[FieldTrip] 2 (between subject) x 2 (between subject) x 3 (within subject) ANOVA for Time Frequency Plots
Maris, E.G.G. (Eric)
e.maris at donders.ru.nl
Tue Jun 23 15:22:46 CEST 2015
From: russ port <russgport at gmail.com<mailto:russgport at gmail.com>>
Subject: [FieldTrip] 2 (between subject) x 2 (between subject) x 3 (within subject) ANOVA for Time Frequency Plots
Date: 22 Jun 2015 17:37:07 CEST
To: <fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>>
Reply-To: FieldTrip discussion list <fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>>
After reading the wiki (http://www.fieldtriptoolbox.org/faq/how_can_i_test_an_interaction_effect_using_cluster-based_permutation_tests), it appears that I cannot do a cluster based permutation test on a set of data I have because of the fact that I have two between subject factors (genotype and sex). There has been a lot of discussion on the mailing list about how to run these kinds of analyses, but they have always (or at least what I could find), been about data set containing only 1 between subjects factor. As such, I have a question in two parts A) could I look at the data with a proper test of the between subject factors, and the within subject factor AND their interaction using a permutation test that use holms or bonferroni correction (though I would probably use holms because of bonferroni being to conservative) B) should I set this up by either i) testing geno, sex and their interaction separately like I think the wiki suggests above or ii) try and use the cfg.cvar input as previously suggested else where on this discussion list (though again that was for data with only 1 between subject variable)?
With respect to question A:
I hope I understand you correctly here: you want to analyse a 2 (between) x 3 (within) design, separately for the the data selected according to the level of the other between-subjects factor (thus, 2 analyses of a 2x3 design). Yes, this is possible in the permutation framework. Using Bonferroni correction to account for the fact that you perform 2 tests instead of 1 is a good idea.
With respect to question B:
You cannot test the interaction geno x sex in the permutation framework because these are 2 between-subjects factors. Also, the cfg.cvar option is to be used in a within-subjects design (or a matched-pairs design, which is statistically equivalent to a within-subjects design). Thus, the cvar must be within-subjects variable.
fieldtrip mailing list
fieldtrip at donders.ru.nl<mailto:fieldtrip at donders.ru.nl>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the fieldtrip