Assessing the Validity of Respondents’ Reports of Their Partners’ Ages in a Rural South African Population-Based Cohort

Guy Harling, Harvard University
Tinofa Mutevedzi, Africa Centre for Health and Population Studies

Relationship age disparities are of interest for research and for public health intervention targeting, but evidence on how accurately respondents assess the age disparity in their relationships is scant. We tested partner age report accuracy in a population-based cohort in rural KwaZulu-Natal, South Africa by matching reports of sexual relationship age disparity to conjugal records, allowing comparison of ‘actual disparity’ (each partner’s report of their own age) to ‘perceived disparity’ (one partner’s report of the disparity). Despite considerable age differences, mean differences between actual and mean disparity were small, and correlated at ?=0.78 for women and 0.62 for men. These data suggest that cautious use of partner age reports for research and intervention targeting is justified in this setting. However, the observed heaping of perceived disparity reports at round values (e.g. 5, 10 years) cautions against relying strongly on dichotomous measures of age disparity which use such values as cut-points.