![统计模型:理论和实践(英文版.第2版)[图书]](http://images.china-pub.com/ebook195001-200000/197117/zcover.jpg)
基本信息
- 原书名：Statistical Models: Theory and Practice
- 原出版社： Cambridge University Press
- 作者： (美)David A. Freedman
- 丛书名： 华章数学.统计学原版精品系列
- 出版社：机械工业出版社
- ISBN：9787111317975
- 上架时间：2010-11-3
- 出版日期：2010 年9月
- 开本：32开
- 页码：442
- 版次：2-1
- 所属分类：数学 > 数学实验与数学建模 > 数学模型
【插图】
内容简介
作译者
（美）弗里德曼,是加州大学伯克利分校的统计学教授、杰出的数理统计学家。其研究范围包括鞅不等式分析、Markov过程、抽样、自助法等。他是美国科学学院士。在2003年。美国科学院授予他John J.Carty科学进步奖，以表彰他对统计理论和实践做出的贡献。
目录
Foreword to the ReVised Edition<BR>Preface<BR>1 Observational Studies and Experiments<BR> 1.1 Introduction<BR> 1.2 The HIP trial<BR> 1.3 Snow on cholera<BR> 1.4 Yule on the causes of poverty<BR> Exercise set A<BR> 1.5 End notes <BR>2 The Regression Line<BR> 2.1 Introduction<BR> 2.2 The regression line<BR> 2.3 Hooke\'s law<BR> Exercise set A<BR> 2.4 Complexities<BR> 2.5 Simple vs multiple regression<BR> Exercise set B<BR> 2.6 End notes<BR>3 Matrix Algebra<BR> 3.1 Introduction<BR> Exercise set A<BR> 3.2 Determinants and inverses<BR> Exercise set B<BR> 3.3 Random vectors<BR> Exercise set C<BR> 3.4 Positive definite matrices<BR> Exercise set D<BR> 3.5 The normal distribution<BR> Exercise set E<BR> 3.6 If you want a book on matrix algebra<BR>4 Multiple Regression<BR> 4.1 Introduction<BR> Exercise set A<BR> 4.2 Standard errors<BR> Things we don\'t need<BR> Exercise set B<BR> 4.3 Explained variance in multiple regression<BR> Association or causation?<BR> Exercise set C<BR> 4.4 What happens to OLS if the assumptions break down?<BR> 4.5 Discussion questions<BR> 4.6 End notes<BR>5 Multiple Regression: Special Topics<BR> 5.1 Introduction<BR> 5.2 OLSisBLUE<BR> Exercise set A<BR> 5.3 Generalized least squares<BR> Exercise set B<BR> 5.4 Examples on GLS<BR> Exercise set C<BR> 5.5 What happens to GLS if the assumptions break down?<BR> 5.6 Normal theory<BR> Statistical significance<BR> Exercise set D<BR> 5.7 The F-test<BR> "The" F-test in applied work<BR> Exercise set E<BR> 5.8 Data snooping<BR> Exercise set F<BR> 5.9 Discussion questions<BR> 5.10 End notes<BR>6 Path Models<BR> 6.1 Stratification<BR> Exercise set A<BR> 6.2 Hooke\'s law revisited<BR> Exercise set B<BR> 6.3 Political repression during the McCarthy era<BR> Exercise set C<BR> 6.4 Inferring causation .by regression<BR> Exercise set D<BR> 6.5 Response schedules for path diagrams<BR> Selection vs intervention<BR> Structural equations and stable parameter:Ambiguity in notation<BR> Exercise set E<BR> 6.6 Dummy variables<BR> Types of variables<BR> 6.7 Discussion questions<BR> 6.8 End notes<BR>7 Maximum Likelihood<BR> 7.1 Introduction <BR> Exercise set A<BR> 7.2 Probit models<BR> Why not regression?<BR> The latent-variable formulation<BR> Exercise set B<BR> Identification vs estimation<BR> What if the Ui are N?<BR> Exercise set C<BR> 7.3 Logit models<BR> Exercise set D<BR> 7.4 The effect of Catholic schools<BR> Latent variables<BR> Response schedules<BR> The second equation<BR> Mechanics: bivariate probit<BR> Why a model rather than a cross-lab?<BR> Interactions<BR> More on table 3 in Evans and Schwab<BR> More on the second equation<BR> Exercise set E<BR> 7.5 Discussion questions<BR> 7.6 End notes<BR>8 The Bootstrap<BR> 8.1 Introduction<BR> Exercise set A<BR> 8.2 Bootstrapping a model for energy demand<BR> Exercise set B<BR> 8.3 End notes<BR>9 Simultaneous Equations<BR> 9.1 Introduction<BR> Exercise set A<BR> 9.2 Instrumental variables<BR> Exercise set B<BR> 9.3 Estimating the butter model<BR> Exercise set C<BR> 9.4 What are the two stages?<BR> Invariance assumptions<BR> 9.5 A social-science example: education and fertility<BR> More on Rindfuss et al<BR> 9.6 Covariates<BR> 9.7 Linear probability models<BR> The assumptions<BR> The questions<BR> Exercise set D<BR> 9.8 More on IVLS<BR> Some technical issues<BR> Exercise set E<BR> Simulations to illustrate IVLS<BR> 9.9 Discussion questions<BR> 9.10 End notes<BR>10 Issues in Statistical Modeling<BR> 10.1 Introduction<BR> The bootstrap<BR> The role of asymptotics<BR> Philosophers\' stones<BR> The modelers\' response<BR> 10.2 Critical literature<BR> 10.3 Response schedules<BR> 10.4 Evaluating the models in chapters 7-9<BR> 10.5 Summing up<BR>References<BR>Answers to Exercises<BR>The Computer Labs<BR>Appendix: Sample MATLAB Code<BR>Reprints<BR> Gibson on McCarthy<BR> Evans and Schwab on Catholic Schools<BR> Rindfuss et al on Education and Fertility<BR> Schneider et al on Social Capital<BR>Index
