Dear colleagues, I am learning to use spm for analysis of fmri. My design includes two groups, one of patients and one of healthy controls. I have carried out the first level analysis, establishing the contrasts that interest me, for example the "Habits" condition. Now I’m at the point where I have to do the second level analysis to compare the results of the groups. I performed a two sample t-test with the following contrasts; matlabbatch{3}.spm.stats.con.consess{1}.tcon.name = 'PD > HC (GDS)'; matlabbatch{3}.spm.stats.con.consess{1}.tcon.weights = [1 -1]; matlabbatch{3}.spm.stats.con.consess{1}.tcon.sessrep = 'none'; matlabbatch{3}.spm.stats.con.consess{2}.tcon.name = 'HC > PD (GDS)'; matlabbatch{3}.spm.stats.con.consess{2}.tcon.weights = [-1 1]; matlabbatch{3}.spm.stats.con.consess{2}.tcon.sessrep = 'none'; matlabbatch{3}.spm.stats.con.consess{3}.tcon.name = 'HC (GDS)'; matlabbatch{3}.spm.stats.con.consess{3}.tcon.weights = [0 1]; matlabbatch{3}.spm.stats.con.consess{3}.tcon.sessrep = 'none'; matlabbatch{3}.spm.stats.con.consess{4}.tcon.name = 'PD (GDS)'; matlabbatch{3}.spm.stats.con.consess{4}.tcon.weights = [1 0]; matlabbatch{3}.spm.stats.con.consess{4}.tcon.sessrep = 'none'; matlabbatch{3}.spm.stats.con.delete = 0; when I look at the main effect in HC or PD there are results and they make sense but when I look at the comparison between PD> HC or HC>PD the result is null or I don’t get any activity. For honesty I have to say that if I lower the level of significance to 0.05 uncorrected I can see some activity. Do you think that the lack of activity is due to a real lack of difference between groups or that there may be some mistake?
I think it'd be hard to see if therey's a mistake in the analyses only from those few lines from the job file (which look okay). Could you upload your scripts on github and share the link?
Hi Davide Aloi, thanks for the suggestion. I uploaded the scripts as you suggested, this is the link if you can take a look at it I could be very helpful. thanks!
https://github.com/LinaGuida/Script_fmri.git
Everything looks correct to me. What happens if you lower the threshold even further than 0.001 uncorrected? Are you able to see something?
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis