Wednesday, August 5, 2015

Testing a Class of Models that Includes Majority Rule and Regret Theories: Transitivity, Recycling, and Restricted Branch Independence

BIRNBAUM Michael, DIECIDUE Enrico
Access the publisher's website
Decision 2, 3 (2015) 145-190

Six experiments compared 4 classes of decision-making models that make different predictions for tests of 3 diagnostic behavioral properties. These properties are transitivity of preference, recycling of intransitivity, and restricted branch independence. Two experiments tested decisions based on advice from friends, a situation in which majority rule was previously thought to apply. Both experiments rejected the hypothesis that more than a small percentage of participants might have used majority rule. Four experiments with choices between gambles tested a very general family of models that includes majority rule and regret theory as special cases. Such models not only allow but in certain cases require intransitive preferences and recycling, and they require that restricted branch independence must be satisfied. Two experiments used regret theory fitted to previous data to predict where to find violations of transitivity and recycling. No evidence was found to confirm these predictions; instead, a few participants showed the opposite pattern of intransitivity and recycling from that predicted. Two experiments with monetary incentives used a new experimental design in which a general model including regret theory, salience weighted contrasts, and perceived relative arguments must violate transitivity. Most participants did not show the predicted response patterns required by these models. Combining results of 6 studies, we conclude that the family of models that includes regret theory, perceived relative arguments, and majority rule is not an adequate descriptive model of how most people make decisions. Transitive, configural weight models best described most participants’ data.