European opinion about immigration: The role of identities, interests and information

British Journal of Political Science

Volumn 37, Issue 3, 2007, Pages 477-504

  Sides, John ^a   Citrin, Jack ^b  

^a GEORGE WASHINGTON UNIVERSITY   (United States)

^b UNIVERSITY OF CALIFORNIA   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 34250204546     PISSN: 00071234     EISSN: 14692112     Source Type: Journal    
DOI: 10.1017/S0007123407000257     Document Type: Article

References (81)

1. United Nations Population Division, Replacement Migration (New York: UNO, 2000).
2. C. Joppke, 'European Immigration Politics at the Crossroads', in P. Heywood, E. Jones and M. Rhodes, eds, Developments in Western European Politics 2 (Basingstoke, Hants: Palgrave, 2002), pp. 259-76, at p. 259.
3. For discussion, see: J. Citrin, D. P. Green, C. Muste and C. Wong, 'Public Opinion Towards Immigration Reform: The Role of Economic Motivations', Journal of Politics, 59 (1997), 858-81; V. M. Esses, L. M. Jackson and T. L. Armstrong, 'Intergroup Competition and Attitudes Toward Immigrants and Immigration: An Instrumental Model of Group Conflict', Journal of Social Issues, 54 (1998), pp. 699-724; G. Lahav, Immigration and Politics in the New Europe (Cambridge: Cambridge University Press, 2004); P. M. Sniderman, L. Hougendoorn and M. Prior, 'Predispositional Factors and Situational Triggers: Exclusionary Reactions to Immigrant Minorities', American Political Science Review, 98 (2004), 35-50.
4. J. Fetzer, Public Attitudes Toward Immigration in the United States, France, and Germany (Cambridge: Cambridge University Press, 2000); R. W. Jackman and K. Volpert, 'Conditions Favouring Parties of the Extreme Right in Western Europe', British Journal of Political Science, 26 (1996), 501-21.
5. M. Golder, 'Explaining Variation in the Success of Extreme Right Parties in Western Europe', Comparative Political Studies, 36 (2003), 432-66.
6. J. Money, Fences and Neighbors: The Political Geography of Immigration Control (Ithaca, N.Y.: Cornell University Press, 1999).
7. K. F. Scheve and M. J. Slaughter. 'Labor Market Competition and Individual Preferences over Immigration Policy', Review of Economics and Statistics, 83 (2001), 133-45.
8. Citrin et al., 'Public Opinion Toward Immigration Reform'; Lahav, Immigration and Politics in the New Europe.
9. It is important to note that from a theoretical perspective, prejudice does not enter into the perception of threat or the calculation of self-interest. Immigrant ethnicity, religion, etc. should not matter. For the interest-based explanation, all that matters is the existence of a threat to one's economic circumstances. If there is a preference for immigrants of one ethnicity over another, this should reflect differences in the estimates, presumably accurate, of the impact of these groups on one's material interests.
10. D. O. Sears, 'Symbolic Politics: A Socio-psychological Theory', in S. Iyengar and W. McGuire, eds, Explorations in Political Psychology (Durham, N.C.: Duke University Press, 1996), pp. 113-49; D. Chong, Rational Lives (Chicago: Chicago University Press, 2000).
11. H. Tajfel, Human Groups and Social Categories (Cambridge: Cambridge University Press, 1981).
12. L. Huddy, 'From Social to Political Identity: A Critical Examination of Social Identity Theory', Political Psychology, 22 (2001), 127-56; Sniderman et al., 'Predispositional Factors and Situational Triggers'.
13. J. Citrin and J. Sides, 'More than Nationals: How Identity Choice Matters in the New Europe', in R. Herrmann, T. Risse and M. Brewer, eds, Identities in Europe and the Institutions of the European Union (London: Rowman and Littlefleld, 2004), pp. 161-85; J. Citrin, C.Wong and B. Duff, 'The Meaning of American National Identity', in R. D. Ashmore, L. Jussim and D. Wilder, eds, Social Identity, Intergroup Conflict and Conflict Resolution (Oxford: Oxford University Press, 2001). pp. 71-100; R. de Figueiredo and Z. Elkins, 'Are Patriots Bigots'?' American Journal of Political Science, 47 (2003), 171-88; J. Sidanius and J. Petrocik, 'Communal and National Identity in a Multiethnic State', in Ashmore, Jussim and Wilder, eds, Social Identity, Intergroup Conflict and Conflict Resolution, pp. 101-27; Sniderman et al., 'Predispositional Factors and Situational Triggers.'