前言
This book is primarily intended for advanced undergraduates or begin-ning graduate students in statistics. It should also be of interest to many students and professionals in the social and health sciences. Although writ-ten as a textbook, it can be read on its own. The focus is on applications of linear models, including generalized least squares, two-stage least squares,probits and logits. The bootstrap is explained as a technique for estimating bias and computing standard errors.
The contents of the book can fairly be described as what you have to know in order to start reading empirical papers that use statistical models. The emphasis throughout is on the connection--or lack of connection--between the models and the real phenomena. Much of the discussion is organized around published studies; the key papers are reprinted for ease of reference.Some observers may find the tone of the discussion too skeptical. If you are among them, I would make an unusual request: suspend belief until you finish reading the book. (Suspension of disbelief is all too easily obtained,but that is a topic for another day.)
The first chapter contrasts observational studies with experiments, and introduces regression as a technique that may help to adjust for confounding in observational studies. There is a chapter that explains the regression line,and another chapter with a quick review of matrix algebra. (At Berkeley, half the statistics majors need these chapters.) The going would be much easier with students who know such material. Another big plus would be a solid upper-division course introducing the basics of probability and statistics.
Technique is developed by practice. At Berkeley, we have lab sessions where students use the computer to analyze data. There is a baker's dozen of these labs at the back of the book, with outlines for several more, and there are sample computer programs. Data are available to instructors from the publisher, along with source files for the labs and computer code: send email to solutions @ cambridge.org.
A textbook is only as good as its exercises, and there are plenty of them in the pages that follow. Some are mathematical and some are hypothetical,providing the analogs of lemmas and counter-examples in a more conven-tional treatment. On the other hand, many of the exercises are based on actual studies. Here is a summary of the data and the analysis; here is a specific issue: where do you come down? Answers to most of the exercises are at the back of the book. Beyond exercises and labs, students at Berkeley write papers during the semester. Instructions for projects are also available from the publisher.
A text is defined in part by what it chooses to discuss, and in part by what it chooses to ignore; the topics of interest are not to be covered in one book, no matter how thick. My objective was to explain how practitioners infer causation from association, with the bootstrap as a counterpoint to the usual asymptotics. Examining the logic of the enterprise is crucial, and that takes time. If a favorite technique has been slighted, perhaps this reasoning will make amends.
There is enough material in the book for 15-20 weeks of lectures and discussion at the undergraduate level, or 10-15 weeks at the graduate level.With undergraduates on the semester system, I cover chapters 1-7, and in-troduce simultaneity (sections 9.1--4). This usually takes 13 weeks. If things go quickly, I do the bootstrap (chapter 8), and the examples in chapter 9.On a quarter system with ten-week terms, I would skip the student presenta-tions and chapters 8-9; the bivariate probit model in chapter 7 could also be dispensed with.
During the last two weeks of a semester, students present their projects,or discuss them with me in office hours. I often have a review period on the last day of class. For a graduate course, I supplement the material with additional case studies and discussion of technique.
