help
help > RE: Remove subjects from analysis
Jul 16, 2015 05:07 PM | Alfonso Nieto-Castanon - Boston University
RE: Remove subjects from analysis
Hi Darren,
There are several (at least three) equivalent procedures for exclusing some subjects from your second-level analyses (see this post https://www.nitrc.org/forum/message.php?...), could you please tell me more details about how exactly you set up your second-level analysis to see if I can figure out the source of the problem?
Thanks
Alfonso
Originally posted by Darren Yeo:
There are several (at least three) equivalent procedures for exclusing some subjects from your second-level analyses (see this post https://www.nitrc.org/forum/message.php?...), could you please tell me more details about how exactly you set up your second-level analysis to see if I can figure out the source of the problem?
Thanks
Alfonso
Originally posted by Darren Yeo:
Hi
Alfonso,
I tried to follow the procedure that you specified below to exclude subjects with outlying scores. However, there is no change to the degrees of freedom in the output of my second-level analyses. Do you happen to know what might have gone wrong?
Thanks!
Best,
Darren
Originally posted by Alfonso Nieto-Castanon:
I tried to follow the procedure that you specified below to exclude subjects with outlying scores. However, there is no change to the degrees of freedom in the output of my second-level analyses. Do you happen to know what might have gone wrong?
Thanks!
Best,
Darren
Originally posted by Alfonso Nieto-Castanon:
Hi
Brian,
Yes, you may simply create a Second-level covariate (in Setup.Covariates.Second-level) indicating your subject groups (e.g. PAT and CON variables, dummy coding using 1/0's to indicate which subject belongs to each group). Then in your second-level results tab, simply select the CON covariate in the Between-subject effects list, and that will perform a one-sample t-test restricted only to your control group subjects (check the degrees-of-freedom of the analyses, that should be equal to your number of subjects minus 1). Similarly, you may select only the PAT covariate for looking at connectivity for the pathology group only, or select both PAT and CON, and enter a contrast [1 -1], to look differences in connectivity between the two groups.
Hope this helps
Alfonso
Originally posted by Brian Cechmanek:
Yes, you may simply create a Second-level covariate (in Setup.Covariates.Second-level) indicating your subject groups (e.g. PAT and CON variables, dummy coding using 1/0's to indicate which subject belongs to each group). Then in your second-level results tab, simply select the CON covariate in the Between-subject effects list, and that will perform a one-sample t-test restricted only to your control group subjects (check the degrees-of-freedom of the analyses, that should be equal to your number of subjects minus 1). Similarly, you may select only the PAT covariate for looking at connectivity for the pathology group only, or select both PAT and CON, and enter a contrast [1 -1], to look differences in connectivity between the two groups.
Hope this helps
Alfonso
Originally posted by Brian Cechmanek:
Hello everyone!
I was given a dataset of 240 subjects (98 pathologic, 142 control), pre-setup in the Conn toolbox for analysis. Looking at the results, we want to check just the control group versus one of our conditions, that is remove the 98 pathologic subjects. Is there a way that this can be done without beginning setup from scratch?
Happy to hear any good advice!
Thanks
B
I was given a dataset of 240 subjects (98 pathologic, 142 control), pre-setup in the Conn toolbox for analysis. Looking at the results, we want to check just the control group versus one of our conditions, that is remove the 98 pathologic subjects. Is there a way that this can be done without beginning setup from scratch?
Happy to hear any good advice!
Thanks
B
Threaded View
| Title | Author | Date |
|---|---|---|
| Brian Cechmanek | Aug 28, 2014 | |
| Alfonso Nieto-Castanon | Aug 31, 2014 | |
| Darren Yeo | Jul 12, 2015 | |
| Alfonso Nieto-Castanon | Jul 16, 2015 | |
| Darren Yeo | Jul 16, 2015 | |
| Alfonso Nieto-Castanon | Jul 16, 2015 | |
| Darren Yeo | Jul 24, 2015 | |