Self-reported health: reliability and consequences for health inequality measurement

Self-reported health (SRH) is one of the most frequently employed measures for assessing income-related health inequalities between counties. A previous study has shown that 28% of respondents changed their assessment of their health status when asked a SRH question on two occasions in the same survey (first as part of self-completed questionnaire and then in a personal interview). This study re-examines this issue using another survey where SRH was again asked twice of respondents, but this time the personal interview was first and self-completion second. We find the same variation in responses, but the predominant direction is away from the 'extreme' categories 'Excellent' and 'Poor' which is the opposite direction to the previous study. We therefore conclude that the most likely explanation is a mode of administration effect that makes people less likely to choose the extreme categories in a self-completion questionnaire, but not a personal interview. However, this effect has a relatively minor impact on measures of inequality. This is due to a large proportion of the movement (i.e. movement to the middle) not being related to income and hence does not systematically impact on the cumulative distribution of health across this measure of socio-economic status.