Journal of the American Statistical Association

Journal covers statistical science and its applications, theory, and methods in economic, social, physical, engineering and health sciences.

Articles from Vol. 89, No. 427, September

A Brownian Motion Model for the Progress of Sports Scores
The difference between the home and visiting teams' scores in a sports contest is modeled as a Brownian motion process defined on t [element of] (0, 1), with drift [mu] points in favor of the home team and variance [sigma.sup.2]. The model obtains a...
A Case Study of an Adaptive Clinical Trial in the Treatment of Out-Patients with Depressive Disorder
1. INTRODUCTION Randomized clinical trials play a pivotal role in experiments to determine which of a number of treatments is superior. In most clinical trials, patients enter the trial sequentially over time, and their responses can be monitored....
Adaptive Mixtures
1. INTRODUCTION AND SUMMARY This article discusses nonparametric, or distribution-free, maximum likelihood density estimators. Modem engineering practice has exploited nonparametric density estimates in a wide variety of settings. One application...
Adjusting for Differential Rates of Prophylaxis Therapy for PCP in High- versus Low-Dose AZT Treatment Arms in an AIDS Randomized Trial
1. INTRODUCTION The AIDS Clinical Trials Group (ACTG) study number 002 compared the effect of high-dose AZT (3-azido-3-deoxythymidine) therapy (1,500 mg/day) vs. low-dose AZT therapy (1,200 mg/ day for 4 weeks and 600 mg/day thereafter) on the...
Analysis of Olympic Heptathlon Data
An incremental clustering algorithm is described and applied to 1992 olympic heptathlon data to produce characterizations of groups of the leading athletes. Results are compared with those obtained by classical clustering techniques. The incremental...
A Random-Effects Probit Model for Predicting Medical Malpractice Claims
1. INTRODUCTION The continuing effort to create a reasonable and equitable solution to the problem of medical malpractice litigation has remained unsuccessful for almost an entire generation. The relative failure of the tort liability system...
Assessing Secular Trends in Blood Pressure: A Multiple-Imputation Approach
1. INTRODUCTION The U.S. National Center for Health Statistics (NCHS) conducts periodic surveys of the health and nutrition status of the nation. Public-use data tapes have been made available from the first three such surveys: the National Health...
A Stochastic Model for Analysis of Longitudinal AIDS Data
1. INTRODUCTION The concept of tracking for longitudinal data is popular with epidemiologists and has been discussed in the statistical literature. Foulkes and Davis (1981) and McMahan (1981) developed indices of tracking based on the concept...
A Two-Stage Adaptive Group-Testing Procedure for Estimating Small Proportions
1. INTRODUCTION Group testing has long been recognized as a sampling scheme that can provide substantial benefits. Under this sampling scheme, rather than obtaining measurements on individuals, simultaneous measurement is obtained for a group...
Bayes Optimal Designs for Two- and Three-Level Factorial Experiments
1. INTRODUCTION Many scientific investigations study processes in which the exact nature of the relationship between the process outcome and a number of explanatory variables is not known. For such situations, a collection of statistical techniques...
Bootstrap Recycling: A Monte Carlo Alternative to the Nested Bootstrap
1. INTRODUCTION The nested bootstrap algorithm is a fundamental method of computation in various frequentist inference procedures. Bootstrap iteration (Hall 1992; Hall and Martin 1988), prepivoting (Beran 1987, 1988), calibration (Loh 1987),...
Computable Robust Estimation of Multivariate Location and Shape in High Dimension Using Compound Estimators
1. INTRODUCTION Active research continues into methods of robust estimation of multivariate location and shape (Campbell 1980, 1982; Davies 1987; Devlin, Gnanadesikan, and Kettenring 1981; Donoho 1982; Hampel, Ronchetti, Rousseeuw, and Stahel...
Cox Regression in a Markov Renewal Model: An Application to the Analysis of Bone Marrow Transplant Data
1. INTRODUCTION We consider estimation and prediction in a Cox proportional hazard model where observations on each individual under study form a Markov renewal process with a finite state space, say {1, . . . , r}. In the absence of covariates,...
Cross-Validation of Multivariate Densities
1. INTRODUCTION Most work in data-driven methods for kernel density estimation has focused on the univariate case, while multivariate kernel density estimation has been somewhat neglected. Stone (1984) showed the strong theoretical result that...
Data-Driven Version of Neyman's Smooth Test of Fit
1. INTRODUCTION Neyman introduced the smooth goodness-of-fit test for testing uniformity in 1937. Rayner and Best (1990), in reviewing several tests of fit, including those of Neyman (1937), D'Agostino and Stephens (1986), and Rayner and Best...
