Using cluster analysis to classify farms for conventional/alternative systems research

A multidisciplinary project, Agriculture in Concert with the Environment, is being conducted at the University of Nebraska. The project's goals are to use a farming systems approach to estimate and compare economic, environmental, and sociological performance/characteristics of conventional versus alternative production systems. Although a systems approach is often advocated, operationalizing initial classification of farm systems can be problematic. This is especially true in a multidisciplinary project where incentives for team members' active involvement are influenced by the appropriateness of methods and results for disciplinary publication. The initial classification of farming system observational units has to be robust enough to withstand critical review of each discipline. This study describes problems of classifying observational units for multidisciplinary conventional versus alternative agriculture research and presents cluster analysis, along with other farming systems concepts, to solve the problem. Farming systems research recognizes that decisions are made for a "system" of reasons and economic, environmental, and sociological outcomes result from a "system" of factors. The approach meets many objectives of a multidisciplinary team studying conventional versus alternative production systems -- fields for the soil scientist, economic flows for the economist, and people for the sociologist. However, standards for operationalizing this type of research effort are not well developed. The article describes the specific classification goals that team members had, how well cluster analysis met those goals, and the solution to various problems encountered. The final grouping is based on 59 cropping variables cluster analyzed into five groups ranging from an irrigated monocrop corn system to one that is near organic.