9 of 9 people found the following review helpful:
3.0 out of 5 stars
Examples -- Just the Examples, April 11, 2000
This review is from: Logistic Regression Examples Using the SAS System, Version 6, First (Paperback)
I purchased this book because I needed to do a large number of logistic regression runs for my dissertation. It does an excellent job in going through all the SAS code you need in order to write good logistic regression equations. However, I was disappointed by the very limited discussion surrounding the application of the models. Fortunately, it does provide useful references.
The book is useful because it goes beyond the SAS user manuals in explaining how to program logistic regressions and what SAS's output is describing. However, if you do not already have a good understanding of logistic regression, (i.e. you had one lecture on it in your research methods class) you might be better off with something along the lines of Logistic Regression Using the SAS System: Theory and Application by Paul Allison. (I have no affiliation with Dr. Allison.)
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
6 of 6 people found the following review helpful:
3.0 out of 5 stars
3 1/2 Typically Terse; Mostly for Experts, March 9, 2005
This review is from: Logistic Regression Examples Using the SAS System, Version 6, First (Paperback)
Generally speaking, logistic regression is a statistical technique that tries to explain or predict a dichotomous outcome (e.g., two levels, as in yes/no, succeed/fail, heal/don't heal) from a set of independent variables. For example, we could code quitting smoking as a '1' and not quitting as a '0'. Your outcome (a 1 or a 0 on smoking) may be related (in this hypothetical example) to frequency of smoking, smoking history, physiological and/or psychological features, treatment, or other factors. Logistic regression is used both to describe and to predict such outcomes. The book states this definition once, though not as clearly as the preceeding, then takes off into the mathematics:
"This logistic regression equation models the logit transformation of the ith individual's event probability, p(subscript i) as a linear function of the explanatory variables in the vector, x (subscript i)."
The text gets much more difficult than that.
The SAS Institute's manual on 'Logistic Regression' is most useful for people who already understand a great deal of the rationale and the statistics behind logistic regression. For them, the book's main advantage is its explanation of printed output, and coverage of several related topics. Most importantly, it provides problem examples related to logistic regression, and annotated SAS programs to solve them.
Even the easiest sections of the book assume at least a college-level background in stats, and many sections seem to require post-graduate expertise. There is very little explanation or teaching going on here, and the chapters on various log. reg. applications are simply too brief to learn from. At its most general level, then, it's a user guide for understanding output. For the the very experienced, it's a reference for interpreting output, as a template for writing task-specific SAS log. reg. programs, and for understanding and choosing among various SAS logistic regression techniques. Even experienced users will find some of the explanations lacking, however. For example, the chapter on ROC analysis is out of date, and cites two refences only (the most recent of which is from 1982). Given that this is a growing statistical area, this is just not enough. Other areas are give adequate, although short explanations, and would be quite useful for someone wanting a quick reference.
The book provides nine sample datasets dealing with comsumer choice and medical data. Topics and "examples" include fitting a binary logistic regression model, computing confidence limits, computing customized odds ratios, computing predicted probabilities and classifying observations, creating classification tables, using model selection methods in logistic regression, computing fit tests, producing regression diagnostics, correcting for overdispersion, displaying an ROC curve, ordinal responses, 1:1 matched Data, N:M matched data, fitting interactions, estimating discrete choice probabilities with a mulitnomial logit model, probit analysis for estimating an LD50, and fitting a Bradley-Terry model for paired comparisons. Hopefully, the preceeding list will give you an idea of whether this book is right for you.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
0 of 3 people found the following review helpful:
3.0 out of 5 stars
This is a great start, April 14, 2000
This review is from: Logistic Regression Examples Using the SAS System, Version 6, First (Paperback)
To start a career in clinical biostatistics field, with minimal computer background, this book is a great help !
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No