Distribution Theory: A Markov Chain Approach
The statistics of the number of success runs in a sequence of Bernoulli trials have been used in many statistical areas. For almost a century, even in the simplest case of independent and identically distributed Bernoulli trials, the exact distributions...
Down to Ten: Estimating the Effect of a Red Card in Soccer
1. INTRODUCTION Professional soccer (known outside the United States as football) is popular all over the world; in Europe and South America it is the dominant spectator sport. Because soccer is a low scoring game, the rules have been often revised...
Efficient Semiparametric Estimation in a Stochastic Frontier Model
1. INTRODUCTION The model that we analyze in this article assumes N independent observations (X.sub.i], [Y.sub.i]), which are written as (1) [Y.sub.it] = [X'.sub.it][beta] + [[alpha].sub.it] + [[epsilon].sub.it], t = 1, . . . , T, where [X.sub.i]...
Estimation of Regression Coefficients When Some Regressors Are Not Always Observed
1. INTRODUCTION In applied problems it is common to specify a model [Mathematical Expression Omitted] for the conditional mean of the response [Y.sub.i] of a subject i given a set of regressors [Mathematical Expression Omitted], where a is an...
Estimation with Selected Binomial Information or Do You Really Believe That Dave Winfield Is Batting .471?
1. INTRODUCTION Sports announcers--in particular, baseball announcers--often use hyperbolical descriptions of a plaver's ability. For example, when Dave Winfield, a popular baseball player, is batting, rather than report his current batting average...
Exploring Baseball Hitting Data: What about Those Breakdown Statistics?
1. INTRODUCTION After the end of every baseball season, books are published that give detailed statistical summaries of the batting and pitching performances of all major league players. In this article we analyze baseball hitting data that was...
Importance-Weighted Marginal Bayesian Posterior Density Estimation
Markov chain sampling schemes generate dependent observations [Mathematical Expression Omitted] from a full joint posterior distribution [pie]([teta]\ data). Frequently, only certain marginals of this full posterior density are of interest; thus an interesting...
Inference for Shift Functions in the Two-Sample Problem with Right-Censored Data: With Applications
1. INTRODUCTION Suppose that F and G denote the distribution functions of random variables X and Y. For the distributions F and G, the horizontal shift (or translation) function is defined by (1) [delta](t) [equivalent] [G.sup.-1] [multiplied...
Logit and Multilevel Logit Modeling of College Graduation for 1984-1985 Freshman Student-Athletes
1. INTRODUCTION Academic success in college has become an important indicator of the American educational system. Acceptance or rejection to a particular college can have effects on the individual and the school for many years to come (Manski...
Measuring Change in Latent Subgroups Using Dichotomous Data: Unconditional, Conditional, and Semiparametric Maximum Likelihood Estimation
1. THE LINEAR LOGISTIC MODEL WITH RELAXED ASSUMPTIONS Statements about changes in qualitative data are desired in many situations. For example, a politician may wish to know. the effect of a media campaign on the attitude towards a socially discriminated...
On Quasi-Independence and Quasi-Dependence in Contingency Tables, with Special Reference to Ordinal Triangular Contingency Tables
1. INTRODUCTION The concept of quasi-independence (QI) in an I X J contingency table was introduced for the analysis of tables in which the entries (frequencies) in some of the cells of the table are omitted from the analysis because these entries...
Optimal Recursive Estimation of Dynamic Models
1. INTRODUCTION In time series analysis, system identification, and econometrics, there is a growing interest in statistical methods, for nonstationary and nonlinear stochastic processes. Realizations of such processes, in the form of finite...
P Values Maximized over a Confidence Set for the Nuisance Parameter
1. INTRODUCTION Testing problems are often complicated by the presence of a nuisance parameter vector [theta]. Consider first a model in which there is no nuisance parameter. Suppose that the data X have a probability distribution [P.sub.[nu]],...
Rating Skating
1. INTRODUCTION Early during the 1992 Winter Olympics, Scott Hamilton, the former Olympic champion now working as an announcer for CBS, made a valiant but unsuccessful try to explain how judges' marks are aggregated to determine the placements...
Rationality and Unbiasedness in Hypothesis Testing
1. INTRODUCTION The purpose of this article is to encourage discussion of the concepts of "unbiasedness" and what we shall call "rationality" as they relate to testing statistical hypotheses when one has incomplete data. Censored data provides...
