Work

Medical Statistics — a common sense approach

Year: 2000
Type: book-review
Abstract: by M. J. Campbell & D. Machin. Published by J. Wiley and Sons, 1999, Price :pound; 16.99 ISBN 0-471-98721-2 I had not read the earlier editions of this book and my previous reading has been of traditional texts such as ‘Berry and Armitage’. ‘Medical Statistics’ therefore came as something of a surprise. The emphasis has dramatically shifted from how to do a statistical analysis to when to do it. Reflecting upon this change it is of course perfectly logical as users of statistics no-longer have to spend hours with paper, pencil, slide-rule or even pocket calculator. A myriad of computer programs has put complex statistical analysis within the reach of all of us. Problems of incorrect analysis have been replaced by excessive and often inappropriate analysis. The layout of the chapters follows this intention of the book classifying according to the problem faced rather than the class of test to be used. The second surprise on reading the book was just how much has been squeezed into 200 pages. Not only the basic techniques are covered but also some more exotic items such as ROC curves, logistic regression, cluster randomization, and Cronbach’s alpha. The text also includes sections on questionnaire design and analysis, areas often not covered in main-stream statistical texts despite these being commonly used tools. The frequent examples illustrate how different statistical techniques can be used to answer practical questions and more importantly how to choose between apparently similar tests. An additional feature useful to ensure that readers of the scientific literature are catered for as well as investigators is the paragraph at the end of each chapter ‘points when reading the literature’. The book ends with technical appendices providing the mathematics for those who really feel the need and a series of MCQs. I have always associated MCQs in books with readers intending to sit an examination and it is difficult to imagine what the target readership could be facing. Have I any criticisms? Yes, but they are very minor with some terms such as multiple regression and analysis of covariance being introduced (page 91) without explanation and other concepts such as the difference between interval and ratio scale being unduly laboured. In addition certain of the examples were not particularly helpful with the description of the table top chemical analyser (page 24) leaving me no wiser. The chapter on statistical inference ignores the current trend away from hypothesis testing. Although confidence intervals are mentioned their position in the text implies that they are of only secondary importance. Overall this is an excellent book which I would recommend to a wide audience. In particular for those who have to make sense of the literature but don’t want to learn statistical methodology this book will direct you to the key points to look for in papers and the common pitfalls. For investigators performing analyses this book will not replace your software manual but it will direct you to the correct test for a particular application and remind you of the assumptions implicit in using it. Nearly all consumers of statistical analysis will benefit from reading this book at some stage.
Author: P. R. Jackson
Language: English
Open Access status: green
OpenAlex ID: W1951908143
Cites:
Cited by: 95
Related to: 10