Evidence on the equivalence of the strategic and extensive form representation of games

This paper reports an experiment testing whether strategically equivalent representations of a social situation produce equivalent behavior when actually played by human subjects. The investigation was limited to representative members of the class of generic 2×2 extensive form games of perfect information, which include widely studied games in the experimental literature, and the resulting class of 2×2 strategic form games. We find a systematic difference between subjects’ choices in the strategic and extensive forms of these games. In particular, subjects in the extensive form are consistently more likely to choose a branch that allows the other player to make a meaningful choice. The observed behavioral difference between game forms cannot be attributed to differences in subjects’ ability to do backwards induction, differences in expected payoffs between the two game forms, or differences in interpersonal preferences. We hypothesize that the extensive form elicits inclusive behavior.