Posterior Probabilities of the Independence Axiom With Non-Experimental Data

Robert C. Marshall, Jean-Fancois Richard, Gary A. Zarkin
Duke University, Research Triangle Institute

May 31 1989

This paper addresses the issues associated with the construction of posterior probabilities for the independence axiom of expected utility from non-experimental data. To illustrate the methodology of analyzing non-experimental evidence we consider seat belt usage data. We find a posterior probability close to one of an Allais type paradox in this data set. In addition, the data is consistent with Machina's (1982) Hypothesis II but inconsistent with the "light" hypothesis of chew and Waller (1986).


PDF icon 1989-09.pdf