Computing Bayes factors from robust tests
Hi all,
Since neither JASP nor the BayesFactor package include robust tests, I am considering to compute Yuen's robust t-test and robust multiple regression through a M-estimator and then enter the data of t and R-squared in the "summary stats" of JASP to compute the Bayes factors of the robust tests. Does this procedure makes sense?
Comments
I am not 100% sure.
The "real/royal" Bayesian way is to change the data-generating process, for instance by assuming that the data are t-distributed instead of normally distributed, or assuming . Some classical robust methods throw away data, and that is not really something that Bayesians like to do.
E.J.