We capitalise on an opportunity in the UK Household Longitudinal Study, which asks respondents the same SAH question with identical wording two times. This is done once with a self-completion and once with an open interview mode within the same household interview over four waves. We estimate multivariate models to explore which individual and household-level characteristics are systematically relevant for the likelihood and frequency of inconsistent reporting across the two modes. We find evidence of some inconsistency in reporting; 11%-24% of those who reported a particular SAH category in the self-completion mode reported inconsistently in the open interview. The probability of inconsistency is systematic and influenced by an individual's demographics, education, income, employment status, cognitive and non-cognitive skills. The same characteristics are also systematically associated with the frequency of inconsistent reporting across four UKHLS waves. Analysis of the implications of reporting inconsistencies shows no impact of SAH measurement on the magnitude of the association between income and health, estimated separately using the two SAH measures. A set of dimensions of people's physiological and biological health, captured using biomarkers, is associated equally with both SAH measures, suggesting that reporting inconsistencies in SAH may be driven by mechanisms other than people's underlying health.