Introduction

Hotly debated issues about immigration in many countries focus on the policies of recognition and integration, and on the extent to which newcomers deserve and are entitled to equal rights. Immigrants and minorities may challenge the existing social and cultural order of the nation and thereby raise questions of national unity and cohesion. Threats to society and national identity are the key issues of debate. Hence, the aim of this research was to investigate how national identification is related to out-group threat and the support for recognition and rights of minority groups among ethnic Dutch adolescents and adults. In three studies, support for immigrant rights was examined as a consequence of perceived out-group threat and national identification. Three different models of this process are tested: the ‘group identity lens model,’ the ‘group identity reaction model,’ and the ‘group identity moderator model’ (Eccleston & Major, 2006). In addition, Study 3 examined whether authoritarianism predicts support for multicultural recognition and minority rights independently or rather indirectly via its effect on national identification.

Out-group Threat

In the 1970s, Barker (1981) showed that the fear that foreign cultures will swarm over England and override the British way of life was a main argument to oppose immigration and immigrants. Recent studies in European countries have found a similar public discourse of threat and fear to oppose immigration, multiculturalism, and minority rights (e.g., Triandafyllidou, 2000). In Britain, Lynn and Lea (2003), for example, showed that current ideas of citizenship and national identity are part of a public discourse that contributes to a ‘new Apartheid.’

A variety of theories suggest that fear and perceptions of threat play an important role in generating prejudice toward out-groups in general, and toward immigrant groups in particular (e.g., Esses, Dovidio, Jackson, & Armstrong, 2001; Jackson, Brown, Brown, & Marks, 2001). Different types of threat have been identified, such as realistic and symbolic threat (Stephan & Stephan, 1996) and the perception of these threats can lead to negative reactions toward immigrants and minorities.

Realistic threats can be conceptualized in physical, economic, and political terms. Ethnic minority groups can be perceived as a safety threat and conflicts between groups and negative group reactions are often rooted in a clash of political and material interests. The desire to protect in-group interests is often found to be responsible for negative attitudes and discriminatory behavior toward immigrants and minorities (e.g., Bizman & Yinon, 2001; Esses et al., 2001).

Symbolic threats are based on perceived group differences in values, norms, and beliefs. Out-groups that have different worldviews may be seen as threatening the cultural identity of the in-group and as overriding one’s way of life. Many studies have shown that perceived threats to in-group values by immigrants and minorities predict more negative attitudes toward these groups (e.g., Esses, Hodson, & Dovidio, 2003; Ward & Masgoret, 2006). Studying 17 European countries, McLaren (2003), for example, found that perceptions of realistic threat as well as beliefs that immigrants challenge or undermine national values were the predictors of negative attitudes toward immigrants. Thus, both realistic and symbolic out-group threats have been associated with less support for multicultural recognition and rights of immigrants and ethnic minorities.

National Identification

National identity and national identification are central in debates on immigration and the integration of ethnic minority groups. This is the case in settler countries such as Australia and the United States (e.g., Huntington, 2004) and also in nonsettler European societies that have a historically established majority group (Joppke, 2004).

National identification is important for understanding how a native population responds to and reacts toward newcomers. Theoretically, the question is how exactly national identification is involved in these reactions. National identification can be an antecedent of perceived out-group threat, a consequence of perceived threat, and can also moderate the relationship between the out-group threat and the support for multicultural recognition and minority rights. In addition, an independent negative relationship between the national identification and the support for minorities might exist. These different possibilities have not been tested in previous research on minority rights. It has also been suggested that group identification is the result of personality variables such as authoritarianism (Perrault & Bourhis, 1999), but there is quite a debate about this suggestion (Reynolds & Turner, 2006). Given the importance of national identification in debates about immigrants and the different ways in which identification may affect the support for multicultural recognition and equal rights, I tested three different models of the relationships among national identification, perceived out-group threat, and support for minority rights.

The first ‘group identity lens’ model hypothesizes that national identification is an antecedent of perceived out-group threat and will affect the support for minority rights indirectly, via its association with threat. This model is in line with self-categorization theory (Turner & Reynolds, 2001) that posits that when a particular social identity is salient it provides a ‘lens’ through which the perceiver sees the world and makes sense of it. Group identity functions as a group lens that make people sensitive anything that concerns or could harm their groups. Thus, higher group identification will lead to greater threat perceptions and these perceptions result in a particular response.

Different studies on attitudes toward immigrants and ethnic minorities have found supporting evidence for this model. In their meta-review, Riek, Mania, and Gaertner (2006) found that in-group identification had a significant impact on realistic and symbolic threat. In the Netherlands, Van Oudenhoven, Prins, and Buunk (1998) showed that individuals who identify strongly with the Dutch in-group were more likely to perceive the presence of ethnic minorities as a threat to Dutch culture and society. In agreement with the group-identity-lens model and these empirical results, it can be expected that perceived out-group threat mediates the relationship between the national identification and the support for multiculturalism and minority rights. A Dutch person may be unsupportive of minority rights because she perceives relatively high levels of out-group threat, and she may perceive more threat because of the importance or centrality of the Dutch national identity to her sense of self.

