A1424
Title: A widely applicable sampling method for spreading the survey burden in business surveys
Authors: Marc Smeets - Statistics Netherlands (Netherlands) [presenting]
Harm Jan Boonstra - Statistics Netherlands (Netherlands)
Abstract: Due to the heavy perceived Survey Burden on businesses, Statistics Netherlands (SN) has started a pilot to prevent individual businesses from being sampled disproportionally often. Small businesses with disproportionately high Survey Burden (hotspots) are excluded from new samples or replaced by other units where possible. An important drawback of this pragmatic approach is the risk of biased estimates. SN operates a Sampling module for business surveys, which allows Survey Burden to be spread among businesses without the risk of bias. A reduction in the Number of hotspots could be achieved by spreading the Survey Burden across more surveys using the Sampling module. However, to spread the Survey Burden across surveys, the current Survey Burden Sampling (SBS) method in the Sampling module requires that stratifications of these surveys be aligned, which proves difficult in practice. A new Sampling method, Rotating Random Number Sampling (RRNS), is investigated, which is similar to SBS but with the advantage that stratifications need not be aligned. Different versions of RRNS are examined in a simulation study, where the Survey Burden is spread across multiple surveys including both cross-sectional and Rotating panel surveys, and dynamic populations are considered. The results demonstrate that RRNS could be a useful addition to the Sampling module for further reducing the Survey Burden on businesses.
