11. Some Alternatives to Traditional ANCOVA
Published Online: 30 SEP 2013
Copyright © 2011 John Wiley & Sons, Inc.
Anova and Ancova, Second Edition
How to Cite
Rutherford, A. (2011) Some Alternatives to Traditional ANCOVA, in Anova and Ancova, Second Edition, John Wiley & Sons, Inc., Hoboken, New Jersey. doi: 10.1002/9781118491683.ch11
- Published Online: 30 SEP 2013
- Published Print: 7 OCT 2011
Print ISBN: 9780470385555
Online ISBN: 9781118491683
- general linear models (GLMs);
- heterogeneous regression;
- traditional ANCOVA
This chapter focuses on heterogeneous ANCOVA general linear models (GLMs), and discusses two further alternatives to traditional ANCOVA and the ability of heterogeneous regression ANCOVA to ameliorate the problems caused by a relationship between the covariate(s) and the experimental conditions. The basic problem posed by heterogeneous regression ANCOVA is that experimental effects vary as a function of the covariate. Applying traditional ANCOVA when a relationship exists between the covariate and the experimental conditions will provide an analysis which is statistically correct, but it is unlikely to apply easily or usefully to the real world. By avoiding covariate distribution conflation, the relation between the experimental conditions and the covariate in heterogeneous regression ANCOVA changes from a condition that seriously complicated the interpretation of adjusted experimental effects and questioned the real world validity of the analysis, to the familiar regression topic of multicolinearity.
Controlled Vocabulary Terms
analysis of covariance; covariate; regression