To finish a false debate: ethnic statistics. Rahsaan Maxwell. (Originally published in French as

Transcription

1 To finish a false debate: ethnic statistics Rahsaan Maxwell (Originally published in French as Pour en finir avec un faux débat: les statistiques ethniques En Temps Réel Cahier no. 40, September 2009, with preface by Patrick Weil) France is in the midst of a vigorous debate over how demographic statistics should document immigrant-origin people. Some use the Republican tradition to argue that data should only distinguish between citizens and foreigners and should not further divide the population, especially along socially-constructed and subjective categories like race and ethnicity. Others argue that even if the categories are socially-constructed and subjective, they are important for understanding life outcomes among immigrant-origin individuals. Therefore, these advocates claim that a combination of ethnic, racial, religious, and national-origin categories are necessary for accurately documenting the status of immigrant-origin individuals.

2 The debate between these two perspectives has developed over several decades of academic and political controversies in France. On the one hand, academic researchers or politicians who have attempted to institute some form of ethnic statistics have often been condemned as threats to the Republic who evoke memories of the racist Nazi-collaboration past. On the other hand, activists who support ethnic statistics often claim that their opponents are racist and attempting to hide all evidence of the inequalities in French society. 1 Further complicating the matter is the fact that several distinct issues are involved, which at times means that different sides of the debate are speaking past each other. One issue is the legality of collecting data on subjective categories like race or ethnicity. Another issue is the political interest of collecting such data and whether it would help combat discrimination or perpetuate stigmatization. Yet another issue is how long categories of immigrant-origin remain relevant, whether French citizens should be classified according to the origins of their parents, grandparents, or great-grandparents, and when these descendants should be considered perfectly integrated. In recent years, there is some evidence that a middle-ground consensus may be emerging. Official and obligatory surveys (e.g. the Population Census) look like they will remain off-limits for collecting data on race, ethnicity, or religion. Yet, optional privatelysponsored surveys which pose a variety of questions about identity are increasingly common. Privately-sponsored surveys may offer the benefits of independence from government restrictions on ethnic and racial categories but they are very expensive and not feasible for most researchers in France. Therefore, this article explores the possibility for studying migrant-origin populations in France with existing data. I present results from a longitudinal study connected to 1 The debate has gone through several stages and public conflagrations over the years, which are not necessary to recount in detail here. For a recent overview of the main controversies see the May 2009 special issue of Esprit.

3 the Census called the Échantillon démographique permanent (EDP). This report will interest scholars who seek ways of conducting research on migrant-origin people, politicians who need useful and reliable information on migrant-origin populations, and the general public who wants a better understanding of French society. However, this report also describes the limits of the EDP and the current system for data diffusion in France and presents alternative solutions for supporting better research in the future. EXISTING OPTIONS A central issue in the debate is how to distinguish between French-born citizens of native lineage and French-born citizens of immigrant lineage. Both populations have French citizenship and are equal under the law but researchers and politicians often want to know whether the two groups have different demographic, behavioral, and attitudinal characteristics. One option has been to use data on indirect characteristics that do not directly measure immigrant origin but may be correlated with foreign-origins, e.g. name, native language, or language spoken at home. These characteristics are widely-available in many data sources and have the advantage of effectively measuring individuals who are easily identifiable as having foreign-origins. However, one major disadvantage to this method is that it cannot capture individuals who have assimilated to the point that they primarily speak French and have names that are indistinguishable from the rest of the population but whose life-experiences are still heavily marked by their parents national origins, their religion, or their skin color. Another option is for researchers to conduct their own specialized surveys with questions that directly ask about race, ethnicity, or religion. An advantage to this option is that while it is

