This is an open access article distributed under the terms of the Creative Commons
Attribution License (Europe has encountered large-scale immigration over the last decades (Castles & Miller, 1994). Fears about immigration and threats to national identity are at the core of public debates, and they have inspired research into the social identities of immigrants (Verkuyten et al., 2019). Public discourses on immigration and diversity emphasize that societies need a sense of common belonging reflected in a shared national identity to (re-)enforce social cohesion (Joppke, 2004). Yet little is known about the development of host-national identification among recent immigrants.
From the immigrant perspective, migration triggers deep adjustments within their social identities (Amiot et al., 2007). Arriving in a new country, immigrants’ sense of identity is primarily defined by their origin country, which distinguishes them from members of the receiving society, and, consequently, becomes chronically salient (Phinney, 1993). When entering a country as a religious minority, religion becomes even more influential for migrants’ identity since it serves as a strong boundary marker between their in-group and out-group (Verkuyten, 2007). This is particularly true in Europe, where Islam has been described as a ‘bright boundary’ and an obstacle to the integration of Muslim immigrants (Alba, 2005). This juxtaposition of several identities stemming from their country of origin, their religion, and their new country of residence, might be difficult to reconcile within individuals’ self-concept, particularly when these identities are perceived to be incompatible. Identity incompatibility refers to individuals’ sense that it is impossible to integrate multiple identities within one overall identity (Sindic & Reicher, 2009). The challenge of managing several, potentially conflicting identities is particularly relevant for recently arrived immigrants (Deaux & Verkuyten, 2014), raising questions as to when newly arrived immigrants develop a sense of host-national belonging, and how this development relates to important minority identities.
Against this background, this study has two aims. First, we investigate changes in levels of identification of recent immigrants during the first years after arrival in relation to three identity categories: (1) country of settlement, (2) origin country, and (3) religion. Using a three-wave cross-lagged panel design, we examine how these three identifications are related to each other and whether these relations change over time. Second, we study whether identification patterns are affected by perceived value incompatibility. We study recent arrivals to the Netherlands from Bulgaria, Poland, Spain, and Turkey, to replicate theoretical expectations across a range of immigrant groups.
We extend previous research in three ways. First, we study the development of host-national identification among recently arrived immigrants in conjunction with two important minority identities. Covering two distinct yet overlapping minority identities, and their relation to the development of host-national identification, enables us to identify different compatibility and identification patterns across time and immigrant groups. Second, we investigate how implicit and explicit measures of identity (in-)compatibility are related. Third, our longitudinal approach, utilizing three waves of panel data, enables the analysis of within-person changes and provides a stronger basis for causal inferences.
Immigrants’ Multiple Identities
According to Social Identity Theory (SIT; Tajfel & Turner, 1979), people develop a social identity based on their membership in a group. Individuals try to achieve a positive and distinct self-understanding by dividing the world into in- and out-groups. Self-Categorization Theory further emphasizes that different identities can become salient in different situations (Turner et al., 1987).
Contexts of change like moving to a new country increase the salience of group membership and, consequently, trigger categorization processes that can lead to the reformation of one’s identity (Amiot et al., 2007). According to the acculturation literature, immigrants adopt different strategies in redefining their origin-national identity and developing a host-national identification (Hutnik, 1991). Some immigrants solely preserve their origin-national identification (separation), others adopt the host-national and relinquish their origin-national identification (assimilation). Still others combine their origin and host-national identification (integration) or turn away from both (marginalization).
Previous literature suggests integration is the strategy most favored by immigrants and most adaptive for their well-being (Nguyen & Benet-Martínez, 2013). However, how such integrated or dual identities develop among adult immigrants has not been studied longitudinally (Wiley et al., 2019). Cross-sectional research among immigrants suggests dual identity develops when a significant host-national identity is added to an already established ethnic identity (Fleischmann & Verkuyten, 2016). Our three-wave panel data allows us to capture the dynamics of identity development directly after migration. We set out to study changes in host-national, origin-national and religious identification, and examine their associations to understand when and for whom these identifications are compatible.
Immigrants’ Identity After Migration
Developing a sense of national belonging is an important aspect of successful immigrant integration from both the minority and the majority perspective. Adopting the core norms and values of the new country as their own and passing them on to future generations is contingent on immigrants’ identification with the new country (Nesdale & Mak, 2000). Additionally, research has shown that a shared national identity can reduce negative feelings between minority and majority groups because the former out-group has become part of the national in-group (Gaertner & Dovidio, 2009). Social cohesion and unity in society are additional positive outcomes of a shared national identity (Verkuyten & Martinovic, 2012).
