David L. Banks, Len Stefanski, and Dalene Stangl

A Message from the Editor
David Banks

It is an unalloyed delight to report that the December issue of JASA is the last of my editorial duties. I have certainly enjoyed this job, but it is relentless and wears one down over time. I am deeply and completely grateful to the associate editors, whose unflagging service makes JASA so outstanding; to the insufficiently thanked referees, who do most of the hard work; and to the ASA staff, whose support has made this journal possible. It has been a wonderful ride, and I’m very glad it is over.

I’d like to thank the authors, too. Partly I thank them for the quality of their work, but even more for their professionalism and collegiality. Reviewing is an imperfect process, despite the best will in the world and a sincere desire on the part of all parties to find the good, rather than nit-pick the bad. It is an asymmetric relationship, and few authors feel entirely well used and fully appreciated. Despite these difficulties, I treasure the uniformly positive and constructive interactions I have been lucky to enjoy.

I have no concerns for the future of JASA; Hal Stern, Len Stefanski, and Dalene Stangl have it well in hand. In the words Stephen Vincent Benét attributes to Daniel Webster’s nameless interlocutor regarding the United States, JASA “stands as she stood, rock-bottomed and copper-sheathed.” It is one of the great journals in statistics and deserves its prominence in the history of our profession.

The Applications & Case Studies section starts with a discussion paper by Rob Scharpf, Håkon Tjelmeland, Giovanni Parmigiani, and Andrew Nobel titled “A Bayesian Model for Cross-Study Differential Gene Expression.” There are two discussions, one by Debashis Ghosh and Hyungwon Choi and the other by Xiaodan Fan and Jun Liu.

The section fills out with “Analysis of Multifactor Affine Yield Curve Models,” by Sid Chib and Bakhodir Ergashev; “Bayesian Calibration of Microsimulation Models,” by Caroline Rutter, Diana Miglioretti, and Jim Savarino; “Option-Pricing with Model-Guided Nonparametric Methods,” by Jianqing Fan and Loriano Mancini; “Semi-Parametric Efficient Estimation for Incomplete Longitudinal Binary Data with Application to Smoking Trends,” by Jamie Perin, John Preisser, and Paul Rathouz; and “Modeling and Inference for Measured Crystal Orientations and a Tractable Class of Symmetric Distributions for Rotations in 3 Dimensions,” by Melissa Bingham, Daniel Nordman, and Steve Vardeman.

Theory and Methods

Paul R. Rosenbaum and Jeffrey H. Silber lead the section with “Amplification of Sensitivity Analysis in Matched Observational Studies,” wherein they develop methods for assessing the impact of an unobserved covariate not controlled for matching. For Bayesians at heart, Y. Chung and D. B. Dunson use a probit stick-breaking process to develop a methodology for flexibly characterizing the relationship between a response and multiple predictors in “Nonparametric Bayes Conditional Distribution Modeling with Variable Selection.” Mr. Rogers knows you can learn a lot from your neighbors and so do R. V. Craiu, J. Rosenthal, and C. Yang. In “Learn from Thy Neighbor: Parallel-Chain and Regional Adaptive MCMC,” these authors address the problem of adapting MCMC samplers to multi-modal distribution.

From multiple modes to multiple haystacks, “Simultaneous Testing of Grouped Hypotheses: Finding Needles in Multiple Haystacks” by T. T. Cai and W. Sun develops a compound decision theoretic framework for testing grouped hypotheses and introduces an oracle procedure that minimizes the false nondiscovery rate subject to a constraint on the false discovery rate. The impact of dependence among multiple test statistics is the topic of “A Factor Model Approach to Multiple Testing Under Dependence” by C. Friguet, M. Kloareg, and D. Causeur. A. Gandy’s “Sequential Implementation of Monte Carlo Tests with Uniformly Bounded Resampling Risk” introduces an open-ended sequential algorithm for computing the p-value of a test using Monte Carlo simulation that guarantees the resampling risk is uniformly bounded by an arbitrarily small constant. D. Paindaveine studies two types of multivariate runs—an elliptical extension of spherical runs and a new notion of matrix-valued runs—in “On Multivariate Runs Tests for Randomness.” The Food and Drug Administration will soon issue guidance on multiple endpoints. When they do, Y. Liu and J. Hsu have the methods needed in “Testing for Efficacy in Primary and Secondary Endpoints by Partitioning Decision Paths.”