Modeling the effects of decision complexity on choice behavior under risk

This paper introduces a flexible behavioral error model for binary risky choices, enabling the evaluation of multiple hypotheses on the relationship between choice inconsistencies, commonly known as behavioral errors, and task complexity. I compare the new model's descriptive and predictive power to those widely used in structural estimation of risk preferences, namely Fechner errors, Contextual Utility, and Decision Field Theory. My analyses find that the model, when attached to the Rank Dependent Utility Theory, captures the heterogeneity of risk preferences more comprehensively than its counterparts. The error model also reveals that a U-shaped function best represents the effect of task complexity on error standard deviation, challenging the assumptions of homoscedasticity under Fecher and a monotonous relationship between error standard deviation and task complexity postulated by Contextual Utility and Decision Field Theory. Finally, the model outperforms the other alternative specifications in describing and predicting choices under risk.