Previous cross-sectional work examined correlates of immigrants’ identification with their country of residence and consistently found length of stay to be important (e.g., Nesdale & Mak, 2000; de Vroome et al., 2014). These studies were not able to test the assumption that host-national identification increases within individuals with increasing length of stay. Our longitudinal design allows us to investigate within-individual change and test the hypothesis that the longer immigrants live in the Netherlands, the more they identify with the host country (H1).
In addition to host-national identity, we study immigrants’ origin-national and religious identity. Both ethnicity and religion are important markers of minority identity serving many functions: a sense of belonging, positive self-evaluation, and feelings of certainty (Verkuyten, 2018). Being raised and socialized in their origin country, these identities are usually already consolidated in first-generation immigrants. These identities become especially salient after migrating to a new country, and function as a fundamental element of self-understanding in the new context (Birman & Trickett, 2001).
While there is a large body of research concerned with origin-national identity, religion, which has proven to be equally important for immigrants’ identity (Deaux & Verkuyten, 2014), is less well-studied. For many, religion is important because beliefs, values, and practices give meaning in life, and help individuals cope with stressful circumstances, such as migration (Ysseldyk et al., 2011). Especially in Europe, debates about immigration and integration of immigrants have increasingly become linked to questions about religion in general, and about Muslims’ religious identity in particular (Deaux & Verkuyten, 2014). Islam is often presented as threatening and incompatible with national values such as freedom of speech and gender equality, fueling the perception that there is a schism between natives and Muslim immigrants (Vasta, 2007).
In dealing with their multiple identities, immigrants develop different strategies leading to changes in their host-national, origin-national and religious identification. The acculturation literature assumes these identities to be independent (Hutnik, 1991). However, previous empirical research has documented the full range of associations from positive, via unrelated, to negative correlations (e.g., Fleischmann & Phalet, 2016).
Negative associations between immigrants’ (ethnic or religious) minority and host-national identification are often interpreted as manifestations of identity incompatibility, implying a forced choice between both identities. In intergroup contexts where immigrants find themselves confronted with this forced choice, two distinct developments can be theorized. First, an emphasis on assimilation and national loyalty in public debates may create uncertainty about immigrants’ position in the new society and make them feel rejected and devalued. According to the uncertainty theory of social identity (Hogg, 2000) and the rejection-identification model (Branscombe et al., 1999), this may lead to a stronger emphasis on immigrants’ minority identity or, from the lens of acculturation research, a separation strategy. These theories suggest that in a more hostile and unaccepting environment, immigrants will turn towards their (religious or ethnic) in-group and identify stronger with it over time.
Second, SIT also proposes that individuals respond to the devaluation of their social identity by distancing themselves from the devalued group and becoming a member of a higher-status group (Tajfel & Turner, 1979). In the context of immigration, the host-national group occupies the higher-status position, and immigrants might therefore follow an assimilation strategy and increase their host-national identification while de-emphasizing their minority identification. In a context of pressure to adjust to the norms and values of the new country, this seems a favorable strategy to keep a positive and distinct social identity. However, this individual mobility pathway requires permeable group boundaries (Tajfel & Turner, 1979).
Incompatibility as Predictor of Changing Identification
Perceiving one’s origin-national and/or religious identity to be incompatible with the host-national identity contributes to a sense of impermeability and is therefore likely to trigger diverging developments over time, where one identity is strengthened at the expense of the other, such that they become increasingly negatively correlated. By contrast, immigrants who perceive their identities to be compatible, should be more likely to develop integrated dual identities by adding a sense of national identification to their persisting ethnic and religious identity, thus displaying increasingly positive associations between these identities. One previous cross-sectional study tested this claim and found the correlation between Turkish-Dutch Muslims’ religious identification and their Dutch national identification was more negative among participants who perceived greater identity incompatibility (Martinovic & Verkuyten, 2012). Similarly, permeability was positively correlated with perceived compatibility of origin and host-national identities and related to greater cultural adoption among first-generation Hungarian and Palestinian immigrants in Germany (Sixtus et al., 2019). However, this research was cross-sectional, leaving the question of whether perceptions of identity compatibility result from failed attempts to integrate multiple identities and cultures, or whether these perceptions prevent the emergence of integrated identities over time, unanswered.
Applying a longitudinal perspective, we expect that immigrants who perceive their specific ethnic and religious identities to be incompatible with host-national identification will either increase their host-national identification at the expense of their origin-national and religious identification over time (individual mobility/assimilation), or turn away from the host-national group and reactively increase their group-specific identities (Rejection-Identification/separation). In the context of adult first-generation immigrants, the second option seems more plausible because the minority (i.e., origin and religious) identities are already consolidated and an important part of migrants’ self. At the same time, since we are investigating the first years after arrival, a sense of host-national belonging most likely is still developing. Thus, we expect that immigrants who perceive their minority identity and their identification with the host country as incompatible will identify less with the Netherlands over time (H2) and will increase their origin-national and religious identification (H3).
