Statistical Considerations for On-Farm Research.

The implementation of on-farm research requires thorough planning to adequately address the defined research objectives.  Experimental design, data collection, statistical analysis, and documentation of results are all important components that need to be considered in the planning phase.  A typical goal in applied on-farm research is to apply the results of the controlled studies to a larger landscape beyond the local conditions of an experimental site.  This talk will address the concepts of inference and replication within this context, and examples will be provided to showcase true replication, sub-sampling and pseudo-replication to assist the researcher in decision making for future experiments.  The design schemes and analysis options presented in the talk will be applicable to small demonstration plots, which are typically unreplicated plots, and large-scale whole-field applications which encompass spatial heterogeneity issues which are often disregarded during the analysis phase.