The present paper aims to contribute to the current discussion on how to make gender category operative in epidemiological studies. We argue that when working with multivariate analysis, the variables "sex" and "gender" represent distinct objectives and should thus be treated differently. We begin with a description of hierarchical approaches to analysis, using determinants for infant malnutrition as an example. Thereafter, through a simple example from data on obesity in adults, we posit the analytical difference in the application of sex and gender variables. The paper also points out the relevance of certain epidemiological concepts for this discussion, such as: sufficient, component, and necessary cause; confounding factors; modifying effect; and the classification of variables. We conclude by attempting to help in the complex epidemiological task of making the wide-ranging social aspects of "gender", which originated from a genetically-based differentiation.
Gender; Sex; Epidemiology studies; Analysis, multivariate
