|Cole, Noel - Andy|
Submitted to: Journal of Animal Science
Publication Type: Abstract Only
Publication Acceptance Date: 4/22/2003
Publication Date: 6/22/2003
Citation: Richardson, C. R., Nunnery, G. A., Wester, D. B., Cole, N. A., Galyean, M. L. Power of the Test Considerations for Beef Cattle Experiments. Journal of Animal Science. 2003. v. 81(Suppl.1). p. 153. Abstract No. 680. Interpretive Summary:
Technical Abstract: The inherent value of using power of a test procedure in beef cattle experiments is similar to that for other species. However, because of major differences in the methods and conditions involved in conducting beef cattle experiments as compared to other species, considerations for the use of power test procedures are distinct and specific for this specie. Some of these major differences include: lack of similar research facilities which leads to wide fluctuations in the number of animals used per experimental unit by researchers; keeping experimental animals in totally or partially enclosed indoor pens, open outdoor pens, enclosed fields, or open ranges; use of individual animal data from Pinpointers, Calan gates, and metabolism studies; seasonal weather effects by region on animals kept outside; more variation in the performance of control groups from different locations because of differences in diet composition and animal genetics. When power tests are used in the planning and experimental design phase of a research study, they provide useful information to researchers on the sample sizes necessary to determine the probability of detecting a treatment effect at a predetermined ¿ level. In using power tests across different experimental designs, attention should be given to the consequences of both type I and type II errors. If tests are designed to lower either the type I error or the type II error rate, then the error rate of the other one increases and those consequences should be considered. For several common statistical procedures and experimental designs, tables of power values are available; however, none specifically addresses beef research and available software is not readily available. This presentation will review the body of data available in beef cattle research in which comparisons can be made to determine the effects that experimental design, size of the experimental unit, number of replications, type of housing, regional effects, feed composition, and genetics have on the merits of selecting appropriate power tests to improve the overall effectiveness of beef cattle research studies. In conclusion, the use of a suitable power test prior to conducting an experiment is useful, but post-experimental power calculations are inappropriate to use as an aid to interpret treatment effects.