Statistical comparison of ERP waveforms

[Eric Maris]

Hi Litvak,


> I would like to perform statistical comparison between whole ERP waveforms
> (rather than just peaks). In my case I need paired ttest. My question is
> how should one solve the problem of false positives arising from multiple
> comparisons (all the timeframes)? Is far as I understood from the code the
> timelocked statistics function of Fieldtrip does not address this issue. I
> also looked in EEGLAB and it seems the same. Are there any good solutions
> (preferrable already implemented in Matlab)?


In FIELDTRIP, we use randomization statistics to control the family-wise
error rate.
We have combined randomization statistics with a cluster-finding algorithm
that
drastically improves power in exploratory analyses. This has been
implemented in the
routine CLUSTERRANDANALYSIS which, unfortunately for you, is only available
to users
inside the FC Donders Center. We will make it available to the rest of the
world in the
summer of 2005.


Eric Maris