Applied Logistic Regression

出版者:Wiley-Blackwell

作者:David W. Hosmer Jr.

页数:528

出版时间:2013-4-26

价格:GBP 109.00

装帧:Hardcover

isbn号码:9780470582473

具体描述

This new edition provides a focused introduction to the LR model and its use in methods for modeling the relationship between a dichotomous outcome variable and a set of covariables. It presents expanded coverage on random effects models, estimation in the presence of interaction, and fractional polynomials; offers discussions on Bayesian logistic regression, likelihood based confidence interval estimates, tests for non-nested models, and multivariable fractional polynomials; includes R language and updated SAS, STATA, and BUGS computer code for analyzing data sets; and more.

目录信息

Preface to the Third Edition
1 Introduction to the Logistic Regression Model
1.1 Introduction
1.2 Fitting the Logistic Regression Model
1.3 Testing for the Significance of the Coefficients
1.4 Confidence Interval Estimation
1.5 Other Estimation Methods
1.6 Data Sets Used in Examples and Exercises
2 The Multiple Logistic Regression Model
2.1 Introduction
2.2 The Multiple Logistic Regression Model
2.3 Fitting the ultiple Logistic Regression Model
2.4 Testing for the Significance of the Model
2.5 Confidence Interval Estimation
2.6 Other Estimation Methods
3 Interpretation of the Fitted Logistic Regession Model
3.1 Introduction
3.2 Dichotomous Independent Variable
3.3 Polychotomous Independent Variable
3.4 Continuous Independent Variable
3.5 Multivariable Models
3.6 Presentation and Interpretation of the Fitted Values
3.7 A Comparision of Logistic Regression and Stratified Analysis for 2x2 Tables
4 Model-Building Streategies and Methods for Logistic Regression
4.1 Introduction
4.2 Purposeful Selection of Covariates
4.2.1 Methods to Examine the Scale of a Continuous Covariate in the Logit
4.2.2 Examples of Purposeful Selection
4.3 Other Methods for Selecting Covariates
4.3.1 Stepwise Selection
4.3.2 Best Subsets Logistic Regression || SAS: PROC LOGISTIC
4.3.3 Multivariable Fractional Polynomials
4.4 Numerical Problems
5 Assessing the Fit of the Model
5.1 Introduction
5.2 Summary Measures of Goodness of Fit
5.2.1 Pearson Chi-Square Statistic, Deviance, and Sum-of-Squares
5.2.2 The Hosmer-Lemeshow Tests
5.2.3 Classification Tables
5.2.4 Area Under the Receiver Operating Characteristic Curve
5.2.5 Other Summary Measures
5.3 Logistic Regression Diagnostics
5.4 Assessment of Fit via External Validation
5.5 Interpretation and Presentation of the Results from a Fitted Logistic Regression Model
6 Application of Logistic Regression with Different Sampling Models
6.1 Introduction
6.2 Cohort Studies
6.3 Case-Control Studies
6.4 Fitting Logistic Regression Models to Data from Complex Sample Surveys
7 Logistic Regression for Matched Case-Control Studies
7.1 Introduction
7.2 Methods For Assessment of Fit in a 1-M Matched Study
7.3 An Example Using the Logistic Regression Model in a 1-1 Matched Study
7.4 An Example Using the Logistic Regression Model in a 1-M Matched Study
8 Logistic Regression Models for Multinomial and Ordinal Outcomes
8.1 Multinomial Logistic Regression Model
8.1.1 Introductino to the Model and Estimation of Model Parameters
8.1.2 Interpreting and Assessing the Significance of the Estimated Coefficients
8.1.3 Model-Building Strategies
8.1.4 Assessment of Fit and Diagnostic Statistics
8.2 Ordinal Logistic Regression Model
8.2.1 Introduction to the Models, Methods for Fitting and Interpretation of Model Parameters
8.2.2 Model-Building Strategies
9 Logistic Regression Models for the Analysis of Correlated Data
9.1 Introduction
9.2 Logistic Regression Models for the Analysis of Correlated Data
9.3 Estimation Methods for Correlated Data Logistic Regression Models
9.4 Interpretation of Coefficients from Logistic Regression
9.4.1 Population Average Model
9.4.2 Cluster-Specific Model
9.4.3 Alternative Estimation Methods for the Cluster-Specific Model
9.4.4 Comparison of Population Average and Cluster-Specific Model
9.5 An Example of Logistic Regression Modeling with Correlated Data
9.5.1 Choice of Model for Correlated Data Analysis
9.5.2 Population Average Model
9.5.3 Cluster-pecific Model
9.5.4 Additional Points to Consider when Fitting Logistic Regression Models to Correlated Data
9.6 Assessment of Model Fit
9.6.1 Assessment of Population Average Model Fit
9.6.2 Assessment of Cluster-Specific Model Fit
9.6.3 Conclusions
10 Special Topics
10.1 Introduction
10.2 Application of Propensity Score Methods in Logistic Regression Modeling
10.3 Exact Methods for Logistic Regression Models
10.4 Missing Data
10.5 Sample Size Issues when Fitting Logistic Regressino Models
10.6 Bayesian Methods for Logistic Regression
10.6.1 The Bayesian Logistic Regression Model
10.6.2 MCMC Simulation
10.6.3 An Example of a Bayesian Analysis and Its Interpretation
10.7 Other Link Functions for Binary Regression Models
10.8 Mediation
10.8.1 Distinguishing Mediators from Confounders
10.8.2 Implications for the Interpretation of an Adjusted Logistic Regression Coefficient
10.8.3 Why Adjust for a Mediator?
10.8.4 Using Logistic Regression to Assess Mediation: Assumptions
10.9 More About Statistical Interaction
10.9.1 Additive versus Multiplicative Scale-Risk Difference versus Odds Ratios
10.9.2 Estimating and Testing Additive Interaction
References
Index
· · · · · · (收起)

