ABSTRACT
The conceptual difference between null hypothesis testing and the Bayesian alternative is that predictions about mean differences are stated a priori in a hierarchy of differences as motivated by theory-driven claims. Bayesian confirmatory analysis of variance provides a robust alternative to conventional ANOVA. The BMS tool used is a new analysis technique in language testing. As framework-based test development gains momentum around the world, it examines the evidence that tests can be written to frameworks such as the Common European Framework of Reference (CEFR) and those developed by the Interagency Language Roundtable (ILR) and American Council on the Teaching of Foreign Languages (ACTFL). The Bayesian approach may offer a cost-effective intermediate step between test development and expensive and subjective standard setting methods. The confirmatory approach outlined in this chapter is likely to provide a useful analytic tool for testing framework-generated hypotheses about language proficiency, as well as a viable alternative to null hypothesis testing of group mean differences in general.
