What are the Characteristics of Respondents using Different Devices in Mixed‐device Online Surveys? Evidence from Six UK Surveys

Abstract

Summary There is a move towards online data collection across the world. Online survey response is complicated by respondents using different devices. So far, no research has been conducted in the UK to study characteristics of people using different devices in mixed‐device online surveys. This analysis uses all publicly available UK social surveys with an online component: Understanding Society Innovation Panel, Community Life Survey, European Social Survey, 1958 National Child Development Study and the Second Longitudinal Study of Young People in England. Descriptive analysis and logistic regressions are used to explore significant correlates of device use in online surveys. The results of bivariate analysis suggest that age, gender, marital status, employment, religion, household size, children in household, household income, number of cars and frequency of internet use are significantly associated with device used across surveys. The associations with age, gender, employment status, household size and education are consistent with the findings from other countries. The knowledge about respondents' characteristics using different devices in online surveys in the UK will help to understand better the response process in online surveys and to target certain subgroups more effectively. It is also important for designs of online surveys, understanding of data quality and post‐survey adjustments.

Characteristics of Respondents to Web-Based or Traditional Interviews in Mixed-Mode Surveys. Evidence from the Italian Permanent Population Census

“…Finally, the estimate indicates that the effect of education in choosing a CAWI prevails over the other determinants. The results are coherent with those of Keusch et al (2019), Maslovskaya et al (2019), and Mohorko et al (2013). In fact, education in general been found to be among the determinants of response in surveys, with a higher level linked to a higher propensity to answer through the CAWI option (Hox et al 2015).…”

“…One of the most common methods for controlling for differential nonresponse in mixedmode surveys is to use the estimated propensity of a respondent to participate in each mode. These propensity scores are typically estimated from a generalised linear model (e.g., probit, logit), where a given mode is treated as a possible outcome, conditioned on available covariates information, often limited to participant demographics, as is done by Hox et al (2015), and more recently by Maslovskaya et al (2019) and Rivero et al (2019). However, in this article, we propose a procedure that would allow survey design researchers to control the distance from the CAWI-H profile in the actual data collection and also before proceeding to a new phase of data collection, while propensity scores can only be estimated after collecting the data.…”

Behavioral and Demographic Predictors of Exercise During A COVID-19 Lockdown in Three Ghanaian Cities

“…By utilizing a non-powered sample in this study, our results may not be generalizable to other cities. In harmony with previous studies [39,40], most of our sample is made up of highly educated people, which means that residents who were uneducated or poorly educated were underrepresented in the study. As a result, studies that employ representative samples and more resilient designs (e.g., randomized controlled interventions) are needed in the future.…”

“…Older adults were underrepresented in the study possibly because older adults in Ghana hardly use the internet and social media platforms [40]. With most older adults in Ghana having poor English skills [39,40], our reliance on a survey administered solely in English could have prevented some older adults from participating. Our results are ideally applicable to educated populations aged 18 to 55 years.…”

Comparing Internet and phone survey mode effects across countries and research contexts*

“…In particular, the chosen survey mode may introduce biases into the study that stem from various behavioral consequences of the survey-respondent ecosystem. Past research in a variety of academic disciplines, contexts, and geographies has checked for these forms of bias between different survey modes by comparing obtained responses and corresponding parameter estimates across modes and even across variants of the same mode (Maslovskaya et al, 2019;Struminskaya et al, 2015). The past research has inconclusive results, as discussed in detail in section 1.2 below.…”