14. Lahav, Immigration and Politics in the New Europe. The relationship between the size of the immigrant population and attitudes towards immigrants may also depend on the economic circumstances in the country.
15. G. P. Freeman, 'Modes of Immigration Politics in Liberal Democratic States', International Migration Review, 24 (1995), 881-902, p. 883.
16. R. Nadeau, R. G. Niemi and J. Levine, 'Innumeracy about Minority Populations', Public Opinion Quarterly, 57 (1993), 332-47; L. Sigelman and R. G. Niemi, 'Innumeracy about Minority Populations: African-Americans and Whites Compared', Public Opinion Quarterly, 65 (2001), 86-94; E. Theiss-Morse, 'Characterizations and Consequences: How Americans Envision the American People' (paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, 2003).
17. The ESS also included samples in Israel and Slovenia, which we do not utilize here. OECD country-level data, which we rely on here, are not available for Slovenia. We do not examine Israel since our focus is on European attitudes towards immigration.
18. Studies of attitudes in the Western European EU members include: Lahav, Immigration and Politics in the New Europe; L. McLaren, 'Immigration and the New Politics of Inclusion and Exclusion in the European Union', European Journal of Political Research, 39 (2001), 81-108; L. Quillian, 'Prejudice as a Response to Perceived Group Threat: Population Composition and Anti-Immigrant and Racial Prejudice in Europe', American Sociological Review, 60 (1995), 586-611.
19. V. O. Key, Southern Politics in State and Nation (Knoxville: University of Tennessee Press, 1949); H. M. Blalock, Toward a Theory of Minority-Group Relations (New York: Wiley, 1967).
20. Sears, 'Symbolic Politics'; Chong, Rational Lives.
21. M. V. Hood III and I. L. Morris, '¿Amigo o Enemigo? Context, Attitudes, and Anglo Public Opinion toward Immigration', Social Science Quarterly, 78 (1997), 309-23; Citrin et al., 'Public Opinion Toward Immigration Reform' ; Scheve and Slaughter, 'Labor Market Competition and Individual Preferences over Immigration Policy'; McLaren, 'Immigration and the New Politics of Inclusion and Exclusion in the European Union'.
22. Quillian, 'Prejudice as a Response to Perceived Group Threat'.
23. The ESS immigration battery began with this preamble: 'People come to live in [country] from other countries for different reasons. Some have ancestral ties. Others come to work here, or to join their families. Others come because they are under threat. Here are some questions about this issue.'
24. Whenever we report results for the entire ESS sample, we weight responses so that each country's sample is represented in proportion to that country's actual population (using the ESS variable pweight). Results for individual countries are weighted to account for an unequal probability of selection into the sample within those countries (the ESS variable dweight).
25. McLaren, in 'Immigration and the New Politics of Inclusion and Exclusion in the European Union', examines similar Eurobarometer questions and finds that 'most individuals [appear] not to make adistinction between EC/EU and non-EC/EU immigrants' (p. 85). However, Lahav's examination of other Eurobarometer indicators, presented in Immigration and Politics in the New Europe, does find differences between attitudes towards African and Asian immigrants and attitudes towards European immigrants. Sniderman et al., in 'Predispositional Factors and Situational Triggers', randomize survey respondents into different experimental conditions that describe immigrants in different ways, and find that respondents react more harshly to immigrants who are described as less educated or as culturally dissimilar (see their Figure 1, p. 43). As Pettigrew and Meertens discuss in T. F. Pettigrew and R. W. Meertens, 'Subtle and Blatant Prejudice in Western Europe', European Journal of Social Psychology, 25 (1995), 57-75, there is a developing European norm against blatant prejudice, though this does not obviate the existence of what they term 'subtle prejudice'. Moreover, subtle prejudice is in fact strongly linked to view of immigrants (see, for example. T. F. Pettigrew, 'Reactions to the New Minorities of Western Europe', Annual Review of Sociology, 24 (1998), 77-103.)
