Further Analysis of Non-Response Biases in the 2004 Workplace Employment Relations Survey

WERS experienced a substantial decline in the response rate for the main management interview between 1998 and 2004 (from 80% to 64%). Whilst non-response does not necessarily introduce bias into the achieved sample for a survey, higher rates of non-response necessarily provide greater leeway for biases to occur. The purpose of this short project was to investigate the extent of workplace-level non-response bias in the WERS 2004 cross-section by looking at a wider range of information on sample members than had previously been examined. This involved matching on paradata collected during survey fieldwork and also data items from ONS business datasets held in the Virtual Microdata Laboratory.