External Email - Use Caution
Hello Doug,
We are doing one test I supposed. Because we inputted all covariates into glm_fit.
This is our command line.
We entered all behavioral scores as covariates and obtained each folder named as covariate's name with sig.mgh in each folder.
Does it mean we only have just one test, or several tests if we did the below way?
Best regards, Jahanvi
mri_glmfit \ --glmdir GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2 \ --y lh.testretest_only_TP2.thickness-pc1.stack.fwhm15.mgh \ --label lh.testretest.fsaverage.cortex.label \ --fsgd GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/TP1_TP2.fsgd \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/main.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_age.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_CesD_score.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_STATI_state.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_STATI_trait.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_total_medi_min.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_FFMQ_f1.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_FFMQ_f2.mtx \ --surf fsaverage lh
Message: 4 Date: Mon, 25 Jun 2018 10:45:42 -0400 From: Douglas Greve dgreve@mgh.harvard.edu Subject: Re: [Freesurfer] Question about GLM results in Freesurfer To: freesurfer@nmr.mgh.harvard.edu Message-ID: 10e16d65-9ac1-c85c-875d-405e852ce9a5@mgh.harvard.edu Content-Type: text/plain; charset="utf-8" Are you doing 5 different tests or just one test? It is a tricky question as to whether you should correct across? 5 tests or not. Ideally, one should, but I don't think I've ever seen it done in practice. As a reviewer, I might not demand it for only 5 tests.
On 6/20/18 9:17 AM, jahanvi patel wrote:
Hello,
We have a question about using glm_fit in freesurfer.
When we model several covariates (behavioral scores) in glm, and we wanted to know the correlation between cortical thickness changes with the changes of behavioral scores.? We obtained the correlation results nicely after using freesurfer's monte carlo correction (eg. increase of behavioral scores is correlated with the increase of cortical thickness changes).
Should we still perform the bonefferi corrections on the results we got ? (because we have entered?5 covariates into the glm?model, so we are wondering if it is still needed to divide the?alpha-value by 5 to?determine the final significant findings)?
Thank you so much for helping us,
Jahanvi
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Each contrast is a test, so you have 8 tests. Like I said, the ultimate right thing to do is to correct for all your tests, but this might not leave you with anything if you do. This is often a judgement call. I rarely, if ever, see people correct over multiple tests. However, if you try to publish with a bunch of tests and only one is sig, that might raise some flags with the reviews. Often people have a priori tests and then other tests that they are vaguely interested in.
On 6/26/18 5:15 AM, jahanvi patel wrote:
External Email - Use Caution
Hello Doug,
We are doing one test I supposed. Because we inputted all covariates into glm_fit.
This is our command line.
We entered all behavioral scores as covariates and obtained each folder named as covariate's name with sig.mgh in each folder.
Does it mean we only have just one test, or several tests if we did the below way?
Best regards, Jahanvi
mri_glmfit \ --glmdir GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2 \ --y lh.testretest_only_TP2.thickness-pc1.stack.fwhm15.mgh \ --label lh.testretest.fsaverage.cortex.label \ --fsgd GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/TP1_TP2.fsgd \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/main.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_age.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_CesD_score.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_STATI_state.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_STATI_trait.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_total_medi_min.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_FFMQ_f1.mtx \ --C GLM_dir_corr_behavior_lh_CT_tp1_tp2_plus_FFMQ_f1_f2/contrasts/Delta_FFMQ_f2.mtx \ --surf fsaverage lh
Message: 4 Date: Mon, 25 Jun 2018 10:45:42 -0400 From: Douglas Greve <dgreve@mgh.harvard.edu <mailto:dgreve@mgh.harvard.edu>> Subject: Re: [Freesurfer] Question about GLM results in Freesurfer To: <freesurfer@nmr.mgh.harvard.edu <mailto:freesurfer@nmr.mgh.harvard.edu>> Message-ID: <10e16d65-9ac1-c85c-875d-405e852ce9a5@mgh.harvard.edu <mailto:10e16d65-9ac1-c85c-875d-405e852ce9a5@mgh.harvard.edu>> Content-Type: text/plain; charset="utf-8" Are you doing 5 different tests or just one test? It is a tricky question as to whether you should correct across? 5 tests or not. Ideally, one should, but I don't think I've ever seen it done in practice. As a reviewer, I might not demand it for only 5 tests.On 6/20/18 9:17 AM, jahanvi patel wrote:
Hello,
We have a question about using glm_fit in freesurfer.
When we model several covariates (behavioral scores) in glm, and we wanted to know the correlation between cortical thickness changes with the changes of behavioral scores.? We obtained the correlation results nicely after using freesurfer's monte carlo correction (eg. increase of behavioral scores is correlated with the increase of cortical thickness changes).
Should we still perform the bonefferi corrections on the results we got ? (because we have entered?5 covariates into the glm?model, so we are wondering if it is still needed to divide the?alpha-value by 5 to?determine the final significant findings)?
Thank you so much for helping us,
Jahanvi
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
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
freesurfer@nmr.mgh.harvard.edu