1) General setting:
Subjects play a battle-of-the-sexes game.
In each session, they are in groups of 12 subjects (6 “blue” subjects and 6 “red” subjects). They stay in the same group during the entire session.
At each round, one “blue” subject is randomly matched with one “red” subject (imperfect stranger matching). Subjects will never know who they are matched with, and play 20 rounds of the game (they are randomly rematched after each round).
Subjects are informed that their role is to simultaneously choose an action. Their payoff depends on their choice as well as the choice of the other subject they are matched with. The payoffs are presented in a matrix form.
The payoffs of the matrices are determined by an extension of Akerlof (1997)’s model in which agents have opposite private preferences.
Subjects are informed that their final payoff is the sum of a fixed participation fee (€5) and their gain in one round of the game randomly selected at the end of the experiment by the computer.
Per session, we will consider two groups of 12 subjects. There will be no interaction between these two groups.
We consider 4 treatments that differ regarding the number of actions each subject can simultaneously choose, and the presence, or not, of an unbiased equilibrium. We consider that an equilibrium is “unbiased” if it corresponds to the action that coincides with the average private preferences (see Michaeli and Spiro, 2017).
In all the following matrixes, each subject can choose either the action that maximizes his/her private preferences, or the one that maximizes the private preferences of the other subject he/she is matched with. Thus, all matrixes include two equilibria such that the private preference of one of the two players is maximized, as well as the alienated issue.
The four treatments are:
a) 2x2 matrix: where the unbiased equilibrium is not included.
b) 3x3 matrix: this treatment is similar to the previous one (2x2 matrix), except that the unbiased equilibrium is included.
c) 4x4 matrix: this treatment is an extension of the 2x2 matrix, with additional choices of actions, but without the unbiased equilibrium.
d) 5x5 matrix: this treatment is similar to the previous one (4x4 matrix), except that we add the unbiased equilibrium.
Akerlof, G. A. (1997). Social distance and social decisions. Econometrica, 1005-1027.
Michaeli, M., & Spiro, D. (2017). From peer pressure to biased norms. American Economic Journal: Microeconomics, 9(1), 152-216.