The revised text organizes the chapters somewhat differently, which makes the teaching much easier. The exposition has been improved in a number of other ways, without (I hope) introducing new difficulties. There are many new examples and exercises.
Acknowledgements
I've taught graduate and undergraduate courses based on this material for many years at Berkeley, and on occasion at Stanford and Athens. The students in those courses were helpful and supportive. I would also like to thank Dick Berk, Mfiire Nf Bhrolchfiin, Taylor Boas, Derek Briggs, David Collier, Persi Diaconis, Thad Dunning, Mike Finkelstein, Paul Humphreys, Jon McAuliffe,Doug Rivers, Mike Roberts, Don Ylvisaker, and PengZhao, along with several anonymous reviewers, for many useful comments. Russ Lyons and Roger Purves were virtual coauthors; David Tranah was an outstanding editor.
The contents of the book can fairly be described as what you have to know in order to start reading empirical papers that use statistical models. The emphasis throughout is on the connection--or lack of connection--between the models and the real phenomena. Much of the discussion is organized around published studies; the key papers are reprinted for ease of reference.Some observers may find the tone of the discussion too skeptical. If you are among them, I would make an unusual request: suspend belief until you finish reading the book. (Suspension of disbelief is all too easily obtained,but that is a topic for another day.)
The first chapter contrasts observational studies with experiments, and introduces regression as a technique that may help to adjust for confounding in observational studies. There is a chapter that explains the regression line,and another chapter with a quick review of matrix algebra. (At Berkeley, half the statistics majors need these chapters.) The going would be much easier with students who know such material. Another big plus would be a solid upper-division course introducing the basics of probability and statistics.
Technique is developed by practice. At Berkeley, we have lab sessions where students use the computer to analyze data. There is a baker's dozen of these labs at the back of the book, with outlines for several more, and there are sample computer programs. Data are available to instructors from the publisher, along with source files for the labs and computer code: send email to solutions @ cambridge.org.
A textbook is only as good as its exercises, and there are plenty of them in the pages that follow. Some are mathematical and some are hypothetical,providing the analogs of lemmas and counter-examples in a more conven-tional treatment. On the other hand, many of the exercises are based on actual studies. Here is a summary of the data and the analysis; here is a specific issue: where do you come down? Answers to most of the exercises are at the back of the book. Beyond exercises and labs, students at Berkeley write papers during the semester. Instructions for projects are also available from the publisher.
A text is defined in part by what it chooses to discuss, and in part by what it chooses to ignore; the topics of interest are not to be covered in one book, no matter how thick. My objective was to explain how practitioners infer causation from association, with the bootstrap as a counterpoint to the usual asymptotics. Examining the logic of the enterprise is crucial, and that takes time. If a favorite technique has been slighted, perhaps this reasoning will make amends.
There is enough material in the book for 15-20 weeks of lectures and discussion at the undergraduate level, or 10-15 weeks at the graduate level.With undergraduates on the semester system, I cover chapters 1-7, and in-troduce simultaneity (sections 9.1--4). This usually takes 13 weeks. If things go quickly, I do the bootstrap (chapter 8), and the examples in chapter 9.On a quarter system with ten-week terms, I would skip the student presenta-tions and chapters 8-9; the bivariate probit model in chapter 7 could also be dispensed with.
During the last two weeks of a semester, students present their projects,or discuss them with me in office hours. I often have a review period on the last day of class. For a graduate course, I supplement the material with additional case studies and discussion of technique.
The revised text organizes the chapters somewhat differently, which makes the teaching much easier. The exposition has been improved in a number of other ways, without (I hope) introducing new difficulties. There are many new examples and exercises.
Acknowledgements
I've taught graduate and undergraduate courses based on this material for many years at Berkeley, and on occasion at Stanford and Athens. The students in those courses were helpful and supportive. I would also like to thank Dick Berk, Mfiire Nf Bhrolchfiin, Taylor Boas, Derek Briggs, David Collier, Persi Diaconis, Thad Dunning, Mike Finkelstein, Paul Humphreys, Jon McAuliffe,Doug Rivers, Mike Roberts, Don Ylvisaker, and PengZhao, along with several anonymous reviewers, for many useful comments. Russ Lyons and Roger Purves were virtual coauthors; David Tranah was an outstanding editor.
