In this paper, I present a comprehensive analysis of two decision heuristics that permit intransitive preferences:the lexicographic semiorder model and the similarity model. I also compare these two intransitive decision heuristics with transitive linear order models and two simple transitive heuristics. For each decision theory, I use two types of probabilistic specifications: distance-based models (which assume deterministic preferences and probabilistic response processes), and mixture models (which assume probabilistic preferences and deterministic response processes). I test 26 such probabilistic models on datasets from three different experiments using both frequentist and Bayesian order-constrained statistical methods. The frequentist goodness-of-fit tests show that the distance-based models with modal choice and the mixture models for all of the decision heuristics explain the participants' data fairly well for all stimulus sets. The frequentist analysis generates little evidence against transitivity. Model selection using Bayes factors suggests extensive heterogeneity across participants and stimulus sets.
【 预 览 】
附件列表
Files
Size
Format
View
Rationality or irrationality of preferences? A quantitative test of intransitive decision heuristics