I have done tPAC calculating on multiple trials for one condition, and want to get the averaged z-score result, so that I can compare the result with that of another condition. I am wondering which pipeline is appropriate:
① Z-score using surrogate for each trial, and then average the z-scored result across trials.
② Average the tPAC results of trials and surrogate data across trials, and then z-score using averaged surrogate data.
Could you give me an advise? Thank you.
I would go with the first option, but @Sylvain may provide more insightful suggestions.
If you average z-scores across trials, the outcome is not interpretable as a z-score anymore, right?
It depends on the reference mean and std you pick and the rationale behind standardizing with z-scores. What are they in your case?
Thanks Samiee and Sylvain.
As for z-score, I generate the surrogate data for each trial following the recommendations in this paper (Aru, J. 2015), and intend to calculate z-score using the mean and std of the surrogate database.
In your opinion, which option is more rational in this case?
Then I would recommend you z-score the average of tPAC across trials based on the mean and std observed across surrogates.
Many thanks for your reply.
If I understand correctly, you recommend to analyze as the following procedures:
- To calculate tPAC for each trial, and then get the average tPAC of all trials.
- For each trial, to calculate tPAC of multiple surrogates (100 times), and get the mean and std across surrogates.
- To average the surrogate mean and std of all trials.
- Z-score the average tPAC based on the result from step 3.
Did I get it right?
You may also want to refer to the following paper for more practical guidelines: