Author Information

James R. Schwenke

Abstract

Of primary concern in the statistical analysis of the results of an experiment is to quantify the mean response to treatment and to accurately quantify the experimentation error variance. The traditional approach to account for nuisance sources of variation or a heterogeneous population is to group or block the population (or sample) into homogeneous groups with respect to a concomitant variable. A blocking term then is included in the statistical analysis. Alternatively, concomitant variables can be used as covariate information in a statistical analysis. A statistical analysis incorporating blocks assumes that the magnitude of difference in treatment response is equal across all blocks. Covariate information is used in an analysis to describe individual differential treatment effects on response. Covariate information used in conjunction with blocks may allow for a more realistic and appropriate estimate of the experimentation error variance and, thus, a more powerful analysis.

A series of examples is presented to demonstrate the potential advantage to utilizing both block and covariate information in an analysis of variance.

Creative Commons License


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

Share

COinS
 
Apr 27th, 9:30 AM

COMPARING THE USE OF BLOCK AND COVARIATE INFORMATION IN ANALYSIS OF VARIANCE

Of primary concern in the statistical analysis of the results of an experiment is to quantify the mean response to treatment and to accurately quantify the experimentation error variance. The traditional approach to account for nuisance sources of variation or a heterogeneous population is to group or block the population (or sample) into homogeneous groups with respect to a concomitant variable. A blocking term then is included in the statistical analysis. Alternatively, concomitant variables can be used as covariate information in a statistical analysis. A statistical analysis incorporating blocks assumes that the magnitude of difference in treatment response is equal across all blocks. Covariate information is used in an analysis to describe individual differential treatment effects on response. Covariate information used in conjunction with blocks may allow for a more realistic and appropriate estimate of the experimentation error variance and, thus, a more powerful analysis.

A series of examples is presented to demonstrate the potential advantage to utilizing both block and covariate information in an analysis of variance.