Failures of contingent reasoning in screening decisions

Abstract Screening decisions—choosing a costly action that reveals information about an unknown quality—can be crucial in various situations, such as hiring workers or designing policies. Choosing whether to screen can be difficult because it requires recognizing and trading off the benefits of information with the costs of getting it. We investigate whether people recognize when the benefits of screening dominate its costs and what causes them to choose otherwise. Our online experiment simulates a hiring scenario with an initial trial task. Participants can first reveal candidates’ quality at a small cost and then choose which candidate to hire. Most participants choose the dominated task that does not reveal the candidates’ quality, and this mistake persists even with experience and feedback. We show that this insufficient screening is driven by failures to anticipate inference and plan the full strategy.