Pairs of trustors play finitely repeated Trust Games with the same trustee in a laboratory experiment. We study trustfulness of the trustor and trustworthiness of the trustee. We distinguish between learning and control effects on behavior. Learning effects are related to the trustor’s information on past behavior of the trustee. Control effects are related to the trustor’s opportunities for sanctioning a trustee in future interactions. The design of the experiment, with respect to trustfulness, allows for disentangling learning effects from a trustor’s own experience with the trustee and learning effects through third-party information. Also, the design enables disentangling control effects on trustworthiness and trustfulness through a trustor’s own sanction opportunities and opportunities for third-party sanctions.
Subjects play repeated Trust Games with 15 rounds. Subjects are matched in groups of three for these 15 rounds, one trustee and two trustors, which we call triads. In each of the 15 rounds, the trustee plays one Trust Game with each of the two trustors. Therefore, in every round, while each trustor plays one Trust Game, the trustee plays two Trust Games, adding to 30 Trust Games played per repeated game by the trustee. The trustee does not necessarily need to make a choice in all these 30 games: when the trustor does not place trust, the trustee has no choice to make. In every round, the trustee always plays with the same trustor first, while the other trustor has to wait, and always plays second with this trustee. In the “no information exchange between trustors” condition, trustors do not share any information. In this condition, subjects only know what happens in their own games with the trustee. In the “full information exchange between trustors” condition, trustors playing with the same trustee do share all information about each other’s games.
72 subjects participated in the experiment; 28 male and 44 female participants. Subjects were mostly undergraduate students from different fields, most of them students of social sciences. Four sessions were scheduled and 18 subjects participated in each session. Two sessions were played in the condition with no information exchange between trustors and two sessions in the condition with full information exchange. 2160 Trust Games were played in total. There are 485 games in which the trustor did not place trust, leaving 1675 games (78%of the total number of games played) in which the trustee’s behavior is observed.
For a publication based on these data: https://doi.org/10.1016/j.socnet.2010.05.001
