Customer Reviews

R for SAS and SPSS Users (Statistics and Computing)

5 star:		(9)
4 star:		(0)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

I've used and taught SAS and SPSS since about 1982. It seems to me that much of the new statistical developments are coming out in the open-source R language, rather than business-prediction software like SAS or SPSS. The number of new statistical packages in R is rapidly increasing, including packages supported by high quality textbooks. SAS and SPSS offer "business intelligence" -- software to help businessmen predict the future -- rather than cutting-edge tools for serious research.

There are many good books for R experts, and good beginners books are starting to come out. Before Muenchen's book, there was nothing to help the experienced SAS/SPSS programmer learn R. Since R is object-oriented, it "thinks" quite differently from SAS and SPSS, and you spend as much time unlearning old ways of thinking as learning new ones.

The author of R FOR SAS AND SPSS USERS knows how SAS/SPSS programmers think, since he is one of us and has spent decades at UT teaching people to manage and analyze data in SAS, SPSS, and other software. This makes his explanations seem intuitive and natural without the "one hand clapping" feeling you get from R "help" messages. The book is not only a good introduction but it goes into considerable detail to cover basic and intermediate R programming. The style is simple and lucid. Unlike some R material, the book is rich in concrete examples. Each chapter has 3 tables of similar code in SAS, SPSS, and R. These tables may help it serve as a "lookup book" during programming.

I keep a text file of the book's examples open in my editor when I write R code so that I can cut and paste working code from the book rather than doing trial and error on minor details. This same cut-and-paste approach works with SAS, SPSS, and other software.

If you have some years with SAS or SPSS and you want to learn R, this will be your #1 book.

I'm the author posting a few other reviews. You can read many other reviews and download the example programs at the book's web site, r4stats.com. Cheers, Bob Muenchen

"I think the hands down best intro for R (and I have the Dalgaard and Gelman and Hill books) is R for SAS and SPSS Users. The thing that sells this one is that most people who want to get into R are already users of either SAS or SPSS. What Muenchen does is to track what you would normally do in those apps with how to do the same thing in R. That means he has to explain why R does things (often perversely) the way it does and he guides you to packages in R that replicate SAS and SPSS routines very closely."

-Tracy Lightcap, PhD, Professor and Chair, Department of Political Science, LaGrange College

"One of the most clearly written, well designed, books I've read on a programming language (of any variety or type) in my career. And as a computer scientist, I've read quite a few! You seem to have a knack for guessing ahead of time what problems R users will potentially have and explaining to the reader, without talking down, how to get around the problem."

- Andreas Stefik, PhD
Central Washington University
Department of Computer Science

"I've used SAS for 16 years and have found the transition to R to be fairly difficult. This book has helped a lot. It's well organized and I've found myself turning to it as a go to source for how to get things done. The online documentation for R is probably its weakest characteristic and you need a book like this. In all other respects I have found the book quite useful and would buy additional books by the author if they were available."

-David Young, Director at Crisbal Company Limited

"In order to learn R quickly, I would suggest the following sequence: read An Introduction to R, followed by R for SAS and SPSS Users."

- Robert I. Kobakoff, Ph.D., Author of the web site
Quick-R for SAS/SPSS/Stata Users
(Muenchen suggests the reverse order, naturally!)

"With the integration of R and SPSS beginning with version 16 via the R Plug-In, this is a timely book for SPSS users...This book does a great job of leveraging prior knowledge of SPSS (or SAS) to get users started in making the best use of R. R documentation tends to be written by experts and for experts. This book is written by an expert for beginners."

- Jon Peck, Technical Advisor and Principal Software Engineer, SPSS Inc.

"Honorable Mention: Best R Book for existing SAS/SPSS users.
This is an excellent book on doing statistics with R - even if you are not an experienced user of SPSS or SAS. But if you ARE an SPSS or SAS user, this book will really excite you - it puts everything in R in a familiar context, and will help you get going with R much faster than any other book we have seen."

-Human Landscapes

"So you decided to cut down on your statistical software expenses and decided to get R, but the problem is you know SAS/SPSS and you need to learn R fast enough to justify switching over. The ideal book for you is R for SAS and SPSS Users. ...It's a really easy book, you have the SAS Syntax, the corresponding SPSS Syntax and the R Syntax."

