"Theoretical background and objectives
This project aims to shed new light on the causes and consequences of ethnic residential segregation, i.e. ""the degree to which two or more [ethnic] groups live separately from one another in different parts of the urban environment"" (Massey and Denton 1988: 283). While the number of papers dealing with this classic research question is large, the debate whether, to what extent and for whom ethnic residential segregation matters is far from resolved. This project addresses two issues in particular, namely interaction effects between individual and context characteristics, and the question to what extent segregation results from immigrants' deliberate choices to live among co-ethnics, or from such homphily preferences on the side of members of the majority population. Surprisingly, empirical studies investigating the prevalence and causes of immigrants' residential preferences remain scant. Guided by the preference model of residential choices (Charles 2003), we examine under which conditions and how the residential preferences of ethnic minority and majority members reflect a desire for self-segregation and avoidance of other ethnic groups or not. This study is likely to yield critical findings for both theory and applied initiatives, given that investigating the prevalence and the sources of segregation preferences is of key importance for understanding macro-level patterns of ethnic residential segregation.
Research design, data and methodology
In a first study, we applied multilevel generalised linear regression techniques to individual level survey data from a large metropolitan area (Duisburg) in Germany, supplemented with contextual measures of ethnic residential segregation on the neighbourhood level. We examined whether patterns of segregation were related to rates of interethnic contact, and whether this relationship differed for respondents of different socio-economic status. In a second study, we used factorial survey methodology to address majority members' preferences. One key advantage of this design is that it avoids the notorious problem of collinear contextual variables when investigating neighbourhood settings. In two within-subjects experiments conducted over the internet (total N = 1032), participants evaluated schools or residential areas with different levels of ethnic diversity (i.e. proportions of immigrants). In the vignettes describing schools and areas, we additionally varied factors that are ecologically related to diversity (i.e., neighbourhood socio-economic status and crime in residential areas, and quality of education at schools). At the person level, we measured intergroup contact and prejudice and used these variables to predict the level 1 effect of diversity on preferences for residential or school choice. We estimated a two-level random coefficients model with latent variables to explain preferences. In a third study, we will employ also an experimental factorial survey design, but this time to investigate immigrants' residential preferences. We will use quota samples of different ethnic minority groups living in Germany (e.g. Turks). Respondents will evaluate vignettes describing different residential areas which, in addition to the size of the ethnic in-group, vary systematically along additional dimensions known to affect residential choices such as neighbourhood SES, ethnic infrastructure or crime risk.
The first study has been completed, the second is ongoing and the third will be started in early 2011. Controlling for individual characteristics, results from the first study bring new evidence that friendships of immigrants with host society members are less prevalent in residential areas with greater degrees of ethnic segregation. The strength of this negative association, however, proves to be contingent on immigrants' educational attainment: The lower one's educational attainment, the stronger the negative association between ethnic residential segregation and immigrants' interethnic friendships. In other words, residential segregation is in particular detrimental for those sections of immigrant population for whom interethnic contacts are likely to be most important as a source of social capital, namely those of low socio-economic status. Preliminary results of the second study show that diversity had negative effects on evaluations of schools and residential areas, over and above the effects of infrastructure, crime, or quality of education. Furthermore, results indicate that intergroup contact reduced bias against diverse schools or residential areas, mediated by prejudice, but it did not produce a preference for diversity, except for people with prejudice scores as low as the sample minimum."