proportion of immigrants is still small. In 1998, 5.1% of the population in the 15 EU countries had a foreign nationality (Eurostat 2000). In the “new immigrant countries” in southern Europe –Portugal, Spain, and Italy– the percentage is as low as two or three. For this reason, specific immigrant surveys have been designed that over-sample certain immigrant groups. A drawback of specific immigrant surveys is that they are often limited to a few large immigrant groups and sometimes exclude (well-integrated) smaller groups. Translation of surveys or the use of bilingual interviewers is another aspect of surveys. Systematic non-response is likely to be a problem if difficulties with the host language are not taken into consideration in the survey design. Many immigrants are not adequately proficient in the host language, which could lead to systematic attrition of immigrants with lower command of the destination language. Immigrant surveys are designed to treat these problems appropriately. Labor-force surveys do not meet these additional measures and are therefore vulnerable to systematic non-response related to variables under scrutiny, especially when they are culturally oriented. The quality of census data in traditional immigrant countries lies between these extremes. Bilingual interviewers are used, but interviewers do not speak all languages. Another feature of surveys is the amount and detail of information on relevant independent and dependent variables. Labor force surveys, it is already suggested by its name, can be used solely for the study of the economic integration of immigrants. Moreover, these surveys generally provide little background information. Specific immigrant surveys, by contrast, provide extensive information on immigrants. They can be used for a wide range of topics, such as language proficiency, religious participation, and so on. Moreover, immigrant surveys provide more detailed information on immigrants’ background and migration history. Census data in traditional immigrant countries contain detailed information on immigrants’ economic integration as well as language proficiency. However, religion is not available in the census of the United States, and only limited information is provided in the census of Canada and Australia. It should be noted that IFIS does not contain immigrants for each origin-by-destination combination. One reason is, of course, that there are simply no immigrants in a country. Migration flows tend to be

Data and Methods

57

rather origin-by-destination specific (e.g., from former colonies to the former colonial power), and I have to deal with these observational data. A second reason relates to the sample design. It could be that the survey sample was too small, resulting in no respondents of a certain immigrant group –although there are immigrants from that group in the population. Finally, not observing immigrants in each combination is partly due to survey coding. Many surveys do not provide detailed information on immigrants’ country of birth. For instance, IFIS has information on immigrants from only 14 countries of origin in Canada, not because the immigrant population in Canada is not diverse (on the contrary), but because the census data that is used for this country only lists 14 countries of birth and a rest category lumping together all immigrants born elsewhere. Nevertheless, IFIS includes immigrants from a large variety of countries over the world traced in multiple host societies. Solutions to Comparability Problems In the present study, I will use several ways to deal with problems of cross-national comparability. The first way, it has been said, is that I selected only cross-national comparable surveys in IFIS. Surveys are supposed to be conducted face-to-face using standard questionnaires with fixed response categories. Surveys should contain cross-national comparable independent and dependent variables. And regarding the analysis of language proficiency, surveys also had to be translated in the mother tongue of immigrants and/or bilingual interviewers must be used in the field. Face-value inspection of the surveys in IFIS reveals that design, questions, and answer-categories are in general very similar. Important variables are coded according to international standards, such as ISCED-97 (OECD 1999b) for education and ISCO-88 (ILO 1990) for occupation. I also use standardized cross-national surveys, most notably the European Union Labour Force Survey, in which design and measures are made comparable beforehand. Moreover, for specific analyses, I sometimes excluded surveys that do not meet these criteria. For example, in the analysis of language proficiency in Chapter 7, I did not use the census of Canada because it contains a measure of language proficiency that deviated too much from that of other surveys.

58

Immigrant Integration

The second way is to include variables in the analysis that represent methodological differences (Treiman and Ganzeboom 1990). For example, in Chapter 7 I face the problem that IFIS contains some surveys in which respondents were asked to assess their language proficiency, whereas other surveys in IFIS asked the interviewer to assess respondent’s language proficiency. Instead of relying on one of the two sorts of surveys, or conducting separate analyses, I included a dummy-variable that represents this difference in assessment and thereby control for possible bias in that respect.6 A third way of dealing with problems of cross-national comparability is to examine the sensitivity of the findings by conducting additional analyses on different populations. If the findings of the research are robust, the outcomes should not depend on the inclusion of specific surveys, countries, or groups. In order to examine this assumption I perform sensitivity analyses. For example, in Chapter 4 on immigrants’ employment I conduct additional analyses excluding certain surveys, excluding certain countries and excluding small immigrant groups, and subsequently compare the differences with the findings obtained by the full data set. A fourth way in which I treat the problem of comparability, and another way to examine the sensitivity of the findings, is to use different codings of dependent variables. I assume that changing the dependent variables of this study yields similar outcomes. In Chapter 7, I apply this method to the study of language proficiency. I conduct two separate analyses, in which in one analysis language proficiency is measured on a four-point scale, whereas in another analysis a two-point scale is used. When the logit (using a 2-point scale) and linear (using a 4-point scale) analyses of language-proficiency yield similar results, the possible disturbing factor of categorization is ruled out, and we can have more confidence in the results. A fifth way is to examine how sensitive the results are for changes in the methodological design. In the empirical studies on language, religion, and economic outcomes, I conduct two sorts of analyses. First, an analysis in which the country of destination is treated as timeinvariant. Second, an analysis in which the country of destination varies over time. In the latter analysis, the destination level consists of 6

It should be remarked that the inclusion of these dummy variables controls for the main efect. The possibility of interactions remains, however.

Data and Methods

59

“destinations per year” cases. If both analyses yield similar results, we can have more confidence in our conclusions. Solutions to Small N Problems Although it was a specific aim of this book to have a large empirical scope, and the scope of this research is indeed larger than in previous studies, the number of macro-level cases is still small –at least compared to ordinary quantitative individual-level studies. The problem, however, is not equally potentially damaging to all our macrolevel effects. IFIS contains a total of more than 180 origin groups, 19 destination countries, and about 1,000 communities. Hence the small N problem is not relevant with regard to community effects, of little relevance to origin effects, but could be problematic with regard to destination effects. Because in studies on language proficiency (Chapter 7) and religion (Chapter 8) the number of destination countries is even smaller, these effects are more problematic to estimate. There are two additional ways in which I will deal with the small N problem throughout this study. First, a small number of cases is less problematic when there are fewer explanatory variables. Hence I tried to be very parsimonious with including macro-level factors, especially with regard to destination-level effects. For example, in the study of language proficiency (Chapter 7) I use the presence of socialdemocratic parties in the government as an indicator of integration policies instead of including multiple destination-level variables that are indicators of the same phenomenon. Together with an indicator of anti-immigrant sentiments, I have only two variables measured at the level of destinations in that analysis. Second, having only a few observations to estimate the effects of macro-level variables is less problematic when these variables do not correlate too much. When the association between two variables is high, it is more difficult to estimate the effect of these variables, and hence to provide a test of the theories behind them. I therefore computed the bivariate correlations between macro-level indicators, and, when the colinearity between two variables was too high, one of these two was eliminated from the analysis.

60

Immigrant Integration

COMPARATIVE MACRO DATA Problems of Unmeasured Macro Data In order to test macro-level theories, one needs to include macro-level variables in the analysis that correspond to these theories. Previous research, however, has not done this in a systematic way. Quite a number of studies have relied on separate analyses for different macro cases to examine macro effects, whereas others have relied on the inclusion of dummy variables for macro cases (i.e., origin groups, destinations). In both ways, the correspondence between the macrolevel theory and the macro-level measurement is implicit and indirect. This can be observed both in studies of immigrants’ socio-cultural and economic integration. One example will illustrate this point. In a study on immigrants’ language proficiency, Carliner (2000) compared a number of immigrant groups in the United States. He was interested in several macro-level hypotheses, more specifically whether immigrants’ language proficiency depends on the linguistic distance between the official language of immigrants’ origin country and English, on the geographic distance between immigrants’ origin country and the United States, and on the economic development of immigrants’ origin country. Carliner concluded that his results “do not support the hypothesis that linguistic distance is as important as other factors in explaining English skills,” “do not support the hypothesis that immigrants from poor countries have lower skills than immigrants from richer countries,” and “do lend support to the hypothesis that immigrants from nearby countries have weaker language skills than immigrants from more distant countries.” Carliner reached his conclusions by including dummy variables for immigrant groups and subsequently interpreting the differences between these groups in terms of linguistic distance, geographic distance, and economic development. Besides the fact that this approach would not be feasible when the number of macro cases becomes large, this is a weak test of the proposed theories. A better way of testing the macro-level effects would have been to include the corresponding macro-level variables (i.e., linguistic distance, geographic distance, and economic development) in the analysis –not to rely on the names of macro cases.

Data and Methods

61

The Country and Community File What is important, therefore, is that instead of conducting separate analyses for different macro cases, or including dummy variables for different immigrant groups or countries, macro-level information is included in the analysis. Although this argument, as said, has been made in general (Prezworski and Teune 1970), the migration literature shows many examples in which this suggestion is ignored. What is needed is, in addition to the cross-national file at the individual level, a corresponding file with macro-level information. I constructed such a file, using a variety of sources, resembling as closely as possible the theoretical ideas developed in Chapter 2. A total of more than 20 of such macro variables were constructed and pooled into what will be referred to as the Country and Community File (CCF; Van Tubergen 2004b, see Appendix B). With regard to origin effects, I collected information on such things as the dominant religion, the degree of political freedom, and economic development. With respect to destination countries, I assembled data on immigration policy, political parties in the government, and anti-immigrant attitudes. Among the community variables included in CCF are the geographic distance between the country of origin and the country of destination, the relative size of the immigrant group, and the average length of stay of the group. Appendix B provides detailed description of the measurement of the macro-level variables. Appendix B also explains how I dealt with missing data.

COMPARATIVE MACRO TECHNIQUES Ignoring the Multilevel Structure: “Large N problem” Another way this study tries to improve earlier studies is to use more appropriate statistical techniques. In many earlier studies on the role of macro factors in the cultural and economic integration of immigrants, the macro-level effects are estimated in the same, deficient way. Previous research has mostly estimated individual and contextual effects using regression analysis at the individual level. In this way the parameters and error terms at the contextual level are estimated with

62

Immigrant Integration

information of a large number of individuals –thus ignoring the fact that these observations at the individual level are not independent from one another. The result is that the standard errors at the contextual level are underestimated, which, in turn, can result in unjustified support for the contextual hypotheses (Kreft and De Leeuw 1998; Raudenbush and Bryk 2002; Snijders and Bosker 1999). Cross-Classified Multilevel Techniques A solution to the “large N problem” starts with a clear idea of how the data are structured. Table 3.3 gives an imaginary example relevant to earlier research and the present study. First consider researchers who used a comparative origin design, in which multiple origin groups (e.g., Chinese, Pakistani, and Vietnamese) are examined in a single destination country (e.g., United States). In such a design, the data is hierarchical: immigrants (level 1) are nested within origin groups (level 2). Similarly, researchers adopting a comparative destination design have a data structure in which a single origin group (e.g., Chinese) is nested in multiple destinations (e.g., Australia, German, and United States). In that design immigrants are again the lowest level units, but now nested within destination countries. In comparative origin designs and comparative destination designs researchers should rely on hierarchical, nested multilevel models instead of individual-level regression techniques. Table 3.3 Example of cross classified structure Country of destination Country of Australia Germany Netherlands origin China X X Pakistan Turkey X X Vietnam X X

United States X X X

In the double comparative design advanced in the present study, however, the multilevel structure is somewhat more complex. In a hierarchical or nested structure all entries across a row will fall under a

Data and Methods

63

single column or vice versa. As Table 3.3 indicates, however, no pure hierarchy can be found and immigrants are said to be contained within a cross-classification of their country of origin and country of destination. Some rows contain multiple entries, whereas at the same time some columns contain multiple entries. The conceptually and technically correct way of conceiving the situation of immigrants is depicted in Figure 3.1 Figure 3.1 Classification diagram for crossed structure Country of destination

Country of origin

Immigrant

In this study, therefore, I use so-called “cross-classified” multilevel techniques (Raudenbush and Bryk 2002; Snijders and Bosker 1999). I use Markov Chain Monte Carlo (MCMC) techniques provided in the software program MLwiN to estimate these models (Browne 2002). In order to understand these multilevel models and the way I treated them in this study, I start with discussing fairly simple ones and later on introduce more complex ones.7 Consider, first of all, the “empty” or “unconditional” model, which estimates the components of variation, without including predictors at either the micro or macro level. I assume that the outcome variable is linear (e.g., occupational status). The models refer to the situation, as illustrated in Table 3.3, in which immigrants i from country of origin j live in country of destination k. 7

Cross-classified models have been developed recently, and, at the moment, very few empirical applications can be traced in the literature. For more information on these models, see Chapter 11 in Snijders and Bosker (1999) and Chapter 12 in Raudenbush and Bryk (2002). For the mathematics of the MCMC models, and the implementation of these models in the software program MlwiN, see Browne (2002).

64

Immigrant Integration

The combinations of origins by destinations are referred to as communities or “cells.” In that case, the level-1 or “within-community” model is: Yijk = π0jk + eijk

[1]

with the following assumptions: eijk ~ N (0, σ2) where Yijk

is the value of the outcome variable for immigrant i from country of origin j in country of destination k; is the mean value of the outcome variable for immigrants in community jk, that is, immigrants from country of origin j in country of destination k; is the random “immigrant” effect, that is, the deviation of immigrants’ ijk’s score from the community mean. These deviations are assumed normally distributed with mean 0 and a within-community variance σ2.

π0jk eijk

The level-2 or “between-community” model is: π0jk = θp + b00j + c00k

[2]

with the following assumptions: b00j ~ N (0, τb00) c00k ~ N (0, τc00) where θp b00j

is the grand mean; is the random main effect of country of origin j, that is, the contribution of country of origin j averaged over all countries of destination, assumed normally distributed with mean 0 and variance τb00;

Data and Methods c00k

65

is the random main effect of country of destination k, that is, the contribution of country of destination k averaged over all countries of origin, assumed normally distributed with mean 0 and variance τc00.

Substituting Equation 2 into Equation 1 yields the (unconditional linear) cross-classified multilevel model: Yijk = θp + b00j + c00k + eijk

[3]

which is recognizable as a two-way analysis of variance with random origin effects, b00j; random destination effects, c00k; and withincommunity deviation, eijk. It is important to emphasize that Equation 3 does not contain a random interaction effect of the origin by destination combination, d0jk, that is, communities. Although this is possible, I did not assess these random interaction effects in this study. The main reason is methodological. As outlined by Raudenbush and Bryk (2002), the within-cell sample sizes in these kinds of model specifications are often not sufficient to distinguish the variance attributable to the random interaction effect τd00 from the within-cell variance σ2. Unfortunately, this also applies to the present study, in which many combinations of origins by destinations contain few respondents.8 Thus, in the double comparative design adopted in the present study, the variance at the community level is tapped by the random variance among origins and destinations, and no separate random variance component is estimated for their combination. The empty or unconditional model is useful for assessing the components of variance in outcomes that lie between countries of origin, between countries of destination, and between immigrants. However, an important aim of this study is to identify predictors that explain these variance components. These are so-called “conditional” models, in which predictors at level 1 and 2 are added to Equation 3.

8

I tried several models in which immigrants (level 1) were nested in communities (level 2), and communities nested in both countries of origin and countries of destination (crossed factors at level 3). These models, however, did not converge and could therefore not be used.

66

Immigrant Integration

Including predictors to Equation 1, the level 1 or “within-cell” model, becomes: P

Yijk = π0jk + Σ πpjkapijk + eijk p=1

[4]

with the following assumptions: eijk ~ N (0, σ2) The level-1 outcome is Yijk, the value of the outcome variable Y for case i nested within country of origin j and country of destination k of the two-way array, with i = 1, …, njk cases in cell jk; j = 1, …, J countries of origin; k = 1, …, K countries of destination. We see that level-1 predictors are apijk for p = 1, …, P. Examples of such predictors are length of stay in the host country, sex, marital status, and education. In addition to level-1 predictors, there are variables included at level 2, expanding Equation 2. The conditional level-2 or “between-cell” model is: Q

R

S

π0jk = θp + Σ βpqXpqk + Σ γprWprj + Σ δpsZpsjk + bp0j + cp0k q=1

r=1

s=1

[5]

Where: Xpqk are country of origin specific predictors, q = 1, …, Q. Wprj are country of destination specific predictors, r = 1, …, R. Zpsjk are cell-specific (i.e., community) predictors, s = 1, …, S. βpq are the fixed effects of the country of origin specific predictors. γpr are the fixed effects of the country of destination specific predictors. δps are the fixed effects of the community specific predictors.

Data and Methods

67

Two remarks are important about Equation 5. First, in this study, I only consider intercepts to be random. This does not exclude the possibility that slopes could be random as well. For instance, it could be that the effect of immigrants’ education on occupational status differs between countries of destination. But because of space limitations, I will not examine random slope models in this study. Second, although there is only random variance between origin countries and between destination countries, Equation 5 shows that community predictors are included. These community predictors are estimated at the origin-by-destination level, yielding correct standard errors. The variance between origins and between destinations is therefore explained by including predictors that refer to origins, destinations, as well as communities. Substituting Equation 5 into Equation 1 yields the (conditional and linear) cross-classified multilevel model [6]: P

Q

R

S

Yijk = θp + Σ πpjkapijk + Σ βpqXpqk + Σ γprWprj + Σ δpsZpsjk + bp0j + cp0k + eijk p=1

q=1

r=1

s=1

Until now, it was assumed that level 2 consists of j = 1, …, J countries of origins and k = 1, …, K countries of destinations. However, in several occasions in this study, the data structure is such that there are multiple observations within a single destination country. The reason is that in some countries surveys have been conducted in different years. One approach to this situation would be to add an additional random component to the equation, which pertains to “destination-year” cases. However, introducing this additional component would further increase the complexity of the overall model (which is already quite complex), resulting, in turn, in difficulties in model estimates (i.e., less stability, no convergence). An alternative way of dealing with this problem, adopted here, is to substitute the “destination” component for the “destination-year” component. That is to say, at level 2 there are j = 1, …, J countries of origins and k = 1, …, K countries of destinations per year. Thus, for instance, k1 = France in 1990, k2 = France in 1995, k3 = Spain in 1990, and so on. An advantage of this approach is that the cross-classified structure remains intact, and the level 2 variables that refer to time-dependent processes at the destination level are correctly estimated. A drawback of this solution is that the standard errors of the time-independent destination-level variables are underestimated. In

68

Immigrant Integration

order to examine how sensitive the results are to these “violations,” I will estimate and compare multilevel models in which (1) countries make up the destination level, and (2) models in which countries per year make up the destination level. Note that in both specifications, the number of communities is defined by the origin-by-destination combination, and does not expand when destinations vary over time.9 There is another complexity introduced in this study that needs some clarification. Equation 6 assumes that the outcome variable is linear. In several occasions throughout this study, however, the outcome variable is dichotomous, that is, having two values (e.g., employed vis-à-vis unemployed). In these cases, it is unwarranted to rely on the linear model, and one should instead use regression models that take into account the non-normal distribution of the dependent variable. It is important to mention how these regression models deviate from the linear specifications. The following assumptions are made: Yijk ~ Binomial (1, Pijk) Here, Yijk are the observed (0,1) responses, and Pijk is the probability that immigrant i from country of origin j in country of destination k has “success” (e.g., employment). As has become standard practice in multilevel modeling with dichotomous outcomes, it is assumed here that the distribution is binomial, that is, there exists no extra-binomial variance (i.e., no over-dispersion or under-dispersion). Also, there is no separate parameter for the variance at level 1, because the level 1 residual variance eijk is determined by the probability model: Pijk(1 – Pijk). In addition, one needs to transform the probabilities (that range from 0 to 1) in order to obtain results that are understandable. This is because, otherwise, introducing explanatory variables could result in fitted values that are negative or above 1 (i.e., outside the range from 0 to 1). In this study, I use the logit link function. Other link functions are 9

The reason is that the values of the community variables (e.g., group size, geographic distance) do not vary over time. MlwiN detects, for example, that the group size of immigrants from Algeria in France in 1990 equals that of Algerians in France in 1995, and therefore treats this as one observation: Algerians in France.

Data and Methods

69

probit or log-log functions, but the logit is more or less default nowadays. The Equation is as follows: Logit (p) = ln (p / 1 – p)

[7]

In Equation 7, the probability of “success” p (e.g., employment) is, first of all, divided by the probability of “failure” 1 – p. This is called the odds. Subsequently, the logarithm transforms the multiplicative to an additive scale and transforms the set of positive real numbers to the whole real line (Snijders and Bosker 1999). This is called the log odds or logit. The nice feature of the logit link function, is that it ranges from minus infinity to plus infinity. In order to interpret certain effects of the regression models, one can use the anti-logit function to transform the value back to the probability scale (i.e., exp β / 1 + exp β). Note finally that the logistic distribution for the level 1 residual implies a variance of π2/3 = 3.29 (Snijders and Bosker 1999:224). OTHER ISSUES IN COMPARATIVE MACRO RESEARCH The “Galton” Problem Another issue in cross-national research this study has to deal with is the so-called “Galton” problem (Goldthorpe 1997:9). The idea is that macro-level observations are not (always) independent from one another. The idea is that through processes such as cultural and economic diffusion, countries may converge. Most notably since the emergence of world system theory (Wallerstein 1974), the independence of national observations is questioned. Przeworski (1987) argued that the small N problem and Galton’s problem may run together in the end, and we have a world in which N = 1. Perhaps the best way to deal with this problem is to adopt a dynamic approach, in which countries are followed over a long period of time, and processes of diffusion are incorporated in the models. For the present study, however, dynamic data are not available, so this solution is of little use here. However, the severity of the Galton problem to the present study should not be exaggerated. First, this problem is not related to origin and community effects, but possibly only to destination effects. Second, there are many areas of comparative

70

Immigrant Integration

research in which the Galton problem does not occur. The problem, according to Goldthorpe (1997), only pertains to issues on economic and social policy. That would imply that, in this study, only the destination effects that relate to policy making may suffer from the Galton problem. The variable included in this study refers to immigration policies of Western societies, and one could argue that these are not independent from one another. However, this variable contrasts countries with a strict immigration policy (i.e., selecting immigrants on a number of skills) with countries without such selection. Because the two countries that have adopted such a strict policy, Australia and Canada, have done so for a considerable period of time without change (i.e., since the 1960s onwards), it seems safe to conclude that the Galton problem is not relevant for this variable either. The Black Box Problem The final problem associated with cross-national research is called the black box problem (Goldthorpe 1997). The idea is that researchers could account for a significant part of the variation between countries, without actually understanding the relationship they found. It is argued that comparative research identifies dependent and independent variables, and that the results that come out of the black box of statistical models are unable to tell what is going on in terms of individuals who act intentionally. Some authors have suggested that such theoretical interpretations are dependent upon case studies (Huber, Ragin, and Stephens 1993). There are several ways in which the present study deals with the black box problem. First, I do not merely use independent and dependent variables, but, instead, systematically derive relationships between independent and dependent variables from theories. As argued in Chapter 2, these theories are very general (i.e., highly informative), which implies that they move beyond explaining one particular empirical finding. This is an important way to overcome the problem of black boxes (Goldthorpe 1997; Kiser and Hechter 1991). Moreover, whereas these theories account for macro-level effects, they could be phrased perfectly well at the micro level in terms of purposively acting individuals. These statements vary from the assumption that individuals rationally invest in themselves (human capital theory), have different

Data and Methods

71

chances of meeting other people (structural opportunity theory), prefer some people more than others (prejudice theory), or rely on ties and resources of other people (social capital theory). This was shown in Chapter 2. Thus, using general theories that go beyond a single case, and which contain statements on the individual level, is a first way to overcome the problem of black boxes. Second, the interpretation of the relationships found in standard quantitative regression analyses depends on the extent to which the indicators in the models represent the theoretical ideas behind them. It is thus important to realize that there are two “languages” in scientific research: a theoretical language that specifies relationships in terms of individuals acting purposively, and a measurement language that is to a certain degree a representation of these ideas (Blalock 1961). The distinction is a matter of degree: the more direct the measure, the more likely that the interpretation of the effects indeed represents the underlying ideas. In this study, as mentioned above, I have attempted to measure as directly as possible the theoretical concepts that are included in the analysis, which, in turn, is another way to treat the black box problem effectively.

This page intentionally left blank

CHAPTER 4

Employment

An important aspect of immigrant integration is their position in the labor market. The first step towards economic incorporation is the participation in the labor market as opposed to being inactive. A major second transition is that immigrants find a job instead of remaining unemployed. It is well known in the literature that immigrants are more often inactive in the labor market and have higher unemployment rates than natives. This chapter examines to what extent and why immigrants’ labor force participation and unemployment are an outcome of the country of origin, the country of destination and the immigrant community. Figure 4.1. Une mployme nt rate (%) of immigrant groups. Male s, 18 We ste rn countrie s, 1980-2001 35 30 25 20 15 10 5 0 Germany

Italy

Morocco

73

T urkey

UK

74

Immigrant Integration

As a first exploration of their importance, Figure 4.1 presents the unemployment rate of a small sample of the IFIS data set. It uses a comparative origin design, and focuses on five immigrant groups: Germans, Italians, Moroccans, Turks and British. The figures refer to the unemployment rates of males, aged 25 to 54, observed across 18 Western countries, in the period between 1980 and 2001. The unemployment rate indicates the population that is employed, expressed as a percentage of the active population in the labor market, thus excluding inactive persons (e.g., students, housekeepers, disabled). Figure 4.1 presents strong evidence for group differences in unemployment. It appears that almost 30 percent of male immigrants from Morocco are unemployed. The Turks also have a high unemployment rate: about 19 percent. By contrast, immigrants from the United Kingdom, Italy and Germany are more often employed. Slightly more than five percent of immigrants born in Germany are unemployed. From a comparative destination perspective, a different question can be raised: are there also differences in unemployment levels of immigrants across destinations? Figure 4.2 presents the unemployment level of immigrants, subtracted by the unemployment level of native males of the same age. It shows that in Italy the (relative) unemployment rate of immigrants is almost 19 percentage points higher than among natives. The unemployment rates of immigrants are also high in Belgium, France, the Netherlands and Sweden. By contrast, small differences are observed in Australia, Luxembourg, Spain, the United States and Canada. In the United States the unemployment rate of immigrants is less than 1 percentage points higher than among natives; in Canada, immigrants are even slightly less often unemployed than natives. This book combines the comparative origin perspective with the comparative destination perspective into a double comparative approach. To see why a simultaneous study of multiple origin groups in multiple destinations is fruitful, Table 4.1 again presents the unemployment rate of the five selected immigrant groups, but now separately across the 18 Western countries. To obtain reliable estimates, figures are only presented when there are at least 500 respondents.

Figure 4.2 Differences in unemployment rates of male immigrants and natives (%) Canada USA Spain Luxembourg Australia Germany UK Denmark Finland Sweden Netherlands France Belgium Italy -5

0

5

10

15

20

25

76

Immigrant Integration

Implicitly, the comparative origin approach assumes that differences found between immigrant groups “travel” across destinations. And, similarly, the comparative destination perspective assumes that destination differences pertain to all groups. Are these assumption true? The answer is no. Compare, for instance, the unemployment figures of male Turks in Belgium (N = 2,116 ) and Germany (N = 1,936). In Belgium, 33.3% of the Turkish immigrants in the labor market are unemployed, which is above the mean unemployment rate for Turkish immigrants in all destinations (19.4%) and also above the mean for all immigrants in Belgium (18.6%). In contrast, in Germany, 10.6% of the Turks are unemployed. This is lower than the average Turkish unemployment rate in all the destinations observed, and lower than the unemployment rate in Belgium. This difference exceeds the 10% overall difference in immigrants’ unemployment levels between Belgium and Germany. This means that the specific situation of the Turks in Germany and Belgium determines their deviance from the pattern expected from general origin and destination effects. Apparently, the immigrant community plays an important role. The table also shows that some differences across immigrant groups travel across destinations. For example, the unemployment rate of Italians is always lower than that of Turks, for all countries in which both groups are observed. In Austria, 12.8% of the Turks are unemployed, against only 3.8% among the Italians. Similarly, in Belgium, 14.1% of the Italians do not have a job, whereas this is true for 33.3% of the Turks. Because the groups are observed in the same destinations the differences seem to reflect characteristics of the countries of origin. Finally, the double comparative approach suggests that host countries are important for immigrant integration. Comparing the experiences of two origin groups (Germans and Italians) that can be traced in two destinations with diverging employment levels (Belgium and Luxembourg) replicates the pattern found for the total immigrant population: both German and Italian immigrants are more often unemployed in Belgium than in Luxembourg. In summary, this suggests that the country of destination influences the employment of male immigrants, irrespective of the country of origin.

Table 4.1

Unemployed male immigrants 25-54 years of age, 1980-2001 (% of active population) Country of origin Destination Germany Italy Morocco Turkey UK Mean all groups Mean natives Australia 3.8 12.8 4.6 6.2 4.9 Austria 3 12.8 9.6 4.1 Belgium 5.5 14.1 32.6 33.3 18.6 5.3 Canada 6.8 6.8 4.9 8.6 8.8 Denmark 11.4 5.4 Finland 17.2 10.9 France 32.2 21.5 8.2 Germany 5.3 10.6 8.3 5.2 Greece 7.2 5.6 Ireland 3.8 16.9 15.8 10.8 Italy 28.2 28.7 9.1 Luxembourg 1.7 3.3 2.5 1.4 Netherlands 5.7 26.3 21.6 6.9 16 3.2 Portugal 5.4 3 Spain 12.5 11.4 Sweden 17.5 6.8 UK 8.9 7.9 12.4 8 USA 2.9 4.5 3.2 5.3 5 Mean 5.7 8.5 29 19.4 11 10.3 5.6 Note: Figures on communities with fewer than 500 respondents not presented. Source: Van Tubergen, Maas and Flap (2004).

78

Immigrant Integration

The descriptive explorations suggest that the unemployment chances of male immigrants are affected by their country of origin, their country of destination and their immigrant community. Interesting as these results are, they leave a lot of questions unanswered. Are the unemployment chances of females equally affected by their origin, destination and community? Are there similar differences with respect to labor force participation? What about the other origin groups not included in the overview, such as the Indians, Japanese, Mexicans? And, most importantly, why do immigrants’ country of origin, country of destination, and the immigrant community play such a pivotal role? MEASUREMENT EMPLOYMENT STATUS It is important to discuss the measurement of employment status first. To enhance cross-national comparability, only those surveys are selected for which employment status could be classified using international standard classifications, using the standard definitions of labor-force status adopted by the 13th International Conference of Labour Statisticians in 1982. The following three categories are distinguished: (1) Employed:

(2) Unemployed: (3) Inactive:

Those who have paid employment (who were either at work during the reference week or had a job but were temporarily not at work) or are self-employed (at work during the reference week or had a business but were temporarily not at work). Those without work or currently available and seeking work. All others, such as homemakers, students, and retired persons.

Using this scheme, I constructed two dichotomous dependent variables of labor-force status. First, I examine labor-force participation, considering the odds of an immigrant being either employed (category 1) or unemployed (category 2) vis-à-vis being inactive (category 3). Subsequently, I select the active population and use that sub-sample to

Employment

79

examine employment –and the odds of being employed (category 1) versus being unemployed (category 2). DESCRIPTIVE FINDINGS For a further assessment of macro effects on immigrants’ employment status, descriptive figures are presented for male labor force participation (Table 4.2), female labor force participation (Table 4.3) and female unemployment (Table 4.4). Because such information could, of course, not be presented here for the more than 900 communities included in the data set, I again present the figures for immigrants from Germany, Italy, Morocco, Turkey, and the United Kingdom. Table 4.2 shows that, similar to unemployment, male labor force participation rates are lower among Turkish and Moroccan immigrants than among Germans, Italians, and British immigrants. About 86% of the male Turks between 25 and 54 years of age is active in the labor market, compared to 91% of the Italians and 93% of the Germans. Among natives, about 94% is active in the labor market. Contrary to findings on unemployment, the labor force participation rates of immigrants are particularly high in Italy. In Italy, 98% of the male immigrant population participates in the labor market, of which, according to Table 4.1 a high percentage is unemployed. Participation rates of immigrants are lower in Scandinavian countries, where they clearly fall below that of natives. In Denmark 86% of the male immigrant population participates in the labor market, as opposed to 94% of natives. Similar low activity rates are observed in Sweden, Belgium, and the Netherlands. Table 4.2 suggests that the immigrant community is also important. Again, it is observed that the Turkish community in Belgium is badly integrated economically. Whereas 86.6% of all immigrants in Belgium participates actively in the labor market, and that is true for 86.7% of the Turks across the 18 Western countries, in Belgium their activity rate is only 79.5%. The patterns for men are mostly generalizable to women. Similar to findings on men Table 4.3 (labor force participation) and Table 4.4 (unemployment) show that Turkish and Moroccan women are less well incorporated economically than other origin groups. Only 36% of the Moroccan women and 45% of the Turkish women are active in the labor market, whereas this is true for 63% of the total immigrant

Table 4.2 Labor-force participation of male immigrants 25-54 years of age, 1980-2001 (%) Country of origin Destination Germany Italy Morocco Turkey UK Mean all groups Mean natives Australia 93.7 93.0 . 83.1 95.7 90.9 93.1 Austria 91.8 . . 94.1 . 93.9 94.0 Belgium 91.6 85.9 84.2 79.5 . 86.6 93.5 Canada 95.7 92.1 . . 95.9 91.0 92.9 Denmark . . . . 95.1 86.0 93.8 Finland . . . . . 92.1 90.2 France . 93.4 91.8 95.1 . 93.1 95.7 Germany n.a. 96.7 . 95.4 . 95.8 95.0 Greece . . . . . 97.5 94.6 Ireland . . . . 92.4 90.9 92.6 Italy . n.a. 98.2 . . 97.9 92.9 Luxembourg 93.6 92.9 . . . 95.0 94.3 Netherlands 92.6 . 77.7 80.9 92.5 84.6 95.6 Portugal . . . . . 92.3 93.7 Spain . . 94.5 . . 92.5 93.0 Sweden . . . . 87.1 84.2 91.8 UK 93.6 . . . n.a. 89.9 93.5 USA 94.4 94.6 . . 94.9 90.5 92.4 Mean 92.8 91.1 85.6 86.7 93.4 90.9 93.5 Note: Figures on communities with fewer than 500 respondents not presented. Source: Van Tubergen et al. (2004).

Table 4.3 Labor-force participation of female immigrants 25-54 years of age; 1980-2001 (%) Country of origin Destination Germany Italy Morocco Turkey UK Mean all groups Mean natives Australia . 48.8 . . 60.5 58.5 57.2 Austria 72.2 . . 55.7 . 72.3 77.1 Belgium 64.8 48.0 31.0 29.0 . 50.7 73.6 Canada 77.4 69.8 . . 81.5 75.8 78.0 Denmark . . . . . 70.2 86.1 Finland . . . . . 75.2 85.9 France . . 40.5 . . 57.4 81.1 Germany . 56.9 . 48.9 . 63.0 75.4 Greece 67.7 . . . . 68.1 59.9 Ireland . . . . 60.4 60.8 58.6 Italy . . . . . 80.5 58.9 Luxembourg 53.0 50.5 . . . 63.2 56.6 Netherlands 73.3 . 32.8 38.9 . 58.9 74.6 Portugal . . . . . 77.1 76.4 Spain . . . . . 55.0 58.9 Sweden . . . . . 73.5 87.8 UK 73.9 . . . . 62.6 76.6 USA 65.9 55.3 . . 63.8 65.4 70.6 Mean 68.5 55.5 36.4 44.5 63.7 63.4 70.3 Note: Figures on communities with fewer than 500 respondents not presented. Source: Van Tubergen et al. (2004).

Table 4.4 Unemployed female immigrants 25-54 years of age, 1980-2001 (% of active population) Origin Destination Germany Italy Morocco Turkey UK Mean all groups Mean natives Australia . . . . 4.4 7.8 5.5 Austria 6.4 . . 17.9 . 9.5 5.1 Belgium 14.5 26.3 45.5 . . 22.7 10.3 Canada 7.5 8.5 . . 5.6 9.5 8.4 Denmark . . . . . 13.9 7.9 Finland . . . . . 26.6 11.6 France . . 46.6 . . 27.6 12.9 Germany n.a. . . 13.9 . 13.8 10.1 Greece . . . . . 17.9 14.2 Ireland . . . . 16.4 16.0 11.1 Italy . n.a. . . . 20.6 19.5 Luxembourg . . . . . 4.8 3.0 Netherlands 10.8 . . 17.4 . 14.4 7.2 Portugal . . . . . 7.0 4.7 Spain . . . . . 18.2 23.6 Sweden . . . . . 16.4 5.6 UK 7.0 . . . n.a. 8.7 5.9 USA 5.8 . . . 3.9 7.2 5.2 Mean 9.0 13.8 36.3 21.2 11.5 11.3 7.3 Note: Figures on communities with fewer than 500 respondents not presented. Source: Van Tubergen et al. (2004).

Employment

83

population and for 70% of the native population. I also find that – similar to males–labor force activity rates in Italy are high among female immigrants, but that a substantial part of them is unemployed. In Belgium, the Netherlands and the Scandinavian countries participation rates are low and unemployment rates high. A particularly problematic situation is that of Moroccan women in Belgium, of whom only 31% participates in the labor market, and of those who are active in the labor market, 46% is unemployed. Thus, for both men and women, and with respect to both labor force participation and unemployment, I find that immigrants’ country of origin, country of destination and the immigrant community are important. VARIANCE COMPONENTS Another way to assess the importance of macro processes for immigrants’ employment status is to look at the variance components. How much do the labor-force participation and employment of immigrants differ between origin countries and between destination countries? Table 4.5 provides an answer to that question. It should be noted that the logistic distribution for the level-one residual implies a variance of (π2 / 3 =) 3.29 (Snijders and Bosker 1999). I present the findings of two different random intercept models, both without the inclusion of explanatory variables: one in which destinations are treated as time-invariant (1a) and one in which destinations vary over time (1b). The analysis shows that considerable variation exists between origins and between destinations. Because about 180 different origin groups and only 18 (Western) countries exist in the data set, the strong variation of immigrants’ labor-force status across receiving countries is quite remarkable. It appears that for males, labor-force activity varies more among destinations than among origins (Model 1a: 0.463 compared to 0.288), whereas for females, at least in the data used here, conditions in the country of destination are less important than the characteristics of the country of origin (Model 1a: 0.132 compared to 0.305). With respect to employment, this study finds that for females, employment chances are more affected by the situation in the receiving society than by the characteristics of the sending nation. For males, this

Table 4.5

Variance components Labor-force participation Origin Destination

Males 1a Destination time-invariant 1b Destination per year

.288 (.045) .292 (.045)

.463 (.184) .445 (.071)

Employment Origin Destination .427 (.063) .359 (.059)

.378 (.147) .510 (.072)

Females 1a Destination time-invariant .305 (.043) .132 (.049) .276 (.044) .305 (.120) 1b Destination per year .295 (.042) .158 (.024) .264 (.042) .389 (.059) Note: Standard errors in parentheses. Results obtained from cross-classified multilevel logistic regression models without explanatory variables. Source: Van Tubergen et al. (2004).

