This evaluation will be a randomized controlled trial using randomization at the household level. This strategy is feasible in this program setting because resource constraints imply that only a subset of eligible households in the project sites can be served by World Vision programming. Randomly identifying individuals who will be included in the program is a fair and ethical strategy to allocate program benefits. At the same time, the outcomes of those individuals who are randomized to the control arm (who do not receive World Vision services) will be carefully tracked in order to generate evidence around the effectiveness and cost-effectiveness of graduation model programming.
The sampling frame for the evaluation was a sample of approximately 6,050 households identified as eligible for programming in a separate beneficiary census conducted by World Vision. Households will then be randomized to one of two arms. In the first arm, 5,000 households will receive a full set of graduation model services. In the second, control arm, 1,500 households will not be offered any targeted services from World Vision. The evaluation sample targeted in baseline, midline and endline surveys will include a random subsample from households assigned to treatment (3,000 treatment households) and the full sample of households assigned to control. Within the evaluation sample of 3,000 treatment households, 50% will be assigned to receive individual-level coaching, and 50% will be assigned to receive group-level coaching.
Key indicators collected at baseline, midline and endline will include variables summarizing household program participation, economic welfare, food security, and investment in human capital.