The Late Responding Phenomenon in Web versus Mail Surveys: A Systematic Review and Meta-Analysis

Abstract

With the rise of web surveys to the currently most widely used mode in the survey method, a need to systematically review and aggregate the myriad of detailed insights produced by web survey research has developed. Considering the identified representative and response rate differences for web surveys compared to other modes and given the differences in survey practices between modes, a relevant topic, which we do not know whether it is more or less pronounced for the web survey mode, is “late responding.” There is a fundamental concern that respondents who complete the survey with a certain delay only after one or more additional contact attempts are less motivated to provide survey data of high quality. This poses a threat to the accuracy of survey estimates. With a systematic review and a meta-analytic approach, we address the questions of whether and to what extent late responding is different for web surveys compared to other self-administered survey modes, e.g., mail surveys. As there is a broad range of definitions of late responding in the reviewed 74 studies, a categorization attempt identified three distinct categories. With a meta-analytical approach, a mean share of 27% (CI: 23%–31%) of late respondents across modes was quantified, and no mode difference was found. An additional moderator analysis with 18 sample and survey characteristics did not identify a robust moderator across modes. In addition, this research provides a detailed overview of the different survey practices used in web and mail surveys.