Hallo,
by estimating a PPI GLM as described in section 33.5.1 but using a Second Level Regression Model instead of the described First Level Model, the outcome (i.e. PPI-Interaction t-contrast) is totally different. Could anyone describe in principle what is the difference between estimating a regression model in first level mode vs second level mode with the exact same data (same scans, same regressors)? Why is First Level better for PPI Analysis?
Thank you very much, best wishes
Holger Mohr
|