Employment

85

is less straightforward. When countries of destination are treated as time-invariant, employment chances are more affected by the country of origin than by the country of destination. When destinations are varied over time, however, the employment chances of males differ more among destinations than among origins. This is most likely because employment chances change considerably over time. In general, the variance among destinations increases with time-variant rather than time-invariant measures, but the difference between the two is larger with respect to employment than with respect to labor-force activity. All in all, I find that for both male and female immigrants, labor-force participation and employment chances clearly depend on their origin and destination. DETERMINANTS OF IMMIGRANT EMPLOYMENT Human Capital, Discrimination, and Social Capital Why are there differences in employment status across immigrants’ sending nations, host countries and ethnic communities? To answer this question, I rely on three theories that were introduced in Chapter 2. The first theory is that of human capital, developed mainly in the economics literature (Becker 1964; Borjas 1987; Chiswick 1978). According to this theory immigrants’ economic position is positively determined by their human capital, such as educational level, work experience, language fluency, and individual talents. One important distinction is that between general human capital, reflecting general potential productivity (e.g., education, health) and host-country specific human capital, indicating the knowledge and skills specific to a certain country (e.g., language, educational qualifications). Another relevant distinction is that between observable and unobservable skills (Borjas 1987; Chiswick 1978, 1979). Observable skills include such things as educational level, command of the destination language, and labormarket experience. Ability, motivation, and talent are generally assumed to be unobservable skills. Within a human-capital framework, macro effects are assumed to result from the skill composition of immigrant groups. What makes the human-capital framework interesting for the study of origin, destination, and community effects, is that macro factors play

86

Immigrant Integration

a role in the selection or composition of immigrants’ skills. The selection of immigrants can be either positive (or favorable), with immigrants selected for their high skills, or negative (unfavorable), with immigrants selected for their lower productivity and skills (Borjas 1987, 1988). The properties of origins, destinations, and communities determine the selective pattern of migration in terms of human-capital. Within a human-capital framework, the role of immigrants’ country of origin, country of destination and the immigrant community are therefore explained in terms of compositional differences. Composition effects arise when individual characteristics are not distributed equally across these macro units, and when these characteristics also affect the dependent variable (in this case, economic integration). Within a more sociological framework, researchers have argued that macro factors provide opportunities and create restrictions for immigrants, indicating contextual influences (Model and Lapido 1996; Portes and Rumbaut 1996, 2001). Contextual effects occur when the characteristics of macro units have a direct effect on individuals’ outcomes, over and above the effects of individual characteristics. Societal conditions are, first of all, linked to processes of ethnic prejudice and discrimination, which foster or hinder migrants in becoming fully integrated into the labor market. It is generally assumed that discrimination of immigrants depends on the ethnic threat perceived by the native majority. In addition, discrimination can be diminished by government policy. Besides discrimination, immigrants’ social capital determines his employment status. Those with better connections, are more likely to participate in the labor market and to find a job. It is important to emphasize that the hypotheses refer to immigrants’ economic incorporation in general. Whereas economists have primarily used earnings as a measure of economic integration, sociologists also have studied occupational status and labor-force status as economic outcomes. I apply the human capital theory, prejudice theory and social capital theory to both labor-force participation and employment. Assuming that people participate in the labor market to find jobs, there is a certain association between the decision to participate in the labor market and one’s chances of finding employment. From a humancapital perspective, it is assumed that people with more human capital have more opportunities and incentives to become employed, and therefore participate more often in the labor market (Duleep and

Employment

87

Sanders 1993). Within a sociologic perspective, it is assumed that those who experience discrimination and have fewer social capital are more often unemployed and less willing to participate in the labor market. Analysis and Models Because the dependent variables in the analysis are dichotomous, logit regression techniques are used. This involved, first, estimating models predicting the odds of immigrants being active in the labor market and, then, selecting the immigrants who were active to examine the odds of immigrants being employed. Both labor-market outcomes were estimated separately for males and females, because patterns in labormarket position generally differ by sex. I present the results of the cross-classified multilevel logistic regression models for immigrants’ labor-force status in Table 4.6. I estimated models without individuallevel variables to examine the compositional hypotheses (Models 1 and 3) and models that included individual-level variables to test the contextual hypotheses (Models 2 and 4). The two outcome variables analyzed fluctuate with the business cycle, especially unemployment. Instead of treating the role of destinations in a static way, I therefore made destination countries timedependent in the analysis. This resulted in more than 100 “destinationyear” cases as the destination component instead of only 18 destination countries. In this dynamic design, I was able to include precise macrolevel control variables that measure the average labor-force participation and unemployment rates among native males and females for each destination country in each survey year. In addition, it results in a better estimate of the time-dependent predictor in the analysis (i.e., the presence of left-wing parties 10 years before the survey year). In summary, this results in a two-level model, with a random intercept for individuals at the lower level, and random intercepts for countries of origin, and countries of destinations per year at the higher level. To examine the hypotheses this chapter starts with a model with only macro-level variables. This model was used to examine the hypotheses related to the selection of immigrants’ skills. Individuallevel variables were added in a second model. In general, macro-level effects that indicate skill selection should decrease after human-capital variables are included. However, in the present study, it is unlikely that

88

Immigrant Integration

the effects of these macro-level variables would disappear or even be diminished. This is, first of all, because several “observable” skills were not measured (e.g., labor-force experience) or were only partly measured (e.g., language proficiency) and because “unobserved” skills (e.g., talents) were, of course, completely omitted. Second, with multilevel models that have dichotomous outcomes, the coefficients of the macro-level variables as well as the variance at the macro level tend to increase after micro-level variables with strong effects have been included (Snijders and Bosker 1999). The reason for this is that the residual level-one variance is fixed. I therefore refrain from inspecting changes in the effects of macro-level variables and changes in variance components after adding micro- and macro-level variables, nor did I inspect deviance statistics. It should be noted further that models including individual variables were used to test the contextual effects, for which it is relevant to take composition effects into account. Appendix A describes the measurement of the individual-level variables, while Appendix B provides an overview of the macro-level variables. Destination Effects How can we understand the role of receiving nations in the employment status of immigrants? Human capital theory argues that immigration policy is important, because some countries select more strongly on the skills of immigrants than other nations. Two countries, in particular, have a strict immigration policy: Australia and Canada. In these countries, migrants who apply for an entry visa must meet specific requirements. This so-called “point-system” rates migrants according to their language fluency, job experience, education, job opportunities, and several other characteristics (Borjas 1988; Reitz 1998). Immigrants who pass are therefore assumed to be more skilled, more talented and more needed for the economy than those who fail. Like the United States, European countries did not have a point system during the period I study and immigrants in these countries are less favorably selected than immigrants in Australia and Canada. However, Table 4.6 provides no support for the predicted positive impact of host countries that have a point system. Measured in terms of labor-force

Employment

89

activity and employment, these countries (Australia and Canada) do not seem to select more employable immigrants to their economy. The results do highlight the role of another contextual factor in immigrants’ employment: labor market integration policies. The presence of left-wing parties in the government during the 10 years before the survey year has a positive effect on the economic position of immigrants, except for labor-force participation of females. In general, then, the longer left-wing parties have been in the government of a specific country, the more immigrants in that country are active in the labor market and the more often they are employed. This confirms the suggestion that left-wing parties take more measures than other parties (i.e., liberal, conservative, and Christian-democratic) to reduce inequalities within a country (Hewitt 1977; Lenski, Lenski and Nolan 1991; Smits, Ultee and Lammers 1998), which also includes diminishing ethnic inequalities. These measures (“equal opportunities,” “positive discrimination”) are directly concerned with reducing ethnic discrimination in the labor market. Origin Effects My results show that the influence of the country of origin is an outcome of both selection and discrimination processes. Higher levels of political suppression in the sending country are associated with lower levels of immigrants’ labor-force activity and employment. The relation is significant for both males (Model 1) and females (Model 3). Political suppression and instability in the origin country induce people to migrate for other than pure economic reasons, leading to less talented and less skilled immigrants at arrival in the host country (Chiswick 1978, 1979, 1999). Furthermore, many immigrants from politically suppressive societies are “refugees,” who have experienced traumatic events and who have resided in refugee camps, hindering the acquisition of (host-country specific) human capital. Differences across immigrant groups that travel across receiving nations are also associated with discrimination. The analysis shows that immigrants from predominantly Christian countries participate more often in the labor market and are more often employed than immigrants from non-Christian countries, with the exception of the employment of males. Non-Christian immigrant groups are perceived by natives

Table 4.6 Cross-classified multilevel logistic regression of immigrants’ economic incorporation in 18 Western countries, 1980–2001 Labor force activity Employment Males Females Males Females Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4 Constant -3.265 -2.516 -1.659 -.491 3.319 2.567 3.225 1.554 Destination Point system -.340 -.436* -.030 .150 .185 -.105 .112 .074 Left-wing parties government .080** .089** .033** .024 .068** .067** .097** .096** Origin Political suppression -.054** -.030** -.013* .000 -.044** -.044** -.072** -.050** Predominantly Christian .384** .453** .240** .259** .168* .129 .255** .240** Labor-force activity females .010** .008** Community Gini ratio (origin/destination) -.361** -.454** .031 .057 -.586** -.359** -.828** -.601** GDP per capita ratio -.440** -.432** -.129** -.171** -.304** -.269** -.449** -.411** (origin / destination) Geographic distance -.006 -.011** -.008** -.010** .030** .024** .025** .022** (1000s km) Official language -.089 -.076 .015 .052* .167** .144** .018 .031 Relative group size (%) .032** .022* .065** .054** .018 .008 .039** .023 Mean educational level,group -.023 -.163** .351** -.070* .455** .204** .392** .188*

Micro-level control variables Age -.021** -.016** Duration of stay 0-5 years ref. ref. 6-10 years .735** .384** 10+ years .712** .591** Education Low ref. ref. Middle .484** .481** High .645** 1.022** Language proficiency Not fluent ref. ref. Fluent .250** .105** Unknown -.335** .009 Macro-level control variables Migration survey .316** -.264 -.367** -.654** Labor-force activity natives (%) .070** .067** .018** .011** Unemployment natives (%) * p < .05, ** p < .01 (two-tailed tests). Source: Van Tubergen et al. (2004).

.007**

.013**

ref. .360** .453**

ref. .437** .630**

ref. .333** .819**

ref. .307** .710**

ref. .301** -.042

ref. .323** .409**

-.349**

-.619**

.273**

.417**

-.138**

-.121**

-.078**

-.074**

92

Immigrant Integration

as more culturally threatening to their predominantly Christian culture, values and norms (Lieberson 1980; Portes and Rumbaut 1996, 2001), leading to stronger discrimination against immigrants with such culturally distinct origins (Evans and Kelley 1991; Lindley 2002; Model and Lin 2002; Portes and Rumbaut 1996, 2001; Wanner 1998). Surprisingly, pre-migration exposure to the destination language only affects the employment status of males positively. Male immigrants who have moved from a country in which the destination language is official have a greater chance of being employed than do males who were not exposed to the host language. Language exposure does not affect the employment chances of females, however, nor does it have any effect at all on labor-market participation. Because language proficiency measured at the individual level has a positive effect on immigrants’ labor market opportunities according to my findings, it suggests that those who were exposed before migration to the host language are less favorably selected in other ways – in a way not incorporated in the present analysis. Another interpretation is that language proficiency is somewhat less important for the decision to participate in the labor market and for finding a job, but essential for the quality of the job and their earnings. Indeed, earlier research in the United States has shown that immigrants from countries in which English is an official language have significantly higher earnings (Borjas 1987; Jasso and Rosenzweig 1990a). Another context effect relates to the labor-force activity of females. The labor-force activity of females in the country of origin has a positive effect on their labor-force activity in the destination country, whatever that destination may be. Cultural patterns of female laborforce participation in the origin country are carried over to the destination country (Antecol 2000), which reflects the “discrimination” within an immigrant group against women. In more traditional origin countries, immigrants are socialized with more unequal gender roles, leading to an enduring lower participation rate of women from such countries –even when they move to modern Western countries. Community Effects Community differences are first of all the result of skill selection. The higher the income inequality in the country of origin relative to that in

Employment

93

the country of destination, the less often immigrants participate in the labor market and the less often they are employed. The exception to this pattern is the labor-force activity of females. In general, however, the less talented are the most likely to emigrate from countries with a more dispersed income distribution, and they are most likely to migrate to countries with less income inequality. On the other hand, emigration from societies with more equally distributed incomes tends to be largely concentrated at the upper end of the home country’s income distribution. According to Borjas (1988: 25), this is because “the home country is taxing high-ability workers and insuring low-ability workers against poor labor market outcomes.” Similarly, host countries with high-income inequality attract migrants with high-unobserved abilities (Borjas 1987; Cobb-Clark 1993). Skill selection is also related to the economic development in origin and destination, although the relationship is complex. My results show that the higher the GDP per capita in the sending nation relative to that in the host society, the more often immigrants are outside the labor market, and if they are in the labor market, they are more often unemployed.10 This contradicts the idea that people from relatively advanced economies are more favorably selected in terms of observable skills and the transferability of their skills (Borjas 1987, 1988; Chiswick 1978, 1979; Cobb-Clark 1993; Jasso and Rosenzweig 1990a). Other things being equal, migrants from developing countries should have lower human-capital skills than those originating from more advanced economies. In addition, migrants from rich countries have an edge over migrants from poorer countries regarding the transferability of their skills and subsequent rewards. Why then, do I find a negative impact of the GDP ratio? One possibility is that the disadvantages for immigrants from poorer countries are offset by their post-migration investments in human capital. Immigrants from poorer nations have more to gain from permanent settlement and may, in turn, invest more in human capital. That may include learning the host language, but also obtaining additional education after migration. Geographic distance between origin and destination countries also could affect skill selection. It is assumed in the literature that greater 10

Note that a negative effect of the GDP ratio remains after excluding a possibly associated factor, the average educational level of the immigrant community (analysis not presented here).

94

Immigrant Integration

distance increases migration costs (Borjas 1987; Jasso and Rosenzweig 1990a). As a consequence, individuals with more unobserved human capital are overrepresented among migrants who have moved over long distances. Geographic distance also diminishes the likelihood of return migration (Borjas 1987), which, in turn, increases the incentives to invest in human capital, such as acquisition of the destination language (Chiswick and Miller 2001). It is expected that greater geographic distance between origin and destination countries has a positive impact on immigrants’ economic performance. In line with this argument, I find that a greater distance is associated with higher odds of an immigrant being employed. However, I find the opposite pattern with respect to labor-force activity. Migrants, especially females, who have moved a longer distance are more likely to remain outside the labor market. Next to skill selection, the importance of the immigrant community is related to social capital and discrimination. One contextual characteristic, informed by both notions, is the average educational level of the community. When an immigrant community is highly educated, it has more resources that could benefit the members of the community. Higher educated people have more financial resources and more information about job opportunities. Furthermore, members of higher educated communities are less discriminated by natives. These communities are more favorably perceived by natives, because their norms and values are more similar to their own than those of lower educated communities, and education is associated with economic productivity as well. In accordance with this line of reasoning, I find that the higher the educational level of an immigrant group, the more likely the members of that group are to have a job. With respect to labor-force activity I find, unexpectedly, the opposite pattern: lower education of an immigrant community is associated with higher levels of labor-market activity among the members of that group. Differences across immigrant communities are also related to their size. I find that members of larger immigrant groups have a higher probability of being active in the labor market. Group size has no effect on employment chances, however. The partially positive effect of group size on employment chances can be interpreted in terms of social capital (Evans 1989; Portes and Bach 1985; Portes and Rumbaut 1996; Zhou and Logan 1989). In the same way that natives prefer in-group members, immigrants are more willing to help co-ethnics. Members

Employment

95

from the same immigrant group help each other by offering jobs, buying goods, and lending money. In larger groups, there are naturally more opportunities for immigrants to be hired by co-ethnics. Note that the positive effect of group size on employment is not in line with the suggestion that prejudice and discrimination increase with the relative size of the immigrant group (Blalock 1967; Frisbie and Neidert 1977; Quillian 1995; Tienda and Lii 1987; Tolnay 2001). Possibly, larger immigrant groups are indeed perceived as more threatening, but their own immigrant community provides a shelter to discrimination by providing job opportunities within the own ethnic community. Sensitivity Analyses Because a cross-national analysis using different surveys raises questions about comparability, I performed a sensitivity analysis. Although this study carefully selected surveys that are comparable across nations, and the core of the data set had already been standardized (i.e., the European Union Labour Force Surveys), the sensitivity analysis provides additional insight into the robustness of the findings. I examined how sensitive the findings were to (1) differences between surveys (excluding specific immigrant surveys), (2) the selection of samples (excluding Finland and communities with fewer than 50 respondents), and (3) different multilevel designs (i.e., treating destinations as time-constant instead of looking at destinations by year). In general, I find some instances in which previously significant effects are insignificant or marginally significant, but there are also findings that are closer to the hypotheses than in my original analysis (results not shown). There are few deviations, however, and they do not systematically pertain to specific findings. The overall conclusion is that there is a strong similarity between the original and the additional analyses, suggesting that the findings are quite robust for differences in surveys, samples, and multilevel designs. ILLUSTRATING THE RESULTS This chapter has shown that skill selection (“composition”), social capital and discrimination (“context”) can explain, to a certain extent,

96

Immigrant Integration

the impact of the country of origin, the country of destination, and the immigrant community on the labor-force status of immigrants. To illustrate how these processes work, I have simulated the findings and give concrete examples, focusing on the risk of unemployment for male immigrants. To begin with, the importance of composition effects, as presented in Table 4.7, Model 1 (employment), should be considered. Besides the unexpected finding of the point system and the GDP ratio, four selection factors play a role: political suppression, Gini ratio, geographic distance, and official language. Communities can be distinguished that are favorably selected in all these dimensions from groups that are unfavorably selected. According to the estimates of Model 1 (Table 4.6), and using the range of the variables (not shown) among the male sample, the difference in the predicted chances for employment between these groups could be as high as 55%.11 In a similar vein, the maximum difference across groups could be computed on the grounds of the two significant context variables found in Model 2: left-wing parties in the government and mean education level. The difference in the likelihood of employment could amount to 23%.12 However, in reality, groups with such extreme properties in all dimensions are rare. To illustrate more realistically the magnitude of selection and context effects, I used scores above and below the mean to distinguish between groups with favorable or unfavorable selection and social context. The results are presented in Table 4.7 and, again, refer to the unemployment chances of male immigrants. An example is the role of selection mechanisms. Of the 922 immigrant communities in the male sample, 27 are favorably selected. That is, the immigrants who moved from countries with little political suppression (i.e., below the mean of 8.33), have a low Gini ratio (below 1.16), have a greater distance 11

Computed as follows: e -.044 x 2 + -.586 x .057 + .030 x 19.84 + 1 x .167 / ( e -.044 x 2 + -.586 x + 1) - e -.044 x 14 + -.586 x 2.57 + .030 x 0.17 + 0 x .167 / ( e -.044 x 14 + -.586 x 2.57 + .030 x 0.17 + 0 x .167 + 1). 12 From this example, it should not be concluded that selection effects in general are more important than context effects for the economic status of immigrants. With regard to female labor-force participation, for instance, I find three significant context effects, which are more important than the one selection effect. .057 + .030 x 19.84 + 1 x .167

Employment

97

between origin and destination (more than 5.57 thousand kilometers), and have an official language in the destination country resembling that in their country of origin. There are 97 communities unfavorably selected in all four dimensions and 798 groups that scored favorable on some dimensions and unfavorable in others. Table 4.7 shows that among the 27 favorably selected communities, the mean percent employed is 93 (i.e., 7% unemployed) and in the 97 unfavorably selected communities the mean percentage employed is 74% (i.e., 26% unemployed).13 Table 4.7 also presents concrete examples of communities with a favorable or unfavorable selection or social context. I illustrate the findings for selection effects with an example of a favorably selected group: the British immigrant community in Australia. It is interesting to contrast this group with Iraqi immigrants who have settled in Sweden. The British immigrants already had a head start in leaving a country with the least political suppression (score 2), which indicates no political reasons for migrating, whereas the Iraqi immigrants come from a very suppressive society (score 14) and presumably many of them are political refugees. In addition, the Gini ratio for British immigrants in Australia is 1.03, but for immigrants from the unequally distributed Iraq who moved to the more equally distributed Sweden, it is 1.52. Furthermore, the British immigrants in Australia have traveled over a far greater distance (16,970 kilometers) than the Iraqis in Sweden (3,480 kilometers). Finally, the British community in Australia was exposed to the official language before migration, whereas the Iraqi immigrants in Sweden were not. All in all, this results in employment rates of 95% among the favorably selected group of British immigrants in Australia and 56% among the unfavorably selected group of Iraqi immigrants in Sweden. This difference of 39 percentage points comes very close to what one would expect from imputing the values for both groups in Model 1 (i.e., 30%).

13

The differences are replicated when contrasted with the unemployment level of natives. The unemployment level of favorably selected communities is 6% above that of the native population in the destination country; the unfavorably selected communities are 25% more often unemployed.

Table 4.7 Illustration of composition and context effects on unemployment among male immigrants Composition Context Favorable Unfavorable Favorable Unfavorable Political Low High Left-wing parties High Low suppression Gini ratio Low High Mean educational High Low level group Geographic High Low distance Official language Yes No Groups (N) 27 97 171 232 Mean % employed 93% 74% 90% 86% Examples US in the UK Albanians in Germans in Mexicans in (% employed) (97%) Austria (81%) Sweden the US (92%) (96%) UK in Poles in Germans in Vietnamese in Australia Belgium France Canada (86%) (95%) (75%) (91%) Indians in the Iraqi in Italians in Pakistani in UK (89%) Sweden (56%) Greece the UK (78%) (85%) Note: Average scores used to distinguish between “low” (i.e., below average) and “high” (i.e., above average). Source: Van Tubergen et al. (2004).

Employment

99

CONCLUSIONS AND DISCUSSION This chapter showed that immigrants’ labor force participation and unemployment depend on the country of origin, the country of destination, and the immigrant community. These macro differences are partly the result of skill selection: immigrants with more human capital have more economic opportunities, and the selection of such skills is related to macro characteristics (e.g., political suppression, income inequality). Part of their role goes unnoticed when looking only at education and language proficiency. Even when such observable skills are taken into account, unobserved skills (talents, motivation) are important as well. Next to the selection of individual skills, macro differences reflect pure contextual effects. Immigrants are discriminated in the labor market, and the degree to which they are discriminated is related to perceived ethnic threat, policy measures towards reducing discrimination, and within group norms on gender roles. After taking observed and unobserved human capital into account this chapter shows that immigrants from non-Christian origin countries less often participate in the labor market, and when they do, they are more often unemployed than immigrants from Christian countries. Apparently, these differences across religions mean that natives feel more threatened by culturally distinct groups. Policy measures towards reducing discrimination seem effective. In societies with a strong presence of left-wing parties in the government, that are assumed to facilitate incorporation of immigrants into the labor market, immigrants are indeed more active in the labor market, and when they are active, they are more likely to be employed. In addition, this chapter shows that the cultural patterns for the labor-force activity of females in the origin country are carried over to the destination country. Exclusion of women from the labor market in the country of origin therefore plays an important role in the labor-force participation of female immigrants in their country of destination. Social capital is important for immigrants’ economic incorporation. I find that the relative size of the immigrant group is positively related to immigrants’ labor-force participation. Immigrants profit from the ethnic capital available in their community, which increases with group size. Furthermore, immigrants from communities with higher aggregate levels of education have better chances for employment. These

100

Immigrant Integration

communities are also less often discriminated against than communities of lower status. Reviewing the findings of this chapter shows that only one hypothesis on skill selection was confirmed for female labor-force participation. This suggests that selection effects are not as important for female labor-force participation as they are for male labor-force participation and male and female unemployment. However, it should be emphasized that the labor-force participation of women is a particularly complex phenomenon, which also depends on a number of factors not considered here, such as the presence of young children, marital status, and husband’s economic success. Because the findings of this chapter on women’s labor-force participation are probably sensitive to the many variables not included in the analysis, the failure to find support for a number of hypotheses concerning this particular outcome does not call into question the more general hypotheses on skill selection.

CHAPTER 5

Occupational Status

Participating in the labor market and having a job are only the first steps towards full incorporation in the labor market. A logical follow up question, which is the focus of this chapter, is to study the status of the jobs that immigrants occupy. It is well known that immigrants are less active in the labor market and have higher unemployment rates than natives, but it is also observed that immigrants occupy lower quality jobs. It is tempting to examine whether the country of origin, the country of destination and the immigrant community equally affect immigrants’ occupational status. This chapter again relies on theories of human capital, prejudice and social capital, but also adds some new hypotheses. One theoretical extension that I make in this chapter is to look at the influence of racial prejudice. According to prejudice theory, not only religious characteristics of the country of origin are important, racial distinctions can be influential too. Research has shown that natives’ social distance towards ethnic groups not only overlaps with a distinction in religion but also race, ranking nonwhites, Islamic, Buddhist, and other non-Christian groups at the top of the social distance scale (Owen, Eisner, and McFaul 1981; Pineo 1977). Because the Western countries examined in this study are predominantly white, I expect that non-white groups will be more discriminated against in the labor market, leading to lower status jobs than white groups. Another extension I make in this chapter relates to the role of social capital. In Chapter 4, I examined the role of social capital by looking at the size and average education of the immigrant community. In this chapter, I also look at the percentage of the population that is 101

102

Immigrant Integration

employed. When the immigrant community consists of a large group that is unemployed, there are fewer people who can help co-ethnics in finding higher quality jobs. Migrants who belong to economically powerful communities could benefit more from their economic potential than migrants from disadvantaged communities. The dependent variable is the status of the current job, as measured in terms of the International Socio-Economic Index (ISEI) (Ganzeboom, De Graaf and Treiman 1992). To obtain ISEI scores for the occupations, I used tools that convert the ISCO-88 classification into ISEI (Ganzeboom and Treiman 1996, 2003). The EULFS provided three digit ISCO-88 scores. The original codings of occupations in the US census and the Australian survey were translated into four digit ISCO-88 using conversion tools (Ganzeboom 2003; Lampert 2003). For the 14 occupational categories provided in the Census of Canada, I relied on the weighted ISEI scores provided by Model and Lin (2002).14 DESCRIPTIVE RESULTS Table 5.1 (males) and Table 5.2 (females) present the mean occupational status (in ISEI) of immigrants by origin, destination, and community. I illustrate how seven origin groups fare in the 17 destination countries. Table 5.1 shows that the mean occupational status (in ISEI scores) of all immigrants, averaged over the 17 destination countries, is 43.1. Like labor force participation and unemployment, there are strong differences across origin groups in their occupational status. Immigrants from the United States have an average ISEI of 53.2, which is ten points above the mean ISEI of immigrants from all countries, and almost 20 points higher than the average ISEI of immigrants from Morocco (35.0) and Turkey (35.2). There are also pronounced differences in immigrants’ occupational standing between destination countries. I find a low occupational status among immigrants in Austria (average ISEI is 37.8), France (37.2), Greece (37.8), and Germany (37.8), whereas the occupational status is considerably higher in the United Kingdom (50.1).

14

See the Appendix in Wanner (1998) for the specific ISEI scores assigned to the categories.

Table 5.1 Occupational status (ISEI) of male immigrants Origin Destination China Germany Italy Morocco Poland Australia . 46.9 a 44.0 a . . a Austria 40.8 51.9 48.1 a . 39.1 Belgium 52.7 a 49.0 37.2 35.9 44.5 a Canada 47.2 45.7 41.3 . 41.2 Denmark . 48.8 a . . 45.3 a Finland . 59.3 a . . . a France 46.5 51.4 39.1 34.0 39.6 a a Germany . n.a. 34.8 29.7 40.5 Greece . 41.1 49.4 a . 32.8 Ireland . 51.9 48.1 a . . a Luxembourg 39.3 52.9 42.0 53.2 a 46.4 a a Netherlands 46.9 49.1 41.6 36.8 42.0 a a Portugal . 44.4 . . . Spain 40.5 a 50.3 43.8 a 28.2 35.2 a a a Sweden 66.8 49.8 . 33.9 50.2 a a UK 50.6 50.5 45.5 38.4 50.6 a a USA 47.8 48.1 39.6 48.7 41.4 Mean 47.4 49.1 40.3 35.0 40.5 a . = N < 25 ; = 25 < N < 200. Source: Van Tubergen (2006a).

Turkey

USA

. 32.9 34.9 . 35.9 a . 31.8 32.0 43.0 . 45.1 36.4 . . 37.3 a 44.8 a 49.4 35.2

. 51.2 60.5 50.7 53.3 . 62.7 46.9 51.8 50.9 59.7 53.0 . 58.2 53.6 56.2 n.a. 53.2

Mean all groups 40.1 37.8 42.6 43.6 44.9 46.9 37.2 37.8 37.8 46.1 42.7 44.8 43.3 39.5 41.6 50.1 45.1 43.1

Mean natives 47.3 43.1 45.0 43.1 45.0 44.3 42.9 43.8 41.5 42.1 44.7 49.0 39.8 40.5 44.7 46.9 44.3 44.2

Table 5.2 Occupational status (ISEI) of female immigrants Origin Destination China Germany Italy Morocco Poland Australia . . . . . Austria 35.9 a 46.7 54.0 a . 38.9 Belgium 46.9 a 48.7 37.6 33.9 41.3 a Canada 42.7 47.6 42.6 . 42.7 Denmark . 44.4 a . . 41.1 a Finland . . . . . France 40.7 a 50.1 a 41.9 a 28.9 37.5 a Germany . n.a. 32.7 a 34.8 41.7 Greece . 47.8 56.7 a . 31.2 Ireland . 50.9 51.0 a . . a Luxembourg 38.9 49.8 39.4 . 40.9 a Netherlands 42.1 a 47.2 47.2 a . 40.2 a a Portugal . 55.2 . . . Spain 35.0 a 49.4 a 53.7 a 23.2 36.7 a a a Sweden 50.6 47.3 . . 42.4 UK 49.5 48.6 45.2 . 46.7 a a USA 44.2 46.9 42.4 49.5 41.2 Mean 43.7 47.9 41.7 32.6 40.5 a . = N < 25 ; = 25 < N < 200. Source: Van Tubergen (2006a).

Turkey

USA

. 27.2 32.5 . 28.0 a . 33.2 a 28.9 45.1 . . 34.8 . . 36.1 a 47.6 a 50.9 a 32.4

. 53.0 a 57.0 a 52.4 52.8 a . 60.1 a 51.0 a 54.5 a 53.6 55.8 a 54.9 a . . 54.4 a 55.5 n.a. 54.1

Mean all groups 42.3 34.2 44.0 44.7 41.3 45.7 34.1 37.8 37.4 48.7 38.7 44.1 46.6 39.7 41.5 48.3 43.7 42.3

Mean natives 47.4 42.7 45.6 47.7 42.2 44.9 42.4 43.7 42.7 47.3 45.9 46.8 38.7 42.0 43.7 46.1 47.1 45.6

Occupational Status

105

Table 5.1 also provides some clues for the impact of immigrant communities. Compare, for instance, the occupational status of Turks in Germany and the United States. In Germany, Turks have an average ISEI of 32.0, which is below the mean ISEI of Turks in general (35.2) and below the mean of all immigrants in Germany (37.8). In contrast, in the United States, Turks have a much higher occupational status (49.4), which is far above their general rate and also above the overall pattern observed among immigrants in the United States (45.1). Apparently, then, the specific situation of Turks in Germany and the United States determines their deviance from the pattern expected from general origin and destination effects. MULTIVARIATE ANALYSES I constructed three cross-classified multilevel regression models of immigrants’ occupational status to test the hypotheses. Model 1 includes the macro composition factors plus a macro-control variable for the occupational status of natives. In Model 2, individual controls are included. Model 3 adds the contextual factors. The results are presented separately for males and females (Table 5.3). Note that in these models, I use destinations per year as the destination level.15 To examine the results for the hypotheses on composition effects, we have to look at Models 1 and 2. Because Model 1 contains only macro-composition factors (and a macro control), it is logically to examine the results of this model for the discussion of composition effects. Adding human capital factors in Model 2 normally reduces the strength of composition effects. However, by no means should the macro effects in the present study disappear. This is because several observable skills were not measured (e.g., labor-force experience) or only partly measured (e.g., length of residence) and unobserved skills (i.e., talents, productivity, ambitions) were, of course, completely omitted. Moreover, some macro composition factors may be positively 15

In additional analyses (not shown) I examined how sensitive the findings were to different multilevel designs, i.e., treating destinations as time-constant instead of looking at destinations by year. The overall conclusion is that of strong similarity between the original and the additional analyses, suggesting that the findings are robust for differences in multilevel designs.

Table 5.3 Cross-classified multilevel regression of socioeconomic status (ISEI) in 17 Western countries, 1980–2002, immigrants between 25 and 54 years old Males Females Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Constant 14.518 9.008 -6.900 20.051 17.233 6.392 Origin Political suppression Predominantly Christian origin White (%) Destination Point system Mean occupational status, native reference group Social-democratic parties in government

-.222

-.259**

-.233** -.929 .021**

-.281*

-.187*

-.129 -.230 .021**

-4.137** .753**

-5.965** .617**

-6.621** .585**

-.765 .610**

-2.832** .436**

-4.359* .379**

Community Gini origin/Gini destination -3.644** GDP origin/GDP destination .942** Geographic distance (per 1000 km) .521** Official language .085

.117

-3.610** .659** .391** .341*

-2.094** .420 .289** .699**

-.101

-5.181** 3.381** .426** .726**

-3.703** 2.363** .357** .950**

-2.046* 1.968** .240** 1.433**

Relative group size (%) Mean educational level group Employment rate group (%) Individual Age Duration of stay 0-5 years 6-10 years 10+ years Education Low Middle High Married

-.284** 6.366** .027**

-.687** 6.422** -.010

.086**

.078**

-.042**

-.050**

Ref. -.207 .602**

Ref. -.097 .930**

Ref. .615** 2.262**

Ref. .731** 2.613**

Ref. 4.973** 21.501** .374**

Ref. 4.621** 20.756** .450**

Ref. 7.634** 20.893** -.618**

Ref. 7.157** 20.098** -.569**

Number of observations Destination year 126 126 126 Origin 181 181 181 Community 859 859 859 Individual 138,472 138,472 138,472 * p < .05; ** p < .01 (two-tailed tests). Source: Van Tubergen (2006a).

126 170 809 100,697

126 170 809 100,697

126 170 809 100,697

108

Immigrant Integration

(or negatively) selective, but their effects may be suppressed due to associations with other selection mechanisms that are negatively (or positively) selective. Taking individual-level factors into account could therefore show an increase in some macro-level effects. Human Capital: Composition Effects The results show that the role of immigrants’ country of origin and community are partly explained by skill selection. With respect to immigrants’ origins, we see that people from more politically suppressive societies, who are less well prepared for participating in the labor market, have a lower occupational standing. The results are significant for males in Model 2, and for females in Models 1 and 2. With regard to the immigrant community, a number of characteristics affect skill selection. It appears that immigrants who migrated from countries with a more unequal income distribution to countries with less inequality have a lower occupational position than immigrants who moved in the opposite direction. The effect of the ratio of the Gini score of the country of origin to the Gini score of the country of destination is significantly negative in Model 1. It becomes less strong after individual controls are included, but remains significant. These findings indicate that the less talented and less productive move from unequal to more egalitarian societies, whereas the opposite is true for the more talented and productive immigrants. I also find that the less economically advanced the country of origin relative to the country of destination, the lower the occupational status of immigrants. People who moved from less economically advanced nations to more advanced nations have a lower position in the labor market than immigrants who moved from relatively rich countries. For both males and females, Model 1 shows a significantly positive effect of the GDP ratio on occupational status. In Model 2, the GDP ratio is still significant, but the coefficient has become somewhat smaller. This suggests that part of the relationship is due to favorable selection of educational diplomas, whereas the remaining effect underscores the idea that immigrants’ human capital obtained in advanced economies is valued more in the receiving countries than skills obtained in more developing nations.

