Design and Estimation of Surveys to Measure Data Quality Aspects of Administrative Data

Abstract

  National Statistics Institutes (NSIs) have been increasingly seeking to replace or enhance traditional survey-baseddata sources with administrative data sources; with the aim to improve overall quality in the absence of a definitive register ofthe population. The Beyond 2011 Census Programme in England and Wales is an example of looking to replace a traditionalcensus with administrative data collected for another purpose by a different organisation, when there is no definitive registeras a starting point. There are also similar projects across NSIs within the area of business surveys looking to useadministrative sources to reduce cost and burden. In this paper we start with considering all aspects of a quality frameworkfor administrative data and then focus on the elements relevant to data quality such as accuracy and coherence. We fit theseconcepts into the framework for total survey error highlighting the components an NSI needs to measure to produce estimatesbased on the administrative data. We then explore the use of both dependent and independent quality surveys to adjust theadministrative data for ‘measurement’ and ‘coverage’ aspects to improve the quality of estimates produced from theadministrative data.