The second ‘group identity reaction’ model is based on the idea that perceiving group threat leads individuals to identify more strongly with their in-group and that stronger identification leads to more negative out-group attitudes. People can cope with out-group threats by adopting group-based strategies that increase in-group identification. Some experimental evidences have shown that threats can indeed increase group identification (Jetten, Branscombe, Schmitt, & Spears, 2001). Further, survey research among racial and ethnic minority groups shows that increased perceptions of discrimination predicts increased in-group identification (Schmitt & Branscombe, 2002; Verkuyten & Yildiz, 2007). Discrimination against minorities presents a threat to minority group identity, making minorities increasingly turn toward the minority in-group. Ethnic and racial identities differ from national identity, however, and the position of majority groups is also different. To the best of my knowledge, there is no systematic evidence that threats posed by immigrants and minorities leads to stronger national identification. In most western European countries, indigenous majority groups have a relatively secure position and ethnic group boundaries are rather impermeable. These conditions imply that threats posed by immigrants and ethnic minorities are unlikely to lead to stronger national identification (Verkuyten & Rijerese, 2008).

The third ‘group identity moderator’ model predicts that national identification interacts with out-group threat to predict the support for multiculturalism and minority rights. This model is in line with social identity theory (Tajfel & Turner, 1986). The idea is that perceived threat has different effects depending on national identification because the motivational meanings of perceived threat are different. Compared with low identifiers those with high in-group identification are more likely to be concerned about their group, especially when the position and value of the group identity is at stake. For example, high identifiers have been found to react more negatively toward the out-group under threat than do low identifiers (e.g., Branscombe & Wann, 1994). Among Israeli citizens, Bizman and Yinon (2001) found that realistic threat to the in-group, but not symbolic threat, was a more important predictor of attitudes toward immigrants for high than for low identifiers. In two studies in the context of Northern Ireland, Tausch, Tam, Hewstone, Kenworthy, and Cairns (2007) found that symbolic threat, but not realistic threat, predicts out-group attitudes and trust for high identifiers only. They argue that the divergent findings regarding the two types of threat are due to the relative importance of realistic and symbolic threats in the context of Israel and Northern Ireland, respectively. Identification should moderate only effects of threats that are actually relevant in a given context. It is also possible, however, that the group moderation model only finds support in divided contexts of sectarian and violent conflicts in Israel and Northern Ireland. In three studies in the context of the Netherlands, Verkuyten and Yildiz (2007), for example, found no evidence for this model among Turkish–Dutch participants.

Authoritarianism

In social psychology there is a long tradition of theories and research that explains negative out-group attitudes and behaviors in terms of personality variables. The study on authoritarianism and social dominance orientation is extensive and recently these constructs have been conceptualized as the sets of ideological beliefs (Duckitt, 2001). Authoritarianism and social dominance have been used to explain reactions toward immigrants and the support for multiculturalism and minority rights (e.g., Danso, Sedlovskaya, & Suanda, 2007; Jackson & Esses, 2000). The emphasis in this line of study differs from the social identity perspective which focuses on group memberships and intergroup relations (Reynolds & Turner, 2006). The social identity perspectives suggest that in an intergroup context social identity produces intergroup behavior independent of personality or individual differences in ideological beliefs. There is experimental evidence that supporting this idea (e.g., Reynolds, Turner, Haslam, & Ryan, 2001; but see Heaven & St. Quintin, 2003). For example, Verkuyten and Hagendoorn (1988) showed that in an intergroup context, in-group stereotypes and not authoritarianism was related to out-group attitudes.

In trying to reconcile a social identity perspective with the role of these individual different factors it has been suggested that variables such as authoritarianism and social dominance orientation affect processes of group identification (Perrault & Bourhis, 1999; Sidanius, Pratto, & Mitchell, 1994). Authoritarian individuals, for example, may be ready to categorize themselves and others as in-group and out-group members leading them to identify more strongly with relevant in-groups. The implication is that personality variables do not only affect intergroup reactions independently, but also indirectly, via their effect on in-group identification. In Study 3, I examined this possibility by focusing on authoritarianism.

Overview

The three studies presented here examine the relationships among national Dutch identification, perceived out-group threat, and the support for multicultural recognition and rights for immigrants and ethnic minority groups. Three different models of these relationships were tested. In agreement with self-categorization theory and considering the existing research, the most support was expected for the group-identity-lens model in which national identification is an antecedent of perceived out-group threat and identification is related to the support for multiculturalism and minority rights via threat. In the first two studies these relationships and expectations were examined among relatively large samples of ethnic Dutch adolescents, and in Study 3 among a national sample of ethnic Dutch adults.

The three studies use similar measures for national identification and for out-group threat, but use somewhat different measures of multicultural support. There are an increasing number of studies that examine people’s stereotypes and prejudicial attitudes toward immigrants and ethnic minority groups. However, the hotly debated issues in society are usually about particular policies and specific group rights. Feelings and beliefs toward out-groups can differ from support for minority policies and rights. For example, tolerance of minority practices is not the absence of prejudice but rather a separate construct that emphasizes forbearance or ‘putting up with’ without interference (Vogt, 1997).

There is a widespread use of terms such as ‘multiculturalism’ and ‘minority rights’ and there is no single view or strategy implied. At the very least, multiculturalism offers a positive view of cultural and identity maintenance for ethnic minority groups and, as such, a concomitant need to accommodate diversity in an equitable way. European multiculturalism has always been targeted at immigrants and minorities rather than the majority group (Joppke, 2004). For minority groups, multiculturalism offers the possibility of maintaining their own culture and obtaining equal rights and status in society. In this research the focus is on multicultural recognition (Studies 1 and 2), on tolerance (Study 2), and on equal rights (Study 3).