Occupational Status

109

Immigrant communities are also important, because they are selective in terms of the travel distance between origin and destination. The greater the distance between the country of origin and the country of destination, the higher the occupational status of immigrants. Model 1 shows that the distance between the capital cities of the origin and destination countries has a significantly positive impact on the occupational status for males and females. The effect diminishes once individual variables are taken into account, but remains significant. Travel distance favorably selects immigrants and also increases the benefits of human capital investments after migration –people who traveled over longer distances are less likely to return. A final selection mechanism that plays a role at the community level is the exposure to the language. Model 1 indeed shows a significantly higher score for this group among females, but not for males. However, in Model 2, after taking individual factors into account, the results are significant for males. The effect in Model 2 is stronger for females as well, and inspection of Model 3 reveals very significant effects of language exposure for both males and females. This suggests that immigrants who moved to a destination with the same official language tend to be negatively selected on other skill factors –suppressing the positive skill selection in terms of language proficiency. All in all, the analysis confirms the idea that language exposure improves the occupational standing of immigrants. The results reject an influential idea in the literature on the importance of host countires in the selection of the immigrant population. It is hypothesized in the literature that immigrants in countries having a point system have a better economic position (Borjas 1988), which includes having a higher occupational status. The analysis shows that the opposite is true: immigrants in countries having a point system have a lower occupational status than immigrants in countries without such immigration policy. Social Capital and Discrimination: Context Effects Differences in occupational status across origin groups, host countries and immigrant communities are not only the result of skill selection. Discrimination and social capital are important as well. Some origin groups have a higher occupational status than other groups because

110

Immigrant Integration

they are predominantly white. The percent white of the country of origin has a positive impact on the occupational status. Non-white groups meet stronger discrimination in the labor market than white groups in the predominantly white receiving nations, which results in a lower occupational status. I do not find an effect of religious origin, however. Immigrants from non-Christian origin do not have the expected lower occupational status than immigrants from mainly Christian societies. Apparently, then, social distance and the resulting discrimination is stronger on racial than on religious grounds. Ethnic prejudice and discrimination are important for explaining the role of the immigrant community. I find that the larger the size of the immigrant group relative to the total population in a country, the lower the occupational status of the immigrants of that group. Larger groups are perceived by the native majority as more economically powerful and threatening, and are more discriminated in the labor market. It opposes the alternative suggestion that immigrants of larger groups could profit from the ethnic capital available in their own group. One interpretation of this anomaly is to question the idea of ethnic solidarity and the supposed beneficial outcomes of belonging to a large immigrant group (Li 1977; Sanders and Nee 1987). However, not finding a positive effect of group size on occupational status is not a strong case against the ethnic capital idea. Group size only indicates how many co-ethnics are available for help, but does not measure the resources available in the community, which are crucial for providing positive social support. I find that the higher the education of the immigrant community, the higher the occupational status of the immigrants of these groups. It is important to realize that this is a true contextual effect, because the education of immigrants at the individual level is controlled. In addition, the results for the male sample show a positive effect of the percent employed of the community on immigrants’ occupational status. These findings support the idea that immigrants in more resourceful communities are more succesfull in the labor market. Immigrants in such communities are less discriminated and they also have more social capital. I find no support for a contextual effect of receiving nations: the idea that the presence of social-democratic parties in the government has a beneficial impact on the occupational status of immigrants. The

Occupational Status

111

analyses for the male and female samples show that the presence of social-democratic parties in the government in the 10 years preceding the survey year had no significant effect on immigrants’ occupational standing. Explained Variance Analysis of random intercept models without explanatory variables (“empty model”) shows that the variance at the individual level for males is 83% of the total variance observed, the variance between origins is 14.5% and between destinations only 2.4% (figures not presented). Comparing the empty model with the full model for males (i.e., Table 5.3, Model 3) shows that 36% of the total variance is explained. More specifically, the final model explained 6% of the variance among destinations, 28% of the variance among individuals, and 84% of the variance among origins. The results for the female sample are very similar. Apparently, then, from a macro perspective, I was better able to explain the variance between origin countries than between destination countries. CONCLUSIONS In this chapter, I studied macro effects on the occupational status of immigrants. The analysis shows that the country of origin, the country of destination, and the ethnic community all play a role. More precisely, about 17% of the total variance of immigrants’ occupational status was observed at the macro level. It appears that the occupational standing of immigrants varies quite strongly among origin groups. I find little variation among receiving nations, which suggests that the occupational status of immigrants is quite similar across nations. Macro differences in immigrants’ occupational status can be interpreted in terms of the selection of human capital, discrimination, and social capital. Comparing the results of this chapter with the findings of the previous chapter on labor market participation and unemployment yields some more general interesting insights into immigrants’ economic incorporation.

112

Immigrant Integration

To begin with, the findings suggest that the occupational status of immigrants is more strongly affected by human capital factors than immigrants’ labor force participation and employment. Chapter 4 found no support for the expected positive effect of language exposure, no confirmation for the influence of geographic distance, and no support for the hypothesized effect of economic development. Findings on occupational status are more in line with human capital theory. As hypothesized, I find that immigrants’ occupational status is higher when immigrants traveled over a longer distance, when they were exposed to the host language before migration, and when they come from more developed nations. Apparently, other forces than skill selection are operating with respect to labor force activity and employment. Looking at characteristics of receiving nations, my study has not supported the idea that immigrants perform better in the labor market in countries that use a pointsystem. With regard to immigrants’ laborforce activity and employment, Chapter 4 finds no significant difference between countries with and without such a selective immigration policy, whereas this chapter shows that immigrants’ occupational status was even significantly lower in countries with a point system. Thus, my study suggests that although Australia and Canada adopted policies to select the more favorable immigrants to their economy, these policies seem to have failed. Several explanations for this unexpected finding could be advanced. One interpretation is that immigrants in Australia and Canada are not incorporated well economically, despite the selective immigration policy. That is to say, it could be that other characteristics of these countries are unfavorable to the situation of immigrants, but these were not considered in the present study. For example, illegal migration may be less widespread in Australia and Canada, assuming that illegal immigrants have higher participation rates, lower unemployment figures and higher quality jobs than documented immigrants. Unfortunately, information on legal status is not available in my data set, and aggregate estimates are available for only a few countries. Nevertheless, illegal entry in any country probably is more likely to happen via land than via sea or air. The results concerning the point system may therefore reflect differential populations of undocumented migrants.

Occupational Status

113

On the other hand, this explanation is less plausible because undocumented immigrants are probably less well incorporated economically. In addition undocumented migrants belong to specific groups. For instance, the Immigration and Naturalization Service of the United States (1996) reports that of the estimated 5 million illegal immigrants in 1996, 2.7 million (54%) are from Mexico. This would imply that the estimate of the economic integration of the members of this community may be biased (but in an unknown direction and to an unknown extent), but need not affect the estimate of the destination effect, which is computed for all groups in a country. Another possibility is that countries other than Australia and Canada have adopted similar selective policies. For example, in the 1940s, the United States signed labor contracts with the Mexican government to import agricultural workers (Galarza 1964). Similar labor contracts were signed by Western European countries with Mediterranean countries in the 1960s and 1970s (Fassman and Münz 1994). As a consequence, many migrants from these countries were employed immediately after their migration. However, these policies were only applied to specific origin groups (e.g., Mexicans), a specific sex (males), and for specific periods (e.g., the 1960s), and are therefore unlikely to be as selective as those of countries with a point system. Another explanation for not finding any effect is that the point system “works not by attracting more skilled immigrants from each source country, but by changing the national origin mix of the immigrant population” (Borjas 1999:59). If this is the case, one would not find an effect from immigration policy because, in the present study, the impact of destinations was examined independently of the influence from the country of origin. Australia and Canada attract immigrants from more developed nations, but immigrants from those countries do not perform better than comparable immigrants in other countries. Although this seems a plausible explanation, it does not tell us why the occupational status of immigrants in Australia and Canada is lower. Perhaps the reason why immigrants have a lower occupational status is that the pointsystem partly selects immigrants that are attractive to the economy in that particular period. In other words, immigrants are recruited to fulfill short term labor market shortages in certain segments of the economy. Because of economic fluctuations and shifts in economic activities, immigrants in Australia and Canada might be less

114

Immigrant Integration

well prepared to anticipate such changes. Thus, short term policies selecting particular immigrants might have unintended consequence at the longer term. Comparing the results of this chapter with those of Chapter 4 also provides insight into the importance of integration policies. This chapter finds no support for the idea that the presence of socialdemocratic parties in the government has a beneficial impact on the occupational status of immigrants. In Chapter 4, I found such a positive effect for immigrants’ labor-force participation and employment. These combined findings suggest that an active state policy towards integrating immigrants succeeds in increasing levels of participation and employment, but not in their socioeconomic standing. This may not seem all too surprising, since government policies are more often directed towards changing levels of social welfare, schooling, and positive labor-market discrimination than on the level of immigrants’ occupation once they are active in the labor market. From the perspective of the country of origin, two factors, in particular, deserve special attention. The first is the religious’ makeup of the home country, measured in terms of being either predominantly Christian or otherwise. Immigrants from Christian countries have an economic advantage over immigrants from other countries: they are more active in the labor market and they are less often unemployed. The reason that immigrants from Christian countries do not occupy the better jobs, is possibly the result of unobserved selection: only the very talented non-Christian immigrants have a job and those immigrants are compared with the (less selected) Christian group. In short, I find strong evidence for the idea that immigrants from Christian nations meet less discrimination in the labor market than those from nonChristian countries. A second important characteristic is the political situation in the home nation. Immigrants who come from countries with few civil rights and political liberties are more inclined to migrate for noneconomic reasons than those from less suppressive countries. Because political migrants or refugees are less prepared to participate in the economy in the host society, they are negatively selected. As a consequence, they are less active in the labor market, are more often unemployed, and have lower quality jobs. In summary, the political conditions under which immigrants migrate play an important role in

Occupational Status

115

the economic integration of immigrants, irrespective of the country they move to. From the perspective of the immigrant community, I examined a number of factors, but the size of the immigrant community deserves special attention. My study shows that immigrants from larger communities are more active in the labor market, but this by no means implies that they are well economically incorporated. Community size has no influence on the employment chances of immigrants, and even a negative impact on the quality of jobs that immigrants occupy. The divergent outcomes of group size possibly reflect differential processes that are at work simultaneously. On the one hand, immigrants belonging to large ethnic communities could benefit from job options within their own community. Presumably, ethnic communities are indeed able to generate employment opportunities for their own members, resulting in higher labor force participation. On the other hand, larger groups are perceived as more threatening to the native population, leading to stronger discrimination outside the ethnic community. As a consequence, the higher quality jobs in the normal labor market are more difficult to access for immigrants belonging to larger communities.

This page intentionally left blank

CHAPTER 6

Self-Employment

In the sociological literature, self-employment is considered an important avenue for immigrants’ economic mobility (Raijman and Tienda 1999b; Sanders and Nee 1996). Among white males in the United States in 1980 16.5% of the foreign-born and 11.7% native-born were self-employed (Borjas 1986). It is generally believed that rates of self-employment are higher among the foreign-born than among natives (Borjas 1986). Recent figures in several Western societies suggest an increase in self-employment among immigrants (Waldinger, Aldrich and Ward 1990). An important issue in the literature on the self-employment of immigrants is the role of the group. Earlier studies have shown that self-employment rates between immigrant groups differ considerably. In 1990, the rate of self-employment for males in the United States, by ethnic group, ranged from 3.2% for Laotians and 6.8% for Mexicans to 27.9% for Koreans and 28.6% for Israelis (Fairlie and Meyer 1996). The self-employment experience among Asian immigrant groups, especially Koreans, in the United States has gained much scholarly attention (Kim and Hurh 1985; Kim, Hurh and Fernandez 1989; Light and Bonacich 1988; Yoon 1991). Research has shown that differences between groups remain even after human capital and demographic variables at the individual level have been taken into account (Borjas 1986). In this chapter, I pursue this contextual approach to immigrants’ self-employment. To begin with, I take the contextual perspective in a new direction by asking to what extent immigrants’ self-employment differs between receiving nations. Cross-national empirical studies on 117

Table 6.1 Classification of self-employment Survey Variable [Label] Self-employed EULFS US 1980 Census

Stapro [Professional status] Class [Class of worker]

-self-employed with employees -self-employed without employees -self-employed workers -employee of own corporation

US 1990 Census

Class [Class of worker]

-self-employed in own not incorporated business, professional practice, or farm -self-employed in own incorporated business, professional practice or farm

Wage/salary workers, family workers, etc. -employee -family worker -private wage and salary workers -federal government workers -state government workers -unpaid family workers -employee of a private for profit company or business or of an individual, for wages, salary, commissions -employee of a private not-forprofit, tax-exempt, or charitable organization -local government employee -state government employee -federal government employee -working without pay in family business or farm

Canada 1991 Census

Canada 1996 Census

COWP [Class of worker]

-paid workers (self-employed incorporated) -self-employed without paid help unincorporated -self-employed with paid help unincorporated COWP [Class of -paid workers (self-employed worker] incorporated without paid help) -paid workers (self-employed incorporated with paid help) -self-employed without paid help unincorporated -self-employed with paid help unincorporated STC01 -self-employed [Occupational status] -employer*

-paid workers (wage and salary earners) and unpaid family workers

-paid workers (wage and salary earners) and unpaid family workers

Australia -in the labor force: employed1981 wage, salary earner Census -helper, unpaid * = a person who operates his or her own unincorporated economic enterprise or engages independently in a profession or trade. Source: Van Tubergen (2005).

120

Immigrant Integration

self-employment among immigrants are rare. The studies that have been conducted have consisted of separate country reports on immigrants’ self-employment, which have subsequently been combined and compared (Kloosterman and Rath 2003; Rath 2002; Waldinger, Aldrich and Ward 1990). In this chapter, I combine the perspective on the role of the immigrant group with that of the role of the receiving nation. To what extent do self-employment rates vary across immigrant groups? Do these group differences travel across nations? What is the influence of the immigrant community? Are self-employment rates in some countries higher than in other nations? Before examining the descriptive and multivariate outcomes, it is important to see how self-employment was measured in the different data sets. Self-employed persons include those with or without employees, in their own business, either incorporated or unincorporated. They are contrasted with wage/salary employees and (unpaid) family workers. Table 6.1 provides the original coding and description of employment status from these surveys. Included in the International File of Immigration Surveys are the European Union Labour Force Survey, and census data from the United States, Canada, and Australia. Apparently, there are no problems involving a lack of comparability between the surveys used in this study. DESCRIPTIVE RESULTS To give a descriptive account of the role of contextual effects, Table 6.2 presents the percent of self-employed immigrants by origin, destination, and community. The results suggest that origins, destinations, and communities all play a role in the self-employment decisions of immigrants. Consider, first, the differences observed between the countries of origin. The mean self-employment rate of all immigrants, averaged over the 17 destination countries, is 14.9. The percent selfemployed among the Chinese is 24.5, almost 10 percentage points higher. Filipinos, also from the Asian continent, have a much lower rate (4.8). The self-employment figures of the three other groups fall in between (Italians, 19.4%; Poles, 11.4%; Turks, 10.8%). There are also pronounced differences in immigrants’ selfemployment between destination countries. Of all immigrants observed in the data set in Austria, about 7.6% were self-employed.

Table 6.2 Self-employment (%) of male immigrants of selected origin groups by destination country Origin Destination China Italy Philippines Poland Turkey Other All Natives Australia 30.1 26.8 2.3 12.6 8.2 14.5 14.9 19.3 Austria . 14.4 . 7.8 4.0 8.3 7.6 14.0 Belgium . 16.1 . . . 18.1 17.7 17.0 Canada 20.5 20.4 5.7 15.4 14.2 16.6 16.3 15.2 Denmark . . . . . 11.6 11.6 12.2 Finland . . . . . 11.9 13.9 21.1 France . 21.0 . . 14.4 12.3 12.9 14.6 Germany . 15.5 . 7.9 5.2 9.0 8.7 10.9 Greece . . . 8.8 35.1 16.8 17.8 39.2 Ireland . 42.0 . . . 26.5 26.9 27.9 Luxembourg . 12.1 . . . 7.4 8.2 11.5 Netherlands . 23.8 . . 8.7 11.2 11.3 13.2 Portugal . . . . . 23.7 25.2 23.9 Spain . . . . . 24.5 25.6 22.6 Sweden . . . . . 14.6 15.8 18.8 UK 27.0 27.1 . . . 20.6 20.9 16.8 United States 16.1 18.7 5.5 12.4 18.4 13.7 13.6 12.5 Mean 24.5 19.4 4.8 11.4 10.8 15.0 14.9 17.0 Note: . = N < 200. Source: Van Tubergen (2005).

122

Immigrant Integration

This chapter also finds low self-employment rates among immigrants in Germany (8.7%) and Luxembourg (8.2%), whereas self-employment rates are considerably higher in Ireland (26.9%), Spain (25.6%), and Portugal (25.2%). It is important to note that in most countries, immigrants’ self-employment rates are lower than those of natives in the same age category. In only six (Belgium, Canada, Portugal, Spain, the United Kingdom, and the United States) of the 17 countries in the data set are the self-employment figures of immigrants higher than those of natives. This finding is in conflict with the literature, which suggests that immigrants generally have higher self-employment rates than natives (e.g., Borjas 1986). I also find that the community is important. In Austria, Turks have a self-employment rate of 4%, which is below the mean self-employment rate of Turkish immigrants in general (10.8%) and also below the mean of all immigrants in Austria (7.6%). In contrast, in Greece, Turks have a very high self-employment rate (35.1%), which is far above their general rate and also above the overall pattern observed among immigrants in Greece (17.8%). Apparently, then, the specific situation of Turks in Austria and Greece determines their deviance from the pattern expected from general origin and destination effects. VARIANCE COMPONENTS How strong do self-employment rates differ across origins and destinations? Table 6.3 shows the variance components of multilevel logistic regression models with random intercepts for the country of origin and the country of destination. In one model, destinations are treated as time invariant, in the alternative model they vary over time. The table shows that in the models in which destinations are treated as time constant the variance is .377 among origins and .227 among destinations. Because there are 179 origin countries in the analysis, and only 17 destinations, which are, in addition, also quite homogenous, the large variation between destinations is quite remarkable. Analysis of time variant models shows no change with respect to the origin component, but the destination component differs somewhat. It appears that the variance remains about the same (it changes from .227 to .234), but the standard error reduces considerably (from .107 to .033).

Table 6.3

Variance components from cross-classified multilevel logistic regression models with random intercepts for country of origin and country of destination; male immigrants, 25-54 years old Destinations time invariant Destinations time variant Variance component S.E. Variance component S.E. Country of origin 0.377 (0.058) 0.376 (0.058) Country of 0.227 (0.107) 0.234 (0.033) destination Individual 3.290 3.290 Total 3.894 3.900 Source: Van Tubergen (2005).

124

Immigrant Integration

THEORIES ON IMMIGRANTS’ SELF-EMPLOYMENT Due to the complex nature of entrepreneurship, a large number of theories on self-employment have been suggested, such as the notions on middleman minorities (Bonacich 1973), blocked mobility (Light 1979), protected markets and ethnic enclaves (Aldrich et al. 1985; Portes and Bach 1985), mixed embeddedness (Kloosterman and Rath 2001), ethnic strategies and interaction theory (Aldrich and Waldinger 1990; Light and Rosenstein 1995), stepladders (Raijman and Tienda 1999a), and social capital and social networks (Flap, Kumcu and Bulder 2000; Renzulli, Aldrich and Moody 2000). Although these notions highlight different aspects and causes of immigrants’ entrepreneurship, none of them provide a comprehensive theoretical perspective that covers the effects of the country of origin, the country of destination, and the immigrant community. In this chapter, I rely on the three general theories used in the previous chapters. These are: human capital theory, social capital theory, and prejudice theory. All three theories assume, although for different reasons, that immigrants “normally” become a paid worker. Because of certain individual characteristics or societal circumstances, however, immigrants use self-employment as an alternative route for economic mobility.16 According to human capital theory, immigrants make such a career change because they are less productive when working as a paid employee in the normal labor market. The main reason for being less productive is that immigrants lack host-country specific skills. Many immigrants do not speak the host language well upon arrival, and for that reason have fewer opportunities of finding a job. Furthermore, the educational qualifications obtained in the country of origin are less valued, particularly for immigrants originating from poorer nations. When immigrants do not speak the language well and do not have the 16

It has become standard practice in the literature on the self-employment of immigrants to study persons who are employed, and not to compare selfemployment with no employment. It is important to emphasize that it is also assumed in my theoretical approach adopted here that no employment is not a third option for immigrants. In addition, I do not distinguish factors that determine the decision to become entrepreneur vis-à-vis factors of maintaining a business.

Self-Employment

125

appropriate qualifications, it might be beneficial to them to start their own business. Furthermore, some immigrants are familiar with being self-employed, and for that reason, starting a business in the host country can be especially attractive. Prejudice theory agrees that immigrants use self-employment as an alternative route to economic mobility. However, prejudice theory does not assume that immigrants make this decision because of a lack of skills. Instead, the theory argues that immigrants are discriminated in the normal labor market, pushing them into self-employment (Light 1979). Even when immigrants speak the host language well and when they do have the appropriate educational qualifications, immigrants are discriminated on racial, ethnic, and religious grounds. The theory assumes that the level of discrimination immigrants experience in the labor market directly affects self-employment. Social capital theory argues that the role of personal ties with coethnics and the resources they have are central for understanding selfemployment (Flap, Kumcu and Bulder 2000; Portes and Bach 1985). Similar to other theories, it assumes that immigrants normally do paid labor, but that depending on their social network they might become self-employed instead. Immigrants in communities with much financial and human capital can more easily start a business than immigrants belonging to less resourceful communities. In more resourceful groups, co-ethnics provide information on how to begin and maintain a business, and they can help financially. In addition, immigrant communities with much social capital also have a better consumer market for the goods and services provided by the ethnic entrepreneur. This chapter uses these three theories, supplemented with other notions, to explain the effects of origins, destinations, and communities. To test these theories, I constructed two cross-classified multilevel logistic-regression models of self-employment. The dependent variable are the log-odds of an immigrant being self-employed (score 1), versus a wage/salary employee or (unpaid) family worker (0). The results are presented in Table 6.4.17 Model 1 presents the findings when 17

I did not inspect changes in the effects of macro-level variables and changes in variance components after adding micro- and macro-level variables, nor did I inspect deviance statistics. This is because, in multilevel models with dichotomous outcomes, the residual level-one variance is fixed and the coefficients of the macro-level variables as well as the variance at the macro

126

Immigrant Integration

destinations are treated as time-invariant (resulting in 17 destination cases). In this model, I treated all variables that pertain to the host society as time-invariant and included the average country scores for the two time-variant variables. Model 2 gives the results for the timevarying analyses (estimated to examine the time-variant destination variables). This results in 126 “destination-year” cases. Whereas the first model is more suited to testing time-invariant destination variables, the second is a better test of time-variant destination variables. Origin Effects How can we explain differences between origin groups that “travel” across destinations? One possible factor is the self-employment rate in the country of origin (Fairlie and Meyer 1996; Yuengert 1995). This idea is informed by more general notions on the role of immigrants’ cultural attitudes, values, and beliefs in their economic incorporation. Sowell (1996), in particular, maintains that immigrants of culturally similar background have the same economic position across multiple destinations. One aspect of immigrants’ home country culture that is transmitted to the country of destination is the practice of selfemployment. Sowell (1996:112-13) illustrates this point: “Japan had a long tradition of entrepreneurship (…). This economic initiative also became apparent in the histories of Japanese emigrants who settled overseas.” When immigrants come from a country with a relatively large self-employment sector, they are socialized into practices that facilitate self-employment, having skills favorable for starting a business. These immigrants are more likely to have self-employed parents, to have been exposed to training in small business, or to have been self-employed themselves in the country of origin. It has been documented in the literature that the children of self-employed parents are more likely to be self-employed themselves (Blau and Duncan 1967; Hout and Rosen 2000), and this tendency increases with earlier experiences with self-employment (Nee, Sanders and Sernau 1994). In

level tend to increase after micro-level variables with strong effects have been included (Snijders and Bosker 1999). The multilevel models are appropriate, however, for testing macro-level hypotheses.

Self-Employment

127

view of this, one would expect that the self-employment rate in the home country has a positive effect on the odds of self-employment. The analysis does not support this hypothesis, however. Model 1 shows that the relationship is positive, but not significant. The results of this study suggest that self-employment practices of immigrants in the destination country are not merely a reflection of patterns observed in the origin country. In an earlier study, Yuengert (1995), found a significantly positive relationship for immigrant groups in the United States, but his analysis was criticized for ignoring a group-level component of the error term in his individual-level equation. In a replication, using more appropriate techniques, Fairlie and Meyer (1996) could not find such a significant effect for immigrant groups in the United States. The analysis of this chapter does not find that political reasons for migration affect rates of self-employment. Theoretically, it is assumed that political suppression and instability in the origin country may induce people to migrate for other than economic reasons (Chiswick 1999). These non-economic migrants, mostly refugees, are not as prepared as people who move for mainly economic reasons to participate as a salaried worker in the normal labor market. Indeed, this book showed that people who move for predominantly political reasons are more often inactive and more often unemployed (Chapter 4), and have jobs of lower quality (Chapter 5). However, the results of this chapter show that non-economic migrants do not use self-employment as an alternative route to mobility. The religious and racial characteristics of the country of origin could be important as well. Because the Western countries examined are predominantly white and Christian, immigrants from non-Christian and non-white origins experience more discrimination in the labor market than white and Christian groups (Bogardus 1959; Evans and Kelley 1991; Model and Lin 2002). Non-Christian and non-white groups are therefore pushed out of the normal labor market; hence, one would expect that the odds of self-employment are higher among immigrants from non-Christian countries and non-white groups. As predicted, the analysis finds that immigrants from origin countries that are predominantly Christian are less often self-employed

Table 6.4 Cross-classified multilevel logistic regression of self-employment in 17 Western countries, 1980–2002, male immigrants 25-54 years old Destinations Destinations time-invariant time-variant Model 1 Model 2 Constant -4.871 (.159) -5.615 (.181) Origin Self-employment rate (%) Political suppression Predominantly Christian origin White (%) Destination Classic immigration country Self-employment (%), native reference group Unemployment (%), native reference group Community GDP origin/GDP destination Official language

.001 -.012 -.312** .006**

(.004) (.014) (.097) (.002)

.004 -.010 -.322** .007**

(.003) (.012) (.109) (.001)

-.274 .009

(.160) (.009)

-.192 .029**

(.128) (.004)

.021

(.012)

.058**

(.007)

.055 -.010

(.044) (.027)

.203** .035

(.042) (.031)

Educational heterogeneity Relative group size (%) Duration of stay group (% 10+ years) Sex imbalance Geographic distance (per 1000 km) Employment group (%) Mean educational level group

.367** -.080** .003**

(.080) (.009) (.001)

.348** -.082** .003**

(.070) (.009) (.001)

.320** .020**

(.097) (.003)

.375** .019**

(.100) (.003)

-.009** .341**

(.001) (.054)

-.008** .325**

(.001) (.034)

Individual Age .029** (.001) .030** Duration of stay 0-5 years ref. ref. 6-10 years .411** (.026) .406** 10+ years .546** (.022) .545** Education Low ref. ref. Middle .181** (.020) .186** High .118** (.022) .123** Married .102** (.019) .101** Note: Standard errors in parentheses. * p < .05; ** p < .01 (two-tailed tests). Source: Van Tubergen (2005)

(.001) (.026) (.022) (.019) (.022) (.019)

130

Immigrant Integration

than those from mainly non-Christian countries. According to Model 1, the odds of self-employment among immigrants from Christian societies are (1 - e -.312 =) 27 percent lower than among immigrants from non-Christian nations. Inspecting the raw data reveals that selfemployment rates among immigrants from the Philippines (5%) and Mexico (6%), two Catholic origins, are considerably lower than those from Pakistan (26%) and Israel (29%), predominantly Muslim and Jewish origins, respectively (figures not presented). Unexpectedly, the percent white in the origin country has a positive impact on the rate of self-employment. One possible explanation for this anomaly is invoked by the idea of “consumer discrimination,” which argues that white consumers, who make up the majority of the countries examined in this chapter, tend to prefer buying from whites and dislike purchasing goods and services from non-whites (Borjas and Bronars 1989). It is possible that non-white immigrants experience discrimination from both employers (pushing them into selfemployment) and consumers (making it more difficult to be selfemployed), and that the latter process is stronger than the former. Destination Effects How are differences between destination countries to be interpreted? First of all, opportunities for starting a business may depend on policies concerning immigrants. Several authors have maintained that it is more difficult and costly for immigrants to start a business in a European country than in classical immigrant countries (Light 1994; Waldinger, Aldrich and Ward 1990). It is argued that the labor market is strongly regulated, which raises obstacles to entrepreneurship among immigrants. By contrast, in classical immigrant countries, like Australia, Canada, and the United States, there are virtually no formal barriers to the geographical or economic mobility of immigrants, which facilitates ethnic entrepreneurship (Aldrich and Waldinger 1990). Are self-employment rates of immigrants in classic immigrant countries higher than in the new European immigrant countries? The results do not support this idea. Model 1 shows that self-employment figures for immigrants in traditional immigrant countries are not significantly different from those of similar immigrants in new

Self-Employment

131

immigrant nations.18 These multivariate outcomes concur with the descriptive findings presented in Table 6.2, in which a higher average self-employment rate cannot be detected among immigrants in traditional immigrant countries. I do find that other characteristics of receiving nations play a role. Self-employment rates of immigrants tend to adjust to the selfemployment rate of natives, reflecting the general opportunity structure for entrepreneurship in the receiving economy (Razin and Langlois 1996). Model 2 shows a direct relationship between the percentage of self-employed natives in a specific country at a specific time and the odds of self-employment among immigrants in that country and at that time. The magnitude of the effect is .029 in Model 2, showing that for one percentage point increase in the self-employment rate of natives, the expected odds of immigrants’ self-employment increases by 2.9 percent. An illustration of this process provides the case of Greece, where self-employment rates in the period 1992-2002 dropped among the native population, from 41.2% to 37.0%, and among the immigrant population from 23.0% to 13.0%. Second, cross-national variation in self-employment rates of immigrants is related to the unemployment rate among natives. Model 2 shows that the percent unemployed natives in a specific country in a specific year has a positive and significant effect on the odds of selfemployment among immigrants in that country at that time. One illustration of this effect is the situation in the United Kingdom, where the unemployment rate of native males decreased in the period 19922002, from 10% to 4%, and the self-employment rate of male immigrants decreased accordingly, from 23% to 19%. The relationship between unemployment and self-employment reflects the role of discrimination. Immigrants are given second preference to natives as employees. This hierarchy may be especially harmful to immigrants in times of high unemployment, since in such times, they are the first to become unemployed and they experience the strongest decline in wages (Model and Lapido 1996; Thurow 1975). In countries with a high unemployment rate among natives, immigrants are pushed out of the labor market and become entrepreneurs. 18

With regard to the role of receiving nations, we have to look at Model 1 for time-invariant variables and Model 2 for time-variant variables.

132

Immigrant Integration

Community Effects How could we understand the role of immigrant communities, that is, effects that arise from the specific combination of origin and destination? For a start, the odds of self-employment depend on settlement intentions. Although no direct measure of “settlement intentions” is included in the analysis, several more indirect measures show higher rates of self-employment among communities with a longer settlement history in the host society. The most direct indicator of settlement intentions, the average length of stay of a community, shows that communities with a higher proportion of immigrants who have remained more than 10 years in the destination country have higher self-employment figures. The higher the average length of stay of the members of a community, the longer the settlement intentions of that group. Another, more indirect indicator of settlement intentions, geographic distance between the country of origin and country of destination, also supports the idea that groups that plan to stay in the destination country for a considerable period of time have a greater likelihood of self-employment. The analysis shows that groups who have traveled over a longer distance have higher self-employment rates. It is assumed that immigrants who move over a longer distance are more likely to settle permanently (Borjas 1987). However, the third indicator of settlement intentions, the sex balance, is positive, suggesting that communities with either many males or (quite uncommon) many females, which is indicative of a short settlement history and a short stay, have a larger propensity for self-employment. Overall, however, the findings are more in favor of the idea that immigrants who plan to stay for a longer period are more likely to be entrepreneurs. These results show that for communities that plan to stay for a short period in the destination country, self-employment is not an attractive option because entrepreneurship involves risks and costs. Self-employment is riskier than a wage/salary job and, hence, less attractive for sojourners than for immigrants who plan to settle permanently in the host society (Aldrich and Waldinger 1990). This study also examines the relationship between relative group size and self-employment. There is some disagreement in the literature about whether group size has a positive or a negative effect on the odds of self-employment. Evans (1989) argues that ethnic entrepreneurs

Self-Employment

133

have lower information costs regarding the consumption preferences of their own group, making it more attractive to become self-employed when the own group is larger. In addition, immigrants from larger groups find easier access to start-up capital in their ethnic community (Yuengert 1995). These notions would lead one to expect a positive relationship between group size and self-employment. Alternatively, it has been argued that supplying goods exclusively for one’s own ethnic group is a weak basis for running a small business (Aldrich and Waldinger 1990). Even for larger immigrant groups, the clientele base is usually quite small. Producing goods for the non-ethnic market is therefore considered crucial for the continuity of ethnic firms. In this respect, the presence of a sizable group of co-ethnics in the environment increases the level of competition by co-ethnics and, in turn, limits the attractiveness of starting a business. These ideas suggest that it is less attractive for immigrants of larger groups to be selfemployed. In addition, Razin and Langlois (1996) argue that immigrants from sizable groups can find ample opportunities as normal salaried workers in ethnic enclaves and, therefore, need not use selfemployment as an alternative strategy for economic mobility. In view of these arguments, one would expect that group size has a negative effect on the likelihood of immigrants’ self-employment. I find evidence for the latter idea: the analysis shows a negative relationship between the size of an immigrant community relative to the total population and the odds of self-employment. This supports the idea that self-employment is a less-attractive option for immigrants in larger immigrant communities. Immigrants from large communities find ample opportunities as normal salaried workers in ethnic enclaves and therefore need not use self-employment as an alternative strategy for economic mobility. And although larger groups have a larger market for ethnic goods, the ethnic consumer market is, even for a more numerous community, quite small. Furthermore, competition between co-ethnics for small markets is stronger in more sizable groups, which makes self-employment in larger groups less attractive. Another community factor that affects immigrants’ selfemployment is the heterogeneity of the group. The analysis shows that the effect of the group’s educational heterogeneity on self-employment is positive and significant. The magnitude of the effect is 0.367 in the time-invariant model, showing that the expected odds of selfemployment among immigrants in perfectly heterogeneous

134

Immigrant Integration

communities are 28 percent higher than among immigrants in groups with concentrated educational qualifications (i.e., 1 – e .367 x .67). The heterogeneity of the immigrant community leads to more selfemployment for two reasons. First, groups with a diverse composition could have greater access to resources and information than homogeneous groups (Granovetter 1973; Renzulli, Aldrich, and Moody 2000). In order to run a business, immigrants need to have sufficient resources and information about market conditions and business opportunities. Second, highly diverse groups probably have a higher niche diversity. The lack of niche diversity in an immigrant group is an obstacle for immigrants in that group to become self-employed (Lieberson 1980). When a group is composed of members with the same skills, valuable for the same kind of small business, the competition among members of that group for the consumer market is high, which makes it less attractive to be an entrepreneur. I further find that the unemployment rate of a group’s total active population has a positive impact on the likelihood of self-employment of that group. The unemployment rate at the ethnic community level indicates the degree of discrimination: communities that meet less discrimination will have a higher proportion of their members participating in the labor market. In these groups, immigrants can find jobs as normal wage/salary workers more easily, and for that reason are less likely to be self-employed. This study also finds an effect of the mean educational level of the immigrant group. The analysis shows that groups with higher levels of educational generally have higher self-employment rates. In higher educated groups, there is more ethnic capital (Borjas 1992). Immigrants can benefit from the financial resources and information to start a business. In addition, groups with high ethnic capital also have a better consumer market for the goods and services provided by the ethnic entrepreneur. This study found evidence that refutes two ideas proposed in the literature. The analysis does not show the predicted negative effect of the immigrant group’s language skills on self-employment. Groups who have been exposed in their home country to the receiving nation’s official language, and are assumed to speak the language fluently, do not appear to have lower self-employment rates than groups who have not been exposed. This contradicts the idea that immigrants in groups with lower language skills find it attractive to start a business, as more

Self-Employment

135

co-ethnics in these groups are willing to work for them. Evans (1989) suggested that the lower the average language proficiency of a certain group, the more attractive it is for immigrants from that group to become an entrepreneur. Evans assumed that immigrants with few language skills do not perform well in the open (non-ethnic) labor market and may find it profitable to work for co-ethnics. “The ethnic capitalist can profit by hiring workers, paying them more than the broader market offers but — at least initially — less than majority group workers with equivalent skills earn in the broader market” (Evans 1989:952). One explanation for not finding such an effect, is that I was unable to control for language proficiency at the individual level. Earlier research found that immigrants who have few command of the destination language have lower probabilities of being selfemployed (Fairlie and Meyer 1996; Lofstrom 2002). Hence, not taking into account the individual language skills of immigrants suppresses the contextual language effect. Second, I do not find that immigrant groups that have difficulties “transferring” their human skills are more likely to become selfemployed. People who move from developing nations to more economically advanced societies do not have a greater likelihood of being self-employed than people who move from more advanced societies. In Model 1 the GDP ratio has no effect, whereas in Model 2, I find a significant positive effect on self-employment. One would expect that immigrants from poorer nations have more problems in transferring their skills, leading them to become more often selfemployed (Aldrich and Waldinger 1990; Sanders and Nee 1996). These are groups that have moved from developing economies to more advanced nations, who find that their skills, most notably educational qualifications, are not valued by the native population at the same level as similar skills among immigrants from more developed nations (Borjas 1987). My research, however, refutes the idea that immigrants from poorer nations are more likely to become self-employed. CONCLUSIONS This chapter shows that the probability of immigrants to be selfemployed depends on their origins, destinations, and communities. Four main arguments can be drawn from this chapter.

