Abstract
This paper used a blended approach that involves multiple techniques to, first, test a set of assumptions around a health behavior change communication intervention theory of change (ToC) and, second, surface some unidentified assumptions involving the local context. The intervention was integrated with women's self-help groups (SHGs) in Uttar Pradesh, India. The key assumption tested in this paper was the linkage between SHG membership, program exposure, and maternal, newborn, and child health practices. Learnings were substantiated through empirical investigations, including structural equation modeling and mediation analysis, as well as 'co-learning' workshops within the community. The workshops aimed to capture and interpret the heterogeneity of local contexts through deep dialogs with the community and program implementers at various levels. Statistical analyses indicated a significant association between the amount of women's program exposure and their health practices. SHG membership was shown to affect maternal health practices; however, it did not have a direct effect on neonatal or child health practices. The 'co-learning' workshops revealed crucial aspects, such as prevailing socio-cultural norms, which prevented pregnant or recently delivered women from participating in SHG meetings. This paper encourages evaluators to work with the community to interpret and co-construct meaning in unpacking the contextual forces that seldom appear in the program ToC.