Thanks for your helpful feedback, Vladimir.
To further clarify, I have one factor with two levels - A1 and A2, and I have distributions that model functions fitted to A1 and A2 (B1 and B2, respectively).
I would like to take the parameter estimates from each participant to a second-level analysis where each participant serves as a random effect in a 2 x 2 ANOVA with factor A (two levels: A1 and A2) and function (B1 and B2). I would like to contrast B1>B2 for A1, and B2>B1 for A2, in order to test for timepoints and electrodes correlated with the level of B appropriate to the level of A.
Based on your previous feedback and section 41.5 of the manual, is the correct approach to initially convert each trial to an image and include this in a second-level analysis with factor A (level A1 and A2), and then two covariates corresponding to model B1 (the covariate followed by zeroes) and model B2 (zeroes followed by the covariate), and then to take the resulting beta weights after estimation into another 2nd level analysis where I would include appropriate linear contrasts as outline earlier? I'm assuming mean-centring is not necessary but I could be wrong.
Forgive me if this is unclear. I am happy to clarify otherwise.
Regards,
Chase Sherwell.
|