p. armitage and t. colton (eds): encyclopedia of biostatistics
TRANSCRIPT
Genes, Brain and Behavior (2007) 6: 304 # 2007 The AuthorJournal compilation # 2007 Blackwell Munksgaard
Book review
Encyclopedia of Biostatistics
P. Armitage and T. Colton (eds)
John Wiley and Sons Ltd, 2nd edn., 2005. $3775, eightvolumes, lxxxiv þ 6022 pp. ISBN 0-470-84907-X
Some of the most frequent mistakes made in submitted
manuscripts (and, unfortunately, also in many published ar-ticles) concern matters of statistics and statistical inference. A
classical example is the situation where researchers testa treatment in two groups of animals (say, males and females),
obtain a significant effect of treatment in only one group andsubsequently conclude that their groups react differently to the
treatment. (Of course, such a conclusion would only bejustified if the interaction termof a two-way analysis of variance
– ANOVA – would be significant.) Although such fundamental
issues should be addressed during the training of researchers,this is often not the case: whereas psychologists generally get
a good foundation in statistics during their education, biologistsoften do not, and they frequently seem to regard statistics as
just a necessary nuisance to be used solely to pacify finickyreferees. To complicate matters, the statistical knowledge of
referees (and editors) is often not much better than that ofauthors. In short, therefore, an encyclopedia like the present
one should be part of the library of any institution engaged inbiological or biomedical research.
The present work is of an astonishing completeness.Subjects covered vary from pure statistical theory (but always
with practice in mind), through generally used biomedicalmethods and experimental designs, to biographical items on
prominent statisticians (geneticists will recognize several ofthem, from Francis Galton to Sir Ronald Fisher). True, some
methods are not presented. The Hayman–Jinks ANOVA fordiallel tables is an example (the words ‘diallel cross’ are not
even listed in the index), but much as I personally like thisparticular method, it has to be admitted that it is only rarely
used any more nowadays.
The encyclopedia is very general and covers all of thebiosciences, including, for instance, agriculture. Nevertheless,
geneticists will find a wealth of information on subjects such aslinkage and segregation analysis, path analysis, twin analysis,
etc. To avoid giving the impression that this encyclopediawouldbe of use only for researchers working with humans, I have to
mention that even a topic such as ‘animal screening methods’can be found in the detailed subject index. All general statistical
techniques that are commonly, or even less commonly, used inbehavioral neurogenetics are covered. This varies from detailed
treatments of multivariate methods such as the much-abuseddiscriminant and factor analyses to lemmas discussing reliable
workhorses such as the t-test, ANOVA and nonparametricstatistics like the Mann–Whitney U-test.
Although some lemmas are, by necessity, rather technical,others are also understandable for researchers less familiar with
statistical theory. The lemma on P values and statistical signif-icancemerits to be read by anyone involved in analyzing research
data, be it as an author, journal referee or grant reviewer.The price is steep (but at $0.50/page not excessive), so
not many individual researchers will be able to afford thiscarefully produced encyclopedia. However, I hope and trust
that this Encyclopedia will find its way into many institutionallibraries and would like to encourage any student or
researcher to make the trip to their libraries to consult itcarefully and often.
Wim E. Crusio
Universite de Bordeaux ICentre de Neurosciences Integratives et Cognitives
CNRS UMR 5228
Bat B2 – Avenue des Facultes33405 Talence
FranceE-mail: [email protected]
304 doi: 10.1111/j.1601-183X.2006.00300.x