Analyzing Ranking and Rating Data from Participatory On- Farm Trials

Participatory methods have been widely adopted by researchers working on applied agricultural problems including crop breeding. This change in paradigm has implications for the design and analysis of methods used. Some of these are summarized in a companion paper by this author (see Coe 1, this proceedings). An assumption of this paper is that formal analysis of quantitative data collected systematically from trials is still an important part of the process. Without this it is difficult to see how the research activity can generate information relevant to anyone other than the small number of farmers directly involved. Breeders that have adopted participatory methods have generally recognized this, but have faced three difficulties: 1. The experimental designs used are often irregular in layout due to farmer participation (e.g., in choosing which varieties to test on their farms) or constraints arising from trials being located in farmers fields. 2. The focus of analysis often shifts from overall varietal selection to understanding the variation in varietal response across farms. This is genotype x environment interaction (GEI), where the E may include social or economic variables in addition to biophysical environments. 3. Much of the quantitative data collected may be ratings and rankings, for which the more usual methods of analysis may not be appropriate.