Survey mode change by necessity: Evaluating survey dataset performance in a global pandemic

Abstract

The COVID-19 pandemic affected how surveys were administered. Web based designs were used instead of face-to-face (F2F) interviewing, and data collection was more frequent. Changes were made with limited information about impacts survey estimate non-response biases and their causes. We examine the extent to which information on such questions in pre-pandemic surveys was predictive of similar in pandemic surveys in two Understanding Society: the UK Household Longitudinal Study (UKHLS) datasets: the main survey and the COVID-19 Study. We find that main survey information was of limited predicted value for the COVID-19 Study. The implications of these findings are then discussed.