Abstract
Are multiculturalism policies associated with a rejection of the receiving society’s identity by immigrants? Is there a policy arrangement that makes identification with the receiving society more likely? This article addresses head-on whether there is a trade-off between ethnic identification and relationships with the host society and whether this trade-off is associated with certain policy regimes. Looking at immigrants in 10 Western democracies, the results show that despite the fact that inclusive policies decrease identification with the majority, they do so while being associated with integrationist orientations on issues such as language choice and cultural traditions. Immigrants in these countries value relationships with both their ethnic group and the majority to a greater extent. In countries with inclusive policies, ethnic identification is also associated with commitment to the receiving society and, contrary to more restrictive regimes, does not lead to greater perceived discrimination. These findings are robust to additional models looking at Muslim immigrants.
Despite statements from politicians and academics across Western countries declaring that multiculturalism has failed (Vertovec & Wessendorf, 2010), the actual impact of incorporation policies—policies whose aim is to incorporate immigrants in their new society—on immigrants remains unclear. Studies examining the effect of these policies on different sociopolitical outcomes have come to mixed conclusions (Dancygier & Laitin, 2014; Koopmans, 2013). For instance, some find that multiculturalism policies mixed with a generous welfare state lead to poor job market integration (Koopmans, 2010) or argue that cultural assimilation measures might be necessary to promote sociocultural integration (Ersanilli & Koopmans, 2010). Yet, others find that multiculturalism policies are associated with engagement and political inclusion (Wright & Bloemraad, 2012) and that citizenship access—at least in Switzerland—leads to more social integration (Hainmueller, Hangartner, & Pietrantuono, 2017). These disparate, often contradictory, findings are in part a consequence of the disagreements over the theoretical and empirical value of different outcomes. They are also stemming from an important issue that researchers examining the effects of policies on immigrants in a comparative framework face: the lack of reliable data. Immigrants represent a constituency that can be hard to reach and there are often too few of them in nationally representative samples, which limits the possibilities for statistical analysis.
Given the central role they play, however, it is crucial to evaluate comparatively whether incorporation policies are fulfilling their objectives or whether, as their critics argue, they are failing. These critics often focus on questions of identity and of immigrants’ attachment to their new society while stressing the necessity of creating a common “we” (e.g., Barry, 2002; Bissoondath, 1994; Miller, 1995). At the same time, proponents of inclusive policies argue that they are necessary to ensure that newcomers engage with their new society and feel accepted (e.g., Kymlicka, 1995). In the end, the debates hinge on the role played by ethnic identification: Critics of cultural recognition and inclusive policies see ethnic identity as leading to segregation and disengagement, whereas proponents of these policies see ethnic identity as a path to meaningful participation and sense of membership. Empirically, however, it remains unclear whether policies are associated with identity choices and how these choices affect relationships between immigrants and the majority.
Therefore, this study asks, “Are some policy arrangements associated with greater ethnic identification within immigrant populations?” How do these identification processes reflect on larger attitudes toward language choice, cultural traditions, and perceived discrimination when comparing across policy regimes? Ultimately, the question becomes whether there is a trade-off between ethnic identification and relationships with the host society and whether certain policy regimes create this trade-off. To answer these questions, I use a data set developed by social psychologists whose original goal was to study immigrant youth well-being, a survey that contains unique measures of identification and acculturation orientations in 12 countries (Berry, Phinney, Sam, & Vedder, 2010). I then combine the individual-level data from 10 of these countries with well-known country-level indices measuring two important incorporation policies: multiculturalism policies and citizenship policies.
The study offers three main results. First, in line with critiques of more inclusive policies, countries with extensive multiculturalism policies and more open access to citizenship do have immigrants with higher levels of ethnic identification and lower level of identification with the majority. Second, this higher level of ethnic identification does not lead immigrants in these countries to segregate or isolate themselves. On the contrary, in countries with more inclusive policies, ethnic identification is associated with a desire to engage both with the ethnic group and with the majority on issues such as marriage, language choice, and social activities. Third, looking at perceived discrimination, the study shows that immigrants in restrictive countries are more likely to feel discriminated and that ethnic identification increases this perception. Strikingly, ethnic identification is not associated with perceived discrimination in countries with extensive multiculturalism policies and more open citizenship rules.
The study makes four contributions to the literature on incorporation policies and immigrants’ attitudes and the political debates surrounding these issues. First, it investigates important outcomes that allows for a direct test of the often-heard arguments against more inclusive policies made by political theorists and politicians. Focusing on identification, acculturation, and perceived discrimination, the study addresses head-on common worries about ethnic identification, commitment, and sense of belonging among immigrants. By doing so, it improves on past studies that were limited in terms of geographical coverage and outcome measures. More specifically, the results add to those of Wright and Bloemraad (2012) by showing across a larger set of countries and additional outcomes that critics of policies who see them as a source of division are actually targeting policies that are associated with engagement and sense of belonging.
Second, I go beyond a comparison of outcomes across regimes and also focus on how the association between ethnic identity and these outcomes varies. This is important because critics and supporters of inclusive policies agree that they lead to more ethnic identification, but disagree on the consequences of this identification. It is one thing to look at how ethnic identity and perceived discrimination vary between inclusive regimes and more exclusionary ones, but it is also important to look at what kind of attitudes ethnic identity is associated with in countries with different incorporation policies.
Third, this study represents an important advancement from a measurement standpoint. Although other studies have investigated how policy regimes are associated with identity (e.g., Wright & Bloemraad, 2012), they are often limited by the questions available to them in large cross-country surveys. Concepts such as identity are complex and require attention to their multifaceted nature. The results presented here are based on multiple items scales that better reflect this complexity.
Finally, the study also examines Muslim immigrants more specifically, an important subgroup when it comes to debate over integration (Koopmans, 2013). Past comparative studies have either been unable to examine Muslims specifically due to lack of data (Wright & Bloemraad, 2012) or have had to use country of origin a proxy (Hainmueller et al., 2017). The study shows that patterns observed across regimes are robust to an analysis that concentrates solely on self-identified Muslims.
