I've been struggling for the past few days trying to figure out how to correctly run a mixed model ANCOVA.
I have a continuous dependent variable, a continuous independent variable, and a categorical independent variable. My categorical variable should be a random factor. I'd like to know the effect of each independent variable and the interaction term on the dependent variable. To me, it looks like I need a mixed model ANCOVA.
I've tried doing this in JMP, as well as using the anovan and aoctool functions in MATLAB. Each gives me a different result. Honestly, I can't even interpret the output from JMP because it doesn't give you a convenient effects table when you include a random factor in your model. I hate to be one of those people who simply trusts the output of a stat program, but I don't know enough about the math to understand why I would get different results from the different programs.
Is there a mathematical difference between an "ancova" and an anova specifying a continuous factor (this is how the anovan function works)? Should be using a different program?
In some places I've read that an ANCOVA removes the effect of the continuous variable so that you can test the difference in means between the categorical groups. This is information that I need, but I also need to know the effect of the continuous variable when you remove the effects of the categorical variable. Will this test even give me the information I'm looking for?
I'm basically stuck until I can figure this out, so I really appreciate any help you can offer!
Advertise on Talk Stats