统计方法学：2025–2026 年度综述

An annual survey of the stat.ME arXiv category, covering research published from September 2025 – August 2026. Built on a close-reading set of 144 papers and organised into 40 chapters, it maps the year by the field's research sub-directions.

Book framework (share of the year's output):

• Robust Estimation and Optimal Experimental Design for Causal Effects (18.9%) — Focuses on robust methods for estimating treatment effects and optimizing experimental designs to handle partial measurements and complex covariate structures.
• Causal Inference and Discovery under Unmeasured Confounding (12.6%) — Addresses challenges in identifying causal effects and discovering causal structures when confounding variables are unobserved or misspecified.
• Conformal Prediction and Uncertainty Quantification (12.1%) — Develops methods for providing distribution-free uncertainty guarantees, such as conformal prediction intervals, for predictive models.
• Machine Learning Frameworks for Statistical Inference and Policy Learning (8.7%) — Integrates deep learning and representation learning techniques into statistical frameworks for tasks like policy learning and distributional outcome modeling.
• High-Dimensional Statistical Inference and Change Point Detection (7.8%) — Tackles inference problems in high-dimensional settings, including change point detection, covariance estimation, and adaptive modeling.
• Bayesian Nonparametric and Hierarchical Modeling Approaches (7.5%) — Utilizes Bayesian nonparametric and hierarchical models to handle complex data structures, latent risk pathways, and spatial heterogeneity.
• Functional Data Analysis and Spatio-Temporal Modeling (6.9%) — Focuses on modeling functional and spatio-temporal data, addressing issues like missing data, clustering, and physics-informed regression.
• Linear, Generalized, and Mixed Models with Latent Structures (6.7%) — Advances the theory and application of linear, generalized linear, and mixed models, including graphical and latent variable extensions.
• Model Selection, Averaging, and Variable Selection (6.6%) — Develops criteria and methods for selecting models and variables, including Bayesian model averaging and tests for parameter constancy.
• Regression Methods including Quantile and Robust Techniques (4.5%) — Explores various regression techniques, such as quantile regression, symbolic regression, and robust semiparametric inference.
• Meta-Analysis, Sensitivity Analysis, and Mediation Analysis (4.3%) — Provides frameworks for meta-analysis, sensitivity analysis, and mediation analysis to assess robustness and decompose effects in observational studies.
• Time Series Forecasting and Multivariate Time Series Analysis (3.4%) — Focuses on forecasting techniques, change-point detection, and inference for multivariate and functional time series data.

Each sub-direction is covered across three chapters — how it framed its questions, the strength of its key findings, and its open debates and frontiers — so the reader gets a traceable picture of what this part of the field actually produced over the year.