4 not feasible for such questions to appear on the French census, independent researchers may be able to collect more detailed information about identity among immigrant-origin people. 2 A disadvantage to this strategy is that it can be very time-consuming and expensive for individual researchers to conduct their own surveys. In addition, when individual researchers are affiliated with public institutions (e.g. INSEE or INED) there have been several controversies over whether it is appropriate for the French government to collect such information. 3 A third option is to use existing data sources that identify the country of birth for respondents parents. There are several longitudinal surveys in France that include this information and allow researchers to distinguish between French-born citizens with native parents and French-born citizens with foreign-born parents. 4 This strategy is well-suited for identifying respondents with foreign-born parents but the information is relatively crude and does not include additional details on parents citizenship. As a result, this strategy cannot distinguish between the descendants of French colonial settlers and French colonial subjects because both have parents who were born abroad. 5 While these two sub-groups are both French citizens and equal under the law, much of the recent debate around immigrant integration has focused on the descendants of former colonial subjects. Not being able to clearly identify this population is a significant drawback to the use of these data for understanding the nuances of immigrant integration. 2 Prominent examples of this strategy include the research of Vincent Tiberj on French Muslims or the short opinion polls of the Le Conseil représentatif des associations noires de France (CRAN) on French blacks. 3 Two prominent examples include the 1992 Mobilité géographique et insertion sociale (MGIS) survey and the 2009 Enquête Trajectoires et Origines (TeO) survey, both conducted by INED. 4 For example, the 1993 and 2003 Formation et Qualification Professionnelle (FQP) survey, the 1999 Histoire Familiale survey, the Enquête Emploi since 2005 and the 2006 Enquête Logement. 5 In the case of Algeria, this strategy also fails to distinguish between the Jewish Algerian migrants and Muslim Algerian migrants who may have very different integration experiences in metropolitan France.

5 L'ÉCHANTILLON DÉMOGRAPHIQUE PERMANENT Another option for studying immigrant-origin populations in France that is lesscommonly used is the Échantillon démographique permanent (EDP). The EDP offers the advantage of data on respondents parents place of birth as well as parents citizenship. This allows researchers to make the crucial distinction between the descendants of French colonial settlers (who had French citizenship) and the descendants of French colonial subjects (most of whom did not have French citizenship). The EDP began in 1968 with a sample of 485,009 residents of metropolitan France. The sample was constructed using information from the état civil to identify individuals born on the 1 st, 2 nd, 3 rd or 4 th days of October. This provided a sample of roughly 1% (4/365) of the French population. The EDP sample was part of the 1968 Population Census and provided standard demographic census data for each respondent. In addition, Census data for the EDP sample were combined with état civil records to provide more detailed information on their family history, including the place of birth and the citizenship of their parents. The same procedure was conducted for the 1975 Census (526,014 people), the 1982 Census (551,211 people), the 1990 Census (528,428 people), and the 1999 Census (617,195 people). Each successive wave included all respondents from the previous waves who were not deceased as well as new respondents who had been born in between surveys. The EDP files currently contain data on

6 almost 900,000 respondents, 260,605 of whom have been present in each Population Census since The EDP is a rich source of data but it is also a complicated dataset, which has limited its use among social science researchers. As with any large dataset the EDP contains non-responses across different questions and survey years. To improve the quality of the data, INSEE staff has imputed responses to certain variables but not in a uniform manner across survey years. In addition, there have been slight changes to certain variables across the years which make it difficult to compare outcomes over time on certain questions. Both issues are common among all longitudinal studies but the EDP has the additional barrier of containing sensitive information on private French citizens. Unlike public opinion surveys conducted by private polling firms, the EDP data are based on official government files and contain the full name and address of each respondent. This creates security concerns which limit the conditions under which researchers may access the EDP. Due to the sensitive personal information included in the files, I was required to submit a proposal and once it was accepted I was only able to access the data under limited conditions. I worked under the supervision of Guy Desplanques and Catherine Borrel along with the Research Group of Statistics and Studies on Immigration at INSEE. I was only able to analyze the data at INSEE s Paris offices during normal business hours and was prohibited from leaving the building with any data files. Despite the EDP s numerous difficulties, several researchers have accessed its unique data. The EDP has primarily been used to study demographic developments (e.g. social mobility or internal migration around metropolitan France) among the general population. Fewer people 6 For more details see Cristine Couet (2006) L échantillon démographique permanent de l Insee Courrier des statistiques no

