It seems surprising that when applying widely used random number generators to generate one million random numbers on modern architectures, one obtains, on average, about 116 collisions. This article explains why, how to mathematically compute such a number, why they often cannot be obtained in a straightforward way, how to numerically compute them in a robust way and, among other things, what would

We present a method of generating functions to compute the distributions of the first-arrival and inter-arrival times of random patterns in independent Bernoulli trials and first-order Markov trials. We use segmentation of pattern events and diagrams of Markov chains to illustrate the recursive structures represented by generating functions. We then relate the results of pattern time to the probability

Publication bias can occur for many reasons, including the perceived need to present statistically significant results. We propose and compare methods for adjusting a single published estimate for possible publication bias using a truncated normal distribution. We attempt to estimate the mean of the underlying normal sampling distribution using only summary data readily available in most published

Binge-watching behavior is modeled for a single season of an original program from a streaming service in order to understand and make predictions about how individuals watch newly released content. Viewers make two choices in binge watching. First, the onset when individuals begin viewing the program is modeled using a change point between epidemic viewing with a non-constant hazard rate and endemic

Micha Mandel (2020), “The Scaled Uniform Model Revisited”, The American Statistician, 74:1, 98-100: Comment by Taraldsen. The American Statistician. Accepted 10 May 2020.

We explore the arguments for maximizing the ‘evidence’ as an algorithm for model selection. We show, using a new definition of model complexity which we term ‘flexibility’, that maximizing the evidence should appeal to both Bayesian and Frequentist statisticians. This is due to flexibility’s unique position in the exact decomposition of log-evidence into log-fit minus flexibility. In the Gaussian linear

The (bivariate) Poisson distribution is the most common distribution for (bivariate) count random variables. The univariate Poisson distribution is characterized by the famous Stein–Chen identity. We demonstrate that this identity allows to derive even sophisticated moment expressions in such a simple way that the corresponding computations can be presented in an introductory Statistics class. Then

In the Anglo-American legal system, courts rely heavily on experts who perform an essential social function in supplying information to resolve disputes. Experts are the vehicles through which facts of any technical complexity are brought out. The adversarial nature of this legal system places expert witnesses in a quandary. Enjoined to serve the court and their profession with unbiased, independent

Comment on Szabo (2019).

(2020). R Markdown: The Definitive Guide. The American Statistician: Vol. 74, No. 2, pp. 209-210.

(2020). Model-Based Clustering and Classification for Data Science: With Applications in R. The American Statistician: Vol. 74, No. 2, pp. 208-209.

(2020). Capture-Recapture Methods for the Social and Medical Sciences. The American Statistician: Vol. 74, No. 2, pp. 207-208.

(2020). The Art of Statistics: How to Learn From Data. The American Statistician: Vol. 74, No. 2, pp. 207-207.

(2020). Benjamin, D. J., and Berger, J. O. (2019), “Three Recommendations for Improving the Use of p-Values”, The American Statistician, 73, 186–191: Comment by Foulley. The American Statistician: Vol. 74, No. 1, pp. 101-102.

Sufficiency, conditionality, and invariance are basic principles of statistical inference. Current mathematical statistics courses do not devote much teaching time to these classical principles, and even ignore the latter two, in order to teach modern methods. However, being the philosophical cornerstones of statistical inference, a minimal understanding of these principles should be part of any curriculum

The study of conditions under which a two-dimensional random variable (X, Y) will have the property that X/(X+Y)=dY/(X+Y) was initiated by Bhattacharjee and Dhar. Some additional perhaps unexpected examples related to this phenomenon are provided. Discrete and absolutely continuous cases are discussed in detail. Singular continuous cases are briefly mentioned.