In addition, the three studies differ in their target minority group. In public debates and in research many different labels are used to define categories of minorities. Minorities are discussed as abstract broad categories (e.g., immigrants, migrants, ethnic minorities) or as more narrow, ethnic or religion-specific groups (e.g., Blacks, Turks, Muslims). The level of abstraction might affect people’s attitude. For example, two studies in different countries, found that attitudes toward abstract categories (‘ethnic minorities’), serve more symbolic functions than attitudes toward more specific target groups (‘Indians’ or ‘Chinese’), which seem to result more from realistic threats (Watt, Maio, Rees, & Hewstone, 2007). In Study 1 the target groups were relatively specific (Turks and Moroccans). In Study 2 the target groups were abstract (‘immigrants) and specific (‘Muslims’). In Study 3 the target group was abstract (‘ethnic minorities’). The use of these different target groups was not meant to examine the effects of the level of abstraction systematically, but rather to assess whether the relationships among national identification, out-group threats and support for multiculturalism, and minority rights are similar across different categories of minorities.

Study 1

Study 1 examines the relationships among national identification, out-group threat, and support for multiculturalism among a sample of Dutch adolescents. Following self-categorization theory I expected most support for the group-identity-lens model.

Method

Sample

In 2005 a sample of 784 ethnic Dutch students were recruited from schools in the middle and east of the country to participate in research on ‘Current issues in Dutch society.’ It took about 20 min to complete the anonymous questionnaire. Of the participants 47% were women and 53% were men. The age of the participants ranged from 13 to 17 years and the mean age was 15.03 (SD = .95). Following the Dutch educational system, four levels of education were analyzed: preparatory vocational training and lower general secondary education (vmbo, 32.9%), middle general secondary education (havo, 31.9%), upper general secondary education (atheneum, 21.7%), and highest general secondary education (gymnasium, 13.6%).

Measures

Support for multiculturalism was measured with six items that have been used in previous research in the Netherlands (Verkuyten & Yildiz, 2007). The items focused on the Turkish and Moroccan Muslims living in the Netherlands. Three sample items are: ‘Turks and Moroccans may keep their own traditions and culture,’ ‘Turks and Moroccans have the right to have their own Muslim schools,’ ‘Turks and Moroccans should simply adapt to the Dutch norms and values’ (reverse scored). Answers were given on 5-point rating scales: strongly disagree (1) to strongly agree (5). Cronbach’s alpha was 0.79 with higher scores indicating stronger endorsement of multiculturalism.

National identification was assessed by asking the participants to respond to six items (same 5-point scales) that were taken from previous studies in the Netherlands (see Verkuyten, 2005). These items measure the importance attached to one’s national group membership and two sample items are ‘Being Dutch is an important part of how I see myself,’ and ‘My Dutch identity is an important part of my self.’ Cronbach’s alpha was 0.88 with higher scores indicating greater national identification.

Symbolic threat was measured using items adapted from scales used by Stephan and Stephan (1996) and Stephan, Diaz-Loving, and Duran (2000). The items focused on threats posed by Muslim immigrants, because this group has emerged as the focus of immigration and diversity debates in Europe and is at the heart of what is perceived as a ‘crisis of multiculturalism’ (Modood & Ahmad, 2007; Zolberg & Long, 1999). Participants were presented with the following three statements: ‘Dutch identity is being threatened because there are too many Muslims,’ ‘Dutch norms and values are being threatened because of the presence of Muslims,’ and ‘Muslims are a threat to the Dutch culture’ (similar 5-point scales).

Safety threat was also assessed using three items. The items were: ‘I am afraid of increasing violence and vandalism of Muslims in my city,’ ‘I am afraid of increasing violence and vandalism in the Netherlands of Muslims,’ and ‘I am afraid of terrorist attacks of Muslims in the Netherlands.’ The response scales were identical to those used for measuring symbolic threat. Maximum likelihood estimation with oblique rotation was used to determine whether the participants make a distinction between the two types of threat. Only one factor emerged that explained 56.5% of the variance. All six items loaded above 0.56 on the factor. Hence, the items were averaged to compute one scale for out-group threat. Cronbach’s alpha was 0.87 with higher scores indicating stronger feelings of threat.

Results

Mean Scores and Intercorrelations

On average, participants exhibited moderate levels of national identification (M = 3.44, SD = 0.83) and out-group threat (M = 3.18, SD = 0.92), both around the mid-point of the scale. Support for multiculturalism (M = 2.48, SD = 0.89) was below the mid-point of the scale indicating relatively little endorsement of multiculturalism.

Table 1 shows the intercorrelations among the three measures. Greater national identification related to more out-group threat and to lower endorsement of multiculturalism. Further, out-group threat and multiculturalism were negatively related.

Table 1 Intercorrelations for the different measures in Studies 1 and 2
Full size table

Multiculturalism

In order to examine the associations in more detail, a hierarchical regression analysis was conducted to predict the support for multiculturalism. The effects of gender, age, and educational level were entered on Step 1, and the main effects of national identification and perceived out-group threat (centered scores) were entered on Step 2. The model in the first step was significant, F change(3, 781) = 24.55, p < 0.001. All three predictors had significant independent effects. Females were more supportive of multiculturalism than males, beta = 0.10, t = 2.79, p = 0.005. Participants with higher levels of education endorsed multiculturalism more strongly, beta = 0.26, t = 7.55, p < 0.001. Further, greater age predicted reduced support for multiculturalism, beta = −0.10, t = 2.93, p = 0.005.

The addition of the two measures on Step 2 significantly increased the explained variance, R 2change  = 0.16, F change(2, 779) = 80.40, p < 0.001. Greater out-group threat predicted significantly less support for multiculturalism, beta = −0.40, t = 11.53, p < 0.001. National identification had no independent significant effect on multiculturalism, beta = −0.04, t = 0.99, p > 0.05. The full regression model explained 24% of the variance in the support for multiculturalism.