Incompatibility as Predictor of Associations Between Identities
Next to these main effects of perceived incompatibility on the development of host-national, origin-national and religious identification, we expect incompatibility to interact with these identities in the prediction of other identities over time. In other words, to what extent host-national, origin-national and religious identification are positively or negatively associated over time, should depend on perceived compatibility. This expectation is based on previous literature showing a negative relationship between host-national and minority identification is partly explained by individuals’ beliefs about incompatibility (e.g., Sabatier, 2008). We refer to such beliefs as explicit measures of identity (in-)compatibility, as they measure individuals’ assessment that it is impossible to express multiple identities at the same time (Sindic & Reicher, 2009). This explicit approach takes into account that the relationship between host-national, origin-national and religious identification is complex, may differ within and between groups, and should be understood from the way individuals perceive their multiple identities (Chryssochoou & Lyons, 2011). Especially for newly arrived immigrants, perceived identity incompatibility might influence the development of host-national identification as perceptions of loyalty conflict; incompatible values and hostility have been found to lower national identification with the country of settlement (Verkuyten & Martinovic, 2012).
Another approach to identity incompatibility is to examine the associations between different identifications, which can range from positive, via zero, to negative. In this implicit approach, identification patterns are defined as incompatible when origin-national or religious identification is negatively correlated with host-national identification (Fleischmann & Phalet, 2016). In contrast to explicitly stated perceptions of identity incompatibility assessed at the individual level, this approach is located at the group level as patterns of associations cannot be observed for an individual (Chryssochoou & Lyons, 2011). Nevertheless, from a theoretical point of view the implicit approach aims to capture the same identity process as the explicit approach to identity incompatibility. Similar to Martinovic and Verkuyten (2012), we empirically examine the overlap between explicit and implicit measures of identity incompatibility by estimating interactions between explicit incompatibility and national identification in the prediction of origin and religious identification (i.e., implicit incompatibility), and vice versa. In our cross-lagged panel analysis, we expect that those immigrants who perceive more incompatibility will also have more negative associations between origin country (H4a) and religious (H4b) identification on the one hand and host-national identification on the other.
The Current Study
The Netherlands have historically been a welcoming society encouraging immigrants to maintain their distinct cultures. Since the beginning of the 21st century, however, issues of immigration and diversity have become focal points in public debates, with an extreme turn towards demanding conformity and assimilation from newcomers (Vasta, 2007) and electoral success of explicitly anti-immigrant/Islamophobic parties (Akkerman et al., 2016).
Our panel data allow us to compare identification patterns across four immigrant groups coming from Bulgaria, Poland, Spain, and Turkey. Since the enlargement of the European Union in 2004 and 2007, many Poles and Bulgarians have immigrated to the Netherlands. High unemployment following the financial crisis of 2007 led to an increase in immigration from Spain (CBS, 2018). While those three groups migrated relatively recently, we will also study recent arrivals from the more established but culturally/religiously more distant Turkish immigrant community.
Data and Method
Participants
The current study uses three waves of data from the New Immigrant Survey Netherlands (NIS2NL; Lubbers et al., 2018). The survey targeted immigrants from Bulgaria, Poland, Spain, and Turkey within the first five to ten months after their registration in the Netherlands.1
Of the full sample (N = 4,808), we excluded those who did not complete all three waves (N = 3,474) and who did not consistently state the same gender and birth year (N = 124). Additionally, not all immigrants arriving in the Netherlands registered immediately. Therefore, participants who stated that they were living in the Netherlands for longer than five years at Wave 1 (N = 182) were excluded from the sample.2 The final analytical sample consisted of 1,028 immigrants (151 Bulgarians, 358 Poles, 298 Spaniards, and 221 Turks, MAge = 31 (SD = 7.95) at Wave 1, 61% female).
Measures
We treated the three identification measures as latent constructs to effectively account for measurement error:
Host-national identification and origin-national identification were each measured with two items: “How important is the following to your sense of who you are: [the Netherlands]/[country of origin]?”; and “I have a strong sense of belonging to [the Netherlands]/[country of origin]”. After reverse coding and harmonizing, the answer scales ranged from 0 (not important at all) to 6 (very important) for the first question and from 0 (totally disagree) to 6 (totally agree) for the second.
Religious identification was measured with two items. Participants rated the statements “My religion is an important part of myself” and “I have a strong sense of belonging to my religion” on a 5-point scale ranging (after recoding) from 0 (totally disagree) to 4 (totally agree). Non-religious participants who did not answer the statements (NWave1 = 188; NWave2 = 199; NWave3 = 197) received the value 0 (totally disagree) to avoid missing values.
