Carefully consider the necessity and type of baseline survey data to be collected when evaluating adaptive pilots. The CBRLM evaluation conducted a baseline household survey and cattle assessment that were not ultimately used for the evaluation analysis. These wasted resources are an example of the challenges that resulted from insufficient integration and lack of a common understanding between the CBRLM implementer and independent evaluator. If the evaluator had better understood the adaptive nature of the intervention and uncertainty about where implementation would occur, we could have saved a lot of time and money on the baseline surveys and adapted the evaluation design much earlier. In reality, large baseline surveys may not always represent good investments. While they may boost precision and allow heterogeneity to be examined, they (i) can be expensive; (ii) sometimes get discarded as happened here; and (iii) are not necessary for an RCT. Given the uncertainty about where implementation would occur, an alternative to a full baseline survey might have been a grazing area-level baseline of a much longer list of grazing areas in anticipation of the implementation grazing areas changing over time. The funding that would have gone to a traditional baseline survey could instead have gone to repeated behavioral audits during implementation as multiple endline measures can offer more precision than baseline measures.
Lesson Learned