Friday, June 29, 2012

Searching for the Grail

For the past year I've been searching for the Grail, the R textbook that students need. Note, not an R textbook that students need, the. Here are two contenders.

Discovering Statistics Using R
Andy Field's Discovering Statistics Using SPSS was a great textbook (read the reviews at Amazon) with only one problem - SPSS. He has now fixed that problem by publishing Discovering Statistics Using R. You can get the general idea by watching his online statistics lectures.

This is a great book, probably the best single book on R I've read, and I know I'm going to be using it a lot. I will be recommending it to students next year, but as a background reference as it goes far beyond the needs of the courses I teach. And I also have some niggles with it. For example, chapter 4 on graphs concentrates almost exclusively on ggplot2, which was very useful for me, but I feel it would be better if the simpler examples at the later in the chapter came before the complicated Facebook example.
Also an obvious problem that should have been picked up at the editorial stage: colour="Red" doesn't work with two colour printing when everything is blue! In general, I love the irreverent writing style, but I felt that the musical analogies were overdone, and probably inaccessible to younger (sorry Andy) and overseas students. The writing is a bit lumpy too - e.g. the sudden swerve from the "I'm just one of the lads" style into the intricacies of multiple regression models. I'm delighted to have an R-based textbook with both in, but the writing could do with a bit of smoothing out. R Commander seems to appear from nowhere at regular intervals with insufficient explanation. Personally, I avoid external shells when I'm teaching using R, but if I was going to use one, it would be R Studio these days. Something for the next edition I hope. It would also be nice to have functions listed in the index, e.g. to be able to quickly find out where lm(), glm(), etc, occur in the text.

Getting Started With R Our second contender is Getting Started With R, An Introduction for Biologists. This is a different kettle of fish, a slim 100 pages (at a not so slim price). Again, I learned some useful tips from this book (especially regarding plotting error bars), which (usefully) concentrates on generating graphs in R. The big downside is a devastating sentence in chapter 5, the only general statistics chapter, which says "If you have never actually used these statistics before, or never taken a course in statistics ... now is a good time to go and do this".

Right. Not suitable for a statistics course then.

Still searching for the Grail.

No comments:

Post a Comment