Intergroup Distinctiveness and Discriminatory Immigration Attitudes: The Role of National Identification

Abstract

We examined the moderating role of national identification in understanding when a focus on intergroup similarity versus difference on ingroup stereotypical traits—manipulated with scale anchors—leads to support for discriminatory immigration policies. In line with intergroup distinctiveness research, national identification moderated the similarity–difference manipulation effect. Low national identifiers supported discriminatory immigration policies more when intergroup difference rather than similarity was made salient, whereas the opposite pattern was found for high national identifiers: They trended toward being more discriminatory when similarity was made salient. The impact of assimilation expectations and national identity content on the findings is discussed.

ACKNOWLEDGMENTS

Both authors contributed equally to this article. We thank Christian Staerklé for helpful comments on a previous draft.

Notes

¹In preliminary analyses, two scores—a restrictive and an empowering immigration policy attitude—were distinguished. The results patterns were very similar and the scores were highly correlated (r = .65, p < .001, when empowering immigration policy items were reversed). Therefore, for the sake of parsimony, the scores were combined in the final analyses.

²The distribution of residuals in the presented regression analyses do not meet the homoscedasticity requirement, which might result in incorrect estimates of the variance, leading to inconsistent and uninterpretable t statistics for the parameters (White, 1980). Therefore, using robust standard errors ensures inference consistency (for a discussion, see Antonakis, Bendahan, Jacquart, & Lalive, 2010). The regression coefficients are identical to the estimates of ordinary least squares regression analysis, but the standard errors are robust against failure to meet homoscedasticity assumptions. In our study, ordinary least squares regression analyses revealed essentially identical results.

³In preliminary analyses, Intergroup Comparison Ratings × Type of Intergroup Comparison and Intergroup Comparison Ratings × National Identification interactions were also separately included in the model. These interaction terms, however, were not significant. It is important to note that the predicted National Identification × Type of Intergroup Comparison interaction pattern remained significant.

⁴That intergroup comparison ratings did not differ between the experimental similarity–difference conditions may at first seem surprising. This finding indeed suggests that drawing attention to similarities does in fact not make a group seem more similar than drawing attention to differences. However, this does not undermine our reasoning and interpretation of our main findings. The responses to the two types of scales (“how similar” vs. “how different” are ingroup members from outgroup members) are not strictly comparable even when the responses are reverse coded. Identical strength of comparison ratings does not imply that the retrieved representations of immigrants (i.e., exemplars) are the same (Tourangeau, Rips, & Rasinski, 2000). That is, the scale anchors do not necessarily carry the same psychological meaning for participants (e.g., that “very similar” would be identical to “not at all different”). Instead, different processes may be at work in the two conditions. Because “East European immigrants” are a vast and highly diverse group including several nationalities, the retrieved representations of immigrants (i.e., the exemplars) in the two conditions may not be the same. In the similarity condition, Swiss participants may more easily think of “good” (i.e., hardworking, competent, and ambitious) East Europeans who are similar to them, whereas in the difference condition they may activate representations of “bad,” lazy, and incompetent immigrants. It seems plausible that activation of stereotypes of different subgroups of East European immigrants and the corresponding distinct representations of similarity and difference drive the effect rather than the actual magnitude of perceived intergroup similarity and difference. Importantly, this explains why the effect of the experimental manipulation occurred over and above the (covariate) effect of comparison ratings.