[FieldTrip] Between-condition permutation analysis

[Raghavan Gopalakrishnan]

Dear Fieldtrippers,

 

I have a scenario as depicted below with a oversimplified example.

Condition-1 has subjects-A (n=30),B (n=45),C(n=100)and D(n=50)

Condition-2 has subjects-X (n=90), Y(n=75) and Z(n=60)

 

Basically there are two conditions with 

1. totally different subjects (ABCD in condition-1, XYZ in condition-2)

2. different number of subjects (4 in condition-1, 3 in condition-2)

3. different number of trials per subject (as indicated by n next to each
subject)

 

I would like to perform cluster based time frequency comparisons between the
two conditions.

I have come across studies where same subjects are studied over two
conditions and have differing trial numbers. But, here I have different
subjects in each condition, different number of subjects percondition and
different number of trials per subject. 

Does group comparison of condition-1 vs condition-2 using cluster based
permutation analysis based on differing t-statistic hold valid in the above
mentioned experimental design? Any input would be of great help.

 

I was following some advice from a previous thread and paper from Groppe et
al. that permutation statistics are insensitive to population variance.

http://www.cogsci.ucsd.edu/~dgroppe/PUBLICATIONS/mass_uni_preprint2.pdf

https://mailman.science.ru.nl/pipermail/fieldtrip/2015-March/021934.html

 

Thank you.

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20200514/416610b1/attachment.htm>