Katie Lane, Central Indiana Chapter Secretary, and Eric Sampson, ASA Journals Manager
It’s certainly no ordinary short course when the “Paul Shaffer of Bayesian statistics” is in the house and calling the tunes. Sure, Brad Carlin is Mayo Professor in Public Health and professor and head of the division of biostatistics at the University of Minnesota, but he is also a musician and band leader who brings a performer’s sense of timing to his courses. While the tempo can be fast, Carlin makes sure to hit every note along the way.
For members of the Central Indiana Chapter, the feature act was Carlin’s short course, “Bayesian Methods for Data Analysis, Meta-Analysis, and Adaptive Trials,” part of the ASA’s Council of Chapters Traveling Courses. The course was held May 17 at the Lilly Corporate Center in Indianapolis, Indiana.
“When the going gets rough, the rough go Bayesian,” Carlin said. “It’s not that the classical approach is wrong, but it is, let’s say, limited in scope!”
Biostatisticians in the drug and medical device fields are often faced with data that are highly multivariate, temporally correlated, costly, and difficult to obtain. Bayesian methods, however, are well suited to these types of data, as they help combine a lot of information in naturally hierarchical ways. Advances in computing and Markov chain Monte Carlo (MCMC) methods help researchers conduct safety and efficacy studies, equivalence studies, and meta-analyses. They also help create hierarchical models that account for all sources of uncertainty. In addition, regulatory agencies are starting to encourage hierarchical Bayesian approaches, making these methods an important tool in the analyst’s bag.
Carlin continued with a quick, but thorough, primer on Bayesian inference and computation, followed by several examples using R and WinBUGS to perform analyses.
The second half of the short course covered the basics of Bayesian clinical trial design, Bayesian adaptive methods (particularly important to statisticians working on clinical trials), and Bayesian meta-analysis.
Chapter members in attendance said the course was well organized and Carlin was a great speaker, keeping the audience engaged until the end. For those who have limited background in Bayesian statistics, Carlin provided a good contrast on the difference between frequentist and Bayesian philosophies.
Tuan Nguyen, president of the Central Indiana Chapter, said, “Carlin provided a great overview as well as a refreshing perspective on the theory and applications of Bayesian statistics with particular motivating examples in clinical trial research. It is encouraging to hear that the FDA is more open to the use of Bayesian statistical approaches. For applied statisticians, these short courses provide a good review of the rigorous training acquired in graduate school as well as novel applications pertinent to their work.”
Ed Brizendine (past president of the chapter) said, “These traveling courses offered by the ASA are a great value and service to the local chapters. They allow for chapters to host nationally known statisticians who speak on current and relevant topics.”
A number of chapter officers pitched in to make the short course a reality. In addition to Nguyen and Brizendine, Sameera Wijayawardana (chapter representative), Peipei Shi (treasurer), Katie Lane (secretary), Zhangsheng Yu (past president), and Huiping Xu (vice president) all assisted in organizing the course.
For more Bayesian-inspired tunes recorded by Carlin and his intrepid band, visit YouTube and watch the video “Bayesian Cabaret, What I Like About You”. Oh what a prior!
Leave a Reply