Title: Why Bayesian Models Tests of Different Dimensionality Are Even Worse
Than Null Hypothesis Significance Testing

Authors: Timo von Oertzen, Angelika Stefan

Abstract:
In this presentation, we present some general maxims one may have when testing
or analyzing parameters of a model. We review why the general idea of
comparing a parameter space to a restricted version of itself with less
dimensions (e.g., fixing a parameter to a point) is not a valid hypothesis. We
specifically demonstrate in a simulation that Bayesian Model testing does not
solve the problem but, in comparison to Null Hypothesis Significance Testing,
even makes it worse.