Effects of Grouping Information on Decision Making Under Risk*

Abstract The ability of decision makers to deal with information in terms of category labels rather than as precise data points is hypothesized as an explanation of how complex choices are made within the limits imposed by human information‐processing capacity. Twenty‐five decision makers placed bets under varying conditions of grouping of cues (probability of winning/losing, amount to be won, amount to be lost) as a test of this hypothesis. The results indicate that experimental pre‐grouping of cues has: (1) a statistically significant but practically unimportant impact on amounts bet; (2) no statistically significant effect on number of different bets made; and (3) no statistically significant effect on the fit of the bets to those predicted by an expected value model, except when grouping categories are very wide. These results support the contention that decision making occurs through the manipulation of category labels rather than exact values. Study of processes by which numerous exact cues are reduced to a smaller number of category labels is suggested as a complement to the study of sequential processing of alternatives, satisficing, the use of heuristics, and other means by which human beings make complex choices with limited cognitive capacity.