Are anchoring vignettes ratings sensitive to vignette age and sex?

Anchoring vignettes are commonly used to study and correct for differential item functioning and response bias in subjective survey questions. Self-assessed health status is a leading example. A crucial assumption of the vignette methodology is 'vignette equivalence': The health status of the person described in the vignette must be perceived by all respondents in the same way. We use data from a survey experiment conducted with a sample of almost 5000 older Americans to validate this assumption. We find weak evidence that respondents' vignette ratings may be sensitive to the sex and, for older respondents, also to the age (implied by the first name) of the person described in the vignette. Our findings suggest that vignette equivalence may not hold, at least if the potentially subtle connotations of vignette persons' names are not fully controlled.