序言
Some books are correct. Some are clear. Some are useful. Some are entertaining. Few are even two of these. This book is all four. Statistical Models: Theory and Practice is lucid, candid and insightful, a joy to read.We are fortunate that David Freedman finished this new edition before his death in late 2008. We are deeply saddened by his passing, and we greatly admire the energy and cheer he brought to this volume--and many other projects---during his final months.
This book focuses on half a dozen of the most common tools in applied statistics, presenting them crisply, without jargon or hyperbole. It dissects real applications: a quarter of the book reprints articles from the social and life sciences that hinge on statistical models. It articulates the assumptions necessary for the tools to behave well and identifies the work that the as-sumptions do. This clarity makes it easier for students and practitioners to see where the methods will be reliable; where they are likely to fall, and how badly; where a different method might work; and where no inference is possible--no matter what tool somebody tries to sell them.
Many texts at this level are little more than bestiaries of methods, pre-senting dozens of tools with scant explication or insight, a cookbook,numbers-are-numbers approach. "If the left hand side is continuous, use a linear model; fit by least-squares. If the left hand side is discrete, use a logit or probit model; fit by maximum likelihood." Presenting statistics this way invites students to believe that the resulting parameter estimates, standard errors, and tests of significance are meaningful--perhaps even untangling complex causal relationships. They teach students to think scientific infer-ence is purely algorithmic. Plug in the numbers; out comes science. This undervalues both substantive and statistical knowledge.
To select an appropriate statistical method actually requires careful thought about how the data were collected and what they measure. Data are not "just numbers." Using statistical methods in situations where the un-derlying assumptions are false can yield gold or dross--but more often dross.
Statistical Models brings this message home by showing both good and questionable applications of statistical tools in landmark research: a study of political intolerance during the McCarthy period, the effect of Catholic schooling on completion of high school and entry into college, the relation-ship between fertility and education, and the role of government institutions in shaping social capital. Other examples are drawn from medicine and epidemiology, including John Snow's classic work on the cause of cholera--a shining example of the success of simple statistical tools when paired with substantive knowledge and plenty of shoe leather. These real applications bring the theory to life and motivate the exercises.
The text is accessible to upper-division undergraduates and beginning graduate students. Advanced graduate students and established researchers will also find new insights. Indeed, the three of us have learned much by reading it and teaching from it.
And those who read this textbook have not exhausted Freedman's ap-proachable work on these topics, Many of his related research articles are collected in Statistical Models and Causal Inference: A Dialogue with the Social Sciences (Cambridge University Press, 2009), a useful companion to this text. The collection goes further into some applications mentioned in the textbook, such as the etiology of cholera and the health effects of Hormone Replacement Therapy. Other applications range from adjusting the census for undercount to quantifying earthquake risk. Several articles address the-oretical issues raised in the textbook. For instance, randomized assignment in an experiment is not enough to justify regression: without further assump-tions, multiple regression estimates of treatment effects are biased. The col-lection also covers the philosophical foundations of statistics and methods the textbook does not, such as survival analysis.
Statistical Models: Theory and Practice presents serious applications and the underlying theory without sacrificing clarity or accessibility. Freed-man shows with wit and clarity how statistical analysis can inform and how it can deceive. This book is unlike any other, a treasure: an introductory book that conveys some of the wisdom required to make reliable statistical inferences. It is an important part of Freedman's legacy.
David Collier, Jasjeet Singh Sekhon, and Philip B. Stark
University of California, Berkeley
This book focuses on half a dozen of the most common tools in applied statistics, presenting them crisply, without jargon or hyperbole. It dissects real applications: a quarter of the book reprints articles from the social and life sciences that hinge on statistical models. It articulates the assumptions necessary for the tools to behave well and identifies the work that the as-sumptions do. This clarity makes it easier for students and practitioners to see where the methods will be reliable; where they are likely to fall, and how badly; where a different method might work; and where no inference is possible--no matter what tool somebody tries to sell them.
Many texts at this level are little more than bestiaries of methods, pre-senting dozens of tools with scant explication or insight, a cookbook,numbers-are-numbers approach. "If the left hand side is continuous, use a linear model; fit by least-squares. If the left hand side is discrete, use a logit or probit model; fit by maximum likelihood." Presenting statistics this way invites students to believe that the resulting parameter estimates, standard errors, and tests of significance are meaningful--perhaps even untangling complex causal relationships. They teach students to think scientific infer-ence is purely algorithmic. Plug in the numbers; out comes science. This undervalues both substantive and statistical knowledge.