Perceived incompatibility was measured with a single item. Respondents rated the statement “The values of Dutch people and [country of origin people] are irreconcilable” on a 5-point scale ranging (after recoding) from 0 (strongly disagree) to 4 (strongly agree).
As control variables, participants’ gender (male = 0, female = 1), age at migration (in years) and level of education (dummies for Bachelor’s and Master’s degree or higher, lower educated migrants are the reference category) were included. To replicate our theoretical expectations across groups, we also controlled for country of origin and religious affiliation, distinguishing between Christians, Muslims, non-affiliates and other religions. Finally, we included months since migration at Wave 1 as control. All control variables were treated as time-invariant.
Analytical Approach
We estimated a cross-lagged panel model3 using Mplus version 7.3 (Muthén & Muthén, 1998-2017). To deal with skewed variables, maximum likelihood estimation with robust standard errors (MLR) was applied. By endogenizing the independent variables, missing values were dealt with by Full Information Maximum Likelihood assuming that these were missing at random.
Although we initially considered a multigroup model with origin countries as grouping variable, the small sample sizes of the four groups in combination with the complex three-wave latent-variable model made the estimation of this model impossible. Therefore, we analyzed all immigrant groups jointly. However, we replicate the analyses for each group separately and report these results – which do not include significance tests of parameter estimates across groups, for the reason just described – in the Supplementary Materials.
Host-national, origin-national, and religious identification were regressed on their respective scores at the preceding wave. Autoregressive paths, i.e., the paths between one variable at one timepoint to the same variable at the next timepoint, were included to assess stability over time (Adachi & Willoughby, 2015). All three identifications were additionally regressed on perceived incompatibility and the controls (see Figure 1). The assumption of stationarity was tested by constraining all path coefficients and residual (co)variances to be equal between waves.
Figure 1
Conceptual Model Displaying the Hypothesized Relations
Note. For readability, correlations (between the residuals) and the moderator are only shown for the transition between Wave 2 and 3 but are also included for previous waves.
Measurement Model
A Confirmatory Factor Analysis (CFA) and a test of longitudinal measurement invariance was performed to determine whether the latent constructs host-national, origin-national and religious identification were distinct and provide adequate measures that are comparable over time.
First, a CFA was conducted. Only using Wave 1 data resulted in a not optimal model fit, χ2(6) = 74.747, p < .001, RMSEA = .106, CFI = .965, SRMR = .024. Based on modification indices, the covariance of the residuals of two items was freed (Kline, 2010) resulting in a satisfactory model fit, χ2(5) = 1.551, p = .907, RMSEA = .000, CFI = 1.000, SRMR = .004. Subsequently, the model was extended by including repeated measures from all three waves. Moreover, the Multi-Trait Multi-Method structure regarding longitudinal analysis was applied by using Correlated Uniqueness (Kline, 2010). The final model fit was good, χ2(78) = 148.814, p < .001, RMSEA = .030, CFI = .994, SRMR = .023.
Second, we tested for measurement invariance comparing an unconstrained model (i.e., loadings, intercepts and residual variances of the items were allowed to vary across waves) with a constrained (fully constrained or scalar invariant model). The Satorra-Bentler scaled Chi-square difference test (Satorra & Bentler, 2010) showed that the constrained model was an equally good fit to the data, Δχ2(12) = 17.762, p = .123. More details can be found in the Supplementary Materials.
Results
Descriptive Findings
The first aim of this study was to examine the degree to which recently arrived immigrants in the Netherlands identify with their host country, and how this changes within the first years after arrival. The top section of Table 1 shows no significant change in host-national identification was observed, either for the sample as a whole or for any of the immigrant groups. On average, respondents identified with the Netherlands above the midpoint already in Wave 1.
