Figure 3
From: Reconciling yield gains in agronomic trials with returns under African smallholder conditions
Source: authors calculation, using Stata version 16 (www.stata.com) and Microsoft Excel (https://www.microsoft.com/es-co/microsoft-365/microsoft-office?rtc=1).
Yield and yield increment of farmers who stayed in all trials and those who missed at least one season. Average yields over 3 seasons, based on observed yield values for planted seasons, and predictions of yield for the seasons missed, as described in “Methods”. As farmers’ participation decisions likely depend on performance of all trial plots, values are averaged over all treatment and control plots. Green bars indicate the 95% confidence interval for each estimation. The proportion of farmers who participated three seasons is 33% for maize and 41% for soybean. Differences between those who missed at least one season and those that stayed 3 seasons are significant in all cases: p-values of the difference are 0.000, 0.018, 0.000 and 0.017 for maize yield, maize yield increment, soybean yield and soybean yield increment, respectively (two-sided test).
