Abstract

To explore determinants of data quality in the European Social Survey. We use two measures of data quality: an internal measure: gender ratios and an external measure: how well the ESS data replicates LFS estimates which are assumed to be of higher quality. Only the second measure relates to the data we request with this application. We study how these two measures of data quality are related to response rates (a traditional measure of survey quality) and sample type. In our research to date, it appears that response rates rae not a good indicator of data quality, in contrast to what many reserachers believe. Sample type, on the other hand, does relate to data quality -- some sample selection methods, such as the random walk method, are associated with lower quality data.