Exposition on the Selection of Appropriate Experimental Design and Statistical Analysis for Pasture Improvement Research

Abstract

Selection of appropriate treatment and experiment designs are essential elements in research. However, the expense and variability associated with pasture renovation studies creates unique problems in the application of standard statistical techniques. Pasture-size renovation studies are restricted by expense, requiring the use of grazing exclosures (subsamples). Treatment design must include an adequate control for treatment comparison. Controls for pasture renovation practices cannot be limited to untreated areas within a grazing exclosure. The true measure response is found in the difference between treated areas and a typical grazed pasture situation. Criteria for exclosure selection (homogeneity) and heterogeneity of the grazed pasture may result in unequal variances or nonnormal error distributions, thus restricting the use of an analysis of variance. The experiment design must recognize the requirements for making reliable inferences. Pasture-to-pasture variability generally demands that pastures should be replicated in renovation studies to allow general inferences. Within pasture variability would support the need for multiple exclosures within each pasture. Costs associated with this kind of research limit the utility of idealized experimental designs. Several alternative experimental designs are discussed. Limitations in interpretation and risks of drawing erroneous or weak conclusions are reviewed.