Information Criteria, Bayesian Methods, and Their Implications for Statistical Literacy

• Activity Number: 436
• Type: Roundtables
• Date/Time: Wednesday, August 3, 2011 : 7:00 AM to 8:15 AM
• Sponsor: Section on Teaching of Statistics in the Health Sciences
• Author(s): Rochelle Elaine Tractenberg*+
• Companies: Georgetown University Medical Center
• Keywords: statistical literacy ; information criteria ; bayesian methods ; multi-model inference ; null hypothesis significance testing

Abstract

The use of information criteria (IC) or Bayesian methods (BM) for modeling, analysis, or decisionmaking, are treated as "advanced topics" and are at odds with null hypothesis significance testing (NHST). Can statistical literacy be achieved if NHST is the sole object of these literacy efforts? Options for integrating IC, BM, and NHST in statistical literacy curricula can reinforce current efforts, while not condemning the newly statistical literate to ignorance of these important advances in applied statistics, include: 1. pare down the NHST to accommodate incorporation of the other two topics. 2. choose IC or BM and integrate just that into the NHST training. 3. maintain the current emphasis on NHST in the development of statistical literacy, mentioning (leaving) IC and BM for advanced students. Considerations for revising/developing new statistical literacy programs include the preparation of NHST-oriented instructors to teach IC and BM, and the ramifications of limiting exposure to NHST and focusing on IC and/or BM. We will discuss these options, and others that participants may be contemplating for their own efforts to improve/achieve statistical literacy at their institutions.