Search

This paper quantitatively analyzes the spatial heterogeneity of district‐level correlates of open defecation in rural India. We employ standard non‐spatial regression, spatially explicit regressions and multi‐scale geographically weighted regression to compare the stability of measurable correlates of open defecation across these different methods as well as across analyzed spatial units. Attributes like ownership of household assets, drinking water inaccessibility and prevalent literacy rates were identified as the most stable district‐level correlates of open defecation. Our results also demonstrated the relevance of our hypotheses about (a) possible negative sanitation externalities stemming from the co‐concentration of Scheduled Caste communities and other communities in densely populated rural districts, and (b) possible positive sanitation externalities stemming from the co‐concentration of Muslim and non‐Muslim communities in densely populated districts. Overall, however, our analyses demonstrate notable spatial clustering and significant spatial non‐stationarity of examined variables. Therefore, in our opinion, research findings that ignore spatial heterogeneity of sanitation drivers provide incomplete information for policy development and implementation. [ABSTRACT FROM AUTHOR]