Book Reviews

To understand God's thoughts we must study statistics, for these are the measure of his purpose (Florence Nightingale)

Statistics, as you know, is the most exact of false sciences (Jean Cau)

Almost all human life depends on probabilities (Voltaire)

With seasonally adjusted temperatures, you could eliminate winter in Canada (Robert Stanfield)

The mathematicians are a sort of Frenchmen: when you talk to them, they immediately translate it into their own language, and right away it is something utterly different (Goethe)

If your experiment needs statistics, you ought to have done a better experiment (Lord Rutherford)

Oratory is dying; a calculating age has stabbed it to the heart with innumerable dagger-thrusts of statistics (W Keith Hancock)

Probabilities direct the conduct of wise men (Cicero)

These quotations, which highlight the generally held ambivalence toward statistics, are several from a larger set on the back cover of this slim volume entitled Version 4 (Health and Sickness) of Essentials of Statistical Methods (HSESM). The book admirably meets its modest goal: namely, to provide a cheap, clearly written and concise discussion of those methods of statistics and probability that are typically taught in a good one-semester introduction to statistics.

HSESM is meant to supplement the voluminous textbooks that are typically prescribed for such courses. Such textbooks, besides usually providing a wealth of examples and of problems useful for creating assignments, often contain substantial sections of extraneous material that serve mainly to differentiate one textbook from another, but that also allow authors to gallop their hobby horses several times around the track before a captive audience.

The student who would find HSESM useful would be one taking a calculus level introduction to statistics and who has purchased a standard textbook. HSESM, which is intermediate between a fact or formula sheet and a full-blown textbook, would serve as a readable and well-organized résumé of course material. A student who had a set of course notes that had the material and organization of HSESM would indeed be fortunate. Perhaps this suggests that lecturers might find this volume a useful addition to their bookshelves.

HSESM is divided into the three areas customary for introductions to statistics: descriptive statistics, probability and statistical inference. The teaching of descriptive statistics is likely to be greatly influenced by the availability of good software and computing environments, something that is not emphasized in HSESM. The section on probability requires some knowledge of the calculus, a point that does not seem to be made clear in the preface. The section on statistical inference is interspersed with a number of insightful comments and discussions on various aspects of inference that are subject to some controversy. Specific reference to p values would have been of value to those who might need to interpret published analyses of health sciences data. The author regards the word "data" as being singular; although he is not unique in this usage, I find it somewhat distracting.

A possible addition to the list of quotations mentioned would be the following:

Statistics means never having to say you're certain (American Statistical Association T-shirt)

This volume would "almost certainly" be a valuable purchase by a student of introductory statistics.

Ian B MacNeill
Cancer Bureau
Laboratory Centre for Disease Control
Health Canada, Tunney's Pasture
Address Locator: 0602E2
Ottawa, Ontario K1A 0L2

Order directly from
Rumsby Scientific Publishing
PO Box Q355, Queen Victoria Building
Sydney, New South Wales 2000
Australia

This book examines the social determinants of health from the viewpoint of the emerging "society-and-health" paradigm, comprising a series of reviews written mainly by social scientists. Schor and Menaghan begin by describing family pathways to health. Patrick and Wickizer address community and health, presenting the CIAR (Canadian Institute for Advanced Research) model of health determinants and referring to both A New Perspective on the Health of Canadians and (without attribution) Achieving Health for All. They offer a very comprehensive tabular summary of the relevant literature, noting the lack of evidence for effectiveness of the "Healthy Communities" approach and (to a lesser extent) of community organization.

King and Williams examine the relationship between race and health, rejecting the concept of race as a social construct and concluding that "racial" differences in health can be explained by confounding variables. Walsh, Sorenson and Leonard address sex and health, with specific reference to cigarette smoking. They find that sex is also socially constructed, and recommend that future studies of the relationship between sex and health should use a broad society-and-health "lens," encompassing social processes like the division of labour and the distribution of power. Some of this material is very heady reading for the lay (non-behavioural scientist) reader.

The chapter by Marmot, Bobak and Davey Smith on social inequalities and health offers a breather for the epidemiologist reader: a very clear presentation in a more familiar style. Brenner departs from the established format by presenting the results of his "economic-change model of population health" of the relationships between mortality and business cycles. The sensitivity of the model is awesome, but makes one wonder about overspecification. He also offers the first hint of policy solutions, by admitting a role for government action in moderating the health-damaging effects of economic trends.

Johnson and Hall offer a fine analysis of class, work and health, incidentally illustrating what an insightful sociologist was Karl Marx. Corin's chapter on culture is unfortunately inaccessible to this reader. Bunker, Frazier and Mosteller offer a quantitative assessment of the contribution of selected preventive and curative medical care to life expectancy (1.5 years and 3.5-4 years respectively, with the potential for an additional 7-8 months and 1-1.5 years respectively), along with semiquantitative estimates of their impact on quality of life.

Miller closes the volume with a chapter on thinking strategically about health. He categorizes the social determinants of health as based in resource position (absolute wealth), relational position (social roles, respect) and relative position (inequalities). Public policy must work to reduce resource inequalities and to achieve solidarity through narrowing economic distances and involving people in common pursuits (e.g. local decision making). The case is strong, but it is hard to have much hope in the current political climate.

The introductory chapter by the editors (three social scientists and a physician) and the closing commentary provide some synthesis, but there is not a lot to tie the various chapters together. Personally, I would like to have seen more on possible solutions, to balance the detailed analysis of the problems.

The book exemplifies the cross-fertilization but also the tension between epidemiology and social science. Conventional epidemiology comes in for considerable criticism, especially for its focus on individuals, individual diseases and one-to-one causal relationships, and for not studying issues like the functioning of the tobacco industry (which seems to this reader a more appropriate topic for political science and the law). A recurring theme is the need to distinguish variables that act at the individual level from those that act at the community level, and to treat them appropriately.

Society and Health covers much the same ground as Why Are Some People Healthy and Others Not? by Evans et al., although with a sociological slant as distinct from the political economy of Evans et al. (Michael Marmot and Ellen Corin contribute to both books). So which one to read, if you can only read one? Most readers will find the Evans book more accessible, broader and more immediately relevant to Canada. But try to read both, since they are fairly complementary.

Overall rating: Good

Strengths: Authoritative, comprehensive, up-to-date

Weaknesses: Heavy reading for the non-behavioural scientist
Not much integration or synthesis

Audience: Epidemiologists, policy makers

Robert A Spasoff
Department of Epidemiology and Community Medicine
University of Ottawa, Faculty of Medicine
451 Smyth Road
Ottawa, Ontario K1H 8M5
[Currently on sabbatical at Universiteit van Amsterdam]