Thanks Doug,

I have two follow-up questions.

1. I am using QDEC and my discrete variables are: (a) TYPE (PATIENTS/CONTROLS) and (b) GENDER
I include age as a continous co-variate and  then select all of the variable before running the design.

I believe QDEC is generating the various possible contrast matrices and the corresponding research question.

When I run QDEC, I see the following questions:

A. Does the average thickness, accounting for Gender differ between Patients and Controls?
B. Does the thickness-age correlation, accounting for Gender differ between Patients and Controls ? (I think this is looking at whether the slopes are different for the two groups)

 I was wondering why I dont see a question like "Does the average thickness, accounting for Gender AND AGE differ between Patients and Controls?"

2. It is likely I dont have a full understanding of the GLM theory. Could you please suggest some good references describing the GLM theory ?
Thanks
Mehul




On Thu, Oct 29, 2009 at 10:57 AM, Douglas N Greve <greve@nmr.mgh.harvard.edu> wrote:


Mehul Sampat wrote:
Hi FS folks,

I have a basic GLM question. I went through the tutorials online but I was not sure and wanted to check with someone.

I am trying to compare the cortical thickness between a group of patients (n = 166) and controls (n = 76).

For patients mean age is 49.8 +/- 9.1 and there are 55 Male; 111 Female
For controls mean age is 40.5 +/- 11.4 and there are 26 Male; 50 Female

If include gender as a fixed factor does the output of the GLM answer the following:
1. "Is the cortical thickness different between the patients and controls accounting for gender"
All the factors in mri_glmfit/QDEC are random factors. But, yes, it would answer that question.


2. As I understand the GLM setup one can control for gender and other discrete factors but not for continuous co-variates such as age ?
That is one can find the association between thickness and age and see if it is different for the two groups.
However if the age distributions for the two groups are different one cannot control for it with GLM.  is this interpretation correct ?
Not quite. You can always put age in as a continuous covariate. If there are effects of age or an interaction between age and group, then there are some subtle statistical issues.

doug

If so, how would one control for age in such an analysis ?

Any help is much appreciated.

Thanks
Mehul




2. For the correction of multiple comparisons, when should one use FDR as compared to monte-carlo simulations ?

Thanks
Mehul
 
------------------------------------------------------------------------

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

--
Douglas N. Greve, Ph.D.
MGH-NMR Center
greve@nmr.mgh.harvard.edu
Phone Number: 617-724-2358 Fax: 617-726-7422

In order to help us help you, please follow the steps in:
surfer.nmr.mgh.harvard.edu/fswiki/BugReporting