26. Other ESS items suggest a somewhat more positive view of immigrants. For example, only 38 per cent agree that 'Average wages and salaries are generally brought down by people coming to work and live here', while 60 per cent agree that 'People who have come to live and work here help to fill jobs where there are shortages of workers'.
27. We computed this index by scaling each individual item to range from 0 to 1 (such that 1 indicates allowing no immigrants into the country), and then averaging these items. The reliabilities of the two indices are very high overall (for perceived consequences, α = 0.84; for preferred levels, α = 0.94) and is comparably high in each individual country (for a discussion of the a statistic, see L. J. Cronbach, 'Coefficient Alpha and the Internal Structure of Tests', Psychometrika, 16 (1951), 297-334). Exploratory factor analyses of the items in each index suggest that each set of items taps a single dimension.
28. Exploratory factor analysis also suggests that these two sets of items are tapping different underlying constructs.
29. We remain agnostic as to the relationship between the perceived consequences of immigration and preferred levels of immigration. It may be that beliefs about consequences lead to preferred levels. Or beliefs about consequences may only be rationalizations of preferred levels. Given this ambiguous causation, we do not include consequences in our model of preferred levels, and vice versa.
30. These two items are: 'Which of the descriptions on this card comes closest to how you feel about your household's income nowadays?: living comfortably, coping, finding it difficult, finding it very difficult' and 'If for some reason you were in serious financial difficulties and had to borrow money to make ends meet, how difficult or easy would that be?' These items were combined into a scale (α = 0.59). More information about variable coding is available from the authors.
31. Scheve and Slaughter, 'Labor Market Competition and Individual Preferences over Immigration Policy'.
32. The ESS included other items that speak to this general idea but that tap different specific aspects of a nation's identity. An overwhelming majority of European publics believe in the benefits of a common language: 91 per cent agree that 'It is better for a country if almost everyone is able to speak at least one common language.' When asked if 'It is better for a country if there are a variety of different religions', 36 per cent of the ESS respondents agree and 32 per cent disagree. There is majority opposition (54 to 30 per cent) to the idea that 'communities of people who have come to live here should be allowed to educate their children in their own separate schools if they wish'. There is, then, a general belief in the benefits of cultural unity for social harmony and a desire that immigrants acculturate, particularly by learning their new country's language and participating in a country's educational institutions. However, there is no consensus on the need for religious homogeneity in the abstract. Given the different foci of these items, it is not surprising that they do not scale together readily. We therefore rely on the single indicator of preferences for cultural unity because it reflects the broadest endorsement of an 'ethnic' conception of the nation.
33. Citrin and Sides, 'More than Nationals'; see also A. Luedtke, 'European Integration, Public Opinion and Immigration Policy: Testing the Impact of National Identity', European Union Politics, 6 (2005), 83-112.
34. This scale has a reliability of 0.73.
35. Approximately 17 per cent of the ESS sample could not provide a numerical estimate of the immigrant population. To avoid throwing out these cases, we imputed values on a country-by-country basis, drawing on respondent's level of education and frequency of political discussion as predictors. One or both of these variables was significantly associated with estimates in all of these twenty countries. Their substantive effect was to reduce the estimates, since more educated or politically engaged people are in general less likely to overestimate immigrant population size. Most importantly, the multivariate results discussed below are substantively the same if we exclude those who could not provide an estimate.
36. The actual percentage of foreign-born residents derives from OECD data to be found at 〈http://www.oecd.org/dataoecd/18/23/34792376.xls〉. This dataset contains a count of the number of residents in each country, broken down by their country of origin. To calculate the number of immigrants, we simply summed up the number of residents and subtracted the native-born. Some residents' countries of origin could not be determined and they were coded as 'unknown'. For the purposes of this analysis, we count them as native-born. In practice, this has little effect on the estimates except in Germany, where a large number of residents were classified as 'unknown'. Considering the 'unknowns' in Germany to be foreign-bom generates an estimate of the proportion of foreign-born residents of 19 per cent, while considering them native-born generates an estimate of 11 per cent. Given that this latter figure better corresponds to other OECD analysis (see J. C. Dumont and G. Lemaître, 'Counting Immigrants and Expatriates: A New Perspective', OECD Social, Employment, and Migration working paper, available at 〈http://www.oecd.org/ dataoecd/27/5/33868740.pdf〉, accessed 6 July 2005), we employ the measure that counts the 'unknowns' as native-born.