Games can be a powerful tool for learning about statistical methodology. Effective game design involves a fine balance between caricature and realism, to simultaneously illustrate salient concepts in a controlled setting and serve as a testament to real-world applicability. Striking that balance is particularly challenging in response surface and design domains, where real-world scenarios often play

In a test of significance, it is common practice to report the p-value as one way of summarizing the incompatibility between a set of data and a proposed model for the data constructed under a set of assumptions together with a null hypothesis. However, the p-value does have some flaws, one being in general its definition for two-sided tests and a related serious logical one of incoherence, in its

This article explores models for geostatistical data for situations in which the region where the phenomenon of interest varies is partitioned into two disjoint subregions. This is called a binary map. The goals of the article are 3-fold. First, a review is provided of the classes of models that have been proposed so far in the literature for geostatistical binary data as well as a description of their

(2020). Comment on “A Note on Collinearity Diagnostics and Centering” by Velilla (2018) The American Statistician: Vol. 74, No. 1, pp. 68-71.

The article revisits univariate and multivariate linear regression models. It is shown that least-square estimators (LSEs) are minimum risk estimators in general class of linear unbiased estimators under some general divergence loss. This amounts to the loss robustness of LSEs.

Online consumer product ratings data are increasing rapidly. While most of the current graphical displays mainly represent the average ratings, Ho and Quinn proposed an easily interpretable graphical display based on an ordinal item response theory (IRT) model, which successfully accounts for systematic interrater differences. Conventionally, the discrimination parameters in IRT models are constrained

The idea of transforming one random variate to another with a more convenient density has been developed in the first half of the 20th century. In his thesis, Norman L. Johnson (1917–2004) developed a pioneering system of transformations of the standard normal distribution which gained substantial popularity in the second half of the 20th century and beyond. In Johnson’s 1949 Johnson, N. L. (1949)

The correlation coefficient (CC) is a standard measure of a possible linear association between two continuous random variables. The CC plays a significant role in many scientific disciplines. For a bivariate normal distribution, there are many types of confidence intervals for the CC, such as z-transformation and maximum likelihood-based intervals. However, when the underlying bivariate distribution

This article describes a new algorithm for generating correlation matrices with specified eigenvalues. The algorithm uses the method of alternating projections (MAP) that was first described by Neumann. The MAP algorithm for generating correlation matrices is both easy to understand and to program in higher-level computer languages, making this method accessible to applied researchers with no formal

Although there is a significant literature on the asymptotic theory of Bayes factor, the set-ups considered are usually specialized and often involves independent and identically distributed data. Even in such specialized cases, mostly weak consistency results are available. In this article, for the first time ever, we derive the almost sure convergence theory of Bayes factor in the general set-up

ABSTRACT Data Science is one of the newest interdisciplinary areas. It is transforming our lives unexpectedly fast. This transformation is also happening in our learning styles and practicing habits. We advocate an approach to data science training that uses several types of computational tools, including R, bash, awk, regular expressions, SQL, and XPath, often used in tandem. We discuss ways for undergraduate

Abstract Over the last three decades, data have become ubiquitous and cheap. This transition has accelerated over the last five years and training in statistics, machine learning, and data analysis has struggled to keep up. In April 2014, we launched a program of nine courses, the Johns Hopkins Data Science Specialization, which has now had more than 4 million enrollments over the past five years.

For several decades, the resampling based bootstrap has been widely used for computing confidence intervals (CIs) for applications where no exact method is available. However, there are many applications where the resampling bootstrap method cannot be used. These include situations where the data are heavily censored due to the success response being a rare event, situations where there is insufficient

Abstract This study describes a novel method to reweight a comparison group used for causal inference, so the group is similar to a treatment group on observable characteristics yet avoids highly variable weights that would limit statistical power. The proposed method generalizes the covariate-balancing propensity score (CBPS) methodology developed by Imai and Ratkovic (2014) to enable researchers

The test of Yang and Prentice, based on the short-term and long-term hazard ratio model for the presence of a regression effect appears to be an attractive one, being able to detect departures from a null hypothesis of no effect against quite broad alternatives. We recall the model on which this test is based and the test itself. In simulations, the test has shown good performance and is judged to

As a convention, p-value is often computed in frequentist hypothesis testing and compared with the nominal significance level of 0.05 to determine whether or not to reject the null hypothesis. The smaller the p-value, the more significant the statistical test. Under noninformative prior distributions, we establish the equivalence relationship between the p-value and Bayesian posterior probability of

