Hi,

You need to model a mean or you run the risk of fitting your data very poorly as a lack of a mean (intercept) in the model implies that the fitted line goes through the origin. This is rarely the case so an intercept (column of 1s) is always added. If you can find a basic linear regression book it will explain this further.

The only reason we don't have an intercept in the level 1 analysis is because Feat mean centers both the data and the design. For higher level analyses this is not done and so we must put the intercept in on our own.

Hope that helps,
Jeanette

On Thu, Feb 23, 2012 at 9:15 AM, Noam Schneck <[log in to unmask]> wrote:

Hi,

This is a basic question, but here goes:

1) In a higher level FSL analysis looking at activity for some cope related to a behavioral or clinical covariate, is it necessary to covary for the group mean in the higher level analysis? Why?
2) In a PPI analysis in which the cope is physiological activity of some seed region, is it necessary to control for group mean?

Thanks,
Noam