Planned Comparisons in Two-Way ANOVAs

The researchers associated with Excerpt 14.29 [not shown here] deserve credit for having specific plans in mind when they designed their study and when they analyzed their data. Their research questions guided their statistical analysis, and as we have noted, their analysis did not follow the conventional "rules" for making statistical comparisons in a two way ANOVA. Far too many applied researchers erroneously think that (1) F-tests for main and interaction effects must be computed and (2) comparisons of cell means can be made only if the interaction is significant. This is unfortunate for several reasons, the main one being that planned comparisons of cell means can sometimes produce interesting findings that remain undetected by the "standard" F-tests of a two-way ANOVA.

(From Chapter 14, pp. 415-416)

Copyright © 2003 Schuyler W. Huck
All rights reserved.