This paper considers the role of survey design and question phrasing in evaluating the subjective health assessment responses using the Survey of Health, Ageing and Retirement in Europe (SHARE) dataset. A unique feature of this dataset is that respondents were twice asked during the survey to evaluate their health on a five-point scale, using two different sets of descriptors to define the five points, with the ordering of which set was first given determined randomly. We find no evidence to refute the assertion that the order was determined by random assignment. Yet we document differences in the response distributions between the two questions, as well as differences in inference in comparing the two populations (those that were asked one question first versus those that were asked the other). We then consider determinants of the degree of concordance between the two questions, as well as the determinants of individuals that provide conflicting responses. There appears to be evidence to suggest that individuals' assessments of their health in response to the second question may be influenced by the battery of health questions that were asked following the first assessment. We find that information in self-assessed health responses is useful in examining health outcomes. Our results suggest that adjusting such responses to take into account framing and sequencing of questions may improve inference. In addition, we show that accounting for survey design may be important in models for predicting outcomes of interest, such as the probability of a major health event.
↧