Incorporation Policy Regimes and Immigrants: Cultural Accommodation and Citizenship
Cultural accommodation and access to citizenship have been the focus of most of the extant literature on the empirical consequences of policy regimes and for good reasons. The choices that countries make on these two dimensions are expected to have important consequences for immigrants’ identity choices and ultimately for their incorporation in their new society. How easy should it be to become a citizen? How much should a country accommodate the different cultural practices of newcomers? The policies that serve as answers to these questions send a signal to immigrants about who is in and who is out, and about which identities are encouraged and accepted.
Cultural accommodation policies are often referred to as “multiculturalism policies.” The term “multiculturalism” has taken on different related meanings, which can sometimes lead to confusion. For instance. it has been used to refer to a normative political theory on how societies should deal with ethnic diversity (e.g., Kymlicka, 1995) or to simply state a sociological fact (a multicultural country). In the context of the present study, multiculturalism takes on a third definition: a set of governmental policies aimed at recognizing and accommodating cultural diversity (multicultural policies or MCPs).
Access to citizenship has also been hypothesized as influencing immigrants’ sense of membership and identity (for a recent example see Hainmueller et al., 2017). Citizenship here is understood in a limited sense, that is, as a legal status. Access to citizenship matters because when laws governing citizenship are more open, they send a welcoming signal to newcomers and perhaps reinforce a sense of membership. Understood in that sense, citizenship is also distinct from MCPs as granting citizenship does not necessarily come with a recognition of different cultural practices. Again, this distinction is made clear by recent trends where MCPs have been heavily criticized (Vertovec & Wessendorf, 2010) while citizenship laws have become less restrictive (Howard, 2006).
The recent empirical studies of the consequences of these policies for immigrant integration have offered contradictory findings. This is due in part to scholars disagreeing on the outcome that should be studied. Bloemraad (2007), for example, argues that the aim of a policy like multiculturalism is the sociopolitical integration of immigrants in the form of citizenship and political inclusion rather than economic outcomes such as labor market integration. An opposing view is that of Koopmans (2010), who presents economic outcomes as valid and shows, for instance, that the combination of MCPs and a generous welfare state leads to poor labor market participation. Others have taken a more pragmatic view and have measured social integration with questions about club membership and newspaper consumption (Hainmueller et al., 2017).
In this study, I follow Bloemraad and argue that in many cases the outcomes of interest are too far removed from the policies to lend themselves to meaningful analysis in a correlational framework. In cross-country analyses of complex policies of the types that immigration scholars study, the number of confounding variables rapidly becomes problematic, the possibility of omitted variable bias increases, and readers are often left trying to unpack the “blackbox” of causality. Koopmans (2010) does show that multiculturalism is associated with outcomes, such as high level of segregation and a strong overrepresentation of immigrants within the convicted criminal population, but he does not investigate the causal mechanism that would lead from multiculturalism policies to criminality or to ethnic enclaves. Although he does hypothesize that these outcomes are due to multiculturalism not creating strong incentives for language acquisition and interethnic contacts, he never gets at this mechanism directly.
When confronted with survey data that only lend themselves to correlational analysis, it is perhaps preferable to focus on more immediate outcomes. This research concentrates on three such outcomes: identity (national and ethnic), acculturation, and perceived discrimination. In addition to being causally closer to incorporation policies, these outcomes are also part of the mechanisms that are often proposed to explain more distant ones. The suggestion is not that the following analysis fully unpacks the causal mechanism linking policies to identification and acculturation, but that by looking at more immediate outcomes, the role of policies becomes more conceivable.
Ethnic and National Identity
Critics of multiculturalism view it as a vector of division, reinforcing ethnic identities to the detriment of a sense of belonging or, in the words of Schlesinger (1991), as a “disuniting” force. On the other side of the debate, political theorists argue that making these identities disappear would also mean making claims related to these identities impossible, create a less just society, and establish barriers to full participation in the public sphere (e.g., Kymlicka, 1995; Taylor, 1994; Young, 1990). 1 In other words, both critics and supporters agree that multiculturalism should lead to more identification with the ethnic group, but they disagree on the consequence of this identification. One groups sees it as leading to segregation and disengagement, the other as a path to meaningful participation and sense of membership.
A preconception that often tints these theoretical debates is that identification with the host and identification with the ethnic group are mutually exclusive. National and ethnic identity are are both multidimensional in nature (see Abdelal, Herrera, Johnston, & McDermott, 2008). One useful understanding of identity proposed by Citrin, Wong, and Duff (2001) and applicable to both is of a concept that has three dimensions: cognitive (identification as), affective (identification with), and normative (accepted criteria for group membership; see also Brewer, 2001; Brubaker & Cooper, 2000; Citrin & Sears, 2009). When studying identity choices, the focus is often on the first two. Clearly, however, it is the second dimension, identification with, that is seen as crucial for both critics and proponents of open incorporation policies. 2
Research in social psychology sees the relation between national identity and ethnic identity in the context of immigration through a two-dimensional model that recognizes that preservation of one’s culture and identification to the host society are conceptually distinct and vary independently (Berry et al., 2010; Citrin & Sears, 2009; Phinney, Horenczyk, Liebkind, & Vedder, 2001). Choices as to the importance given to each will be influenced not only by individual characteristics but also by policies such as cultural accommodation policies or citizenship access. Citrin and Sears (2009) write that these policies are likely to influence the distribution of identity choices, but because they only look at the ethnic and national identification in the United States, they cannot isolate the role of policies. At the same time, these single-country or small-N studies do give us some information on minorities’ identification choices. For instance, ethnic minorities do tend to have lower levels of attachment to the country compared with the majority (Elkins & Sides, 2007; Verkuyten & Martinovic, 2012). Canada, often given as an example of the successes of multiculturalism, does not seem to match this pattern. Wright and Bloemraad (2012) show that ethnic identification is more salient in Canada compared with the United States and that this attachment to the ethnic group is not linked with a rejection of the host society. Their conclusion is quite to the contrary: Immigrants in Canada feel more attachment to their new country than their American counterparts. In that case, a policy like multiculturalism seems to be encouraging a double identification. More easily attainable citizenship should also send a welcoming message to newcomers and incite identification with the host. However, Ersanilli and Koopmans (2010) find that it is not the case. Comparing France, the Netherlands, and Germany, they find that it is the former which is able to foster more national identification, whereas Germany is the least capable of doing so. The country with the most open citizenship laws, the Netherlands, finds itself in the middle. Ersanilli and Koopmans do not account for cultural accommodation policies where these three countries also significantly differ. What if a country sends a welcoming message associated with open citizenship without the deleterious effects attributed to multiculturalism by its critics? One expectation then is that countries with open citizenship but without MCPs should see a stronger identification with the host society and a weaker one with the ethnic group.
