Gaia, Alessandra; Brown, Matt; Adali, Tugba; Fleetwood, Stella; Lai, Christy; (2024) Response burden and survey participation: experimental evidence on the effect of interview length on non-response conversion. (CLS Working Paper 2024/5). UCL Centre for Longitudinal Studies: London, UK.

Preview

Text CLS-Working-Paper-2024-5-Response-burden-and-survey-participation.pdf - Accepted Version Download (305kB) | Preview

Abstract

In a context of declining response rates, identifying best practices for non-response conversion is of key importance for achieving adequate sample sizes and augmenting representativity in both cross-sectional and longitudinal studies. This paper examines whether re-contacting non-respondents after completing face-toface fieldwork in the latest wave of a longitudinal study, which uses a web and faceto-face mixed-mode approach, and inviting them to participate in a web-based nonresponse conversion survey, can increase participation. In addition, we examine whether reducing interview length at this final stage can lead to a more significant boost in response. We discuss the trade-off in loss of information collected versus higher sample size and representation. In order to address our research aims we use experimental data from the Next Steps Age 32 Survey. Three months after the end of fieldwork, a random half of those who had not participated to the survey were re-invited to participate in the full 60 minute version of the survey while the remaining half of the sample was allocated to a ‘short’ 20 minute version. The full survey instrument included a cognitive assessment, requests for data linkage consents and request to provide a saliva sample for DNA extraction while, the “short” survey instrument did not include these additional requests. Overall, compared to the longer version, the shorter non-response conversion survey led to a higher response rate. Analysing differences by sociodemographic characteristics and previous wave response behaviour, we find that response rates increase significantly for previous wave non-respondents allocated to the “short” version (versus “full” version), while no difference on response rates based on survey length is observed among previously cooperative sample members. This finding suggests that the offer of shorter instruments seems a promising practice for obtaining at least some information from the sample members which are particularly hard to convince to take part otherwise.

Download activity - last month

Export as JSONCSVXML

Download activity - last 12 months

Export as CSVJSONXML

Downloads by country - last 12 months

Export as XMLCSVJSON

Archive Staff Only

View Item