Subjects responded to risk and caution life-dilemma problems in either a group discussion or a control procedure. Before and after experimental treatment, subjects filled in a questionnaire involving whether or not the character in the dilemma should take the risk, the lowest probability
they would accept before recommending to the character that he take the risk, and utilities and subjective probabilities of the various outcomes within the dilemmas. It was found that, while the discussion condition yielded shifts on the “lowest probability” question, only very
meager shifts were found on the dichotomous “yes-or-no” question, suggesting that it may be difficult to generalize the results of group shift research to real-life groups who must make decisions on a yes-or-no basis. In addition, it was found that the behavioral decision theory
concept of subjective expected utility was very accurate in predicting the subjects' dichotomous choices on the dilemmas.