I’m trying to look at differences between two behavioral conditions in terms of power topography of N=185 eeg electrodes. I have ~6 samples per session split between A/B at about a 1:2 ratio. In total, I have data from ~160 sessions (~100 subjects; some subjects had multiple sessions).
I’m applying a mixed-effects linear model in the following way:
import statsmodels.formula.api as smf md = smf.mixedlm('y ~ Q1_DUMMY', df_nona, groups=df_nona["ID"]) mdf = md.fit()
where y is the normalized power (per electrode), and Q1 is a dummy variable for conditions A or B I’m interested in and groups variable (random intercepts) refers to the SUBJECT ID.
While this only provides univariate statistics and no cluster-correction, does this seem like an appropriate test? Is there a different test to apply here?