Amazon.com: Customer Reviews: Logistic Regression Models (Chapman & Hall/CRC Texts in Statistical Science)
Amazon Vehicles Editors' Picks Amazon Fashion Learn more Discover it Songs of Summer Fire TV Stick Health, Household and Grocery Back to School Totes Summer-Event-Garden Amazon Cash Back Offer TheKicks TheKicks TheKicks  Amazon Echo  Echo Dot  Amazon Tap  Echo Dot  Amazon Tap  Amazon Echo Starting at $49.99 All-New Kindle Oasis GNO Shop Now

Customer Reviews

4.3 out of 5 stars
15
Format: Textbook Binding|Change
Price:$84.98+ Free shipping with Amazon Prime
Your rating(Clear)Rate this item


There was a problem filtering reviews right now. Please try again later.

on July 5, 2009
LOGISTIC REGRESSION MODELS (2009, Chapman & Hall/CRC - 656 pages)
Joseph M Hilbe, Jet Propulsion Laboratory, CalTech and Arizona
State University

Prof Hilbe and I have coauthored several journal articles together in the past and plan to write more together in the future on quasi-least squares regression. However, I was delighted to have the opportunity to read an almost completed version of "Logistic Regression Models" prior to its publication. When I concluded reading it, I came to the clear conclusion that it is a must-have text for those with an interest in logistic regression and related models. Whether the reader is just learning about the area, or is an experienced statistician, there is something for readers of every statistical background.

Some of the reasons I recommend this book with 5 stars are:

1) The text is a comprehensive coverage of the subject - It discusses more logistic-based models than any other single text.

2) Prof. Hilbe's writing style is aimed to make the concepts involved with logistic regression as clear and understandable as possible.

3) The text offers a host of fully worked out examples, demonstrating the background, construction, interpretation, and evaluation of each model discussed. Both real and simulated data is used, with instructions on how to create and appropriately use synthetic data and synthetic models.

4) The book presents clearly stated guidelines on how to decide between models.

5) Examples are provided that use Stata, but with discussion of the capabilities of other major software applications, and how they deal with particular models and options.

6) At the end of each chapter, R code is provided that duplicates the Stata examples used in the text, whenever possible.

7) Numerous end of chapter questions are provided, with an accompanying Solutions manual that contains fully worked out answers to all 237 questions. Qualified instructors will be given the 186 page Solutions Manual free of charge.

8) A full chapter is devoted to the construction and interpretation of interactions, as well as their graphical representation.

9) A full chapter is devoted to the nature of binomial over dispersion and how it is defined, identified, and handled. Comparisons with count model over dispersion are discussed at some length.

10) Controversial topics such as the interpretation of odds ratios as risk ratios, goodness of fit tests for panel models such as GEE and QLS and other similar subjects are fully discussed.

11) The book includes a 29 page tutorial on basic data management, functions, and logistic modeling using Stata for those readers who are not familiar with Stata. Sufficient background is provided in order to fully understand the examples used in the text.

12) Some 60 data sets used in the text for examples and for end-chapter questions are provided for download on the text's web site. Download sites are given for the data sets in the following formats: Stata, Excel, SAS, SPSS, R, Limdep. The three most used datasets are also formatted to ASCII comma delimited, JMP, Systat, Minitab, and Statistica. An appendix identifies the location in the text for the first use of all data sets.

13) Over 40 User/author written commands are also available for download. Many useful commands and functions have been created to assist readers and researchers in their modeling and evaluation tasks.

14) The price is excellent price for a textbook this large (656 pages), especially one that is published in Chapman & Hall/CRC's respected "Texts in Statistical Sciences" series.