7 have used the data to study immigrant-origin populations, but that number is sure to grow. The first migration scholar to use the EDP was Jean-Luc Richard who wrote his Ph.D. dissertation at INED. Richard used the EDP to study employment disadvantages among second-generation migrants with different origins, 7 and he combined the EDP with the Enquête Participation Électorale (EPE) to analyze voting behavior in the 1995 presidential election among secondgeneration migrants. 8 The EDP and the EPE allowed me to identify first-generation migrants as respondents born in a foreign country without French citizenship. Second-generation migrants were born in France to parents born abroad without French citizenship. Respondents whose parents were born in France with French citizenship were coded as native French metropolitans. Using these definitions, In the rest of this article I present results from my research with the EDP which analyzed migrant socio-economic outcomes as well as voting behavior in three 2004 elections. THE EDP AND SOCIO-ECONOMIC OUTCOMES Some of the major questions concerning immigrant integration are the extent to which immigrants and their descendants are able to access labor market opportunities and whether they face discrimination or other disadvantages. In France these concerns have been shaped by concerns over the high unemployment rates in the suburban areas where many migrant-origin 7 Jean-Luc Richard (2007) Youth of foreign origin entering the labor market in France: An approach of discrimination against children of African immigrants Revista Economica, 31(1): Jean-Luc Richard (2004) Partir ou rester? Destinées des jeunes issus de l immigration. Paris: Presses Universitaires de France; (1999) Comment Votent Les Jeunes Français Issus de l Immigration, Ville-École- Intégration, 118(3): ; (1998) Rester en France, devenir français, voter: trois étapes de l'intégration des enfants d'immigrés, Economie et statistique, :

8 individuals live as well as the violent disturbances of fall 2005 which were seen as a way for migrant-origin youth to protest their lack of opportunities in mainstream French society. Furthermore, much of the public discourse in France assumes that second-generation migrants from North African, Sub-Saharan African, or Caribbean origins face severe socio-economic disadvantages. However, due to the lack of systematic data on these populations much of the evidence for these claims has relied on small (and statistically unreliable) surveys, anecdotal evidence, or dramatic (and unrepresentative) events or individuals. 9 Given the mismatch between the seriousness of the issues and the paucity of concrete data, the EDP is a useful option for examining these issues with a nationally-representative sample from across metropolitan France. Table 1 presents the EDP sample sizes used to analyze socio-economic outcomes among first and second-generation respondents from nine national-origin groups: Native French metropolitans, Caribbeans, Maghrebians, Spanish, Italians, Polish, Portuguese, Africans, and Turkish. 10 Unfortunately there were not enough Asian respondents in the survey to present reliable data on their socio-economic outcomes although they are a growing and increasinglyimportant segment of the French population. 9 I do not mean to suggest that there are no reliable studies of second-generation migrant socio-economic outcomes in France. On the contrary, there are many reliable studies which have provided a wealth of useful information on the difficulties faced by these populations. However, these studies tend to be focused either on small samples or a limited geographical frame and the EDP is a useful alternative that provides a broader view of these issues across metropolitan France. 10All respondents from all groups are residents of metropolitan France. Native French metropolitans are classified as individuals born in metropolitan France with parents born in metropolitan France. The other categories include migrants and children of migrants from the respective origins. These are not the only national-origin categories included in the EDP but they are the ones with the largest samples and the ones most suitable for statistically-reliable analysis The full EDP dataset has 900,000 respondents but fewer are used in this analysis because not all respondents were available (or living) to be matched with variables from the 1999 Census.

9 Table 1: Group Sample Sizes in the 2003 EDP N Native Metropol. 267,407 Caribbeans 11 4,357 Maghrebians 7,598 Spanish 4,793 Italians 7,878 Polish 1,788 Portuguese 4,429 Africans 12 1,108 Turkish 455 Educational Outcomes Figure 1 presents results from the EDP for educational outcomes among the different groups. The bars indicate the percentage of respondents from each group with each level of qualifications. The results indicate that most groups have roughly the same percentage of respondents with higher education degrees (~20%), although Africans have the highest percentage of respondents with higher education degrees at almost 40%. There is a bit more variation across groups for the percentage of respondents with no qualifications. Again, Africans have the best outcomes with less than 10% with no qualifications. The European migrant groups have some of the largest percentages of respondents with no qualifications (over 30%) and native metropolitans are roughly in the middle with just over 20%. 11 Respondents are classified as Caribbean if they were they were born in Guadeloupe, Guiana, or Martinique with French citizenship or if they were born in metropolitan France with at least one parent who was born in Guadeloupe, Guiana, or Martinique with French citizenship. Although migrants from the Caribbean are technically internal migrants without the same legal issues as immigrants from other countries they face many of the same general integration issues. 12 Countries coded as African are Cameroun, Central African Republic, Congo, Cote d Ivoire, Benin, Ethiopia, Gabon, Guinee, Liberia, Volta, Madagascar, Mali, Mauritania, Nigeria, Senegal, Chad, Togo, Nigeria, Zaire, Maurice, Somalia, and the Comores.

