On the Role of Role Reversal

Remarks on Charness and Rabin

Abstract

The short paper is a comment on some recent work by Charness and Rabin (Quarterly Journal of Economics, Vol 117, 2002), who intend to gain insight into the structure of individual preferences by using 'simple' experimental games. In the course of their experiments, Charness and Rabin use an experimental technique called 'role reversal' in order to increase the number of their laboratory observations. They claim that the use of this technique does not influence the nature of their experimental results. We present results of a series of experiments conducted in the Magdeburg laboratory (MaxLab) in order to find out whether this claim of Charness and Rabin is correct. Our results give clear evidence that indeed Charness and Rabin are mistaken. Because of role reversal, participants seem to believe they are playing a game very different from the 'simple' games Charness and Rabin want them to play. Consequently, Charness and Rabin analyze a game different from the on their participants were apparently playing. This means that the analysis by Charness and Rabin, though important and interesting, has to be treated and understood with the greatest of caution and care.