Assessment of respondent acceptability of preference measures: Discriminatory power of graphic positioning scale versus traditional scaling measures

Abstract

Objective: The objective of this study was to compare discriminatory power of two different measures - graphic positioning scale (GPS) versus traditional scale (TS) - in assessing respondent acceptability of three preference measures: visual analog scale (VAS), standard gamble (SG), and willingness to pay (WTP). Methods: Two face-to-face interviews were conducted at least 1 week apart in a convenience sample of women aged 22 to 50 years with no history of breast cancer or cancer requiring chemotherapy. Study participation required completion of two surveys: one evaluating health preferences for an acute condition (chemotherapy-induced nausea and vomiting) and one evaluating a chronic condition (breast cancer). Data were collected from March 2000 to June 2000 at Ohio State University. Respondents were randomized to either GPS or TS surveys. Data analysis was a two-step process. First, a four-way multivariate repeated-measures analysis of variance (MANOVA) was conducted to assess respondent acceptability of three-preference metrics - VAS, SG, and WTP - in health-care decision making. Each of the four dependent variables, difficulty, clarity, reasonableness, and comfort in use in decision making, was measured on 9-point Likert scale. Second, a mixed design univariate analysis of variance (ANOVA) was performed for each dependent variable to optimize MANOVA analysis. Univariate ANOVAs, 2 × (2 × 3), were composed of three independent variables: assessment (GPS/TS), condition (acute and chronic), and preferences (VAS, SG, WTP). Results: Of 126 respondents, 119 were usable and complete. MANOVA results showed (P < .05) for two main effects, condition (F4,114 = 6.375) and preferences (F8,110 = 9.290), and two significant interactions, condition × assessment (F4,114 = 3.421) and condition × preferences (F8,110 = 2.087). Conclusion: GPS has higher discriminatory power than TS in assessing respondent attitudes toward health preference measures. Results showed that respondents had more difficulty and less comfort when making decisions for chronic than for acute conditions. Results also show that respondents regard WTP as a more reasonable decision-making tool when assessing acute interventions in preference to SG and vice versa for chronic conditions. Of VAS, SG, and WTP methods, VAS was perceived as being the easiest to understand. These results can be explained by direct versus indirect comparisons made with GPS and TS methods, respectively.