The impact of commitments on longitudinal survey attrition

Abstract

Attrition is a pervasive problem in longitudinal research. While attrition often stems from commonly reported on demographic characteristics, attrition can also be influenced by study design features that invoke behavioural responses in participants. One study feature that could impact attrition is the use of commitments (asking respondents to commit to a behaviour). In this paper, for the first time, we explore the impacts of using commitments on attrition in a field experiment. We ask survey respondents whether they would like to pre-commit to attending a nature restoration volunteering event. Surprisingly, we find that the attrition rate is 16 percentage points worse for those who are pre-committed to an event. Using linear probability models, we explore this finding and theorise that ingenuine commitments provoke feelings of guilt and respondents actively avoid these feelings by dropping out. These results have important implications, especially if feelings of guilt drive longer-term disengagement with policy and research.