Content area
Stat Papers (2009) 50:673675
DOI 10.1007/s00362-008-0131-6
BOOK REVIEW
Phillip I. Good (2005): Introduction to Statistics Through Resampling Methods and R/S-PLUS (Paperback)/
Phillip I. Good (2005): Introduction to Statistics Through Resampling Methods and Microsoft Ofce Excel (Paperback)
Wiley-Interscience, 248 pp., GB 34.50, e 59,90, US $70.50, ISBN: 978-0471715757/Wiley-Interscience, 231 pp., e 62,90,
US $74.95, GB 34.50, ISBN: 978-0471731917
Wolfgang Polasek
Published online: 2 April 2008 Springer-Verlag 2008
The author of the book is described by AMSTAT, where he held courses for practitioners, as follows (http://www.amstat.org/chapters/detroit/tc2001.html
Web End =http://www.amstat.org/chapters/detroit/tc2001.html ): Dr. Phillip Good (Ph.D. Berkeley) is from Information Research, a California based rm that provides product testing services and statistical consulting services. Phillip Good has over 30years experience in aerospace, computer, medical devices, pharmaceutical and petroleum industries and is the author of two popular textbooks on resampling methods.
Indeed, this book has the ability to be popular among practitioners. It is written in a special style that is rarely seen for statistical textbooks. Probably we should dene a new word for such attempts to explain what certain elds of sciences are doing. It is a type of translation of the scientic jargon, that statisticians use everyday but is pidgin to laymen. Trans-meta-scription or trans-popularizing scientic results might be appropriate. Thus the book aims to explain empirical reasoning that has emerged out of statistical thinking in the last decades.
Statistics has the reputation of being incomprehensible for many students of descriptive sciences. When new sciences start to become empirical, then suddenly researchers are confronted with inductive thinking in a rigorous abstract and deductive language, called mathematics. Many people want to know why empirical work has to be so complicated. I guess this book has found a way to introduce laymen into the challenges of empirical reasoning. One problem is that chance comes in many disguises when empirical problems have to be put on a statistical test bench.
The author tries to open the door to statistics through resampling by the way of the computer programs R and S-PLUS. Resampling seems to be an intuitive way to explain the classical, frequentist interpretation of statistical inference procedures.
W. Polasek (B)
Institute of Advanced Studies, Vienna, Austria e-mail: [email protected]
123
674 Book Review
Chapter 1 is called Variation and explains how to get a feeling for data coming from populations and shows how to use the computer programs R and Splus e.g. on bootstrapping. Chapter 2 continues with Probability and explains permutations, the binomial and conditional distributions. The third chapter is on Distributions and is rather short, since it covers only discrete and continuous distributions. Chapter 4 explains 3 types of Hypothesis Testing, one sample, tow-sample tests and on correlations. Chapter 5 describes Designing an Experiment or Survey. It explains how to formulate hypotheses, calculate sample sizes and sequential sampling.
Chapter 6 explains Analyzing Complex Experiments and covers a large spectrum from testing dose responses, rank test to categorical tables. Chapter 7 is on Developing Models and explains bivariate und multivariable regression.
Chapter 8 is on Reporting your ndings and gives a list of useful dos (and some donts) including a warning to use p values: is this a useful advice, given all the computational conveniences of R and Splus? Chapter 9 discusses Problem Solving in the sense of solving practical problems that occurs in the course of data collection. An Appendix explains briey the interactive version of Splus.
In summary, the book is a nice attempt to give honest advice, offers an introduction to statistical methods across different subjects (non-parametrics, biometrics). The book alone will be not enough to give a novice in statistics enough condence to make an analysis on its own. He would need more books and skills in the high-level program languages. A book like this one should give a warning that one cannot become an expert in statistics after 200pages of introduction.
Now to the meta-aspects of the book: The chosen approach (using R or Splus) gives a fresh start to statistics, but it can not serve as a real short cut to become a statistical expert. The user has to be familiar with a good deal of mathematical or computer science thinking, like matrices, vectors or arrays. Simple statistical problems can easily turn into complex monsters that even expert nd hard to solve. But introductory books like this one help to widen the circles of educated clients who understand the problems of statistics. It also shows the practitioners the wide range of possible answers in statistics. The exercises of the book are a good supplement, only the solutions (or at least some of them) are missing.
Introductory statistics books have not yet found a drawing line to separate the swimmer and non-swimmer zone: What area in statistics can be safely done after an introductory course and what expeditions need an experienced guide. Book like this one should also give advise on good practice and should present standard examples that beginners could follow. But it should give some caveats as well. Unfortunately, there seems no agreement even among experts on these issues.
Finally, I found this an interesting book, but more need to be done to close the gap of understanding between the growing body of statistical knowledge and the laymen battling with simple questions: p-values or signicance levels for my data set?
Notes on the Excel-Version of the book.
First of all, the 2 version of the book, briey called Excel and R versions, are equally structured, only the gures and the explanations how to run the examples are changed. Thus, the main comments from above on the approach of the books to explain statistics stay the same. Only, I found the Excel version much less appealing. First of all, it takes a day to install all the add-ins into Excel. And everybody who has
123
Book Review 675
done similar things knows that these add-ins dont work at the rst time. Thus you are already exhausted with nuisance problems before it comes to apply the ideas of statistics and resampling. Some add-ins had to be run with other add-ins that are only available on the Ofce installation disk etc.
Thus I found the R-approach much easier and much more straightforward to explain the main statistic ideas. This seems surprising, given all the beauties of spreadsheet calculation you can do in Excel, which is very closein data operationsto descriptive statistics, but has no features to store these ndings in a modern statistical way. Maybe it will take another generation of spreadsheet programs to marry modern basic statistics with book-keeping, spreadsheets and cell-calculations.
Also, there is a big advantage in using R: It is free of charge, while for Excel version you have to pay for at least 3 programs if you want to do run and keep the examples for longer than in 1month. The trial versions are only free for 1month.
Furthermore, when doing the Excel-embedded programs, I was never sure, if the program was doing the right thing. It was much harder for me to think of ways as how to verify the results I got in Excel than in R.
Summarising, I recommend the R approach: I guess this opinion will be shared by many professionals, and students who were exposed to modern informatics concepts and statistics in some university courses. For some practitioners this might not be the case, they feel more familiar with the Excel version. Some universities offer introductory statistics courses based on Excel. I certainly do not support this approach, because Excel misses all the elegance of modern statistics, and given all the problems I had in running the examples of the Excel-book, my opinion did not change.
PS: Interestingly, the Excel version of the book gets a 2-star rating while the R/Splus gets a 3-star rating at amazon.com.
Reference
http://www.businessbookmall.com/free-stuff-statistics.htm
Web End =http://www.businessbookmall.com/free-stuff-statistics.htm
123
Springer-Verlag 2009