Speaker: Alyssa Counsell, York University
Department of Psychology

Title: Bayesian versus Traditional Approaches for Testing Mean Equivalence

Abstract: Procedures such as the two-one sided tests (TOST) method have been popular for examining mean equivalence, as it is inappropriate to declare means equivalent through nonrejection of the traditional null hypothesis (Ho: μ1 = μ2). This study evaluated two Bayesian methods for equivalence testing of two group means; Kruschke's (2013) Bayesian estimation method compares a high density interval to the equivalence region, whereas Morey and Rouder's (2011) Bayes Factor method compares an interval-based null hypothesis to the complement of this hypothesis. The current simulation study compared the probabilities of concluding equivalence for the TOST and two Bayesian methods across several conditions. The presentation will discuss both empirical and theoretical differences between frequentist methods (equivalence tests) and Bayesian methods (two different procedures). Furthermore, I will discuss the differences between the two Bayesian approaches.

Suggested Readings:

Kruschke, J. K. (2013). Bayesian estimation supersedes the t -test. Journal of Experimental Psychology: General, 142, 573-603.

Morey, R. D., & Rouder, J. N. (2011). Bayes factor approaches for testing interval null hypotheses. Psychological Methods, 16, 406-419.

Schuirmann, D. J. (1987). A comparison of the two one-sided tests procedure and the power approach for assessing the equivalence of average bioavailability. Journal of   Pharmacokinetics and Biopharmaceutics, 15, 657-680.