A Tailor-Made Test of Intransitive Choice

We performed a new test of transitivity based on individual measurements of the main intransitive choice models in decision under uncertainty. Our test is tailor-made and, therefore, more likely to detect violations of transitivity than previous tests. In spite of this, we observed only few intransitivities and we could not reject the hypothesis that these were due to random error. A possible explanation for the poor predictive performance of the intransitive choice models is that they only allow for interactions between acts, but exclude within-act interactions by retaining the assumption that preferences are separable overstates of nature. Prospect theory, which relaxes separability but retains transitivity, predicted choices significantly better than the nontransitive choice models. We conclude that descriptively realistic models need to allow for within-act interactions, but may retain transitivity.

Subject classifications: Utility/preference: Estimation. Decision analysis: Risk.

Area of review: Decision Analysis.