To select an appropriate statistical method actually requires careful thought about how the data were collected and what they measure. Data are not "just numbers." Using statistical methods in situations where the un-derlying assumptions are false can yield gold or dross--but more often dross.
Statistical Models brings this message home by showing both good and questionable applications of statistical tools in landmark research: a study of political intolerance during the McCarthy period, the effect of Catholic schooling on completion of high school and entry into college, the relation-ship between fertility and education, and the role of government institutions in shaping social capital. Other examples are drawn from medicine and epidemiology, including John Snow's classic work on the cause of cholera--a shining example of the success of simple statistical tools when paired with substantive knowledge and plenty of shoe leather. These real applications bring the theory to life and motivate the exercises.
The text is accessible to upper-division undergraduates and beginning graduate students. Advanced graduate students and established researchers will also find new insights. Indeed, the three of us have learned much by reading it and teaching from it.
And those who read this textbook have not exhausted Freedman's ap-proachable work on these topics, Many of his related research articles are collected in Statistical Models and Causal Inference: A Dialogue with the Social Sciences (Cambridge University Press, 2009), a useful companion to this text. The collection goes further into some applications mentioned in the textbook, such as the etiology of cholera and the health effects of Hormone Replacement Therapy. Other applications range from adjusting the census for undercount to quantifying earthquake risk. Several articles address the-oretical issues raised in the textbook. For instance, randomized assignment in an experiment is not enough to justify regression: without further assump-tions, multiple regression estimates of treatment effects are biased. The col-lection also covers the philosophical foundations of statistics and methods the textbook does not, such as survival analysis.
Statistical Models: Theory and Practice presents serious applications and the underlying theory without sacrificing clarity or accessibility. Freed-man shows with wit and clarity how statistical analysis can inform and how it can deceive. This book is unlike any other, a treasure: an introductory book that conveys some of the wisdom required to make reliable statistical inferences. It is an important part of Freedman's legacy.
David Collier, Jasjeet Singh Sekhon, and Philip B. Stark
University of California, Berkeley
媒体评论
“统计的第二门课是严肃的、正确的和有趣的。本书介绍了回归、因果建模、最大似然和自助法。分析现实数据的每个人都应该阅读本书，并且我们也很荣幸现在能有这个修订版。”
——Persi Diaconis，斯坦福大学数学与统计学教授
“在本书中，作者解释了因果建模中主要使用的统计方法，通过有趣的实例清晰而生动地描述了复杂的统计思想。初学者和实践者都将从本书中获益。”
——Alan Krueger，普林斯顿大学经济与公共政策学教授
“回归方法经常应用于观测数据，目的是获得因果结论。在什么环境下这是合理的？分析背后的假定是什么？本书回答了这些问题。对于不仅仅使用回归来总结数据的任何人，本书都是必读的。本书的写作风格非常好，对于社会科学中相关研究论文的讨论极具洞察力。对于从事统计建模或者讲授回归的每个人，我强烈推荐此书”
——Aad van der Vaart，阿姆斯特丹自由大学统计学教授
“本书是该学科的一个现代导论，讨论了图形模型和联立方程等主题。书中有许多富有启发性的练习和计算机实验。特别有价值的是关于应用统计中主要“哲人石”的关键评论。这是一本鼓舞人心的而又易读的书，无论是老师还是学生都会从中受益。”
——Gesine Reinert，牛津大学统计学教授
——Persi Diaconis，斯坦福大学数学与统计学教授
“在本书中，作者解释了因果建模中主要使用的统计方法，通过有趣的实例清晰而生动地描述了复杂的统计思想。初学者和实践者都将从本书中获益。”
——Alan Krueger，普林斯顿大学经济与公共政策学教授
“回归方法经常应用于观测数据，目的是获得因果结论。在什么环境下这是合理的？分析背后的假定是什么？本书回答了这些问题。对于不仅仅使用回归来总结数据的任何人，本书都是必读的。本书的写作风格非常好，对于社会科学中相关研究论文的讨论极具洞察力。对于从事统计建模或者讲授回归的每个人，我强烈推荐此书”
——Aad van der Vaart，阿姆斯特丹自由大学统计学教授
“本书是该学科的一个现代导论，讨论了图形模型和联立方程等主题。书中有许多富有启发性的练习和计算机实验。特别有价值的是关于应用统计中主要“哲人石”的关键评论。这是一本鼓舞人心的而又易读的书，无论是老师还是学生都会从中受益。”
——Gesine Reinert，牛津大学统计学教授
