Findings suggest that simple causal diagrams can structure data collection and interpretation in ways aligned with farmers’ goals.
This study explores how scientists can support on-farm experiments using analytical methods that align with farmers’ endogenous learning processes to inform their management decision. Four maize (Zea mays L.) farmers across 10 site-years in New York participated in this study to evaluate the effectiveness of a nitrogen-fixing inoculant (NFI) applied with a reduced side-dress nitrogen rate. Farmers designed and implemented their own experiments using a range of layouts, including side-by-side comparisons and strip trials.
Two analytical approaches were compared: a quantitative yield analysis using spatial regression, and a causal pathway analysis based on mechanistic steps informed by field sampling (e.g., quantitative polymerase chain reaction detection of NFI organisms, nitrogen nutrition index, and yield). While yield data suggested positive or neutral treatment effects at all sites when simply comparing yield average, the spatial regression analysis and causal pathway analysis identified positive outcomes in only seven or four of 10 site-years, respectively, reflecting a more conservative interpretation of efficacy.