- Ajay Ohri, Author of DecisionStats

This book is absolutely excellent. The focus is on the data manipulation and processing that goes on before analysis. As a longtime SAS user, this is the major stumbling block for me using R. The parallels and discrepancies across the languages are clearly pointed out with solid code examples. The book covers basic syntax but more importantly it goes way beyond saying this is the syntax for an "if" statement in SAS and this is an "if" statement in R. The author goes into the important fundamental differences in how the two languages think about and process data.[...]

There is also very good coverage of R graphics (especially the set of functions in ggplot2 that are wildly useful and rarely mentioned in other books). The coverage of statistics is limited to only one chapter. So, do not get the book if you only want to learn the ins-and-outs of R stats. Happily that chapter covers the most commonly done statistics. So even in its short presentation it should help everyone.

While the book is geared toward someone with experience in SAS or SPSS, I think it would be excellent for anyone learning R. The links to the point and click versions of R (R commander, Rattle or JGR) are invaluable for anyone starting out.

The author is actively maintaining the book's website. So be sure to grab the errata and his notes.

I wanted to write this book but Robert Muenchen did a much better job. The people who developed S at Bell Labs and then the group that copied it for R didn't think much about the messy data of real studies in research and industry. Also working with row-based packages like SAS and SPSS puts one in a certain mindset. Muenchen successfully guides people in this mindset to take advantage of the flexibility of R.

The sequence of chapters lets a person start running R, access and modify data, and then graph or analyze. He answers questions that come up in daily work such as selecting variables and observations, changing names, and handling missing values. Too many books and classes show interesting functions without prioritizing their use. If it is beyond the scope of the book, he refers to websites and other books.

The connection to SAS and SPSS made this book like a bright light for me. For example for years I struggled with the idea of a 'factor'. But when he explained that it was like a 'formatted variable' it clicked. He keeps the other languages in the background so I think someone who didn't know them could still use the book.

I have a few small criticisms. There are a few editing errors in the book. Someone looking to do a specialized analysis would be better off with a different resource since the statistics section is short. The ggplot package looks interesting but a more extensive traditional graphics section would be more useful.

This book really is a superb reference for looking up how to do things in R. As an experienced SAS user - an ordinary guy using statistics for work, not a statistician - who recently branched out, I found that R's very different mindset made for a formidable learning curve. My discovery of this book flattened the learning curve dramatically and has saved me dozens of hours. I found the book to be a far more accessible treatment of R than other resources and I have little doubt that those coming to R from backgrounds other than SAS or SPSS will similarly find it valuable. Although it is worth reading the book cover to cover, sections are structured so that it is easy to jump in wherever some help is needed. The table of contents effectively points the way to major topics and the index is implemented well. Explanations are clear and examples are abundant: Muenchen generally shows multiple ways to accomplish the same or similar tasks. These varied approaches not only help cement understanding of how R works, but give the reader an abundance of models from which to work.

I've worked at learning R language off and on for about 2 years, with limited success. As is well known, there is much help available for R, but it assumes that you already know about it. Not the best way to learning something. I have found Robert Muenchen's to be a godsend in learning R. Most of us in the statistics and psychometrics areas know and use SAS and/or SPSS. Robert Muenchen saw a need of people who knew SAS/SPSS to learn R and created his excellent book. His concurrent explanation of how these three different languages do the same operations is the Rosetta Stone for learning R. I strongly recommend this book to anyone who desires to learning R, especially if they have some familiarity with SPSS and/or SAS. I also compliment his putting an abbreviated free version of the book on the internet. I found that quite useful, too. Plus, it encouraged me to buy his book.

I am a competent beginner in SAS and SPSS but don't have any programming background and very little math knowledge (words like matrix and vector aren't part of my vocabulary.) Despite this, I decided to try to learn R since it's free. I started with "A Beginners Guide to R" and "Introductory Statistics with R." While both are well written I couldn't grasp the unfamiliar R concepts without some corollary to something familiar (i.e. SAS or SPSS.) Muenchen's book is just what I needed. I highly recommend it for users with backgrounds similar to mine.

