Hi Marta,

There is no reason for you not to present the results using all these methods (especially since most software packages automatically provide all methods). That is particularly valuable if you find discrepancies between them. If they all concur, you can simply state that you ran the analyses using the various approaches and achieved similar (if not identical) values.

Ariel

Since I don't know your data, I can only speculate. My guess is that the difference has to do with the inclusion of the covariate (electrodes) vs running the standard weighted difference in means. It is the equivalent of comparing a t-test to a regression that has an additional covariate. If that covariate is meaningful, then the results will obviously differ.