Abstract Estimators based on influence functions (IFs) have been shown to be effective in many settings, especially when combined with machine learning techniques. By focusing on estimating a specific target of interest (e.g., the average effect of a treatment), rather than on estimating the full underlying data generating distribution, IF-based estimators are often able to achieve asymptotically optimal

Previously published statistical analyses of NCAA Division I Men’s Basketball Tournament (“March Madness”) game outcomes since the 64-team format for its main draw began in 1985 have uncovered some apparent anomalies, such as 12-seeds upsetting 5-seeds more often than might be expected, and seeds 10 through 12 advancing to the Sweet Sixteen much more often than 8-seeds and 9-seeds—the so-called middle-seed

There is a long debate in experimental design between the classic randomization design of Fisher, Yates, Kempthorne, Cochran, and those who advocate deterministic assignments based on notions of optimality. In nonsequential trials comparing treatment and control, covariate measurements for each subject are known in advance, and subjects can be divided into two groups based on a criterion of imbalance

(2020). Did Phlegon Actually Use a Stem-and-Leaf Display? The American Statistician: Vol. 74, No. 2, pp. 211-211.

Abstract In statistical practice, researchers commonly focus on patterns in the means of multiple dependent outcomes while treating variances as nuisance parameters. However, in fact, there are often substantive reasons to expect certain patterns in the variances of dependent outcomes as well. For example, in a repeated measures study, one may expect the variance of the outcome to increase over time

(2020). Correction. The American Statistician: Vol. 74, No. 2, pp. 212-212.

Quetelet’s data on Scottish chest girths are analyzed with eight normality tests. In contrast to Quetelet’s conclusion that the data are fit well by what is now known as the normal distribution, six of eight normality tests provide strong evidence that the chest circumferences are not normally distributed. Using corrected chest circumferences from Stigler, the χ2 test no longer provides strong evidence

Data for 30-day readmission rates in American hospitals often show that patients that receive Home Health Care (HHC) have a higher probability of being readmitted to hospital than those that did not receive such services, but it is expected that when control variables are included in a regression we will obtain a “sign reversal” of the treatment effect. We map the real-world situation to the binary

(2020). Connecting and Contrasting the Bayes Factor and a Modified ROPE Procedure for Testing Interval Null Hypotheses. The American Statistician. Ahead of Print.

Many modern experiments, such as microarray gene expression and genome-wide association studies, present the problem of estimating a large number of parallel effects. Bayesian inference is a popular approach for analyzing such data by modeling the large number of unknown parameters as random effects from a common prior distribution. However, misspecification of the prior distribution can lead to erroneous

The United States held 13 draft lotteries between 1917 and 1975, and a contingency procedure is in place for a selective service lottery were there ever to be a return to the draft. In 11 of these instances, the selection procedures spread the risk/harm evenhandedly. In two, whose anniversaries approach, the lotteries were problematic. Fortunately, one (1940) employed a “doubly robust” selection scheme

We examine the discrete distributional form that arises from the “classical occupancy problem,” which looks at the behavior of the number of occupied bins when we allocate a given number of balls uniformly at random to a given number of bins. We review the mass function and moments of the classical occupancy distribution and derive exact and asymptotic results for the mean, variance, skewness and kurtosis

Significance testing is often criticized because p-values can be low even though posterior probabilities of the null hypothesis are not low according to some Bayesian models. Those models, however, would assign low prior probabilities to the observation that the p-value is sufficiently low. That conflict between the models and the data may indicate that the models needs revision. Indeed, if the p-value

“Big data” is a buzzword these days due to an enormous amount of data-rich applications in different industries and research projects. In practice, big data often take the form of data streams in the sense that new batches of data keep being collected over time. One fundamental research problem when analyzing big data in a given application is to monitor the underlying sequential process of the observed

It is important to reduce fire department response times to incidents to improve communities’ general safety, to make the allocation of emergency resources more efficient, and to improve situational awareness. In this article, we identify which factors affect turnout times and travel times for the Arlington County Fire Department in Virginia by applying both linear and spatial models to the U.S. National