The argument behind countries making citizenship harder to access while also rejecting cultural accommodation is that this should incite immigrants to assimilate to the host society. Accordingly, if this logic holds, we should see immigrants in these countries adopt the host society’s identity to a greater extent. Conversely, scholars in the social identity theory tradition (e.g., Tajfel & Turner, 1979) argue that when group boundaries are considered impermeable, an individual will choose to solidify relationships with his in-group, in this case the ethnic group. According to social identity theory then, rather than fostering assimilation, countries with closed boundaries in the form of restrictive citizenship and low MCPs should increase identification with the ethnic group.
Competing expectations are also present when it comes to countries with relatively open borders and important MCPs. Opponents of such policies argue that they reinforce particular identities and offer no incentive to identify with the larger group. If they are right, countries in that group should see higher identification with the ethnic group and lower identification with the host. On the other side of the debate, proponents argue that cultural accommodation policies encourage participation in the larger society and make it possible for ethnic minorities to recognize themselves in the larger group. Consequently, ethnic identity and national identity should both be strong in these countries.
Acculturation
The assumption in most of the literature is that identification with the host is necessary for immigrants to want to engage with the majority. What if we were to look at more concrete attitudes that are directly linked to the relationships an immigrant wishes to have with his own ethnic group and with the majority? Perhaps identification with a group is not a prerequisite for a desire to interact with it, learn its language, or even get involved in relationships like friendship or marriage. It is these types of relationships and choices made about them by immigrants that scholars in social psychology call acculturation orientations.
The concept of acculturation has seldom been used in political science and when it has, it is mostly as a synonym for assimilation. For instance, Branton (2007) writes, “Acculturation, or cultural assimilation, refers to the process whereby immigrants assume the cultural values and norms of mainstream U.S. society (p. 301)” (see also De la Garza, Falcon, & Garcia, 1996; Lien, 1994). 3 In other fields, its meaning is quite different. The concept was first developed in anthropology dating back to Redfield, Linton, and Herskovits (1936) who define it as “those phenomena which result when groups of individuals having different cultures come into continuous firsthand contact, with subsequent changes in the original culture patterns of either or both” (p. 149).
This is also the understanding in social psychology, where most of the contemporary immigration-related work on acculturation is done. Although some conceptual ambiguity remains, scholars in social psychology have come to a consensus on its meaning and the processes behind it (Sam, 2006). For them, acculturation refers to the changes that occur at the individual level when an immigrant experiences intercultural contact with his or her new culture (e.g., Berry et al., 2010). According to the model of acculturation developed by Berry (1980), immigrants face two “choices” when entering in intercultural contact: whether to maintain their ethnic culture and identity and whether relationships with the host should be pursued or avoided. Similar to identification, these choices about relationship with the host and with the ethnic group are considered to be independent of one another and may be conscious or not. 4
The immigrant’s strategies on these two orthogonal dimensions will situate her in the realm of acculturation orientations. Figure 1 maps the four possible acculturation orientations based on these choices. Viewed through this model, assimilation becomes one of the four acculturation orientations, one where an immigrant rejects her own ethnic culture while embracing the majority culture and seeking relationships with the host only. Conversely, an immigrant is separating herself from the host if he only seeks relationships with his own group. The third orientation, integration, occurs when an immigrant embraces relationships with both her ethnic group and with the majority. Finally, rejecting both the ethnic group and the host leads to marginalization. 5

Acculturation orientations.
The case for acculturation as an outcome of interest partly rests on the argument that acculturation orientations can be conceived as measures of commitment and attachment to the host society that are distinct from a direct measure of identification. Most critics of multiculturalism stress the need for a strong common identity that will generate a commitment to the larger society and trust in conationals (Barry, 2002; Miller, 1995). Perhaps, is it possible to see commitment and attachment as distinct from identification, an argument made by Mason (2010). If identification with the host is not necessary to foster a desire to enter in relationship with the larger society, it is conceivable that some policy regimes are more conducive to such relationships.
In a revised model of Berry’s original acculturation model, Bourhis et al. (1997) suggest that these orientations do not emerge from a vacuum and they introduce policies as one of the explanatory factors driving acculturation orientations (p. 373). Bourhis and his colleagues divide incorporation policies into an ideological continuum with four ideal types of state ideologies (pluralist, civic, assimilation, and ethnist). However, they do not derive specific expectations about the effects of these different ideologies on acculturation orientations and do not ground them empirically. These ideologies are analogous to the early work in political science based on conception of national definition as civic or ethnic and suffer from the same shortcomings. What are the expectations for the relationship between policy regimes and acculturation orientations? First, we should observe more variation across policy regimes within each orientation than between them: Integration should be the favoured orientation for all policy regimes and marginalization should be the least favoured one (Verkuyten & Martinovic, 2012). When looking within orientation, expectations differ whether one adheres to the arguments of proponents of MCPs and open borders or with those of partisans of more restrictive policies. If the former are right, open policies should help minorities feel part of the larger society while encouraging the preservation of ethnic ties. Accordingly, these policies should increase integrationist attitudes. On the contrary, if partisans of more restrictive policies are right and open policies do not incentivize intergroup contact, we should see immigrants in these countries separate themselves from the larger society and opt to have relationships primarily with their ethnic group. Conversely, more restrictive countries should promote assimilation more than other types of regimes because they reinforce national identity and do not recognize ethnic groups’ cultural differences.
The Interplay Between Ethnic Identity and Acculturation
One possibility is that some policy regimes might be able to create a sense of belonging and commitment among their minorities without it being necessarily coupled with a strong identification with the majority. This is an important aspect of the relationship between ethnic group identification and attitudes toward the host society. If ethnic identity is strong but not coupled with a sense of commitment with the host, this could lead to backlash among the majority. A country that is able to help minorities preserve their identity while also fostering a sense of belonging and commitment to the host society might be able to make this resistance from the majority less severe.
