Behavior observed in many experimental games is inconsistent with the joint
hypotheses of rational play and complete information of the game. I assume
that the experimental session is a Bayesian game of incomplete information
and that the observed actions are supported by an equilibrium of this game
. This structure permits estimation, within the theoretical model, of the d
istribution of player characteristics which determines the equilibrium outc
omes of the game. The model developed for the coordination games allows a t
ype that follows a myopic learning rule as well as strategic types with eit
her selfish or altruistic preferences. Journal of Economic Literature Class
ification Numbers: C72, C92. (C) 1999 Academic Press.