Do Self-Reported Strategies Match Actual Behavior in a Social Preference Experiment?

According to many models, social preferences are influenced by properties of payoff distributions like the payoff to the group member who is worst off, or higher payoffs to other members possibly causing envy. In this paper, we explore if subjects in a social preference experiment consciously take these elements into account. To study this question, we performed an experiment in which subjects stated indifference values of unequal payoff distributions, and explicitly reported on the strategies they used in stating these values. This approach allows us to obtain a cardinal measure of their social preferences, and to study the impact both of subject characteristics and stated strategies on these values. Our results indicate that although experimental factors, like group composition or field of study of subjects, are significantly related to self-reported strategies, they barely have a statistically significant impact on the indifference values given. On the other hand, we find highly significant relationships between self-reported strategies and actual behavior. Furthermore, our results indicate that the salience of different payoff distribution characteristics, and the impact of different motives on actual decisions, is context dependent.