37. There is a statistically significant relationship between the estimated and actual percentage of foreign-born residents. Nevertheless, that there is some regularity to popular estimates of immigrant numbers does not detract from the level of misperception in these estimates.
38. These two questions about the size of immigrant populations occurred relatively late in the battery of questions about immigration. Thus, it is possible that respondents had been 'primed' to think about immigration and to consider it a more serious issue, thereby leading them to give higher estimates than they would have otherwise. However, we do not think that the question ordering is a serious confound. The amount of overestimation is comparable to that found in other studies (such as Theiss-Morse, 'Characterizations and Consequences'). Moreover, other polling in Britain by the MORI organization has found that respondents estimated the proportion of foreign-born residents to be 23 per cent, which is very close to the estimate reported in Figure 1 (23.8 per cent). See 〈http://www.mori.com/polls/2003/community.shtml〉.
39. Our measure of social trust is an index combining three questions: 'Generally speaking, would you say that most people can be trusted, or that you can't be too careful in dealing with people?'; 'Do you think that most people would try to take advantage of you if they got the chance, or would they try to be fair?'; and 'Would you say that most of the time people try to be helpful or that they are mostly looking out for themselves?' These items were all 0-10 scales, and the resulting index has a reliability of 0.77.
40. Quillian, 'Prejudice as a Response to Perceived Group Threat'.
41. Pettigrew and Meertens, 'Subtle and Blatant Prejudice in Western Europe'.
42. L. Hagendoorn and S. Nekuee, eds, Education and Racism: A Cross-National Inventory of Positive Effects of Education on Ethnic Tolerance (Brookfield, Vt.: Ashgate, 1999).
43. J. Zaller, The Nature and Origins of Mass Opinion (Cambridge: Cambridge University Press, 1992).
44. M. Gabel, 'Public Support for European Integration: An Empirical Test of Five Theories', Journal of Politics, 60 (1998), 333-54; McLaren, 'Immigration and the New Politics of Inclusion and Exclusion in the European Union'.
45. We weight the sample by the ESS variable pweight. We also estimated these models using hierarchical or multilevel modelling techniques (see A. S. Byrk and Stephen W. Raudenbush, Hierarchical Linear Models: Applications and Data Analysis Methods, 2nd edn (Newbury Park, Calif.: Sage Publications, 2002); D. A. Luke, Multilevel Modeling (Thousand Oaks, Calif.: Sage Publications, 2004); M. R. Steenbergen and B. S. Jones, 'Modeling Multilevel Data Structures', American Journal of Political Science, 46 (2002), 218-37). The results obtained were substantively very similar to those obtained with the simpler estimation strategy employed here - a strategy some have argued can be employed successfully as long as the standard errors are adjusted (see C. Kam and R. Franzese, 'Modeling and Interpreting Interactive Hypotheses in Regression Analysis: A Refresher and Some Practical Advice' (Ann Arbor: University of Michigan Press, forthcoming)). Replication of the analysis using hierarchical models is available from the authors.
46. The models also include a dummy variable for each country except one, though we do not report the coefficient estimates.
47. Citrin et al., 'Public Opinion Toward Immigration Reform'; Lahav, Immigration and Politics in the New Europe.
48. The maximum substantive impact of sociotropic evaluations in this model is equivalent to about one-half of a standard deviation in the perceived consequences index.
49. Scheve and Slaughter, 'Labor Market Competition and Individual Preferences over Immigration Policy'.
50. It is possible that this measure of preference for cultural unity is at least somewhat endogeneous to attitudes about immigration. Perhaps those who advocate cultural unity are doing so as a consequence of their opposition to immigration. The design of the ESS module on immigration, in which the measure of preferences for cultural unity occurred roughly in the middle of the various questions about immigration, may have exacerbated this possibility, if respondents were explicitly thinking about immigrants when they answered this question. A typical strategy to deal with any endogeneity would be a two-stage least-squares regression, drawing on other variables as 'instruments' for preferences for cultural unity. Unfortunately, ideal instruments do not exist in the ESS survey - i.e., there are no measures that strongly correlate with preferences for cultural unity but not with either of the dependent variables we employ here. However, other analysis (for example, de Figueiredo and Elkins's 'Are Patriots Bigots'?') that draws on a two-stage estimator still finds a relationship between attitudes towards the nation - in particular, a more aggressive nationalism - and hostility towards immigrants. Moreover, Sniderman et al. show in 'Predispositional Factors and Situational Triggers' that cultural threats are more strongly associated with views on immigration when statements about these threats are 'decoupled' from any reference to immigrants. Thus, if our measure, even though its wording does not reference immigrants, is nevertheless not fully 'decoupled', this may actually attenuate its impact.