We revisit the classical problem of testing whether a normal mean is zero against all possible alternatives within a Bayesian framework. Jeffreys showed that the Bayes factor for this problem has a drawback with normal priors for the alternatives. He showed also that this deficiency is rectified when one uses a Cauchy prior instead. Noting that a Cauchy prior is an example of a scale-mixed normal prior

Abstract The problem of dividing an estate among creditors, when their claims total more than the value of the estate, was posed in the Talmud and has been analyzed in the game theory literature. Here, we reveal a close connection between schemes for estate division and linear regression solution paths obtained by least angle regression or by the lasso. We focus primarily on the division scheme known

Several scientific fields including psychology are undergoing a replication crisis. There are many reasons for this problem, one of which is a misuse of p-values. There are several alternatives to p-values, and in this article we describe a complement that is geared toward replication. In particular, we focus on confidence intervals for the probability that a parameter estimate will exceed a specified

Abstract Scholars have documented the failed randomization in 1969’s inaugural Vietnam Selective Service Lottery, but the story of how statisticians fixed that problem remains untold. Here, as the 50th anniversary of these events approaches, we recount how John W. Tukey, a team of statistical luminaries, and a graduate student from the University of Chicago repaired the draft lottery.

The correlation coefficient is a commonly used criterion to measure the strength of a linear relationship between the two quantitative variables. For a bivariate normal distribution, numerous procedures have been proposed for testing a precise null hypothesis of the correlation coefficient, whereas the construction of flexible procedures for testing a set of (multiple) precise and/or interval hypotheses

The "New Statistics" emphasizes effect sizes, confidence intervals, meta-analysis, and the use of Open Science practices. We present 3 specific ways in which a New Statistics approach can help improve scientific practice: by reducing over-confidence in small samples, by reducing confirmation bias, and by fostering more cautious judgments of consistency. We illustrate these points through consideration

Bayesian model selection provides a natural alternative to classical hypothesis testing based on p-values. While many articles mention that Bayesian model selection can be sensitive to prior specification on parameters, there are few practical strategies to assess and report this sensitivity. This article has two goals. First, we aim to educate the broader statistical community about the extent of

As statisticians and scientists consider a world beyond p < 0.05, it is important to not lose sight of how we got to this point. Although significance testing and p-values are often presented as prescriptive procedures, they came about through a process of refinement and extension to other disciplines. Ronald A. Fisher and his contemporaries formalized these methods in the early twentieth century and

Stepwise regression building procedures are commonly used applied statistical tools, despite their well-known drawbacks. While many of their limitations have been widely discussed in the literature, other aspects of the use of individual statistical fit measures, especially in high-dimensional stepwise regression settings, have not. Giving primacy to individual fit, as is done with p-values and R2

This article examines the evidence contained in t statistics that are marginally significant in 5% tests. The bases for evaluating evidence are likelihood ratios and integrated likelihood ratios, computed under a variety of assumptions regarding the alternative hypotheses in null hypothesis significance tests. Likelihood ratios and integrated likelihood ratios provide a useful measure of the evidence

This article resulted from our participation in the session on the "role of expert opinion and judgment in statistical inference" at the October 2017 ASA Symposium on Statistical Inference. We present a strong, unified statement on roles of expert judgment in statistics with processes for obtaining input, whether from a Bayesian or frequentist perspective. Topics include the role of subjectivity in

The score statistic continues to be a fundamental tool for statistical inference. In the analysis of data from high-throughput genomic assays, inference on the basis of the score usually enjoys greater stability, considerably higher computational efficiency, and lends itself more readily to the use of resampling methods than the asymptotically equivalent Wald or likelihood ratio tests. The score function

Many Bayes factors have been proposed for comparing population means in two-sample (independent samples) studies. Recently, Wang and Liu (2015) presented an "objective" Bayes factor (BF) as an alternative to a "subjective" one presented by Gönen et al. (2005). Their report was evidently intended to show the superiority of their BF based on "undesirable behavior" of the latter. A wonderful aspect of