Open c-kroeger opened 2 months ago
Hi Charlotte,
Thanks for pointing out this issue, The changing is pvalues is because emmeans_test()
interprets the variable provided in group_by(variable)
as a covariate of the model and it calculates expected groups means after controlling for the covariates.
We are currently looking into this issue and plan to release a definitive fix with the next cyCONDOR
version (0.2.1). Updates will follow on this thread.
Thanks again @c-kroeger and @skmueller for looking into this!
P-values are changing depending on which cluster/cell types are provided to the emmeans test after filtering the anova results. Seems like emmeans_test interprets the grouping by cluster/cell type strangely.