Hierarchical Maximum Likelihood Parameter Estimation for Prospect Theory

What is it about?

An individual’s tolerance of risk can be quantified by using decision models with parameters. Those models need to be individualized such that they maximally fit a set of risky choices that the individual has made. A goal of this model fitting procedure is to identify parameters that correspond to stable underlying risk preferences, an important individual difference. Using hierarchical statistical methods, we show significant improvements in the reliability of individual risk preference parameter estimates over other common methods. This hierarchical procedure uses population-level information to break “ties” (or near ties) in the fit quality for sets of possible risk preference parameters. By breaking these statistical ties in a sensible way, researchers can avoid overfitting choice data and thus more resiliently measure individual differences in people’s risk preferences. This paper shows that using this estimation method one can extract more “signal” from a set of noisy choices, and thus yield a better measure of people’s innate risk preferences and also be useful for making better predictions of future behavior.

Featured Image

Why is it important?

This new estimation method provides an improved way to measure people's risk preferences and therefore make better predictions about future risky choice behavior.