2020-12-02, 13:45–14:00, Zoom Breakout Room 1
Survey data collection underpins a large proportion of social science research across multiple disciplines, but is increasingly difficult: response rates are decreasing, and methods to sample national populations are growing more expensive and complex. In this presentation, we will present results of a data collection experiment to test optimal ways to recruit survey respondents via mobile phones.
Our approach uses random-digit selection of mobile phone numbers combined with SMS invitations, with respondents asked to complete the survey online and a URL directing them to an online questionnaire, consisting of primary and secondary socio-demographic questions, as well as questions on the use of internet, health, technology, life satisfaction, and political attitudes.
The benefits of this approach to sample recruitment is its simplicity and cost effectiveness and could be used in the future by students, academics, and social and market research companies. Traditionally, cross-sectional general population surveys use many other recruitment approaches: mail outs, telephone calls, or face-to-face contacts. Text messaging is, generally speaking, predominantly used as an additional communication channel and for sending reminders.
The main aim of the project is to causally identify through random assignment practices affecting response rates in survey research using this sampling type and online survey mode. A number of data collection characteristics are randomized and later used as predictors of survey (non)response. To test for the effects of different incentives, quotas are set based on the incentives offered: one third of the final sample with no incentives offered, one third to enter a price draw, and one third $5 monetary incentives or charity donations. Further, our approach randomly allocates invites to different days (2 groups: weekends, weekdays), at a different time (3 groups: morning, afternoon, evening), and with 3 different SMS invitation texts.