10 Figure 1: Highest education level among different groups Sources: INSEE 1999; 2003 EDP The results in figure 1 may run contrary to some popular notions of educational outcomes among migrant groups. One of the standard stereotypes is that European migrants and native French metropolitans are better-qualified than non-european migrants. These results do not support those views and suggest that Maghrebians, Caribbeans, and Turkish respondents have educational outcomes that are roughly similar to those of native French metropolitans and better than those of Southern and Eastern European migrants. In addition, the data suggest that overall levels of educational disadvantage for non-european origin migrants in France might not be as high as some expect. Furthermore, Sub-Saharan Africans appear to have better qualifications

11 than any other group which is a fact often noted in scholarly circles but less well-known among the general public. 13 Figure 2: Highest male education level among different groups Figures 2 and 3 present educational data across groups with results according to sex and generation. There are two main trends across the two figures. First, comparisons across groups correspond more or less to the group comparisons in figure 1. Second, there is a general trend towards better educational outcomes in the second generation but that is more pronounced among women as opposed to men. The largest gains across generations have come among 13 One explanation for the relatively-high levels of education among Sub-Saharan African-origin migrants is the fact that many arrived in France to escape the stifling conditions of living under political dictators at home. These migrants tended to have resources which gave them the option of migrating to Europe in search of more professional and entrepreneurial economic opportunities but were not the same profile as the stereotypical working-class migrant. In addition, highly-educated Sub-Saharan African migrants are often less visible in the public sphere than the working-class and unemployed migrants often see in the streets and who distort public perceptions of the whole African-origin population. See: Jean-Christophe Dumont (2008) A Profile of Immigrant Populations in the 21 st Century: Data from OECD Countries. OCED: Paris, France or MarcTardieu (2006) Les Africains en France: De 1914 à nos jours. Le Rocher: Monaco.

12 Spanish and Italian women for whom the first generation was not very likely to have degrees past collège-level but by the second generation the outcomes have converged with most other groups. Figure 3: Highest female education level among different groups The results in figures 1, 2, and 3 indicate better educational for non-european migrants in comparison to both native French metropolitans and European-origin migrants than is commonly assumed. To some extent, this is evidence of the strength of the French educational system and its relatively broad reach across groups with different migrant origins. However, these results should be interpreted with a few notes of caution. First, the European migrant groups in this EDP analysis are from some of France s traditional sources of working-class migrants so the comparison between European and non-european origins migrants might look different if more

13 Northern and Western European countries (e.g. the United Kingdom, Germany, Netherlands, or Scandinavian countries) were included. Second, non-european migrant groups may have larger percentages of undocumented lower-skilled migrants who do not appear in these official data and may skew the actual and publically-observed percentages for these groups towards in a lesssuccessful direction. Finally, these data are for rough categories of educational attainment but conceal significant diversity. There is evidence that non-european migrants are more likely than native French metropolitans to get technical BACs as opposed to traditional BACs and are less likely to attend the prestigious grandes écoles even if they do attend higher education. As a result, while the overall educational outcomes look fairly good across immigrant-origin groups and speak to the strength of the broad French educational system, there may still be significant socio-economic disadvantages across groups when occupational and employment outcomes are considered. Occupational Attainment Table 2 presents results across groups for occupational categories among active individuals. 14 In comparison to the educational outcomes in figure 1, the results in table 2 appear to indicate greater diversity across groups in terms of occupational profiles. For example, Turkish and Portuguese migrants are especially concentrated in manual-labor occupations, Caribbeans have high levels of employees, and Africans, Polish and Native French Metropolitans have the highest percentages of professionals. Much of this conforms to conventional wisdom about migrant employment patterns, as Turkish and Portuguese migrants have been known to 14 Students, retirees, and unemployed individuals are excluded from these data and do not factor in to the percentages. In addition, results are not presented for the percentage of respondents employed in agricultural work because the numbers are so small among most migrant groups. As a result, some of the columns may not sum to 100% because of the agricultural workers not included in these tables.

