While there is reason to hypothesize both educational attainment and selection effects, their specific impact has not been addressed yet. Previous work relied exclusively on cross-sectional analyses, thus confounding the two mechanisms. If educational affects attitudes, there should be changes in attitudes as people pass through educational levels. If, on the other hand, selection into education is an important explanation, the direct effect of education should be much smaller.
Drawing on the Swiss Household Panel (1999-2011), we find that virtually all variation in education disappears when within-person variance is modeled. While we find large differences in attitudes toward immigrants between people with different levels of education, we do not observe changes in attitudes as people pass through education. In addition, we find evidence that parental background affects individuals’ attitudes. Overall, our results suggest that future research on education and attitude formation would benefit from addressing selection into education.