Selection into experiments: New evidence on the role of preferences, cognition, and recruitment protocols

Abstract

We study selection into lab experiments based on data from two cohorts of first-year university students. We combine two experiments: a classroom experiment in which we elicit measures of time and risk preferences, overconfidence, trust, reciprocity, altruism, and cognitive reflection and a recruitment experiment with four treatment conditions that randomly vary the information provided in the e-mail sent to recruit lab participants. We find that students with higher cognitive skills are more likely to participate in experiments. By contrast, we find little evidence of selection along time and risk preferences, overconfidence, trust, and reciprocity, and our evidence of selection along altruism is inconclusive. In terms of recruitment conditions, mentioning financial incentives boosts the participation rate in lab experiments by 50 percent. Although recruitment conditions affect participation rates, they do not alter the composition of the participant sample in terms of the elicited characteristics. Finally, students who repeatedly participate in lab experiments are more patient than those who participate only once.