136

Immigrant Integration

First, arguments about immigrants’ opportunities for selfemployment in traditional immigrant countries are probably overstated. In Australia, Canada and the United States about 15 percent of the employed male immigrants have their own businesses, which is about the average of immigrants in European countries. Multivariate analysis does not show a significant difference either. Hence, it is not the distinction between the “old” and “new” immigrant countries that is relevant cross-nationally. In line with this conclusion, Min and Bozorgmehr (2003) remark that although the government of the United States tried to promote minority business, it has failed to achieve that goal. They list a number of ways in which the U.S. government hinders the activities of small business owners. Instead of contrasting traditional and new immigrating countries, my study shows that the differences between European countries are more substantial. Whereas in some European countries more than 25 percent of the immigrants are self-employed (e.g., Ireland, Portugal and Spain), in other countries less than 10 percent of the immigrants are self-employed (e.g., Austria, Germany and Luxembourg). This study shows that the cross-national variation in the self-employment experience of immigrants concurs with the variation in selfemployment of natives and the unemployment rate in a country. The self-employment rate of immigrants tends to adjust to the selfemployment rate of natives, reflecting the general opportunity structure for entrepreneurship in the receiving economy, and it increases with the unemployment rate, reflecting stronger discrimination of immigrants in the labor market. Second, I find evidence against the idea that immigrants who are unfavorably selected in terms of human capital – and for that reason have difficulty finding employment in the normal labor market – use self-employment as an alternative strategy of economic mobility. It does not matter whether political suppression in the sending nation is high or low, indicating that the reasons for migration do not play a role in the decision to start a business. Thus, there is evidence to refute the suggestion that people, who move for predominantly political reasons and who were assumed to be less-well prepared to participate in the labor market as employees, are more likely to become entrepreneurs than people who move for economic reasons. My study also disproves the suggestion that immigrants from countries with relatively large selfemployment sectors are socialized into practices that facilitate self-

Self-Employment

137

employment in the destination country. This finding does not support the idea that immigrants’ cultural background and the specific skills to become an entrepreneur influence their economic standing across multiple destinations, as argued by Sowell (1996). In addition, my results show that people who move from developing nations to more economically advanced societies are less likely to be self-employed than people who move from more advanced societies. This opposes the idea that immigrant communities that have difficulties “transferring” their human skills are more likely to be self-employed. One explanation for these unexpected findings is that less human capital makes immigrants not only less successful in finding work as a wage/salary earner, but also limits their mobility as ethnic entrepreneurs. Possibly, immigrants who migrated for non-economic reasons, who have country-specific expertise in entrepreneurship, and who have difficulties transferring their skills are indeed more likely to start a business in the host country. But at the same time, those immigrants are also less likely to succeed in maintaining a business, due to the lack of human and financial capital (Bates 1997). One way further research could address this issue is to disentangle two processes not considered here: factors involved in starting a business and determinants of maintaining a business. It is assumed that those with unfavorable human capital more often start a business, but less often succeed in maintaining their business. Third, this study shows that immigrants use self-employment as an alternative strategy for economic mobility when opportunities to work in the labor market as a wage/salary earner are blocked due to discrimination. I find that higher unemployment rates among natives in a certain country increase the likelihood of immigrants’ selfemployment in that country. Thus, in countries with a high unemployment rate among natives, immigrant employees are pushed out of the labor market and become entrepreneurs. The same pattern is observed at the community level. The unemployment rate of an immigrant group’s total active population in a certain country has a positive impact on the likelihood of self-employment in that group. The amount of discrimination experienced by a community pushes immigrants of that community towards self-employment. Furthermore, immigrants from predominantly non-Christian countries of origin tend to be more often self-employed than immigrants from Christian nations. This supports the idea that non-Christian immigrant groups, who are

138

Immigrant Integration

more culturally distant from the predominantly Christian host countries, experience more discrimination in the labor market. Surprisingly, however, this study finds that the percentage of whites in the country of origin has a positive impact on the rate of selfemployment. This finding does not support the idea that people from predominantly white countries would experience less discrimination in the labor market and would therefore have lower rates of selfemployment. One possible explanation is invoked by the idea of “consumer discrimination,” which argues that white consumers, who make up the majority of the countries examined, tend to prefer buying from whites and dislike purchasing goods and services from non-whites (Borjas and Bronars 1989). It is possible that non-white immigrants experience discrimination from both employers (pushing them into selfemployment) and consumers (making it more difficult to be selfemployed), and that the latter process is stronger than the former. Fourth, this study provides insight in the role of social capital, suggesting that co-ethnics can both help and hinder each other in starting and maintaining a business. The results show a negative relationship between the relative size of an ethnic community and the likelihood of self-employment. This finding indicates that competition between co-ethnics for small markets is stronger in more sizable groups, which makes self-employment in larger groups more difficult. In a similar way I find that co-ethnics hinder each other in homogeneous groups. Communities composed of members with largely the same educational level show lower rates of self-employment than more diverse groups. In more diverse communities, the level of competition between co-ethnics is smaller because members are looking for different kinds of jobs. Furthermore, co-ethnics help each other in heterogeneous groups, because the educational heterogeneity of the ethnic community facilitates the availability of resources and information important for starting and maintaining a business. Finally, it appears that immigrants who belong to communities with a higher average education generally have higher self-employment rates. In these ethnic communities, more financial and human capital is shared among immigrants, facilitating the start and maintenance of a business.

CHAPTER 7

Language Proficiency

Due to the growing share of immigrants in many Western societies, there has been increasing concern for the degree to which immigrants acquire the language that is spoken in the destination country. The reasons for this concern are clear: Language skills are a form of human capital that positively affect immigrant earnings and labor market opportunities (Kossoudji 1988; Shields and Price 2002), and language fluency of immigrants is associated with better interethnic relations in a society (Espenshade and Calhoun 1993; Gordon 1964). Immigrants’ second-language proficiency has been typically studied from a micro perspective in a single country. Effects of individual characteristics on language proficiency, such as age of migration, duration of residence, and educational level, have been widely documented (Carliner 2000, Espenshade and Fu 1997; Espinosa and Massey 1997; Solé 1990; Stevens 1999). Persons migrating at a young age, who have been resident in the destination country for a considerable amount of time, and persons with a higher education generally have better language proficiency. Systematic differences in second-language fluency have also been observed by marital status, gender, and migration motive (Chiswick and Miller 1996; Stevens 1986). These micro-level effects are substantial and have been observed in several countries (Chiswick and Miller 1995). This chapter studies immigrants’ language proficiency crossnationally and looks at the influence of the sending country, the receiving country, and the ethnic community. Previous chapters have shown that these macro level components determine immigrants’ 139

140

Immigrant Integration

economic incorporation. Now, I will examine to what extent this is also true for language skills. LANGUAGE: EXPOSURE, EFFICIENCY, INCENTIVES Both sociologists and economists have studied immigrants’ language proficiency (Chiswick and Miller 1995; Espenshade and Fu 1997). Sociologists have mostly focused on the impact of immigrants’ exposure to the language and social opportunities for language learning (Stevens 1992, 1999). Economists, on the other hand, have noted the importance of immigrants’ self-selection and the difficulty to learn a language. Both disciplines have paid attention to the role of economic incentives in learning a new language. In a recent contribution to the literature on immigrants’ language proficiency, Chiswick and Miller (2001) summarized the various theoretical ideas with three general notions: “exposure,” “efficiency,” and “incentives.” It is important to emphasize that these notions are connected to the general theories discussed in Chapter 2. The idea of exposure is related to the structural opportunity theory. The notions on efficiency and incentives can be subsumed under the human capital theory. Below, I first review how these notions were used in previous research to explain individual-level determinants of immigrants’ language proficiency. Subsequently, I use these ideas to formulate contextual hypotheses that pertain to the role of origins, destinations, and communities. First of all, immigrants’ proficiency in the destination-language is considered a function of the amount of exposure to that language. Immigrants learn a new language by opportunities to hear, study, and use the language (Stevens 1999). Such opportunities typically depend on the language skills and usage of the people with whom immigrants interact, such as the partner, colleagues, neighbors, and friends. Language proficiency is also an outcome of immigrants’ efficiency to learn a new language. Efficiency is defined as the degree to which immigrants improve their language proficiency given a certain amount of exposure. It is assumed that difficulties to learn a new language are greater for people who are less favorably selected in terms of observed and unobserved human capital, as well as for people who have to learn

Language Proficiency

141

a language that is linguistically distant from their mother tongue (Chiswick and Miller 2001). Economic incentives also determine the language proficiency of immigrants. Language skills are a form of human capital that may improve one’s economic position, as illustrated by the strong effect of language fluency on earnings (Chiswick and Miller 1995; Espinosa and Massey 1997; Kossoudji 1988; Shields and Price 2002). As all forms of human capital, language skills are embodied in a person and immigrants are assumed to invest deliberately in learning the second language after arrival, depending on the expected costs of language investments and the benefits in terms of employment chances and earnings. The notions of exposure, efficiency, and incentives have been used with considerable success to explain a number of individual-level effects on immigrants’ language proficiency. In line with the ideas on exposure, it is found that the longer immigrants stay in the country of destination, and hence the more exposed they are to the official language, the better they speak that language (Jasso and Rosenzweig 1990a). In a similar way, researchers have studied immigrants’ language proficiency as an outcome of language exposure in the family, including the impact of the language of the partner (Espenshade and Fu 1997; Stevens 1985). The notion of efficiency is often used to explain why people who migrated at a younger age speak the language better. Young people are more capable of learning a new language than older people (Stevens 1999). The concept of investments, finally, has been used as an explanation for the positive relationship between school attainment and language proficiency. Immigrants with little education can find employment in ethnic enclaves and would therefore have few incentives to invest in language learning (Carliner 2000). Although these notions have been applied mainly to individual effects, they also can be used to understand contextual effects. The notion of exposure is, perhaps, the most obvious example. Because a language is learned and used in interaction with others, it is an inherently social phenomenon and exposure to the language will therefore depend on macro-level characteristics. The notions of efficiency and incentives, however, have contextual implications as well. The difficulties to learn a second language can be group-specific, depending, for instance, on the “linguistic distance” between the mother tongue of the group and the official language of the receiving

142

Immigrant Integration

nation. Similarly, economic incentives of investing in a second language can be higher or lower for immigrants in different social contexts. Below, I use the notions of exposure, efficiency, and incentives to systematically develop contextual hypotheses about the effects of origins, destinations, and communities on immigrants’ language proficiency. MEASUREMENT LANGUAGE PROFICIENCY This chapter focuses on speaking abilities. The dependent variable is destination-language proficiency, i.e., the extent to which respondents are able to speak the destination language. I constructed a variable containing four categories, and classified all surveys according to the same metric: (1) (2) (3) (4)

not at all not well well very well

(5.7%) (19.9%) (26.1%) (48.2%)

Table 7.1 shows how I classified the answering categories. Because the surveys were mostly done independently, there are several potential survey differences that may bias the estimates. I discuss three important differences: in questioning, in coding, and in reporting. A first and most obvious cause of measurement error could be associated with the wording and number of the response categories. Although the surveys were done independently, the labeling of the answering options is in fact quite similar across surveys (see Table 7.1). The lowest category appears to be quite similar in the different countries (e.g., none, can’t speak it, not at all, very bad). The highest category is also comparable (fluently, English only, very well, perfect). Note that the sometimes added top category “English only” is unproblematic. Probably more doubts may arise about the middle categories. In some countries, a distinction is made between two middle categories, in other countries, there are three middle categories. Moreover, the wording of the lower middle category varies among surveys.

Language Proficiency

143

I tried to solve this problem using additional logit models where the four-point scale is replaced by a two-point variable that combines categories 1, 2, and 3: (1) (2)

not at all, not well, well very well

(51.8%) (48.2%)

An additional advantage of this variable is that the distribution becomes more homogeneous (48 percent fluent and 52 percent less well). I assess to what extent the regression results change when using the twopoint scale instead of the four-point scale. A second potential source of bias is associated with the wording of the questions rather than with the wording of the answers. The surveys conducted in Australia, Belgium and the United States use a two-step procedure to measure language proficiency. In Australia and the United States, respondents are asked if they speak a language other than English at home, and only those who do are asked to report their English speaking abilities. In Belgium, people are asked which languages they speak, and only then do they report on their language proficiency. Because direct or indirect questioning can affect the assessment of language proficiency, I included a dummy variable in the analysis representing this difference. A third survey characteristic that may bias cross-national analyses, is the source of report. In most surveys, respondents themselves were asked to report their proficiency to speak the second language. By contrast, in surveys collected for Denmark, Germany, and Great Britain, destination-language proficiency was assessed by the interviewer.19 To examine whether these survey differences may obscure true differences in language proficiency, a dummy is included in the analysis representing this difference. I also analyzed a Danish survey conducted in 1986, that included both respondent and interviewer reports. This survey allows one to assess the implications 19

Objective assessment of language skills would be desirable, but self-reported or interviewer-assessed language skills have become standard practice in studies on language skills. Little is known to what extent and in which direction bias in self-assessed measurements occurs. However, Carliner (2000) discusses this issue and concludes that self-report and test based measures highly correlate.

Table 7.1 Measurement of destination-language proficiency Question AUS81 Do you ever speak a language other than English at home now? If not, how well do you think you speak English? AUS88 Is English first language spoken at home? If not, how well do you think you speak English? BEL93 Which languages do you speak? Could you tell me how well you speak them? BEL96 Which languages do you speak? Could you tell me how well you speak them? DEN88 How do you assess respondent’s proficiency in Danish? DEN99

How do you assess respondent’s proficiency in Danish?

GER88

Knowledge of German language? (speaking)

GER91

Knowledge of German language? (speaking)

GER94

Knowledge of German language? (speaking)

GER99

Knowledge of German language? (speaking)

GB94

Assess respondent’s English ability.

Cat 1 -not at all

Cat 2 -not well

-very poor -poor -fair -none -little -reasonable -none -little -reasonable -not at all -poor -reasonable -not at all -poor -reasonable -none -little -sufficient -none -little -sufficient -none -little -sufficient -none -little -sufficient -not at all -slightly

Cat 3 -well

-well

Cat 4 -very well -English only -very good -English only -very well

-well

-very well

-good

-perfect

-good

-perfect

-well

-perfect

-well

-perfect

-well

-perfect

-well

-perfect

-fairly well

-fluently

-good

ITA94 ITA98 NET91

NOR83

How is your actual knowledge of Italian? (speaking) How is your actual knowledge of Italian? (speaking) When you speak Dutch, do you have difficulties with that language? When you speak Dutch, do you have difficulties with that language? How well do you speak Norwegian?

-none -none -can’t speak it -can’t speak it -not at all

NOR96

How well do you speak Norwegian?

-very bad

NET94

USA80

Do you speak a language other than English at home? If yes, how well does this person speak English? USA90 Do you speak a language other than English at home? If yes, how well does this person speak English? Source: Van Tubergen and Kalmijn (2005).

-little -little -always

-well -well -sometimes

-very well -very well -never

-always

-sometimes

-never

-well

-very well

-well

-very well

-not at all

-poor -moderate -poor - moderate -not well

-well

-not at all

-not well

-well

-very well -English only -very well -English only

146

Immigrant Integration

of such differences directly. The findings, of which a cross-tabulation is presented in Table 7.2, show that there is a strong correlation between the two measures: r = 0.71. More importantly, when both variables are coded on an interval scale ranging from 1 to 3, I find virtually the same means: 2.13 for both respondent and interviewer assessment. A t-test for paired variables turns out to be not statistically significant (t = – .08). Hence, it can be concluded from this particular case where measures can directly be compared, that levels of language proficiency do not differ between sources of report. Table 7.2 Cross-tabulation of interviewer by respondent assessment of destination-language proficiency among immigrants in Denmark, 1986 Interviewer assessment Respondent Good Reasonable Poor/not at all Total assessment Good 78 29 3 110 Reasonable 45 176 48 269 Poor/not at all 2 33 148 183 Total 125 238 199 562 Source: Van Tubergen and Kalmijn (2005) MODELS This chapter employs linear- and logit-regression techniques to estimate destination-language fluency. Because the dependent variables in the analyses are of ordinal level, ordered-logit or multinomial-logit estimates are more suited. However, several authors remark that with destination-language fluency as a four-point dependent variable, the coefficients from ordered-logit and linear regression have the same sign, relative size, and statistical significance (Carliner 2000; Chiswick 1991). Also, linear and logit regression are somewhat easier to interpret and can be used more easily in a multilevel framework. Important to emphasize is that this chapter compares immigrants with respect to their language proficiency at the time of the survey. Although the duration of stay in the destination country is controlled, it should be recognized that the acquisition of a second language is a dynamic process. Immigrant groups enter their destination with a

Language Proficiency

147

certain amount of skills in the second language, they gradually learn the language, and they ultimately reach a certain level of proficiency. Differences between groups—or more precisely, between combinations of origin and destination—can then arise in three ways: Groups may have different initial language levels, they may differ in the speed with which they learn the language, and they may differ in the level they ultimately have after a substantial number of years in the destination (i.e., reach different plateaus). If Indian immigrants in the United States, for example, on average have better second language skills than Chinese immigrants, is this because they assimilated better or is it because Indians were already more proficient in English to begin with? With a cross-sectional design, it is impossible to separate these components very well. As a solution, some researchers have used a synthetic cohort approach, but this design is potentially biased without a separation of immigrant cohort effects and duration of stay effects (Borjas 1985; Carliner 2000). Although duration of stay is included in the models, I cannot separate duration effects from immigrant cohort effects because the surveys considered are generally not far apart in time. This makes a synthetic cohort design less feasible. However, this study follows an alternative approach to solve this problem. First, I exclude by design all groups in the data set that speak the language fluently already upon arrival. These are groups for which the dominant language in their origin resembles that of their destination, such as British immigrants in the United States. Second, for the remaining (non-fluent speaking) groups, I develop hypotheses about the degree to which they were exposed to the destination language in the country of origin. More specifically, I develop two measures for prior language exposure (i.e., degree of globalization and similarity of official languages) and these are included in the multivariate models. By examining these prior exposure measures, the larger part of the contextual variation in initial language ability is taken into account.20 Hence, the (other) hypotheses tested refer to the speed of language acquisition and to the level that immigrants ultimately reach. Both of 20

Studies measuring change in language ability in a prospective fashion are still in its infancy (Jasso, Massey, Rosenzweig, and Smith 2003), and do not consider more than one destination. In addition, there are no studies that do contain direct measures of prior ability at the contextual level, all studies use (at best) proxies for ability at arrival and this is also the approach advanced here.

148

Immigrant Integration

these will be reflected in the level of language proficiency at the time of the survey. A DESCRIPTIVE PROFILE For an initial assessment of the impact of the social context on immigrants’ language proficiency, Table 7.3 presents immigrants’ observed mean language score on a four-point scale by origin, destination, and community. Because it is not possible to present the findings on all 182 origin groups included in the data set, I selected for descriptive purposes 12 well-known immigrant groups that are observed in at least three destinations. The observed language scores are presented for communities, and for origins and destinations (total). I divided Belgium in a French speaking part, a Dutch speaking part, and a region where both languages are official and dominant. The analysis therefore includes 11 destinations. Such a description does not take into account the role of composition effects, nor does it estimate the role of each contextual factor independent of the other contextual factors. Table 7.3 therefore also presents the results of a multivariate cross-classified multilevel model with dummy variables for origins and destinations, in which I controlled for the individual-level variables age at migration, duration of residence, duration of residence squared, schooling, and sex. The 170 other origin groups were combined in one category in this model. The adjusted means present the predicted level of language ability for each destination country and for each origin country, with all individual factors centered on the mean and males as the reference category. Predictions for origin countries are evaluated holding constant destination. Similarly, predictions for destination countries are evaluated holding constant origin (set to the value of the 170 “other” groups). The results suggest that origins, destinations, and communities all play a role in the language skills of immigrants. The average language score of all immigrants in the data set is 3.17, slightly more than speaking the language “well” (score 3). Of the 12 selected groups, immigrants from the Philippines (3.64) and India (3.58) have particularly good language skills, whereas those from China (2.68), Turkey (2.64), and especially Vietnam (2.58) have little proficiency of

Language Proficiency

149

the destination language. The total adjusted estimates show that the differences between these origin groups remain after taking relevant individual characteristics into account. The difference between an average male immigrant from India, who has a language score of 3.53, and a comparable immigrant from Vietnam (2.10) is almost 1.5 point. This suggests that characteristics of the country of origin have contextual effects. Differences in immigrants’ language skills are also pronounced between destination countries. Host societies in which few immigrants speak the language well are the Dutch speaking part of Belgium (2.32) and Norway (2.51). Controlling for composition effects, these countries remain at the bottom of the receiving regions, accompanied by the French speaking part of Belgium, which ranks second lowest. Clearly better proficiency of the destination language have immigrants in the United States (3.21), Australia (3.27), and Denmark (3.43). The good performance of immigrants in Denmark is due to composition effects and should not be exaggerated, for it ranks only fourth best in the list of adjusted totals. The top ranking of immigrants in the United States and Australia remains the same, however. Next to differences between origins and destinations, Table 7.3 also gives some clues for the role of community effects. Compare, for instance, the language skills of Pakistani in Great Britain and the United States. In Great Britain, Pakistani have an average language score of 2.60, which is below the average score of Pakistani in all destinations (3.09) and also below the mean language score of all immigrants in Great Britain (3.07). By contrast, in the United States, the average language skills of Pakistani (3.53) is far above their general score (3.09) and also above the average of immigrants in the United States (3.21). Apparently, then, the specific situation of Pakistani in Great Britain and the United States determines their deviance from the general pattern expected by origin and destination effects. Although these initial descriptive figures are interesting, they primarily serve as an illustration that the country of origin, the country of destination, and the combination thereof are important for immigrants’ language proficiency. To assess and interpret such differences in a more systematic way, however, it is important to examine all 182 origin groups, all origin-by-destination combinations, and to take survey effects into account. I now turn to such an analysis.

Table 7.3 Language proficiency by destination and origin country: means for 12 selected origin groups Destination Origin AUS BEL BEL BEL DEN GB GER ITA NET NOR USA Total Total Dutch French Mixed adj. China 2.77 . . . . 2.40 . 2.77 . . 2.64 2.68 2.35 Greece 2.81 . . . . . 2.69 . . . 3.17 2.96 2.39 India 3.91 . . . . 2.94 . 2.75 . . 3.63 3.58 3.53 Italy 3.06 . . . . . 2.75 . . . 3.15 3.02 2.46 Morocco . 2.53 3.03 2.94 . . . 2.86 2.85 . 3.71 2.90 2.79 Pakistan . . . . 3.60 2.60 . . . 2.47 3.53 3.09 2.93 Philippines 3.71 . . . . . . 2.94 . . 3.61 3.64 2.94 Poland 3.13 . . . . . . 2.88 . . 3.15 3.10 2.49 Spain 3.00 . . . . . 2.74 . . . 3.02 2.92 2.53 Turkey 2.59 2.23 2.56 2.37 3.14 . 2.57 . 2.77 2.39 3.20 2.64 2.31 Yugoslavia 2.98 . . . 3.61 . 2.66 3.13 . 2.89 3.25 3.05 2.52 Vietnam 2.35 . . . . 2.52 . . . 2.20 2.79 2.58 2.10 Total 3.27 2.32 2.71 2.76 3.43 3.07 2.68 2.91 3.02 2.51 3.21 3.17 Total adj. 2.89 1.93 2.25 2.30 2.64 2.50 2.40 2.63 2.65 2.27 2.84 Note: adjusted total computed with models including age migration, duration, duration squared, schooling and sex (male=1). The scale ranges from 1 (not at all proficient) to 4 (very well proficient). Source: Van Tubergen and Kalmijn (2005).

Language Proficiency

151

DECOMPOSITION OF VARIANCE How much does immigrants’ language proficiency vary between individuals, origins, destinations, and communities? And to what extent do the micro and macro variables included in the analyses explain the variance at each level? To answer these questions, Table 7.4 presents the variance at different levels with and without the inclusion of macrolevel variables. To begin with, I estimated an empty model, defined as a model with random intercepts only, without the inclusion of explanatory variables. The analysis reports a variation of 0.101 between destinations, 0.177 between origins and 0.667 between individuals. The fraction of the total variance due to macro level is ([0.101 + 0.177]/ [0.101 + 0.177 + 0.667] = ) 0.294. Hence, more than a quarter of the individual differences in second-language proficiency of immigrants can be attributed to the country where immigrants came from and the country they migrated to. This suggests that macro-level factors play an important role in immigrants’ language proficiency. Decomposing this macro-level variance into variance among origins and variance among destinations yields the following results. The fraction of the total variability that is due to differences among destinations can be calculated as (0.101 / [0.101 + 0.177 + 0.667] = ) 0.107. The fraction due to the origin level is 0.187. Hence, origin plays a more important role than destination, which is probably related to the larger number of origin than destination cases, which, in addition, are all Western. Nevertheless, both immigrants’ country of origin and country of destination are associated with language skills. To assess the role of composition effects, I examine to what extent individual variables can explain macro-level variations. When only individual-level variables are added, the explained variance for destinations is 11 percent, i.e., (0.101 – 0.090) / 0.101. For origins it is 41 percent. This indicates that differences between origins and destinations can be explained partially by differences in individual characteristics. More than half of the variance remains, however, suggesting that contextual effects are involved as well. When the contextual approach advanced in this book is correct, the amount of variation of a certain component should diminish when context-variables are added to the model. This turns out to be the case. For example, the percentage reduction in error variance among

Table 7.4 Decomposition of variance cross-classified multilevel models Empty model Individual Individual + Individual + variables destination origin variables variables Destination .101 (.053) .090 (.048) .076 (.042) .092 (.051) Origin .177 (.021) .104 (.012) .104 (.012) .083 (.010) Individual .667 (.002) .476 (.002) .476 (.002) .476 (.002) Total .945 .670 .656 .651 Note: standard errors in parentheses. Source: Van Tubergen and Kalmijn (2005).

Individual + community variables .093 (.048) .065 (.008) .475 (.002) .633

All variables

.073 .051 .475 .599

(.042) (.006) (.002)

Language Proficiency

153

destinations is ([0.090-0.076] / 0.090) 16 when destination variables are added to the model with only individual variables. Similarly, the proportional reduction in variance is 20 percent among origins when origin variables are included. When community variables are introduced in models with only individual variables, the total macrolevel variance is reduced with 19 percent. MULTIVARIATE ANALYSES The results of the cross-classified multilevel analyses of immigrants’ destination-language proficiency are presented in Table 7.5 (linear regression) and Table 7.6 (logit regression). Model 1 includes all micro-level and macro-level variables. Because earlier studies have found some curve-linear macro-level effects (Espenshade and Fu 1997), Model 2 adds quadratic specifications. In general, the linear and logit models yield similar results, but the logit estimates are more often statistically significant. Models 3 and 4 are estimated for sensitivity analyses. Destination Effects What explains the role of receiving nations? This study finds two important factors. First, in societies with a left-wing legacy, immigrants have poorer command of the destination language. Both linear and logit estimates in Model 1 show that the more strongly left-wing parties are represented in the government in the 1980s, the less well immigrants speak the destination language. In Model 2 of Table 7.6 (logit estimates), the relationship becomes just below significance levels when quadratic specifications are added. All in all, however, the findings show a clear negative impact of the presence of left-wing parties in the government on immigrants’ language skills. The political climate in countries with a left-wing government is more tolerant towards immigrants, resulting in fewer incentives to learn the host language. Another, related, mechanism is that left-wing parties are in favor of a more “linguistic-pluralism” model of integration, whereas parties at the center or to the right of the political spectrum are more inclined to “laissez-faire” or “assimilationist” language policies.

Table 7.5

Constant

Cross-classified multilevel linear regression analysis of immigrants’ destination-language proficiency on individual and contextual characteristics in nine Western countries, 1980-1999 Countries as origins and destinations Countries as origins and surveys as destinations Model 1 Model 2 Model 3 Model 4 2.887 (.133) 3.090 (.145) 2.774 (.092) 3.074 (.170)

Destination effects Left-wing parties Prejudice Origin effects National imports/GDP (%) Non-Christian origin Political suppression GDP per capita (in 1000s USD) (GDP per capita)2 Community effects Official language Language distance Same language branch Indo-European, other branch

-.078** -.025

(.017) (.014)

-.063** -.047**

(.015) (.010)

-.062** -.020**

(.021) (.006)

-.055** -.035**

(.013) (.009)

.003** .116* -.010 .006

(.001) (.048) (.006) (.004)

.003** .106* -.013* -.013 .001

(.001) (.046) (.006) (.010) (.000)

.003** .074 -.011* -.012 .001

(.001) (.043) (.005) (.009) (.000)

.003** .077* -.011* -.016* .001*

(.001) (.039) (.005) (.008) (.000)

.247**

(.028)

.245**

(.027)

.211**

(.028)

.211**

(.030)

(.019)

.000 -.067**

(.019)

.000 -.068**

(.020)

.000 -.049*

(.020)

.000 -.047*

Non-Indo-European Relative group size (%) (Relative group size)2 Geographic distance (1000s km) (Geographic distance)2 Individual variables Male Age at migration Years since migration (Years since migration)2 Years of schooling Survey variables Respondent assessment Indirect questioning

-.158* -.108**

(.063) (.008)

-.004**

(.001)

.043** -.019** .042** -.001** .062**

(.003) (.000) (.001) (.000) (.000)

-.134* -.316** .098** .017** -.001**

(.055) (.021) (.009) (.005) (.000)

-.016** -.318** .103** .024** -.001**

(.052) (.022) (.009) (.005) (.000)

-.168** -.316** .102** .025** -.001**

(.048) (.022) (.009) (.005) (.000)

.044** -.019** .042** -.001** .062**

(.003) (.000) (.001) (.000) (.000)

.043** -.019** .042** -.001** .062**

(.003) (.000) (.001) (.000) (.000)

.043** -.019** .042** -.001** .062**

(.003) (.000) (.001) (.000) (.000)

-.086 -.008

(.128) (.102)

Number of observations Destination 11 11 22 22 Origin 182 182 182 182 Community 360 360 360 360 Individual 186,091 186,091 186,091 186,091 Note: standard errors in parentheses; * p < .05; ** p < .01. Source: Van Tubergen and Kalmijn (2005).

Table 7.6 Cross-classified multilevel logistic regression analysis of immigrants’ destination-language proficiency on individual and contextual characteristics in nine Western countries, 1980-1999 Countries as origins and destinations Countries as origins and surveys as destinations Model 1 Model 2 Model 3 Model 4 Constant .166 (.177) .239 (.001) .865 (.172) .569 (.121) Destination effects Left-wing parties Prejudice Origin effects National imports/GDP (%) Non-Christian origin Political suppression GDP per capita (in 1000s USD) (GDP per capita)2 Community effects Official language Language distance Same branch Indo-European, other branch

-.154** -.112**

(.032) (.018)

-.094 (.054) -.106** (.009)

-.207** -.101**

(.030) (.009)

-.203** -.095**

(.022) (.017)

.013** .386** -.050** -.009

(.002) (.119) (.009) (.016)

.015** .407** -.039** -.085** .004**

.013** .305 -.076** -.120** .005**

(.002) (.176) (.007) (.035) (.002)

.016** .529** -.108** -.042** .001*

(.002) (.164) (.011) (.010) (.001)

.616**

(.094)

.375** (.114)

.297*

(.129)

.305**

(.093)

.000 -.190**

(.066)

.000 -.359** (.069)

.000 -.373**

(.058)

.000 -.370**

(.066)

(.003) (.090) (.011) (.020) (.001)

Non-Indo-European -.983** Relative group size (%) -.297** (Relative group size) 2 Geographic distance (1000s km) -.041** (Geographic distance) 2 Individual variables Male Age at migration Years since migration (Years since migration)2 Years of schooling Survey variables Respondent assessment Indirect questioning

.000 -.054** .113** -.002** .162**

(.141) (.030) (.002)

(.012) (.001) (.003) (.000) (.002)

-1.216** -.925** .275** -.122** .004**

(.204) (.069) (.028) (.014) (.001)

-.850** -.943** .288** -.078** .002**

(.105) (.073) (.031) (.011) (.000)

-1.015** -.953** .291** -.104** .003**

(.017) (.084) (.035) (.011) (.000)

.000 -.054** .114** -.002** .162**

(.012) (.001) (.003) (.000) (.002)

.003 -.054** .114** -.002** .162**

(.012) (.001) (.002) (.000) (.002)

.003 -.054** .114** -.002** .162**

(.012) (.001) (.002) (.000) (.002)

-.071 .365**

(.132) (.071)

Number of observations Destination 11 11 22 22 Origin 182 182 182 182 Community 360 360 360 360 Individual 186,091 186,091 186,091 186,091 Note: standard errors in parentheses. * p < .05; ** p < .01. Source: Van Tubergen and Kalmijn (2005).

158

Immigrant Integration

In a linguistic-pluralism model of integration, immigrants and their children are offered opportunities to speak and learn their mother tongue (Castles and Miller 2003; Koopmans and Statham 1999; Vermeulen 2000). This includes mother tongue courses at school and translation of official documents into minority languages. In view of both arguments, the election of left-wing parties in the government (unintentionally) reduces immigrants’ exposure to the second language and the incentives of acquiring that language. Language learning not only depends on how the government of the receiving society approaches the immigrant population, it also depends on how the members of the receiving society treat immigrants. An important factor in this respect is anti-immigrant prejudice. Attitudes of native citizens towards immigrants affect immigrants’ exposure to the official language. If natives have strong anti-immigrant sentiments, it is more difficult for immigrants from all origins to interact with members of the receiving society. Previous research found that prejudice towards immigrants differs considerably among countries, due to such factors as the percentage of immigrants, immigration flows, and (change of) unemployment levels (Fetzer 2000; Scheepers, Gijsberts, and Coenders 2002). My results show that such anti-immigrant attitudes negatively affects immigrants’ language proficiency. The effect is statistically significant in the logit models. The magnitude of the effect is -.112 in logit Model 1, showing that for one percentage point increase in negative attitudes of the native population, the expected odds of speaking the language very well declines by 11 percent for all immigrants (i.e., 1 – e -.112). Anti-immigrant sentiments vary from three to almost 20 percent of the population, which suggests that this is an important factor that accounts for differences between countries. Origin Effects How are differences among origin groups to be interpreted? Factors are included that relate to the level of globalization, as well as to the political, economic, and religious conditions in the sending country. This study finds that, whatever their destination, immigrants from countries with more globalized economies speak the language better. The results of both the linear and logit models indicate that the degree of globalization in immigrants’ origin country is important for their

Language Proficiency

159

language proficiency. People who migrated from countries with a higher level of economic globalization have better language skills. Logit Model 1 shows that the effect of globalization (i.e., national imports / GDP) is substantial. One standard deviation increase in the globalization measure (i.e., 26.9) is associated with a 42 percent increase in language fluency, which is a substantial effect (i.e., 1 – e 26.9 x .013 ). This underscores the idea that immigrants from more economically globalized societies are more strongly exposed to the foreign language before immigrating, through such things as business relations, the media, or foreign-language learning at school. The greater integration in the organization of production, distribution, and consumption of commodities in the world economy inevitably resulted in more exposure to foreign goods, media, and also to a growing emphasis on foreign-language learning at schools in countries throughout the world (Massey et al. 1993). This primarily includes learning the world language (English), but also other languages that are spoken in many countries (e.g., French, Spanish), or languages that are important due to (commercial) relations to other countries. Furthermore, differences among origin groups are due to political conditions at the time of migration. I find that those who moved from politically suppressed societies have a poorer command of the destination language. The effect is significant in the logit models. This finding concurs with the suggestion that political migrants are less efficient in learning a new language. Political migrants, typically referred to as “refugees,” leave their country mainly because of war, discrimination, oppression, or other violations of political rights and civil liberties. Refugees are less-well prepared and therefore less-well selected for the labor market than economic migrants, which includes fewer abilities to learn the language of the destination country (Chiswick and Miller 2001). Moreover, refugees have more often experienced traumatic events and have more stress than economic migrants (Marsella, Bornemann, Ekblad, and Orley 1994), which will also hamper their efficiency of language learning. For both reasons, the stronger the suppression in the country of origin, the lower the language proficiency of immigrants. The fourth characteristic of the country of origin considered, is the level of modernization. Several authors have argued that economic incentives to invest in the destination language depend on the likelihood of return migration (Chiswick and Miller 2001; Espenshade

160

Immigrant Integration

and Fu 1997). In this reasoning, it is assumed that longer expected duration in the receiving country makes it more beneficial for immigrants to invest in second-language learning. One measure of long-term commitments to living in the destination country is the degree of modernization in the country of origin (Espenshade and Fu 1997). If economic opportunities in the origin country are less favorable, it is argued, all emigrants from that country have a uniformly greater expected reward of learning the language in the destination country. The shadow of a shared future may then be longer and economic incentives to invest in learning the language will correspondingly be higher. In view of this argument, one would expect that immigrants from more developed nations have lower language skills than immigrants from less advanced economies. However, Model 1 of the logit and linear estimates find no significant effects of GDP per capita. Logit Model 2 (Table 7.6) shows that the main effect of GDP per capita is negative and the quadratic term is positive. Both effects are significant. The turning point of the effect in logit Model 2 is located at about $ 11,500 GDP per capita (i.e., 1000 x .0855 / [2 x .0037] ). Language fluency declines with GDP before that point and increases with GDP after that point. A graphical examination of these effects, however, shows that the initial declines are very small (not shown). The increases after the minimum point are more substantial in size, but there are relatively few cases in that part of the data (i.e., 13 percent). Hence, it can be concluded that the level of modernization in the origin country has no important general effect on language proficiency. One possible explanation for not finding a clear negative effect of modernization on language proficiency is that the effect is suppressed: immigrants from wealthier countries received higher quality schooling, which increased their efficiency to learn a new language. Religious characteristics of the origin country may also be relevant for the process of language acquisition. One reason to believe that religion plays a role is that the frequency of daily interactions between natives and immigrants depends on social distance, the willingness to be associated with other groups (Bogardus 1959). Research in Canada and the United States showed that natives’ social distance towards ethnic groups partly coincides with a distinction in religion, ranking Islamic, Buddhist, and other non-Christian groups at the top of the social distance scale (Owen, Eisner, and McFaul 1981; Pineo 1977).

