Adding interactions in order to model intersectionality: an empirical study on self-perceived health status in Argentina

Abstract

In recent decades, intersectionality has been at the center of both feminist debates and the theory of gender. In the United States, Canada and Europe, it has achieved a hegemonic status, strengthened by its multiple possible applications, precisely because it does not meet the necessary requirements to become a theory or conception with defined contours. Intersectionality was mainly incorporated in qualitative studies, favoring methodologies that were deemed to be best suited to address the complexity which lies within (e.g., ethnography, deconstruction, genealogy, ethnomethodology and case studies). In the field of population health research in particular, new approaches to model intersectionality in quantitative studies are still emerging. One way of making progress in multivariate analysis has been to calculate logistic regressions models separately for men and for women. Other authors work with additive models from multiple linear regressions, where different “levels of intersectionality” are included in different steps of the regression. Another possible approach, when applying multiplicative models, is the inclusion of interaction terms in conventional regression models. This chapter aims at contributing to these theoretical-methodological discussions about intersectionality throughout an empirical analysis of health inequalities in Argentina. More specifically, we use one of the many multiplicative statistical models to analyze the self-perceived health status of the population aged 18 and older, living in urban areas of Argentina. We address the effect of different sociodemographic and geographical variables on self-perceived health status, and then we add interactions within the regression: between gender and educational level, between gender and the income quintile and between gender and the age group. We work with the data of the National Survey of Risk Factors (ENFR for Encuesta Nacional de Factores de Riesgo in Spanish, 2013), provided jointly by the National Institute of Statistics and Censuses (INDEC) and the National Ministry of Health in Argentina (MSAL). This survey was carried out based on a probabilistic design (by conglomerates and stratified), throughout four stages (department, area, housing and household member). The final database is made up of 32,365 cases nationwide.