Framing Allais: Is the paradox robust to the pictorial framing of probabilities?

The Allais paradox refers to a choice problem in which individuals face two pairs of lotteries, A-B and C-D, and typically prefer A to B and D to C, a choice pattern that violates Expected Utility theory. Prior theoretical and experimental work suggests that the AD pattern is sensitive to features of experimental design and that framing the probabilities of outcomes pictorially rather than numerically may reduce its incidence. We test this hypothesis in an online experiment (N=595) with three treatments. In Baseline, outcome probabilities are presented numerically as fractions. The two pictorial treatments are Pie, where probabilities are displayed using pie charts, and Grid, which represents probabilities through a grid of colored balls. Contrary to our expectations, we find that the frequency of the AD pattern is substantially the same across the three treatments, indicating that the Allais paradox is robust to the pictorial framing of lottery-outcome probabilities. We then discuss some implications of our findings.