Let me start with a piece of particularly good news. Thanks to the immense effort and enthusiasm of Professor Vincent Goulet from Université Laval, The Canadian Journal of Statistics now has a brand new, modern LaTeX template which provides full support for both English and French. The class cjs-rcs-article is now available on CTAN and has its own home page, designed by Vincent:

https://cjs-rcs.gitlab.io/

You will surely enjoy the ability of the new class to move seamlessly between the final look, an anonymized version for review, and a draft without the journal details that is suitable for a repository or your own personal use. If you are planning on attending the SSC annual meeting, don’t miss Vincent’s presentation about the class to learn all about its numerous features that can make your article shine and the typesetting process much more fun.

On behalf of the CJS editorial team, let me extend my deepest thanks to Vincent for designing, programming, documenting, and maintaining the new class, and to my editorial assistant Julie Falkner for her tireless and constant support throughout the process and much helpful input. The efforts of Wiley, notably their production team led by Julio Espin, to accommodate and transition to the new template are also gratefully acknowledged and much appreciated, as is funding from the SSC without which this entire endeavour would have been impossible.

At the end of the year, I also wish to express my sincere gratitude to the current and former managing editors, Llwellyn Armstrong and Bouchra Nasri, my editorial assistant Julie, and all authors, associate editors, referees, and copyeditors for their valuable contributions to CJS. I am likewise indebted to Belkacem Abdous for translating the abstracts, and to my husband Christian Genest for additional assistance in French. This generous collective effort is essential to the success of our journal. Please consider supporting it further, by choosing it as a venue for your work and by citing recently published articles whenever meaningful.

And now for something completely different: let me introduce the December 2023 issue. It comprises 14 research articles and is already available online on the journal’s website.

The issue opens with two contributions to network analysis and classification, respectively. Fan, Jiang, Yan, and Zhang [1] target degree heterogeneity in affiliation networks. They propose a class of bipartite graph models, consisting of nodes of actors and events connected by edges representing affiliation relationships, and develop moment-based inference for them. Jiao, Frostig, and Ombao [2] try to identify features that discriminate between neurological conditions on which local field potentials were recorded in the brain. As these potentials have zero mean, a variation pattern functional classifier is developed that utilizes the second moment structure and can adaptively identify the discriminative feature functions that account for the major differences.

The next pair of articles addresses specific challenges in generalized regression models. Shen and Wu [3] develop automatic structure recovery for generalized additive models, by extending Wu and Stefanski’s approach for additive models. Their method can identify predictors that have nonlinear effects or polynomial effects up to a given degree. Hossain, Mandal, and Lac [4] improve inference for generalized partially linear models in that they use inactive covariates as auxiliary information and form pretest and Stein-type shrinkage estimators.

Four articles are devoted to various testing problems. Huang, Liu, Zhou, and Feng [5] study two-sample tests for location in data whose dimension can be much larger than the sample size and find a locally most powerful weighted spatial sign test whose weight is the inverse norm function. Two-sample tests for multivariate data are also at the heart of the work of Wang, Fan, and Wang [6], who focus on complex settings in which the conventional Euclidean or spherical spaces are not sufficient. Their proposal uses a hyperbolic divergence rooted in hyperbolic geometry. Chen, Gong, and Wang [7] develop joint tests for the drift and volatility functions in diffusion models. Founded on empirical processes, their Kolmogorov-Smirnov and Cramér-von Mises type procedures can account for different convergence rates of the drift and volatility estimators. Derumigny, Fermanian, and Min [8] consider the problem of testing whether the copula of a conditional distribution is affected by the conditioning event(s). This is akin to assessing the validity of the so-called simplifying assumption that underlies the popular vine copula models. Here, the authors condition on the random vector being in a specific set and develop tests of the hypothesis that conditional Kendall correlations are constant over these sets.

Copulas are also inherent to the work of Koike and Hofert [9] who characterize all transformations of random variables such that the resulting Pearson’s correlation is a measure of concordance. They are then led to study matrices of pair-wise generalized Gini’s gamma.

Several articles contribute to lifetime and medical data analysis. To handle the challenge of ranking left- or right-censored data for the purpose of rank-based inference in quantile regression, Sun and He [10] introduce the notion of censored quantile rank scores and use them to develop tests for quantile regression coefficients both at a single level and over a quantile region. Wang, He, Ma, Bandyopadhyay, and Sinha [11] use semiparametric generalized odds rate models with a subject-specific random frailty effects to deal with clustering in current status data and develop a minorize-maximize algorithm for estimation that allows to separate the parametric and nonparametric components of the model. Ge, Peng, and Tu [12] aim to identify patients who benefit from a given treatment, which is relevant in personalized medicine. In the setup when the effectiveness of the treatment is assessed through longitudinal measurements and numerous baseline covariates need to be accounted for, they propose a generalized single-index linear threshold model and develop inference based on generalized smoothed estimating equations. In medical studies and beyond, missing data are frequent and must be accounted for. Zheng, Zhang, and Zhou [13] consider the case when the response and some of the covariate values are missing at the same time and not at random. They study the identifiability of the observed likelihood using instrumental variables and estimate the parameters of interest along with the tilting parameter without requiring external data.

The issue ends with a paper on experimental design by Sun and Zhao [14], who extend the general minimum lower-order confounding theory to the case of three level fractional factorial split-plot designs with more important whole plot factors under the orthogonal components system.

Here’s to inspirational readings and new beginnings in 2024,

Johanna G. Nešlehová

Editor-in-Chief, The Canadian Journal of Statistics

Table of Contents of the December 2023 Issue of The Canadian Journal of Statistics

1. Asymptotic theory in bipartite graph models with a growing number of parameters by Yifan Fan, Binyan Jiang, Ting Yan, and Yuan Zhang
2. Variation pattern classification of functional data by Shuhao Jiao, Ron D. Frostig, and Hernando Ombao
3. Automatic structure recovery for generalized additive models by Kai Shen and Yichao Wu
4. Pretest and shrinkage estimators in generalized partially linear models with application to real data by Shakhawat Hossain, Saumen Mandal, and Le An Lac
5. A high-dimensional inverse norm sign test for two-sample location problems by Xifen Huang, Binghui Liu, Qin Zhou, and Long Feng
6. A hyperbolic divergence based nonparametric test for two-sample multivariate distributions by Roulin Wang, Wei Fan, and Xueqin Wang
7. Empirical-process-based specification tests for diffusion models by Qiang Chen, Yuting Gong, and Xunxiao Wang
8. Testing for equality between conditional copulas given discretized conditioning events by Alexis Derumigny, Jean-David Fermanian, and Aleksey Min
9. Matrix compatibility and correlation mixture representation of generalized Gini’s gamma by Takaaki Koike and Marius Hofert
10. From regression rank scores to robust inference for censored quantile regression by Yuan Sun and Xuming He
11. Minorize-maximize algorithm for the generalized odds rate model for clustered current status data by Tong Wang, Kejun He, Wei Ma, Dipankar Bandyopadhyay, and Samiran Sinha
12. A generalized single-index linear threshold model for identifying treatment-sensitive subsets based on multiple covariates and longitudinal measurements by Xinyi Ge, Yingwei Peng, and Dongsheng Tu
13. Likelihood identifiability and parameter estimation with nonignorable missing data by Siming Zheng, Juan Zhang, and Yong Zhou
14. General minimum lower-order confounding three-level split-plot designs when the whole plot factors are important by Tao Sun and Shengli Zhao