Coverage and non-response errors in the UK New Earnings Survey

The methods underpinning the UK's annual structural earnings survey-the New Earnings Survey-have remained largely unchanged since the survey's inception in 1970. Gradual changes in the labour market over recent years have led to coverage errors in the survey; non-response may also introduce error if it is non-random. The paper describes some of the New Earnings Survey's main non-sampling errors, the effect on survey results and the research that was undertaken by the Office for National Statistics as part of the development of the new Annual Survey of Hours and Earnings to be able to remove or otherwise to account for these errors. Copyright 2007 Royal Statistical Society.