What should we expect as identification with the ethnic group grows? The answer to that question is at the basis of much of the criticism directed at policies that recognize cultural diversity. In terms of acculturation orientations, an increase in ethnic identification can have two competing consequences. On one hand, it can lead to an increase in the value given to relationships with the ethnic group and a rejection of the host, that is, it can increase separation attitudes. On the other hand, it can increase the value given to relationships with both groups, that is, it can increase integration attitudes.
Policies are likely to come in play in determining which of these two outcomes an increase in ethnic identity will lead to. More restrictive regimes, for example, are likely to make identification with the ethnic group and relationships with the host antithetical by not recognizing ethnic identities as legitimate in the public sphere and by making it hard to gain legal membership. If proponents of MCPs are right, countries with more open policies should see ethnic identity being associated with integration. On the contrary, if it is the critics that are right, a heightened sense of ethnic identity should lead to separation.
Data and Method
Measuring Policy Regimes
For multiculturalism policies, Banting and Kymlicka (2006) have built a comprehensive index composed of scores on eight different policy areas (multicultural policy index [MPI]). These policy areas are official affirmation of multiculturalism, presence of multiculturalism in schools curriculum, inclusion of ethnic representation in public media and licensing, exemption from dress code for minorities, funding of ethnic organizations to support cultural activities, funding of bilingual and mother tongue instruction, affirmative action, and dual citizenship. A country receives 0 point if it has not adopted a given policy, half a point if it has done so only partially, and a full point if it has fully adopted it.
The citizenship policies index (CPI) was elaborated by Howard (2006) and measures three aspects of citizenship policies: whether a country allows dual citizenship, whether it grants jus soli citizenship, and the minimum number of years of residency that are required to acquire citizenship (see also Howard, 2009). In a comparison of different indices measuring citizenship policies, Helbling and Bauböck (2011) find that they are highly correlated even if they are based on different indicators and that Howard’s measure is the simplest and most parsimonious.
In addition to these two policy areas, Goodman (2010) has also developed an index, civic integration index (CIVIX), that measures civic integration requirements in terms of country knowledge, language, and culture. Goodman sees the CIVIX as complementary to a simple measure of access to citizenship like Howard’s. Her data show that, as of 2009, a series of European countries had adopted more restrictive requirements. However, earlier results from 1997 show almost no variation among the 14 countries she studies (Goodman, 2010, p. 764). Because the individual data used here come from this earlier period and because Howard’s measure of citizenship access has been recognized for its parsimony and validity (Helbling & Bauböck, 2011), the analysis concentrates on citizenship as legal status coupled with multiculturalism.
Once we plot immigration countries based on these two indices, the picture that emerges in Figure 2 is one of three distinct groups of countries. One group with relatively low (or closed) citizenship policies as well as low MCPs (Low-Low), one with few MCPs but more open citizenship policies (Low-High), and finally a third group displaying both open citizenship policies and comparatively more MCPs (High-High).

Incorporation policies across contexts: A typology.
These three different policy regimes are at the basis of the empirical analysis and serve as the Level-2 variable. 6
This graph of policy regimes is similar to the one found in Wright and Bloemraad (2012) and based on a comparable strategy used by Koopmans, Statham, Giugni, & Passy (2005). The fact that the International Comparative Study of Ethnocultural Youth (ICSEY) data set only contains data for two countries in the Low-Low category—Norway and Germany—is less than ideal. However, these two countries represent a good range of values in this group with Germany being the upper boundary and Norway close to the lower one.
The ICSEY
In addition to the country-level data, the analysis employs individual-level data from a data set that has yet to be used in political science. The ICSEY is a survey of 26 groups of immigrant youth (aged 13 to 18 years) in 12 countries and was conducted from the end of the 1990s to the early 2000s (for more details on the survey methodology see Berry et al., 2010). What makes this survey unique, in addition to its large sample of immigrants (N = 5,366), is its focus on identity and different acculturation orientations among teenagers of immigrant background.
One issue when using cross-country immigrant data is that characteristics specific to a given immigrant group might influence the outcome. Ideally, a data set would sample the same group in all countries or all groups in the countries that are part of the studies. In sampling groups, the ICSEY data set does a bit of both. It samples immigrants with a Vietnamese or Turkish background when possible—seven countries for the former, six for the latter—but also includes a large number of different groups to better reflect the immigrant composition of a given country. The only exception is the United Kingdom that only sampled immigrants of Indian origin. Consequently, the country was dropped from the analysis. 7 Israel is also excluded because of its singular history. The analysis is consequently based on 10 of the 12 countries contained in the original data set.
The objective behind the ICSEY study was to look at the effect of identification and acculturation on a series of psychological and sociocultural adaptation measures. Here, instead of using these measures as explanatory factors, the study uses them as the outcome and evaluates how they vary according to incorporation policies contexts when controlling for other individual-level factors.
Identification
The first two outcomes are additive measures of identification with the host society (α = .90) and identification with the respondent’s own ethnic group (α = .86). Both are additive index (see the Supplemental Materials for descriptive statistics and question wording of the different index used in the analysis.) Conceptually, these identification measures are a mix of two of the three dimensions proposed by Citrin et al. (2001). The first question—I feel that I am part of . . .—can be conceived as being about both the cognitive dimension of identity (identification as) and the affective dimension (identification with). All the other questions are measuring the affective dimension. It is possible that some of these questions are more likely to depend on context in ways that bias our results. For example, some might argue that it is less embarrassing and uncomfortable to be ethnic in High-High countries, and that, as such, the embarrassed and uncomfortable people in these countries are different from the embarrassed and uncomfortable people in Low-Low countries. The results that follow are robust to versions of the dependent variables that exclude these more “context-dependent” questions.
Acculturation
Acculturation orientations measures consist of four additive indices assessing the extent to which respondents display attitudes in line with each of the four acculturation orientations outlined in Figure 1: assimilation (α = .61), integration (α = .48), separation (α = .68), and marginalization (α = .59). These orientations are measured based on respondents’ preferences in five domains: cultural traditions, language, marriage, social activities, and friendship (Berry, Kim, Power, Young, & Bujaki, 1989).
