Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
The Harless–Camerer (HC), Hey–Orme (HO) and random preference (RP) models of stochastic variation in choice under uncertainty are compared. Implications of these models, including some that are independent of the deterministic theory with which they are combined, are tested in an experiment in which participants respond to decision problems twice. The HC model generally performs poorly; the HO model predicts more violations of dominance than are observed; while the RP model fails to account for those few violations which do occur. Additional regularities are observed which are inconsistent with all three models when combined with expected utility theory.