14 work in manual-labor jobs and Caribbeans have been known to concentrate on public sector employee positions. The relatively high percentage of African professionals may surprise some readers, although this follows logically from the relatively high percentage of African migrants with higher education credentials. Table 2: Occupational category distributions among active members of different groups Manual Employ. Artisan Intermed. Prof. Workers Propriet. Native Metropol. 25% 30% 7% 21% 11% Caribbeans 26% 48% 1% 19% 6% Maghrebians 31% 35% 6% 20% 9% Spanish 29% 34% 8% 20% 9% Italians 30% 32% 8% 21% 8% Polish 25% 29% 6% 27% 13% Portuguese 45% 32% 6% 13% 3% Africans 20% 40% 4% 21% 16% Turkish 59% 22% 9% 8% 3% Sources: INSEE 1999; 2003 EDP Tables 3 and 4 present occupational data across groups with results according to sex and generation. For the most part, women are less likely than men to be employed in manual labor positions and are more likely to work as employees. In addition, there is a general trend towards smaller percentages of manual workers and larger percentages of intermediary and professional workers among the second generation as opposed to the first generation. This is to be expected as first-generation migrants were more likely to have lower educational qualifications and second-generation migrants are likely to greater familiarity with French society and therefore more capacity to achieve high occupational status. Maghrebian men are a notable exception to this trend because 54% of the second generation is employed in manual occupations compared to only 39% of the first generation.

15 Moreover, 17% of first-generation Maghrebian men were professionals compared to only 6% of second-generation men. This appears to be evidence of a general trend towards downward mobility among Maghrebian men, which is consistent with some of the popular media reports about the problems in Maghrebian communities. Further evidence of this problem can be found in the data on unemployment rates which is presented in the next section. However, it is also possible that these data may overstate the extent of downward mobility among Maghrebian men because students are excluded from the calculations. Data from the EDP on age structure across migrant groups suggest that Maghrebians have an age distribution that is heavily skewed towards young people in comparison to other groups. 15 Maghrebians have a larger percentage of people who are still in school and have not yet entered the workforce. In theory, as these younger Maghrebians enter the workforce they may improve on the data presented in table 3 because of their greater familiarity with French culture and society in comparison to first-generation migrants. Nonetheless, the labor market difficulties faced by many young Maghrebians in France are a significant problem for social cohesion and the future of migrant integration and are compounded by the unemployment difficulties outlined in the next section. Table 3: Male occupational category distributions among different groups Manual Employ. Artisan Intermed. Prof. Workers Propriet. Caribbeans First Gen 44% 30% 2% 18% 7% Sec Gen 33% 36% 2% 22% 7% Maghrebians First Gen 39% 15% 10% 18% 17% Sec Gen 54% 15% 5% 20% 6% 15 For example, in the EDP sample 22% of Maghrebians were under 30 years old compared to 5% of native French metropolitans. This is not the youngest distribution of all groups but when combined with the fact that Maghrebians are more likely than other groups to live in disadvantaged neighborhoods where educational and upwardly-mobile professional occupational opportunities are limited, this helps explain why the data show larger percentages of manual labor workers among the second generation.

16 Spanish First Gen 50% 10% 12% 18% 7% Sec Gen 41% 15% 9% 21% 12% Italians First Gen 51% 8% 15% 16% 7% Sec Gen 43% 13% 9% 23% 9% Polish First Gen Sec Gen 39% 13% 7% 27% 15% Portuguese First Gen 61% 8% 12% 14% 5% Sec Gen 55% 14% 5% 20% 6% Africans First Gen 32% 16% 6% 25% 21% Sec Gen Turkish First Gen 64% 7% 17% 8% 3% Sec Gen Sources: INSEE 1999; 2003 EDP Note: Sample sizes were too small (i.e. <100) for first-generation Polish respondents and secondgeneration Turkish and African respondents so their results are omitted from the table Table 4: Female occupational category distributions among different groups Manual Employ. Artisan Intermed. Prof. Workers Propriet. Caribbeans First Gen 9% 68% 1% 17% 5% Sec Gen 11% 55% 0% 29% 5% Maghrebians First Gen 14% 55% 5% 18% 8% Sec Gen 13% 58% 2% 22% 5% Spanish First Gen 20% 58% 6% 11% 3% Sec Gen 12% 53% 6% 20% 7% Italians First Gen 29% 47% 6% 11% 3% Sec Gen 15% 53% 5% 19% 6% Polish First Gen Sec Gen 17% 48% 6% 23% 7% Portuguese First Gen 24% 58% 4% 12% 2% Sec Gen 17% 61% 3% 14% 4% Africans First Gen 9% 66% 2% 14% 9% Sec Gen Turkish First Gen Sec Gen Sources: INSEE 1999; 2003 EDP Note: Sample sizes were too small (i.e. <100) for first and second-generation Turkish, firstgeneration Polish, and second-generation African respondents so their results are omitted from the table

