Selection effects on dishonest behavior

Abstract

In many situations people behave ethically, while elsewhere dishonesty reigns. Studies of the determinants of unethical behavior often use random assignment of participants in various conditions to identify contextual or psychological factors influencing dishonesty. However, in many real-world contexts, people deliberately choose or avoid specific environments. In three experiments (total N = 2,124) enabling self-selection of participants in two similar tasks, one of which allowed for cheating, we found that participants who chose the task where they could lie for financial gain reported a higher number of correct predictions than those who were assigned it at random. Introduction of financial costs for entering the cheating-allowing task led to a decrease in interest in the task; however, it also led to more intense cheating. An intervention aimed to discourage participants from choosing the cheating-enabling environment based on social norm information did not have the expected effect; on the contrary, it backfired. In summary, the results suggest that people low in moral character are likely to eventually dominate cheating-enabling environments, where they then cheat extensively. Interventions trying to limit the preference of this environment may not have the expected effect as they could lead to the selection of the worst fraudsters.

Affidavit aversion: Public preferences for trust‐based policy instruments

“…Although some research shows that pledges can sometimes prevent unethical behavior (e.g., Beck et al, 2020; Jacquemet et al, 2019; Peer & Feldman, 2021), other evidence suggests that trust‐based instruments, and pledges specifically, may be perceived as an easy way to increase gains without substantial risks which might, in some contexts, signal people the possibility of behaving dishonestly (Tyran & Feld, 2006). Indeed, research shows that when participants could self‐select one of two tasks, those who chose the task where they could lie for financial gain, claimed higher gains than those who were assigned to the same task at random (Houdek et al, 2021). If indeed, people would choose trust‐based instruments to exploit the system, then that would defeat the purpose of these instruments.…”