Bayesian Methods Widespread in February Issue of TAS
The February 2021 issue of The American Statistician (TAS) is available online and features 13 articles and one letter to the editor. One of the benefits of ASA membership is free access to the online issues of TAS.
The General section has six articles. The first uses a known result in time series to revisit the common practice of estimating posterior expectations from MCMC samples. The second implements Bayes factors to test ordering constraints among multinomial parameters. The third presents an analysis of how players in Contestant’s Row should bid on the game show The Price Is Right. The fourth examines empirical likelihood ratio hypothesis tests for positive quadrant dependence between two random variables. The fifth proposes a framework to perform hypothesis testing in the presence of adversaries. Finally, the sixth article describes a Bayesian approach to make inference on the correlation coefficient under divergence-based priors.
The Statistical Computing and Graphics section has two articles. The first article discusses sampling strategies for fast updating of Gaussian Markov random fields. The authors describe how one can sample from target distributions more efficiently when compared to existing methods. Their approach is illustrated by using data sets in imaging analysis and election voting. The second describes an R Shiny app to plan a Bayesian two-stage, “drop-the-losers” design in clinical trials. The authors describe this design and then illustrate the app by using data from a weight-loss randomized trial.
There are three articles in the Statistical Practice section. The first presents an “assumption lean” framework for regression analysis and analyzes criminal offense data using a Poisson generalized linear model. The second deals with time-dependent bias that arises in survival analysis and presents a mortality study of politicians in Germany, the UK, and the USA. The third article also presents a Poisson regression analysis, but to model the response time to structure fires while incorporating spatial information.
The Teacher’s Corner section has one article that presents a probabilistic approach to find higher-order moments of the binomial distribution. In doing so, the author describes new ways to calculate finite sums of integers raised to common exponents.
Finally, the Short Technical Note section has one article. This article revisits the p-value discussion but from a Bayesian point of view in the context of performing meta-analyses.
The February issue concludes with a Letter to the Editor, which clarifies certain claims made in a hypothesis testing article published last year.