Mediation and Alternative Models

According to Baron and Kenny (1986), the critical test for mediation is that the relationship between the independent variable (national identification) and the dependent variable (support for multiculturalism) must be significantly reduced when the mediator variable (out-group threat) is controlled. Table 1 shows that national identification was significantly related to multiculturalism and to out-group threat, and that out-group threat was negatively related to the support for multiculturalism. In addition, when in the second step multiculturalism was regressed onto out-group threat and national identification, the effect of national identification was no longer significant (from beta = −0.22, p < 0.001, to beta = −0.05, p > 0.10). The Sobel test for mediation confirmed that the mediational path was reliably >0, z = 8.61, p < 0.001. The result of this mediational analysis is consistent with the group-identity-lens model.

Two alternative models were tested. First, the group-identity-reaction model argues for reversed mediation: national identification mediating the relationship between out-group threat and multiculturalism. However, the effect of out-group threat was not significantly reduced in an analysis in which in the second step multiculturalism was regressed onto national identification and out-group threat. The Sobel test for this reversed mediation was not significant, z = 0.54, p > 0.10.

Second, the group-identity-moderator model predicts that national identification moderates the relationship between out-group threat and the endorsement of multiculturalism. Moderation was tested by examining whether the interaction term between national identification and out-group threat significantly predicts the endorsement of multiculturalism (Aiken & West, 1991). In an additional regression analysis, this interaction was not significant, beta = 0.01, t = .21, p > 0.10.

Discussion

The results of Study 1 indicate that the ethnic Dutch participants had moderate levels of national identification and of perceived out-group threat, and little support for multiculturalism regarding Turkish and Moroccan minorities. In addition, stronger national identification predicted reduced support for multiculturalism mediated by increases in out-group threat. Thus, there was statistical evidence for the group-identity-lens model: national identification is associated with higher out-group threat and, via threat, to more strong rejection of multiculturalism. There was no evidence for reverse mediation in which national identification mediates the relationship between out-group threat and the endorsement of multiculturalism. There was also no evidence for the group identity moderator model that predicts that national identification moderates the relationship between threat and multiculturalism.

Study 2

A second study was conducted to examine whether these findings could be generalized to another sample. In so doing, I focused on the support for multiculturalism in relation to immigrants and ethnic minorities in general and also on tolerance toward Muslims in particular. Thus, the emphasis was on immigrants and ethnic minorities more generally and whether participants supported their cultural recognition and rights. Further, research has shown that the endorsement of abstract principles such as freedom of speech or the support for cultural diversity can differ from the evaluation of concrete practices and acts. Principle considerations tend to differ from (the lack of) support for practical implications and situations (see Vogt, 1997). Most debates on cultural diversity are not about principles per se, such as equality and freedom, but rather about whether specific acts and actors should be tolerated. In Study 2, I tried to maximize the relevance and validity of the research by using two concrete cases related to the participants’ school experiences.

Method

Sample

In 2006, a sample of 393 ethnic Dutch participants were recruited at schools, this time in the south of the country, to participate in a research on ‘Current issues in Dutch society.’ It took about 20 min to complete the anonymous questionnaire. Of the participants 52.5% were women and 47.5% were men. The age of the participants ranged from 13 to 17 years and the mean age was 14.81 (SD = 0.85). In the analyses and for this sample, three levels of education were used: preparatory vocational training and lower general secondary education (vmbo, 42.1%), middle general secondary education (havo, 21.8%), and upper general secondary education (vwo, 36%).

Measures

Support for multiculturalism was measured with six items that were taken from Berry and Kalin’s (1995) Multicultural Ideology Scale. These items focus on immigrants and minority groups in general and have been used in previous research in the Netherlands (Arends-Tóth & Van de Vijver, 2003; Verkuyten, 2005). Three sample items are: ‘The more cultures there are, the better it is for the Netherlands,’ ‘Allochthones (Dutch general terms for immigrants and ethnic minorities) should be supported in their attempts to preserve their own cultural heritage in the Netherlands,’ and ‘Allochthones should forget their cultural background as soon as possible’ (reverse scored). Answers were given on 5-point rating scales: strongly disagree (1) to strongly agree (5). Cronbach’s alpha was 0.76 with higher scores indicating stronger endorsement of multiculturalism.

Tolerance was examined with two items that involved Muslims in the school context. The first item was on a new Muslim teacher and the second on a political speech by a Muslim at one’s school. The items were, ‘Would you accept it when a Muslim is appointed as a new teacher at your school?’ and ‘Would you accept it when a Muslim makes a public speech at your school?’ Answers were given on 5-point rating scales: no, certainly not (1) to yes, certainly (5). The responses on both questions were highly correlated (r = 0.67, p < 0.001). Hence, the two items were averaged with higher scores indicating higher tolerance.

Symbolic threat and safety threat were measured with the same items and rating scales that were used in Study 1. Again, maximum likelihood estimation with oblique rotation was used to determine whether the participants make a distinction between the two types of threat. As in Study 1, only one factor emerged that explained 54.5% of the variance. All six items had a high factor loading (>0.57). Cronbach’s alpha for the six items was 0.88 with higher scores indicating stronger feelings of out-group threat.

National identification, or the importance attached to one’s national background, was assessed by the same six items used in Study 1. Cronbach’s alpha was 0.89.

Results

Mean Scores and Intercorrelations

Similar to Study 1, on average, participants exhibited moderate levels of national identification (M = 3.44, SD = 0.86) and out-group threat (M = 3.01, SD = 0.95), both around the mid-point of the scale. On average, support for multiculturalism (M = 2.71, SD = 0.68) was again below the mid-point of the scale indicating relatively little endorsement for multiculturalism. The mean score for tolerance was above the mid-point (M = 3.43, SD = 1.13) and indicates some acceptance.

