Urbanization effects on survey nonresponse: a test within and across cities

This paper addresses differences in survey response across city size by applying an explanation from House and Wolf to areas within cities. The data are from Toronto, Hamilton and Kitchener, in Canada. Consistent with other literature, response falls sharply from largest to smallest place, but variation in response rates within the cities is negligible. The intra-city analysis is accomplished by factor analyzing variables such as population density, crime rate and ethnic structure, mostly computed for census tracts, into a variable labelled social disorganization. What House and Wolf termed compositional variation in response is handled by record linking with municipal assessment information. The analysis is based on micro data taking the people listed for the sample as unit of observation.