Use of anchoring vignettes to evaluate health reporting behavior amongst adults aged 50 years and above in Africa and Asia--testing assumptions

Abstract

Background: Comparing self-rating health responses across individuals and cultures is misleading due to different reporting behaviors. Anchoring vignettes is a technique that allows identifying and adjusting self-rating responses for reporting heterogeneity (RH).

Objective: This article aims to test two crucial assumptions of vignette equivalence (VE) and response consistency (RC) that are required to be met before vignettes can be used to adjust self-rating responses for RH.

Design: We used self-ratings, vignettes, and objective measures covering domains of mobility and cognition from the WHO study on global AGEing and adult health, administered to older adults aged 50 years and above from eight low- and middle-income countries in Africa and Asia. For VE, we specified a hierarchical ordered probit (HOPIT) model to test for equality of perceived vignette locations. For RC, we tested for equality of thresholds that are used to rate vignettes with thresholds derived from objective measures and used to rate their own health function.

Results: There was evidence of RH in self-rating responses for difficulty in mobility and cognition. Assumptions of VE and RC between countries were violated driven by age, sex, and education. However, within a country context, assumption of VE was met in some countries (mainly in Africa, except Tanzania) and violated in others (mainly in Asia, except India).

Conclusion: We conclude that violation of assumptions of RC and VE precluded the use of anchoring vignettes to adjust self-rated responses for RH across countries in Asia and Africa.

Keywords: anchoring vignettes; cognition; mobility; reporting heterogeneity; response consistency; self-rating; vignette equivalence.

Fig. 1

Predicted vignette locations (relative to vignette severity level 5) for mobility (N=9,375) and cognition domain (N=8,788) identified from HOPIT model 4. Reference category is Navrongo, Ghana: (a) mobility – difficulty in moving around; (b) mobility – difficulty in vigorous activity; (c) cognition – difficulty with memory; (d) cognition – difficulty with learning. Y-axis is standardized to SD units of vignette severity level 5 to allow comparison of perceived vignette locations.

Fig. 2

Predicted threshold locations for mobility (N=293) and cognition (N=373) identified from vignettes (HOPIT model 1) and from objective measures (HOPIT model 2): (a): mobility – difficulty in moving around; (b) mobility – difficulty in vigorous activity; (c) cognition – difficulty with memory; (d) cognition – difficulty with learning. Y-axis is standardized to SD units of vignette severity level 5 identified from HOPIT model 1 to allow comparison of perceived threshold locations.