As shown in Table 1, higher national identification was significantly related to greater out-group threat and reduced support for multiculturalism and tolerance. Greater out-group threat reduced support for multiculturalism and tolerance, and support for multiculturalism predicted greater tolerance.

Multiculturalism

A hierarchical regression analysis was conducted to predict the support for multiculturalism. The effects of gender, age, and educational level were entered on Step 1, and the measures for national identification and out-group threat were entered on Step 2. The model in Step 1 was significant, F change(3, 390) = 6.92, p < 0.001. Replicating Study 1, females were more supportive of multiculturalism than males, beta = 0.17, t = 3.38, p = 0.001, and increasing age again predicted lower support for multiculturalism, beta = −0.10, t = 2.12, p = 0.034.

The addition of the measures in Step 2 significantly increased the explained variance, R 2change  = .32, F change(2, 388) = 98.98, p < 0.001. Reduced support for multiculturalism was predicted by greater national identification (beta = −0.15, t = 3.38, p = 0.001) and greater out-group threat (beta = −0.49, t = 10.88, p < 0.001). The full regression model explained 37% of the variance in multiculturalism support.

Table 1 shows that national identification was significantly related to multiculturalism and to out-group threat. In addition, out-group threat was negatively related to the support for multiculturalism. In order to examine whether out-group threat mediates the relationship between national identification and multiculturalism, endorsement of multiculturalism was regressed onto out-group threat and national identification, and the effect of identification was significantly reduced (from beta = −0.37, p < 0.001, to beta = −0.11, p < 0.05). The Sobel test for mediation confirmed that the mediational path was reliably >0, z = 7.03, p < 0.001. The result of this mediational analysis is consistent with the group-identity-lens model.

I again examined the reversed mediation in which national identification mediates the relationship between out-group threat and multiculturalism. Here, the Sobel test was significant, z = 3.21, p < 0.001. This indicates that the group-identity-reaction model also fits the data, although less strongly.

Further, replicating results of Study 1, additional regression analysis indicated that national identification did not moderate the relationship between out-group threat and endorsement of multiculturalism. The interaction terms was not significant, beta = 0.02, t = 0.26, p > 0.10. Thus, there was, again, no support for the group identity moderator model.

Tolerance

In order to examine whether tolerance judgments operate similarly to support for multiculturalism, a hierarchical regression analysis was conducted in the effects of gender, age, and educational level which were entered on Step 1, and the measures for national identification and out-group threat were entered on Step 2. The model in Step 1 was not significant, F change(3, 390) = 1.01, p > 0.05, indicating no gender, age, or educational level differences. The addition of the two measures in Step 2 significantly increased the explained variance, R 2change  = 0.31, F change(2, 388) = 90.08, p < 0.001. Reduced tolerance was predicted by greater out-group threat, beta = −0.53, t = 11.44, p < 0.001, whereas the effect for national identification was also negative but not significant, beta = −0.07, t = 1.51, p > 0.05. This suggests that out-group threat fully mediates the relationship between national identification and tolerance. The Sobel test for mediation confirmed that the mediational path was reliably >0, z = 7.12, p < 0.001. The result of this mediational analysis supports the group-identity-lens model.

Again, I examined the reversed mediation in which national identification mediates the relationship between out-group threat and multiculturalism. The Sobel test was not significant, z = 1.72, p > 0.05. Thus, there was no evidence that the group-identity-reaction model fits the data for the tolerant judgments.

In an additional regression analysis there was also no support for the group-identity-moderator model. The interaction between national identification and out-group threat was not significant, beta = 0.01, t = 0.08, p > 0.10.

Discussion

The results of this second study are not only quite similar to those of the first, but also go beyond the first study by focusing on multiculturalism in relation to a more general target group as well as on tolerance toward Muslims in particular. The mean scores for national identification and out-group threat were similar to those in Study 1 and the association between these two measures was also similar. National identification and out-group threat were negatively associated with the support for multiculturalism. Furthermore, the negative effect of national identification on multiculturalism was partly mediated by out-group threat. However, the reverse model in which national identification mediates the relationship between out-group threat and the support for multiculturalism also fitted the data, although less strongly.

The findings for tolerance also strongly support the group-identity-lens model. Out-group threat fully mediated the relationship between national identification and tolerance toward Muslims in the school context, and there was no evidence for the reversed mediation.

Furthermore, in examining the endorsement of multiculturalism and also tolerance, no support for the group-identity-moderator model was found. National identification did not moderate the relationship between out-group threat and either support for multiculturalism or tolerance.

Study 3

In Study 3, I examined whether these findings generalize to an older sample by analyzing data collected among a national sample of Dutch adults. In this study the focus was on participants’ endorsement of equal rights of immigrants and ethnic minority groups. The topic of equality and equal rights is central in debates on managing cultural diversity (Vermeulen & Slijper, 2003). Multiculturalism is, typically, closely linked to the notion of equality and is seen as an important ideology and policy approach for addressing inequality and structural discrimination. In addition, the role of authoritarianism was also considered in Study 3. The idea that personality and individual differences in ideological beliefs explain reactions toward minority groups has been put forward by many social psychologists. Personality-based explanations have also been criticized. The social identity perspective (Reynolds & Turner, 2006), for example, argues that a personality account tends to overlook people’s group memberships and the intergroup relations that shape people’s attitudes and beliefs. Experimental research has shown that individual differences in authoritarianism tend to become irrelevant to prejudice where group membership is salient (e.g., Reynolds et al., 2001; Verkuyten & Hagendoorn, 1998, but see Heaven & St. Quintin, 2003). While trying to reconcile personality and social identity accounts some researchers have suggested that variables, such as authoritarianism, may not only affect out-group perceptions and evaluations independently, but also indirectly by their influence on processes of group identification (e.g., Perrault & Bourhis, 1999; Sidanius et al., 1994). High authoritarians tend to see the world in ‘black’ and ‘white’ terms which would make it likely that they define themselves and others as members of in- and out-groups. Thus, authoritarianism may affect in-group identification that, in turn, shapes the perception of the intergroup situation (i.e., threat) and the reaction toward immigrants and ethnic minority groups. To date, this possibility has been examined in a few experimental studies using the minimal group paradigm. Perrault and Bourhis (1999) found that ethnocentrism, but not authoritarianism, predicted intergroup discrimination through in-group identification. In another study, Sidanius et al. (1994) found that the effect of group identification on in-group favoritism was greater among participants high than low in social dominance orientation.

