Working Papers 2004 – Abstracts
18/2004
View paper
Further evidence on game theory, simulated interaction, and unaided judgement for forecasting decisions in conflicts.
Kesten C. Green
If people in conflicts can more accurately forecast how others will respond, that should help them to make better decisions. Contrary to expert expectations, earlier research found game theorists' forecasts were less accurate than forecasts from simulated interactions using student role players. To assess whether the game theorists had been disadvantaged by the selection of conflicts, I obtained forecasts for three new conflicts (an escalating international confrontation, a takeover battle in the telecommunications industry, and a personal grievance dispute) of types preferred by game theory experts. As before, students were used as role-players, and others provided forecasts using their judgement. When averaged across eight conflicts including five from earlier research, 102 forecasts by 23 game theorists were no more accurate (31% correct predictions) than 357 forecasts by students who used their unaided judgement (32%). Sixty-two percent of 105 simulated-interaction forecasts were accurate, providing an average error reduction of 47% over game-theorist forecasts. Forecasts can sometimes have value without being strictly accurate. Assessing the forecasts using the alternative criterion of usefulness led to the same conclusions about the relative merits of the methods.
Keywords: accuracy, conflict, forecasting, game theory, judgement, methods, role playing, simulated interaction.
