Dear Amir,
This question can be answered quite well in SPM framework. When you
are doing TF analysis and export to images you can choose to average
over frequency. That'll give you a scalp x time image of power in the
band of interest. You then do a t-test between the two conditions and
for the condition where the activity starts earlier you'll get higher
activation at this time. If you use FWE correction that'll control
your multiple comparisons. You can then plot the waveforms of power
(that will appear in an additional dataset that will be created with
the images) for the two conditions, mark the significant bit and show
that indeed it's because in one of the conditions the activity starts
earlier. Unfortunately there is not way yet to do this kind of
plotting automatically but it will be possible in the future.
Best,
Vladimir
On 13 Dec 2009, at 09:57, Amir H Javadi wrote:
> Dear Vladimir :-)
>
> This time, I have a statistical question rather than SPM. Sorry if
> it is too irrelevant.
>
> OK. I have 48 electrodes and the experiment has two conditions, say
> Remembered and Forgotten. The onset of frequency activity of a
> frequency band for these two conditions are different. From visual
> investigation it is clear that the onset of frequency activity of a
> specific frequency band for remembered condition is earlier than the
> forgotten condition almost for all the electrodes.How can I
> statistically analyse data as the electrodes are not independent and
> i have to do lots of corrections?
>
> Have a good time :-)
> Amir
>
|