Strategic interactions and belief formation: an experiment

Abstract

Traditional models of belief formation in repeated games assume adaptive players who do not take strategic interactions into account. We find that these approaches are limited in the sense that people think more strategically and realize that, in contrast with the classical view, their own actions are likely to influence their opponents' behaviour.

Acknowledgement

Financial support from SMU (URC Grant), the CNRS and Louis Lévy‐Garboua (PPF project) is gratefully acknowledged. All remaining errors are our own.

Notes

¹The experiment was programmed using ‘Regate’ (Zeiliger, 2000).

²In games HL, HH, LL and LH we have 34, 32, 38 and 30 subjects, respectively. ψ _r is 2.33 in games HL and HH and 0.67 in games LL and LH. θ _r is 5 in games HL and LL and 15 in games HH and LH. ψ _c and θ _c are 0.41 and 40, respectively, in all games.

³See Nyarko and Schotter (2002) for a detailed description of this procedure.

⁴Note that , so our conclusions are valid across actions.

⁵The number of observations varies within games as some players never played X or never played Y. These players are thus excluded from this analysis.

⁶The full set of models are not estimated on row players and column players only because ψ _i and θ _i (respectively ψ _j and θ _j ) vary only across row players (respectively column players).