Daniel R. Jeske, The American Statistician Executive Editor
The May 2020 issue of The American Statistician (TAS) is now available online. Print copy subscribers will be receiving their issue when the publisher is able to resume from printing delays caused by the COVID-19 crisis.
The General section has two articles. The first discusses issues related to sports analytics research and offers guidance on selecting the right publication venue. The second article investigates the use of multiple imputation methods when computing a confidence interval for the binomial p with missing data and shows that the Wilson interval has better sampling properties than the Wald interval in this context.
The Statistical Practice section has three articles. The first revisits an old friend, Simpson’s Paradox. A new summary statistic is proposed that ameliorates the effects of the paradox. The second article looks into the general question of how many imputed data sets are needed when using imputation techniques to handle missing data. The answer might be more than you think. The final article in this section discusses Bayesian t-tests. The article provides an accessible gateway to Bayesian hypothesis testing and proposes a new Bayesian t-test that offers additional flexibility.
The Statistical Computing and Graphics section has two articles. The first article discusses approaches for testing the equality of covariance matrices and offers graphical methods for revealing how they might differ. The second presents an improved algorithm for plotting a likelihood ratio-based confidence region for general two-parameter families.
The Data Science section includes an article that provides an interesting introduction to archetype analysis, a data mining technique for representing instances of a data set as a mixture of archetypes (i.e., exemplars) constructed from training data. It proposes new methods for handling missing data in this type of analysis.
The Short Technical Note section has an article that discusses different parameterizations of time when designing a cluster randomized trial to be conducted over different time periods. The article shows that the choice of time parameterization does not affect the variance of the treatment effect estimator in many interesting cases.
The History Corner has two articles that discuss the role of statistical thinking in recognizing and fixing problems in the design of war draft lotteries. One of the articles focuses specifically on the draft lottery used by the US 50 years ago for the Vietnam War while the other looks broadly at 13 draft lotteries used by the US between 1917 and 1975.
The May issue concludes with a Letter to the Editor commenting on a previous TAS article about Phlegon’s use of stem and leaf displays.
The editorial board asks you to consider the journal as a potential outlet for your work. Learn more about submitting an article to TAS on the journal’s website.
Leave a Reply