Repeated Response versus Strategy Method : Experimental Evidence from an Oversight Game

Abstract

Existing laboratory research suggests that the Nash equilibrium in mixed strategies has limited capacity to predict subjects’ behavior in an Oversight Game (OG). We propose that this inconsistency between the game-theoretical solution and subjects’ actual behavior may be due to the elicitation method used in previous laboratory experiments. To test this conjecture, we design a laboratory experiment in which subjects play the OG either using the conventional repeated response method, in which subjects are informed of their opponent’s action after each period, or using a novel strategy method, in which subjects choose their mixed strategy and let the computer randomize. Comparing the elicitation methods shows that it has no effect on subjects’ decisions in the OG, regardless of whether the repeated response or strategy method has been implemented. Under both conditions, subjects uniformly deviate from the Nash equilibrium in mixed strategy.