Language Proficiency

161

Because all host societies I examine are predominantly Christian, it can be argued that immigrants from predominantly non-Christian societies are less likely to develop close personal ties with members of the host society than immigrants from Christian societies. Because infrequent interaction with natives makes it more difficult to learn the language, one would expect that immigrants from a predominantly non-Christian origin country are less proficient in the destination language than immigrants from Christian origins. However, I find no support for the idea that immigrants from nonChristian origin speak the destination-language less well than those from mainly Christian societies. On the contrary, my analysis shows that immigrants from non-Christian origins have significantly better language skills. According to logit Model 1, the odds that immigrants from non-Christian societies speak the language very well are 47 percent higher than the comparable odds for immigrants from Christian societies. One explanation for this unanticipated finding is that the distance between the cultures of the home and host societies involves migration costs, hence selecting the more favorable immigrants from distant cultures. That would concur with the efficiency approach to language learning and needs to be further researched. Community Effects The third group of contextual effects stems from immigrant communities. A first characteristic of the community that may account for immigrants’ language proficiency, is a pre-migration relationship with the host society (Chiswick and Miller 2001). Several Western countries had colonies for extended periods of time. In most colonies, the native citizens were assimilated to the culture of the mother country and an integral part of this learning process was acquiring the language of the mother country. Even after de-colonization, these languages sometimes have the status of an official language (i.e., a language used in schools and formal settings), or even of a primary or dominant language (i.e., a language also widely used in informal contexts –at home, on the street). Immigrants for whom the dominant language in the origin country resembles that of the country they settle in, naturally speak the destination language perfectly already upon arrival. Because these groups are uninteresting to examine, they are excluded from my

162

Immigrant Integration

study. Nevertheless, important differences in pre-migration exposure remain. One would expect that groups for which the language of the country of destination was official in the home country have a higher level of language proficiency than groups for which this was not the case. Both linear and logit analyses indeed show that those originating from a country in which the destination language is official have better language skills than those born in a country in which the destination language is not official. Model 1 in Table 7.5 shows that, measured on a four-point scale, the difference between these groups is .247. This supports the idea that these immigrants were exposed to the destination language before migration, and therefore had higher language skills upon arrival. A second and related community effect is suggested by the notion of “linguistic distance,” the resemblance between the languages spoken in the origin and destination country (Finegan 1999; Grimes 2000). If the linguistic distance between two languages is small, the efficiency to learn the other language is higher (Beenstock, Chiswick, and Repetto 2001; Carliner 2000; Chiswick and Miller 2001; Espenshade and Fu 1997). When it is less difficult for immigrants to learn the destination language, exposure to that language will yield higher levels of improvement. In all the destination countries examined in this chapter, the official languages belong to two branches of the Indo-European language family: the Romance branch (e.g., French, Italian, Spanish) or the Germanic branch (e.g., English, Dutch, German). Linguistic distance is generally assumed highest when languages belong to different families (e.g., Afro-Asiatic vis-à-vis Indo-European or Uralic vis-à-vis Indo-European). When two languages belong to the IndoEuropean language family, they are considered more distant when they belong to different branches of that family (e.g., French vis-à-vis English is more distant than French vis-à-vis Italian). I find that linguistic distance plays an important role in language proficiency. As expected, Tables 7.5 and 7.6 show that people who moved from countries having an official language that does not belong to the Indo-European family, have the lowest destination-language skills. Those who moved from an Indo-European language speaking country, but not of the same Germanic or Romance branch as the destination country rank second lowest. The best destination-language skills are observed among language combinations that are linguistically

Language Proficiency

163

most similar: Indo-European languages within the same Germanic or Romance language group. The odds of speaking the language very well for these groups is 2.67 times as great as for immigrants from nonIndo-European speaking countries. Another community characteristic is the size of the immigrant group relative to the total population. For two reasons, group size may be important for language learning. First and foremost, the larger the immigrant group, the more likely are day-to-day interactions within the own group and the less likely are day-to-day interactions with the receiving group (Blau 1977). Immigrants from larger groups have therefore fewer social opportunities of exposure to the destination language, and will be less likely to learn the second language (Clyne 1991; Dustmann 1994; Evans 1986; Lopez 1996; Stevens 1992; Veltman 1983). A second mechanism has to do with incentives. Language learning may be less attractive if immigrants find themselves in an ethnic enclave that provides labor-market opportunities and for which destination-language skills are not required (Evans 1989; Portes and Bach 1985). A precondition to the formation of such ethnic enclaves or economies is the presence of a sizable group of countryfellows. Thus, additional to the argument that group size decreases the exposure to the destination language, it is argued that group size also lowers the economic incentives of learning the destination language. Both linear and logit analyses show that relative group size has the expected negative impact on immigrants’ language proficiency. Thus, the larger the immigrant group in a particular country, the poorer the language skills of that group. The effect is statistically significant in both the linear and the logit model. The magnitude of the effect is -.297 in the logit model, showing that for one percentage point increase in relative group size (which is a considerable range for minority groups), the expected odds of speaking the language very well declines by 26 percent. I also find that a quadratic specification of the group size effect is statistically significant (Model 2). Using a four-point scale, the turning point is located at 1.6 percent, which is at the high end of the scale (the mean relative group size is .09). Increases in relative size up to 1.6 percent are associated with a decline in language skills, but after that point, there is an increase in language skills associated with increases in size. Because there is only one group that is more than 1.6 percent of the population, this means that language skills decrease with group size, but that it declines faster at smaller sizes than at larger sizes.

164

Immigrant Integration

A final characteristic of communities considered in this study is the geographic distance between origin and destination countries. Geographic distance between origin and destination increases the costs of migration, which, in turn, affects immigrants’ self-selection (Borjas 1987; Chiswick 1999). Those who migrated over longer distance will be more favorably self-selected, which implies higher cognitive skills in general, including the ability to learn a new language. In addition, it is argued in the literature that immigrants who have traveled over greater distances are less likely to return in view of the higher costs of migration (Chiswick and Miller 2001). This would increase the economic incentives to invest in learning the host language. Thus, for reasons of both efficiency and incentives, one would expect that greater geographic distance between origin and destination country is associated with better second-language fluency. Model 1 of both linear and logit estimates shows that the relationship between distance and language proficiency is negative instead. Adding a quadratic term in Model 2 yields significant results as well. In the linear regression model, the main effect is positive and the quadratic term is negative (Table 7.5). Further graphical inspection, however, shows that the magnitude of the effect is negligible (the maximum predicted level is 3.16 whereas the minimum predicted level is 3.09). In the logit model, the main effect is negative and the quadratic term is positive (Table 7.6). The extreme is located at about 16 thousand kilometers, which is also the maximum of the scale. Hence, this model suggests that language ability generally declines with distance, although it declines faster at small distances than at large distances. The magnitude of the effect is more substantial here: Immigrants coming from a place about 500 kilometers away have a 2.5 times greater odds to speak the language fluently compared to immigrants who needed to travel 16 thousand kilometers. Perhaps this anomaly may be explained by migration and remigration patterns. When distances are small, groups can more easily travel between home and host locations. Because length of residence in the receiving nation is measured since the last entry, this would imply that geographically more nearer groups have longer resided in the destination country than groups from larger distances. Hence, these groups have been more strongly exposed to the destination language than this study was able to control.

Language Proficiency

165

Sensitivity Analyses Because a cross-national analysis of different surveys raises questions about comparability, I performed sensitivity analyses. I estimated the linear and logit Models 3 and 4 again, using surveys as destinations. This results in 22 destination-year cases. Model 4 includes two survey characteristics that could bias cross-national analyses: (a) Whether the second-language ability of the respondent was assessed by the interviewer or by the respondent, and (b) whether the question on language ability was direct or indirect (i.e., after an initial question about the languages the respondent speaks). These characteristics vary among surveys within destination countries. The findings in Table 7.5, using linear regression, do not suggest that the assessment of immigrants’ language proficiency differs by source of report, nor do I find that direct versus indirect questioning affects the assessment of language skills. Logit analyses presented in Table 7.6, however, suggest that direct versus indirect questioning affects the assessment of language skills. I find that surveys using indirect questioning yield higher language scores than surveys using direct questioning. More importantly, however, is that the macro-level effects in Tables 7.5 and 7.6 remain the same after controlling for these survey characteristics. This suggests that the findings are quite robust. CONCLUSIONS This chapter applied cross-classified multilevel techniques and finds, in accordance with the macro-level perspective pursued in this book, that immigrants’ language proficiency varies between origins, destinations, and communities. More than a quarter of the total variability in language skills can be attributed to the country of origin and the country of destination. Controlling for individual-level correlates of language proficiency, more than half of these macro-level differences between macro-units remain. This implies that, besides composition effects, contextual effects play an important role in the second-language proficiency of immigrants. In order to understand these contextual effects, I relied on three theoretical ideas about immigrants’ language proficiency that have been suggested in the literature. According to these ideas, immigrants’

166

Immigrant Integration

destination-language skills are a function of exposure to that language (both prior and after migration), of the difficulties to learn a new language, and of the economic incentives to invest in learning a new language. The role of exposure is connected to the general theory on structural opportunities, whereas the ideas on efficiency and incentives can be subsumed under the human capital theory. I used these ideas to develop hypotheses on the role of contextual factors that pertain to origins, destinations, and communities. Including these theoretically informed macro-level variables, I was able to explain a substantial part of the macro-level variance. Most of the results concur with the hypotheses. I also examined the robustness of the findings. First of all, I used two different classifications of language proficiency. The analyses, using a four-point scale and a two-point scale, generally yield similar results. Second, additional models were analyzed in which survey characteristics were taken into account. I found no difference between interviewer and respondent assessment of language skills. The logit analyses showed that indirect questioning yields somewhat higher scores of language skills than direct questioning. More important, however, is that the contextual effects found remain the same after these survey characteristics are taken into account. All in all, the sensitivity analyses suggest that the findings of this chapter are robust.

CHAPTER 8

Religion

This book is concerned with the role of the country of origin, the country of destination, and the community in the integration of immigrants. This chapter focuses on the religion of immigrants. One reason for doing such a study is that immigration flows increased dramatically in many Western societies after World War II. Nowadays, immigrants and their offspring make up a sizable part of Western populations, and their religious practices contribute to the religious profiles of these countries (Smith 2002). The present chapter differs in several important ways from previous chapters. Whereas in previous chapters I focused exclusively on the influence of the macro level, while controlling for individual effects, in this chapter I also hypothesize on effects at the individual level. The reason for doing so is that very few studies have studied the impact of individual characteristics on religiosity. Within the sociology of religion and the sociology of immigration, surprisingly little large-scale empirical research has been done on immigrants. The reason is that the study of the religion of immigrants has been hampered by the availability and quality of data (Ebaugh and Chafetz 2000; Warner and Wittner 1998; Yang and Ebaugh 2001a). As a result, several studies have focused exclusively on the religion of a single immigrant group, such as Greeks (Veglery 1988) or Koreans (Hurh and Kim 1990) in the United States. Other studies have been restricted to immigrants with a specific religion, such as Nelsen and Allen’s (1974) study of Catholic immigrants in the United States. More recently, several small-scale studies on the religion of immigrants have 167

168

Immigrant Integration

been conducted in specific regions of the United States. Examples of these include the “Religion, Ethnicity, and New Immigrant Research” project carried out in Houston, Texas (Ebaugh and Chafetz 2000; Yang and Ebaugh 2001b), and the ethnographic case studies of the “New Ethnic and Immigrant Congregations” project (Warner and Wittner 1998). In summary, little is known on both individual and contextual determinants of immigrants’ religion. Another important difference of this chapter compared to previous chapters is that religion is less indicative of integration. Whereas language skills and labor market incorporation are well-known indicators of immigrant integration, this is less clear in the case of religion. When are immigrants religiously “integrated” or “assimilated”? Answering this question is quite difficult, both conceptually and methodologically. Conceptually, religious assimilation probably means that, with length of stay in the host country, immigrants become less attached to the religious practices in their home country and adapt to the religious patterns of the host population (Van Tubergen 2007). This would mean that immigrants switch their religion in case the main religion in the home and host societies differ (e.g., from Muslim to Christian). Because religious switching is a rather rare event, it is probably better to look (as in the this chapter) at religiosity in terms of religious affiliation and religious attendance. Stronger religious assimilation would then imply that, when immigrants from a highly religious society (e.g., Morocco) move to a more secular society (e.g., the Netherlands), immigrants become less religious over their life-course. In contrast, when people move from a rather secular country (e.g., China) to a more religious nation (e.g., the United States), religious assimilation means that they become more religious. Methodologically, it is thus important to have information about immigrants’ length of residence in the host country, the religiosity of immigrants before migration (or as a proxy, parental religious’ involvement) and their religiosity after migration (i.e., at the time of the survey). In the present study, I have only information on the latter. Strictly speaking, then, I refrain from studying religious assimilation or integration in this chapter. Instead, I describe and explain the religiosity of immigrants from cross-national and cross-sectional perspective: why are some immigrants more religious than others? How do these

Religion

169

differences relate to their home and host nations and the interplay between these two? A final difference compared to previous chapters pertains to theory. In the previous chapters, I used theories on human capital, social capital, structural opportunities, and prejudice to understand macro differences in language acquisition and labor market incorporation. These theories are of little usage for the study of religion. Thus, rather than deriving hypotheses from the four general theories, I use other notions discussed in the sociology of religion and the immigration literature. MEASUREMENT RELIGION This chapter analyses two aspects of religiosity: religious affiliation (whether people think of themselves as members of a religious community, denomination, or religion) and religious participation (the frequency with which people attend religious meetings). Of all the surveys included in the International File of Immigration Surveys (IIFS), 20 contain comparable information of religious affiliation. Religious affiliation was standardized into a dichotomous variable: (1) Affiliated with a religion (0) Not affiliated with a religion

(88.8%) (11.2%)

Out of the eight countries and 20 surveys in IFIS, seven countries and 14 surveys contain information on religious participation (the surveys for Denmark did not have questions on religious attendance). To render the results as comparable as possible, I standardize the detailed answer categories in these surveys into two more-general categories: (1) Attending religious meetings once a week or more (29.7%) (0) Attending religious meetings less than once a week (70.3%) I did not detect problems in comparing the answer categories of both variables, because all surveys in the meta-file contain a fixed response category for no religious affiliation and for attending religious meetings once a week or more.

Table 8.1 Overview of survey questions regarding religious affiliation and religious attendance Survey Religious affiliation Religious attendance AUS84 What is your religious denomination? Is it How often do you attend religious services? Protestant, Catholic, some other religion, no religion, or what? AUS88 Do you think of yourself as having a religion or n.a. faith? AUS90 What is your religious denomination now? How often do you attend religious services? AUS94 What is your religious denomination now? How often do you attend religious services now? AUS95 What is your religious denomination now? How often do you attend religious services now? BEL93 What is your religious affiliation? Do you attend the mosque sometimes? BEL96 What is your religious affiliation? Do you attend the mosque sometimes? CAN86 What if any is your religion? How often do you attend services? CAN91 What is this person’s religion? n.a. DEN88 What is your religion? n.a. DEN99 What is your religion? n.a. GB74 What is your religion or church? How often do you go to church/mosque/temple? GB94 Do you have a religion or church? How often do you attend services or prayer meetings or go to a place of worship? ITA94 What’s your religion? n.a. ITA98 What’s your religion? Do you practice your religion in a place of worship (church, mosque, synagogue)?

NET94 NET98 USAgss USAnsfh

Do you think of yourself as part of a particular religious community, church, or religion? Do you think of yourself as part of a particular religious community, church, or religion? What is your religious preference? Is it Protestant, Catholic, Jewish, some other religion, or no religion? What is your religious preference?

n.a. How often do you attend religious services now? How often do you attend religious services?

How often do you attend religious services? (number of times per year) USAnes Do you ever think of yourself as part of a Do you ever attend religious services, apart from particular church or denomination? occasional weddings, baptisms, or funerals? Note: Questions have been translated into English where necessary. Source: Van Tubergen (2006b).

172

Immigrant Integration

Some difficulties, however, arose with regard to the comparability of question formulations. Table 8.1 provides an overview of the wording of the survey questions regarding religious affiliation and participation. It shows that there are no apparent differences that may systematically affect the comparability of responses with regard to religious participation.21 It also shows that the more or less standard way of asking immigrants about their religious affiliation was “What is your religion?”. However, there are two differences in the wording of the question that deviate from this standard and may affect the reliability of making cross-national comparisons. First of all, in most of the surveys, the question tends to assume that people have a religion (“What is your religion?”), but there are surveys in which the questions do not assume this (“Do you have a religion?”). The first kind of survey uses a so-called “one-step method” to ask directly for religious affiliation (e.g., Catholic, Protestant, Muslim, other religion, no religion). The second uses a “two-step method”: first, respondents are asked if they have a religion, and only if they answer affirmatively are they asked to specify what their religion is. Although both kinds of surveys provide a fixed answer category for having “no religious affiliation,” the wording of the question could have a systematic effect on the response. It is possible that one-step questions yield higher levels of religious affiliation than two-step questions. In order to deal with this bias when comparing surveys, I include a dummy variable in the analysis that exactly represents this difference. A second possible source of bias is related to the difference between “religion” vis-à-vis “religious denomination.” In most surveys, respondents were asked for their religion, but in some surveys, respondents were asked for their religious denomination. Whereas questions regarding religion are more of an indication of self-identified religious affiliation, questions in respect to denomination are more directed towards actual membership. Although people who identify with a religion are likely to be members of a certain religious community and people who are members of a religious community are likely to consider themselves religiously affiliated, these variables are not necessarily the same. I therefore include a dummy variable that represents this difference. 21

The surveys used for Belgium asked for mosque attendance, because only Muslims participated in that survey.

Religion

173

DESCRIPTIVE RESULTS I start with a descriptive overview of immigrants’ religiosity. Table 8.2 presents the results for religious affiliation, Table 8.3 for religious participation. Both tables suggest pronounced differences in religiosity among immigrants of different countries of origin. For example, Table 8.2 shows that the mean percent affiliated to a religion is above 95 for immigrants from India, Italy, Morocco, Poland, and Turkey. Religious affiliation is much lower among immigrants from Germany (83%), the United Kingdom (81%), as well as those from the former Yugoslavia (84%). Origin differences are also found with regard to religious participation (Table 8.3). Except for immigrants from Italy, the five origin groups with high religious affiliation (India, Morocco, Poland, Turkey) also attend religious meetings frequently. In particular immigrants from Poland, of whom 51% attends religious meetings at least once a week. In contrast, only 13% of immigrants from the UK attend religious meetings once a week or more. The descriptive figures also suggest that immigrants’ religiosity differs between host societies. Religious affiliation among immigrants is particularly high in Belgium (98%), Denmark (93%), Great Britain (95%), and Italy (94%). It is much lower in Australia (80%) and Canada (81%). In Australia, levels of religious participation among immigrants are also much lower. Of all immigrants in that country, only 13% attends religious meetings at least once a week. Much higher levels of religious participation are observed in Great Britain (34%), the Netherlands (33%), and the US (35%). I also find that the immigrant community is important. An example is the German community in Australia. Of the Germans who migrated to Australia, 70% is affiliated to a religion. This is below the average religious affiliation of Germans across all destinations (which is 83%), and also below the average religious affiliation of all immigrants in Australia (which is 80%). Thus, the religiosity of the German community in Australia deviates from the general differences among origin countries and destination countries. To give another illustration, consider the Moroccan and Turkish immigrant communities in the Netherlands. In both communities, 49% attends religious meetings at least once a week. This is clearly above the mean attendance of these origin groups across all destinations (34 to 35%), and also above the average attendance of all immigrant groups in the Netherlands (33%).

Table 8.2 Religious affiliation among immigrants in eight Western countries, 1974-2000 (%) Country of origin Country of Germany India Italy Morocco Poland Turkey UK Yugoslavia Mean all destination (ex-) groups Australia 70 91 93 . 93 . 77 86 80 Belgium . . . 98 . 99 . . 98 Canada 87 . 98 . 94 . 85 . 81 Denmark . . . . . 96 . 83 93 Great Britain . 98 . . . . . . 95 Italy . . . 96 98 . . 83 94 Netherlands . . . 98 . 97 . . 87 United States 85 87 87 . 96 . 82 . 89 Mean 83 98 96 97 96 Note: . = less than 50 respondents. Source: Van Tubergen (2006b).

98

81

84

89

Table 8.3

Religious participation among immigrants in seven Western countries, 1974-2000 (% attending religious meetings at least once a week) Country of origin Country of Germany India Italy Morocco Poland Turkey UK Yugoslavia Mean all destination (ex-) groups Australia 12 . 11 . . . . 8 13 Belgium . . . 31 . 28 . . 29 Canada 30 . 29 . . . 20 . 25 Great Britain . 35 . . . . . . 34 Italy . . . 20 54 . . 26 29 Netherlands . . . 49 . 49 . . 33 United States 20 24 35 . 38 . 21 . 35

Mean 21 34 22 34 51 Note: . = less than 50 respondents. Source: Van Tubergen (2006b).

35

13

22

30

176

Immigrant Integration

VARIANCE COMPONENTS Table 8.4 presents the variance components, obtained from crossclassified multilevel logistic models with random intercepts for immigrants’ origin and destination. I computed two different models: one in which countries are treated as destinations, and another in which surveys make up the destination level. Note that these are empty models, that is, without the inclusion of explanatory variables. The total variance is composed of variance between individuals σ2, variance between countries of origin τb00, and variance between destinations (i.e., countries or surveys) τc00. The macro-level variance can be obtained by computing the intraunit correlation coefficient ρbc = (τb00 + τc00) / ( τb00 + τc00 + σ2), where σ2 is fixed to 3.29. In the case that countries make up the destination component τc00, ρbc is 0.31 [i.e., (1.311 + 0.158) / (1.311 + 0.158 + 3.29)] for religious affiliation, and 0.21 for religious attendance. This implies that almost a third in the individual differences in religious affiliation, and almost a quarter of the individual differences in religious attendance can be attributed to the country of origin and the country of destination. This means that the correlations between outcomes of two (randomly chosen) immigrants who are from the same country of origin and who live in the same country are 0.31 (religious affiliation) and 0.21 (religious attendance). When surveys make up the destination level, the figures are 0.33 and 0.26, respectively. Apparently, macro factors are more important for understanding religious affiliation than for religious attendance. Decomposing the macro variation into two components results in a proportion of the total variation that is due to the country of origin ρb (i.e., τb00 / ( τb00 + τc00 + σ2)), and a proportion of the total variation that is due to the country of destination ρc (i.e., τc00 / ( τb00 + τc00 + σ2)). When countries make up the destination level, ρb (0.28) is much higher than ρc (0.03) with respect to religious affiliation, but ρb (0.10) is slightly smaller than ρc (0.11) for religious participation. When surveys are used as destinations, the difference with respect to religious participation is more pronounced (i.e., ρb = 0.09, ρc = 0.17). This suggests that the country of origin is more important than the country of destination for understanding immigrants’ religious affiliation, and the opposite is true for understanding the religious attendance of immigrants.

Table 8.4 Variance components from random intercept models without explanatory variables, crossclassified multilevel logistic regression of religious affiliation and religious attendance Religious affiliation Religious attendance Variance component Standard error Variance component Standard error Countries as destinations Country of origin 1.311 (0.231) 0.402 (0.089) Country of destination 0.158 (0.150) 0.448 (0.379) Individual 3.290 3.290 Total 4.759 4.140 Surveys as destinations Country of origin Surveys Individual Total Source: Van Tubergen (2006b).

1.286 0.343 3.290 4.919

(0.227) (0.145)

0.392 0.738 3.290 4.357

(0.085) (0.328)

178

Immigrant Integration

DETERMINANTS OF RELIGION Table 8.5 presents the results of the multivariate multilevel logistic regression analyses of religion. Model 1 shows the results for religious affiliation and Model 2 and Model 3 pertain to religious attendance. In Model 3, religious affiliation is included as a predictor of religious attendance, for two reasons. First, in addition to examining the role of having a Christian background at the contextual level (i.e., predominantly Christian origin country vis-à-vis mainly non-Christian origin country), this model allows to assess the influence of a Christian background at the individual level (i.e., Christian affiliation vis-à-vis non-Christian). This is important, because the macro-level concept does not consider patterns of selective emigration: it is possible that minorities having a non-Christian religion are migrating from predominantly Christian countries (or vice versa). Second, Model 3 provides the opportunity to examine whether the effects presented in Model 2 persist after religious affiliation is taken into account (i.e., affiliated to a religion vis-à-vis not affiliated to a religion). In other words, it is assessed whether the effects are either indirect (i.e., influencing the religious affiliation of immigrants) or direct (i.e., influencing the religious attendance of immigrants). Because I have included survey characteristics, and there are multiple surveys within destination countries, I have used country of origin as the origin level and surveys as the destination level. Individual Effects What determines the religiosity of immigrants? To begin with, I examine correlates at the individual level. The religion literature has documented a number of empirical regularities at the individual level, including such factors as age, sex, education, and marital status. These patterns have been observed earlier among native, Western populations, and the strategy of this chapter is to examine whether these individuallevel factors equally apply to the immigrant populations. One important individual-level factor associated with religion is age. The general idea advanced in the religion literature is that people’s religiosity increases with age, although different interpretations have been proposed. Stark and Bainbridge (1987) argue that at a higher age,

Religion

179

people have stronger fears of death, and religion may therefore become more important in providing the promise of an afterlife. Other authors (Chaves 1991; Hout and Greeley 1990) have proposed a life course model, maintaining that “over the lifespan individuals typically marry, settle down in a community, and have children. Presumably they are more inclined to attend church at each successive stage” (Firebaugh and Harley 1991: 495). Although the role of age is difficult to estimate due to associated cohort or period effects, most researchers conclude that age has a positive effect on religiosity (Argue, Johnson, and White 1999; Campbell and Curtis 1994; Firebaugh and Harley 1991). In view of these arguments and findings, one would expect that age has a positive impact on the religiosity of immigrants. My analysis supports this hypothesis. Age has a significant positive effect on the chance of religious affiliation, and on the likelihood of attending a religious meeting once a week or more. The magnitude of the effect is 0.018 (Model 1) for religious affiliation, and 0.026 (Model 2) or 0.025 (Model 3) for religious attendance. In other words, with each successive year, the expected odds of religious affiliation increases with 1.8 percent (i.e., 1 – e .018), and the odds of weekly religious attendance with 2.5 to 2.6 percent. Labor-force status is another individual-level characteristic associated with religiosity. One argument made in the religion literature is that employed people have less time to be active members of a religious community, and are therefore less religious than those who are unemployed or inactive in the labor market (Iannaccone 1990). In accordance with this idea, a study conducted in 22 countries found that those who are employed have weaker religious beliefs and attend church less often than those who do not have a job (Campbell and Curtis 1994). Hence, I predict that employed immigrants are less religious than unemployed or inactive immigrants. The analyses of religious affiliation and religious attendance are in line with this prediction. Employed immigrants are significantly less often affiliated with a religion and significantly attend religious meetings less often than immigrants who are unemployed or inactive. A third individual-level factor is sex. Researchers have hypothesized that women are more religious than men. Some authors have maintained that, because females are more risk-averse than males, females are more religious (Miller and Hoffmann 1995).

Table 8.5

Constant

Cross-classified multilevel logistic regression of religious affiliation and religious attendance among immigrants in eight Western countries, 1974-2000 Religious Affiliation Religious Attendance Model 1 Model 2 Model 3 Coefficients S.E. Coefficients S.E. Coefficients S.E. 2.606** (.207) -2.922** (.287) -5.877** (.385)

Individual effects Age Employed Male Schooling Married Religious affiliation No religion Christian Other religion Origin effects GDP per capita (in 1,000 USD) Predominantly Christian Political and religious suppression

.018** -.233** -.423** -.028** .340**

-.065** -.030 -.021

(.002) (.040) (.040) (.004) (.043)

(.023) (.179) (.019)

.026** -.228** .067 -.017** .380**

-.046** .514** -.001

(.001) (.041) (.043) (.004) (.040)

(.017) (.172) (.023)

.025** -.220** .123** -.015** .363**

(.002) (.040) (.043) (.004) (.044)

Ref. 3.476** 3.143**

(.145) (.161)

-.023 .215 .023

(.018) (.150) (.028)

Destination effects Religious concentration Religiosity (% religious attendance at least once a week) Social-democratic history (past 10 years) Community effects Relative group size (%) Survey effects Migration survey (vs. general survey) Two-step question (vs. one-step) Denomination (vs. religion) Number of observations Origin Destination (surveys) Community Individual

-.002 .015*

(.006) (.007)

.002 .025*

(.008) (.011)

-.017** .025

(.006) (.013)

-.108*

(.049)

-.146**

(.043)

-.090

(.090)

-.030

(.021)

.008

(.028)

.008

(.028)

.851

(.527)

.582

(.367)

1.104**

(.250)

-.613*

(.264)

.086

(.339)

140 20 272 38,244

* p < .05, ** p < .01, (two-tailed tests). Source: Van Tubergen (2006b).

113 14 189 19,548

113 14 189 19,548

182

Immigrant Integration

It is also suggested that socialization might predispose women more towards expressive values that are congruent with religious values and practices, whereas males learn more instrumental values, which are less consonant with religion (De Vaus and McAllister 1987). In line with both arguments, it is found that women are more religious than men, irrespective of the measure of religiosity (Miller and Hoffman 1995). Because both ideas would apply equally to immigrants, I predict that female immigrants are more religious than male immigrants. The differences observed by sex are intriguing. In line with theoretical expectations, I find that religious affiliation is indeed considerably lower among males. The odds that female immigrants are affiliated with a religion are 53 percent higher than the comparable odds for male immigrants. However, the supposed lower levels of attendance among males cannot be supported. According to Model 2, male and female immigrants participate in weekly meetings at an equal rate. Moreover, Model 3 shows that immigrant men more often attend religious meetings than immigrant women. Combined together, these results show that religious attendance among immigrant men is higher than that of immigrant women, once the higher percentage unaffiliated to a religion of immigrant men is taken into account. Education is another possible determinant of religiosity. One influential idea in the literature is that at schools, people are taught a mechanistic world-view, trained in critical thinking, and that this mechanistic, critical world-view is difficult to wed with the traditional, religious world-view (Bruce 1999; Lenski, Lenski, and Nolan 1991; Need and De Graaf 1996; Weber 1993 [1922]). According to this idea, one would expect a negative relationship between schooling and religiosity. Although a number of studies indeed found such a negative association (e.g., Kelley and De Graaf 1997), some studies showed no or even a positive relationship between schooling and religiosity (e.g., Smith, Sikkink, and Bailey 1998; Te Grotenhuis and Scheepers 2001). Veglery (1988) could not find any relationship between schooling and church-membership or attendance among first-generation Greek immigrants. Although the empirical support is somewhat weak for the supposed negative association between schooling and religiosity, I hypothesize such an inverse relationship for the immigrant population. This study indeed finds a significant inverse relationship between schooling and religiosity. This means that immigrants with a higher education are less often affiliated with a religion and attend religious

Religion

183

meetings less often than less-educated immigrants. Note that the magnitude of the effect is somewhat larger with respect to religious affiliation than with respect to religious attendance, and that the effect of schooling on attendance persists even after religious affiliation is taken into account. A final individual-level characteristic considered in this study is marital status; more specifically the distinction between married and unmarried people. It is generally argued that, while people’s religiosity influences the likelihood of marriage, union formation also influences people’s religiosity. Thornton, Axinn and Hill (1992) argue that cohabiting people –as opposed to married people– attending at religious services could receive sanctioning by religious leaders and other adults attending services. Furthermore, Sherkat and Wilson (1995) maintain that religious endogamy is more common than exogamy and that the partner constitutes a constraint on the choice of new religious options. Thus, although people partly choose their partner on religious grounds, they argue that religious norms of the partner also provide an enduring, independent force in an individual’s religious behavior. In sum, both arguments predict that married people are more religious than unmarried people. This hypothesis has received ample empirical support in general population studies (Campbell and Curtis 1994; Iannaccone 1990; Smith, Sikkink, and Bailey 1998; Thornton, Axinn and Hill 1992), and in a study of Greek immigrants in New York (Veglery 1988). In view of these ideas and observations, one would expect that married immigrants will be more religious than unmarried immigrants. In accordance with this hypothesis, I indeed find significantly higher levels of religious affiliation and participation among married immigrants. Origin Effects In line with the macro approach of this book, the religiosity of immigrants could also depend on factors associated with their home country, over and above their individual-level characteristics. One possibly relevant factor is the country’s level of modernization. One argument is that people who grow up in a modern country are less religious than those who were born in a more traditional country. With higher levels of education, technology, and more activist ideologies,

184

Immigrant Integration

principles like a spirit of free inquiry or freedom of thought are stimulated and an active, mechanistic world-view would be more dominant, leading, in turn, to a lower level of religious commitment (Bruce 1999; Lenski, Lenski, and Nolan 1991; Need and De Graaf 1996; Weber 1993 [1922]). In line with this idea, a cross-national study of 15 nations showed that the country’s modernization reduces people’s religious orthodoxy (Kelley and De Graaf 1997). I assume that modernization has an enduring influence on people’s religion, and hypothesize that immigrants who were born in a modern country have less religious commitment than people who grew up in a less-developed nation. In line with this idea, I find that immigrants from countries with a higher GDP per capita are less often affiliated with a religion. The magnitude of the effect is substantial. For instance, the odds that immigrants from the least economically developed country (i.e., $100 in 1980) are affiliated with a religion is almost seven times larger than the comparable odds for immigrants from the most developed country (i.e., $29,100). Model 2 shows that GDP per capita also has a significantly negative influence on religious attendance. Interestingly, when taking religious affiliation into account, the effect halves and becomes insignificant (Model 3). This suggests that GDP per capita has no direct influence on attendance, rather the effect is indirect: immigrants from more modern nations are less often affiliated to a religion, and for that reason attend religious meetings less frequently. I also consider immigrants’ religious upbringing. Because the host countries examined in this study are all predominantly Christian, social integration in the religious community of the host society will be presumably higher among immigrants who have a Christian background. Christian immigrants are probably stimulated in their religion by the native — Christian — majority, and also have ample opportunities for practicing their religion. By contrast, immigrants with a non-Christian background may, due to lack of groups reinforcing their religion and fewer structural opportunities, gradually loose their attachment to their religion. It is therefore predicted that immigrants from Christian origins are more often affiliated with a religion and attend religious meetings more frequently than immigrants from nonChristian origins. This hypothesis is partly confirmed in the analysis. I do not find that immigrants from predominantly Christian countries are significantly

Religion

185

more often affiliated with a religion. However, Model 2 shows the predicted effect on religious attendance. Immigrants from predominantly Christian countries attend religious meetings significantly more frequently than immigrants from non-Christian societies. Because it could be objected that emigration flows are selective, and that non-Christian people are emigrating from predominantly Christian countries (and vice versa), it is important to compare the results of Model 2 and Model 3. The effect of religious origin becomes insignificant once religious affiliation at the individual level is taken into account. The reason for this is that immigrants from predominantly Christian countries are mainly Christians and Christian immigrants participate more often than non-Christian immigrants. This finding is in line with theoretical expectations. The conditions in the home country at the time of migration can also be important for understanding the religious commitment of immigrants. It is maintained in the migration literature that some immigrants move for religious reasons, because of persecution and suppression in their country of origin (Chiswick 1999). In many nonWestern countries, religious freedom is limited (Marshall 2000), so migrating to a Western country might be induced by the possibility of gaining religious freedom. Based on these ideas, I predict that immigrants from religiously suppressive societies are more religious and attend religious meetings more frequently than immigrants from religiously more open nations. I do not find evidence for this hypothesis, however. Possibly, this is because I have to rely on a rather indirect measure of religious suppression (i.e., violation of political rights and civil liberties). In countries with fewer political rights and civil liberties, not only people with more religious commitment are more likely to move. In fact, all people with different opinions are more likely to emigrate, including those who have more secular thoughts. Because of these methodological problems, not finding the predicted effect of religious suppression does not imply that it is not there. Destination Effects Receiving countries can also play a role in the religion of immigrants, irrespective of the country they come from. One possibly relevant

186

Immigrant Integration

characteristic of host societies is the degree of religious pluralism. It is suggested in the religion literature that, similar to other types of markets, competition among religious firms tends to lead to the production or supply of religious goods or services of the kind consumers demand (Stark and Bainbridge 1987). In regulated and monopolized religious economies, it is maintained, religious firms produce unattractive religious products, badly marketed. Consequently, religious consumption is expected to be higher in a free, competitive market than in a monopolistic or oligopolistic religious situation. Although the empirical support for this idea is, at present, still open to debate (Chaves and Gorski 2001; Stark and Finke 2000; Voas, Olson, and Crockett 2002), tests have been mainly based on general populations. One exception is the study of Chaves, Schraeder, and Sprindys (1994). They found that the more competitive and unregulated the religious environment in an industrialized nation, the more likely it is in that country for Muslims (many of whom are assumed to be immigrants) to undertake the hajj (pilgrimage) to Mecca. I assume that in more religiously competitive and pluralistic host countries, immigrants are more likely to find a religion that suits their needs, and that the religious “products” will also be of higher quality. Hence, I predict that in more religiously competitive and pluralistic host countries, immigrants will be more religious. I find support for this hypothesis with respect to religious attendance, but not for religious affiliation. The significantly inverse relationship between religious concentration and attendance appears after religious affiliation is taken into account. Thus, I find that in more pluralistic and religiously competitive host societies, immigrants are not more often affiliated with a religion, but, once religious affiliation is controlled, immigrants participate more frequently than do immigrants in religiously monopolized countries. Note that it is not surprising that religious concentration has no effect on affiliation, since affiliation is less constrained by structural forces. Host societies may also be important in providing a more or less sacred canopy. In the eight countries I examine, the religiosity of the native population varies from relatively secular, as in the Netherlands (where 59% stated they had a belief in God in 1991), to more religious nations, such as the United States (94%) (De Graaf and Need 2000). In the religion literature, it is argued and indeed found that social groups (e.g., family, friends, school, media, neighborhood) shape one’s

Religion

187

religious environment and are therefore important for determining one’s religion (Berger 1967; Durkheim 1961 [1897]; Kelley and De Graaf 1997; Myers 1996; Need and De Graaf 1996; Te Grotenhuis and Scheepers 2001). Similarly, in the migration literature, it is maintained and observed that people who migrate from one region in a country to another region “accommodate” their beliefs to the religiosity of their destination (Bibby 1997; Stump 1984; Smith, Sikkink, and Bailey 1998; Welch and Baltzell 1984; Wuthnow and Christiano 1979). In view of these ideas, I assume that the religious commitment of immigrants tends to adapt to the religious context of the receiving nation. Thus, it is hypothesized that the religiosity of immigrants is directly related to the religiosity of the native population. I find some evidence for this hypothesis. Model 1 shows that the higher the percentage of natives in a country that attends religious meetings once a week or more, the more often immigrants in that country are affiliated to a religion. The religious participation of the native population also has the predicted positive effect on religious attendance of immigrants, although the effect is marginally significant in Model 3 (t = 1.92). The role of the political makeup of receiving societies can also be important for the religion of immigrants. It is argued in the literature that social-democratic parties have more activist and secular ideologies than Christian and liberal parties (Lenski, Lenski, and Nolan 1991). The more activist and secular ideologies of social-democratic parties are assumed the lower people’s attachment to traditional, religious world-views. Based on these ideas, I predict that immigrants in societies with a predominantly social-democratic legacy are less religious and attend religious meetings less frequently than immigrants in societies with predominantly Christian and liberal parties in the government. I find that immigrants in countries with a social-democratic legacy are indeed less often affiliated to a religion than immigrants in countries with a dominant Christian and/or liberal history. In addition, Model 2 shows that the presence of social-democratic parties in the government reduces the religious attendance of immigrants. However, on close inspection, it appears that political parties influence the attendance of immigrants only indirectly, by decreasing the rate affiliated to a religion. Model 3 shows that, once religious affiliation is

188

Immigrant Integration

controlled, social-democratic legacy has no direct effect on religious attendance of immigrants. Community Effects Next to the general role of immigrants’ country of origin and country of destination, the interplay between origin and destination could also determine the religiosity of immigrants. One such community factor is the size of the immigrant group. In a way, the size of the immigrant group is indicative of the cohesiveness of the community. Immigrants maintain their religion through interactions with other immigrants from their origin country and the presence of a large group of co-ethnics in the direct environment strengthens religious beliefs and practices (Berger 1967; Durkheim 1961 [1897]; Kelley and De Graaf 1997). Furthermore, to establish a religious community and to fund places of worship, a sufficiently large number of co-religionists in the direct environment is necessary. Based on these ideas, I predict that the larger the relative size of the immigrant community, the higher the level of religiosity of their members. However, my analysis does not support this idea. I do not find a significant effect of the relative size of the immigrant group on religious affiliation and weekly church participation. In various other models (not presented here), I examined quadratic specifications of group size as well as the bivariate relationship between group size and religion. None of these models showed a significant effect, however. One reason for not finding the predicted effect is that group size is only indirectly related to the more relevant idea of the cohesiveness of the immigrant community. In more close-knit communities, religious behavior can be better controlled and sanctioned. Although the cohesiveness is partly influenced by the size of the immigrant group, it is also strongly determined by spatial segregation. Sensitivity Analyses In order to examine how robust the findings are to differences in surveys, I performed sensitivity analyses. It appears that surveys that are designed to examine immigrant populations show significantly

Religion

189

higher levels of religious attendance, but not religious affiliation. Thus, a methodological conclusion can be drawn: general population surveys, such as the General Social Survey of the U.S., underestimate the religious attendance of the immigrant population. In addition, Model 1 shows that surveys that use a two-step question to ask for religious affiliation have lower levels of affiliation than surveys using a one-step method. According to the analysis, asking for denomination or religion does not play a role. Comparing the findings with and without the inclusion of survey effects, shows similar results (not shown). CONCLUSIONS In the sociology of religion, little attention has been paid to the study of immigrants. In this chapter I developed a specific migration framework, in which the religiosity of immigrants is an outcome of individual characteristics (individual effects), the country of origin (origin effects), the country of destination (destination effects), and the combination of origin and destination (community effects). Using notions discussed in the religion and immigration literature, I proposed a number of factors that fit this conceptual apparatus. In contrast to case studies that focus on a single immigrant group in a single country, the specific migration framework adopted here looks at multiple groups in multiple countries. In that way, the religious experience of such diverse groups as the Mexicans in the U.S., Turks in the Netherlands, and Pakistani in Great Britain were compared. One valuable insight that came out of this comparison is that immigrants’ country of origin is more important than the country of destination for understanding immigrants’ religious affiliation, but that the opposite is true for understanding the religious attendance of immigrants. Another conclusion of this chapter is that across the different immigrants, ethnic groups, and countries several general patterns of immigrants’ religiosity emerge. It is important to confront these observations with ideas proposed in the existing literature. One way of assessing theories in the sociology of religion is to apply them to a new research area or population and examine their empirical success (Jelen 2002). Whereas the sociology of religion has focused mainly on JudeoChristian beliefs in Western nations (Turner 1983), immigrants originate from all over the world, including both highly religious and

190

Immigrant Integration

more secular nations, poor and rich, Christian and non-Christian. The patterns observed in my study generally concur with accepted insights of the sociology of religion, and these insights therefore have a broad empirical scope. On the other hand, some findings of my study contradict with earlier observations in the sociology of religion and thereby challenge the generalizability of well-known ideas. To start with the confirmations, I find that a number of individuallevel factors are important. It appears that immigrants’ religiosity increases with age and decreases with schooling. I also find that religiosity is lower among employed immigrants than among inactive and unemployed immigrants, and that married immigrants are more religious than unmarried immigrants. Because these relationships have been documented in previous studies on general populations, one could argue that they reflect certain “general” mechanisms at the individual level. Thus, while these patterns have been observed earlier among native, Western populations (i.e., predominantly Christian, wealthy) this study finds that they can be extended theoretically and empirically to the immigrant population (i.e., including non-Christian, poorer groups). This chapter also finds evidence for several theories in the literature. To begin, some findings are in line with modernization theory. Modernization theory argues that people have a certain “world-view” or “ideology,” which has a more or less active and rationalized representation of people and their environment (Lenski, Lenski, and Nolan 1991). In more passive world-views, people envisage themselves more or less fatalistically as part of a holy world. By contrast, in more active representations of humankind and the world, people actively participate and understand their environment. With higher levels of education, technology, and more activist ideologies, principles like a spirit of free inquiry or freedom of thought are stimulated and an active, mechanistic world-view would be more dominant, leading, in turn, to a lower level of religious commitment (Bruce 1999; Lenski, Lenski, and Nolan 1991; Need and De Graaf 1996; Weber 1993 [1922]). This theory is confirmed in my study by the observation that religiosity is lower among immigrants who were born in modern countries, lower among immigrants with a higher education, and lower among immigrants who live in receiving countries with a stronger presence of social-democratic parties in the government.

