Post hocs / planned contrasts in Bayesian
We're currently stuck with a four-way mixed ANOVA to analyse some EEG data, but found out that JASP does not seem able to do post tests once factors have more than two levels. Unfortunately, this is often the case in EEG since at the most basic level you at least want to include multiple frequency bands. The analyses look in general like this:
Pathology (diagnosed, undiagnosed) * frequency band (delta, theta, alpha, beta, gamma) * location (frontal, central, posterior) * (median split on some behavioural measure that is hypothesized to influence the relationship).
We thought it might be possible to just do many many many separate analyses, because there's is no inflation of alpha errors in Bayesian so there is no need to correct for that (as most post tests do). However, this would be an aweful lot of work and we're not sure if this truly solves the problem in a proper way. Would we still be able to interpretate the results as if we had performed a four-way mixed ANOVA with post tests?
Following this all, what would be the best (Bayesian please) option for doing a multiple-way mixed/RM ANOVA when at least one factor has more than two levels?
Kristel de Groot