Control

In the discussions of both hierarchical multiple regression and logistic regression, we saw that researchers often incorporate control or covariate variables into their analyses. Try to remember that such control is very likely to be less than optimal. This is the case for three reasons. First, one or more important confounding variables might be overlooked. Second, potential confounding variables that are measured are likely to be measured with instruments possessing less than perfect reliability. Finally, recall that the analysis of covariance undercorrects when used with nonrandom groups that come from populations that differ on the covariate variable(s). Regression suffers from this same undersirable characteristic.


(From Chapter 16 in the 6th edition, p. 400)

 

Copyright © 2012

Schuyler W. Huck
All rights reserved.

| Book Info | Author Info |

Site URL: www.readingstats.com

Top | Site Map
Site Design: John W. Taylor V