51. More specifically, the marginal effect is computed as the change associated with a shift from the minimum to maximum value of misperception. The logic of these graphs and the STATA code used to make them is discussed in T. Bomber, W. R. Clark and M. Golder, 'Understanding Interaction Models: Improving Empirical Analyses', Political Analysis, 13 (2005). 1-20. A graph of the same interaction from the model of perceived consequences produces a substantively similar result.
52. As Bomber, Clark and Golder, 'Understanding Interaction Models', point out, the statistical significance of the interaction term does not necessarily indicate whether the independent variable's effect on the dependent variable can be distinguished from zero at all levels of the conditioning variable. This is why Figure 2 is particularly important.
53. There is also the potential for endogeneity here: perhaps people who are opposed to immigrants over-estimate their numbers (consciously or not). However, the findings of the research cited previously - which finds that overestimates of minority populations are related to opposition to programmes that would benefit these minorities - gives us confidence that our results are not spurious.
54. O. Allport, The Nature of Prejudice (Cambridge, Mass.: Addison-Wesley, 1954).
55. J. Bowers and K. Drake, 'EDA for HLM: Visualization When Probabilistic Inference Fails', Political Analysis, 13 (2005), 301-26. See also the other articles in this same issue of Political Analysis, which focus on 'multilevel modelling for large clusters' and analyse multi-country surveys similar in structure to the ESS.
56. It is telling that Quillian, who uses hierarchical models to investigate opinion about immigration in only twelve countries in 'Prejudice as a Response to Perceived Group Threat', finds 'fragility' in his estimates (see his fn. 21, p. 603; see also I. Kreft, 'The Estimation of Cross-Level Interactions' at 〈http://www.calstatela.edu/faculty/ikreft/quarterly/node13.html〉), accessed 17 February 2005.
57. To make the graph more readable, we do not indicate the 95 per cent confidence interval around the estimate of the intercept. All of these intercepts are statistically distinguishable from zero.
58. The unemployment data were compiled by the OECD. See 〈http://www. oecd.org/dataoecd/41/13/18595359.pdf〉.
59. There is a positive relationship between the size of the foreign-born population and the intercepts from the model of perceived consequences, but it is driven by a single outlier (Luxembourg), suggesting that this relationship is not broadly representative of all of the data.
60. See Lahav, Immigration and Politics in the New Europe; Quillian, 'Prejudice as a Response to Perceived Group Threat'.
61. See also C. Dustmann and A. Glitz, Immigration, Jobs, and Wages: Theory, Evidence and Opinion (London: Centre for Research and Analysis of Migration, 2005).
62. We also found no evidence that these intercepts were related to the combination, or interaction, of the size of the immigrant population and any of these measures of economic health (see Golder, 'Explaining Variation in the Success of Extreme Right Parties in Western Europe' ; Quillian, 'Prejudice as a Response to Perceived Group Threat'). They are also not related to membership in the EU or length of tenure in the EU. It is also worth noting that estimating a hierarchical model with both individual-level and country-level measures produces insignificant coefficient estimates for the country-level measures. Though we are cautious about the value of these hypothesis tests, for the reasons stated above, this at least confirms the descriptive results presented in Figure 4.
63. Hood and Morris (in '¿Amigo o Enemigo? Context, Attitudes, and Anglo Public Opinion toward Immigration') find that respondents in the United States who live in counties with larger Asian and Hispanic populations tend to feel more positively towards immigration. Scheve and Slaughter ('Labor Market Competition and Individual Preferences over Immigration Policy') do not find the same, however, using the metropolitan statistical area as the contextual unit. Perrineau finds that voting for the National Front is actually more strongly related to immigrant presence in a larger contextual unit (the départment) than in a smaller unit (communes) - see P. Perrineau, 'Le Front National: un électoral autoritaire', Revue Politique et Parlementaire, 87 (1985), 24-31. Specifying the correct contextual unit and isolating its effect is a significant task for future research.
