Not disputing the paper or the scientific method applied here, but how is "bayesian factor" different than p-values?
The fact that you can incorporate (and reflect on) a priori assumptions about the likelihood of one or the other model is another advantage imho.