76

This article explores the way in which the instrument for measuring health impacts the scope of social health inequalities by using structural equations with latent variables. A synthetic indicator of latent health is created from perceived health indicators, limitations of activity, chronic illnesses and mental health. The effect of these socio-demographic variables on latent health is then separated from its direct effect on each health indicator. This method highlights social differences in declaration by presuming that the contribution of socio-demographic characteristics in the explanation of latent health reveals determinants in state of health, while with the latent health factor, the direct effects of these characteristics on health indicators reflect a heterogeneity in declaration. Thus, we show a social heterogeneity in declaration with latent health factors: women and the elderly, the highly educated and the upper classes declare chronic illnesses more often; mental health problems appear over-declared by women and people living alone and under-declared by the elderly; the unemployed, retirees and executives declare activity limitation more often. Finally, people with a lower level of education under-declare perceived ill health. Although the four indicators in this study can be faulted with heterogeneity in declaration, the chronic illness indicator is the one that most reduces the scope of social health inequalities.