[Eeglablist] Warning: change STUDY design

[Makoto Miyakoshi]

Dear Jet,

> ICA has been performed for the two datasets separately.

Good.
The basic rule is that removing an electrode cap and new ICA is necessary,
unless you can replicate the original cap/electrode location with
sub-millimeter precision.
https://sccn.ucsd.edu/wiki/Makoto's_preprocessing_pipeline#General_tips_for_performing_ICA_.2806.2F26.2F2018_updated.29

> I understand that that is why I get to see the warning I mentioned.

There is EEGLAB 'session' which, I believe, is supposed to use 'same
subject but different capping'. However, I often simply regard the same
subject different sessions as different subjects. This way you can avoid
troubles you experienced.

> Should this be possible?

No. You can't 'merge' different sessions (i.e., different capping) from the
same subject. EEGLAB's 'merge' only works if and only if ICA-related
matrices are shared and identical (i.e.,  paired t-test is possible).

> However, since the two datasets have been aqcuired on two separate days I
understand that ICA analyses should be performed on the datasets
separately.

You should treat them as if the data were from different subjects (i.e.,
you have to use two-sample t-test)

Makoto

On Tue, Jul 17, 2018 at 10:34 PM Jet Zoon <jetzoon at gmail.com> wrote:

> Dear eeglablist,
>
> In using eeglab to perform analyses for a study with a rather complex
> design I ran into the following warning: Warning: ***** change STUDY design
> as it combines datasets with different ICA decompositions.
>
> A short description of the design:
> testday 1: participants receive a high-energy savoury meal
> testday 2: participants receive a high-energy sweet meal
> Before and after meal intake we measured neural responses (using EEG) to
> visual and olfactory stimuli of the following categories:
> High-energy savoury food
> High-energy sweet food
> Low-energy savoury food
> Low-energy sweet food
> Non-food
> Baseline (no product/no odour)
> Each stimulus category was presented 18 times before and 18 times after
> meal intake.
>
> To keep the statistical power as high as possible, with the limited amount
> of items we presented, I would like to combine stimulus categories between
> the two test sessions. I think that it is best to end up with the stimulus
> categories: fully congruent (between meal and stimulus: e.g. High-energy
> sweet meal - high-energy sweet odour), taste congruent, energy congruent,
> not congruent, non-food and baseline.
>
> I now included data from two sessions on two separate days in the study
> dataset. ICA has been performed for the two datasets separately. I
> understand that that is why I get to see the warning I mentioned. However,
> in contrast to what is mentioned in an earlier conversation on this topic (
> https://sccn.ucsd.edu/pipermail/eeglablist/2011/003669.html), I can run
> the analyses and get my results.
>
> Should this be possible? Can I just interpret the results as usual or does
> combining items from two datasets provide invalid results? If so, is there
> another option to combine the items from the two datasets? In the message I
> referred to, it is suggested to only merge sessions with the same ICA
> decomposition. However, since the two datasets have been aqcuired on two
> separate days I understand that ICA analyses should be performed on the
> datasets separately.
>
> I really appreciate your help in resolving this problem.
>
> Kind regards,
> Jet Zoon
>
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu



-- 
Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20180718/e318cad9/attachment.html>