Table 1
Mean Differences and Correlations of the Main Variables Across Waves
| Means and Correlations | Bulgarians (n = 151) |
Poles (n = 358) |
Spaniards (n = 298) |
Turks (n = 221) |
All (N = 1028) |
|---|---|---|---|---|---|
| Means (SD) | |||||
| HI | |||||
| Wave 1 | 3.78 (.12)a, x | 3.91 (.06)a, x | 3.17 (.08)b, x | 3.68 (.10)c, x | 3.63 (.04)x |
| Wave 2 | 3.93 (.11)a, x | 3.88 (.07)a, x | 3.27 (.07)b, x | 3.70 (.10)c, x | 3.67 (.04)x |
| Wave 3 | 4.05 (.11)a, x | 3.82 (.07)b, x | 3.28 (.08)c, x | 3.80 (.10)b, x | 3.69 (.04)x |
| OI | |||||
| Wave 1 | 4.07 (.14)a, x | 4.12 (.09)a, x | 3.85 (.09)b, x | 4.28 (.12)c, x | 4.07 (.05)x |
| Wave 2 | 4.13 (.14)a, x | 4.18 (.08)a, x | 4.20 (.09)a, y | 4.31 (.11)a, x | 4.20 (.05)y |
| Wave 3 | 4.18 (.13)a, x | 4.29 (.08)a, x | 4.16 (.09)a, y | 4.08 (.12)a, y | 4.19 (.05)y |
| RI | |||||
| Wave 1 | 2.09 (.11)a, x | 2.18 (.06)a, x | .68 (.06)b, x | 2.70 (.10)c, x | 1.84 (.05)x |
| Wave 2 | 1.93 (.11)a, x | 2.05 (.06)a, y | .67 (.06)b, x | 2.58 (.10)c, y | 1.75 (.05)y |
| Wave 3 | 1.83 (.11)a, x | 2.14 (.06)b, z | .64 (.06)c, x | 2.46 (.11)d, z | 1.73 (.05)y |
| Incomp | |||||
| Wave 1 | 1.82 (1.14)a, x | 1.71 (.98)a, x | 1.26 (.95)b, x | 2.28 (1.07)c, x | 1.70 (1.07)x |
| Wave 2 | 1.70 (1.15)a, x | 1.80 (.95)a, x | 1.35 (1.03)b, x | 2.08 (1.25)c, y | 1.71 (1.10)x |
| Wave 3 | 1.71 (1.10)a, x | 1.72 (.99)a, x | 1.32 (.97)b, x | 2.10 (1.22)c, y | 1.68 (1.09)x |
| Correlations | |||||
| HI & OI | |||||
| Wave 1 | .03a | -.03a | .27**b | -.17c | .03 |
| Wave 2 | .09a | .05a | .22*a | -.35**b | .01 |
| Wave 3 | .24a | .08a | .33***b | -.36**c | .06 |
| HI & RI | |||||
| Wave 1 | .21*a | .17*a | .18**a | -.03a | .16 |
| Wave 2 | .12a | .12a | .09a | -.05a | .13 |
| Wave 3 | .11a | .20**a | .07a | -.09a | .12 |
| OI & RI | |||||
| Wave 1 | .46***a | .38***a | .35***a | .57***b | .42*** |
| Wave 2 | .32**a | .36***a | .35***a | .68***b | .38*** |
| Wave 3 | .47***a | .33***a | .36***a | .61***b | .38*** |
Note. All means (Wave 1) across all groups are significantly (p < .05) different from the midpoint of the scale except for religious identification and identity incompatibility among Bulgarians. All correlations do not significantly (p < .05) differ from each other within the groups across waves. HI = host-national identification; OI = Origin-national identification; RI = religious identification; Incomp = Perceived identity incompatibility.
a, b, c, drefer to horizontal comparisons across groups within waves: means and correlations that do not significantly differ across groups (based on the Wald Test with p < .05) are labeled with the same letter. x, y, zrefer to vertical comparisons across waves within groups: means and correlations that do not significantly differ over time (based on the Wald Test with p < .05) are labeled with the same letter.
*p < .05. **p < .01. ***p < .001.
The second aim of this study was to examine how the identification with one’s origin country as well as with one’s religion changed over time. Table 1 shows that, on average, participants identified with their origin country above the midpoint and increased their identification significantly between the first and second wave. Regarding religious identification, participants identified on average below the midpoint and decreased their identification significantly between the first and second wave. Both origin-national and religious identification stayed stable between the second and third wave.
The most noticeable group differences were that Spanish participants identified on average more with their country of origin shortly after arrival but stayed stable over time, while the longer Turkish participants lived in the Netherlands, the weaker they identified with Turkey. Respondents from Poland and Turkey had a rather strong identification with their religion already shortly after arrival. While Polish respondents became less attached to their religion at first (between Wave 1 and 2) but more over time (between Wave 2 and 3), Turkish respondents’ religious identification became less important over time.
Looking at perceived incompatibility, participants stayed on average slightly below the midpoint. Incompatibility was stable across all waves. Bulgarians perceived their identities to be neither compatible nor incompatible, Polish and Spanish participants considered their identities as rather compatible, while Turks perceived their identities as rather incompatible, although this attenuated somewhat from Wave 1 to Wave 2. These group differences were significant except for the contrast between Turks and Bulgarians (see Table 1).
