Do children cheat to be honored? A natural experiment on dishonesty in a math competition

We use data from a children mathematics contest in Israel that involved a first unmonitored online stage at home and a second monitored stage in class, both with the same difficulty level. The performance deterioration from the first to the second stage allows to estimate the dishonesty in the unmonitored first stage (mostly in the form of being helped by the parents or older siblings). We also analyze dishonesty using a set of 3–4 problems that appeared in both tests. Contrary to much of the literature on gender effects in dishonesty, in our data girls were not more honest than boys. The sample consists of children in second to sixth grades, and we find that older children are significantly more honest. A stronger socio-economic level of the city was associated with more cheating. Children from religious schools tended to be more honest than children from secular schools. We also discuss other potential reasons for differences between performance in the two stages, such as pressure and stress, but conclude that they are secondary to the effects of dishonesty.