15) The book is authored by a leading scholar and professor in the area. Prof Hilbe teaches the "Logistic Regression" and "Advanced Logistic Regression" courses for Statistics.com, the foremost web-based continuing education site for professional statisticians and researchers worldwide.
0Comment| 11 people found this helpful. Was this review helpful to you?YesNoReport abuse
on September 1, 2009
An otherwise good book is marred by problems in exposition and editing. An example of the former is that the derivation of the IRLS algorithm in section 3.1 relies on the definition of the exponential family not given until section 3.2 and on properties of the Bernoulli distribution not given until section 4.1. Examples of the latter are errors in equations 3.6 (missing equals sign), 3.10 (dL/dBeta should be L), 3.12 (extraneous theta as coefficient of y), and 3.17 (inadvertent exchange of mu and eta) in the space of a just over a page in section 3.1. (The editing problems should gradually be mitigated somewhat as the publisher's errata page is updated, and if further printings and editions are published.)
11 comment| 8 people found this helpful. Was this review helpful to you?YesNoReport abuse
Joseph Hilbe not only is familiar with logistic regression, but has written much of the code used to do this in Stata and other analysis tools. In this book he presents the fundamental concepts of logistic regression, walks readers through examples of the most common types of analysis, and then explores variations and frontiers of rapid innovation and change. Hilbe's book has a new edition every few years, reflecting his grasp and discussion of the moving edge of the field.

The author's walkthrough style is one of the book's strengths. He uses examples based on the same data sets, allowing readers to benefit from a growing familiarity with the data's structure and idiosyncrasies. The presentation is technical and does require close attention. But the focus is practical, discussing frequently-encountered challenges and solutions. Hilbe acknowledges that there is no "right" answer to advanced regression problems, but steers readers away from wrong approaches and highlights the trade-offs of different strategies.

I encountered this text in an online course taught by the author. A natural procrastinator, I did assignments mostly during the last day before the deadline, so did not benefit fully from the discussion-board's give-and-take with Hilbe about our problem sets. This meant that I relied more on the text and the author's recommendations to other students that they look for answers in particular sections. While I would not have wanted to take the course using the book alone, I was impressed with how much it has to offer. It is a good text and an excellent reference.

A friendly warning--the text focuses on analysis using Stata and supports R with appendices and end-of-chapter R code. I attempted to use NCSS 8.0 for the class and did not have a good experience. In hindsight I wish I had become familiar with R--it's free!--before working with the book. The author does recommend this, but I did otherwise. It's my own fault for making it hard on myself.

Note to Kindle users: The Kindle version of this text is in .pdf format, so does not support all of the navigation features of a native-format Kindle book. Although it displayed well on my iPad, Kindle Fire, and Kindle for PC, it would not load on my iPhone. A message explains that this platform is not supported. Plan accordingly.
0Comment| 3 people found this helpful. Was this review helpful to you?YesNoReport abuse
on June 18, 2009
"Logistic Regression Models" appears to be the most comprehensive text on the subject, and it seems the author took great effort to write it in a manner that is as understandable as possible. There are many worked out examples for the logistic models discussed, ranging from binary and grouped logistic models, to ordered, unordered, and partially ordered categorical response models, panel models such as fixed, random and mixed effects models, GEE and QLS, survey models, discriminant logistic analysis, and exact logistic models.

Complete discussions are also given for the construction and interpretation of interactions, as well as seldom discussed topics such as overdispersion and when logistic model estimates may be interpreted as risk ratios instead of odds ratios. Simulation methods are explained and used.

Stata is used with most of the examples, but R code is also provided, and the main examples are also available in STATISTICA datasets for those following the examples using STATISITCA software. I understand a SOLUTIONS MANUAL will be provided to qualified instructors who adopt the book as a textbook; this "Instructors Manual" provides completely worked out answers for the 237 questions found at the ends of the chapters.

