10.25394/PGS.12735752.v1 Laura E. Leavens Laura E. Leavens After the Project is Over: Measuring Longer-Term Impacts of a Food Safety Intervention in Senegal Purdue University Graduate School 2020 food safety aflatoxins Randomized Controlled Trial Longer-term follow-up studies Agricultural technology adoption post-harvest practices Senegal hermetic storage PICS Agricultural Economics 2020-07-30 00:08:46 Thesis https://hammer.purdue.edu/articles/thesis/After_the_Project_is_Over_Measuring_Longer-Term_Impacts_of_a_Food_Safety_Intervention_in_Senegal/12735752 <p>We followed up with about 2,000 smallholder households in Senegal, two years after these households participated in a randomized controlled trial (RCT) aimed at reducing levels of aflatoxins in smallholders’ stored maize. In the initial intervention, treated households were provided with training on proper post-harvest practices, low-cost moisture meters for testing if maize was sufficiently dry to store, plastic tarps for drying maize of the ground, and hermetic (airtight) storage bags to mitigate aflatoxin development in stored maize. Using cross-sectional follow up data on aflatoxins levels and drying and storage practices from 2019 along with baseline demographic data from 2016, we estimate both the longer-term intention-to-treat (ITT) effects and the treatment on the treated (TOT) effects that the four inputs provided on households’ aflatoxins levels in stored maize. The ITT analyses estimate the intervention’s average effect by treatment group, but this may underestimate the true impact for households who complied with recommended post-harvest practices and adopted the recommended technologies. The TOT analyses estimate the local average treatment effects (LATE) of the intervention, that is its impacts on those who were driven by the intervention to follow best practices or use a given technology. Since the decision to follow these practices or adopt a technology was not random, we instrumented the usage decision with the exogenous, random treatment group assignment to get an unbiased estimate. Outside of our main models, we conducted a heterogeneity analysis to test if households with different characteristics benefit differently from the intervention. We interacted each treatment assignment with various household characteristics, including the woman’s level of involvement in the intervention. Additionally, we estimate the cost-effectiveness of providing training and a tarp, according to WHO guidelines for public health interventions. </p>