Hi Paul,
Manually calculating the difference images and then running the stats on those is sound ( and makes the permutation strategy easier to understand ).
Kind Regards
Matthew
--------------------------------
Dr Matthew Webster
FMRIB Centre
John Radcliffe Hospital
University of Oxford
On 5 Mar 2019, at 15:15, paul geha <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Dear Mathew,
yes, actually I noticed after sending the email that the nuisance variables would be correlated to the mean of each subject. For example, for each subject I have the age repeated twice in the age nuisance variable column; this would be correlated to the EV modeling the mean of this subject.
But how then can we enter the nuisance variables ?
I am thinking of calculating the difference in the seed maps for each subject separately and then run an ANOVA between the groups with nuisance variables.
Please let me know if you think this is a sound statistic.
thanks a lot!
Paul
On Tue, Mar 5, 2019 at 9:01 AM Matthew Webster <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Dear Paul,
Do any of the 3-nuisance EVs look highly correlated to any of the mean/difference EVs or each other?
Kind Regards
Matthew
--------------------------------
Dr Matthew Webster
FMRIB Centre
John Radcliffe Hospital
University of Oxford
On 4 Mar 2019, at 22:07, paul geha <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Dear FSL experts,
I am trying to examine group differences using PALM. I have in total 15 subjects and 3 groups, with 5 subjects per group. Each subject has 2 different seed connectivity maps. I set a design.mat map and design.con to examine group differences for the difference between the 2 seed maps as such:
EV1: set to 1 for G1 seed 1 and -1 for G1 seed 2
EV2: set to 1 for G2 seed 1 and -1 for G2 seed 2
EV3: set to 1 for G3 seed 1 and -1 for G3 seed 2
EV(4-19) to model each subject's mean.
I set the contrasts C1: 1 -1 0 and C2: 1 0 -1 and C3: 0 1 -1 and use an F-test 1 1 0 (1 for C1 and 1 for C2 and 0 for C3)
This design works well and gives me results.
However, I want to add 3 nuisance variables to the design. If I just add 3 more EVs with the demeaned nuisance variables for all 15 subjects I get a Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND…and "rank deficient design ...." .
I also get a similar error message if separate the nuisance variables by group (so use 9 EVs instead of 3 only).
Please let me know how I can fix this design.
thanks!
Paul
________________________________
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
________________________________
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
________________________________
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
