Customer Reviews

SAS for Linear Models, Fourth Edition

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

 

 

This is the fourth edition of this primer on linear models by Littell and Freund. As with the previous editions it gives a thorough treatment of the models and their application through the use of SAS procedures. In the fourth edition, Walter Stroup has been added and Phil Spector dropped as coauthor. Also this edition drops the word "System" in the title after "SAS". The main addition to this edition is the inclusion of a lot of detailed information on the application of mixed effects models through use of the MIXED procedure. There is a great deal of comparison of the differences between PROC MIXED and PROC GLM in the analysis of these models. Many interesting examples are presented and care is taken to show the user how to specify the models so as to get the appropriate analysis. They also teach you enough to aid in interpreting the output.
Littell, Milliken, Stoup and Wofinger have also written a very nice book titled "The SAS System for Mixed Models" and my only question would be to ask "which book offers more?" This fourth edition seems to now cover many of the same topics that highlight that book.

Readers should be aware of the two books and should investigate for themselves the differences and overlap before deciding to purchase either one. One clear difference is the date of publication. This book, published in 2002, is more current and has several references from 1997 and after whereas the mixed models book was published in 1996.

I give this book two stars from an econometrician's point of view. This book focuses on linear models related to ANOVA and hence may have a good audience in disciplines such as statistics, biostatistics, etc, but it has ABSOLUTELY nothing useful for people interested in econometrics or financial statistics.

For example, the book doesn't even mention heteroskedasticity at all, and there's absolutely no time series coverage. In econometrics we also do weighted regressions a lot, and this book doesn't have anything.

I wish the book title had been more descriptive. Linear models are many stripes and used in many different disciplines. By using such a catch-call title, the book is misleading.

So, if you want to do linear models in economics or finance, do NOT buy this book.