Perceived discriminaiton
Finally, the index for perceived discrimination is composed of nine different questions measuring whether the respondent has been teased because of his or her ethnic background, whether his or her ethnic groups is treated unfairly, and so forth (α = .85).
In addition to these measures, the data set contains other sociodemographic variables such as religion. With Muslim immigration being the target of most of the backlash against cultural accommodation and open borders, the presence of respondents’ religion in the data set allows for a more targeted investigation of this group in particular. The fact that respondents are teenagers is arguably another strength of the data set. Teenagers are in the midst of forming their identity and their choices will be influenced both by their family—thus reflecting their parents’ orientations—and by educational programs, more likely to be affected by policies, especially multiculturalism.
Analytical Strategy: Visualizing Multilevel Data
Analyzing the relationship between policies and individual attitudes across countries is inherently a multilevel proposition. Respondents are nested within countries and their respective policy contexts and the goal is to identify the impact of these policies once we control for individual-level predictors of the outcome. In the present case however, two important assumptions of multilevel modeling are problematic. First, the asymptotic properties of maximum likelihood estimators (MLEs) on which multilevel models rely necessitate a good number of cases (Bowers & Drake, 2005). 8 With 10 countries as the Level-2 sample size, this is far from ideal (Maas & Hox, 2005; Stegmueller, 2013). Second, MLE also assumes that these Level-2 cases are a representative random sample of a given population. The 10 countries in the ICSEY cannot be considered a random sample of immigration countries. Rather, they are more akin to a convenience sample: they are the countries for which we have individual-level data.
Given these limitations, I follow the suggestions of Bowers and Drake (2005) and use graphical visualization of these multilevel data as the analytical strategy. 9 Variants of this “two-stage” strategy are employed by Wright and Bloemraad (2012) and (Goodman & Wright, 2015) to study closely related topics to the one under investigation here and by Sides and Citrin (2007) to look at opposition to immigration in European countries. Like them, this study investigates the relationship between policy contexts at the country level and a series of dependent variables at the individual level. For this reason, the dependent variables are regressed on individual-level predictors when the data are pooled within policy contexts identified in Figure 2. The estimates and confidence intervals of the intercepts for each of the policy contexts are then graphed. The different individual-level variables are coded such that the intercepts correspond to the predicted outcome for an immigrant who is 17 years old, has been in the country for 5 to 10 years, does not have citizenship, and has parents whose training is in an unskilled profession. 10 The justification for using such a hypothetical immigrant is similar to Wright and Bloemraad’s: It is exactly the type of immigrant about which incorporation concerns are more prevalent. This type of immigrant—without citizenship, short residence in the country, and lower education/unskilled— is also less likely to identify with the majority, making it a harder case for double identification (e.g., Zimmermann, Zimmermann, & Constant, 2007). 11
Controlling for those sociodemographic variables is also essential in making sure that we are comparing apples with apples. Some countries, Australia and Canada, for instance, select a large share of their immigrants based on education and skills. If the analysis did not control for parents’ skills, it would overestimate the role that inclusive policies play, given that those two countries are in the High-High category. By comparing immigrants with unskilled parents in all three regimes, I make sure that the results are not due to a variation in immigrant selection based on education or skills. One important point on which I diverge from Bloemraad and Wright’s empirical strategy is that they restrict their analysis to the models’ intercepts because they are not interested in the individual-level relationship between two variables and the variation in this relationship across Level-2 units. Here, however, one of the main objectives is to investigate how ethnic identity translates into different acculturation orientations and into perceived discrimination across contexts. To look at this relationship more specifically, I follow the procedure used by Bowers and Drake (2005) and also graph the slope coefficient for ethnic identity for each Level-2 unit. Again, this is similar to the “two-stage” procedure used by Sides and Citrin (2007). In the first stage, I estimate an individual-level model for each policy regime. These models include the same covariates specified earlier but also add ethnic identity. I then extract the intercept and the ethnic identity coefficient to form a second “data set” where the unit of analysis becomes policy regimes. Finally, I plot these policy regimes’ intercepts and slopes for each acculturation orientation and for perceived discrimination.
In the analysis, I address issues of missing data by generating six data sets through conditional multiple imputation (Kropko, Goodrich, Gelman, & Hill, 2014; Su, Yajima, Gelman, & Hill, 2011). All estimates presented in the “Results” section are consequently calculated by fitting a model to each imputed data set, after which estimates and standard errors are calculated according to Rubin’s rules (Rubin, 1987).
Results
Ethnic and Majority Identity
Figure 3 plots the intercept values for identification with the majority and with the ethnic group when these identity measures are regressed within policy contexts. Throughout the analysis, all the dependent variables are continuous indices (from 0 to 1) and the estimator is an ordinary least squares (OLS) regression.

Identification with ethnic and majority identity by policy contexts.
The first thing that stands out when looking at Figure 3 is that, across policy regimes, immigrants tend to identify more with their ethnic group (top panel) than with the majority group (bottom panel). This would indicate that no regime is able to foster a higher level of identification with the majority than with the ethnic group. Given that the hypothetical immigrant represented by this intercept does not have citizenship and has only been in the country for 5 to 10 years, this does not come as a surprise.
As Bowers and Drake (2005) point out, when plotting these intercepts, if the dots and their confidence intervals are clearly stacked on top of one another, we would then have to conclude that there is no variation across policy regimes. This is not what we observe for immigrants’ identification. For both ethnic and majority identity, there are differences across regimes, but these difference are small, especially in the case of ethnic identity with an intercept value of .64 for High-High countries and of .58 for Low-Low countries. This would indicate that more open policies are indeed associated with a higher level of ethnic identification but only slightly. Policies that are designed to decrease ethnic identification (Low-Low), do not appear to fulfill all their promises either. Moving to the second panel that plots intercepts for majority identification, we see that these Low-Low countries are associated with less identification with the majority than the other two types of regimes. As hypothesized, when the welcoming message of open citizenship is coupled with few MCPs, countries are able to foster more identification with the host society, but this identification still goes hand in hand with a high level of ethnic identification. In other words, double identification is more likely in Low-High regimes than in the other two.