In contrast to these studies, Reynolds et al. (2007) examined the role of authoritarianism and social dominance orientation and they found no evidence of significant relationships between these measures and either in-group identification or discrimination. They also found no evidence for the role of group identification being moderated by participants’ endorsement of these individual difference measures. In addition, there was evidence that authoritarianism and social dominance orientation shape out-group behavior independently in some situations, namely where a comparison between egalitarian and hierarchical social structures is introduced (Reynolds et al., 2007) or when there is intergroup threat (Pratto & Shih, 2000).

Hence, Study 3 was designed to assess whether authoritarianism explains intergroup perceptions independently or rather through or in interaction with processes of in-group identification. Following self-categorization theory I expected authoritarianism to have an independent positive effect on the perception of out-group threat. Thus, national identification was not expected to mediate the relationship between authoritarianism and the perception of out-group threat and this perception was also not expected to be explained by the interaction between authoritarianism and national identification. The use of national survey data for examining these relations adds to the study using the minimal group paradigm. Further, a large-scale national survey provides a dataset in which the different relationships should be readily observable.

Method

Sample

In 1998, a national representative survey among the ethnic Dutch populations was conducted by randomly selecting telephone numbers and randomly selecting an adult participant from each household. The interview was approximately 45-min long and used computer-assisted interviewing for conducting randomized experiments in large-scale survey research (see Sniderman & Hagendoorn, 2007). In total 2007 people were interviewed and half of them were asked questions about threats posed by ethnic minorities. Using this sub-sample and after excluding participants with more than one missing value the working dataset for the current analysis consists of 887 participants.

Measures

Support for equal rights was measured with seven items using the 4-point scales. Four sample items are ‘Do you agree or disagree that ethnic minorities should have the right for equal treatment as the Dutch,’ ‘Considering the high unemployment rate among ethnic minorities, do you agree or disagree that there should be extra job training courses for ethnic minorities,’ ‘All foreigners who live on welfare should leave the country’ (reverse), and ‘Ethnic minorities should have the same political and social rights as the Dutch people.’ Cronbach’s alpha was 0.79 with higher scores indicating stronger support for equal rights.

National identification was assessed by asking the participants to respond to four items (4-point scales) that were similar as the one’s used in the first two studies. The items were, ‘I often think of myself as Dutch,’ ‘I consider myself a typical Dutchmen,’ ‘I am proud that I am Dutch,’ and ‘If someone said something bad about Dutch people I feel almost as if they said something bad about me.’ Cronbach’s alpha for the 4-item scale was 0.69.

Out-group threat was measured with five items (4-point scales) asking about different forms of threat by ethnic minority groups. The statements were, ‘I am afraid of increasing violence and vandalism in my neighborhood by ethnic minorities,’ ‘I am afraid that my economic prospects will get worse by ethnic minorities,’ ‘I am afraid of increasing violence and vandalism in Dutch society by ethnic minorities,’ ‘These days, I am afraid that the Dutch culture is threatened by ethnic minorities,’ and ‘I am afraid that the economic prospects of Dutch society will get worse because of ethnic minorities.’ The answers to these five statements were correlated and maximum likelihood estimation yielded a single factor that explained 51.78% of the variance. Cronbach’s alpha was 0.84 with higher scores indicating stronger feelings of out-group threat.

In this research authoritarianism was measured with four items and using 4-point scales. The items were taken from a large-scale study conducted in Italy (Sniderman, Peri, De Figuerido, & Piazza, 2000) and were ‘Whenever a private or public employer finds it necessary to reduce the number of employees, the first to be let go should be women who have a husband who is working,’ ‘Only the elderly, children, and handicapped should receive public assistance,’ ‘It is better to live in an orderly society in which the laws are vigorously enforced than to give people too much freedom,’ and ‘Rules are there for people to follow, not to try to change.’ Cronbach’s alpha for these four items was 0.59.

Results

Mean Scores and Intercorrelations

The participants indicated moderate levels of national identification (M = 2.71, SD = 0.81) and of authoritarianism (M = 2.31, SD = 0.75), a low level of perceived out-group threat (M = 1.94, SD = 0.85), and some support for equal rights for minority groups (M = 3.12, SD = 0.64). Table 2 shows the intercorrelations among the four measures. Higher national identification was related to greater authoritarianism and out-group threat but to a lower level of support for equal rights. Further, authoritarianism and out-group threat were positively associated and both measures predicted reduced support for equal rights.

