Does endogeneity matter? A comparison of empirical analyses with and without control for endogeneity

PURPOSE: Using data from the Cebu Longitudinal Health and Nutrition Survey, we perform an empirical investigation of the effects of endogeneity and unobserved heterogeneity in the analysis of health outcomes. METHODS: First, we lay a theoretical background for this analysis and develop a set of expectations regarding the effects of ignoring endogeneity. Then, by modeling the effect of infant-feeding patterns on time to resumption of menses, we perform parallel analyses with and without control for endogeneity. RESULTS: We show that in this analysis, as far as the effects of endogeneity are concerned, empirical results do accord with theoretical expectations. There are differences in parameter estimates between models, that lead to somewhat different interpretations. CONCLUSIONS: We discuss the importance and implications of these findings for epidemiological studies of health outcomes. We outline the steps involved in such an analysis and discuss the practical limitations of the method, as well as the possible gains. (C) 1997 by Elsevier Science Inc.