Nonparametric models for longitudinal data provide a flexible platform for evaluating the tenability of parametric assumptions on the functional forms of the change processes of central focus in longitudinal studies. Given the increasing popularity of longitudinal designs such as intensive longitudinal designs in the social and behavioral sciences (Walls & Schafer, 2006), the book, Nonparametric Models for Longitudinal Data by Drs. Colin O. Wu and Xin Tian, is a timely addition to the collection of books on nonparametric modeling approaches. This book features the applications of nonparametric modeling techniques to a rich array of modeling contexts: unstructured nonparametric models for modeling time trends; structured nonparametric models, which include time-varying coefficient models, shared-parameter change-point models, and mixed-effects models; and nonparametric models for conditional distribution functions (e.g., conditional cumulative distribution function, conditional quantiles, and other functionals). The authors provided illustrative examples based on four publicly available data sets: the Baltimore Multicenter AIDS Cohort Study, the National Growth and Health Study, the Enhancing Recovery in Coronary Heart Disease Patients Study, and the Hematopoietic Stem Cell Transplantation data. All of the models considered are regression models with a univariate outcome (dependent variable) measured longitudinally and as linked to both time-invariant and timevarying covariates. Chapters 1 and 2 of the book provide an overview of the models and statistical smoothing techniques covered in subsequent chapters. Distinctions are made between parametric, nonparametric, and semiparametric models. In the context of this book, parametric models are defined as models that impose confirmatory, scientifically driven postulates on the functional relationships between the outcome and covariates. In contrast, nonparametric models are models that uncover Journal of Educational and Behavioral Statistics 2020, Vol. 45, No. 3, pp. 369–373 DOI: 10.3102/1076998620915291 Article reuse guidelines: sagepub.com/journals-permissions © 2020 AERA. http://jebs.aera.net

中文翻译：

---

纵向数据的非参数模型综述

纵向数据的非参数模型提供了一个灵活的平台，用于评估纵向研究中集中关注的变化过程的功能形式的参数假设的持久性。鉴于纵向设计的日益普及，例如社会和行为科学中的密集纵向设计（Walls＆Schafer，2006年），《 Drs。Longstd纵向数据的非参数模型》一书。Colin O. Wu和Xin Tian是非参数建模方法书籍的及时补充。本书将非参数建模技术应用于大量建模环境中：用于建模时间趋势的非结构化非参数模型；结构化的非参数模型，包括时变系数模型，共享参数变化点模型，和混合效应模型；条件分布函数的非参数模型（例如，条件累积分布函数，条件分位数和其他函数）。作者基于四个可公开获得的数据集提供了示例性例子：巴尔的摩多中心艾滋病队列研究，国家成长与健康研究，增强冠心病患者康复研究和造血干细胞移植数据。所考虑的所有模型都是回归模型，具有纵向测量的单变量结果（因变量），并与时不变协变量和时变协变量相关。本书的第1章和第2章概述了后续各章中涉及的模型和统计平滑技术。在参数，非参数，和半参数模型。在本书的上下文中，参数模型定义为对结果和协变量之间的函数关系施加确定性，科学驱动的假设的模型。相反，非参数模型是揭示《教育与行为统计杂志2020年》第1卷的模型。45，第3号，第369–373页DOI：10.3102 / 1076998620915291文章重用指南：sagepub.com/journals-permissions©2020 AERA。http://jebs.aera.net 3102/1076998620915291文章重用指南：sagepub.com/journals-permissions©2020 AERA。http://jebs.aera.net 3102/1076998620915291文章重用指南：sagepub.com/journals-permissions©2020 AERA。http://jebs.aera.net