Strategic Reasoning in Hide-and-Seek Games: A Note

Aggregate behavior in two-player hide-and-seek games deviates systematically from the mixed-strategy equilibrium prediction of assigning all actions equal probabilities (Rubinstein and Tversky, 1993, Rubinstein et al., 1996, Rubinstein, 1999). As Crawford and Iriberri (2007) point out, this deviation can be explained by strategic level-k reasoning. Here we provide empirical evidence that, indeed, it is non-equilibrium beliefs that lead to the behavior observed in the earlier studies: when a player's opponent is forced to play the equilibrium strategy, the player's choices are uniformly spread over the action space. At the same time, we find robust evidence of an unexpected framing effect.