Religion

191

The results also shed light on social-integration theory. According to social-integration theory, people who are strongly integrated into a social group will be more likely to comply with the norms of that group (Berger 1967; Durkheim 1961 [1897]; Kelley and De Graaf 1997; Myers 1996; Need and De Graaf 1996; Te Grotenhuis and Scheepers 2001; Van Tubergen, Te Grotenhuis, and Ultee 2005). Social groups shape one’s religious environment and are therefore important for determining one’s religion. Initially, people are predominantly affected by the religiosity of their parents (Myers 1996). Later in life, however, people are exposed to other values in school and to the ideas of their teachers. Furthermore, people acquire new friends outside the family, and when they establish their own family, they are influenced by their partner. People also interact with colleagues at work and with neighbours and are exposed to a variety of attitudes and norms through the media. In accordance with social-integration theory, I find that Christian immigrants more often attend religious services than nonChristian immigrants, and that the religiosity of the receiving context positively affects the religiosity of immigrants. Finally, I find some support for the religious market theory (Iannaccone 1991; Stark and Bainbridge 1987; Stark and Finke 2000). This theory maintains that there are “religious markets,” which are similar to other types of markets in that competition among producers or firms tends to lead to the production or supply of goods or services of the kind consumers demand. Religious firms compete with one another for religious consumers. In regulated and monopolized religious economies, religious firms produce unattractive religious products, badly marketed. Consequently, religious consumption is expected to be higher in a free, competitive market than in a monopolistic or oligopolistic religious situation (Chaves and Gorski 2001; Stark and Bainbridge 1987; Stark and Finke 2000; Voas, Olson, and Crockett 2002). In line with this theory, this chapter shows that the religious concentration of the receiving nation is inversely related with immigrants’ religiosity. This study also finds some unexpected results, however, which challenge theoretical insights in the sociology of religion. Perhaps the most important finding of this study that contradicts theoretical considerations, is the male-female pattern. In general population studies it is well-known and consistently found that females are more religious than males, irrespective of the measure of religiosity. Although I find

192

Immigrant Integration

that female immigrants are indeed more often affiliated with a religion than male immigrants, female immigrants did not attend religious services more often than male immigrants. On the contrary, once the higher percentage of immigrant men not affiliated to a religion is taken into account, I find that immigrant men more frequently attend religious meetings than immigrant women. One explanation for this unexpected finding is that in my study all religions were taken together without considering differences between them. It is possible that specific religious practices and prescriptions are relevant here. For instance, some studies suggest that in Muslim communities, males attend religious services more often than females (Horrie and Chippindale 1990). Further research that examines the impact of these and other theological differences on the religious practices of immigrants is therefore to be encouraged.

CHAPTER 9

Conclusions

In Chapter 1, I raised a simple question: why are Mexican immigrants in the United States and Moroccan immigrants in the Netherlands poorly integrated? Why do they speak the host-country language less well and why are they disadvantaged in the labor market? The assimilation theory (Gordon 1964) would have a simple answer to this question. According to this theory, immigrant adjustment is just a matter of time. The theory, I showed in Chapter 2, is partly right: previous studies found that immigrants gradually learn the host language, they find employment after a few years and they improve their labor market position over time. Such processes occur both within the first generation and across immigrant generations. In brief, immigrants become more integrated over time. The assimilation theory does not tell the whole story, however. The theory was challenged empirically by studies showing that Mexicans in the United States and Moroccans in the Netherlands are less well integrated than other immigrant groups, even after differences in the length of stay of the groups and differences in demographic characteristics (e.g., gender, marital status) are taken into account. Why do such differences across immigrant groups exist, while they have been in the country for the same time? Why do some immigrant groups assimilate less quickly than other groups? The assimilation theory was confronted with another finding it could not explain. Some studies suggested empirically that immigrants from the same origin country who migrated to different countries or regions, were not equally integrated. Why would immigrants in one receiving context be more integrated than immigrants -from the same origin country- in another context? In Chapter 2 I argued that the 193

194

Immigrant Integration

assimilation theory could not explain why important macro differences persist after controlling for assimilation factors and other individual characteristics. More specifically, empirical findings suggested that the socio-cultural and economic integration of immigrants differ between immigrant groups and between receiving contexts. The main aim of this book was to examine such macro differences in the incorporation of immigrants. I argued, first of all, that recent demographic changes provided new opportunities for examining the influence of the immigrant group and the receiving country. Since the 1960s, migration flows to Western countries have increased substantially and have affected more countries than before. Classic immigration countries, such as Australia, Canada, and the United States, are again experiencing large-scale immigration, whereas new immigrant countries are emerging in Europe. As a result of this process, one could fruitfully compare the integration of immigrants crossnationally. At the same time, migration flows within countries have become more diverse, that is, immigrants today originate from a variety of countries: Western and non-Western, poor and rich, Christian and non-Christian (Castles and Miller 2003). This increasing heterogeneity of the immigrant population enables researchers to study the influence of the immigrant group. Furthermore, I argued that previous studies on immigrant integration have either studied multiple origin groups in a single destination country, or a single origin group in multiple destination countries. As a result of these single comparative research designs, the interplay between immigrants’ country of origin and country of destination has not been discussed. This book integrated both approaches into a double comparative design. This resulted in two different components of immigrant groups not clearly distinguished before. First, immigrant groups originate from a certain country, and the characteristics of their home country could affect their integration, irrespective of their destination. When, for instance, Mexican immigrants are poorly integrated not only in the United States but also in other societies, characteristics of the home countries could be advanced to explain these differences. I called these explanations “origin effects.” Second, differences between immigrant groups could also indicate, what I called, “community” effects, which refer to properties of the combination of the country of origin and the country of destination. It

Conclusions

195

could be, for example, that Mexican immigrants are less well integrated in the United States than other groups, but that the opposite is true in Canada. These differences cannot be accounted for by the characteristics of either the country of origin or the country of destination, but seem to arise from a combination of both. The double comparative perspective also provides new insights into the importance of host countries. What is the role of the destination country, irrespective of immigrants’ country of origin? What is the influence of immigration and integration policies? These kind of macro-level effects were labeled “destination effects” and they reflect the role of the receiving society in immigrant integration. In Chapter 3 it was argued that the single comparative designs are inadequate to study macro differences. As a result, there is uncertainty about the “empirical challenges” to the assimilation theory. In this book, I used a double comparative design, in which multiple origin groups were studied in multiple destination countries simultaneously. This provides a better perspective on macro differences. Empirically, I examined immigrant integration in terms of laborforce activity, employment, occupational status and self-employment, language proficiency, religious affiliation, and religious attendance. The first four issues measure economic integration (i.e., the degree of equality between immigrants and natives), whereas the other three are more or less indicative of cultural integration (i.e., the degree to which cultural patterns and values are shared among immigrants and natives). I analyzed the integration of immigrants from more than 180 countries of origin, in 19 Western destination countries, in almost 1,000 communities. The immigrants in this study originate from a variety of countries, both Western and non-Western. The destination countries include both classic immigrant countries (Australia, Canada, and the United States) as well as new immigrant countries in Europe (Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, and the United Kingdom). MAIN EMPIRICAL FINDINGS One general conclusion of this book is that the integration of immigrants not only depends on individual characteristics, but also on

196

Immigrant Integration

the country of origin, the country of destination, and the immigrant community. These macro components play an important role in different aspects of integration, even after controlling for individual factors. Thus, immigrants’ source country affects immigrants chances to participate in the labor market, their likelihood of being unemployed, their occupational status and their chances of self-employment. Immigrants’ proficiency in the host country language and religiosity are related to immigrants’ country of origin as well. This book equally showed that immigrants’ position in the labor market and their cultural integration depend on the host country and on the immigrant community. Thus, differences across origin groups observed in a single country are not always generalizable to other countries. In such cases, origin differences do not travel across destinations, highlighting the importance of the immigrant community. The major empirical conclusion of this book, then, is that in order to understand the integration of immigrants, we need to look at their country of origin, their country of destination, and their interplay. Although this general conclusion is interesting, it can be formulated more precisely. Throughout the empirical chapters of this book, I have presented descriptive figures that show how various aspects of integration differ by country of origin, country of destination, and the immigrant community. Looking at these results comparatively provides us with an interesting follow-up question: in which host country are immigrants integrated most? To answer this question, I ranked the receiving countries on a scale from low to high integration. Because of their somewhat ambiguous interpretation in this respect, I leave out three of the seven indicators of integration: religious affiliation, religious participation, and self-employment. Note further that this descriptive overview only ranks receiving countries, not origin groups or communities, which would be, due to their large numbers, impossible to present. Table 9.1 shows the results. It appears that immigrants in Belgium, in particular, are not integrated well culturally and economically. They do not speak the language well (especially in the Dutch speaking parts), have high inactivity rates, and high unemployment rates. Equally problematic seems to be France, where, except for the labor-force participation of males, immigrants have a low economic position. In two countries (Portugal and the United States) the situation of immigrants appears to be more favorable. In these countries, immigrants are more often active

Conclusions

197

in the labor market, they have high employment rates, and their jobs are of reasonable quality. In addition, in the United States, immigrants speak the official language quite well. In some countries, patterns of integration are less consistent across several dimensions. An example is Italy, where immigrants are very often active in the labor force, but, once they are active, less often succeed in finding a job. There is a second way to be more precise in concluding that we need to look at immigrants’ origin, destination, and their interplay. In each empirical chapter, I discussed how strong the integration of immigrants differs across the countries of origin and countries of destination. When we examine these results comparatively, we may ask whether different aspects of integration of immigrants are always to the same degree affected by these macro components. To answer this question, Table 9.2 presents the variance components obtained from the cross-classified models with random intercepts for country of origin and country of destination, estimated for all indicators of immigrants’ cultural and economic integration examined in this study. These are empty models, without explanatory variables. In all models presented in Table 9.2 destinations are treated as time constant. The total variance φ is composed of the variance between individuals σ2, the variance between countries of origin τb00, and the variance between countries of destination τc00. 22 Several important conclusions can be drawn from comparing the variance components. Consider the variation at the macro level across all outcomes, as a proportion of the total variance. This is the intraunit correlation coefficient ρbc = ((τb00 + τc00) / φ). Table 9.2 shows that across all outcomes studied, ρbc varies between 0.12 (labor-force activity of females) and 0.31 (religious affiliation). This indicates that between 12% and 31% of the total variance in the cultural and economic integration of immigrants is due to the variance among origin and destination countries. In other words, the correlations between

22

It should be remarked that (1) I did not compute a random variance of the origin by destination interaction τd00, that is, communities (for reasons explained in Chapter 3), and (2) in the models that use a logistic link function, the level-one residual implies a variance of π2/3 = 3.29 (Snijders and Bosker 1999:224).

Table 9.1 Overview of ranking of destination countries in the cultural and economic integration of immigrants: four indicators Cultural Economic integration integration Language Labor force activity Employment Occupational status proficiency Males and Males Females Males Females Males Females females Australia Austria Belgium Dutch part French part Mixed part Canada Denmark Finland France Germany Greece Ireland Italy

2 . . 11 8 7 . 1 . . 9 . . 6

11 5 15 . . . 10 16 9 6 3 2 11 1

15 6 18 . . . 3 7 4 16 11 8 13 1

4 8 16 . . . 7 9 14 17 6 5 12 18

4 6 17 . . . 6 9 16 18 8 13 11 15

12 14 10 . . . 7 5 2 17 14 14 3 .

9 16 7 . . . 5 11 4 17 14 15 1 .

Luxembourg Netherlands Norway Portugal Spain Sweden UK United States

. 5 10 . . . 4 3

4 17 . 8 7 18 14 13

10 14 . 2 17 5 12 9

1 13 . 3 11 15 10 2

Number of 11 18 18 18 destinations Note: . = no information. Figures go from high (1) to low integration.

1 10 . 2 14 12 5 3

9 6 . 8 13 11 1 4

13 6 . 3 12 10 2 8

18

17

17

200

Immigrant Integration

outcomes of two (randomly chosen) immigrants who are from the same country of origin and who live in the same country of destination are between 0.12 and 0.31, depending on the outcomes studied. This seems substantial compared to, for instance, results in educational research (i.e., pupils nested in schools), where values between 0.05 and 0.20 are common (Snijders and Bosker 1999). Furthermore, the findings give evidence to suggest that macro factors are more important for immigrants’ cultural integration than for immigrants’ economic integration. The intraunit correlations ρbc are between 0.21 (religious participation) and 0.31 (religious affiliation) among indicators of cultural integration. This is clearly higher than the intraunit correlations ρbc of economic integration, which vary between 0.12 (labor-force activity females) and 0.20 (employment males). Note that, on closer inspection of Table 9.2, it appears that the macro-level variation for economic outcomes is generally smaller for immigrant women than for immigrant men. All in all, we can conclude that immigrants who are from the same country of origin and who live in the same country of destination are quite similar to one another, but they are more similar culturally than economically. Are the country of origin and the country of destination equally important? Decomposing the macro variation into two components results in a proportion of the total variation that is due to the country of origin ρb (i.e., τb00 / φ), and a proportion that is due to the country of destination ρc (i.e., τc00 / φ ). It appears that ρb varies between 0.07 (labor-force activity males, employment females) and 0.28 (religious affiliation). This means that immigrants from different countries of origin differ in both their cultural and economic integration. Whether immigrants come from, say, Argentina, Japan, Pakistan, or Turkey, significantly influences their labor-force activity, employment chances, occupational status, the likelihood of self-employment, language proficiency, religious affiliation and attendance. However, some outcomes, such as religious affiliation, seem to depend more strongly on immigrants’ country of origin than other outcomes, such as laborforce activity. The same is true for the country of destination, but the range is much smaller. The intraunit correlation ρc is 0.02 for immigrants’ occupational status and 0.11 for immigants’ language proficiency, religious participation, and the labor-force activity of males. Thus, host societies play a role in the cultural and economic integration of

Conclusions

201

immigrants, but the importance is not the same. Immigrants who come from the same country of origin, but move to different nations clearly differ in their cultural and economic integration, but the differences are more pronounced with regard to some aspects (e.g., language proficiency) than to other aspects (e.g., occupational status). Another relevant issue to examine, is whether the integration of immigrants differs equally among origins and destinations. For that reason, Table 9.2 presents the ratio’s of the intraunit correlations, that is ρb / ρc. It appears that for 7 out of 10 outcomes studied, the variance between origins is larger than that among destinations. This suggests that, in general, the country of origin is more important for the integration of immigrants than the country of destination. Perhaps this is because the 19 Western countries examined in this study are more or less the same in important aspects, adopting, for instance, more or less the same integration and immigration policies. In that sense, it could be that these policies are important for the integration of immigrants, but because all countries have similar policies, I do not find large differences across nations. Another possibility is that countries differ from one another (e.g., in terms of institutions and policy making), but that these differences are simply not important for the integration of immigrants. I cannot disentangle these different possibilities, but they certainly deserve further attention. Although the general conclusion is that there is more variation between immigrants from different countries who traveled to the same destination than between immigrants from the same origin who moved to different countries, it is interesting to consider three marked “deviations.” First, Table 9.2 suggests that the country of origin is more important than the country of destination for understanding immigrants’ religious affiliation (ρb / ρc = 9.33), but that the opposite is true for understanding the religious attendance of immigrants (ρb / ρc = 0.91). Second, the occupational status of immigrants clearly depends on their country of origin, but hardly differs between destination countries. Third, whereas the labor-force activity of male immigrants is more strongly affected by the country they moved to than by their sending nation, the opposite is true for women immigrants. Although it is beyond the scope of the present study, these observations induce new questions, and call for further research.

Table 9.2 Overview of intraunit correlations of cross-classified multilevel models with random intercepts Origin + Origin Destination Origin / Sex Method Destination Destination Labor force activity .19 .07 .11 .64 Males Logit .12 .08 .04 2.00 Females Logit Employment

.20 .15

.10 .07

.09 .08

1.11 .88

Males Females

Logit Logit

Occupational status

.17 .15

.15 .13

.02 .02

7.50 6.50

Males Females

Linear Linear

Self-employment

.16

.10

.07

1.43

Males

Logit

Language proficiency

.29

.19

.11

1.73

Both

Linear

Religious affiliation

.31

.28

.03

9.33

Both

Logit

Religious attendance

.21

.10

.11

.91

Both

Logit

Conclusions

203

THEORETICAL CONCLUSIONS It is important to know empirically that immigrant integration differs across receiving nations. More specifically, it is relevant to see in which countries immigrants in particular have few language skills and are poorly incorporated in the labor market, and which countries provide contrasting cases. It is equally valuable to observe that immigrant integration differs across origin groups and to see examples of origin groups that are poorly integrated and origin groups that are successfully integrated. And, finally, we can learn from immigrant communities that are badly integrated in one country, but are more successful than other communities in another country. Interesting as these descriptive findings are, they leave an important question unanswered: Why do we see these differences? Why are immigrants in one country better integrated than in another country? Why are some origin groups more integrated culturally and economically than other origin groups? Observering these differences is one thing, to explain them yet another. To what extent can differences in the cultural and economic integration of immigrants across countries of origin, countries of destination, and communities be explained? Chapter 2 showed that the idea of contexts of reception was specifically proposed to understand macro differences. This idea focused on the role of government policy, labor market conditions, and ethnic community characteristics. I argued in Chapter 2 that this idea rests on assumptions from four general theories, and that by making these assumptions more explicit, the empirical scope of the theory is expanded. The four general theories are: human capital theory, structural opportunity theory, prejudice theory, and social capital theory. Although these theories were developed outside the migration literature, I have shown in Chapter 2 that they can be used to explain macro differences. The first major theoretical conclusion of this book is that macro differences in immigrant integration are partly the result of skill selection. In line with human capital theory, this study finds strong evidence that immigrants who arrive in the host country with fewer general human capital (i.e., education, talents, ambition), are less well integrated. They have more difficulties in learning the host language, they participate less often in the labor market, and when they do

204

Immigrant Integration

participate, they are more often unemployed and occupy less prestigious jobs. Besides these general skills, host-country specific human capital is important too. Immigrants who speak the host language less well, who have fewer information on the host country labor market, and who have problems in transferring their educational qualifications, are less well incorporated economically. Macro differences in immigrant integration are therefore to a certain extent attributable to skill selection, over and beyond observable skills at the individual level. This book systematically related a series of characteristics of immigrants’ country of origin, country of destination, and the immigrant community to skill selection processes. In line with human capital theory, for example, it was demonstrated empirically that immigrants who come from politically suppressive societies speak the language less well, are more often inactive in the labor market, more often unemployed, and have jobs of lower quality. Those immigrants, mostly called ‘refugees,’ predominantly migrate for non-economic reasons, making them less efficient in learning new languages and less well prepared for participating in the host country labor market. To give another example of skill selection, this book finds that immigrants from countries with an unequal income distribution who moved to more egalitarian nations are less well incorporated economically. The reason is that the less talented and less productive have much to gain by migrating to welfare states, which have more social security benefits. Skill selection is not the only explanation of macro differences. Even after considering individual characteristics and macro-level characteristics that select the skills of immigrants, this book finds that macro differences remain. Such macro differences reflect pure contextual effects, which play a role over and above individual characteristics of the immigrant. Prejudice theory provides insight in why context matters. According to prejudice theory natives have a positive attitude towards their own group and a negative attitude towards immigrants. Immigrants are not considered as a homogeneous out-group, however, and the degree to which natives have negative sentiments towards immigrants varies across origin groups, receiving countries, and immigrant communities. The theory assumes that negative attitudes towards immigrants are particularly strong when they are perceived as a threat to the native population, both culturally and economically.

Conclusions

205

Negative attitudes of natives towards immigrants lead to ethnic exclusion. For example, this book shows that in countries having strong anti-immigrant sentiments, immigrants have fewer language skills than in countries with a more immigrant friendly population. The reason is that in countries with more hostile attitudes, natives avoid social interaction with immigrants, leading to less exposure to the host country language, and thus fewer language learning. Exclusion of immigrants also takes place in the labor market. As an illustration, this book shows that immigrants from predominantly Christian origin countries are less well incorporated economically: they participate less often in the labor market, and when they do, they are more often unemployed than comparable immigrants from mainly Christian countries. The reason for the stronger discrimination of nonChristian immigrants is that they are perceived as more culturally threatening to the predominantly Christian native population. As a result of ethnic discrimination in the labor market, immigrants use selfemployment as an alternative route to economic mobility. This book also demonstrates that receiving nations are to a certain extent successful in reducing discrimination. In countries with a stronger presence of social-democratic parties in the government, immigrants are more active in the labor market, and they have higher employment levels. Social-democratic parties take more active measures (positive discrimination, equal opportunity policy) to reduce inequalities between immigrants and natives than conservative, liberal and other political parties. Contextual effects are also the result of structural opportunities. In particular, this book shows that immigrants’ language acquisition is considerably lower in large immigrant communities. In such communities, there are fewer interactions between immigrants and natives, leading to fewer exposure to the destination language. It is important to say that this book did not explore the influence of other macro characteristics besides group size that are associated with contact opportunities. Such an extension might show that structural opportunities play a major role in immigrant adaptation. The findings of this study shed light on the importance of social capital. Immigrants who have more social capital are better incorporated in the labor market. The degree of social capital depends on the willingness of people within their social network to offer help and on the resources available to them. Thus, within their network

206

Immigrant Integration

people could rely on others to obtain information on job opportunities or to help them financially in starting an ethnic business. In line with this theory, this book finds that immigrants who belong to higher educated communities are less likely to be unemployed, more likely to have jobs of higher quality, and more likely to have an ethnic business. The influence of the average education of an immigrant community is not attributable to immigrants’ own education, but plays a role over and above characteristics of the immigrant. This suggests that lower educated communities face a double disadvantage: immigrants in those communities have fewer human and social capital at the same time. In summary, the theories on human capital, prejudice, structural opportunities, and social capital help us interpreting origin, destination, and community differences in immigrant integration. They specify conditions under which immigrants from one origin country are better integrated than immigrants from other countries. They explain why a certain host country provides a favorable context for one origin group, but not for other groups. And they help us understand why in some nations the immigrant population is better integrated in general. Relying on the general insights of this book we can easily understand particular cases. To come back to the example introduced in Chapter 1, the poor language skills of Mexicans in the United States and Moroccans in the Netherlands are partly the result of similar community characteristics. Because English is not an official language in Mexico, and Dutch is not used in Morocco either, immigrants in both communities were not exposed to the host language before migration. As a consequence, both groups arrived with few language skills. In addition, Mexicans in the United States and Moroccans in the Netherlands belong to the largest immigrant communities in their countries. Immigrants in both communities have few daily interactions with natives at work and in their neighborhood, reducing the exposure to the host language and the incentives of learning that language. The theories and findings of this book also explain the economic difficulties that immigrants in both communities experience. Because both groups arrive with a lack of language skills, economic opportunities are severely limited. In addition, education qualifications obtained in Mexico and Morocco are not easily portable to the more modern host countries. Besides the lack of host-country specific human capital, both groups also consist of immigrants with fewer general human capital. The income distributions in Mexico and Morocco are

Conclusions

207

much more unequal than in the United States and the Netherlands. As a result, both communities consist of less talented and less productive people. Finally, immigrants in both communities have few social capital. The immigrants in their community are (equally) low educated, and therefore limited in the resources to help. FUTURE RESEARCH There are several ways, in which future research could extend the macro approach of this book to immigrant integration. Ethnic Intermarriage, Earnings, and Spatial Segregation This book examined the influence of macro-level factors on the economic and cultural integration of immigrants. However, I have not studied each aspect of these two more general dimensions. This means that questions on the role of the country of origin, the country of destination, and the immigrant community could be raised with regard to such issues as language use and ethnic identity (i.e., indicators of cultural integration), as well as immigrants’ earnings and quality of housing (economic integration). In addition, other dimensions of immigrant integration could be examined within this perspective. Examples are: ethnic intermarriage and inter-ethnic friendships (social integration), geographical segregation (spatial integration), smoking, drinking, life-expectancy (health status) and voting patterns (political integration). In each of these research fields, researchers may address the role of the country of origin, country of destination, and the immigrant community. Moreover, it would be relevant to examine the degree to which the theories applied in this study are empirically successful in these unexplored areas. The Fate of the Second Generation Another logical follow-up study would be concerned with the children of immigrants, the so-called “second generation.” For demographic

208

Immigrant Integration

reasons alone this group is of particular interest. Migration flows increased after the 1960s to the old and new immigrant countries. Naturally, at the beginning of the 21st century, Western countries consist of a sizable group of second-generation immigrants. How well are they incorporated in the labor market? How do they score on language tests? What happens to their religious belief and practices? The fate of the second generation has gained recent interest in the migration literature, especially the offspring of the 1960s migration flows to the United States (Portes 1996; Portes and Rumbaut 2001; Rumbaut and Portes 2001; Zhou 1997). In addition, several crossnational studies appeared on the children of immigrants. In these innovative studies separate analyses were conducted for different countries and compared subsequently (Crul and Vermeulen 2003; Heath 2006). These studies are the first that explored the role of the country of origin and country of destination in the position of the second generation. However, no cross-national studies have been conducted at the present time that rely on a single cross-national data file and that contain information on children of immigrants from a variety of countries in a variety of destinations. For theoretical reasons, it is interesting to confront the origindestination-community perspective adopted here with studies of the second generation. The theoretical question to be answered is whether the origin-destination-community factors that affected the integration of first-generation immigrants persist in the second generation. With respect to human capital theory, it is important to acknowledge that the children of immigrants did not choose to emigrate, and they did not choose to migrate to the country where they were born. Hence, the selectivity processes of immigration of their parents are of no direct significance to the children of immigrants. However, do they play a role indirectly? For example, does the geographic distance between origin and destination affect the economic integration of immigrant children due to the selection of unobserved human capital of their parents? Regarding prejudice and social capital theories, similar theoretical questions can be raised. For example, are second-generation immigrants from predominantly non-white, non-Christian countries less integrated culturally and economically? Is it true that children of immigrants belonging to less resourceful communities have a lower economic status?

Conclusions

209

Regional Differences Follow-up studies could use the origin-destination-community perspective in which the “destination” component switches from the macro-level to the meso-level. These are within-nations units, which have a meaningful geographical and theoretical interpretation. Examples are the 52 states in the US, the 12 provinces in the Netherlands, or the 16 “Bundesländer” in Germany. Meso-levels could be chosen at a lower level as well, such as the more than 3,000 counties in the United States, or the approximately 600 municipalities in the Netherlands. Extending the origin-destination-community approach in this direction is promising in at least three ways. First of all, it generates a number of interesting questions on the importance of local receiving contexts. Second, meso-level analysis increases the number of destination cases considerably. This is an important achievement with respect to the problem of “Small-N’s-Big Conclusions” (Lieberson 1991, 1994), so much associated with macro-level analyses. A third advantage of extending the origin-destination-community design to meso-level research, is that aggregate factors can be measured more directly. Destination countries are not homogeneous entities in certain aspects. For example, the size and segregation of immigrants groups generally tends to differ strongly within nations. Cross-level interactions Future research could also extend the macro approach to immigrant integration by hypothesizing on cross-level interactions. These are interactions between macro-level and micro-level factors, for example between relative group size (a macro factor) and education (a micro factor). Throughout this study (except for Chapter 8), I examined the role of macro-level factors while “controlling” for relevant micro-level factors. Hence, I distinguished a micro approach from a macro approach, without bringing these approaches together. Further research can combine both perspectives and examine macro-micro interactions. A natural way of doing this is to consider an important individual characteristic that determines integration and subsequently hypothesize how certain macro-level factors may affect

210

Immigrant Integration

this relationship. Consider, for example, the relationship between educational level and economic status. It is well known that immigrants with better educational qualifications perform better in the labor market. May this relationship at the individual level be mitigated by factors at the macro level? One possible argument is that the role of educational diplomas depends on the economic status of the immigrant community. It could be argued that the value of immigrants’ individual diploma is centered round the mean educational level of the community, thereby down-valuing immigrants with above average educational skills, and up-valuing immigrants with below average educational skills. A Dynamic Perspective Another promising way to extend the macro perspective of this book is to integrate it with a dynamic approach. From a dynamic point of view, macro differences in immigrant integration can come from different sources. First, differences arise at the time of migration (i.e., selective immigration). Second, immigrants can differ in their speed of change (i.e., assimilation). Third, some immigrants decide to leave the destination after a specific time, while other immigrants remain in the country (i.e., selective emigration). These issues play a role in the subjects examined in this study. For example, with regard to language proficiency (Chapter 7), some immigrants speak the language perfectly upon arrival whereas other immigrants have no knowledge of the official language. Furthermore, some immigrants more quickly acquire the language of the host country than other immigrants. Finally, differences in language proficiency arise because of selective remigration to the home country. With crosssectional data one cannot disentangle these effects. Combining the double comparative research of this book with a more dynamic design provides a better understanding of the role of the country of origin, the country of destination, and the immigrant community. Characteristics of immigrants’ origins, destinations, and communities could affect selective immigration, differences in assimilation, and selective emigration. For example, immigration policy should have an effect on selective immigration, whereas integration

Conclusions

211

policy should primarily influence the assimilation process, but possibly also selective remigration. To integrate double comparative and dynamic research, one needs to have cross-national comparable panel data for a large number of countries, in which immigrants from multiple origins are followed from the very moment they arrive in the destination country and for a considerable period of time. Recently, beginnings have been made to study immigrant integration in a prospective fashion (Jasso, Massey, Rosenzweig, and Smith 2003). At present, however, these studies have a fairly short time span and do not consider more than one destination. An alternative approach is to pool multiple cross-sectional surveys, which are conducted over a long period. Using such a synthetic cohort approach (Borjas 1985) enables the researcher to disentangle differences in selective immigration (i.e., immigrant cohorts), and differences in assimilation (i.e., years since migration). One drawback of this design is that one does not follow the same individuals over time, which implies that the estimates could be affected by selective return migration (Jasso and Rosenzweig 1990b). Despite that issue, combining the well known synthetic cohort approach with the double comparative perspective of this book is a fruitful way to further increase our understanding of macro influences on immigrant integration.

This page intentionally left blank

APPENDIX A: MICRO-LEVEL DATA

The International File of Immigration Surveys

I collected and standardized existing surveys containing individuallevel information on immigrant integration. The surveys were combined into one cross-national data set: the International File of Immigration Surveys (Van Tubergen 2004a). To obtain data that were both high quality and comparable across countries, surveys had to fulfill several criteria. First, surveys had to contain a sufficiently large number of immigrants to provide detailed analyses and the survey sample should (approximately) be nationally representative. Second, surveys had to contain standard questionnaires with fixed response categories. Third, surveys had to contain crossnational comparable independent and dependent variables. In addition, for the analysis of language proficiency, surveys had to be translated in the mother tongue of immigrants and/or bilingual interviewers had to be used in the field. Two sorts of surveys were collected: some specifically of immigrants and others that were of the general population. Specific immigrant surveys are designed to study immigrant populations. It is generally acknowledged that these surveys are especially suitable to study immigrant integration (Chiswick and Miller 1996). They make use of bilingual interviewers and may be translated into the language of the immigrants, immigrant groups are over-sampled, and the surveys contain detailed information on issues of migration and integration. A limitation is that they are often limited to three or four main groups, and sometimes exclude (well-integrated) smaller groups. 213

214

Appendix A

For that reason, it is important to include general population surveys as well. These can be, for instance, census data and labor force surveys. Because general population surveys are not designed to interview immigrant populations, some immigrant groups may be underrepresented. For instance, the General Social Survey in the United States does not make use of bilingual interviewers and is considered nationally representative only for adults who speak English well enough to understand the interview (Davis, Smith, and Marsden 2000). An important data source that is included in IFIS are annual laborforce surveys conducted in the European Union in the period 1992 to 2002. These are Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom. These surveys were collected and standardized by Eurostat, the statistical office of the European Union, and pooled into the so-called European Union Labour Force Survey (EULFS). Because Eurostat takes great care in standardizing the labor-force surveys, the designs of these surveys have become much alike in the 1990s, reducing problems of comparability (Eurostat 1998). Part of the entire EULFS data were released for the IFIS data set in the form of two requested multivariate tables. The data set is restricted to first generation immigrants, defined as those born outside the country of residence. To exclude the influence of differential school participation and retirement, I selected the population between the ages of 25 and 54 years for the analysis of employment (Chapter 4). To facilitate comparisons, the analyses of occupational status (Chapter 5) and self-employment (Chapter 6) were also restricted to the same age group. It appeared that for the traditional immigrant countries (Australia, Canada, and the United States) the samples were much larger than those for the new immigrant countries, so I restricted the number of respondents in large immigrant groups to a maximum of 2,000 per survey in order to avoid computational problems. Not all surveys of IFIS are included for each analysis. Table A1 gives an overview. One important reason is that a specific survey does not contain information on a relevant variable. For example, the EULFS provides only information on economic outcomes, and is therefore omitted from the study of language proficiency and religion. Another reason why some surveys have not been used in a particular study is that variables are not comparable.

Appendix A

215

One example is the Census of Canada in the study of language. Because the Canadian census question on language-proficiency is ambiguous, Canada was not included in the final analyses of language proficiency. Chiswick and Miller (1995), for example, equal those who speak French of English well enough to conduct a conversation with those who speak the language “not well,” “well,” or “very well.” On the other hand, Antecol, Cobb-Clark and Trejo (2003), Chiswick and Miller (1992), and Duleep and Regets (1992) classify those able to conduct a conversation as those who speak the language “well,” or “very well.” In a preliminary analysis of my study, I tried both classifications, but the logit estimates deviated too much from comparative analysis excluding Canada. CODING OF INDEPENDENT VARIABLES Some surveys contain precise information on all relevant independent individual-level variables, whereas other surveys have cruder measures or do not contain certain variables. The EULFS, for example, does not provide precise information on duration of residence and schooling and has no information on language skills. Therefore, I had to make some concessions to render the variables cross-nationally comparable (see below), especially in the analyses that include the EULFS. Age: Age was measured in years or by estimations of midpoints for surveys using age categories. Duration of residence: For the analysis of employment (Chapter 4), occupational status (Chapter 5) and self-employment (Chapter 6), I constructed three categories: 0–5 years, 6–10 years, and 11 years or more. The surveys used for the analysis of language proficiency (Chapter 7) contain more precise measures on length of stay, so for that chapter it is included as a continuous variable, measured in years. Age at migration: Measured in years or estimated midpoints for surveys using categories. For the census data of Australia and the United States, the older immigrant cohorts had to be excluded from my analysis, because the exact date of their arrival is unknown.

