Richard Morey - Advances in Bayesian hypothesis testing for general-purpose statistical models

Since its inception, null hypothesis significance testing (inference by p values) has become the dominant method by which statistical hypotheses are tested. However, it has not been without controversy. Both Bayesian and frequentist statisticians have leveled a number of criticisms against null hypothesis significance testing, but in spite of these criticisms, null hypothesis significance testing remains the main statistical inferential method used in the sciences. Part of the reason for this is a lack of alternatives. Recent advances in Bayesian statistical methods have led to alternative methods being viable for use in everyday research situations. I discuss these advances and their advantages, some critiques of their use, and present a software implementation.