Testing the intransitivity explanation of the Allais paradox

This paper uses a two-dimensional version of a standard common consequence experiment to test the intransitivity explanation of Allais-paradox-type vi olations of expected utility theory. We compare the common consequence effe ct of two choice problems differing only with respect to whether alternativ es are statistically correlated or independent. We framed the experiment so that intransitive preferences could explain violating behavior when altern atives are independent, but not when they are correlated. We found the same pattern of violation in the two cases. This is evidence against intransiti vity as an explanation of the Allais Paradox. The question whether violatio ns of expected utility are mainly due to intransitivity or to violation of independence is important since it is exactly on this issue the main new de cision theories differ.