A COMPARISON OF MINIMAX BEHAVIOR WITH OBSERVED BEHAVIOR IN A TWO-PERSON ZERO -SUM GAME INTERACTION.
ARIZONA STATE UNIV TEMPE
Pagination or Media Count:
Three sets of two subjects male-male, female-female, male-female played three two-person zero-sum games involving two or three strategy choices per person in an attempt to compare their behavior with the minimax strategy solutions of the games. Games containing a saddle point thus requiring a pure strategy solution were solved by all subjects within 30 trials or less. The game demanding a mixed strategy solution was not always solved even after 100 trials. In fact, some subjects continued to play a dominated strategy, which theoretically should never have been played. A performance measure, X, was defined as the average over possible courses of action of the percentage point deviation of the observed frequency of each course of action. The performance measure was determined for each player on each game and the effect of the different games, different players, and different combinations of players upon X was determined by an analysis of variance. The following results were found 1 there was no difference in the effect of the games on X 2 there was a slight difference in the effect of the players on X and 3 there was a significant difference in the effect of the different combinations of players on X in that the male-male combination was found to more closely approximate the minimax strategy solutions. Result 1 was invalidated since the order in which the games were played was not randomized.
- Statistics and Probability