OBJECTIVES: To examine whether random utility maximization (RUM) or Random Regret Minimization (RRM) better describes how people make decisions when facing with different levels of risks and survivals.

METHODS: A discrete choice experiment (DCE) was designed with 6 attributes including 5 health domains of the EQ-5D-5L (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) plus the out-of-pocket costs expressed as % of annual household income (5%, 10%, 20% and 50%). Responses were collected using SurveyMonkey in December 2022 (n=150). Modified Federov algorithm was adopted to select the experimental design using Ngene. Ten choice sets were designed, each with 3 options. One dominant test choice set was included for quality screening. Both RUM and RRM models were used for the estimation.

RESULTS: Ninety respondents (60%) passed the screening and were included in the analysis (54 failed to pass the dominant test, 6 gave the same answers to every scenario). Mean age was 40.1 years (SD ±16.0); 42% male; 71.1% white, 11.1% black; 90.0% had college degree; 81.2% had health insurance coverages. All estimated coefficients were significant in both models with pain/discomfort being the most impactful factor in respondents’ decision-making, followed by mobility and selfcare (p<0.01). RRM and RUM showed similar decision rules and, the log-likelihood estimations were also comparable across models suggesting similar data fittings.

CONCLUSIONS: In this experiment, respondents did not yield strong semi-compensatory behavior, indicating traditional RUM better describes choice behavior. Further investigation on choice behaviors under different levels of risks and survivals are warranted.