Dear SPM user,
I have N-subjects and two conditions. I want to compare these conditions at source level (MEG-data, evoked).
First I used accidentally unpaired two-sample t-test. Afterwards I used the paired t-test, the correct t-test to compare both conditions.
The results a curious: For the unpaired two-sample test I get significant results (corrected for FWE, p<0.05) for the contrast Cond1 > Cond2 (t-contrast). For the paired t-test the significant results disappear (only uncorrected p<0.001) for the same contrast Cond1 > Cond2 (t-contrast). The brain coordinates are nearly same for both contrasts.
My question: Why disappeared the significant results for the paired t-test? - What is the reason?- I expected better results for the paired t-test. Is this assumption incorrect?
Best regards,
Cerisa Stawowsky