The ICSEY also contains a sample of majority youth, which can be used to assess whether the pattern observed in identification with the majority among immigrants is just a reflection of how people generally identify with the majority group in the different regimes. These intercepts are represented by the three squared dots that are lined up almost completely on top of each other to the right of the .75 mark. Clearly, the pattern observed among immigrants are not reflecting a more general pattern in their new societies. Unsurprisingly, majority respondents identify more with the majority than immigrants do, but there is no significant difference between policy regimes in the level to which youth majority members identify with their own group. 12
Acculturation Orientations
Whereas identification was measured with questions about membership and pride, acculturation orientations are measured with a series of questions on marriage, cultural traditions, language, friendship, and social activities. Acculturation orientations are based on the importance given to relationships with the ethnic group and the majority. For example, an immigrant is displaying an integrationist orientation if marrying a coethnic or marrying a host are both seen favorably. It is also important to remember that these orientations are not necessarily mutually exclusive. In other words, it is possible for a given individual to see integration as the preferred orientation and to strongly agree with the questions on integration, but to also agree, albeit less strongly, with questions measuring separation.
Figure 4 plots the intercept values for each acculturation orientation when these orientations are regressed within policy regimes. Here again, we see that there is no difference across regimes in terms of the preferred acculturation orientation. For all regimes, immigrants are more likely to see both relationships with the majority and with the ethnic group as important. The second preferred orientation is separation—favoring relationships with the ethnic group over relationships with the majority—followed by assimilation and marginalization. The only exception to this pattern is in Low-Low countries where immigrants show similar levels of marginalization and assimilation orientations.

Acculturation orientations by policy contexts.
Where we see differences, however, is in the extent to which immigrants adopt each acculturation orientation across contexts. For instance, immigrants in Low-Low and Low-High regimes exhibit more separation attitudes than those in High-High regimes. In terms of assimilation, it is not surprising to see immigrants from High-High regimes being the least likely to favor this orientation, given the importance of ethnic identification in these countries. Combining these two results together, offers an interesting picture in light of the criticism against more open policies. The logic behind restrictive policies is to encourage assimilation into the larger society, whereas one criticism against policies like multiculturalism is that it encourages ethnic segregation. Results displayed in Figure 4 indicate that immigrants in High-High countries do in fact assimilate less than in other regimes but also separate themselves less. With the exception of marginalization, it is also in the separation orientation that we find the largest gap between two regimes with a 10 percentage points difference between High-High countries and Low-Low countries. In sum, when we examine separation and assimilation orientations, open policies are not associated with a hunkering down effect among immigrants and more restrictive policies are not associated with more assimilation either. Looking at integration attitudes, it is in Low-High countries that this orientation is more important, on average, followed closely by High-High countries. This is somewhat surprising in light of the previous finding on identification. Figure 3 showed that, compared with immigrants in Low-High regimes, those living in High-High regimes identified less with the host society and more with their ethnic group, but these identification choices do not seem to be associated with a rejection of the host society given how important integrationist attitudes are in High-High regimes. Of course, integration is the preferred orientation across all policy contexts, which again represents an interesting finding when considering the negative consequences often attributed to multiculturalism. This high level of integration orientation across regimes is also interesting in light of what the data on identification showed—that immigrants in all regimes identify more with their ethnic group than with the majority. The lack of identification with the host does not seem to impede or diminish the importance given to relationships with the larger society and more open policy regimes do not seem to be reinforcing these boundaries even if they are in fact associated with more ethnic identification.
Ethnic Identity, Acculturation Orientations, and Perceived Discrimination
The main disagreement between proponents and opponents of more open incorporation policies is not whether it reinforces ethnic identity—there is agreement on this matter—but whether this focus on ethnic identification leads to negative consequences on intergroup attitudes. Critics of multiculturalism, for instance, argue that a reification of ethnic differences makes it impossible to foster a sense of membership with the host society and leads to ethnic segregation. Results presented thus far show that immigrants living in countries with open citizenship and a higher score on the MCP index tend to identify more with their ethnic group than immigrants in other policy regimes. Conversely, they also tend to identify less with the majority than immigrants in countries that do not couple open citizenship with multiculturalism. This could lead one to conclude that critics are vindicated in their fear that MCPs lead to a loss of a common sense of belonging or membership.
Perhaps, however, some policy arrangements make it possible for ethnic identifiers to feel accepted as “members” of their new polity and to seek relationships with the majority even if they identify more with their ethnic group. If critics of open policies are right, these policies should, on the contrary, lead ethnic identifiers away from the majority and reinforce their desire to engage solely with their ethnic group. One way to understand the consequences of ethnic identification across policy regimes is to look at the relationship between ethnic identity and acculturation orientations and between ethnic identification and perceived discrimination.
Acculturation
Like previous figures, Figure 5 plots intercept values for each policy regime, but because the interest now lies in the effect of ethnic identity across policy regimes, it also plots the coefficient for ethnic identity. Recall that these values are coefficients when acculturation orientations are regressed within policy regimes on individual-level predictors. The other variables in the model are the same as those in previous sections. The ethnic identity index is centered at 0 (from –.5 to .5). Consequently, the intercept can be interpreted as the average level of a given acculturation orientation for a hypothetical immigrant at the midpoint of ethnic identification.

Acculturation orientations and ethnic identification by policy contexts.
The results displayed in Figure 5 should alleviate the concerns that cultural accommodation policies, by reinforcing ethnic identity, will lead to a rejection of the host society. Looking at the top left panel, we see that after adding ethnic identification to the model, integrationist orientations are still the most likely orientation across regimes with all three intercepts hovering around .75. It is when we turn to the ethnic identity coefficients that we see an interesting variation: ethnic identity is linked to more integrationist attitudes in Low-High and High-High regimes but not in Low-Low regimes.
But what about separation? Maybe ethnic identifiers are also more likely to want to prioritize relationships with their own group. Indeed this is what we observe in the separation panel (top-right). Across all three regimes, ethnic identity is associated with more separation attitudes and the coefficient increases as policies become more open—albeit only slightly. Looking at the intercepts, the patterns observed in Figure 4 remain once we add ethnic identity: It is in High-High regimes that we see the lowest level of separation attitudes.
