We study the incidence of different equilibria in a three-player "battle-of-the-sexes" type game. Payoff increases in coordination so that player prefer to coordinate on one action. Each player has a preferred coordination action. We study what equilibria are played when we increase the payoff to coordinate on ones preferred action.
Experimental design
In total, we have 3x2 treatments. Between-subjects, we vary (i) whether the marginal return from coordinating on one’s preferred action (PA) is Low (M=1.1) or High (M=3), and (ii) whether --provided M=3-- one player per group receives a choice-independent Lumpsum or not.
We also vary the labeling of the players and actions. In a first treatment we label them A, B and C. In a second treatment, they are labeled ●, ■ and ▲. The reason for this treatment is that the former contains a natural ordering which may cause a rotation strategy more salient to subjects. If we already have close to maximal rotation in the Low treatment then it will be hard to detect differences across Low and High.
Design details
Randomization method
Computerized randomization by experimental software
Randomization unit
Individual participants are randomly assigned to groups (of three individuals), and to the different treatments.
Clusters
150 groups in total
Observations
3 individuals per group, thus 150*3=450 individuals in total.
Arms
150 individuals (50 groups) per treatment
Primary outcomes
Subjects play a three-person Battle-of-the-Sexes game, repeated over 48 periods. We collect data on individual actions in every period.We will construct variables according to whether coordination occurred (i.e. an NE action profile was played). In addition, we will also construct variables that indicate whether a particular on-path equilibrium strategy was played. We focus on three types of equilibria:
(i) a dictator equilibrium, in which one player always prefers their preferred action (PA) x and the other two players also play action x.
(ii) a bully equilibrium, in which two players alternate synchronously between their respective PAs x and y, and the third player also alternates between x and y.
(ii) a rotation equilibrium, in which all three players rotate synchronously over their respective PAs x, y, and z.
In all these equilibria there is full coordination (i.e. all three players choose the same action in a given period), but equilibria differ in the distribution of PAs within a group of three players.
Secondary outcomes
Which type of information (choice history and payoff history) is acquired by the players
Cited by
No citation data available for this registration yet.