17 Unemployment outcomes Table 4 presents results from the EDP on unemployment rates across groups, sex, and generations. The first column on the left indicates that unemployment numbers were roughly similar (10-15%) for most groups but that they were much higher for Africans (23%), Maghrebians (28%) and Turkish migrants (29%). This suggests that generally-speaking these groups face the most difficulties with getting access to employment, although there are some important differences across sex and generation. Generally-speaking, the second-generation (and especially second-generation women) have higher unemployment rates than the first generation. This is part of a broader trend in French society where young people face greater difficulties getting access to the labor market. However, the results in table 4 suggest that there are important differences across migrant groups and non-european origin second-generation migrants in particular are vulnerable to higher unemployment rates. 16 The high unemployment rates for Africans are particularly troubling given the relatively-high educational outcomes presented earlier. This suggests that Africans may face unique hurdles gaining access to the French labor market. Table 4: Unemployment rates across different French groups Overall First Gen Men Sec Gen Men First Gen Women Sec Gen Women Native Metrop % 9% 12% 11% 17% Caribbeans 15% 12% 25% 14% 24% 16 Second-generation Caribbeans and Maghrebians have particularly high unemployment rates. Unfortunately due to small sample sizes for second-generation Africans and Turkish respondents, the results were not reliable enough to present here according to generation and sex. However, the small-sample results did suggest that those groups may face high unemployment rates among the second-generation which are comparable to those for Caribbeans and Maghrebians. 17 The generation comparisons for native French metropolitans are based on age categories because all of the respondents were born in metropolitan France. The full French first generation includes individuals aged 45 and older and the second generation includes individuals aged

18 Maghrebians 28% 22% 30% 29% 29% Spanish 12% 8% 11% 13% 16% Italians 11% 7% 8% 15% 13% Polish 12% - 11% - 14% Portuguese 11% 6% 9% 11% 17% Africans 23% 16% - 31% - Turkish 29% 23% Note: Results are omitted for sub-groups whose sample sizes were too small (i.e. <100) Socio-economic summary This section has used EDP data to present an overview of socio-economic outcomes for various migrant-origin groups. In some cases the results were consistent with the conventional wisdom but in other cases the results may surprise some people. This suggests that the EDP is a valuable resource that should be used to promote better awareness of migrant integration dynamics in France and to help formulate better public policies. The data presented here have been linked with the 1999 Census which unfortunately is not the most recent Census. More research should be done using the EDP with recent Census data to get a more up-to-date picture of migrant socio-economic outcomes. VOTING AND POLITICAL INTEGRATION The second main research question I investigated with the EDP was whether immigrantorigin citizens have different voting behavior from native-origin citizens in France. In addition, I examined whether these dynamics varied across migrant groups. The topic is important because recent unrest in urban and suburban areas across France have raised questions about the extent to which migrant-origin populations are able to access mainstream political channels and whether some feel that anti-systemic violence is the only option.

19 Migrants have traditionally been less active than natives in formal politics because many did not possess French citizenship. Even once migrants obtained citizenship they have often been slow to fully participate in mainstream French politics because of continued interest in homeland politics or socio-economic difficulties that limit their possibilities for engagement. In the 1980s, migrant political participation became an increasingly visible public issue in France as second-generation migrants from the Maghreb used their citizenship status to push for greater inclusion in French society. This new mobilization encouraged mainstream political parties to court the votes of the rising non-european origin electorate. Yet, most observers acknowledge that migrant political incorporation has been slow and uneven and led to many frustrations, such as those exhibited by the fall 2005 riots. Moreover, there is a lack of reliable and systematic data on migrant and migrant-origin political participation in France, due to the various restrictions on data collection. My research therefore represents a unique opportunity to use existing data to gain concrete insight on migrant political incorporation. Combining the EDP and the EPE: Voting Analysis The Enquête Participation Électorale is a report of voter turnout from each precinct across the country and shows whether each registered French citizen voted or not. A personal identification number identified the subjects in both surveys and allowed me to merge the EDP variables with a random sample of voting data from the EPE. While the EDP has been available since the 1960s, the EDP and EPE personal identification numbers have only been compatible since the late 1990s, which is why no analysis of minority migrant voting behavior was possible for earlier elections. I used the EDP and EPE to analyze turnout among seven different groups:

