Basic statistics and epidemiology: a practical guide
Article first published online: 2 AUG 2011
© 2011 The Authors. ANZJPH © 2011 Public Health Association of Australia
Australian and New Zealand Journal of Public Health
Volume 35, Issue 4, page 397, August 2011
How to Cite
(2011), Basic statistics and epidemiology: a practical guide. Australian and New Zealand Journal of Public Health, 35: 397. doi: 10.1111/j.1753-6405.2011.00751.x
- Issue published online: 2 AUG 2011
- Article first published online: 2 AUG 2011
By: AntonyStewart . Radcliffe Publishing , Oxford . 200 pages . ISBN 978-1-84619-411-5 .
Reviewed by Paul Agius
La Trobe University, Victoria
Elementary applied statistics texts are as common as the cold. Less common are well and clearly written examples, crafted in such a way as to allay the anxiety often experienced by the target audience of these texts. Basic statistics and epidemiology: a practical guide takes up this challenge and, to a large extent, succeeds.
This is the third edition of the book and it covers statistical topics that would be of interest to those with little knowledge of the basic principles of applied quantitative methods as they relate to social and health research. The book is made up of 31 chapters covering topics such as populations and sampling, power and sample size, levels of measurement, descriptive statistics, hypothesis testing, the normal distribution and probability. In terms of specific statistical methods, the book offers explanations of t-tests, analysis of variance (ANOVA), chi-square tests, bivariate correlation and linear regression. The book also has a section on epidemiological methods, and gives informative introductory explanations of cohort and case control designs, randomised control trials and common, discipline-specific, measures of association.
I think one of the challenges in writing introductory methods texts is the extent to which elementary discussion of fundamental statistical methods and topics is engaged in at the expense of sufficient theoretical and methodological depth to adequately equip the novice researcher; even at an introductory level. It is in this respect that I feel this book disappoints in some respects. The chapter on ANOVA is a welcome inclusion in this third edition; however I believe the lack of any discussion of repeated measures ANOVA (although the paired t-test is covered and the author does provide references for further reading on ANOVA), or post hoc/planned comparisons with respect to one-way ANOVA, are significant omissions. The failure to include a section on repeated measures design is particularly surprising, given the book is partly aimed at those working in epidemiological settings where a considerable proportion of research involves longitudinal or panel data.
Similarly, although the topic of theoretical assumptions underpinning statistical methods is raised briefly in the book, coverage of the topic is limited. The author discusses the assumption of equal variance between groups in one-way ANOVA, however other basic assumptions, such as independence of observations, the level of measurement and normality of the dependent measure, are not covered with respect to ANOVA and OLS regression. I feel the book would have benefited from a more specific focus on assumptions here, including elementary steps the reader might take when assumptions are not met (e.g. transformations of dependent variables with non-normal distributions in linear regression). Moreover, I found the lack of any discussion relating to missing data – what it is, the different types and how to treat it (i.e. imputation) – surprising and, to some extent, undermined the books applicability.
Despite these shortcomings, the book has much to offer the novice quantitative researcher and indeed the more experienced researcher with little experience of health statistics per se. In particular, I found the chapters covering epidemiological methods and measures of association informative. In the book, discussions of key and often misunderstood epidemiological measures, such as odds ratios and the various risk estimates (absolute, relative, attributable and population attributable), are outlined clearly and in most cases basic examples are used well to communicate the theory underpinning these measures. The author also provides exercises that cover a large proportion of the material outlined in the book and this is useful to further reinforce learning.
As I mentioned earlier in this review, it is difficult to find applied introductory statistical methods texts that are comprehensively written and sufficiently detailed, but, most importantly, also inviting to read for those anxious about learning statistics. This book, on the whole, achieves this and I would recommend it as a worthwhile starting point reference for those new to epidemiological methods or quantitative research generally.