A Study of the Relationship between Education, Literacy, and HealthRenato Carletti
A method of data mining, Reconstructability Analysis (RA) developed by Martin Zwick at the Portland State University Systems Science Program, has been used with the available data with the purpose of 1) identifying complex models of health prediction that could capture non-linearities and interaction effects among predictors and in their relation with health status, 2) isolating interaction effects that could be introduced in the multiple linear regression equation with the purpose of improving its performance. RA identifies a model that performs better than the linear regression model, in terms of average error produced. In this model, health status is predicted by occupational status, intensity of performing strenuous activities, and income level. The interpretation of this model, which is not trivial because no linear relations are implied, has been done using State Based RA. Literacy appears in several models identified by RA, and the best of these models has been analyzed in depth. In this model, health status is predicted by occupational status, intensity of performing strenuous activities, and literacy level. Overall, this study offers no support for a human capital theory: resources, as measured by the literacy level, do not associate linearly with most of the factors that explain differences in health, as well as they do not associate linearly with health status itself, once demographic characteristics of the population are controlled for. Findings suggest that the relation between education and health might not be explained mainly by the skills one learns through the educational process, as human capital theory claims. Education could affect health because of the credentials it provides.
Monday, May 10, 2004
|
