Dear FSL experts,
I am reposting this question because the question wasn't clear enough.
I modeled my 3 groups X age interaction with Glm and entered the desing in randomise.
Extracting the FA values from the resulting p map, the mean and SD can not properly represent these findings.
A statistical software such as SPSS, in ANCOVA tests based on a GLM, produces the "estimated marginal mean" to take into account the covariate effect. In SPSS there is not an equivalent analysis of non-parametric tests using GLM designs, such as Randomise performs.
Is there any way to "derive" a corresponding measure of the estimated marginal mean using Randomise??
Any reply would be greatly appreciated!!
Thank you so much,
Amy
|