Hello!
I have the ERP data (1 condition) of 15 participants and would like to have the Grand Average (across participants). For the same, I went Run>Average>Everything. I would like to know if the method mentioned is correct? If not kindly let us know the solution of the same.
Thanks in Advance!
Yes, this is the correct procedure.
Thank you, @Sylvain though I have a doubt, by using the same methodology, I calculated the Grand Average (15 subjects) and then calculated the PSD.
And Alternatively, I took all the 15 subjects and calculated the PSD (selecting the 'Save Average PSD values(across trials)').
The PSD results from these two methods were significantly different, what do you think what's the underlying reason/issue for the same.
Thanks.
The PSD of the average time series is indeed not equivalent to the average of the time series' PSDs.
The procedure needs to match your research question and hypothesis. By averaging the time series before deriving the PSD, you assume that the time features of brain responses across all subjects are equivalent, regardless of possible differences terms of time delays of ERP responses.
Thank You, @Sylvain for the same.
If we assume that the components of the brain signals are strictly time-locked to the presentation of a stimulus; taking an example of an Oddball paradigm (standard vs deviant), shouldn't we initially average the time series (for all the participants) and then do the PSD.
Also, let me give you an illustration of the dataset/paradigm. I would really appreciate if you could give your input on the same.
We have a choice preference task of various products between two categories.
For the same, we did the initial processing in Brainvision Analyzer sw and imported the epochs (ERPs responses) of all the participants in Brainstorm.
Now, in order to calculate the Average PSD among all the participants, what should be the ideal way to do it.
Our hypothesis is that there will be an increase in the N200 component correlates with a more preferred product.
Looking forward to it.
Thank You!
The PSD should be preferably computed over continuous recordings.
If you don't have access to the continuous recordings, you could compute it from the individual trials, but only if the duration of the epochs is sufficiently long.
In that case, do not try to compute the PSD with the Welch method from each trial. Instead, select all the trials of one condition/one subject in the Process1 list, select the process "Frequency > Fourier Transform (FFT)", with the option "Save average selected".
Then, average across participants the subject-averages of the FFT.
If you have doubts regarding any of these processes, please get back to the introduction tutorials.