Robust Bounded-Influence Tests in General Parametric Models
1. INTRODUCTION In the last decade, the amount of statistical research devoted to robustness has increased considerably. But most of the research effort has focused on robust estimation, especially in linear models, and far less attention has...
Robust Estimation in the Analysis of Complex Molecular Spectra
1. INTRODUCTION Insight into the physical structure of atoms and molecules can be obtained by assessing the pattern of energy levels associated with their quantum mechanical states. Experiments associated with this topic form the scope of atomic...
Saddlepoint Approximation for the Distribution of a Ratio of Quadratic Forms in Normal Variables
In this article, the saddlepoint approximations to the density and tail probability of a ratio of quadratic forms in normal variables are derived. A numerical exposition via the Durbin-Watson test statistic reveals several desirable features. The approximations,...
Screening with Cost-Effective Quality Control: Potential Applications to HIV and Drug Testing
1. INTRODUCTION Group testing, introduced by Dorfman (1943), is now used in a variety of fields (Chen and Swallow 1990) and has been used to screen blood donors in less-developed countries (Emmanuel, Bassett, Smith, and Jacobs 1988; Monzon et...
Self-Validating Computations of Probabilities for Selected Central and Noncentral Univariate Probability Functions
1. INTRODUCTION Self-validating numerical computation, sometimes called automatic error analysis, can be achieved in more than one way. We will use interval analysis to accomplish self-validating computation of probabilities and percentiles....
Some Tests for Comparing Cumulative Incidence Functions and Cause-Specific Hazard Rates
1. INTRODUCTION In the competing risks model, a unit is exposed to several risks at the same time, but it is assumed that the eventual failure of the unit is due to only one of these risks, which is called a "cause of failure." Let a unit be...
Statistics in Sports
Reporting and analysis of sports and sporting activities involves the use of quantitative reasoning by more Americans than in any other aspect of everyday life. Batting averages, team standings, and world records are concepts in such common use that...
The Collapsed Gibbs Sampler in Bayesian Computations with Applications to a Gene Regulation Problem
This article describes a method of "grouping" and "collapsing" in using the Gibbs sampler and proves from an operator theory viewpoint that the method is in general beneficial. The norms of the forward operators associated with the corresponding nonreversible...
The Construction of New Bivariate Exponential Distributions from a Bayesian Perspective
1. INTRODUCTION The exponential distribution is frequently used as a component lifetime distribution in reliability theory because of its distinctive loss-of-memory property or constant failure rate. In other words, the exponential distribution...
Toward a Reconciliation of the Bayesian and Frequentist Approaches to Point Estimation
1. INTRODUCTION The debate between proponents of the Bayesian and of the frequentist approaches to estimation is, in our view, in an inconclusive state. Despite the many worthy arguments one finds in the literature favoring one approach or criticizing...
Transform-Both-Sides Approach for Overdispersed Binomial Data When N Is Unobserved
1. INTRODUCTION 1.1 Overdispersion Due to Variability of N For binomial outcomes, [Y.sub.i], i = 1, 2,. . ., n with parameters [N.sub.i] and [P.sub.i] the observed variation of the response frequently exceeds the nominal variance determined...
Trying out for the Team: Do Exhibitions Matter? Evidence from the National Football League
1. INTRODUCTION In a recent essay in the sports weekly Sports Illustrated, Rick Telander argued that the National Football League (NFL) should eliminate its preseason schedule because the games provide no useful services and the sole purpose...
Uniformly More Powerful Tests in a One-Sided Multivariate Problem
1. INTRODUCTION Numerous articles have reported on anomalies of certain normal likelihood ratio (LR) tests in the sense that uniformly more powerful tests are found (Berger 1989; Gutmann 1987; Menendez and Salvador 1991; Menendez, Rueda, and...
Use of Cumulative Sums of Squares for Retrospective Detection of Changes of Variance
1. INTRODUCTION We study the detection of multiple changes of variance in a sequence of independent observations. There are series, particularly in the area of finance, that do not follow the usual assumption of constant variance underlying most...
Use of Estimating Functions for Estimation from Complex Surveys
1. INTRODUCTION Point and interval estimation of population parameters is a cornerstone of modern statistical theory. For parametric infinite populations, these parameters completely describe the underlying distribution. In finite populations,...
Versions of Kernel-Type Regression Estimators
1. INTRODUCTION Given data ([t.sub.i], [Y.sub.i]), i = 1, 2,..., n, there exists a host of different formulas going by the name of the kernel estimator, m, of the regression mean function, m. With K the kernel function that we take to be a probability...