ABSTRACT
Self-assessed health is often included in general socioeconomic surveys, such as the BHPS and the European Community Household Panel (ECHP). This kind of subjective measure of health has caused debate in the literature concerning its validity. It has been argued by some that perceived health does not correspond with actual health (see Bound 1991), while others have argued that it is a valid indicator of health (see Butler et al. 1987). As a self-reported subjective measure of health, SAH may be prone to measurement error. General evidence of non-random measurement error in self-reported health is reviewed in Currie and Madrian (1999) and Lindeboom (2006). Crossley and Kennedy (2002) report evidence of measurement error in a five-category SAH question. They exploit the fact that a random sub-sample of respondents to the 1995 Australian National Health Survey were asked the same version of the SAH question twice, before and after other morbidity questions. The first question was administered as part of the SF-36 questionnaire on a self-completion form, the second as part of a face-to-face interview on the main questionnaire. They found a statistically significant difference in the distribution of SAH between the two questions and evidence that these differences are related to age, income and occupation. This measurement error could be explained by a mode of administration effect, due to the use of self-completion and face-to-face interviews (Grootendorst et al. (1997) find evidence that self-completion questions reveal more morbidity); or a framing or learning effect by which SAH responses are influenced by the intervening morbidity questions.
