With respect to men’s contributions, censored regression models did not show a significant influence of the category of observer, either during the first period of the PGG or during the subsequent periods. In contrast, the models indicated that women contributed more in the presence of young female observers than in the presence of male observers, during both the first period and the subsequent periods.

Men’s contributions tended to increase across periods in the presence of young female observers, reaching a maximum at the last period, although the typical decreasing pattern was observed for the two types of controls. Similarly, women’s contributions increased globally across the game periods in the presence of male observers and not in the presence of female observers.

Generosity was measured as the probability of donating among (a) male and (b) female participants, and as the amount donated by (c) male and (d) female participants who made a non-zero donation (raw data). Logistic regression models showed that men’s propensity to make a donation in the presence of young female observers was significantly higher than in the presence of male observers, but not significantly different from their probability of making a donation in the presence of old female observers. For the female participants, logistic regression models showed that they were more likely to donate in the presence of female observers than in the presence of male observers. Among the participants who made a non-zero donation, linear mixed models showed that men were more generous in the presence of young female observers than in the presence of old female observers, but not significantly more than in the presence of male observers. No significant difference was found in the amounts donated by women.