Abstract: | Replies to comments by D. A. Sprott (see record 2005-09606-003) on the current author's original article (see record 1968-09688-001). The original purpose of the illustration was to show that analysis of covariance could leave a statistically significant residual treatment effect even though there was no treatment effect in the variate not predictable from the covariate. As to our statement that there is a "significant effect even when none was present," the context of the paragraph should make it clear that the word none refers to "no treatment effect in the variate not predictable from the covariate." (PsycINFO Database Record (c) 2010 APA, all rights reserved) |