How do programs work to improve child nutrition?: Program impact pathways of three nongovernmental organization intervention projects in the Peruvian highlands

This paper examines the program logic of three nongovernmental, community-based programs with different intervention models to reduce childhood stunting. Two programs, Child Nutrition Program (PNI) and Good Start, focused directly on education and behavior change among caregivers, or the short routes to achieve impact, while one program, Sustainable Networks for Food Security (REDESA), focused on upstream factors, such as improving local governance and coordination, improving water and sanitation, and increasing family incomes, or the long routes to achieve impact. We compared the logic of each program as it was explicitly documented to the logic as perceived by the implementers. We elucidated the program impact pathways (PIPs) of key activities by actors at different operational levels in each program to identify congruencies and gaps in the perceptions of causal mechanisms between program activities and their intended outcomes, and analyzed them with the simple program models and logical frameworks to highlight the methodology and utility of PIPs. In a desire to move beyond static input-out models of the three programs, we designed and conducted data collection activities (document review, semi-structured interviews, and observations) with the intention of gaining insights about those aspects of the program that brought causal mechanisms of a given program into clearer focus. We propose that different methods for eliciting PIPs may be necessary at different operational levels. The interview method elicited more complete responses among those who are familiar with programmatic concepts, whereas actors at the local operational level provided sparse and fragmentary responses, even when simple, common language was used during the interviews. Group participatory processes, using visual aids, may be more effective for mapping the perceptions of those who are not accustomed to articulating information about programs. To reduce the length and frequency of interviews with program actors, initial PIPs could also be constructed from program documents, then discussed and revised iteratively with program actors. Although program logic models and the logical frameworks provide a succinct overview of the program (for communication, strategic planning, and management), we found that PIPs provide a better representation of the causal connections between program activities and results, particularly when both upstream and direct intervention activities were part of the same program. PIPs provide a visual tool for tracking how activities were perceived to work and make an impact, bringing into focus the different pathways of the activities and influences along the way. Beyond the logical sequence of program inputs, outputs, and outcomes, the conceptualization of impact pathways is a useful approach for understanding the causal connections required for impact and for identifying where attention and reinforcements may be required within program operation. The utility of this tool warrants its use not only during final evaluation but also during mid-program monitoring and relevant assessments. National- and regional-level program actors had good understanding of the overarching frameworks and principles of their respective programs as well as the program components and activities. They demonstrated a strong coherence to the program documents, provided similar cohesive responses, and were able to articulate the impact pathways. However, program actors at the national level identified fewer facilitators and barriers along the impact pathways than did the local actors, revealing that the practical dimensions of the impact pathways were not as evident to planners and managers farther from the communities. Although program actors at the local level were more apt to provide practical examples of influencing factors or incidents that occur during implementation, they had difficulty fully articulating their perceived PIPs and provided fragmented views of how the activities linked to their outcomes. Similar patterns were found across the three programs. This finding raises the question of desirability of a common understanding of the goals and pathways by which these outcomes are achieved or the acceptability of diversity of perspectives. It is still unclear whether program effectiveness may be improved through greater congruency in the PIPs. Future research should elucidate how congruency of PIPs among program actors across operational levels could be increased, and whether greater congruency would improve program implementation and effectiveness.