HERC Seminar: Using experiments to improve the design of surveys eliciting past health care usage

Self-reported information on past utilization of health care is routinely used by many health researchers. However, the recall period of questions varies widely between surveys and this variation may affect the results of analyses. This seminar examines the role of length of recall period, mainly using a large experiment conducted in Sweden of self-reported hospitalizations of respondents. Respondents were asked a question on the number of nights they spent in hospital in which the recall period varied between one and twelve months. Our findings have conflicting implications for survey design as the preferred recall length depends on the objectives of the analysis. We will conclude by outlining planned experiments to build a better evidence base for designing surveys.