20 native French metropolitans, Spanish, Italians, Polish, Portuguese, Maghrebians, and Caribbeans. As seen in the previous section on socio-economic outcomes, the EDP by itself has a larger sample that would allow comparison of a broader range of national-origin groups but the EPEcompatible files are only for a limited number of respondents. The 2004 EPE has results for three elections: the European Parliament elections, the first round of the Regional Councilor elections, and the second round of the Regional Councilor elections. As these are official data collected from local voting bureaus the only information about electoral behavior is whether or not an individual voted. 18 Table 5 presents the turnout rates for each election across each migrant-origin group. As expected, turnout rates were higher across all groups for the regional elections. In each election, native French metropolitans have the highest turnout rates, followed by the four European migrant groups. The lowest turnout rates for each election are among Maghrebians and Caribbeans. These results suggest that there are important differences in political behavior across migrant-origin groups. Table 5: Vote Turnout among registered citizens in three 2004 French Elections (European Parliament, French Regional Elections Rounds 1 and 2) Native Spanish Italian Polish Portuguese Maghreb. Caribbean Metropol. Euro 46% 45% 42% 37% 34% 30% 25% Region 1 66% 66% 64% 62% 57% 50% 47% Region 2 70% 68% 66% 59% 59% 56% 53% N 21, Source: 2004 EPE, 2003 EDP 18 More detailed information about candidate and party preferences are unavailable from these data because they would violate the right to secret ballots.

21 Statistical analysis indicates that the turnout differences between native French metropolitans and each of the European migrant groups are not statistically significant. This suggests that first and second generation Spanish, Italian, Polish, and Portuguese migrants are just as likely as native French metropolitans to vote. However, turnout differences between both Maghrebians and Caribbeans and native French metropolitans are statistically significant. This suggests that non-european origin migrants are the most disadvantaged politically, which is consistent with the conventional wisdom on migrant integration in France. To explore why turnout rates were lower for Caribbeans and Maghrebians, I conducted statistical analysis which weighed the importance of various factors that are commonly assumed to be important for voting (socio-economic status, age, political context). Caribbeans and Maghrebians are less likely than native French metropolitans to have high socio-economic status and are more likely to be young, both of which might explain their lower turnout rates. However, I find that Caribbeans and Maghrebians still have lower turnout rates than native French metropolitans even when these factors are taken into account. Instead, I focus on the importance of geographic context and argue that the low turnout rates among Maghrebians and Caribbeans in the EPE are best explained by their residential segregation. I claim that this means the underlying determinants of non-european origin migrant voting behavior in France are similar to those of native metropolitans. However, residents of disadvantaged neighborhoods are less likely to vote and because non-european origin migrants are more likely than native metropolitans to live in disadvantaged urban and suburban areas they have lower overall turnout More details about this analysis along with the full statistical models can be found in Rahsaan Maxwell (Forthcoming 2009) Political Participation in France among Non-European Origin Migrants: Segregation or Integration? Journal of Ethnic and Migration Studies.

22 Implications of the voting analysis The EDP and the EPE allow us to examine voting behavior among first and secondgeneration migrant individuals. This is preferable to merely analyzing the first generation because it provides a broader picture of how immigrant integration is proceeding. The EDP and EPE analysis provides concrete data to support the claim that first and second-generation Spanish, Italian, Polish, and Portuguese migrants are well-integrated politically in France. However the analysis also provides data to support the argument that first and second-generation Caribbean and Maghrebian migrants are not well-integrated politically. This is an important finding because it suggests that more work needs to be done in order to ensure that Caribbean and Maghrebian citizens become fully engaged in productive mainstream politics. Moreover, the EPE only includes data on registered voters and since non-european migrants tend to have lower registration rates than native French metropolitans, the overall participation gap is likely to be even lower than what is documented in table 5. Many observers will not be surprised by these findings because the conventional wisdom in France suggests that non-european origin migrants face more integration difficulties than European-origin migrants. However, the EDP and the EPE allow more detailed statistical analysis of the relative importance of different factors for political participation. This dataset makes it possible to highlight residential segregation as the most important explanation for lower turnout rates among Maghrebians and Caribbeans in comparison to native French metropolitans. The findings from the EDP and the EPE join a broader chorus of analysis on how spatial and residential segregation are affecting French society. Disadvantaged urban and suburban neighborhoods are a problem for migrant-origin and native-origin residents alike, because they

