Authors
Abstract
Smartphone apps are increasingly being used for a variety of studies involving intensive measurement. Many of these studies are conducted among small groups of volunteers where selection bias is of less concern and ways to increase participation have been largely ignored. As these app-based studies are being included in large population-based surveys to permit broader inference and extend measurement across a broad range of topics, concerns over participation (or otherwise) in such studies and the implications for selection bias are becoming more salient. Given this, we conducted an experiment to test the effect of various features of an app study protocol on uptake or participation. Respondents in a large nationally-representative longitudinal survey in Great Britain were asked to download an app and use it every evening for 14 days to answer questions about their experiences and wellbeing that day. We experimentally varied three features of the study protocols: Inviting participants to the app study earlier in the annual interview increased participation rates, while a shorter daily questionnaire and bonus incentives for completing all 14 days had no significant effect. We also examine the effects of study protocols on non-participation bias, as well as the problems respondents had with installing the app and reasons for not participating.
Subjects
Link
https://www.understandingsociety.ac.uk/research/publications/547673
Paper download