ADDRESSING ISSUES OF SCALE, RIGOR AND CONTEXT IN THE DESIGN OF EVALUATIONS OF LARGE-SCALE BEHAVIOR CHANGE INTERVENTIONS IN BANGLADESH, VIETNAM AND ETHIOPIA

Multisectoral programs are being launched globally to address the multiple determinants of poor nutrition simultaneously. As these program are designed and rolled out, it is apparent that interventions to support behavior change are a necessary major component of most multisectoral programs. However, the evidence base on how to roll out and achieve impact through large-scale nutrition behavior change programs is limited, partly because of limited program efforts that are truly at scale and partly due to the challenges of rigorously evaluating complex large-scale programs. This presentation and its accompanying presentation (Sanghi et al, on Alive & Thrive) together address issues related to design and implementation of large-scale behavior change interventions and the evaluation of these large-scale interventions. The evaluation of Alive & Thrive (A&T), a multi-year initiative that aimed to improve IYCF practices in Bangladesh, Vietnam and Ethiopia using multiple delivery platforms, addresses several evaluation challenges including: the technical aspects related to evaluating the nutritional impact of interventions targeted during the first two years of life when rapid growth faltering occurs; the identification of proper counterfactuals in the context of rapidly scaling-up programs; and the development of methodologies and tools to assess implementation, utilization and program impact pathways within evolving program portfolios. We designed the impact and process evaluations through a technically robust and programmatically engaged, step-wise process in three countries. In Bangladesh and Vietnam, cluster-randomized probability evaluation designs were used; in Ethiopia, the evaluation used an adequacy design. In addition, different methodological approaches were utilized to answer critical "how" questions and to generate information on the many processes and pathways to program impact. The methodological approach that guided all process evaluation activities included four components: 1) developing detailed program impact pathway models, 2) linking data collection to program impact pathways (PIPs) utilizing mixed methods, and multiple data sources, 3) linking evaluation timelines with program implementation timelines, and 4) engagement with the program implementation and management teams. Using these approaches, in all three countries, theory-driven impact and process evaluations based on program impact pathways, and using mixed methods, generated informati