assumption on mixed designs
On the slides for mixed designs, we are shown to check the box for "equality of covariance matrices"...is this different from the covariance in ANCOVA? Also, it is an assumption in mixed designs correct?
posted by Nikki Frederick @ 8:43 PM
4 Comments:
At 9:23 PM,
KGL said…
Box's test of equality of covariance matrices is the test for assumption of homogeneity of covariance. You actually initially learned about this in the repeated measures lecture (it is part of compound symmetry). Homogeneity of covariance is the assumption that "covariances and correlations between pairs of levels are equal [that is] variance of difference scores are equal" (Lecture notes)
My guess is that this clarifies your question, if not, what part of ANCOVA are you thinking this relates to?
At 5:50 PM,
Nikki Frederick said…
I know the lecture notes as that's where my ? came from...so I guess I wasn't clear...I thought covariance was only in ANCOVA, but I see it's also assumed in RM as well.
At 10:41 AM,
Grace Liu said…
I think it is also important to distinguish covariance and covariate.
Covariance is something that you deal with in RM as Katie said.
Covariate is a confounding variable that is not a variable you are interested in your research model, but it affects your DV.
At 11:21 AM,
Nikki Frederick said…
Thanks!
Post a Comment
<< Home