Regarding the relation of origin-national and religious identification with the development of immigrants’ host-national identification, Table 1 indicates that the higher participants identified with their origin country, the stronger they also identified with their religion. This positive association was particularly strong among Turks. Host-national and origin-national identification was not significantly associated. However, for Spaniards the two identifications were strongly and positively related, but negatively for Turkish participants. There were no significant associations between host-national and religious identification.4
Structural Model
Results of the cross-lagged panel model assuming stationarity are presented in Table 2.5 The autoregressive paths were significant for all three identifications and had high coefficient estimates. This indicates that the level of the variable in the previous wave was a good predictor of the same variable in the next wave, meaning that between-person differences in identification were rather stable. When estimating cross-lagged models, such results are quite common and lead to relatively low coefficient estimates for the cross-lagged relations. Nevertheless, low coefficients in cross-lagged paths are still meaningful (Adachi & Willoughby, 2015).
Table 2
Cross-Lagged Panel Model Predicting Host-National, Origin-National and Religious Identification
| Variables | DV: Host-national identification (t)
|
DV: Origin-national identification (t)
|
DV: Religious identification (t)
|
|||
|---|---|---|---|---|---|---|
| b | SE | b | SE | b | SE | |
| Host-national identification (t-1) | .63*** | .06 | -.11 | .08 | .03 | .05 |
| Origin-national identification (t-1) | .02 | .05 | .78*** | .05 | .01 | .04 |
| Religious identification (t-1) | .08 | .04 | .03 | .05 | .82*** | .03 |
| Perceived identity incompatibility (t-1) | -.03 | .02 | -.02 | .03 | .03 | .02 |
| Host-national identification × perceived identity incompatibility | .09** | .03 | -.03 | .04 | -.03 | .02 |
| Origin-national identification × perceived identity incompatibility | -.01 | .02 | -.01 | .03 | .00 | .02 |
| Religious identification × perceived identity incompatibility | -.04* | .02 | .02 | .03 | .01 | .01 |
| Controls | ||||||
| Female | .05 | .04 | .00 | .05 | .05 | .03 |
| Age | .01* | .00 | .00 | .00 | .00* | .00 |
| Low education (ref.) | ||||||
| Middle education | .02 | .05 | .00 | .06 | -.06 | .04 |
| High education | -.03 | .05 | -.09 | .06 | -.09* | .04 |
| Country of origin (ref. Spain) | ||||||
| Bulgaria | .14* | .07 | -.18* | .09 | .02 | .05 |
| Poland | .02 | .06 | -.15* | .07 | .16*** | .04 |
| Turkey | .04 | .12 | -.35* | .17 | .13* | .06 |
| Religion (ref. No religion) | ||||||
| Christian | .00 | .06 | .14 | .08 | .18*** | .04 |
| Muslim | .09 | .13 | .16 | .17 | .21** | .08 |
| Other | .13 | .21 | -.64* | .29 | -.35** | .11 |
| Months migration | .00 | .00 | .00 | .00 | .00 | .00 |
| Fit statistics | ||||||
| Loglikelihood (df) | -34212.911 (259) | |||||
| AIC | 68943.822 | |||||
| Explained Variancea | 39.9% | 40.1% | 26.9% | |||
Note. Unstandardized results. Coefficients and residual (co)variances were constrained to be equal over time.
abased on the proportional reduction in error.
*p < .05. **p < .01. ***p < .001.
Contradicting Hypothesis 1, and consistent with the descriptive comparison, time since migration was not a significant predictor of host-national identification. This implies that there were no substantial within-person or between-person changes in host-national identification; participants who lived in the Netherlands for a longer period do not identify more with it.
Similarly, Hypotheses 2 and 3 found no support, as there were no main effects of perceived incompatibility on any of the identifications. Thus, immigrants did not lower their Dutch identification, nor did they increase their origin-national or religious identification in response to perceived incompatibility.
However, in line with Hypothesis 4, perceived compatibility moderated the cross-lagged relations between identifications. Specifically, we confirmed H4b that predicted a more negative association between religious and host-national identification at higher levels of perceived incompatibility. A similar moderation was not found for the relationship between origin-national and host-national identification, refuting H4a and providing only partial support for H4 overall. Substantively, this finding implies that perceived incompatibility mainly relates to the incompatibility of host-national with religious, but not with origin-national, identification.
Finally, we also found an unpredicted moderation of perceived incompatibility with an auto-regressive path, such that immigrants who perceived high incompatibility were more stable in their host-national identification over time compared to those who perceived less incompatibility. The auto-regressive relations of origin-national and religious identification were not significantly related to perceived incompatibility. Overall, the findings imply that perceived incompatibility is most influential for the development of host-national identification but has little bearing on the absolute levels of, and changes in, identification with migrants’ origin country and their religion.
