Mobile and mixed-device surveys
|Dr Vera Toepoel (Utrecht University )
|Mrs Marika De Bruijne (CentERdata)
|Mr Arnaud Wijnant (CentERdata)
Although tablets are increasingly used for web survey completion, it remains unclear whether responses obtained in tablet-based web surveys differ from those collected via PC. A lab experiment is conducted to assess the impact of questionnaire design on data quality in tablet surveys. Using a cross-over design, all participants complete the questionnaire both on PC and tablet. The laboratory setting allows controlling for the context of survey administration and holding technical features constant. A subset of respondents participates in cognitive interviews subsequent to the experiment. The findings of the study will be available in April 2015.
Studies that investigate the effects of survey completion via mobile devices using the data from probability-based general population surveys are scarce. It is not clear whether their findings would hold in other probability-based panels. We replicate and extend the recent study on the effects of smartphones and tablets on measurement error by Lugtig & Toepoel (2015, forthcoming), which uses the data from a probability-based general population LISS Panel in the Netherlands. In addition to the indicators of measurement error and the effects of switching between devices used in the original study, we focus on respondent and situation characteristics.
The study reported here uses a crossover experiment in a probability-based Web panel to compare data quality in a conventional Web survey (PC Web) to a version of the same survey that was reformatted for small screens and filled out on smartphones (mobile Web). I found that respondents in the mobile Web survey really were more mobile and more engaged with the other people and things around them. Despite this, I found there to be no terrible dangers in mobile administration. Response quality – conscientious responding and disclosure of sensitive information – was equivalent between mobile and PC Web.
Can we move web surveys from a browser to an app on mobile phones? The Pew Research Center randomly assigned 2,011 members of their American Trends Panel to receive a series of short surveys either on a mobile app or on the web. This experiment used signal-contingent experience sampling to survey respondents twice a day for seven days about their smartphone use. A significantly higher proportion of panelists in the web treatment than the app treatment responded to at least one of the 14 survey invitations. Other advantages and disadvantages of the two modes will be discussed.
This paper examines the use of web-based, automated SMS text messages in public opinion research in an attempt to cost-effectively reach more diverse samples of citizens. A sample of 1,000 mobile subscribers was contacted via a “cold text” asking them to participate in a short survey regarding their opinion on a local park (raffle for iPad mini was incentive). Non-respondents received follow-up phone calls to determine their reason(s) for not responding. Findings suggest automated SMS text messages are an effective way to measure public opinion, and cost effective when compared to other methods.