PIH45 Satisfaction Assessment of Insurance System for Urban Residents (URMS) of University Students in Shenyang

Abstract

OBJECTIVES: Published 'Think Aloud' experiments on EQ-5D elicitation methods suggest that VAS and TTO may capture different constructs of preference. There has been some discussion concerning issues of so called implausible options and the possibility of non-trading in DCEs, the extent to which respondents may be 'aware' of their rationale for choices or valuations, and the use of heuristics in complex tasks. This qualitative study seeks to uncover how respondents think about EQ-5D elicitation tasks, and how they arrive at-and-interpret their responses to the extent that they are aware of these processes. METHODS: Forty respondents carried out a series of valuation exercises (ranking, VAS, TTO and DCE in randomized order) involving 6 EQ-5D states. In a variant of a 'Think Aloud' method, respondents were asked to explain their thought processes and decisions after each exercise. RESULTS: A 'Thurstone Effect' is suggested by Ranking-VAS preference reversals: 58% of respondents reported having 'changed their minds' about the order of preference among states when assigning VAS values to states that they had just ranked (novel task), and declared their VAS order as being 'correct'. Evidence of TTO non-trading thresholds, and inclusion of factors in TTO that are absent in VAS valuations were observed (68%). Differences between valuations are also observed because of the natural comparators in each elicitation method: 11,111 for TTO, one pair for DCE and all 6 states for ranking and VAS. Some evidence of the effect of one elicitation method on another was found. E.g. some respondents report that performing a TTO task in between a VAS and DCE influences the way they compare states in the DCE. Further insight into heuristics used by respondents was obtained using patterns of preference reversals and confidence ratings. CONCLUSIONS: These findings carry implications for the design of EQ-5D Valuation subsets.