Probability weighting and L-moments

Several popular generalizations of expected utility theory - cumulative prospect theory, rank-dependent utility and Yaari's dual model - allow for non-linear transformation of (de-)cumulative probabilities. This paper shows an unexpected connection between probability weighting and the statistical theory of L-moments. Specifically, cubic probability weighting results in a linear tradeoff between the expected value (the first L-moment), Gini (1912) mean difference statistic (the second L-moment, also known as L-scale) and the third L-moment (measuring skewness). Inverse S-shaped probability weighting function crossing the 45° line at a probability ≤0.5 reflects an aversion to the dispersion of outcomes and an attraction to positively skewed distributions