Findings (Nov 2024)

Predicting Response Rates Once Again

  • Daniel Heimgartner,
  • Kay W Axhausen

Abstract

Read online

We investigate the relation between a survey's response burden and response rate, differentiating recruitment efforts and incentives paid. The results indicate that the effect of response burden is more negative than previously expected. Recruitment shifts the response curve upwards, while incentives flatten it. Surveys beyond 2,000 points appear overly burdensome, sustaining high response rates only through recruitment coupled with incentives. Without incentives, the level effect of recruitment quickly vanishes. Contrary to previous findings, we can not identify a negative time-trend. The data, functions and workflow underlying this analysis are organized as an `R`-package to foster a collective effort towards understanding response rates.