Making numbers add up to lie detection

Many sociologists are uncomfortable with quantification and statistics, but even those who align themselves with the qualitative side of the methodological paradigm divide need a basic understanding of what the other side is all about. There are times when statistics simply cannot be avoided. We all know that they can be used to tell lies as well as access the truth, so it could be argued that we all need to equip ourselves with basic lie-detection methods.

Understanding Social Statistics is aimed at undergraduate sociology students, but it would also be useful to mature seekers of statistical expertise. It is well written, attractively laid out and uses a language that demystifies the subject, instead of cloaking it with obscure technical terms. Both authors are sociologists, and this book is based on years of teaching social statistics to sociology students. The experience of transmitting statistical understanding to students, most of whom are probably not much interested in statistics, has undoubtedly influenced the book's tone, content and accessibility.

Two characteristics of this book make it unusual. The first is its practical, how-to-do-it emphasis. Chapter two, for example, discusses in suitably basic, pragmatic terms the use of computers to analyse data. It tells students how to distinguish between different kinds of computers (not all you need to know, but all most of us need to know) before moving on to consider SPSS as the most commonly used social science package for data-analysis. The acronym SPSS used to stand for Statistical Package for the Social Sciences, but now apparently means Statistical Products and Service Solutions - a much less attractive and less memorable, but presumably more marketable, name. The book offers what adds up to a short course on SPSS, which students are likely to find very useful. Here and elsewhere, the authors stress the value of statistical procedures in helping social scientists to answer questions about the real social world.

A second novel feature is the use throughout of the position of women to provide illustrative examples of such questions. Unless I missed it on my quick reviewer's read of the book, there is no explanation for this - but no matter, examples need to come from somewhere and it makes for a more consistent read to find them all coming from the same place. Most of the examples used come from two databases: the UK General Household Survey and the Social Indicators of Development dataset collected by the World Bank and containing data about 171 different countries. In both cases, the subsets of data drawn on have been put on the worldwide web so that students can do their own analyses.

Otherwise Understanding Social Statistics contains a more conventional catalogue of statistical terminology and procedures: univariate and bivariate analysis, measures of central tendency and dispersion, sampling and influence, and techniques for modelling data, including multiple regression and loglinear analysis.

Inevitably, the level of engagement with statistical concepts and manipulations required at the end of the book is considerably higher than the one needed at the beginning. While this may be enough to drive some students back into the qualitative stable, they will at least have a much better idea of what they are retreating from.

Ann Oakley is professor of sociology and social policy, Institute of Education, University of London.