64. We also used hierarchical modelling techniques to 'decompose' the variance in attitudes to determine the proportion of variance explained at the individual and country levels, respectively. For both indices, nearly 90 per cent of the variance was explained at the individual level. This suggests that country-level attributes, while not inconsequential, are substantively a smaller part of the explanatory picture.
65. The coefficients for both measures of economic evaluations have been multiplied by - 1, so that the hypothesized relationship between the magnitude of these coefficients and country-level economic hardship will be positive.
66. We employ GDP per capita (in 2002 US $1000), based on purchasing power parities. See 〈http://www.oecd.org/document/28/0,2340, en_2825_495684_2750044_1_1_1_1,00.html〉.
67. The correlation presented in plot (c) is r = -0.48. The correlation between GDP per capita and the coefficients from the models of perceived consequences are r = -0.71 (satisfaction with personal finances) and r = - 0.60 (satisfaction with economy). It is worth noting that including a cross-level interaction between economic concerns and GDP in a single hierarchical model of both individual-level and country-level attributes generates similar findings: the impact of satisfaction with the economy is larger among respondents who live in countries with a lower GDP per capita. A similar investigation of the relationship between country-level economic indicators and the coefficients for the other individual-level economic measures, income and unemployed status, generated insignificant results.
68. The same is true in plots of the coefficients from the models of perceived consequences, and if we use alternative measures of immigrant presence, such as the proportion of immigrants from outside Western Europe. We got similar findings when we included cross-level interactions in the multivariate model shown in Table 3.
69. We also found no systematic differences among regions within Europe in terms of the magnitude of individual-level coefficients. For example, it does not appear that attitudes in the three East European countries in the ESS sample have a qualitatively different structure than those in Western Europe.
70. Quillian, 'Prejudice as a Response to Perceived Group Threat'. See Scheve and Slaughter, 'Labor Market Competition and Individual Preferences over Immigration Policy', for results similar to ours.
71. Quillian also argues that it is the interaction of country-level economic hardship and immigrant populations that 'prime' individual-level attributes. Though we do not present such tests, other analysis (not shown) finds no evidence of such a relationship.
72. For example, Sniderman et al., 'Predispositional Factors and Situational Triggers'.
73. For example, see M. Gilens, 'Political Ignorance and Collective Policy Preferences', American Political Science Review, 95 (2001), 379-96.
74. D. Hollinger, Post-ethnic America (New York: Oxford University Press, 1995).
75. Y. Soysal. Limits of Citizenship: Migrants and Post-national Membership in Europe (Chicago: Chicago University Press, 1994).
76. Such sensitivity to public opinion inevitably has its limits. First, some leaders believe that immigrant workers benefit countries with declining native-born populations. Secondly, completely closing the door to immigration is probably impossible for governments to implement, constrained as they are by permeable borders and international pressures to accept at least some of the people uprooted by war and persecution. And since migration is largely voluntary behaviour, a 'designer' policy that admits just the 'right' kind of immigrant also may not be feasible. Germany may want Indian software engineers, but these people may prefer to immigrate into the United States or Great Britain.
77. M. Bommes, 'The Shrinking Inclusive Capacity of the National Welfare State: International Migration and the Deregulation of Identity Formation', in G. Brockman, ed., The Multicultural Challenge: Comparative Social Research, vol. 22 (London: Elsevier, 2003), pp. 43-68.
78. For more information, see 〈http://naticent02.uuhost.uk.uu.net/ archive/tech_report/2_sample.doc〉: and 〈http://naticent02.uuhost.uk. uu.net/methodology/monitoring_evaluating_non_response.doc〉.
79. See Table 1 of this ANES analysis at 〈ftp://ftp.nes.isr.umich.edu/ ftp/nes/bibliography/documents/nes010162.pdf〉.
80. E. Scarbrough, 'The British Election Study and Electoral Research', Political Studies, 48 (2000), 391-414.
81. See the studies cited in G. Langer, 'About Response Rates', The Public Perspective, 14 (2003), 16-18.