Table 2 Intercorrelations for the different measures in Study 3 (N = 887)
Full size table

Equal Rights

A hierarchical regression analysis was conducted to predict the support for equal rights. The effects of gender, age, and educational level were entered on Step 1, and the measures for authoritarianism, national identification, and out-group threat were entered on Step 2. The full model explained 58% of the variance. The model in Step 1 was significant, F change(3, 884) = 44.37, p < 0.001. Age and education, but not gender, had significant independent effects. Again higher age was related to lower support for equal rights, beta = −0.09, t = 2.75, p = 0.006, whereas higher education was related to more support for equal rights, beta = 0.34, t = 10.62, p < 0.001.

The addition of the measures in Step 2 significantly increased the explained variance, R 2change  = 0.45, F change(3, 881) = 312.02, p < 0.001. All three measures were significant independent predictors: for authoritarianism, beta = −0.10, t = 3.84, p < 0.001, for national identification, beta = −0.07, t = 2.68, p = 0.008, and for out-group threat, beta = −0.66, t = 25.03, p < 0.001.

Table 2 shows that national identification was significantly related to equal rights and to out-group threat. In addition, out-group threat was negatively related to the support for equal rights. This pattern of results suggests that out-group threat mediates the relationship between national identification and equal rights. In the analysis in which the endorsement of equal rights was regressed onto out-group threat and national identification, the effect of identification was significantly reduced (from beta = −0.36 to −0.10), whereas the effect for out-group threat remained the same. The Sobel test confirmed that the mediational path was reliably >0, z = 10.91, p < 0.001. The result of this analysis supports the group-identity-lens model.

The reversed mediation in which national identification mediates the relationship between out-group threat and equal rights was also tested. The Sobel test was not significant, z = 1.46, p > 0.10. This indicates that the group-identity-reaction model did not fit the data.

In addition, it was examined whether national identification moderated the relationship between out-group threat and support for equal rights. The interaction term was not significant, beta = −0.02, t = 1.01, p > 0.10. Thus, there was, again, no support for the group-identity-moderator model.

The Role of Authoritarianism

Further analyses were conducted to examine the precise role of authoritarianism. The first possibility is that identification mediates the relationship between authoritarianism and out-group threat. Table 2 shows that authoritarianism was significantly related to out-group threat and to national identification. In the analysis in which out-group threat was regressed onto national identification and authoritarianism, the effect of identification was not significantly reduced. The Sobel test indicated that the mediational path was not reliably >0, z = 0.79, p > 0.10. Hence, there was no evidence that national identification mediated the relationship between authoritarianism and out-group threat.

The second possibility is that identification is moderated by authoritarianism. In a hierarchical regression analysis, the addition of the interaction between authoritarianism and national identification on Step 3 did not significantly increase the explained variance in the support for equal rights, F change(1, 880) = 1.66, p > 0.10, beta = −0.03. The interaction between authoritarianism and national identification also was not significant in a regression analysis predicting out-group threat, beta = 0.02, t = 0.68, p > 0.10.

Furthermore, it might be that authoritarianism and threat interact in predicting the support for equal rights (Feldman, 2003) or that there is a three-way interaction among authoritarianism, identification, and threat (Stellmacher & Petzel, 2005). An additional regression analysis indicated that neither of these interactions were significant (beta = −0.02, t = 0.54, p > 0.01, and beta = 0.01, t = 0.45, p > 0.10, respectively).

A Path Model

Figure 1 depicts a path model for influences on the support for equal rights based on results from the regression analyses. The partial correlation coefficients on the paths show the relative effects of the predictor variables on the endogenous variables, with the other variables influencing them held statistically constant. The pathways show independent positive effects of authoritarianism and national identification on perceived out-group threat, which, in turn is negatively associated with the support for equal rights. Further, there are direct negative pathways from authoritarianism and national identification to the support for equal rights.

Fig. 1
figure 1

Results of the path analysis for the support for equal rights in Study 3. Path weights are standardized. ** p < 0.01; *** p < 0.001

Full size image

Discussion

Using a national sample and focusing on the support for equal rights, the findings of the third study are similar to those of the first two studies. Again, and in agreement with self-categorization theory, it was found that perceived out-group threat mediated the relationship between national identification and, this time, support for equal rights for immigrants and ethnic minority groups. No evidence was found for reverse mediation in which national identification would mediate the relation between out-group threat and equal rights. In agreement with the first two studies, there also was no evidence for the group identity moderator model.

Study 3 also examined the role of authoritarianism. The findings show that authoritarianism had an independent positive effect on perceived out-group threat and a negative effect on the support for equal rights. There was no evidence for an indirect effect of authoritarianism via national identification. There also was no evidence that authoritarianism explains out-group threat or the endorsement of equal rights in interaction with national identification. These results are similar to Reynolds et al.’s (2007) experimental findings and strongly support self-categorization theory. It should be noted, however, that authoritarianism was measured with only four items and that the internal consistency of the four-item scale was not very high.

General Discussion

European societies are the reluctant recipients of immigrants and native populations feel threatened by the presence of ethnic minority groups (Jackson et al., 2001). Many consider the logic of the national and the multicultural as incompatible, and immigrants and ethnic minorities are portrayed as posing realistic and symbolic threats. The current research was conducted in the Netherlands and focused on the support for multicultural recognition and equal rights for immigrants and ethnic minorities. The central question was the role played by national identification in people’s perception of out-group threat and the support for minority rights.

Three different models were examined and the results of the three studies were very similar and in line with self-categorization theory (Turner & Reynolds, 2001). The group-identity-lens model fitted the data and there was little evidence for the group-identity-reaction model and no evidence at all for the group-identity-moderator model. It was found that national identification was positively related to perceived out-group threat, and that threat, in turn, was negatively related to the support for multiculturalism and minority rights. Self-categorization theory argues that group identity functions as a group lens that makes people sensitive and vigilant to anything that concerns or could harm their group. Thus, group identification leads to greater threat perception and once threat is perceived it leads to less support for immigrants and ethnic minorities. Thus, the current findings, together with those of other studies (see Riek et al., 2006), support the group-identity-lens model.

