 |
 |
Thanks to everyone for helping with this problem. I've now successfully implemented my 3 x 3 flexible factorial model, by including weightings on the relevant subject columns in the model (ie. Volkmar's solution). Everything works great, and the model performed as expected. All is not completely roses, however. I'm now trying to perform the exact same model, but with an additional covariate included. So it's still a 3 x 3 factorial design, but now I've included one covariate in the design matrix. Unfortunately, with the covariate included, Volkmar's solution no longer seems to work. That is, the contrast manager tells me that any non-balanced contrast (ie. that does not equal zero) is invalid. This seems odd to me, because Volkmar's solution of adding weights to the relevant subject columns works when the covariate is not included. But something about adding the covariate to model unbalances the model again. Does anyone have any idea why the covariate may cause this effect? To remind readers of my model, I have 2 conditions: Group (3 levels), and Trialtype (3 levels). There are 10 participants per group. Using the flexible factorial model, my design matrix thus has 40 columns: 30 Subject columns followed by the 9 Trialtype columns (3 for each group) followed finally by the covariate column And, to recap, the contrast: .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 .1 1 0 0 1 0 0 1 0 0 which follows Volkmar's advice, is a valid contrast as long as the covariate is not also included in the model. Adding the covariate as the 41st column in the matrix prevents this contrast from being valid, however. Any help would be appreciated. Matt