Another book on statistics for biologists. What will it be this time? Cookery book instructions without indication when to use or why, or a mathematical treatise beyond the ability or needs of the apprentice biologist? David Heath has produced a rare phenomenon in biological publication, a readable book on quantitative aspects of the subject.
Biologists work in an increasingly experimental and quantitative discipline but generally come from a less mathematical background than other natural scientists. A course in experimental design and statistical analysis is an essential part of modern biology courses but must be presented in a way that is neither patronising by assuming no mathematical skills nor should it be filled with erudite, largely irrelevant theory. The combination of experimental design and subsequent analysis in one text, delivered by a working biologist, is a happy medium. The deliberate omission of computing techniques, another difficult area for many biology students, and the restriction of the formulaic necessities to exemplary boxes, leaves a readable book with a practical basis to satisfy a beginner.
What is good about this book? It has evolved from a course for biologists by a biologist and contains real examples from field and laboratory experimentation. Physiological and behavioural choice experiments, blood-cell types, leaf phenology, dog-whelk distributions on rocky shores, vegetation surveys and growth plots, simple genetics experiments etc., just the type of examples the student will meet in his course. Most importantly the author introduces the student to flow diagrams for developing experimental design and systematic recording practices for experiments which include not only the observations but also the questions and hypotheses and decisions that lead to the choice of appropriate statistical treatments.
The author does not attempt too much but deals fully with his chosen topics. He explains what are meant by variables, populations and samples in the field and the laboratory, and their limitations. An understanding of normality and normalising transforms gives a basis for deciding whether an estimate is real or not and the often neglected idea of precision is also treated. Comparisons between one, two, three and more "populations" are dealt with and thence to analysis of variance. An important area for field biologists is the introduction of non-parametric tests. For the first time for me the mysterious chi-squared statistic becomes understandable, rather than just used.
I like the use of boxes and the checklists in the final chapter. The diagrams and photographs keep the reader in touch with real problems. The glossary is full and useful and the bibliography while short, is pertinent, contains reference to the indispensable Freshwater Biological Association Publication No. 25. by J. M. Elliot dealing with non-parametric data.
The book does not delve into multivariate techniques or other complex statistical techniques but if everything presented is understood, the next advanced steps and ideas will be a cinch.
Tony Andrew is lecturer in environmental sciences, University of Ulster.
An Introduction to Experimental Design and Statistics for Biology
Author - David Heath
ISBN - 1 85728 131 4 and 1 85728 132 2
Publisher - UCL Press
Price - £40.00 and £14.95
Pages - 372