We tested the robustness of our results by re-estimating our model with three different specifications to address the coding of religious identification for non-religious participants, the issue of (selective) panel attrition and differences across immigrant groups. Details on these analyses can be found in the Supplementary Materials. These robustness checks supported the findings reported here, including the four proposed hypotheses.
Discussion
Previous research emphasizes the importance of dual identities for immigrant integration strategies (e.g., Nguyen & Benet-Martínez, 2013). Our study represents the first to longitudinally track changes in identities (including religious identity) relevant to the formation of dual identities in newly arrived immigrants. Moreover, by considering both implicit and explicit measures of identity (in-)compatibility, we are not only able to identify different identification patterns over time, but can also investigate how these two measures are related. We found immigrants’ host-national identification to be relatively stable over time, whereas origin-national and religious identification underwent more changes, in group-specific ways. This suggests immigrants’ strategies to (re-)define their origin and religious identification may differ from strategies driving identification with their host country.
The first aim of this study was to investigate the degree to which recently arrived immigrants in the Netherlands identify with their host country and how this changes within the first years after arrival. The four immigrant groups we studied (Bulgarians, Poles, Spaniards, and Turks) identified already quite highly with the Netherlands shortly after arrival and their identification was stable over time. There was no significant relationship between length of stay and host-national identification. We propose two possible explanations for this unexpected finding. First, we were only looking at quantitative changes over time. Previous research has shown that the content of identity is crucial as it gives meaning to one’s self-definition (Livingstone & Haslam, 2008). It is possible that there were qualitative differences in the understanding of identification levels. It could be that the content of what participants had in mind when thinking of the importance of the Netherlands for their self, differed with increasing length of stay. Second, a time span of four years might be too short to capture within- as well as between-person changes. Host-national identification is often considered the final step of integration (Gordon, 1964). Thus, even though immigrants already identified quite highly with the Netherlands upon arrival, more time might be needed for major changes to take place.
The second aim was to investigate how the degree of identification with one’s religion and the origin country of recently arrived immigrants changes within the first years after arrival, and how these identifications relate to the development of host-national identification. Regarding origin country identification, all groups identified rather strongly with their origin country to begin with. However, the development over time differed across groups. For example, immigrants’ identification with Spain first became stronger and stayed stable after some time, while for Turkish migrants the development was the opposite – first staying stable and then decreasing.
There were also different developments with respect to immigrants’ religious identification. Bulgarians and Spaniards had low levels of identification with their religion upon arrival and showed no change over time. Poles’ level of religious identification, however, was high upon arrival, initially decreased but became stronger over the total time span. Even though Turkish immigrants identified rather highly with their religion to begin with, they became less religious over time.
When looking at the relationship between immigrants’ different minority identities and the development of host-national identification, there was a clear pattern across all four groups: origin-national and religious identification were strongly and positively related, while host-national and religious identification were not. With respect to the relationship of host-national and origin-national identification, Spaniards who identified strongly with the Netherlands also identified more strongly with Spain, while Turks who identified strongly with Turkey identified less with the Netherlands.
These descriptive group differences confirm the assumption of acculturation research: that immigrants use different strategies to (re-)define their minority identities than they use to come to identify with their host country (e.g., Hutnik, 1991). Moreover, especially for the case of Turks, these findings also support previous literature showing negative relations between religious and host-national identification among Muslim minorities in European immigrant destinations (e.g., Fleischmann & Phalet, 2016; Fleischmann et al., 2019). Seemingly, rather than experiencing identities as independent, a forced choice between one’s religious identity and a strong sense of belonging to traditionally Christian and increasingly secular societies is more frequent in this intergroup context.
Our second hypothesis, that immigrants who perceive more incompatibility will identify less with the Netherlands over time, was not confirmed. However, we found that perceived incompatibility moderated the cross-lagged effect of religious on national identification, such that these identities became more negatively related over time at higher levels of perceived incompatibility. We also found that immigrants who already identified highly with the Netherlands, and perceived the values of their minority identity and host country to be incompatible, identified even stronger with the Netherlands over time. These findings suggest immigrants who perceive their identities to be incompatible do not necessarily reject the host-national identity but rather engage in an acculturative strategy of assimilation. In line with SIT, individuals might decide to turn towards the higher-status group instead of their low-status migrant group to sustain a positive and distinct social identity. This so-called individual mobility pathway is feasible when group boundaries seem permeable (Tajfel & Turner, 1979). Having established a shared sense of host-national belonging might be a sign that for some immigrants group boundaries can be overcome and seem rather permeable.
