All situations with two vignettes: this table gives calculations for the nonparametric estimator C for all possible situations with two vignette responses, v1 and v2, and a self-rated health response, ya.

a The table was adapted from “King G, Wand J. Comparing Incomparable Survey Responses: Evaluating and Selecting Anchoring Vignettes. Political Analysis. 2007;15(1):46–66”. As shown in Table 1, vignette 1 described a person with better health than that in vignette 2. Thus, the response to vignette 1 (v1) was expected to be smaller than that to vignette 2 (v2). The situation of v1 being equal to v2 was called tied. The situation of v1 being bigger than v2 was called inconsistently ordered vignette response. In this analysis, we treated the tied values and inconsistently ordered vignette responses as missing values. Moreover, y represented the response to self-rated health question; ‘/’ represented missing value.All situations with two vignettes: this table gives calculations for the nonparametric estimator C for all possible situations with two vignette responses, v1 and v2, and a self-rated health response, ya.