Mothers’ health knowledge gap for children with diarrhea: A decomposition analysis across caste and religion in India

The access to health networks is an integral part of sustainable development, which has largely been ignored in previous studies of health knowledge production. Additionally, the previous literature is scarce on health knowledge gaps and the intersection of deeply institutionalized marginalization of certain groups—such as by caste or by religious system in India, Bangladesh, or Nepal—and the resources these groups have available. To address these knowledge gaps, we explore the relationship among health knowledge and caste and religion and a number of important mediating factors in India, estimating causal impacts through a combination of instrumental variables and decomposition methods. Five main results are established: (1) the presence of a substantively large “raw” (unconditional) health knowledge caste gap favoring high caste women—though at the same time with an overall relatively low level of health knowledge across castes and religions—thus pointing towards even deeper, more structural, endemic public policy challenges for Indian policy makers); (2) evidence that the endowments and the returns to these endowments increase the health knowledge gaps—indicating that high caste women have higher education and better access to health networks but also higher returns to these characteristics; (3) for Adivasi women network homophily works to decrease the discrimination part of the health knowledge gap—it may therefore not be enough if these women merely get access to health networks (even if they are of high quality) if caste and religion-related gaps in health knowledge are to be reduced; such networks also have to be homophilous, to have an effect; (4) while observed individual characteristics explain a large—indeed, sometimes the major—part of the gaps, in several cases a substantial part of the health knowledge gap is left unexplained—consistent with the presence of discrimination against these systemically marginalized women; and (5) in turn, the substantial dampening of the caste and religion effect once socioeconomic status is controlled for suggests that caste differentials are not independent of class differentials. We also perform similar analysis for child mortality, now including health knowledge as one of the focal explanatory variables and obtain similar results—thus providing additional evidence that health knowledge and health network access, two major factors of sustainable development, should receive more attention by policymakers in the future. Lastly, policy implications and implications and suggestions for future data collection efforts are also discussed.