Measuring agricultural knowledge and adoption

Understanding the tradeoffs in improving the precision of agricultural measures through survey design is crucial. Yet, standard indicators used to determine program effectiveness may be flawed, and at a differential rate for men and women. We use a household survey from Mozambique to estimate the measurement error from male and female self-reports of their adoption and knowledge of three practices: intercropping, mulching, and strip tillage. Despite clear differences in human and physical capital, there are no obvious differences in the knowledge, adoption, and error in self-reporting between men and women. Having received training unanimously lowers knowledge misreports and increases adoption misreports. Other determinants of reporting error differ by gender. Misreporting is positively associated with a greater number of plots for men. Recall decay on measures of knowledge appears prominent among men but not women. Findings from regression and cost-effectiveness analyses always favor the collection of objective measures of knowledge. Given the lowest rate of accuracy for adoption was around 80%, costlier objective adoption measures are recommended for a subsample in regions with heterogeneous farm sizes.