Customer Reviews

Applied Logistic Regression Analysis (Quantitative Applications in the Social Sciences)

5 star:		(4)
4 star:		(1)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

As its title suggests, this book is an excellent guide to using logistic regression in data analysis. I purchased this book because I needed to do several logistic regression runs for my dissertation. It turned out to be an extremely useful book for two reasons. First, it presents logistic regression alongside more traditional ordinary least squares (OLS) models. Therefore, if you already have a good understanding of OLS models, this book is very easy to follow. Second, its discussion of logistic regression issues in the context of SPSS or SAS makes it very easy to follow along with your own data analysis as you move through the book. Since statistical packages are always improving, this does date the book a little. However, this is a very minor concern. I believe Dr. Menard is to be commended for including issues regarding popular software packages in this work.

When compared to SAS's documentation, this book's greatest advantage is explaining in english (rather than mathematical notation) the assumptions and limitations of SAS's (and SPSS'S) algorithms. Its chapter on logistic regression diagnostics is alone worth the price of the book. In short, if you need to use logistic regression analysis and you already understand OLS, you cannot go wrong with this book.

I bought this book to teach myself logistic regression after buying a much much more expensive text . If you've had the experience of trying to learn a stats technique on your own then you know that you'll probably need more than one book. If I could go back, I would buy this one first and then move on to other more expensive and comprehensive texts. I had a good grasp of multiple regression already and found this book's orientation to logistic regression, done by drawing parallels with multiple regression, very understandable. It was easy to read cover to cover and gave great explanations of the background math, without being at all heavy with formulas. If you are taking a logistic regression course and are having a hard time following the explinations in the text assigned for the class, this would likely provide a good alternative for helping you grasp the concepts.

A good, cheap overview of logistic regression analysis.

I bought and I'm glad I did, but I don't refer to it like I do Hosmer and Lemeshow's text.

Menard's little, green Sage paperback is an excellent introduction to logistic regression analysis. In spite of its brevity, it also serves well as a reference, including off-beat topics such as how to compute standardized regression coefficients for logistic regression equations. Moreover, some of the usual output of SPSS logistic regression runs would be uninterpretable, and commonplace questions would be unduly difficult to answer, if it were not for Mendard's text and its effective use of examples of SPSS output.

Before I bought Menard's introduction, I tried to improve my understanding of logistic regression, including proper interpretation of unstandardized coefficients and various measures of goodness of fit, with the first edition of Hosmer and Lemeshow's Applied Logistic Regression. Compared to Mendard's book, however, Hosmer and Lemeshow's presentiation is tedious, plodding, and needlessly dense. Apparently it was written for an audience to which I do not belong.

I use logistic regression fairly often, and I have yet to encounter an issue that I couldn't address through reference to Menard's Applied Logistic Regression Analysis. The explanations are clear, the formulas are easy to follow, and the examples are instructive. An awful lot of useful information is packed into one brief and inexpensive document.

Prof Scott Menard must be commended for writing an excellent book on Logistic Regression. Explaining it in the context of commercially available software packages is a very good idea. I was able replicate some his analysis using SAS on the data set used in this book (available on line from ICPSR, Univ of Michigan).

I eagerly await the next edition of this monograph. Thank you!

Logistic regression analysis is an analogue to multiple regression--with the dependent variable normally (but not always) pitched at the dichotomous level (just two values). It is a pretty sturdy statistical technique, not demanding a lot of assumptions about the nature of dependent and independent variables. This book is not terribly accessible (I get headaches in some sections), but it is a very nice brief introduction to the subject.

The book begins by noting the rigorous assumptions for multiple regression to "work." Logistic regression does not demand so much of the data. Then, step by step, the book lays out the results from logistic regression and how to interpret these. There is no "explained variation" (multiple R squared) as with multiple regression. There is, however, a pseudo-R squared. Other key outputs? Predictive accuracy of the model; Wald's number; Model Chi Square; Goodness of fit; log-likelihood; standardized and unstandardized coefficients; and on and on. The book also lays out a set of diagnostics, to see if there are any threats to accepting the results of the statistical analysis as legitimate (e.g., outliers).

The book is NOT an easy read. But if you want a brief introduction to the subject, this book is quite helpful. At times, I think, matters could be explained more lucidly, but--overall--this is a nice contribution.