The results for assimilation are, to some extent, also in line with what was expected. We again see that, on average, regimes do not differ significantly in the level of assimilation orientations displayed by immigrants. There are, however, differences in the impact of ethnic identification. A higher identification with one’s own ethnic group decreases assimilation orientations—that is, attitudes and behavior that are favoring the host society and rejecting the ethnic group—but this relationship is weaker in Low-Low and Low-High countries than it is in High-High countries.
Finally, a stronger sense of ethnic identity is linked to lower levels of marginalization in High-High and Low-High regimes but not in Low-Low regimes. Recall that marginalization orientations refer to an individual rejecting relationships with both her own ethnic group and with the host society. Immigrants from Low-Low countries also exhibit more marginalization attitudes on average. Another way of putting these results is to say that in regimes that combine low levels of multiculturalism policies with more restrictive citizenship policies, the only acculturation orientation that sees a gain when ethnic identification increases is separation, whereas High-High and Low-High countries also see an increase in integration attitudes.
In interpreting this finding, it is important to keep in mind the small differences in the average level of ethnic identification across regimes. In other words, high identifiers are not much more likely in High-High countries. What immigrants in these countries are more likely to do, however, is to seek relationships with the host society in domains such as language, marriage, and social activities.
Perceived discrimination
Are more open incorporation policies undermining common membership? So far, the evidence points in the opposing direction. The last outcome under investigation considers perceived discrimination an important measure of social inclusion that might also be affected by policy regimes (Hainmueller et al., 2017; Wright & Bloemraad, 2012). As the top panel of Figure 6 makes clear, perceived discrimination increases incrementally as policies become more restrictive. The index moves from .26 in High-High regimes to .31 in Low-High regimes and finally to .37 for countries with low scores on both citizenship and multiculturalism policies. This result is confirmation of the important but limited evidence offered by Wright and Bloemraad (2012) who also found a monotonic decrease in perceived discrimination across regimes.

Perceived discrimination and ethnic identification by policy contexts.
More importantly, this pattern is also reflected in the coefficients for ethnic identity displayed in the bottom panel. That coefficient is not different from zero in High-High countries means that ethnic identity is not associated with a higher perceived discrimination in these regimes. The other two regimes do, however, have a positive coefficient for ethnic identity. In sum, ethnic identifiers do not perceive more discrimination in High-High countries but they do in more restrictive regimes where immigrants are also more likely to perceive discrimination.
Muslims Immigrants
Given their singular position in immigration debates (Koopmans, 2013), Muslims are often seen as a group that is harder, even impossible, to integrate in the larger society (Modood, 2005). 13 Consequently, it is possible that Muslim immigrants explain some of the results above. Perhaps High-High countries are able to foster integrationist attitudes among ethnic identifiers because those countries’ samples have fewer Muslim immigrants.
There is a total of 1,057 Muslim respondents in the original data set, but they are not equally distributed across policy regimes: 497 of these Muslims respondents are in High-High countries, 401 in Low-Low countries, and only 159 in Low-High countries. 14 In other words, they form a larger proportion of the sample in Low-Low regimes (50%), followed by High-High regimes (30%) and then Low-High regimes (9%). This imbalance, in itself, does not necessarily put into question the patterns presented earlier. One way in which the distribution of Muslim respondents would explain the differences across regimes would be if Muslims identify more with their ethnic identity and less with the majority, with such identification leading to more separation and less integration. If this pattern is observed across regimes, then the results in Low-Low countries compared with High-High countries could be explained by a larger Muslim share of the sample in Low-Low regimes. The case for policies as the explanation would be harder to make. To verify that variation in the religious composition of each regime type is not driving previous findings, the analysis is replicated on the Muslim subset.
In terms of identification with the majority and identification with the ethnic group, the patterns are similar to those observed in the full sample. Muslim immigrants in Low-High countries do identify even more with the majority but given the sample size, this identification is not different from the one seen in the full sample. In terms of acculturation orientations, there are no important differences across policy regimes and integration remains the favored orientation. Figures showing coefficients for identification and acculturation can be found in the Supplemental Materials.
Looking at the relationship between ethnic identification and acculturation displayed in Figure 7, we see that ethnic identification does generate more separation than it did in the full sample. However, integration is still high—around .75—for the three regimes. In other words, based on this cursory glance at Muslim immigrants, they display the same attitudes across policy regimes and thus cannot explain by themselves the differences found in previous sections. Muslim respondents do seem to differ from other immigrants in the effect that ethnic identity is having on acculturation orientation, on integration more specifically. We do not observe the same positive relationship between identification with the ethnic group and more integrationist attitudes in High-High countries. That being said, we do not observe a negative relationship either. In Low-High and High-High countries, the slope is not different from 0. Low-Low countries do, however, have a slope different from 0, but a negative one, meaning that in these countries increasing ethnic identification leads to less integrationist attitudes for Muslim immigrants. In fact, when we look separately at non-Muslims and Muslims (see Supplemental Materials), ethnic identification is increasing integration among non-Muslims in High-High countries but having no effect among Muslims, whereas in Low-Low countries, it is having no effect among non-Muslims and decreasing integration among Muslims.

Acculturation orientations and ethnic identification of Muslim respondents by policy contexts.
In sum, Muslims do seem to be different from other immigrants in the effect that identification with the ethnic group is having on integrationist attitudes but there is also an important difference in this difference across regimes. The pattern observed in Low-Low countries is not due to the fact that they have a larger proportion of Muslim immigrants, but rather that unlike other policy regimes, these countries seem to be unable to mitigate the effect of ethnic identification on integration.
