Allais for all: Revisiting the paradox in a large representative sample

We administer the Allais paradox questions to both a representative sample of the Dutch population and to student subjects. Three treatments are implemented: one with the original high hypothetical payoffs, one with low hypothetical payoffs and a third with low real payoffs. Our key findings are: (i) violations in the non-lab sample are systematic and a large bulk of violations is likely to stem from non-familiarity with large payoffs, (ii) we can identify groups of the general population that have much higher than average violation rates; this concerns mainly the lowly educated and unemployed, and (iii) the relative treatment differences in the population at large are accurately predicted by the lab sample, but violation rates in all lab treatments are about 15 percentage points lower than in the corresponding non-lab treatments. This paper was originally entitled “Allais for all: revisiting the paradox”. S. Huck Department of Economics, University College London, Gower Street, London WC1E 6BT, UK e-mail: [email protected] S. Huck WZB, Berlin, Germany W. Müller (B) Department of Economics and VCEE, University of Vienna, Brünnerstrasse 72, 1210 Vienna, Austria e-mail: [email protected] W. Müller CentER, TILEC and Tilburg University, Tilburg, Netherlands 262 J Risk Uncertain (2012) 44:261–293