Anchoring Vignettes in the Health and Retirement Study: How Do Medical Professionals and Disability Recipients Characterize the Severity of Work Limitations?

PURPOSE Recent studies report systematic differences in how individuals categorize the severity of identical health and work limitation vignettes. We investigate how health professionals and disability recipients characterize the severity of work limitations and whether their reporting patterns are robust to demographic, education, and health characteristics. We use the results to illustrate the potential impact of reporting heterogeneity on the distribution of work disability estimated from self-reported categorical health and disability data. METHOD Nationally representative data on anchoring disability vignettes from the 2004 Health and Retirement Study (HRS) are used to investigate how respondents with an occupation background in health and Social Security disability beneficiaries categorize work limitation vignettes. Using pain, cardiovascular health, and depression vignettes, we estimate generalized ordered probit models (N = 2,660 individuals or 39,681 person-vignette observations) that allow the severity thresholds to vary by respondent characteristics. RESULTS We find that health professionals (excluding nurses) and disability recipients tend to classify identical work limitations as more severe compared to non-health professional non-disabled respondents. For disability recipients, the differences are most pronounced and particularly visible in the tails of the work limitations distribution. For health professionals, we observe smaller differences, affecting primarily the classification of mildly and moderately severe work limitations. The patterns for health professionals (excluding nurses) are robust to demographics, education, and health conditions. The greater likelihood of viewing the vignette person as more severely work limited observed among disability recipients is mostly explained by the fact that these respondents also tend to be in poorer health which itself predicts a more inclusive scale. CONCLUSIONS Knowledge of reporting scales from health professionals and disabled individuals can benefit researchers in a broad range of applications in health and disability research. They may be useful as reference scales to evaluate disability survey data. Such knowledge may be beneficial when studying disability programs. Given the increasing availability of anchoring vignette data in surveys, this is a promising area for future evaluation research.