The contact–prejudice relationship among ethnic minorities: examining personal discrimination as a boundary condition

ABSTRACT

This paper contributes to the literature on how perceived discrimination influences the impact of intergroup contact among ethnic minority members. Previous research has shown that perceived group discrimination inhibits the positive outcomes of intergroup contact among ethnic minority members. In extension, this paper examines whether perceived personal discrimination has a similar negative impact as a boundary condition of the contact–prejudice relationship among ethnic minority members. Analyses expand previous intergroup contact research by showing that the impact of intergroup contact is uniform across various levels of perceived personal discrimination. This finding suggests that the otherwise harmful consequences of perceived personal discrimination can be effectively counterbalanced by positive contact experiences among ethnic minority members. Analyses use evidence from a national probability sample, which was fielded in 2006 to examine some of the largest ethnic minority groups (N = 3,462). The concluding section discusses the theoretical implications of the empirical findings.

KEYWORDS:

• Ethnic minority members
• intergroup contact
• prejudice
• discrimination
• moderation analysis
• national sample

Acknowledgements

The authors would like to thank the Editor, the two reviewers and our colleague Alan Würtz for many useful suggestions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. This data set was not designed or funded by us. Rather, the data set was funded by a Think Tank (related to The Ministry of the Interior). The questionnaire was designed under the guidance of leading Danish scholars in intergroup relations, Professor Peter Gundelach (Dept. of Sociology, Copenhagen University) and consultant Esther Nørregård-Nielsen in particular. Peter Gundelach and Esther Nørregård-Nielsen have also published a report (in Danish) about the investigation, which is available on the Web (http://www.rso.dk/wp-content/uploads/2013/03/taenketanken_vaerdier_og_normer_baggrundsrapport.pdf).

2. The personal registration number system is a unique feature in an international context. In Denmark, all persons with legal residence status have a personal registration number. In efforts to meet the criteria of genuine probability sampling, this system is close to ideal because it ensures that each person has an equal and known chance of being selected.

3. Nonetheless, the response rate in our data does not seem to differ from samples among ethnic minority members in other countries (for a similar Dutch investigation, see Teije, Coenders, and Verkuyten 2013).

4. The mean age in our sample (minus native Danes) is 32.9 (std. dev. = 9.4).

5. The so-called mean procedure is one way of handling ‘don't know’ responses while also protecting the representativeness of the sample. In our case, individuals who responded ‘don't know’ to three or four items in the index were excluded from the analysis (implying a sample loss). These respondents do not provide sufficient information about their attitudes. In contrast, and to exemplify, person A has responded ‘don't know’ to one question and responded ‘agree’ to the other three. Assuming that her responses are relatively consistent, the average score of the three genuine responses is transferred as a score to the ‘don't know’ item.

6. As indicated by the standard deviation compared to the mean, the distribution of the prejudice measure is positively skewed. In additional analyses, we examined in depth to what extent our results are driven by outliers by plotting studentized residuals against leverage values in order to identify influential observations. This test was negative as there were no observations in the critical areas. We also reran all analyses in a censored (Tobit) regression model. The results from this additional analysis were fully consistent with those with report in the text on the basis of an OLS (ordinary least squares) regression model. In sum, our reported relationships are not affected by the distribution of the dependent variable.

7. Using an F-test, we examined whether the effect of this metric measure of contact differed from an alternative dummy solution. The difference between the estimates of the two measures was not statistically significant.

8. Unfortunately, the data set does not offer any measure of perceived group discrimination.

9. Formally, our regression equation is specified as follows: