Conditions de collecte et santé subjective : analyse sur données européennes

It is fairly banal to say that subjective health, the most widely-used health variable, is measured with error. In particular, it would seem important to know how subjective health information depends on the way in which it is collected, as this latter varies widely between countries and surveys. In this article we use a quasi-experimental feature in the SHARE questionnaire to evaluate two framing effects: the place of the health question in the questionnaire, and the verbal labels associated with the different possible answers (on a one to five scale). The place effect is particularly homogeneous: in all of the countries, individuals declare better health after having replied to a detailed health questionnaire. We also find that functional health variables mobility and social and sensory limits to daily life play a more important role in determining health when the health question appeared after the detailed health questionnaire. The effect of different verbal labels is strong and less homogeneous than the place effect: it varies widely, in particular, between countries.