23 offer few educational and occupational resources for productive engagement with mainstream society. To the extent that residents of these neighborhoods face multiple disadvantages in their lives, they may not feel comfortable with formal political activities and may continue to turn to the violent disturbances that have plagued French neighborhoods in recent decades. CONCLUSION The EDP is a valuable source of data for analysis of first-generation migrants and secondgeneration migrant-origin individuals in France. Unlike other sources of data on secondgeneration migrants, the EDP can distinguish between the descendants of French colonial settlers and French colonial subjects. This allows more precise analysis of how non-european origin residents are integrating into French society. One of the fiercest objections to ethnic statistics is that they reify subjective categories. In comparison, the EDP allows one to analyze second-generation migrant-origin individuals by focusing on their geographic origin and without resorting to subjective categories like race or ethnicity. In some respects, this facilitates a more precise analysis than self-identification with categories that may mean different things to different people. However, it is important to note that because the EDP analysis is fundamentally geographic it may diverge slightly from how the respective categories are perceived in public life. For example, the native French metropolitan category might include small numbers of third-generation migrants whose parents were born in metropolitan France as French citizens but who are still perceived by mainstream society as migrant-origin because of their skin color or religious practices. Furthermore, the Caribbean category might include small numbers of French metropolitans born in the Caribbean.

24 Therefore, it is important not to make the conceptual slippage between the national and geographic categories of the EDP and the multiple other forms of identity which can be relevant in public life. Despite the many virtues of the EDP, there are three significant limitations to its use for studying migrant-origin populations in France. One limitation is that it is primarily a demographic study and is restricted to the Census variables. In some circumstances the EDP can be combined with other Census-compatible files (e.g. the EPE) but at the moment it is not possible to use the EDP to examine migrants attitudes towards society, opinions about political issues, and qualitative descriptions of their experiences in France or other non-demographic variables of interest. As a result, the EDP offers data on one narrow slice of immigrant integration and cannot be considered a significant source of data for understanding integration more generally. Another limitation is that the EDP only allows examination of second-generation migrants but is unable to distinguish between respondents whose grandparents or greatgrandparents were migrants and those whose grandparents or great-grandparents were natives. In an ideal world, migrant communities would be perfectly integrated by the third generation but the recent disturbances in France among young third-generation migrants suggest that this is not the case. Moreover, detailed research from the United States on integration trajectories suggests that outcomes can improve and deteriorate well into the fourth generation and that more attention should be paid to the diversity of possible integration trajectories See for example: Tomás Jiménez (2008) Mexican-Immigrant Replenishment and the Continuing Significance of Ethnicity and Race. American Journal of Sociology, 113(6): Edward Telles and Vilma Ortiz (2008) Generations of Exclusion: Mexican Americans, Assimilation and Race. New York: Russell Sage Foundation Press.

25 Finally, EDP files are difficult and cumbersome to access. The long application process and the limited conditions of data analysis in INSEE s Paris offices are a significant constraint on the amount of knowledge that can be produced using the EDP and fewer research projects are conducted using the EDP than might otherwise be possible if the data were more easily available. It is understandable that some of the sensitive information in the EDP must be carefully monitored when the data are used for analysis. However, very few social scientific research projects actually need that information so one option for INSEE would be to make a limited version of the EDP (without respondents names or addresses) available to a broader public. In addition, more resources should be devoted to improving the diffusion of the data. Data sources and data archives in other countries (e.g. the UK Data Archive, the Interuniversity Consortium for Political and Social Research, or the European Social Survey) post their datasets online to be downloaded for free to registered users anywhere in the world. This facilitates more detailed analysis by a broader range of scholars and a greater production of knowledge. In comparison to other countries, information about immigrant integration in France is only sporadically available on a limited number of subjects. As a result, the discussion about immigrant integration in France is too often based on anecdotal evidence. The EDP is a tool which can be used to provide more detailed and reliable data on immigrant integration across generations and its access should be expanded to include more researchers. In addition, future studies should be more readily available in a user-friendly fashion (i.e. free downloads of the full dataset on the internet). These measures would generate better knowledge production and facilitate better public policy about the vital issue of immigrant integration.