Sorry, I don't understand the R terminology. in the FSGD, you will have
(4+1)*3 = 15 columns to the design matrix (4=2x2=number of classes, 3 =
no of continuous variables). Each column will mean something. How many
columns will you have in your R analysis and what will they mean? If you
can make R spit out a design matrix, you can just use that via the --X
option to mri_glmfit.
On 2/9/2022 6:59 AM, Marina Fernández wrote:
External Email - Use Caution
Dear experts,
We run the following analysis in Freesurfer:
*Group* (2 levels) x *X1* (continuous IV) x *X2 *(continuous IV) +
*Sex *(categorical covariate) + *Age* (continuous covariate)
We exported the vertex with the maximum statistic after applying FWE
correction to replicate the analysis in R. We want to make sure about
the structure of the model to later do a Bayesian analysis using the
same structure.
We got significant results for most of our analyses but not for all of
them, likely because we are not doing exactly the same thing.
We have fitted several models in R, but the one that gave the closest
results was the following:
model <- lm (vertex ~ NewFactor (4 levels: Group*Sex) + X1X2
(resulting from X1*X2) * Group + Age*NewFactor, data=mydata)
How should we introduce the interaction term in order to make the
models fitted by Freesurfer and R comparable?
Thanks a lot in advance,
Marina
___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
