|
Abstract:
|
This paper will highlight how the UK Office for National Statistics (ONS) has effectively utilised responders who do not object to re-contact at the final wave of the Labour Force Survey, a large UK household rotating panel survey, as a sample frame for future surveys. We will highlight both the practical and statistical impact of using this approach, quoting qualitative feedback and quantitative results from several recent examples.
The major advantages of using an existing survey as a sampling frame include cost-effectiveness, timeliness, and the ability to effectively target sub-populations without the use of a screener survey. However, response rates to later waves of the Labour Force Survey are poor - recently, below 50% - meaning non-response bias is a major concern. This paper will briefly outline the statistical techniques we use to minimise the risk of non-response bias, including in particular two-phase weighting, and the knock-on impact these methods have on the variance of estimates.
|