This latest book from Joe Hilbe continues his many great books written over the past few years giving substance to all of the GLM variations needed in a multitude of "niches" in research and data analysis. A most valuable book on any statistician's bookshelf, as it contains methods not available elsewhere in any one source!
0Comment| 6 people found this helpful. Was this review helpful to you?YesNoReport abuse
on June 16, 2009
This is one of those rare books that skillfully combine
statistics, software, mathematics, history, and hands on
applications that are useful for beginning and more experienced users. I didn't know, for example, anything about Wedderburn's life, and I learned about it from this book. I continue to learn from this book about logit, glm, GEE, AIC, dispersion, QIC, and QLS and more and use this in my current research.
0Comment| 6 people found this helpful. Was this review helpful to you?YesNoReport abuse
on December 27, 2013
If you are a Stata user and want a definitive guide to logistic regression models, then this is the book.

Warning: if you are NOT using the version of Stata that this book is based on, you will need to convert what Hilbe has to the current Stata methodology. It's not a big thing though and the results are pretty much identical.
0Comment|Was this review helpful to you?YesNoReport abuse
on August 29, 2015
This book is a great companion to Professor Hilbre's new book, Practical Guide to Logistic Regression. While Practical Guide to Logistic Regression, offers great practical examples, Logistic Regression Models discusses logistic regression more thoroughly, Readers wishing to have a more in depth understanding of logistical regression should purchase this book and refer to it concurrently with Practical Guide to Logistic Regression.
0Comment|Was this review helpful to you?YesNoReport abuse
on July 2, 2009
Joe Hilbe is a Professor of Statistics at Arizona State University. I know him because we both teach online courses at statistics.com. Joe teaches logistic regression, advanced logistic regression, negative binomial models and also a courses with James Hardin in Generalized Estimating Equations (GEE, a technique for longitudinal data analysis). I teach Bootstrap Methods. Joe and Jim Hardin collaborate and they have coauthored some books. Joe wrote this text to be a more comprehensive, up-to-date and advanced text compared to its competitiors (e.g. Hosmer and Lemeshow). The book includes theory, methods and applications and is quite extensive. It also has an extensive bibliography. There are many examples and data sets and although there is a discussion of the history of computational methods and software, the real strength of the book is that R code and output from R programs is used to illustrate the applications allowing students to inexpensively replicate results and have a free software environment to apply the methods in their own research. Also, since the author had been a contributor to the STATA software for logistic regression models much of the analysis is also done using STATA which the author is an expert in.
11 comment| 3 people found this helpful. Was this review helpful to you?YesNoReport abuse
on July 4, 2013
Hilbe's book is very good. It has a bit too many pages of code for my taste, which could be relegated to an online site. The data sets are. (See [...] While the book is very good, it is getting a bit dated, since it does not move the treatment to a Bayeisan approach. Maybe that's just too hard, but see Section 4.4 of Congdon's BAYESIAN MODELS FOR CATEGORICAL REGRESSSION.
0Comment|Was this review helpful to you?YesNoReport abuse
on July 19, 2012
Hilbe has put together a very thorough and maybe more importantly accessible treatment of logistic regression. It is often difficult to find statistics books which successfully strike a balance between the necessary mathematics, applied interpretations, and (most rarely) accompanying syntax for popular stats software. In my opinion, Hilbe has managed to find this balance. The book covers all necessary topics with enough depth to satisfy the mathematically curious but in a way that most applied researchers with a solid background in basic regression will also appreciate. I found the straight forward interpretations of resulting models to be refreshing and wholly applicable to statistical write-ups for academic journals but also useful to the statistical consultant writing for a lay audience. Chapter 6 deserves special mention. This chapter deals with interaction terms in logistic regression models but can also be generalized to other nonlinear models such as counts and other forms of nominal/ordinal outcomes. Interaction effects are the subject of interesting theoretical propositions but are often poorly applied, incorrectly applied, and ultimately mis-interpreted in practice when used in nonlinear models. Hilbe provides up to date treatments of these effects, provides very detailed (read easy to follow) code for the popular stats programs of Stata and R. I cannot recommend this book highly enough as I think it will appeal to a broad audience inclusive of the statistically adept and self proclaimed statistically challenged.
0Comment| One person found this helpful. Was this review helpful to you?YesNoReport abuse