Group level analyses trouble-shooting


I just ran some group level analyses. Some of them look totally normal, but certain conditions look like this.

Has anyone seen this before? We were planning on going through the data from all the subjects for the conditions that look like this, and see if any particular person has a messed up run.

Any guidance at all would be appreciated! :slight_smile:


Goooootttttta have some details to try to even start to think about addressing the glimmer of an idea as to approach this question.

For starters:

  • how many subjects
  • what kind of processing ( command used would be awesome)
  • what group analysis are you running? (again, the exact command would be useful, if not necessary).



Thank you for your patience! I’m sorry I didn’t include all of the relevant info, but here is some additional information:

  • We see this issue in all three of our subject groups: one is 28ish, the other is 40ish, and so the third total sample has around 68 subjects

  • We are using 3dMVM

  • I have attached the majority of one of our scripts below:
    (Also note that we have identical scripts for other conditions that worked great and did not give us this rainbow confetti, hence our wanting to go through the Individual Level data from all the subjects for the conditions that look like this to see if any particular person has a messed up run)

# Specify the top-most directory associated with the study
set topdir = /data/projects/STUDIES/Social_Doors/Social_Doors_KIDS/fMRI_ANALYSES

# Specify the directory where subjects' neuroimaging data are located
set subjdata = $topdir/IndvlLvlAnalyses

# Specify the directory where the group analysis results will be written
set results = $topdir/GroupLvlAnalyses/3dMVM

cd $results

3dMVM 																											\
 -prefix 3dMVM.FB.Doors.091719 																				    \
 -jobs 11																										\
 -wsVars 'Task*Outcome'																							\
 -num_glt 14																									\
-gltLabel 1 LoseGames 						-gltCode 1 'Task : +1*LoseGames'								    \
-gltLabel 2 LoseGames_Lose 					-gltCode 2 'Task : +1*LoseGames Outcome : +1*Lose'			        \
-gltLabel 3 LoseGames_Win 					-gltCode 3 'Task : +1*LoseGames Outcome : +1*Win'			        \
-gltLabel 4 LoseGames_Win.V.Lose				-gltCode 4 'Task : +1*LoseGames Outcome : +1*Win -1*Lose'	    \
-gltLabel 5 WinGames 							-gltCode 5 'Task : +1*WinGames'								    \
-gltLabel 6 WinGames_Lose 						-gltCode 6 'Task : +1*WinGames Outcome : +1*Lose'				\
-gltLabel 7 WinGames_Win 						-gltCode 7 'Task : +1*WinGames Outcome : +1*Win'				\
-gltLabel 8 WinGames_Win.V.Lose 				-gltCode 8 'Task : +1*WinGames Outcome : +1*Win -1*Lose'		\
-gltLabel 9 Lose 								-gltCode 9 'Outcome : +1*Lose'									\
-gltLabel 10 Win 								-gltCode 10 'Outcome : +1*Win'									\
-gltLabel 11 Win.V.Lose 						-gltCode 11 'Outcome : +1*Win -1*Lose'							\
-gltLabel 12 LoseGames.V.WinGames 			-gltCode 12 'Task : +1*LoseGames -1*WinGames'				        \
-gltLabel 13 LoseGamesWin.V.WinGamesWin		-gltCode 13 'Task : +1*LoseGames -1*WinGames Outcome : +1*Win'	    \
-gltLabel 14 LoseGamesLose.V.WinGamesLose	-gltCode 14 'Task : +1*LoseGames -1*WinGames Outcome :+1*Lose'	    \
-dataTable																										\
Subj         Task			Outcome		InputFile				     											\
s010        LoseGames	    Lose		$subjdata/s010/s010.results.doors.1mm/stats.s010+tlrc'[FB.LoseGames.Lose#0_Coef]'	    \
s010        LoseGames	    Win			$subjdata/s010/s010.results.doors.1mm/stats.s010+tlrc'[FB.LoseGames.Win#0_Coef]'		\
s010        WinGames		Lose		$subjdata/s010/s010.results.doors.1mm/stats.s010+tlrc'[FB.WinGames.Lose#0_Coef]'		\
s010        WinGames		Win			$subjdata/s010/s010.results.doors.1mm/stats.s010+tlrc'[FB.WinGames.Win#0_Coef]'		    \

#continued for all participants 

s5140        LoseGames	    Lose	    $subjdata/s5140/s5140.results.doors.1mm/stats.s5140+tlrc'[FB.LoseGames.Lose#0_Coef]'	\
s5140        LoseGames	    Win	        $subjdata/s5140/s5140.results.doors.1mm/stats.s5140+tlrc'[FB.LoseGames.Win#0_Coef]'	    \
s5140        WinGames		Lose	    $subjdata/s5140/s5140.results.doors.1mm/stats.s5140+tlrc'[FB.WinGames.Lose#0_Coef]'     \
s5140        WinGames		Win	        $subjdata/s5140/s5140.results.doors.1mm/stats.s5140+tlrc'[FB.WinGames.Win#0_Coef]'

tar -cvzf $results/*.BRIK

Let me know what else you need from me!