A V Law1, D S Pathak, M R McCord. 1. Division of Pharmacy Practice and Administration, College of Pharmacy, Ohio State University, Columbus 43210, USA. law.2@osu.edu
Abstract
PURPOSE: Utility values obtained with the standard gamble (SG) method using the probability equivalence approach (PE) have a reported bias due to the "certainty effect." This effect causes individuals to overvalue a positive outcome when it occurs under certainty. Researchers in the decision sciences have proposed an alternative, "lottery equivalence" (LE) approach, using paired gambles, to eliminate this bias. The major objective of the current study was to investigate the certainty effect in health status utility measures and to test our hypothesis that the certainty effect would act in a reverse direction for negatively valued outcomes. METHODS: Fifty-four subjects completed the study by assessing preferences for three health states by rating scale and then by SG using PE as well as LE approaches with assessment lotteries of 0.5 and 0.75. RESULTS: The results from 41 useable responses point towards possible existence of the certainty effect in health in the hypothesized direction: utility values obtained with the PE were significantly lower than with the LEs. There was no significant difference between the LE values indicating elimination of the bias. CONCLUSIONS: The results have important implications since the SG using PE is thought be the "gold standard" in health status utility measurements.
PURPOSE: Utility values obtained with the standard gamble (SG) method using the probability equivalence approach (PE) have a reported bias due to the "certainty effect." This effect causes individuals to overvalue a positive outcome when it occurs under certainty. Researchers in the decision sciences have proposed an alternative, "lottery equivalence" (LE) approach, using paired gambles, to eliminate this bias. The major objective of the current study was to investigate the certainty effect in health status utility measures and to test our hypothesis that the certainty effect would act in a reverse direction for negatively valued outcomes. METHODS: Fifty-four subjects completed the study by assessing preferences for three health states by rating scale and then by SG using PE as well as LE approaches with assessment lotteries of 0.5 and 0.75. RESULTS: The results from 41 useable responses point towards possible existence of the certainty effect in health in the hypothesized direction: utility values obtained with the PE were significantly lower than with the LEs. There was no significant difference between the LE values indicating elimination of the bias. CONCLUSIONS: The results have important implications since the SG using PE is thought be the "gold standard" in health status utility measurements.