By Whom Do You Want Your Wallet to be Found in Ethnically Diverse Neighbourhoods?

Wednesday, July 16, 2014: 6:15 PM
Room: Booth 54
Oral Presentation
Jochem TOLSMA , Radboud University Nijmegen, Netherlands
Tom VAN DER MEER , University of Amsterdam, Netherlands
The claim that ethnically diverse living environments drive down social cohesion between and within ethnic groups, remains disputed, despite abundant empirical tests performed all over the globe. The Netherlands have proven to be an intriguing test site for this so-called ‘constrict proposition’, as numerous studies – using different operationalizations of ethnic diversity and social cohesion, different data, performed in different time era and defining macro-level geographical units differently – reached rather similar conclusions: in general, ethnic diversity does not erode social cohesion. That said, within ethnically diverse neighbourhoods, trust in neighbours is indeed hampered. Yet, various questions remain. First, does ethnic diversity only drive down trust in neighbours of ethnic outgroups or also trust in coethnic neighbours? Second, is the neighbourhood the crucial geographical unit of which one would expect (negative) diversity effects? And related to the second question how should we define the neighbourhood and consequently operationalize neighbourhood diversity? In this contribution we answer these questions through analyses of the Religion in Dutch Society 2011-2012 dataset, supplemented with unique GIS-data. Our dependent variables are so-called wallet-items in which we systematically varied the ethnicity  of the target of trust (co-ethnics versus outgroup members) and the location of the target of trust (neighbours versus non-neighbours). We introduced spatial thinking into the diversity-cohesion discussion and investigated to what extent the impact of ethnic diversity depends on the definition of the neigbhourhood (administratve units versus egohoods) and on the geographical scale of the neighbourhood. We applied spatial regression techniques and multi-level models to test our hypotheses.