Thank you for writing this book! I've been a SAS programmer for around 15 years, a SPSS programmer for around 4 years, and I also program in VBA, and coming to R from this background has been an exercise in frustration. R is just so different in how it handles data, and its command syntax, I found that having previous programming experience was not really an advantage. I was wondering about the value of this book as I already have the MASS book at home and The R Book, and it was the need to recode a lot of variables was the activity that made me purchase this book.

While other books give emphasis on how to do particular statistical and graphing techniques, they tend to omit details on how to import and manipulate variables and observations in order to undertake the statistical analysis. I find that data preparation is around 90% of my analysis time, so not having this information has a major effect on my productivity. This book covers all that missing detail, as well as some facets of statistical analysis as well. The chapters and sections are well laid out in a logical sequence, and the bonus for the kindle is being able to search for terms.

Robert Muenchen is a good writer as well: plain English explanations are given along with the code. He also gives examples of equivalent SAS and SPSS code so you can see the differences between them and R.

If you are coming to R from a SAS or SPSS background, even if you have other R reference material, I recommend you purchase this book.

I've used SAS for 16 years and have found the transition to R to be fairly difficult. This book has helped a lot. It's well organized and I've found myself turning to it as a go to source for how to get things done. The online documentation for R is probably its weakest characteristic and you need a book like this. In all other respects I have found the book quite useful and would buy additional books by the author if they were available.

Disclaimer: I work for SAS. I have been using SAS since 1978. I have been using S/S-Plus/R since 1993. I have not used SPSS since 1982. Most of my "R experience" is with S-Plus. SAS has added functionality to allow users to exploit R within SAS, so I was looking for a book to recommend to SAS customers. I will reluctantly recommend this book. The book was published before SAS released ODS statistical graphics, so while the R material is sound, the SAS graphics used for comparison are not up-to-date.
I will address the R for SAS users aspect of the book. If you are a SAS "power user," you will probably be disappointed in the book. For example, there is no material on converting from SAS date, time, and datetime variables, formats, and functions to the same functionality in R. The timeDate package supplies much of this functionality for R. To be fair, Muenchen's book would be thousands of pages if he tried to address every feature in every package available to R users.
I like the SAS Output Delivery System (ODS). I can control almost every aspect of the results that I want to capture or display. Muenchen gives very sparse treatment to display capabilities of R related to exporting or publishing the results. For example, how do I display the p-values from a regression with a precision of six decimal places? I can also use SAS formats as a mechanism for searching. SAS formats essentially provide an implementation of binary search. I could not find any material on this capability for R. SAS also supports hash tables. I found no reference to this capability in Muenchen's book. SAS power users often exploit the use of formats and informats. I am still not sure how to accomplish most of the common SAS formatting capabilities using R. Muenchen discusses reading data files using R, but quite frankly, if you have both SAS and R, you would almost always use SAS to read the raw data and to convert it to a form that is easy to read into R.
As one reviewer states, professionals use SAS and SPSS for data processing and analysis related to well established business practices. Experts use R for access to the latest technology even if it has not been tested. Of course, R users also want access to traditional methods as well. If someone invents a new predictive model, it can take years for the new methodology to find its way to SAS or SPSS, whereas an enterprising researcher might code the methodology into an R package and make it available immediately. Researchers routinely offer to give (or sell) their code to companies like SAS and SPSS, but SAS and SPSS would still have to spend time and resources making sure the resulting software is up to standards. Technical support would have to be trained on the new software. Documentation would have to be written, etc. Anyone who has used R knows that there is great variability in the quality of the software and the quality of the documentation. Muenchen is a little too optimistic about the quality of R software and r-help technical support. The tested quality of R relates to well established methodologies, not the newer methods that have just been introduced. That R gets ANOVA right is a must, but that doesn't mean it provides the same quality for all functionality.
One reason for professionals using SAS rather than R can be seen by looking at the examples that have SAS code followed by the equivalent R code. The SAS code is never larger than the R code, and is often quite smaller. With the added R code comes added control, a plus for power users. Unfortunately, some of that added control, like customizing output, is not well covered in Muenchen's book.
One of the reviewers suggests that the reader can also use the book to learn SAS if they already know R. I would strongly discourage this. The SAS examples are weak, and much of the functionality of the SAS Data Step is missing. I fear a user who learned SAS this way would be like a driver who learns to use a manual transmission but never shifts out of first gear.