We study the ability of survey-based measures to predict conditional cooperation in an incentivized Prisoner’s Dilemma (PD) game. We assess whether (i) hypothetical game play, (ii) unincentivized social norms, (iii) survey measures of economic preferences, and (iv) personality traits predict conditionally cooperative behavior when monetary stakes are introduced. Our findings reveal that hypothetical PD responses are the strongest predictors of incentivized behavior, with limited evidence of hypothetical bias. Notably, patience is negatively correlated with conditional cooperation, contrary to expectations. Surprisingly, other economic and social preference measures, including reciprocity and normative evaluations, exhibit weak or no predictive power. These results contribute to the debate on the feasibility of survey-based proxies in behavioral research. They suggest that well-designed hypothetical games can reliably substitute incentivized experiments, but that deviations from accurately mirroring the task may weaken the predictive power of survey measures.
Fallucchi et al. (2026) studied this question.