Observation Moderates the Moral Licensing Effect: A Meta-Analytic Test of Interpersonal and Intrapsychic Mechanisms

When someone who initially behaved morally subsequently acts less morally, this is known as moral licensing. We apply reputation-based theories to predict when and why moral licensing occurs. Specifically, our pre-registered predictions were: (1) participants observed during the licensing manipulation would have larger licensing effects, and (2) unambiguous dependent variables would have smaller licensing effects. In a multi-level meta-analysis of 115 experiments (N = 21,770), there was a larger licensing effect when participants were observed (Hedge’s g = 0.65) compared to unobserved (Hedge’s g = 0.13). After accounting for publication bias using robust Bayesian meta-analysis, support remained when participants were observed (Hedge’s g = 0.51; BF10 = 9.14) but not when unobserved (Hedge’s g = -0.01; BF10 = 0.11). Ambiguity did not moderate the moral licensing effect. These results suggest that moral licensing is predominantly an interpersonal effect based on reputation, rather than an intrapsychic effect based on self-image.