Interviewer Effects in Biosocial Survey Measurements

Abstract

Increasingly surveys are using interviewers to collect objective health measures, also known as biomeasures, to replace or supplement traditional self-reported health measures. However, the extent to which interviewers affect the (im)precision of biomeasurements is largely unknown. This article investigates interviewer effects on several biomeasures collected in three waves of the National Social Life, Health, and Aging Project (NSHAP). Overall, we find low levels of interviewer effects, on average. This nevertheless hides important variation with touch sensory tests being especially high with 30% interviewer variation, and smell tests and timed balance/walk/chair stands having moderate interviewer variation of around 10%. Accounting for contextual variables that potentially interact with interviewer performance, including housing unit type and presence of a third person, failed to explain the interviewer variation. A discussion of these findings, their potential causes, and their implications for survey practice is provided.