Author Information

James R. Schwenke
J. Mark Donovan

Abstract

Analysis of covariance is a well-utilized statistical methodology. The procedure involves a series of statistical tests to first construct a most significant analysis model to characterize the effect of the covariate on response. Pairwise comparisons among treatments are then based on the finalized model.

For traditional Normal error assumptions, each step of the process is based on exact statistical tests. However, the series of statistical tests defines a conditional probability scheme with possible multiplicity issues. The question then becomes if the analysis of covariance methodology considered in entirety is able to maintain a nominal level of significance with good power.

Several procedures have been proposed in the literature suggesting different sequences of tests and understandings of analysis of covariance. This simulation study is being conducted to compare among a number of these analysis strategies. The initial goal was to investigate power of detecting treatment differences using the various analysis of covariance strategies. But, before power can be considered, the ability of the methodology to maintain a nominal level of significance must be investigated.

Creative Commons License


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

Share

COinS
 
Apr 25th, 3:30 PM

INVESTIGATING POWER OF ANALYSIS OF COVARIANCE METHODS

Analysis of covariance is a well-utilized statistical methodology. The procedure involves a series of statistical tests to first construct a most significant analysis model to characterize the effect of the covariate on response. Pairwise comparisons among treatments are then based on the finalized model.

For traditional Normal error assumptions, each step of the process is based on exact statistical tests. However, the series of statistical tests defines a conditional probability scheme with possible multiplicity issues. The question then becomes if the analysis of covariance methodology considered in entirety is able to maintain a nominal level of significance with good power.

Several procedures have been proposed in the literature suggesting different sequences of tests and understandings of analysis of covariance. This simulation study is being conducted to compare among a number of these analysis strategies. The initial goal was to investigate power of detecting treatment differences using the various analysis of covariance strategies. But, before power can be considered, the ability of the methodology to maintain a nominal level of significance must be investigated.