“fpref” — 2007/1/29 — page xvi — #2
xvi PREFACE TO THE SECOND EDITION
for multinomial responses, both nominal and ordinal. Chapter 7 discusses loglinear
models for Poisson (count) data. Chapter 8 presents methods for matched-pairs data.
I believe that logistic regression is more important than loglinear models, since
most applications with categorical responses have a single binomial or multinomial
response variable. Thus, I have given main attention to this model in these chapters
and in later chapters that discuss extensions of this model. Compared with the first
edition, this edition places greater emphasis on logistic regression and less emphasis
on loglinear models.
I prefer to teach categorical data methods by unifying their models with ordinary
regression and ANOVA models. Chapter 3 does this under the umbrella of generalized
linear models. Some instructors might prefer to cover this chapter rather lightly, using
it primarily to introduce logistic regression models for binomial data (Sections 3.1
and 3.2).
The main change from the first edition is the addition of two chapters dealing with
the analysis of clustered correlated categorical data, such as occur in longitudinal
studies with repeated measurement of subjects. Chapters 9 and 10 extend the matched-
pairs methods of Chapter 8 to apply to clustered data. Chapter 9 does this with
marginal models, emphasizing the generalized estimating equations (GEE) approach,
whereas Chapter 10 uses random effects to model more fully the dependence. The
text concludes with a chapter providing a historical perspective of the development
of the methods (Chapter 11) and an appendix showing the use of SAS for conducting
nearly all methods presented in this book.
The material in Chapters 1–8 forms the heart of an introductory course in categor-
ical data analysis. Sections that can be skipped if desired, to provide more time for
other topics, include Sections 2.5, 2.6, 3.3 and 3.5, 5.3–5.5, 6.3, 6.4, 7.4, 7.5, and
8.3–8.6. Instructors can choose sections from Chapters 9–11 to supplement the basic
topics in Chapters 1–8. Within sections, subsections labelled with an asterisk are less
important and can be skipped for those wanting a quick exposure to the main points.
This book is of a lower technical level than my book Categorical Data Analysis
(2nd edition, Wiley, 2002). I hope that it will appeal to readers who prefer a more
applied focus than that book provides. For instance, this book does not attempt to
derive likelihood equations, prove asymptotic distributions, discuss current research
work, or present a complete bibliography.
Most methods presented in this text require extensive computations. For the
most part, I have avoided details about complex calculations, feeling that comput-
ing software should relieve this drudgery. Software for categorical data analyses
is widely available in most large commercial packages. I recommend that read-
ers of this text use software wherever possible in answering homework problems
and checking text examples. The Appendix discusses the use of SAS (particu-
larly PROC GENMOD) for nearly all methods discussed in the text. The tables
in the Appendix and many of the data sets analyzed in the book are available at
the web site http://www.stat.ufl.edu/∼aa/intro-cda/appendix.html. The web site
http://www.stat.ufl.edu/∼aa/cda/software.html contains information about the use
of other software, such as S-Plus and R, Stata, and SPSS, including a link to an excel-
lent free manual prepared by Laura Thompson showing how to use R and S-Plus to
我的内容管理
展开
