How to induce honesty: results from a large-scale experiment

Abstract Everyday public behavior is premised on honesty. But the imperative to be honest can conflict with self-interest. This conflict between public good and private gain is one of the largest sources of malignity in policymaking, often resulting in noncompliance, inaction, or outright sabotage. This empirical paper examines whether giving the option to be seen as honest helps to reduce malignity in public behavior. We conduct a large field experiment involving 34,543 adults in nine countries to study what drives people to behave more honestly. In a three-arm randomized control trial, subjects reported the outcome from a single roll of an unbiased, six-sided die. The higher the reported value, the more they earned. In two of the arms (“private cash” and “private lottery”), the roll was observed only by the subject. In the third arm (“public cash”), subjects could opt to have their roll observed by the experimenter. A truthful report will yield an average value of 3.5. In the two “private” arms, we found significant overreporting (from 3.76 in China to 4.43 in Mexico). In the “public cash” arm, 85% of subjects opted to be observed, with zero or minimum overreporting (3.46 in Germany, to 3.61 in Indonesia). We develop and estimate a model of honesty that can account for behavioral regularities across the nine countries well. Our study shows that most people want to be seen as honest and a simple policy of providing the additional option of being observed is sufficient to induce them to behave honestly.