Dinesen, Peter Thisted2; Sønderskov, Kim Mannemar3
1 Department of Political Science, Aarhus BSS, Aarhus University2 Institut for Statskundskab, Københavns Universitet3 Department of Political Science, Aarhus BSS, Aarhus University
The question about how ethnic diversity affects social trust has been a hot topic in recent years. To this point, within-country analyses of this question have been limited by only having data on contextual ethnic diversity at relatively high levels of aggregation. Consequently, the previous analyses suffer from the problem that aggregate contextual diversity likely conceals substantial variation in the ethnic diversity actually experienced at the micro-level in which people live and interact (i.e. contains measurement error). This in turn renders the estimate of ethnic diversity on trust both imprecise and downward biased (due to attenuation bias). In the present paper we add to the literature by analyzing, as the first study ever, how ethnic diversity of the immediate micro-context affects people’s trust in others. In addition, we compare the effect in the micro context to the impact of ethnic diversity at higher levels of aggregation in order to scrutinize how the relationship varies according to the contextual unit in which ethnic diversity is measured. We analyze the question about the impact of ethnic diversity on trust using Danish data from the European Social Survey, which are linked with data from the national Danish registers. The latter data contain detailed information about the ethnic background and the address of everyone living in Denmark. This enables us to obtain precise measures of ethnic diversity of the immediate surroundings in which each respondent lives. In the analysis we include measures of ethnic diversity in contextual units ranging from a radius of 80 meters up to 2750 meters within the address of a given respondent. In line with our expectations, the results show that increased ethnic diversity in the immediate surroundings affects generalized trust negatively, while the effect vanishes in larger contextual units.
Main Research Area:
70th Midwest Political Science Association Conference, 2012