Hi Dr. Zalesky,

I have skimmed through through previous forum issues regarding the proper design matrix. I was hoping you could confirm that I am using the correct design for my 2x2 (group x condition) mixed analysis?
For this particular matrix (please see attached .mat file): column 1 models condition A and B, column 2 is the interaction, group 1 has n = 57, group 2 has n = 52.

Part of my concern is that there is no column to model the group effect. But as I have experienced (and as you have noted in previous forum responses), adding the group column will make my matrix rank deficient. So then would it be correct to assume that the remaining columns in the design matrix (columns 3-111) account for both group and subject variance?

Thanks for your time, any feedback would be much appreciated!

Kim

Hi Kim,

Thanks for your interest. Jeanette Mumford has some nice slides on 2x2 repeated measures ANOVA (among other designs). Here is the link (see slide 78):

mumford.fmripower.org/setting_up_designs_2009.pdf

You are right, in that adding a column modelling the main effect of group will make the design rank deficient. If you are interested in the main effect of group, then a repeated measures design is not really necessary here.

I had a quick look at you design and it looks fine. Don't forget to specify an exchange block.

Please also take a look at the post on this from a few days ago by Frederic, where this issues is discussed.

I hope that helps!

Andrew




Originally posted by Kim Ray: