Statistics for Veterinary and Animal Science, 2nd ed.

Abstract

This introductory textbook on biostatistics and epidemiology is intended for students in the veterinary and animal sciences with “no prior knowledge of statistics.” Indeed it seems to be the only text on the market focusing on veterinarians. The 1st edition from 1999 was well perceived and widely reviewed (Can Vet J 41:143–144). The book has a clear structure; each chapter starts with a statement of learning objectives and then presents the various aspects of the 15 chapter topics. Biostatistical and epidemiological terminology and methods are presented in plain language and without rigorous statistics. Numerous examples explain applications from a veterinary or animal science perspective. Each chapter concludes with exercises in the form of quizzes and small projects. This 2nd edition has grown by about 50 pages and now includes a CD-ROM with data sets used along with the examples throughout the text. There are 2 new chapters on Further Regression Analysis and Evidence Based Veterinary Medicine as well as additional sections on Sequential and Interim Analysis, Meta-Analysis, and Multivariate Analysis. The 1st edition was too short on regression analysis — a central concept in statistics. The additional chapter on Further Regression Analysis explains the basics of multiple regression models, has tripled the space from the former 1-page entry to logistic regression, and reviews advanced concepts like conditional and ordinal logistic regression. This chapter also introduces Poisson regression and mentions the important concept of clustered data, which are ubiquitously found in veterinary epidemiology. Another regression method included in this new edition is the Cox proportional hazards modeling approach, which is a key factor in the analysis of survival time data. This has extended the section on Survival Analysis from 1 to 4 pages and pays tribute to the importance of survival analysis in the health sciences. Other minor modifications to this edition are additions to the former text that often appear in small print, and which indicate topics in the book that can be left out during the 1st reading. I would have liked to have seen the new section on Meta-Analysis incorporated into the new chapter of Evidence Based Veterinary Medicine and an additional account on the Funnel Plot to assess publication bias. I understand, however, the authors’ point of view of considering Meta-Analysis as a special study type and its placement in the respective chapter on Further Aspects of Design and Analysis. The full understanding of statistical methods results not only from reading a textbook but in addition, the application of the learned methods. The CD-ROM with data sets included in this new edition, therefore, will allow the reader to follow book examples using statistical software. The data sets are given in 4 formats: ASCII, Excel, SPSS and Stata. Care must be taken when importing the ASCII and Excel files, however, because the data files are annotated and cannot be used readily. The Stata files may be read without problem. It would have been useful for the novice to have the Stata or SPSS code for the examples included on the CD-ROM. Furthermore, for those without access to Stata or other commercial software, a reference to the statistical freeware package “R” (www.r-project.org) would have been valuable. Overall, the book is very welcome as both an introduction to the basics of statistics and an overview to some more advanced statistical methods for veterinarians.