本书是一本优秀的统计模型教材，着重讲解线性模型的应用问题，包括广义最小二乘和两步最小二乘模型，以及二分变量的probit及logit模型的应用，还包括关于研究设计、二分变量回归及矩阵代数的背景知识。

这还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。

Foreword to the ReVised Edition

Preface

1 Observational Studies and Experiments

 1.1 Introduction

 1.2 The HIP trial

 1.3 Snow on cholera

 1.4 Yule on the causes of poverty

 Exercise set A

 1.5 End notes

2 The Regression Line

 2.1 Introduction

 2.2 The regression line

 2.3 Hooke's law

 Exercise set A

 2.4 Complexities

 2.5 Simple vs multiple regression

 Exercise set B

 2.6 End notes

3 Matrix Algebra

 3.1 Introduction

 Exercise set A

 3.2 Determinants and inverses

 Exercise set B

 3.3 Random vectors

 Exercise set C

 3.4 Positive definite matrices

 Exercise set D

 3.5 The normal distribution

 Exercise set E

 3.6 If you want a book on matrix algebra

4 Multiple Regression

 4.1 Introduction

 Exercise set A

 4.2 Standard errors

 Things we don't need

 Exercise set B

 4.3 Explained variance in multiple regression

 Association or causation?

 Exercise set C

 4.4 What happens to OLS if the assumptions break down?

 4.5 Discussion questions

 4.6 End notes

5 Multiple Regression: Special Topics

 5.1 Introduction

 5.2 OLSisBLUE

 Exercise set A

 5.3 Generalized least squares

 Exercise set B

 5.4 Examples on GLS

 Exercise set C

 5.5 What happens to GLS if the assumptions break down?

 5.6 Normal theory

 Statistical significance

 Exercise set D

 5.7 The F-test

 "The" F-test in applied work

 Exercise set E

 5.8 Data snooping

 Exercise set F

 5.9 Discussion questions

 5.10 End notes

6 Path Models

 6.1 Stratification

 Exercise set A

 6.2 Hooke's law revisited

 Exercise set B

 6.3 Political repression during the McCarthy era

 Exercise set C

 6.4 Inferring causation .by regression

 Exercise set D

 6.5 Response schedules for path diagrams

 Selection vs intervention

 Structural equations and stable parameter:Ambiguity in notation

 Exercise set E

 6.6 Dummy variables

 Types of variables

 6.7 Discussion questions

 6.8 End notes

7 Maximum Likelihood

 7.1 Introduction

 Exercise set A

 7.2 Probit models

 Why not regression?

 The latent-variable formulation

 Exercise set B

 Identification vs estimation

 What if the Ui are N?

 Exercise set C

 7.3 Logit models

 Exercise set D

 7.4 The effect of Catholic schools

 Latent variables

 Response schedules

 The second equation

 Mechanics: bivariate probit

 Why a model rather than a cross-lab?

 Interactions

 More on table 3 in Evans and Schwab

 More on the second equation

 Exercise set E

 7.5 Discussion questions

 7.6 End notes

8 The Bootstrap

 8.1 Introduction

 Exercise set A

 8.2 Bootstrapping a model for energy demand

 Exercise set B

 8.3 End notes

9 Simultaneous Equations

 9.1 Introduction

 Exercise set A

 9.2 Instrumental variables

 Exercise set B

 9.3 Estimating the butter model

 Exercise set C

 9.4 What are the two stages?

 Invariance assumptions

 9.5 A social-science example: education and fertility

 More on Rindfuss et al

 9.6 Covariates

 9.7 Linear probability models

 The assumptions

 The questions

 Exercise set D

 9.8 More on IVLS

 Some technical issues

 Exercise set E

 Simulations to illustrate IVLS

 9.9 Discussion questions

 9.10 End notes

10 Issues in Statistical Modeling

 10.1 Introduction

 The bootstrap

 The role of asymptotics

 Philosophers' stones

 The modelers' response

 10.2 Critical literature

 10.3 Response schedules

 10.4 Evaluating the models in chapters 7-9

 10.5 Summing up

References

Answers to Exercises

The Computer Labs

Appendix: Sample MATLAB Code

Reprints

 Gibson on McCarthy

 Evans and Schwab on Catholic Schools

 Rindfuss et al on Education and Fertility

 Schneider et al on Social Capital

Index