Mixed Mode Design and Respondent Incentives: Effects on Individual Full Response Outcomes in a Longitudinal Survey Context

Abstract

The present study aims to characterize the effects of a sequential mixed mode survey design and different levels of respondent incentives on the individual full response outcome in a longitudinal survey context in the UK. To reach this purpose, the Understanding Society Innovation Panel waves five, six and seven were analysed. Logistic regression, random effects models, and fixed effects models were used as complementary analysis. The results show that there is no statistically significant effect of mode design on the propensity of being respondent when the level of incentive, sample origin and other basic geographical and personal characteristics are controlled. This evidence refutes the first hypothesis of the study, which predicted a statistically significant impact from mixed mode survey design compared to single mode design. In contrast, the same analysis showed that the level of incentive has a statistically significant impact on the likelihood of being respondent. People who received a high level of incentive or, to a lesser extent, a middle level of incentive, were found more likely to be a respondent compared to people who received a low level of incentive. These results support the second research hypothesis, which predicted this effect. Overall, the findings suggest that, in the context of an ongoing longitudinal survey, a higher level of incentive contributes to an increase the individual response rates, and the incorporation of a sequential mixed mode design of web and face-to-face does not affect them, at least in the initial waves.