Author Information

Marjorie E. Bond
James J. Higgins

Abstract

It is common in agricultural research to have experimental units that consist of multiple observational units. For instance, treatments may be applied to pens of animals, pens being the experimental units, while weights are measured on individual animals, the observational units. If there are a small number of experimental units, the power of statistical tests for treatment effects can be small regardless of the number of observational units. We show that it is possible to increase the power of such statistical tests by taking advantage of prior knowledge of the intraclass correlation. Our assertion is that such prior knowledge is often available although infrequently used. We present several simple methods for taking advantage of this prior knowledge and show that the power of tests based on these methods can be substantially greater than the power of conventional tests specially when the number of experimental units is small.

Creative Commons License


This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Share

COinS
 
Apr 27th, 8:15 AM

USING PRIOR KNOWLEDGE OF THE INTRACLASS CORRELATION TO INCREASE THE POWER OF TESTS FOR TREATMENT MEANS

It is common in agricultural research to have experimental units that consist of multiple observational units. For instance, treatments may be applied to pens of animals, pens being the experimental units, while weights are measured on individual animals, the observational units. If there are a small number of experimental units, the power of statistical tests for treatment effects can be small regardless of the number of observational units. We show that it is possible to increase the power of such statistical tests by taking advantage of prior knowledge of the intraclass correlation. Our assertion is that such prior knowledge is often available although infrequently used. We present several simple methods for taking advantage of this prior knowledge and show that the power of tests based on these methods can be substantially greater than the power of conventional tests specially when the number of experimental units is small.