1382 BOOK REVIEWS
ADEQUACY OF SAMPLE SIZE IN HEALTH STUDIES. the t-distribution is more appropriate, a simple
Stanley Lemeshow, David W. Hosmer Jr., Janelle adjustment, recommended by Snedecor and
Klar and Stephen K. Lwanga published on behalf Cochran,' which slightly increases the sample size,
of WHO by Wiley, Chichester, 1990. No. of pages: will provide a more adequate estimate.
xii + 233. Price: E17.50 This book provides a useful tool for health care
researchers who are concerned about a large popu-
As stated in the preface, this book expands a 1973 lation. Pollsters who survey public opinion and
document of World Health Organization (WHO): pharmaceutical statisticians who conduct post
Adequacy ofsample Size (HSM73.1). It is designed marketing surveys may also benefit from this
to help health workers and managers in determin- book. The rich collection of 40 or so examples with
ing sample size in taking sample surveys or in well worked-out solutions provide practitioners
conducting epidemiological studies. with excellent illustrations to follow. On the other
This book contains three parts. The first part hand, clinical and pre-clinical researchers who are
describes the methods with well worked-out exam- used to analysing data with analysis of variance,
ples. The second part gives the theoretical back- regression analysis or t-tests may find only limited
ground in statistics and epidemiology without in- applications. With the general availability of
volving mathematics beyond high school algebra, microcomputers, a software package covering
and also contains 79 briefly annotated references all the methods discussed would be a natural
in its bibliography. The third part contains 17 enrichment of this book.
sample size tables for the various methods. In summary, this book covers both applications
The methods described in this book focus on and theoretical backgrounds for sample size deter-
population proportions, both parameter estimates mination in epidemiology and sample surveys. The
and hypothesis testing. Out of the eight chapters of book accomplishes both goals well. It is a must for
methodology, seven are devoted to population public-health workers and a nice add-on to the
proportions and only one chapter to continuous libraries of statisticians and health research organ-
response variables. Under the general subject of izations.
population proportions, one- and two-sample situ- A few typographical errors were found on page
ations, case-control and cohort studies, incidence 72: Table 11.3 should be 11.4 and in the first and
rate, lot quality assurance, simple random, strati- third paragraphs, n1 and n, should be m , and m,,
fied, and cluster samplings are discussed separ- respectively.
ately. This book does not cover continuous vari-
ables in sample surveys. Continuous response F. Y. HSIEHand ANDREWA . LIU
variables, however, are also used in sample sur- Computer and Statistical Services
veys and in community intervention studies with Anaquest, BOC Health Care
various sampling techniques. For completion, Murray Hill, NJ 07974, U.S.A.
continuous response variables should have been
included. REFERENCE
Since epidemiological studies and survey sam-
pling usually involve large populations, all sample 1. Snedecor, G. and Cochran, W. Statistical
size methods discussed in this book are based on Methods, Iowa State University Press, 1980,
the normal distribution. For a small sample where p. 104.
STATISTICAL METHODOLOGY IN THE PHARMACEUT- pharmaceutical industry has been one of the major
ICAL SCIENCES. D. A. Berry (ed.), Marcel Dekker, employers of biostatisticians. Many other statisti-
New York, 1990. No. of pages: viii + 578. cians in academia, health care organizations, and
Price: $125. government agencies also associate closely with
the industry in their consulting, investigation, or
regulatory capacities. This book aims to provide a
This is, in my knowledge, the fourth book in the reference for 'applied statisticians and others who
Marcel Dekker Statistics Textbooks and Mono- use statistical method in their professional life', as
graphs series that is directly pertinent to the phar- well as a textbook for undergraduate or graduate
maceutical industry. Probably no other single in- courses.
dustry has received such attention from publishers Does the book fulfil the role of both a reference
of books in statistics. This is rightly so since the book and a textbook? How does the book differ-
