Understanding Nonresponse in Live Video Survey Interviews: Evidence from Two Tests Conducted as Part of the American National Election Studies (ANES)

Abstract

Live video interviewing is an attractive mode for survey research because it has the potential to supplement or, someday, even replace in-person interviewing at a reduced cost and without sacrificing data quality. The reasons for this optimism are several: live video interviewing maintains some of the desirable elements of in-person interviews, such as visual cues and rapport from the interaction between respondents and interviewers, while potentially conferring additional advantages, such as access to respondents living in remote locations. However, previous evidence and experience suggest respondents are reluctant to be interviewed via video, and that reluctance is non-ignorable. This paper explores these recruitment and non-response challenges, informed by the experience of two tests conducted by the American National Election Studies (ANES). The first study is the ANES 2020 Time Series Study, where respondents were randomly assigned to three sequential mode conditions, one of which included live video interviewing as the initial mode. The second study is the ANES 2023 Video Interviewing Pretest, in which respondents were recruited to an in-person interview with a one-month follow-up conducted by video. Respondents to the 2020 ANES video mode were more educated relative to benchmarks and more likely to be Democrats than the rest of the sample. Both qualitative and quantitative data suggest video non-response is heavily related to discomfort with technology, lack of skills, and distrust or dislike of video, and lack of access to equipment. Finally, offering prepaid incentives resulted in more video completions than postpaid incentives.