Table A1 Overview of countries and surveys included in the analyses Country Year Survey id Employment Occupational status 1 15 EU countries 1992-2001 EULFS X 2 14 EU countries 1992-2002 EULFS X 3 Australia 1981 AUS81 X 4 Australia 1984 AUS84 5 Australia 1988 AUS88 X X 6 Australia 1990 AUS90 7 Australia 1994 AUS94 8 Australia 1995 AUS95 9 Belgium 1993 BEL93 X 10 Belgium 1996 BEL96 X 11 Canada 1986 CAN86 12 Canada 1991 CAN91 X X 13 Canada 1996 CAN96 X X 14 Denmark 1988 DEN88 X 15 Denmark 1999 DEN99 16 Germany 1988 GER88 X 17 Germany 1991 GER91 X 18 Germany 1994 GER94 X 19 Germany 1999 GER99 X 20 Great Britain 1974 GB74

Selfemployment X X

Language

Religion

X X

X X X X X X X X X X

X X X X X X X X X X X

X

21 22 23 24 25 26 27 28 29 30 31 32 33

Great Britain Italy Italy Netherlands Netherlands Netherlands Norway Norway United States United States United States United States United States

Other information Destinations (N) Origins (N) Communities (N) Immigrants (N) Age groups

1994 1994 1998 1991 1994 1998 1983 1996 1980 1988 1990 1990-2000 1990-1998

GB94 ITA94 ITA98 NET91 NET94 NET98 NOR83 NOR96 USA80 USAnsfh USA90 USAgss USAnes

X X X X X

X X X X X

X

X

X

X X X

X

X

X

X

X X X X X

X X X 18 187 984 347,991 25-54

17 181 859 239,619 25-54

17 179 840 159,844 25-54

8 180 272 38,244 18-

9-11 182 360 186,091 18-

218

Appendix A

Education: In line with the classification of educational level in the EULFS, I used three categories for education: low (primary education and first stage of secondary education), middle (second stage of secondary education), and high (higher education). Surveys using measures of schooling (years of full-time education) were recoded using information on the years needed to obtain certain educational levels. Information was obtained from ISCED-97 (OECD 1999b). This variable is used in Chapters 4, 5 and 6. Schooling: Total years of full-time education. For surveys that had no direct measure of years of schooling, I relied on educational level and computed the average number of years needed to obtain that level, using the International Standard Classification of Education (OECD 1999b). This variable is used in Chapters 7 and 8. Marital status: I constructed two categories: married and unmarried (divorced, separated, single). Sex: Females are the reference category. Language proficiency: Immigrants’ language proficiency is used as an independent variable in Chapter 4. For surveys in seven countries (Australia, Belgium, Germany, Italy, the Netherlands, the United Kingdom, and the United States) I had individual-level information on immigrants’ proficiency in the destination language. I constructed a variable with three categories: 1 (immigrants who speak the language very well), 2 (immigrants who do not speak the language very well), and 3 (immigrants whose language skills are unknown). Labor force status: I constructed a dummy variable for employed versus all other (unemployed and inactive). Religious affiliation: I constructed a variable with three categories: whether people are affiliated to a Christian religion, a non-Christian religion or not affiliated to a religion. These are included as dummy variables. SURVEY REFERENCES Cross-National File Van Tubergen, F. 2004a. The International File of Immigration Surveys. Utrecht: ICS/Department of Sociology [IFIS].

Appendix A

219

National Files Australian Bureau of Statistics. 1981. Australian Census 1981 Matrix Files-Table M048. Distributed by Australian Social Science Data Archive, Canberra, Australia [AUS81]. Australian Office of Multicultural Affairs. 1988. Issues in Multicultural Australia. Distributed by Australian Social Science Data Archive, Canberra, Australia [AUS88]. Bumpass, L.L. and J.A. Sweet. 1997. National Survey of Families and Households: Wave 1, 1987-1988. Produced by Center of Demography and Ecology, University of Wisconsin, Madison, USA. Distributed by Interuniversity Consortium for Political and Social Research, Ann Arbor, USA [USAnsfh]. Danish National Institute of Social Research. 1986. Refugees in Denmark 1986. Distributed by Danish Data Archive, Odense, Denmark. ———.1988. Second Generation Immigrants 1988. Distributed by Danish Data Archive, Odense, Denmark [DEN88]. ———.1999. Second Generation Immigrants 1999. Distributed by Danish Data Archive, Odense, Denmark [DEN99]. Davis, J.A., T.W. Smith, and P.M. Marsden. 2000. General Social Surveys, 1972-2000. Produced by National Opinion Research Center, Chicago, Il, USA. Distributed by Interuniversity Consortium for Political and Social Research, Ann Arbor, USA [USAgss]. Eurostat. 2002. European Union Labour Force Survey, 1992–2001. Distributed by Eurostat, Luxembourg. ———.2003. European Union Labour Force Survey, 1992–2002. Distributed by Eurostat, Luxembourg. Kelley, J., R.G. Cushing, and B. Headey. 1984. Australian National Social Science Survey 1984. Distributed by Australian Social Science Data Archive, Canberra, Australia [AUS84] Kelley, J., C. Bean, and M.D.R. Evans. 1990. Australian National Social Science Survey 1989-1990. Distributed by Australian Social Science Data Archive, Canberra, Australia [AUS90] ———.1994. Australian National Social Science Survey 1994. Distributed by Australian Social Science Data Archive, Canberra, Australia [AUS94].

220

Appendix A

———.1995. Australian National Social Science Survey 1995. Distributed by Australian Social Science Data Archive, Canberra, Australia [AUS95]. Lesthaeghe, R. 1993. Family Formation and Values Patters for Turkish and Moroccan Women. Distributed by Vakgroep Sociaal Onderzoek, Vrije Universiteit Brussel; Vakgroep Bevolkingswetenschappen Universiteit Gent, Brussel, Gent, Belgium [BEL93]. ———.1996. MHSM Surveys Turkish and Moroccan Men in Belgium. Distributed by Vakgroep Sociaal Onderzoek, Vrije Universiteit Brussel; Vakgroep Bevolkingswetenschappen Universiteit Gent, Brussel, Gent, Belgium [BEL96]. Marplan. 1988. Ausländer in Deutschland 1988. Distributed by Zentralarchiv für Empirische Sozialforschung, Cologne, Germany [GER88]. ———.1991. Ausländer in Deutschland 1991. Distributed by Zentralarchiv für Empirische Sozialforschung, Cologne, Germany [GER91]. ———.1994. Ausländer in Deutschland 1994. Distributed by Zentralarchiv für Empirische Sozialforschung, Cologne, Germany [GER94]. ———.1999. Ausländer in Deutschland 1999. Distributed by Zentralarchiv für Empirische Sozialforschung, Cologne, Germany [GER99]. Natale, M., G.C. Blangiardo, and A. Montanari. 1994. La Presenza Straniera in Italia. Distributed by Dipartimento di Scienze Demografiche, l’universita di Roma ‘La Sapienza’, Rome, Italy [ITA94]. Natale, M. and S. Strozza. 1998. Indagine Statistica sulle Caratteristische e sul Comportamento Economico degli Stranieri Immigrati. Distributed by Dipartimento di Scienze Demografiche, l’università di Roma ‘La Sapienza’, Rome, Italy [ITA98]. Sapiro, V., S.J. Rosenstone, and the National Election Studies. 2002. American National Election Studies Cumulative Data File, 19482000. Distributed by Interuniversity Consortium for Political and Social Research, Ann Arbor, USA [USAnes]. Smith, D.J. 1974. Survey of Racial Minorities. Distributed by UK Data Archive, Colchester, England [GB74].

Appendix A

221

Smith, P. and G. Prior. 1994. The Fourth National Survey of Ethnic Minorities 1993-94. Policy Studies Institute. Distributed by UK Data Archive, Colchester, England [GB94]. Statistics Canada. 1986. General Social Survey Cycle 2: Time Use, Social Mobility and Language Use. Distributed by Statistics Canada, Ottawa, Canada [CAN86]. ———. 1991. 1991 Census Public Use Microdata File. Distributed by Statistics Canada, Ottawa, Canada [CAN91]. ———. 1996. 1996 Census Public Use Microdata File. Distributed by Statistics Canada, Ottawa, Canada [CAN96]. Statistics Norway. 1983. Levekår Blant Innvandrere 1983. Distributed by Norwegian Social Science Data Services, Bergen, Norway [NOR83]. ———. 1996. Levekår Blant Innvandrere 1996. Distributed by Norwegian Social Science Data Services, Bergen, Norway [NOR96]. United States Census Bureau. 1980. 1980 US Census 1 Percent Public Use Microdata Sample. Distributed by Interuniversity Consortium for Political and Social Research, Ann Arbor, USA [USA80]. ———. 1990. 1990 US Census 1 Percent Public Use Microdata Sample. Distributed by Interuniversity Consortium for Political and Social Research, Ann Arbor, USA [USA90]. Veenman, J. 1991. Sociale Positie en Voorzieningengebruik van Allochtonen en Autochtonen 1991. Produced by Instituut voor Sociologisch-Economisch Onderzoek. Distributed by Steinmetz Archive, Amsterdam, Netherlands [NET91]. ———. 1994. Sociale Positie en Voorzieningengebruik van Allochtonen en Autochtonen 1994. Produced by Instituut voor Sociologisch-Economisch Onderzoek. Distributed by Steinmetz Archive, Amsterdam, Netherlands [NET94]. ———. 1998. Sociale Positie en Voorzieningengebruik van Allochtonen en Autochtonen 1998. Produced by Instituut voor Sociologisch-Economisch Onderzoek and Sociaal en Cultureel Planbureau. Distributed by Steinmetz Archive, Amsterdam, Netherlands [NET98].

This page intentionally left blank

APPENDIX B: MACRO-LEVEL DATA

Country and Community File

CODING OF VARIABLES Besides individual variables, this book includes variables on the country of origin, the country of destination, and the immigrant community. Furthermore, I include characteristics of surveys, as a way to control for survey differences (see Chapter 3). Country of destination Point system: I set up a dummy to indicate whether destinations had a point system or not. Australia and Canada have such immigration policies (Borjas 1988). Because they have had a point system since the 1960s, I did not vary this dummy over time. Immigration country: I set up a dummy to indicate whether destinations were a traditional immigration country (Australia, Canada, and the United States) or not (the European countries). Presence of left-wing parties in the government: I rated the presence of left-wing parties in the government as 1 when they form a single party, 0.5 when they join a coalition and 0 when they are absent from the government. Information on the presence of left-wing parties in the government was obtained from Internet sources of the national governments and more general Internet sites for a number of countries (e.g., De Zárate 2003; Derksen 2003). Because coalitions change over time, and policy measures need some time to 223

224

Appendix B

become effective, I scored the presence of left-wing parties over a large time period, before the survey year. In the study of language (Chapter 7), I rated the presence of left-wing parties in the government in the 1980s because most surveys were conducted after that time. For the study of religion (Chapter 8), I did the same, except for the survey conducted in 1974, for which I computed averages for the 1970s. In the analysis of employment (Chapter 4) and occupational status (Chapter 5), I computed the presence of leftwing parties in the government 10 years preceding the survey year. Anti-immigrant attitudes: Data on prejudice towards immigrants were obtained from the first wave of the European Values Studies (Barker, Halman, and Vloet 1992) and World Values Studies (Inglehart et al. 2000), which took place in the period 1981-1984. I used the question “Who do you not want to have as neighbors?”, and computed the proportions responding “immigrants” for each country. Religious concentration: I measured religious competition and diversity with the Herfindahl index of religious concentration: Hj=∑P2ij, where P represents each religious family (or specific denomination) divided by the total number of church members in a country, i represents the index of summation that runs over all religious categories in country j. H equals the probability that any two randomly selected persons belong to the same religion. The index was multiplied by 100, and ranges (theoretically) from 0 to 100. The measure and the country scores were taken from Iannaccone (1991) and refer to the religious situation in the 1970s and the beginning of the 1980s. Religiosity of the host society: To measure the religiosity of the native population, I have used figures on religious attendance, obtained from the European and World Values Studies (WVS), conducted in the period 1981-1984 and 1990 (Barker, Halman, and Vloet 1992; Inglehart et al. 2000). I used the question “Apart from weddings, funerals and christenings, about how often do you attend religious services these days” and averaged the percentages that participate once a week or more for the two study periods. Labor-force activity natives: The percentage of all natives that is active in the labor market (employed or unemployed), per sex and per year. Information refers to persons between 25 and 54 years old. Data are obtained from IFIS.

Appendix B

225

Unemployment natives: The annual percent unemployed natives, per sex. Information refers to persons between 25 and 54 years old. Data are obtained from IFIS. ISEI natives: I included aggregate variables that controlled for differences in labor-market opportunities between countries and between time periods. I computed the annual ISEI of native males and females between 25 and 54 years old. Information is obtained from IFIS. Self-employment natives: This is the percent self-employed individuals among employed natives, measured per year. The variable was computed using aggregate information from IFIS, both varying per year and referring to native males between 25 and 54 years old. Country of Origin GDP per capita: I use Gross Domestic Product (GDP) per capita in constant US dollars in 1980 as a measure of economic development and modernization in the country of origin (OECD 2000). Globalization: I focus on trade globalization as a measure of the more general process of economic globalization. Trade globalization refers to the extent to which the long-distance and global exchange of commodities has increased (or decreased) relative to the exchange of commodities within national societies (Chase-Dunn, Kawano, and Brewer 2000). Following Chase-Dunn et al., I use the ratio of the value of national imports to GDP, in percentages, as an indicator of the level of trade globalization in the world-system. Information was obtained from the World Bank (2001). Political suppression: A rating of political rights and civil liberties, based on information collected by Freedom House (Karatnycky and Piano 2002). Political rights vary from 1 (e.g., free and fair elections, power for opposition), to 7 (e.g., oppressive regime, civil war). Civil liberties vary from 1 (e.g., freedom of expression and religion, free economic activity) to 7 (e.g., no religious freedom, political terror, and no free association). I used the sum score for each country (2–14), averaged over a longer year period. For the analysis on language proficiency, I used the 1981–1990 period. For the other analyses, I relied on the period 1972-1980. This measure is

226

Appendix B

also used as a proxy for religious suppression in the country of origin (Chapter 8). Christian origin: Because the destination countries examined are predominantly Christian, I include a dummy variable for origin countries that have a predominant Christian population. Those countries with more than 50 percent Christian adherents in the 1960–1980 period were assumed to be predominantly Christian. This information was obtained from Brierley (1997). White origin: I relied on the racial self-identification question in the one-percent file of the 1990 Census of the United States to obtain figures on the racial composition of countries. I used the proportion identifying as whites of all immigrants from a certain country as a measure of the proportion of whites in that country. Self-employment in origin country: I computed the percentage selfemployed of the total labor-force population in the country of origin. This information was obtained from the International Labour Office (ILO) and refers to various years over the period 1980-1990 (ILO 1980-1990). Female labor-force participation: I collected information on the average labor-force activity among females between the ages of 25 and 54 years in the countries of origin. Data refer to the period 1980–1986 and were obtained from the ILO Yearbook of Labour Statistics, various years (ILO 1981–1990). Community GDP per capita ratio: I used gross domestic product (GDP) per capita as a measure of economic development and calculated GDP ratios for the origin country relative to the destination country. The GDP was measured in constant dollars per capita for 1980 and was obtained from OECD (2000). GINI ratio: This variable measures the income inequality of the origin country relative to that of the destination country. To measure the income inequality in a country, I used the Gini formula. Information was obtained from a publication of the World Bank (2001), which gives Gini scores per country in the 1990s. Relative group size: A variable for the number of immigrants relative to the total population, expressed as a percentage. Information on this

Appendix B

227

dimension is not widely available for earlier times and for smaller groups. However, aggregating information from the International File of Immigration Surveys, supplemented with several other sources (e.g.,; Eurostat 2000; OECD 1994, 1996, 1998, 1999a, 2000, 2001), I managed to estimate averages for all groups in the 1980-1990 period. Official language: To construct this variable, I use information on official language (i.e., language used at school and in formal settings) and dominant language (i.e., language actively spoken by at least 40 percent of the population) (Grimes 2000). I constructed a dummy-variable indicating that the destination-language is the official (but not dominant) language in the origin country. Note that there are no origin countries in the data set of which the destination language was dominant but not official. In the analysis of language proficiency (Chapter 7), communities were excluded from the analysis for which the destination language was the official and dominant language in the origin country. This includes, for example, British immigrants in Australia, who are assumed to speak the destination language perfectly already before migration. Linguistic distance: For those immigrant groups of which the destination language was not official in their country of origin, the linguistic distance between the official origin and destination language was computed. Using the well-known classification of language-families (Finegan 1999; Grimes 2000), I constructed three dummy-variables: same language family and same language branch, same family but different branch, different family and different branch. Geographic distance: This variable is computed by the so-called “great circle distance method,” taking the capital cities as reference points (Byers 2002). The distance, measured in kilometers, is computed for every origin by destination combination, and in case the capital city is ambiguous for a certain country (e.g., former Yugoslavia), I use the main city in that region as a reference. Because for some of groups analyzed this measure will overestimate their actual travel distance, and for other groups the distance will be underestimated, the overall result will indicate the general impact of travel distance. Nevertheless, I also estimated models with modifications for groups for which my measure clearly overestimates the travel distance

228

Appendix B

(e.g., Cubans and Mexicans in the United States), but results are the same. Employment community: This variable indicates the percentage of the community that is employed in the labor market of all economically active members of that group (i.e., both employed and unemployed) who are between 25 and 54 years old. Average education community: I computed the mean educational level of immigrant communities, based on the three-category classification of education, ranging from low to high. Information is obtained by aggregating individual-level information from the International File of Immigration Surveys. Educational diversity: The heterogeneity of the education of the community was computed with the Herfindahl-index (see e.g., Iannaconne 1991), calculated by squaring the proportion of immigrants of each education in a community, and then summing the resulting numbers. Because education is coded in three categories at the individual level, the index ranges from [1.0 - (1.02 + 0.02 + 0.02) =] 0, or perfect concentration, to [1.0 - (0.332 + 0.332 + 0.332) =] 0.67, indicating maximum diversity. Sex imbalance: I constructed a variable measuring the balance between the number of males and the number of females in the community. The variable ranges from 0 (perfect male-female balance) to 1 (perfect imbalance). Length of stay community: I computed the percentage of an immigrant community that had stayed in the destination country more than 10 years. Survey I also included measures of surveys in the analysis, as a way to examine the role of survey effects. Respondent assessment: Whether the second-language ability of the respondent was assessed by the interviewer or by the respondent. Indirect questioning: Whether the question on language ability was direct or indirect (i.e., after an initial question about the languages the respondent speaks).

Appendix B

229

Migration survey: Indicates that the survey was specifically designed to study immigrants, in contrast to general population surveys. One-step question: I include a dummy variable representing the difference between a survey question that implicitly assumes that people have a religion (“What is your religion?”), in contrast to surveys in which the questions do not assume this (“Do you have a religion?”). Denomination: this refers to surveys in which respondents were asked for their religious denomination or membership. The reference category is people who have been asked to report their selfidentified religion or religious affiliation. Correlations Between Macro-Level Variables A potential problem in contextual analysis is high correlation between macro-level variables. In order to see if this applied to the present study, I calculated the bivariate Pearson correlations at the origin level, destination level and community level. It first of all appears that the correlations, when computed at the different levels, are very similar. Second, correlations are generally not higher than .40, and there are only a few relationships that correlate beyond .50. These are the association between political suppression in the country of origin and the GDP per capita ratio (r = –.59), the relationship between the presence of left-wing parties in the government and anti-immigrant prejudice (r = –.57), and between Christian origin and political suppression in the origin country (r = .51). Overall, there is no a priori reason to doubt the results on grounds of multicollinearity between the contextual variables. Finally, because some variables are at a nominal level, Pearson’s r is less adequate. Using other measures of association, such as Cramer’s V, I obtained similar results. Reference Van Tubergen, F. 2004b. The Country and Community File. Utrecht: ICS/Department of Sociology.

230

Appendix B

TREATMENT MISSING DATA I used a total of 25 macro-level variables in the analyses. For some variables, information was missing for a number of countries. Table B1 gives an overview of all variables for which information on some countries was missing. Table B1 shows that of the 25 macro-level variables, seven have missing cases for some countries. It also shows that in the Country and Community File (CCF), the GINI variable (i.e., a measure of income inequality in a country) has the most missing cases. Of the 208 countries included in CCF, I was unable to find GINI scores for 113 countries (54%). There are also many missing cases for the variables on the self-employment rate in the origin country (44%), and the laborforce activity rate of females in the origin country (42%). One option was to exclude all macro cases of which no information was available on at least one variable. I did not use this strategy, because the number of macro cases is already relatively low for hypotheses testing (see Chapter 3), and this solution would throw away the information that is available for these countries. I therefore used all countries for which individual-level data were available. There are two reasons that make the issue of missing data less problematic. First, not all macro-level variables with missing data in CCF are included in each chapter of this study. For example, the study on language proficiency contains only three variables with missing data. This is important, because the estimates in the analysis of a certain variable are less biased when the number of other variables with biased information is lower. Second, I do not analyze all the countries included in CCF. Whereas CCF contains information on 208 countries, the analysis of selfemployment, for example, includes 179 countries. Fortunately, the countries not included in the empirical chapters are also the countries with missing data (these are mainly smaller countries; for example Niue, Saint Helena, and Tokelau). Thus, the number of missing cases in the empirical analysis was lower than in CCF.

Table B1

Percentage of macro cases for which information is missing, by analysis (absolute number of missing cases in parentheses). CCF

Political suppression in country of origin % White in country of origin Globalization in country of origin Self-employment rate in country of origin Labor force activity rate females in country of origin GDP GINI Total number of macro cases in the analysis

Employment status 20 (38)

Occupational status 21 (38)

22

(46)

12

(25)

23

(48)

44

(93)

42

(88)

42

(78)

23

(48)

22

(41)

22

(40)

54

(113)

53

(100)

53

(96)

208

6

187

(10)

181

Selfemployment 18 (32) 4

Language

Religion

16

(30)

7

(10)

17

(31)

19

(34)

12

(17)

(7)

42

(75)

19

(34)

179

182

140

232

Appendix B

Nevertheless, the empirical chapters contained some macro-level variables for which no direct information was available for some countries. One solution used in social science research is to impute the mean value for all cases for which no direct information is available. I used this strategy with one step in between. I constructed 21 regions and computed the regional averages for the seven macro-level variables. Countries with missing data were assigned the scores of their region. Table B2 presents the regions, and the number of countries per region. Table B2 Regions Region Northern Africa Western Africa Eastern Africa Middle Africa Southern Africa

Number of countries per region 7 17 17 9 5

Northern America Central America South America Caribbean

5 8 14 24

Northern Europe Western Europe Eastern Europe Southern Europe

8 9 6 11

Western Asia Eastern Asia South-central Asia South-eastern Asia

15 8 9 11

Australia and New Zealand Melanesia Micronesia Polynesia

3 5 7 10

Appendix B

233

There are several reasons to feel confident about this solution. It appears that the seven macro-level variables with partly missing data generally have the predicted effect, even those with many missing cases. For example, the effect of the GINI variable in the analysis of employment status (Chapter 4) and occupational status (Chapter 5), is, as predicted, significantly negative –despite the fact that for more than 50% of the cases no direct information was available. Only one macrolevel variable with many (i.e., above 25%) missing cases does not concur with expectations: the self-employment rate in the country of origin does not significantly affect the self-employment rate of immigrants (Chapter 6). However, the self-employment rate of immigrants in their home country was also not significant in an earlier study which excluded all missing cases (Fairlie and Meyer 1996). Finally, it should be mentioned that I conducted several additional analyses in which dummy variables indicated whether information was directly obtained or estimated using averages for regions. These analyses did not show a significant effect of these dummy variables.

This page intentionally left blank

References

Alba, R. 1985. Italian Americans: Into the Twilight of Ethnicity. Englewood Cliffs: Prentice-Hall. ———. 1990. Ethnic Identity: The Transformation of White America. New Haven: Yale University Press. Alba, R. and V. Nee. 1999. “Rethinking Assimilation Theory for a New Era of Immigration.” Pp.137–60 in The Handbook of International Migration, edited by C. Hirschman, P. Kasinitz, and J. DeWind. New York: Sage. ———. 2003. Remaking the American Mainstream: Assimilation and Contemporary Immigration. Cambridge, MA: Harvard University Press. Aldrich, H.E., J. Cater, T. Jones, D. McEvoy, and P. Velleman. 1985. “Ethnic Residential Concentration and the Protected Market Hypothesis.” Social Forces 63:996–1009. Aldrich, H.E. and R. Waldinger. 1990. “Ethnicity and Entrepreneurship.” Annual Review of Sociology 16:111–35. Allport, G.W. 1954. The Nature of Prejudice. Cambridge: AddisonWesley Publishing Company. Antecol, H. 2000. “An Examination of Cross-Country Differences in the Gender Gap in Labor Force Participation Rates.” Labour Economics 7:409–26. Antecol, H, D.A. Cobb-Clark, and S.J. Trejo. 2003. “Immigration Policy and the Skills of Immigrants to Australia, Canada, and the United States.” Journal of Human Resources 38:192–218. Argue, A, D.R Johnson, and L.K White. 1999. “Age and Religiosity: Evidence from a Three-Wave Panel Analysis.” Journal for the Scientific Study of Religion 38:423–35. 235

236

References

Baker, M. and D. Benjamin. 1994. “The Performance of Immigrants in the Canadian Labor Market.” Journal of Labor Economics 12:369– 405. Barker, D., L. Halman, and A. Vloet. 1992. The European Values Study 1981-1990: Summary Report. London: The Gordon Cook Company. Basavarajappa, K.G. and F. Jones. 1999. “Visible Minority Income Differences.” Pp. 230–57 in Immigrant Canada: Demographic, Economic, and Social Challenges, edited by S.S.Halli and L. Driedger. Toronto: University of Toronto Press. Bates, T. 1997. Race, Self-Employment, and Upward Mobility: An Illusive American Dream. Washington, D.C.: Woodrow Wilson Center Press. Becker, G.S. 1964. Human Capital. New York: National Bureau of Economic Research. ———. 1976. The Economic Approach to Human Behavior. Chicago: The University of Chicago Press. ———. 1981. A Treatise on the Family (1st edition). Cambridge: Harvard University Press. Becker, G.S. and B.R. Chiswick. 1966. “Education and the Distribution of Earnings.” American Economic Review 56:358–59. Beenstock, M. 1996. “The Acquisition of Language Skills by Immigrants: The Case of Hebrew in Israel.” International Migration 34:3–30. Beenstock, M., B.R. Chiswick, and G.L. Repetto. 2001. “The Effect of Linguistic Distance and Country of Origin on Immigrant Language Skills: Application to Israel.” International Migration 39:33–60. Bell, B.D. 1997. “The Performance of Immigrants in the United Kingdom: Evidence from the GHS.” The Economic Journal 107:333–44. Berger, P. 1967. The Sacred Canopy. New York: Doubleday. Bibby, R.W. 1997. “Going, Going, Gone: The Impact of Geographical Mobility on Religious Involvement.” Review of Religious Research 38:289–307. Blalock, H.M. 1961. Causal Inference in Nonexperimental Research. Chapel Hill: University of North Carolina Press. ———. 1967. Toward a Theory of Minority Group Relations. New York: Wiley.

References

237

Blau, J.R. and P.M. Blau. 1982. “The Cost of Inequality: Metropolitan Structure and Violent Crime.” American Sociological Review 47:114–29. Blau, P.M. 1977. Inequality and Heterogeneity. New York: Free Press. ———. 1994. Structural Contexts of Opportunities. Chicago: University of Chicago Press. Blau, P.M., T.C. Blum, and J.E. Schwartz. 1982. “Heterogeneity and Intermarriage.” American Sociological Review 47:45–62. Blau, P.M. and O.D. Duncan. 1967. The American Occupational Structure. New York: Free Press. Blau, P.M. and J.E. Schwartz. 1984. Crosscutting Social Circles. Orlando: Academic Press. Blumer, H. 1958. “Race Prejudice as a Sense of Group Position.” Pacific Sociological Review 1:3–7. Bobo, L. 1983. “Whites’ Opposition to Busing: Symbolic Racism or Realistic Group Conflict?” Journal of Personality and Social Psychology 45:1196–1210. Bogardus, E.S. 1959. Social Distance. Los Angeles: Antioch Press. Bonacich, E. 1973. “A Theory of Middleman Minorities.” American Sociological Review 38:583–94. Borjas, G.J. 1985. “Assimilation, Changes in Cohort Quality and the Earnings of Immigrants.” Journal of Labor Economics 3:463–89. ———. 1986. “The Self-Employment Experience of Immigrants.” Journal of Human Resources 21:485–506. ———. 1987. “Self-Selection and the Earnings of Immigrants.” American Economic Review 77:531–33. ———. 1988. International Differences in the Labor Market Performance of Immigrants. Kalamazoo: W.E. Upjohn Institute for Employment Research. ———. 1992. “Ethnic Capital and Intergenerational Mobility.” The Quarterly Journal of Economics 107:123–50. ———. 1999. Heaven’s Door: Immigration Policy and the American Economy. Princeton: Princeton University Press. Borjas, G.J. and S.G. Bronars. 1989. “Consumer Discrimination and Self-Employment.” Journal of Political Economy 97:581–605. Bourdieu, P. 1980. “Le Capital Social: Notes Provisoires.” Actes de la Recherche en Sciences Sociales 31:2–3.

238

References

Boyd, R.L. 1991. “A Contextual Analysis of Black Self–Employment in Large Metropolitan Areas, 1970–1980.” Social Forces 70:409– 29. Bratsberg, B. and J.F. Ragan. 2002. The Impact of Host-Country Schooling on Earnings: A Study of Male Immigrants in the United States. Journal of Human Resources 37: 63-105. Brierley, P. 1997. World Churches Handbook. Exeter: BPC Wheaton Ltd. Brown, R. 1995. Prejudice: Its Social Psychology. Oxford: Blackwell Publishers. Browne, W.J. 2002. MCMC Estimation in MLwiN. Available on-line, 2002, at http://multilevel.ioe.ac.uk. Bruce, S. 1999. Choice and Religion: A Critique of Rational Choice. New York: Oxford University Press. Byers, J.A. 2002. Great Circle Distances Calculated between Points on Earth. Available on-line, 2002, at http://www.wcrl.ars.usda.gov/ cec/moregen.htm. Campbell, D.T. 1975. “‘Degrees of Freedom’ and the Case Study.” Comparative Political Studies 8:178–93. Campbell, R.A. and J.E. Curtis. 1994. “Religious Involvement across Societies: Analyses for Alternative Measures in National Surveys.” Journal for the Scientific Study of Religion 33:215–29. Carliner, G. 2000. “The Language Ability of U.S. Immigrants: Assimilation and Cohort Effects.” International Migration Review 34:158–82. Castles, S. and M.J. Miller. 2003. The Age of Migration: International Population Movements in the Modern World. New York: The Guilford Press. Chase-Dunn, C., Y. Kawano, and B.D. Brewer. 2000. “Trade Globalization Since 1795: Waves of Integration in the WorldSystem.” American Sociological Review 65:77–95. Chaves, M. 1991. “Family Structure and Protestant Church Attendance: The Sociological Basis of Cohort and Age Effects.” Journal for the Scientific Study of Religion 30: 501–14. Chaves, M. and P.S. Gorski. 2001. “Religious Pluralism and Religious Participation.” Annual Review of Sociology 27:261–81. Chaves, M., P.J. Schraeder, and M. Sprindys. 1994. “State Regulation of Religion and Muslim Religious Vitality in the Industrialized West.” Journal of Politics 56:1087–97.

References

239

Cheng, Y. 1994. Education and Class: Chinese in Britain and the United States. Aldershot: Avebury. Child, I.L. 1943. Italian or American? The Second Generation in Conflict. New Haven: Yale University Press. Chiswick, B.R. 1978. “The Effect of Americanization on the Earnings of Foreign-born Men.” Journal of Political Economy 86:897–921. ———. 1979. “The Economic Progress of Immigrants: Some Apparently Universal Patterns.” Pp. 357–99 in Contemporary Economic Problems, edited by W. Fellner. Washington DC: American Enterprise Institute. ———. 1982. The Employment of Immigrants in the United States. Washington DC: American Enterprise Institute. ———. 1991. “Speaking, Reading, and Earnings among Low-Skilled Immigrants.” Journal of Labor Economics 9:149–70. ———. 1999. “Are Immigrants Favorably Self-Selected?” American Economic Review 89:181–85. Chiswick, B.R., Y. Cohen, and T. Zach. 1997. “The Labor Market Status of Immigrants: Effects of the Unemployment Rate at Arrival and Duration of Residence.” Industrial and Labor Relations Review 50:289–303. Chiswick, B.R. and M.E. Hurst. 2000. “The Employment, Unemployment, and Unemployment Compensation Benefits of Immigrants.” Pp. 87–115 in Long-Term Unemployment and Reemployment Policies, edited by L.J. Bassi and S.A. Woodbury. Stamford: JAI Press. Chiswick, B.R. and P.W. Miller. 1992. “Language in the Immigrant Labor Market.” Pp. 229–296 in Immigration, Language, and Ethnicity: Canada and the United States, edited by : B.R. Chiswick.. Washington D.C.: American Enterprise Institute. ———. 1994. “Language Choice among Immigrants in a Multi-lingual Destination.” Journal of Population Economics 7:119–31. ———. 1995. “The Endogeneity between Language and Earnings: International Analyses.” Journal of Labor Economics 13:246–88. ———. 1996. “Ethnic Networks and Language Proficiency among Immigrants.” Journal of Population Studies 9:19–35. ———. 1998. “Language Skill Definition: A Study of Legalized Aliens.” International Migration Review 32:877–900.

240

References

———. 2001. “A Model of Destination-Language Acquisition: Application to Male Immigrants in Canada.” Demography 38:391– 409. ———. 2002. “Immigrant Earnings: Language Skills, Linguistic Concentrations and the Business Cycle.” Journal of Population Economics 15:31–57. Clark, K. and S. Drinkwater. 2000. ”Pushed out or Pulled in? SelfEmployment among Ethnic Minorities in England and Wales.” Labour Economics 7:603–28. Clyne, M. 1991. Community Languages: The Australian Experience. Cambridge: Cambridge University Press. Cobb-Clark, D.A. 1993. “Immigrant Selectivity and Wages: The Evidence of Women.” American Economic Review 83:986–93. Coenders, M. 2001. Nationalistic Attitudes and Ethnic Exclusionism in a Comparative Perspective: An Empirical Study of Attitudes toward the Country and Ethnic Immigrants in 22 Countries. Ph.D. thesis: University of Nijmegen. Coleman, J.S. 1988. “Social Capital in the Creation of Human Capital.” American Journal of Sociology 94:S95–S121. ———. 1990. Foundations of Social Theory. Cambridge: Harvard University Press. Cretser, J.A. 1999. “Cross-National Marriage in Sweden: Immigration and Assimilation.” Journal of Comparative Family Studies 30:363– 80. Crul, M. and H. Vermeulen. 2003. “The Second Generation in Europe.” International Migration Review 37:965–86. Dagevos, J., M. Gijsberts and C. van Praag (eds.). 2003. Rapportage minderheden 2003. Den Haag: Sociaal en Cultureel Planbureau. Davis, J.A., T.W. Smith, and P.V. Marsden. 2000. General Social Surveys, 1972–2000: Cumulative Codebook. Chicago: National Opinion Research Center. Derksen, W. 2003. Election World. Available on-line, 2003, at the internet: http://www.electionworld.org/index.html. De Graaf, N.D. and A. Need. 2000. “Losing Faith: Is Britain Alone?” Pp.119–36 in British Social Attitudes: The 17th report: Focusing on Diversity, edited by R. Jowell., J. Curtice, A. Park, K. Thomson, L. Jarvis, C. Bromley, and N. Stratford. London: Sage.

References

241

De Vaus, D. and I. McAllister. 1987. “Gender Differences in Religion: A Test of the Structural Location Theory.” American Sociological Review 52:472–81. De Zárate, R.O. 2003. Zárate’s Political Collections. Available on-line, 2003, at: http://www.terra.es/personal2/monolith. Duckitt, J. 1992. The Social Psychology of Prejudice. New York: Praeger Publishers Duleep, H.O. and M.C. Regets. 1992. “Some Evidence on the Effects of Admissions Criteria on Immigrant Assimilation.” Pp. 410–39 in Immigration, Language and Ethnic Issues: Canada and the United States, edited by B.R. Chiswick. Washington DC: American Enterprise Institute. Duleep, H.O. and S. Sanders. 1993. “The Decision to Work by Married Immigrant Women.” Industrial and Labor Relations Review 46:677–80. Durkheim, E. [1897] 1961. Suicide. New York: Free Press. Dustmann, C. 1994. “Speaking Fluency, Writing Fluency and Earnings of Migrants.” Journal of Population Economics 7:133–56. ———. 1997. “The Effects of Education, Parental Background and Ethnic Concentration on Language.” Quarterly Review of Economics and Finance 37:245–62. Ebaugh, H.R. and J.S. Chafetz. 2000. Religion and the New Immigrants: Continuities and Adaptations in Immigrants Congregations. Walnut Creek: AltaMira Press. Espenshade, T.J. and C.A. Calhoun. 1993. “An Analysis of Public Opinion toward Undocumented Immigration.” Population Research and Policy Review 12:189–224. Espenshade, T.J. and H. Fu. 1997. “An Analysis of English–Language Proficiency among U.S.-Immigrants.” American Sociological Review 62:288–305. Espinosa, K.E. and D.S. Massey. 1997. “Determinants of English Proficiency among Mexican Migrants to the United States.” International Migration Review 31:28–50. Eurostat. 1998. The European Union Labour Force Survey: Methods and Definitions 1998. Luxembourg: Eurostat. ———. 2000. European Social Statistics: Migration. Luxembourg: Eurostat. Evans, M.D.R. 1986. “Sources of Immigrants’ Language Proficiency: Australian Results with Comparisons to the Federal Republic of

242

References

Germany and the United States of America.” European Sociological Review 2:226–36. ———. 1989. “Immigrant Entrepreneurship: Effects of Ethnic Market Size and Isolated Labor Pool.” American Sociological Review 54:950–62. Evans, M.D.R. and J. Kelley. 1991. “Prejudice, Discrimination, and the Labor Market: Attainments of Immigrants in Australia.” American Journal of Sociology 97:721–59. Extra, G. and L. Verhoeven. 1993. Immigrant Languages in Europe. Clevedon: Multilingual Matters. Fairlie, R.W. and B.D. Meyer. 1996. “Ethnic and Racial Selfemployment Differences and Possible Explanations.” Journal of Human Resources 31:757–93. Fassman, H. and R. Münz. 1994. European Migration in the Late Twentieth Century: Historical Patterns, Actual Trends, and Social Implications. Aldershot: Edward Elgar. Fetzer, J.S. 2000. Public Attitudes toward Immigration in the United States, France, and Germany. Cambridge: Cambridge University Press. Finegan, E. 1999. Language: Its Structure and Use (3rd edition). New York: Harcourt Brace. Firebaugh, G. and B. Harley. 1991. “Trends in US Church Attendance: Secularization and Revival or Merely Lifecycle Effects?” Journal for the Scientific Study of Religion 30: 487-500. Flap, H., A. Kumcu, and B. Bulder. 2000. “The Social Capital of Ethnic Entrepreneurs and their Business Success.” Pp.142–161 in Immigrant Business: The Economic, Political, and Social Environment, edited by J. Rath. Houndmills: MacMillan. Fossett, M.A. and K.J. Kiecolt, 1989. “The Relative Size of Minority Populations and White Racial Attitudes.” Social Science Quarterly 70:820–35. Friedberg, R. M. 2000. “You Can’t Take It with You? Immigrant Assimilation and the Portability of Human Capital.” Journal of Labor Economics 18: 221-52. Frisbie, W.P and L.J. Neidert. 1977. “Inequality and the Relative Size of Minority Populations: A Comparative Analysis.” American Journal of Sociology 82:1007–30. Galarza, E. 1964. Merchants of Labor: The Mexican Bracero Story. Charlotte: McNally and Loftin.

