Nicolas Jacquemet

Dishonest behaviours such as tax evasion impose significant societal costs. Ex ante honesty oaths—commitments to honesty before action—have been proposed as interventions to counteract dishonest behaviour, but the heterogeneity in findings across operationalizations calls their effectiveness into question. We tested 21 honesty oaths (including a baseline oath)—proposed, evaluated and selected by 44 expert researchers—and a no-oath condition in a megastudy involving 21,506 UK and US participants from Prolific.com who played an incentivized tax evasion game online. Of the 21 interventions, 10 significantly improved tax compliance by 4.5 to 8.5 percentage points, with the most successful nearly halving tax evasion. Limited evidence for moderators was found. Experts and laypeople failed to predict the most effective interventions, though experts’ predictions were more accurate. In conclusion, honesty oaths were effective in curbing dishonesty, but their effectiveness varied depending on content. These findings can help design impactful interventions to curb dishonesty.

Whether individuals feel confident about their own actions, choices, or statements being correct, and how these confidence levels differ between individuals are two key primitives for countless behavioral theories and phenomena. In cognitive tasks, individual confidence is typically measured as the average of reports about choice accuracy, but how reliable is the resulting characterization of within-and between-individual confidence remains surprisingly undocumented. Here, we perform a large-scale resampling exercise in the Confidence Database to investigate the reliability of individual confidence estimates, and of comparisons across individuals’ confidence levels. Our results show that confidence estimates are more stable than their choice-accuracy counterpart, reaching a reliability plateau after roughly 50 trials, regardless of a number of task design characteristics. While constituting a reliability upper-bound for task-based confidence measures, and thereby leaving open the question of the reliability of the construct itself, these results characterize the robustness of past and future task designs.

We propose a two-stage experiment in which people receive feedback about their relative intelligence. This feedback is a noisy message reminded at every stage, so that subjects cannot forget this ego-relevant information. Instead, we exogenously vary whether the informativeness of the message is reminded in the second stage. We investigate how this treatment variation affects the informativeness reported by subjects, and their posterior beliefs about their intelligence. We show that subjects report informativeness in a self-serving way: subjects with negative messages report that these messages are significantly less informative in the absence of reminder than with it. We also show that the lack of reminder about message informativeness allows subjects to keep a better image of themselves. These results are confirmed by complementary treatments in which we decrease messages informativeness: subjects tend to inflate the informativeness of positive messages that should now be interpreted as bad news.