Joshua M. Tebbs, The American Statistician Editor
The August 2021 issue of The American Statistician (TAS) is available online and features 13 articles across different sections of the journal. One of the benefits of ASA membership is free access to online issues of TAS.
The General section has nine articles. The lead article provides a gentle introduction to agent-based models for statisticians. After giving numerous applications where agent-based models are commonly applied, the authors describe implementation and ways statisticians and data scientists can contribute to research in this area. The second article introduces readers to “prediction tournaments” and presents an interesting paradox that can arise when comparing contestants.
The next three articles revisit various issues arising in hypothesis testing. The third defends hypothesis testing on the basis of Bayesian model checking and offers implications in both Bayesian and frequentist practice. The fourth article establishes an intriguing connection between the standard Bayes factor and the region of practical equivalence (ROPE). The fifth article develops an equivalence relationship between p-values and Bayesian posterior probabilities with one- and two-sided tests; this equivalence is then used to justify the use of p-values from a Bayesian perspective.
The sixth article considers the matching problem in causal inference and proposes an improved method to produce better effect estimates in terms of mean-squared error. The seventh considers the “Ockham Factor” in model selection, espousing both penalty and flexibility in a decision-theoretic framework. The eighth article considers the omnipresent issue of publication bias in the biomedical and social sciences and proposes strategies to adjust published estimates for this type of bias. Finally, the ninth article uses probability-generating functions to determine distributions of arrival times of random patterns in Bernoulli and Markov trials.
The Statistical Computing and Graphics section has one article, which considers the important problem of generating high-dimensional correlated binary data, focusing on exchangeable, decaying-product and 1-dependent correlations. The novelty of the proposed method is its flexibility and computational efficiency when compared to existing approaches.
There are two articles in the Statistical Practice section. The first gives a comprehensive summary of issues one might face when serving as an expert statistician in a legal dispute. The authors describe the ethics of such a role and provide examples from real cases. The second article revisits the replicability crisis and frames relevant questions from within an experimental design perspective.
Finally, the Teacher’s Corner section has one article, which describes hands-on approaches to illustrating spatial randomness for undergraduate students. The authors also propose analytical methods students can use to evaluate spatial point patterns.
Learn more about TAS.
Leave a Reply