References

243

Gans, H.J. 1962. The Urban Villagers: Group and Class in the Life of Italian Americans. Chicago: University of Chicago Press. Ganzeboom, H.B.G. 2003. Tools for Standardizing Occupation Codes into ISCO-88. Available on-line, 2003, at: http://home.scw.vu.nl/%7Eganzeboom/occisco/index.htm. Ganzeboom, H.B.G., P.M. De Graaf, and D.J. Treiman. 1992. “A Standard International Socio-Economic Index of Occupational Status.” Social Science Research 21:1–56. Ganzeboom, H.B.G. and D.J. Treiman. 1996. “Internationally Comparable Measures of Occupational Status for the 1988 International Standard Classification of Occupations.” Social Science Research 25:201–39. ———. 2003. Conversion Tools ISCO-88 into ISEI. Available on-line, 2003, at http://home.scw.vu.nl/%7Eganzeboom/occisco/index.htm. Glazer, N. and D.P. Moynihan. 1970. Beyond the Melting Pot: The Negroes, Puerto Ricans, Jews, Italians, and Irish of New York City. Cambridge: Harvard University Press. Goldthorpe, J.H. 1997. “Current Issues in Comparative Macrosociology: A Debate on Methodological Issues.” Comparative Social Research 16:1–26. ———. 2000. On Sociology: Numbers, Narratives and the Integration of Research and Theory. Oxford: Oxford University Press. Gordon, M.M. 1964. Assimilation in American Life. New York: Oxford University Press. Granovetter, M.S. 1973. “Strength of Weak Ties.” American Sociological Review 78:1360–80. Grimes, B.F. (ed.). 2000. Ethnologue: Languages of the World (14th edition). Dallas: SIL International. Hayfron, J.E. 2001. “Language Training, Language Proficiency and Earnings of Immigrants in Norway.” Applied Economics 33:1971– 79. Heath, A. (ed.). 2006. Ethnic Minority Disadvantage: Comparative Perspectives. Oxford: Oxford University Press. Hewitt, C. 1977. “The Effect of Political Democracy and Social Democracy on Equality in Industrial Societies: A Cross-National Comparison.” American Sociological Review 42:450–64. Hodge, R. 1973. “Toward a Theory of Racial Differences in Employment.” Social Forces 52:16–31.

244

References

Horrie, C. and P. Chippindale. 1990. What is Islam? London: Allen and Co. Hout, M. and A. Greeley. 1990. “The Cohort Doesn´t Hold: Comment on Chaves (1989).” Journal for the Scientific Study of Religion 29: 519–24. Hout, M. and H. Rosen. 2000. “Self-Employment, Family Background, and Race.” The Journal of Human Resources 35:670–92. Huber, E., C.C. Ragin, and J.D. Stephens. 1993. “Social Democracy, Christian Democracy, Constitutional Structure and the Welfare State.” American Journal of Sociology 99:711–49. Hulsen, M. 2000. Language Loss and Language Processing: Three Generations of Dutch Migrants in New Zealand. Ph.D. thesis: University of Nijmegen. Hurh, W.M. and K.C Kim. 1990. “Religious Participation of Korean Immigrants in the United States.” Journal for the Scientific Study of Religion 29:19–34. Hwang, S., R. Saenz, and B. Aguirre. 1997. “Structural and Assimilationist Explanations of Asian American Intermarriage.” Journal of Marriage and the Family 59:758–72. Iannaccone, L.R. 1990. “Religious Practice: A Human Capital Approach.” Journal for the Scientific Study of Religion 29:297–314. ———. 1991. “The Consequences of Religious Market Structure: Adam Smith and the Economics of Religion.” Rationality and Society 3:156–77. ILO. 1980–1990 (various years). Yearbook of Labour Statistics. Geneva: International Labour Organization. ———. 1990. International Standard Classification of Occupations: ISCO–88. Geneva: International Labour Organization. Inglehart, R. et al. 2000. World Values Surveys and European Values Surveys, 1981-1984, 1990-1993, and 1995-1997 [Data file]. Produced by the Institute for Social Research. Distributed by Interuniversity Consortium for Political and Social Research, Ann Arbor, USA. Jasso, G. and M.R Rosenzweig. 1990a. The New Chosen People: Immigrants in the United States. New York: Russell Sage Foundation. ———. 1990b. “Self Selection and the Earnings of Immigrants. Comment.” American Economic Review 80:298–304.

References

245

Jasso, G., D.S. Massey, M.R. Rosenzweig, and J.P. Smith. 2003. “The New Immigrant Survey in the US: The Experience over Time.” Available on-line, 2003, at www.migrationinformation.org. Jelen, T.G. 2002. “Reflections on the ‘New Paradigm’: Unfinished Business and an Agenda for Research.” Pp. 187–203 in Sacred Markets, Sacred Canopies: Essays on Religious Markets and Religious Pluralism, edited by T.G. Jelen. Lanham: Littlefield Publishers. Jones, F.L. and R. Luijkx. 1996. “Post-War Patterns of Intermarriage in Australia: The Mediterranean Experience.” European Sociological Review 12:67–86. Kalmijn, M. 1993. “Spouse Selection among the Children of European Immigrants: A Comparison of Marriage Cohorts in the 1960 Census.” International Migration Review 27:51–78. ———. 1996. “The Socioeconomic Assimilation of Caribbean American Blacks.” Social Forces 74:911–30. Karatnycky, A. and A. Piano (eds.). 2002. Freedom in the World: The Annual Survey of Political Rights and Civil Liberties, 2001–2002. New Jersey: Transaction Publishers. Kelley, J. and N.D. De Graaf. 1997. “National Context, Parental Socialization, and Religious Belief: Results From 15 Nations.” American Sociological Review 62:639–59. Kelley, J. and I. McAllister. 1984. “Immigrants, Socio-Economic Attainment, and Politics in Australia.” The British Journal of Sociology 25:387–405. Kennedy, R.J.R. 1944. “Single or Triple Melting Pot? Intermarriage Trends in New Haven, 1870–1940.” American Journal of Sociology 49:331–39. ———. 1952. “Single or Triple Melting Pot? Intermarriage Trends in New Haven, 1870–1950.” American Journal of Sociology 58:56– 59. Kim, K.C., and W.M. Hurh. 1985. “Ethnic Resources Utilization of Korean Immigrant Entrepreneurs in the Chicago Minority Area.” International Migration Review 19:82–111. Kim, K.C., W.M. Hurh, and M. Fernandez. 1989. “Intra-group Differences in Business Participation: Three Asian Immigrant Groups.” International Migration Review 23:73–95.

246

References

Kiser, E. and M. Hechter. 1991. “The Role of General Theory in Comparative-Historical Sociology.” American Journal of Sociology 97:1–30. Kloosterman, R. and J. Rath (eds.). 2001. “Immigrant Entrepreneurs in Advanced Economies: Mixed Embeddedness Further Explored.” Journal of Ethnic and Migration Studies 27:189–202. ———. 2003. Immigrant Entrepreneurs: Venturing Abroad in the Age of Globalization. Oxford: Berg Publishers. Kogan, I. 2003. “Ex-Yugoslavs in the Austrian and Swedish Labour Markets: The Significance of the Period of Migration and the Effect of Citizenship Acquisition.” Journal of Ethnic and Migration Studies 29:595–622. Koopmans, R. and P. Statham. 1999. “Challenging the Liberal NationState? Postnationalism, Multiculturalism, and the Collective Claims Making of Migrants and Ethnic Minorities in Britain and Germany.” American Journal of Sociology 105:652–96. Kossoudji, S.A. 1988. “English Language Ability and the Labor Market Opportunities of Hispanic and East Asian Immigrant Men.” Journal of Labor Economics 6:205–28. Kreft, I. and J. De Leeuw. 1998. Introducing Multilevel Modeling. London: Sage Publications. Labov, R. and J.A. Jacobs. 1986. “Intermarriage in Hawaii, 1950– 1983.” Journal of Marriage and the Family 48:79–88. Lambert, P. 2003. Conversion Tools Occupations US 1990 Census into ISCO–88. Available on-line, 2003, at: http://www.cf.ac.uk/socsi/CAMSIS/occunits/distribution. Langlois, A. and E. Razin. 1995. “Self-Employment among FrenchCanadians: The Role of the Regional Milieu.” Ethnic and Racial Studies 18:581–604. Lenski, G., J. Lenski, and P. Nolan. 1991. Human Societies: An Introduction to Macrosociology. New York: McGraw-Hill. Lesthaeghe, R. (ed.). 2000. Communities and Generations: Turkish and Moroccan Populations in Belgium. Brussels: VUB University Press. Le Vine, R.A. and D.T. Campbell. 1972. Ethnocentrism: Theories of Conflict, Ethnic Attitudes, and Group Behavior. New York: John Wiley and Sons.

References

247

Li, P. 1977. “Occupational Achievement and Kinship Assistance Among Chinese Immigrants in Chicago.” Sociological Quarterly 18:478–89. Lieberson, S. 1970. Language and Ethnic Relations in Canada. New York: John Wiley. ———. 1980. A Piece of the Pie: Blacks and White Immigrants Since 1880. Berkeley: University of California Press. ———. 1991. “Small N’s and Big Conclusions: An Examination of the Reasoning in Comparative Studies Based on a Small Number of Cases.” Social Forces 70:307–20. ———. 1994. “More on the Uneasy Case for Using Mill-Type Methods in Small-N Comparative Studies.” Social Forces 72:1225– 37. Lieberson, S. and M. Waters. 1988. From Many Strands: Ethnic and Racial Groups in Contemporary America. New York: Russell Sage Foundation. Lievens, J. 1998. “Interethnic Marriage: Brining in the Context through Multilevel Modelling.” European Journal of Population 14:117–55. Light, I. 1979. “Disadvantaged Minorities in Self-employment.” International Journal of Comparative Sociology 20:31–45. ———. 1994. “The Ethnic Economy.” Pp. 647–71 in The Handbook of Economic Sociology, edited by N.J. Smelser and R. Swedberg. Princeton: Princeton University Press. Light, I. and E. Bonacich. 1988. Immigrant Entrepreneurs: Koreans in Los Angeles, 1965–1982. Berkeley: University of California Press. Light, I. and C. Rosenstein. 1995. “Expanding the Interaction Theory of Entrepreneurship.” Pp. 166–212 in The Economic Sociology of Immigration, edited by A. Portes. New York: Russell Sage Foundation. Lindley, J. 2002. “Race or Religion? The Impact of Religion on the Employment and Earnings of Britain’s Ethnic Communities.” Journal of Ethnic and Migration Studies 28:427–42. Lofstrom, M. 2002. “Labor Market Assimilation and the SelfEmployment Decision of Immigrant Entrepreneurs.” Journal of Population Economics 15:83–114. Logan, J.R. and R.D. Alba. 1993. “Locational Returns to Human Capital: Minority Access to Suburban Community Resources.” Demography 30:243–68.

248

References

Longva, P. and O. Raaum. 2003. “Earnings Assimilation of Immigrants in Norway: A Reappraisal.” Journal of Population Economics 16:177–93. Lopez, D.E. 1996. “Language: Diversity and Assimilation.” Pp. 139–63 in Ethnic Los Angeles, edited by R. Waldinger and M. Bozorgmehr. New York: Russell Sage. Lubbers, M. 2001. Exclusionistic Electorates: Extreme Right-Wing Voting in Western Europe. Ph.D. thesis: University of Nijmegen. Luo, S. and R.L. Wiseman. 2000. “Ethnic Language Maintainance among Chinese Immigrant Children in the United States.” International Journal of Incultural Relations 24:307–24. Marsella, A.J., T. Bornemann, S. Ekblad, and J. Orley. (eds.). 1994. Amidst Peril and Pain. The Mental Health and Well-Being of the World’s Refugees. Washington DC: America Psychological Association. Marshall, P. 2000. Religious Freedom: A Global Report on Freedom and Persecution. Nashville: Broadman and Holman Publishers. Martin, P.L. 1991. The Unfinished Story: Turkish Labour Migration to Western Europe. Geneva: International Labour Office. Massey, D.S. 1981. “Dimensions of the New Immigration to the United States and the Prospects for Assimilation.” Annual Review of Sociology 7:57–85. Massey, D.S., J. Arango, G. Hugo, A. Kouaouci, A. Pellegrino, and J.E. Taylor. 1993. “Theories of International Migration: A Review and Appraisal.” Population and Development Review 19:431–66. Miller, A.S. and J.P. Hoffman. 1995. “Risk and Religion: An Explanation of Gender Differences in Religiosity.” Journal for the Scientific Study of Religion 34:63–75. Min, P.G. and M. Bozorgmehr. 2003. “United States: The Entrepreneurial Cutting Edge.” Pp. 17–37 in Immigrant Entrepreneurs: Venturing Abroad in the Age of Globalization, edited by R. Kloosterman and J. Rath. Oxford: Berg Publishers. Mincer, J. 1958. “Investment in Human Capital and Personal Income Distribution.” Journal of Political Economy 66:281–302. Model, S. 1997. “An Occupational Tale of Two Cities: Minorities in London and New York.” Demography 34:539–50. Model, S., G. Fisher, and R. Silberman. 1999. “Black Caribbeans in Comparative Perspective.” Journal of Ethnic and Migration Studies 25:187–212.

References

249

Model, S. and D. Lapido. 1996. “Context and Opportunity: Minorities in London and New York.” Social Forces 75:485–510. Model, S. and L. Lin. 2002. “The Cost of Not Being Christian: Hindus, Sikhs and Muslims in Britain and Canada.” International Migration Review 36:1061–92. Modood, T., R. Berthoud, J. Lakey, J. Nazroo, P. Smith, S. Virdee, and S. Beishnon. 1997. Ethnic Minorities in Britain. London: Policy Studies Institute. Myers, S.M. 1996. “An Interactive Model of Religiosity Inheritance: The Importance of Family Context.” American Sociological Review 61:858–66. Nee, V., J.M. Sanders, and S. Sernau. 1994. “Job Transitions in an Immigrant Metropolis: Ethnic Boundaries and the Mixed Economy.” American Sociological Review 59:849–72. Need, A. and N.D. De Graaf. 1996. “‘Losing my Religion’: A Dynamic Analysis of Leaving the Church in the Netherlands.” European Sociological Review 12:87–99. Neidert, L.J. and R. Farley. 1985. “Assimilation in the United States: An Analysis of Ethnic and Generational Differences in Status and Achievement.” American Sociological Review 50:840–50. Nelsen, H.M. and H.D. Allen. 1974. “Ethnicity, Americanization, and Religious Attendance.” American Journal of Sociology 79:906–22. OECD. 1994. Sopemi: Trends in International Migration, Annual Report. Paris: Organisation for Economic Co-operation and Development. ———. 1996. Sopemi: Trends in International Migration, Annual Report. Paris: Organisation for Economic Co-operation and Development. ———. 1998. Sopemi: Trends in International Migration, Annual Report. Paris: Organisation for Economic Co-operation and Development. ———. 1999a. Sopemi: Trends in International Migration, Annual Report. Paris: Organisation for Economic Co-operation and Development. ———. 1999b Classifying Educational Programmes: Manual for ISCED–97 Implementation in OECD Countries. Paris: Organisation for Economic Co-operation and Development. ———. 2000. International Migration Statistics. Paris: Organisation for Economic Co-operation and Development.

250

References

———. 2001. Sopemi: Trends in International Migration, Annual Report. Paris: Organisation for Economic Co-operation and Development. Owen, C., H. Eisner, and T. McFaul. 1981. “A Half-Century of Social Distance Research: National Replication of the Bogardus Studies.” Sociology and Social Research 66:80–97. Pagnini, D.L. and P.S. Morgan. 1990. “Intermarriage and Social Distance among U.S. Immigrants at the Turn of the Century.” American Journal of Sociology 96:405–32. Park, R. and E. Burgess. [1921] 1969. Introduction to the Science of Sociology (3rd edition). Chicago: University of Chicago Press. Perlmann, J. 1988. Ethnic Differences: Schooling and Social Structure Among the Irish, Italians, Jews, and Blacks in an American City, 1888–1935. New York: Cambridge University Press. Pineo, P.C. 1977. “The Social Standing of Ethnic and Racial Groupings.” Canadian Review of Sociology and Anthropology 14:147–57. Popper, K.R. 1972. Objective Knowledge: An Evolutionary Approach. Oxford: Clarendon Press. Portes, A. (ed.). 1996. The New Second Generation. New York: Russell Sage Foundation. ———. 1998. “Social Capital: Its Origins and Applications in Modern Sociology.” Annual Review of Sociology 24:1–24. Portes, A. and R.L. Bach. 1985. Latin Journey: Cuban and Mexican Immigrants in the United States. Berkeley: University of California Press. Portes, A. and L. Jensen. 1987. “What’s an Ethnic Enclave? The Case for Conceptual Clarity –Comment on Sanders and Nee.” American Sociological Review 52:768–71. Portes, A. and R.G. Rumbaut. 1996. Immigrant America: A Portrait (2nd edition). Berkeley: University of California Press. ———. (ed.). 2001. Legacies: The Story of the Immigrant Second Generation. Berkeley: University of California Press. Portes, A. and J. Sensenbrenner. 1993. “Embeddedness and Immigration: Notes on the Social Determinants of Economic Action.” American Journal of Sociology 98:1320–50. Przeworski, A. 1987. “Methods of Cross-National Research, 1970–83: An Overview.” Pp. 31–49 in Comparative Policy Research, edited

References

251

by M. Dierkes, H.N. Weiler, and A.B. Antal. Berlin: WZBPublications. Przeworski, A. and H. Teune. 1970. The Logic of Comparative Social Inquiry. New York: Wiley. Qian, Z. and D.T Lichter. 2001. “Measuring Marital Assimilation: Intermarriage among Natives and Immigrants.” Social Science Research 30:289–312. Quillian, L. 1995. “Prejudice as a Response to Perceived Group Threat: Population Composition and Anti-Immigrant and Racial Prejudice in Europe.” American Sociological Review 60:586–611. Ragin, C.C. 1987. The Comparative Method. Berkeley: University of California Press. Raijman, R. and M. Semyonov. 1995. “Modes of Labor Market Incorporation and Occupational Cost among New Immigrants to Israel.” International Migration Review 29:375–93. Raijman, R. and M. Tienda. 1999a. “Immigrants’ Pathways to Business Ownership: A Comparative Ethnic Perspective.” International Migration Review 34:682–706. ———. 1999b. “Immigrants’ Socioeconomic Progress Post-1965: Forging Mobility or Survival?” Pp. 239–56 in The Handbook of International Migration: The American Experience, edited by C. Hirschman, P. Kasinitz and J. DeWind. New York: Russell Sage Foundation. Rath, J. (ed.). 2002. Unraveling the Rag Trade: Immigrant Entrepreneurship in Seven World Cities. Oxford: Berg Publishers. Raudenbush, S.W. and A.S. Bryk. 2002. Hierarchical Linear Models: Applications and Data Analysis Methods (2nd edition). Newbury Park: Sage. Razin, E. and A. Langlois. 1996. “Metropolitan Characteristics and Entrepreneurship among Immigrants and Ethnic Groups in Canada.” International Migration Review 30:703–27. Reitz, J.G. 1998. Warmth of the Welcome: The Social Causes of Economic Success for Immigrants in Different Nations and Cities. Boulder: Westview Press. ———. (ed.). 2003. Host Societies and the Reception of Immigrants. San Diego: Center for Comparative Immigration Research. Renzulli, L.A., H. Aldrich, and J. Moody. 2000. “Family Matters: Gender, Networks, and Entrepreneurial Outcomes.” Social Forces 79:523–46.

252

References

Roodenburg, H., R. Euwals, and H. Ter Rele. 2003. Immigration and the Dutch Economy. Den Haag: Centraal Planbureau. Rosenfeld, M.J. 2002. “Measures of Assimilation in the Marriage Market: Mexican American 1970-1990.” Journal of Marriage and Family 64:152–62. Roy, A. 1951. “Some Thoughts on the Distribution of Earnings.” Oxford Economic Papers 3:135–46. Rumbaut, R.G. and A. Portes. 2001. Ethnicities: Children of Immigrants in America. Berkeley: University of California Press. Sampson, R.J. 1984. “Group Size, Heterogeneity, and Intergroup Conflict.” Social Forces 62:618–39. Sanders, J.M., and V. Nee. 1987. “Limits of Ethnic Solidarity in the Enclave Economy.” American Sociological Review 52:745–73. ———. 1996. “Immigrants Self–Employment: The Family as Social Capital and the Value of Human Capital.” American Sociological Review 61:231–49. Scheepers, P., M. Gijsberts, and M. Coenders. 2002. “Ethnic Exclusionism in European Countries: Public Opposition to Civil Rights for Legal Migrants as a Response to Perceived Ethnic Threat.” European Sociological Review 18:17–34. SCP, WODC, and CBS. 2005. Jaarrapport integratie 2005. Den Haag: Sociaal en Cultureel Planbureau. Semyonov, M. 1988. “Bi-ethnic Labor Markets, Mono-ethnic Labor Markets, and Socioeconomic Inequality.” American Sociological Review 53:256–66. Sherkat, D.E. and J. Wilson. 1995. “Preferences, Constraints, and Choices in Religious Markets: An Examination of Religious Switching and Apostasy.” Social Forces 73:993–1026. Shields, M.A. and S.W. Price. 2002. “The English Language Fluency and Occupational Success of Ethnic Minority Immigrant Men Living in English Metropolitan Areas.” Journal of Population Economics 15:137–60. Skocpol, T. and M. Somers. 1980. “The Uses of Comparative History in Macrosocial Inquiry.” Comparative Studies in Society and History 22:174–97. Smith, C., D. Sikkink, and J. Bailey. 1998. “Devotion in Dixie and Beyond: A Test of the ‘Shibley Thesis’ on the Effects of Regional Origin and Migration on Individual Religiosity.” Journal for the Scientific Study of Religion 37:494–506.

References

253

Smith, T. 2002. “Religious Diversity in America: The Emergence of Muslims, Buddhists, Hindus, and Others.” Journal for the Scientific Study of Religion 41:577–85. Smits, J., W. Ultee, and J. Lammers. 1998. “Educational Homogamy in 65 Countries: An Explanation of Differences in Openness Using Country–Level Explanatory Variables.” American Sociological Review 63:264–85. Snijders, T. and R. Bosker. 1999. Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling. London: Sage Publications. Solé, Y.R. 1990. “Bilingualism: Stable or Transitional? The Case of Spanish in the United States.” International Journal of the Sociology of Language 84:35–80. South, S.J. and S.F. Messner. 1986. “Structural Determinants of Intergroup Associations.” American Journal of Sociology 91:1409– 30. Sowell, T. 1996. Migrations and Cultures: A World View. New York: Basic Books. Stark, R. and W. S. Bainbridge. 1987. A Theory of Religion. New York: Peter Lang Publishing. Stark, R. and R. Finke. 2000. Acts of Faith: Explaining the Human Side of Religion. Berkeley: University of California Press. Stevens, G. 1985. “Nativity, Intermarriage, and Mother-Tongue Shift.” American Sociological Review 50:74–83. ———. 1986. “Sex Differences in Language Shift in the United States.” Sociology and Social Research 71:31–36. ———. 1992. “The Social and Demographic Context of Language Use in the United States.” American Sociological Review 57:171–85. ———. 1999. “Age at Immigration and Second Language Proficiency among Foreign-born Adults.” Language in Society 28:555–78. Stump, R.W. 1984. “Regional Migration and Religious Commitment in the United States.” Journal for the Scientific Study of Religion 23:292–03. Taylor, M.C. 1998. “How White Attitudes Vary with the Racial Composition of Local Populations: Numbers Count.” American Sociological Review 63:512–35. Te Grotenhuis, M. and P. Scheepers. 2001. “Churches in Dutch: Causes of Religious Disaffiliation in the Netherlands, 1937–1995.” Journal for the Scientific Study of Religion 40:591–606.

254

References

Tesser, P.T.M., J.G.M. Merens, and C.S. Van Praag. 1999. Rapportage Minderheden 1999. Den Haag: Sociaal en Cultureel Planbureau. Thomas, W.I. and F. Znaniecki. 1958 [1918–1920]. The Polish Peasant in Europe and America. New York: Dover. Thornton, A., W.G. Axinn, and D.H. Hill. 1992. “Reciprocal Effects of Religiosity, Cohabitation, and Marriage.” American Journal of Sociology 98:928–51. Thurow, L.C. 1975. Generating Inequality: The Distributional Mechanisms of the Economy. Basic Books. Tienda, M. and D.T. Lii. 1987. “Minority Concentration and Earnings Inequality: Blacks, Hispanics, and Asians Compared.” American Journal of Sociology 93:141–65. Tolnay, S.E. 2001. “African Americans and Immigrants in Northern Cities: The Effects of Relative Group Size on Occupational Standing in 1920.” Social Forces 80:573–604. Treiman, D.J. and H.B.G. Ganzeboom. 1990. “Cross National Comparative Status Attainment Research.” Research in Social Stratification and Mobility 9:33–78. Tribalat, M. 1995. Faire France: Une Enquête sur les Immigrés et Leurs Enfants. Paris: La Découverte. Turner, B.S. 1983. Religion and Social Theory: A Materialist Perspective. London: Heinemann. Van den Berghe, P. 1967. Race and Racism: A Comparative Perspective. New York: Wiley. Van Tubergen, F. 2005. “Self-Employment of Immigrants: A CrossNational Study of 17 Western Societies.” Social Forces 84:709-32. ———. 2006a. “Occupational Status of Immigrants in Cross-National Perspective: A Multilevel Analysis of 17 Western Countries.” Immigration and the Transformation of Europe, edited by G. Parsons and T. Smeeding. Cambridge: Cambridge University Press. ———. 2006b. “Religious Affiliation and Participation among Immigrants in Eight Western Countries: A Cross-National Study of Individual and Contextual Effects.” Journal for the Scientific Study of Religion 45:1-22. ———. 2007. “Religious Affiliation and Participation among Immigrants in a Secular Society: A Study of Immigrants in the Netherlands.” Forthcoming in Journal of Ethnic and Migration Studies.

References

255

Van Tubergen, F. and M. Kalmijn. 2005. “Destination-Language Proficiency in Cross-National Perspective: A Study of Immigrant Groups in Nine Western Countries.” American Journal of Sociology 110:1412-57. Van Tubergen, F., I. Maas, and H. Flap. 2004. “The Economic Incorporation of Immigrants in 18 Western Societies: Origin, Destination, and Community Effects.” American Sociological Review 69:701-24. Van Tubergen, F., M. Te Grotenhuis, and W. Ultee. 2005. “Denomination, Religious Context, and Suicide: Neo-Durkheimian Multilevel Explanations Tested with Individual and Contextual Data.” American Journal of Sociology 111:797-823. Veglery, A. 1988. “Differential Social Integration among First Generation Greeks in New York.” International Migration Review 22:627–57. Veltman, C. 1983. Language Shift in the United States. New York: Mouton. Vermeulen, H. (ed.) 2000. Immigrant Policy for a Multicultural Society: A Comparative Study of Integration, Language and Religious Policy in Five Western European Countries. Brussels: Migration Policy Group. Voas, D., D.V.A. Olson, and A. Crockett. 2002. “Religious Pluralism and Participation: Why Previous Research is Wrong.” American Sociological Review 67:212–30. Waldinger, R., H. Aldrich, and R. Ward (eds.). 1990. Ethnic Entrepreneurs: Immigrant Business in Industrial Societies. Newbury Park: Sage Publications. Wallerstein, I. 1974. The Modern World System, Vol.1, Capitalist Agriculture and the Origins of the European World Economy in the Sixteenth Century. New York: Academic Press. Wanner, R.A. 1998. “Prejudice, Profit, or Productivity: Explaining Returns to Human Capital Among Male Immigrants to Canada.” Canadian Ethnic Studies 30:24–55. Ware, C. 1935. Greenwich Village, 1920–1930: A Comment on American Civilization in the Postwar Years. New York: Harper. Warner, R.S. and J.G. Wittner. 1998. Gatherings in Diaspora: Religious Communities and the New Immigration. Philadelphia: Temple University Press.

256

References

Warner, W.L. and L. Srole. 1945. The Social Systems of American Ethnic Groups. New Haven: Yale University Press. Weber, M. [1922] 1993. The Sociology of Religion. Boston: Beacon Press. Welch, M.R. and J. Baltzell. 1984. “Geographic Mobility, Social Integration, and Church Attendance.” Journal for the Scientific Study of Religion 23:75–91. Whyte, W.F. 1943. Street Corner Society: The Social Structure of an Italian Slum. Chicago: University of Chicago Press. Wilson, K.L. and A. Portes. 1980. “Immigrant Enclaves: An Analysis of the Labor Market Experiences of Cubans in Miami.” American Journal of Sociology 86:295–319. Worldbank. 2001. World Development Indicators. Washington: CDROM Worldbank. Wuthnow, R. and K. Christiano. 1979. “The Effects of Residential Migration on Church Attendance in the United States.” Pp. 257–76 in The Religious Dimension: New Directions in Quantitative Research, edited by R. Wuthnow. Academic Press. Yang, F. and H.R. Ebaugh. 2001a. “Transformations in New Immigrant Religions and their Global Implications.” American Sociological Review 66:269–88. ———. 2001b. “Religion and Ethnicity among New Immigrants: The Impact of Majority/Minority Status in Home and Host Countries.” Journal for the Scientific study of religion 40:367–78. Yuengert, A. M. 1995. “Testing Hypotheses of Immigrant SelfEmployment.” The Journal of Human Resources 30:194–204. Yoon, I.J. 1991. “The Changing Significance of Ethnic and Class Resources in Immigrant Businesses: The Case of Korean Immigrant Businesses in Chicago.” International Migration Review 25:303– 32. Zhou, M. 1997. “Growing Up American: Challenge Confronting Immigrant Children and Children of Immigrants.” Annual Review of Sociology 23:63–95. Zhou, M. and J.R. Logan. 1989. “Return on Human Capital in Ethnic Enclaves: New York City’s Chinatown.” American Sociological Review 54:809–20.

Index

composition effects, 86 contexts of reception, 13–14, 35–36, 203 contextual effects, 86, 205 country of destination immigration policy, 17, 61, 70, 88, 109, 112–113 social democracy, 87, 89, 96, 99, 153, 158, 223– 224, 229 country of origin globalization, 5, 147, 158, 159, 225 political suppression, 89, 96–97, 99, 127, 136, 159, 229 religion, 39, 89, 92, 114, 127, 138, 161, 178, 184–185, 187, 190, 205, 226 cross-level interactions, 209 cross-national research black box problem, 70–71 comparability, 5, 50–53, 57–58, 78, 95, 120, 165, 172, 214 Galton problem, 69–70

assimilation, 3, 5, 8–10, 13, 16, 20, 35–36, 44, 168, 193, 195, 210–211 Australia, 2–3, 10–13, 30, 47, 54, 56, 62, 70, 74, 77, 80– 82, 88, 97–98, 103–104, 112–113, 119–121, 130, 136, 143, 149, 173– 175, 194, 198, 214–220, 223, 227, 232 Austria, 3, 13, 54, 76–77, 80– 82, 98, 102–104, 120–121, 136, 195, 198, 214 Belgium, 3, 11, 25, 54, 74, 76– 83, 98, 103–104, 121–122, 143, 148–149, 172–175, 195–198, 214, 216, 218, 220 Canada, 2–4, 11–13, 29–30, 47, 54, 56–57, 70, 74, 77, 80–82, 88, 98, 102–104, 112–113, 119–122, 130, 136, 160, 173–175, 194– 195, 198, 214–216, 221, 223 257

258 multilevel, 32, 62–68, 84, 87, 88, 90, 95, 105–106, 122–123, 125–126, 128, 146, 148, 152–154, 156, 165, 176–178, 180, 202 panel data, 211 sensitivity analysis, 58, 95, 153, 165, 166, 188 small-n, 53, 59, 69, 209 Denmark, 3, 54, 77, 79–82, 103–104, 121, 143, 146, 149, 169, 173–174, 195, 198, 214, 216, 219 employment, 5, 7, 73–102, 111–112, 131, 134, 136– 137, 158, 196 ethnic identity, 7, 37, 45, 207 ethnic capital. See social capital ethnic enclave, 33, 163 European Union Labour Force Survey, 57, 120, 214, 219 Finland, 3, 54, 77, 80–82, 95, 103–104, 121, 195, 198, 214 France, 3, 11, 48, 54, 67, 68, 74, 77, 80–82, 98, 102– 104, 121, 195–196, 198, 214 geographical segregation, 25, 207 Germany, 3–4, 11, 47–48, 54, 62, 74, 76–77, 79–82, 102– 105, 121–122, 136, 143,

Index 173–175, 195, 198, 209, 214, 216, 218, 220 Great Britain, 3, 143, 149, 173, 174–175, 189, 216, 217 Greece, 3–4, 54, 77, 80–82, 98, 102–104, 121, 131, 150, 195, 198, 214 human capital, 15–20, 32, 36, 38, 40–41, 70, 85–86, 89, 93–94, 99, 101, 105, 108– 109, 111–112, 117, 124, 125, 136, 138–141, 166, 169, 203–204, 206, 208 host-country specific skills, 16, 20, 85, 89, 124, 204, 206 observable skills, 85, 88, 93, 99, 105, 204 post-migration investments, 16, 93 skill selection, 19, 87, 92, 93–95, 99–100, 108– 109, 112, 203–204 transferability, 93, 135, 137, 204 unobservable skills, 85 immigrant community geographic distance, 22, 60–61, 68, 94, 96, 112, 132, 164, 208 group size, 22–23, 25, 30, 33, 68, 90, 94–95, 99, 107, 110, 129, 132, 133, 155, 157, 163, 181, 188, 205, 209, 226 income inequality, 19, 92– 93, 99, 226, 230

Index linguistic distance, 22, 60, 141, 162, 227 intermarriage, 7, 10–11, 15, 23, 25, 32, 36–38, 45, 207 International File of Immigration Surveys, 53, 120, 169, 213, 218, 227, 228 Ireland, 3, 45, 54, 77, 80–82, 103–104, 121–122, 136, 195, 198, 214 Italy, 3–4, 45, 54, 56, 74, 77, 79, 80, 81, 82–83, 103– 104, 121, 150, 173, 174– 175, 195, 197, 198, 214, 217–218, 220 language proficiency, 6, 11, 20–22, 25, 36–37, 39–40, 45, 46, 47, 53, 56–60, 88, 92, 99, 109, 135, 139–166, 195, 200–201, 210, 213– 215, 218, 225, 227, 230 language use, 25, 36, 37, 47, 207 Luxembourg, 3, 54, 74, 76–77, 80–82, 103–104, 121–122, 136, 195, 199, 214, 219 modernization theory, 190 modes of incorporation. See contexts of reception Netherlands, 1–4, 11, 48, 54, 62, 74, 77, 79–83, 103– 104, 121, 168, 173–175, 186, 189, 193, 195, 199, 206–207, 209, 214, 217– 218, 221 Norway, 3, 11, 54, 149, 195, 199, 217, 221

259 occupational status, 6–7, 12, 32, 36–37, 40, 47, 63, 67, 86, 101–114, 195–196, 200–201, 214–215, 224, 233 point system, 88, 112–113 Portugal, 3, 54, 56, 77, 80–82, 103–104, 121–122, 136, 195–196, 199, 214 prejudice, 15, 25–39, 95, 158, 203, 224, 229 discrimination, 14, 26–40, 86–87, 89, 92, 94–95, 99, 101, 109–111, 114, 125, 127, 130–131, 134, 136–138, 159, 205 religious affiliation, 6, 41, 168––189, 195, 197, 200– 201, 229 religious attendance, 6, 56, 169–187, 200 religious market theory, 191 research design case study, 43–50 comparative destination design, 46–48, 62 comparative origin design, 48, 62, 74 double comparative design, 5, 43–49, 62, 65, 195 synthetic cohort design, 147 second generation, 10, 39, 207–208 self-employment, 6–7, 37–38, 40, 47, 117–138, 195–196,

260 200, 205, 214–215, 230, 233 social capital, 14–15, 32–39, 71, 94, 95, 101, 109–111, 124–125, 138, 169, 203, 205–207 social-integration theory, 191 Spain, 3–4, 54, 56, 67, 74, 77, 80–82, 103–104, 121–122, 136, 150, 195, 199, 214 Sweden, 3, 11, 13, 54, 74, 77, 79, 80–82, 97–98, 103– 104, 121, 195, 199, 214

Index United States, 1–5, 8– 12, 16– 17, 23, 25, 29, 32, 35, 44– 49, 54, 56, 60, 62, 74, 88, 92, 102, 105, 113, 117, 120–122, 127, 130, 136, 143, 147, 149, 160, 167, 173–175, 186, 193–196, 199, 206–209, 214–218, 221, 223, 226, 228 voting, 207