|Home | About | Journals | Submit | Contact Us | Français|
Introduction to statistics for biology, 3rd edn.
RH McCleery, TA Watt, T Hart. 2007.
Boca Raton, FL: Chapman & Hall/CRC. $62.95 (paperback). 296 pp.
The population of textbooks in statistics for undergraduate biology students has grown rapidly over the last decade. I have a sample of a dozen published since 1999 on my shelf, and this is not counting titles originating from outside the UK or aimed at specific subject areas. They all cover the same basic territory but vary in the range of analyses, exposure of the underlying mathematics or extent of linkage to computer packages. The demand for statistics textbooks for biology students must be quite large, but it is becoming increasingly well supplied.
Assessing the place of Introduction to statistics for biology in this crowded market is somewhat presumptuous. It started life in 1993 as Trudy Watt's Introductory statistics for biology students; survival to a third edition is evidence that it has appeal. By its own admission the new edition has changed substantially. It has been revised to emphasize the underlying concepts common to many of the analyses, post hoc tests have been excluded, there is fuller coverage of non-parametric statistics, and a trial copy of Minitab 15 software has been included.
My dozen did not include this edition or its predecessors. I came to it as an ecologist responsible for teaching statistics to a large class of biology students. On reading the preface it appeared the authors shared my general philosophy. They emphasize the value of statistics as a tool for working biologists, the need to develop an understanding of the underlying ideas (but not ‘serious’ mathematics), and the redundancy of most hand-worked calculations in the age of user-friendly computer packages. Notwithstanding the latter, students are warned that they must grasp the concepts in order to point the software in the right direction and make sense of its output. With so much in common I hoped the book would turn out to be an ideal support for my teaching.
The new edition provides a coherent account of the statistical ideas underlying the common tests. It succeeds in drawing attention to the linkages between procedures. The authors advise that it should be read from cover to cover, which is quite the opposite to some of its competitors. Dipping in would be challenging because the assumption is made throughout that preceding sections have been read and understood. Dipping in could also be frustrating because indexing is one of the book's weakest points. The index seems to have been largely constructed from words emboldened in the text, leading to entries such as ‘More common than you think’. There is no ‘Q’ entry for ‘quartile’, but it appears under ‘L’ for ‘lower quartile’ and ‘U’ for ‘upper quartile’. The entries for ‘randomisation’ and ‘replication’ do not refer to the principal sections in which these are addressed. There is no entry for ‘pseudoreplication’ or ‘repeated measures’, which are dealt with in the text, albeit briefly.
Despite the promotion of the computer rather than the calculator there are several places where the text resorts to hand calculations. One of these is estimating confidence limits for group means by using the residual mean square in ANOVA, rather than group standard deviations. This prompted me to look at Minitab's calculation of confidence limits and discover that it is not consistent. The authors may also be unaware of this; they show later, without comment, the output from Minitab's ‘Interval Plot’ facility, which uses the group standard deviations. Overall, the integration with Minitab is good but occasionally it lapses into command line control, which students would rarely contemplate, and uses tables of statistics when Minitab yields precise probabilities. I doubt if the text provides sufficient support to allow someone new to Minitab to carry out analyses, although an appendix provides an introduction to the software.
Parametric tests are covered up to and including factorial ANOVA designs, but the line is drawn before orthogonal contrasts and ANCOVA. It is good to see power analysis included, although I am not sure why it is relegated to an appendix. Correlation and regression are explained well, particularly the analysis of residuals. There is nothing on multivariate methods, which is understandable in an introductory text. Non-parametrics are covered at length in a separate section rather than alongside their parametric equivalents. There is sound advice for students on project management in the final chapter, although a good supervisor will cover much of this in a way that matches the arrangements of a particular department.
Overall, the standard of writing is high, which is important in a text covering material that many students find difficult or unappealing. Unfortunately there are a few errors in prominent places. There is no glossary of terms, an absence that compounds the shortcomings of the index.
It seems appropriate to end with a statistic. In my sample of competitors the new volume lies just above the median in terms of pence per page (excluding appendices). By this crude measure of value for money it is fairly typical. I provide my students with a subset of my sample as suggested reading, encouraging them to select the one that best suits their interests and their pockets. I am not sure that I will be adding this one to the list. Whilst admirable in many ways, I doubt that it will work well for more than the upper fraction of a class of mixed ability and motivation.