Little evidence (Study 2) was found for the group-identity-reaction model in which in-group identification is a consequence of perceived threats. In relation to minority groups there is experimental evidence for this model (e.g., Jetten et al., 2001) and the model is also supported in survey research among ethnic minority groups (Schmitt & Branscombe, 2002). The situation for majority groups is different, however. In European countries there is a long history of an established and large native population that has a powerful and secure position. These conditions do not make it very likely that threats posed by immigrants and ethnic minorities lead to group-based strategies involving increased group identification or the turning to the in-group as a resource for coping with these threats (Verkuyten & Rijerese, 2008).

In all the three studies there was no support for the group-identity moderator model. This model posits that national identification interacts with perceived threat to predict support for multiculturalism and minority rights. It turned out that in all analyses the interactions between national identification and perceived threat were not significant. In the context of Israel, Bizman and Yinon (2001) did find that perceived threat was a more important predictor of immigrant attitudes for high than for low identifiers. A similar result was found by Tausch et al. (2007) in the context of Northern Ireland. One possible explanation for these divergent findings is the type of threat involved. In the context of Israel, it was found that the effect of realistic threat was moderated by group identification, whereas in the context of Northern Ireland moderation was found for symbolic threat. The participants in the current research did not make a clear distinction between both types of threat and a single score was used. It might also be relevant that these previous studies focused on prejudice and trust and not on the support for multicultural recognition and equal rights. Another explanation is the particular contexts in which these previous studies have been conducted. Tausch et al. (2007) argue that the divergent findings for the type of threat in Israel and Northern Ireland are due to the relative importance of both threats in the two intergroup settings. However, both settings concern divided contexts of sectarian and violent conflicts. This is not the case in the Netherlands and in a previous research we also did not find supporting evidence for the group-identity-moderator model (Verkuyten & Yildiz, 2007).

In understanding out-group reactions, the social identity perspective focuses on group processes and intergroup relations rather than individual difference variables like authoritarianism. It is argued and found that in an intergroup setting, identification will produce intergroup behavior independent of these variables (e.g., Reynolds et al., 2001; Verkuyten & Hagendoorn, 1998, but see Heaven & St. Quintin, 2003). Some researchers have suggested, however, that personality-like variables may predispose people to be more or less affected by group conditions (Perrault & Bourhis, 1999; Sidanius et al., 1994). Authoritarianism, for example, would explain out-group reactions through or in interaction with in-group identification. However, findings from Study 3 indicated no evidence that authoritarianism affected perceived out-group threat and support for minority rights through its effect on national identification. There was also no evidence that authoritarianism in interaction with identification significantly predicted out-group threat and support for equal rights. These findings are in agreement with the results of two experimental studies conducted by Reynolds et al. (2007) and support self-categorization theory.

Authoritarianism was found to have an independent positive effect on perceived threat and a negative effect on support for equal rights. A relationship between authoritarianism (and also social dominance orientation) and negative out-group reactions is also found by Reynolds et al. (2007) under the condition of power differences and by Pratto and Shih (2000) under the condition of intergroup threat.

The fact that in the three studies relatively large samples were used and from different time periods, that different age groups were involved and that different questions were used for measuring minority support, provides strength to the conclusions, both in terms of stability and convergence of the results. Overall, there was little support for multicultural recognition (Studies 1 and 2) and more support for tolerance (Study 2) and for equal rights (Study 3). However, similar patterns of relationships were obtained for the different support measures. This similarity suggests that the level of abstraction of the target groups does not affect the associations among national identification, perceived out-group threat, and support for minority recognition and rights (cf. Watt et al., 2007).

In all the three studies it was found that age had a negative effect on minority support. This was found in the two adolescent samples in which the age range was relatively small and in the national representative sample of Dutch adults. In addition, participants with higher levels of education endorsed multiculturalism and minority rights more strongly (Studies 1 and 3) and females had more positive attitudes than males (Studies 1 and 2). These findings are similar to other studies that have examined the association of these factors to out-group attitudes (see Duckitt, 1992).

In evaluating the present results, two points are discussed. First, the research was correlational and therefore no claims about causal sequence can be made. The data provide strong support for the group-identity-lens model. However, it is possible that, for example, multiculturalism affects group identification (Verkuyten, 2005; Wolsko, Park, Judd, & Wittenbrink, 2006). Hence, it would be useful to examine the current relationships using an experimental design. Furthermore, longitudinal research would be helpful to understand the relationships better. For example, it may be the case that over time threat gradually increases national identification as predicted by the group identity reaction model.

Second, the level of national identification was a focus, because I was interested in the strength of attachment to the national category. However, group identification is a multidimensional construct and different dimensions of national identification might be related differently to perceptions of out-group threat and the support for multiculturalism and equal rights. Furthermore, not only is the level of identification important, but also is the content of the national identity. Self-defined multicultural and immigrant countries might react differently to newcomers than nonsettler countries with an established and dominant majority group.

In conclusion, this research suggests that national identity and the perception of threats posed by immigrant groups are important factors in people’s support for multiculturalism and equal rights. The findings of the three studies are in line with self-categorization theory and in agreement with the group-identity-lens model. Public debates on immigrants and ethnic minorities tend to focus on the supposed threat to national identity and culture. High national identifiers feel that national interests, beliefs, and values are threatened by newcomers and minorities, and as a result they tend not to support multiculturalism and minority rights.