Discussion and Conclusion
Examples of politicians blaming policies that recognize cultural pluralism for a supposed decline in social cohesion abound (see Koopmans, 2013; Vertovec & Wessendorf, 2010). What often follows from such statements is a call for measures that are more assimilationist in nature and that make citizenship a “crown” on integration achievements rather than a “catalyst” for that integration (Hainmueller et al., 2017). The results from the present analysis show that, contrary to many of those assertions, a strong common identity does not appear to be necessary to “bring together” immigrants and the receiving society. Many immigrants displayed integrationist attitudes—a desire to have relationships with both the majority and their own ethnic group—while identifying more with their own ethnic group. More importantly, the results also showed that if the main goal of incorporation policies is to foster these integrationist attitudes and socialize immigrants into the larger society, countries with open citizenship laws and extensive multiculturalism policies are the ones best equipped to achieve it.
Focusing on identity and acculturation in three distinct policy regimes (High citizenship-High MCPs, High citizenship-Low MCPs, Low citizenship-Low MCPs), the study shows that these regimes do not differ significantly in the extent to which immigrants identify with their ethnic group. Where they do differ is in their identification with the host. In line with critiques of MCPs, countries with open citizenship policies that also have cultural accommodation policies were less likely to promote national identification among their immigrant population. Countries with a combination of open citizenship—perhaps sending a welcoming message to newcomers—and few MCPs were the ones most able to foster identification with the majority among their immigrant population.
The study then turns to acculturation orientations, a concept capturing attitudes toward relationships with the majority group and with one’s own ethnic group. Considering four different acculturation orientations, results show that High-High countries were the ones most able to foster integrationist attitudes. It is in these countries that immigrants were more likely to see relationships with both the host and their ethnic group as important. These relationships encompass a wide-ranging set of situations, such as friendship, language choice, and marriage.
The most striking finding is that High-High and High-Low countries were able to create an environment in which ethnic identity is associated with more integrationist attitudes. In the Low-Low regimes, the only acculturation orientation that saw an increase with higher ethnic identification was separation. These results should help put to rest the worry that ethnic identity precludes commitment to the larger society. Although the evidence presented here is only a first step in understanding the link between policies and these outcomes, it demonstrates that it is in fact the case that ethnic identity is associated with less commitment to the majority but not in countries with policy regimes that have open citizenship and extensive MCPs. When we combine these results with those of Wright and Bloemraad (2012), who find that immigrants from these same High-High countries were not less likely than their counterparts in Low-High or Low-Low countries to be engaged in their political community, we have a picture at odds with the recurrent rhetoric against cultural accommodation policies.
Of course, it is possible that these results are due to a selection bias causing Low-Low countries to have to deal with immigrants who are different—that is more difficult to integrate—than High-High countries. To test for this possibility, an additional analysis was conducted with the sample restricted to Muslim respondents, a group that has been presented as a hard case for integration. Overall, these immigrants do not differ significantly from the rest of the sample in their identification and acculturation orientations.
In addition to being able to look at Muslims more specifically, the study had another particularity: it focused on immigrant youth. The fact that young people are in more direct contact with some policies through school and are at a moment in their life when they are less resistant to change might mean that the results presented here overestimate the impact of incorporation policies in the large immigrant population. Nonetheless, the impact of policies on immigrant youth is crucial when examining identification, acculturation, and sociopolitical integration more generally. First, the political socialization literature highlights that many predispositions are acquired during those formative years and are durable well into adult life. Second, the effect of policies on young immigrants can also trickle up to their parents. Political socialization in immigrant families can be a bidirectional process (Wong & Tseng, 2008).
One limitation of the study is that the immigrant data are from the late 1990s and early 2000s and the patterns observed might have changed. For instance, new policies not encompassed by citizenship access or MCPs have been enacted and may have an impact on some of the relationships highlighted here. As mentioned earlier, one recent evolution of incorporation policies has been the introduction of a series of civic integration measures of the type studied by Goodman (2010)—for example, a stronger emphasis on language acquisition prior to arrival, citizenship ceremonies reinforcing identity elements, and so forth. Although some recent work suggests that these policies do not affect immigrant integration (Goodman & Wright, 2015), adding new policies to reflect the environment in which immigrants find themselves today is an important path for future research.
Another consequence of the period covered by the data is that it presents a picture of a period where anti-immigrant discourse was perhaps less present or at least less focused on multiculturalism. The recent backlash against multiculturalism, even if primarily in discourse rather than in actual policies (Banting & Kymlicka, 2013; Vertovec & Wessendorf, 2010), potentially has consequences for immigrants’ identification processes and for intergroup relationship more generally. Future studies looking at the impact of policies on immigrants’ attitudes should endeavor to include this change in discourse, perhaps by including political parties manifestos, media content, or majority opinions in the analysis (see, for example, Just & Anderson, 2014).
Even if we were to be able to add more contextual variables to the analysis, an important issue remains: the availability of data when it comes to the comparative study of immigrants. The data used in this study made it possible to examine important outcomes to an extent that previous studies had been unable to do. Yet, although it represents an improvement on previous work, the investigation was limited to immigrants from 10 countries. As is often the case when large-N research is developing, the hope is that we will soon have more individual-level data to combine with the already existing country-level measures to further examine the relationships that the present work has highlighted. Given what is at stake, and the importance of these empirical questions, this is an essential and necessary step.
Supplemental Material
Incorporation_Policies_and_Immigrants-Appendix – Supplemental material for Do Incorporation Policies Matter? Immigrants’ Identity and Relationships With the Receiving Society
Supplemental material, Incorporation_Policies_and_Immigrants-Appendix for Do Incorporation Policies Matter? Immigrants’ Identity and Relationships With the Receiving Society by Charles Breton in Comparative Political Studies
Supplemental Material
Replication_Files_CPS-17-0565.R1 – Supplemental material for Do Incorporation Policies Matter? Immigrants’ Identity and Relationships With the Receiving Society
Supplemental material, Replication_Files_CPS-17-0565.R1 for Do Incorporation Policies Matter? Immigrants’ Identity and Relationships With the Receiving Society by Charles Breton in Comparative Political Studies
Footnotes
Acknowledgements
I thank John W. Berry for making the International Comparative Study of Ethnocultural Youth (ICSEY) data available. I also thank Irene Bloemraad, Richard Johnston, Antje Ellermann, Fred Cutler, and Gregory Eady for comments on previous versions.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was made possible by the financial support of the Social Sciences and Humanities Research Council (SSHRC postdoctoral fellowship—award number 756-2014-0071).
Notes
Author Biography
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
