This review appeared in Anthropological Forum pp. 118-119. Reproduced here with the permission of the author.

Statistics for Anthropology, by Lorena Madrigal. Cambridge University Press, Cambridge, 1998, xiv, 238pp., tables figures appendices, references, index. ISBN 0-521-57786-1. (paperback).

PETER HISCOCK
Department of Archaeology and Anthropology
The Australian National University

Any volume that facilitates clear measurement and analysis of quantitative data in Anthropology is to be applauded. This book fulfils that role by being a well-written introduction to the subject. In the early chapters terms and symbols are defined, making the material unambiguous even to those students with minimal numeracy. Virtually all the statistics commonly used in Anthropology are dealt with, including measures of central tendency and dispersion, data transformation and analysis with z scores, comparison of means with t-tests, ANOVA, non-parametric tests (primarily Mann-Whitney U, Kruskal-Wallis, and Wicoxon signed-rank tests), linear regression, Pearson and Spearman correlations, and the chi-square test. To my mind this is a good selection of the most useful procedures needed in the tool-kit of every researcher. The presentation of material is supplemented in three ways. Firstly, in each chapter there are exercises/questions, with the answers provided in Appendix A. This is probably a benefit, although the amount of work involved in answering these questions is very uneven between chapters, and the questions are irritatingly placed in the middle of chapters rather than at the end, interrupting the presentations. Secondly, reference is made to the processing of statistics with a particular software package (SAS - the Statistical Analysis System). Since most readers will probably not have this software I think this is perhaps not helpful. Thirdly, there is a solid collection of tables in Appendix C (z scores, t-distribution, F-distribution, U values, distribution, Wilcoxon signed-ranks T values, and Pearson r critical values). These tables are valuable in interpreting the statistics, and make the book continuously useful.

This book is certainly useable, and I have no hesitation in recommending it. However, for reasons I will explain, this book would not be my first choice as a text since it contains a number of unsatisfactory features.

First, necessary aspects of some statistics are missing. For example, the discussion of the Chi-squared () statistic makes no reference to or Cramer’s V. This is important because a common misunderstanding of the test sees it as an indicator of the strength of a pattern, whereas it does no such thing. The test merely indicates whether an association between variables is likely to exist, not whether such an association is strong or weak. Other measures such as Cramer’s V measure the strength of the association. In the absence of even a mention of Cramer’s V, I would be worried that the misunderstanding would be repeated in yet another generation of students.

Second, there is a very limited discussion of graphs and their interpretation, something that would be extremely valuable. There is no description of powerful graphical forms such as box and whiskers graphs. More puzzling, Madrigal includes an advocacy of those appalling graphs called pie diagrams, in preference to more coherent and easily interpreted graphs.

Third, there is a minimal discussion of data collection and no discussion of procedures of sampling in the field, the nature and adequacy of samples, and the implication of sampling decisions for analysis. Obviously the author could not be expected to include all matters in a book this size, but the omission of these matters does seem regrettable.

Forth, the framework for significance evaluations involves the development of null hypotheses and their evaluation/rejection using set significance levels (e.g., 0.05), rather than the simpler and more realistic evaluations of significance advocated by statisticians such as Tukey, and now being widely employed in archaeology. This follows normal conventions in some other social sciences, but is a much less friendly framework for undergraduates, and to my mind generally unrealistic. I would much rather train future researchers to use the sliding scale of significance explained in Drennan’s (1996) fabulous introduction to statistics.

In conclusion, this is a good book on statistics for Anthropology, but I believe there are better ones available at the moment.

Reference:

Drennan, R.D. 1996 Statistics for Archaeologists. A commonsense approach. Plenum Press.