Post Hoc Tests, Samples, and Populations

The purpose of a post hoc analysis is to evaluate the null hypothesis associated with each contrast that is investigated. As we have pointed out several times, many applied researchers seem to forget this exceedingly important point. They often talk about their findings with reference only to their sample means, and they discuss their results in such a way as to suggest that they have proven something in a definitive manner. When they do so, they are forgetting that their "discoveries" are nothing more than inferences regarding unseen population means, with every inference potentially being nothing more than a Type I or Type II error.

(From Chapter 13, p. 362)

Copyright © 2003 Schuyler W. Huck
All rights reserved.