Regarding the relationship of explicit and implicit measures of identity incompatibility, we expected that those who perceive more incompatibility will also have more negative associations between origin country and religious identification on the one hand and host-national identification on the other. We did not find a relationship of explicit and implicit measures of identity incompatibility regarding origin-national and host-national identification. This might be due to the wording of the statement measuring perceived identity incompatibility which particularly referred to the incompatibility of values. The concept of norms and values is closely linked to religion but less closely to origin country. Religion often functions as a meaning-system constituted around beliefs, values and traditions that provide certainty and security (Verkuyten, 2007). Thus, it seems plausible that when asking about one’s minority values, individuals’ religious beliefs and norms are more salient than values referring to the origin country. In line with this argument and our expectation, we found that immigrants who identified strongly with their religion and perceived the values of their minority group and the Dutch to be incompatible identified less with the Netherlands over time. This suggests that it is particularly religious identification that is difficult to combine with host-national identification in a secular country (e.g., Fleischmann & Phalet, 2018). Public debates continuously problematize religious diversity and immigration of religious minorities in general and Muslims in particular. Islam is often associated with being threatening and considered to be incompatible with Western norms (Vasta, 2007). This societal context might heighten the pressure for Muslims to assimilate. Future studies might investigate whether our results apply to all religious groups equally due to the Netherlands being a highly secular country or whether this is especially true for Muslims, as Islam is often perceived as undermining Western/Dutch values and norms.
Regarding the robustness of our results, we were able to replicate them with non-religious people being recoded as missing and only using the first two waves. The latter finding is particularly important considering the substantial amount of panel attrition, which was not random but partly conditional on host-national identification (high identifiers were more likely to stay). We additionally found a positive association between religious and host-national identification. This means that only at high levels of perceived identity incompatibility, immigrants who identify highly with their religion identify less with the Netherlands over time. This further supports our reasoning that the process of negotiating multiple identities within one’s self can partly be explained by individuals’ perception of incompatibility (Sabatier, 2008). We were not able to fully replicate the findings with a simplified version of our main model investigating each immigrant group separately, although a lack of statistical power might account for this.
As the discussion of the results implies, this study reveals some limitations that might guide future research. First, the sample was rather young and highly educated, for which there are several potential explanations. To collect the data, immigrants that recently registered at the municipality were approached. While for Turkish migrants this is a necessity regarding their visa requirements, immigrants coming from the EU do not necessarily need to register in the Netherlands. This may bias the likelihood of registration toward more successful EU immigrants (e.g., in terms of having a job). Moreover, our panel attrition analysis (described in the Supplementary Materials) showed that the sample became slightly selective in terms of gender, level of education, country of origin, months since migration and host-national identification over time. Having such a young and highly educated sample means we need to be careful generalizing our results to the whole population of recently arrived immigrants. Nonetheless, the fact that the sample is mainly young and highly educated might genuinely reflect current intra-EU migration (Fries-Tersch et al., 2017). Thus, future research is needed to further investigate how these findings apply to other immigrant groups, e.g., non-Western immigrants and refugees.
Second, we were not able to conduct a multi-group analysis to compare the four different immigrant groups more systematically. The findings of our robustness checks indicate that the results are more reflective of individual rather than group differences. However, the robustness check running the main analysis for each group separately and using mean scores instead of latent constructs was limited in the sense that it allowed for more measurement error. Moreover, the smaller sample sizes additionally lowered statistical power. Future large sample research is needed to more systematically examine whether identification patterns generalize across immigrant groups with varying characteristics (e.g., religious affiliation). Except for the Polish participants, the sample was not drawn at random, leaving opportunity only for improvement in terms of the data gathering technique, e.g., to guarantee a higher participation rate.
Third, although the longitudinal design of this study allowed us to test for causality, many of our key findings took place simultaneously. Thus, this raises the question whether there is a causality in the first place and if yes what comes first – level of identification or perceived incompatibility. Future research is needed to further investigate the remaining issue of causality.
Notwithstanding these limitations, we can conclude that, especially for highly religious immigrants, it seems more difficult to combine several identifications into one social identity which hindered immigrants to develop a shared host-national identity with the Netherlands. At the same time, immigrants who already identify highly with the Netherlands are more likely to “choose” the Netherlands over their minority identity. These findings support the notion that religion is often considered a strong pillar in individuals’ overall identity being of profound importance to individuals’ lives. In addition, this research emphasizes that the negative relationship between host-national identification and minority identification can partly be explained by individuals’ beliefs about incompatibility. Thus, to develop social cohesion and reduce antagonism between minority and majority groups, it is important to address the importance of religion in immigrants’ lives. Moreover, more awareness of minorities’ perception of incompatibility is needed in public debates that portray immigrants as undermining social cohesion by lacking a common sense of belonging to the host society. As Phinney (1999) summarizes: “Increasing numbers of people find that the conflicts are not between distinct groups but between different cultural values, attitudes, and expectations within themselves” (p. 26).