用户评价

天呐，我最近刚读完这本关于应用统计学的巨著，简直是醍醐灌顶！书里对贝叶斯推断的阐述简直是教科书级别的清晰，特别是它如何将复杂的先验信息整合进模型构建过程，让人印象深刻。作者没有仅仅停留在理论的阐述上，而是深入挖掘了实际应用中的各种陷阱和解决方案，比如在小样本情况下如何进行稳健的参数估计，这一点在我的研究生阶段的课题中简直是雪中送炭。而且，书中对模型选择和诊断的讨论非常详尽，什么AIC、BIC的取舍，残差分析的细微差别，都处理得面面俱到，让人感觉作者对这个领域有着极其深刻的洞察力。如果说有什么可以挑剔的，也许是某些高级主题的证明过程略显跳跃，但对于目标是应用而非纯数学推导的读者来说，这反而是一种高效的学习方式。总而言之，这本书为我打开了一扇通往更深层次统计建模的大门，绝对值得所有处理分类数据和概率预测的人员仔细研读。

说实话，这本书的叙事风格非常独特，它不像传统教科书那样枯燥乏味，反而更像是一位经验丰富的数据科学家在跟你分享他的“秘籍”。书中对数据预处理和特征工程的强调令人耳目一新，尤其是在处理现实世界中那些“脏数据”时，作者提供了一套切实可行的、基于经验的指导方针，而不是空泛的理论。我特别欣赏作者在讨论模型解释性时所采取的视角——他认为一个“漂亮”的模型不应该仅仅是预测准确，更要能被业务人员理解。关于交互项的处理，书中的章节提供了很多实用的技巧，如何识别并恰当地引入高阶交互，避免模型过度拟合，这些都是我过去在实践中摸索了很久才领悟的道理。对于初入职场的分析师来说，这本书的实用性远超其理论深度，它更像是一本“武功秘籍”，教会你如何将招式（统计理论）应用到实战（真实数据）中去。

坦白说，我对这本书的排版和图表质量感到非常满意。在处理复杂统计分布和概率图时，清晰的视觉呈现至关重要，而这本书在这方面做得非常出色。图表设计简洁明了，配色专业，避免了视觉上的干扰。更重要的是，作者在论述模型假设检验时，没有采用那种冷冰冰的、纯粹的P值导向的讨论，而是深入探讨了统计功效（Power）和效应量（Effect Size）的重要性。他反复强调，一个“显著”的结果如果效应量太小，在实践中可能毫无意义。这种对统计实践的批判性反思，使得这本书不仅仅是工具书，更是一本培养良好统计思维的哲学读本。读完之后，我感觉自己对统计结果的解读更加谨慎和负责任了。

这本书最让我感到惊喜的是它对因果推断在广义线性模型框架下的融合探讨。很多同类书籍往往将因果推断视为一个完全独立的领域，但这本著作巧妙地将它们联系起来，展示了如何利用这些统计工具来探索“如果…会怎样”的问题，这在政策评估和医学研究中至关重要。作者对混杂因素（confounding variables）的识别和调整给出了非常细致的指导，甚至涉及到了一些非常前沿的因果发现算法的初级介绍。我感觉作者的视野非常开阔，他不仅教你如何建立一个好的预测模型，更引导你思考如何从数据中提取出更深层次的、关于世界运作机制的洞察。对于那些希望从“预测”走向“解释”和“干预”的读者，这本书提供了必要的理论桥梁和实践路径。

我对这本书的结构安排感到非常赞赏，它采取了一种非常逻辑严密的递进方式。从最基础的二元选择模型开始，逐步过渡到多项式、有序响应等更复杂的结构，每一步的衔接都非常自然。作者在介绍每一种模型时，都会先给出直观的动机，然后才进入数学框架，这种“先知其然，后知其所以然”的教学方式极大地降低了学习曲线的陡峭程度。尤其是关于缺失数据处理的那部分内容，书中介绍的几种多重插补技术，配上了清晰的R语言（或者说是某种统计软件）代码示例，让我立刻就能上手操作。这种紧密结合最新软件实现的风格，对于希望快速将知识转化为生产力的读者来说，简直是福音。我必须强调，它的代码示例并非简单的复制粘贴，而是经过深思熟虑，旨在突出核心概念的实现细节。

可以说是关于logistic regression的百科全书了

可以说是关于logistic regression的百科全书了

非常全面完整，囊括当前很流行的逻辑回归方法及应用实例。

非常全面完整，囊括当前很流行的逻辑回归方法及应用实例。

非常全面完整，囊括当前很流行的逻辑回归方法及应用实例。