on April 15, 2004
This is an excellent book on the subject. It covers the subject in detail. I work in a quantitative group of a large financial institution. My boss asked me recently to make a presentation on logistic regression, a subject I knew nothing about at the time.
Thanks to this book, I was able to develop the knowledge and expertise in this esoteric statistical method to make a descent presentation to my group. I was also able to develop basic logistic regression models in Excel.
Personally, unlike many of my colleagues (Russians with formidable backgrounds in math) I don't have an advanced knowledge in math or statistic. The book reviewed and taught me what I needed to know to grasp logistic regression. This included an excellent review of logarithms and partial derivatives.
If you are interested to learn this subject, and you need the equivalent of a most user friendly "Logistic Regression for Dummies" this is the book for you.
By the way, logistic regression is not so well known today outside of academic and quantitatively oriented circle. But, after studying it, I feel like it will take over linear regression big time. It is such a superior method to analyze situations associated with human behavior and human choices. Whether you are a psychologist, social scientist, or developer of financial products you will sooner or later have to grasp logistic regression because your counterparts and competitors will. And, in the statistical evaluation of decision-making, linear regression just can't compete with logistic regression. This book will help you catch the logistic regression wave.
This slim book provides a meaningful introduction to the concepts of logistic regression: what it is, where it outshines linear regression, significance of results, and more. But since this is aimed at users of statistical packages (STATA, SPSS, etc.), it doesn't go into the computational details of parameter estimation or much else beyond hand-calculator levels.
The appendix on logarithms should tell you that this doesn't aim at readers comfortable with (or in need of) a peek under the covers at how things actually work. So decide what your goals are. If you want to gain confidence in meaningful use and interpretation of canned packages, this could be very helpful. But, if you're more a tool-maker than tool-user, this could be disappointing.
on July 6, 2009
This book is a good, quick introduction that will allow you to jump right into logistic regression. I found Pampel's writing to be comprehensible and quite suitable for anyone with a good working knowledge of regular linear regression.
All in all, you could do far worse if you need a basic overview into LR. Think of this as a good book to let you know what you need to know. It'd also make a good reference if you need a primer on logs.