Review

Source: Biometrics, Vol. 44, No. 1 (Mar., 1988), pp. 318-319

Published by: International Biometric Society
Stable URL: http://www.jstor.org/stable/2531933
Accessed: 21-03-2016 05:41 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact [email protected].

Wiley and International Biometric Society are collaborating with JSTOR to digitize, preserve and extend access to Biometrics.

http://www.jstor.org

This content downloaded from 129.96.252.188 on Mon, 21 Mar 2016 05:41:15 UTC
All use subject to JSTOR Terms and Conditions
 318 Biometrics, March 1988

that the treatment is rather formal, but considered the COX, C. P. A Handbook of Introductory Statistical

book to be clear, well-written, and accurate if regarded Methods. Wiley, New York, 1987, 272pp. $34.95/

as an introduction to statistical methods rather than on ?30.50. ISBN 0-471-81971-9.

medical statistics.

The author's aim is to give an economical summary of

This fourth edition maintains the previous approach

basic statistical techniques, with emphasis on method-

and is intended to appeal to the same audience as

ological aspects, to supplement verbal instruction in a

before: the advanced undergraduate and beginning

first course for graduate students intending to apply

graduate student and the health professional in need of

statistical methods in their researches. The compact

a reference book on statistical methodology. Expansion

form of presentation and terse style result from his

and clarification of certain discussion sections has been

attempt to rid the student of the practice of note-

undertaken, and there is greater emphasis on computer

taking-he suggests that this attenuates understanding

applications. The number of student exercises has been

and hinders the whole information transfer process.

increased still more; brief solutions are supplied for

No calculus is used in the book. Nor is a computer.

odd-numbered exercises. The addition of a dozen or so

"Unveridically" simple data are widely used in illustra-

large (n , 1,000) univariate/bivariate/multivariate

tive examples, with the intention of allowing accurate

data sets, which seem plausible, and from which

computational details to be mastered rapidly.

individual students are asked to sample, will further

Normality assumptions underlie much of the book.

enhance the appeal of the book to instructors and

Some 150 short sections are grouped into 11 "chapters,"

lecturers.

going from basic concepts and procedures to matrices,

multiple regression, and multiple correlation.

Even students whose lecturer does not have the

DUNN, 0. J. and CLARK, V. A. Applied Statistics:

author's particular orientation could find this a useful

Analysis of Variance and Regression, 2nd edition.

revision aid.

Wiley, New York, 1987, 445pp. $36.95/f31.95. ISBN

0-471-81269-2.

RIZVI, M. H. (ed.) Modern Statistical Selection,

The aim of the first edition (1974) of this book was to Parts I and II. American Sciences Press, Syracuse,

be both a comprehensive, self-contained textbook, and New York, 1985/1986, 202pp. (Part I), 249pp. (Part II).

also a useful reference work for applied research. Its $49.75 (Part I), $49.75 (Part II). ISBN 0-935950-13-3

presentation of analysis of variance and regression in- (Part I), 0-935950-14-1 (Part II).

volved only very elementary algebra, and avoided math-

These volumes are in fact issues 3 and 4 for 1985 of the

ematical derivations and proofs. Reviewers generally

American Journal of Mathematical and Management

welcomed the authors' explanations of the intuitive

Sciences, and issues 1 and 2 for 1986. They commem-

bases of the techniques discussed in the book, and their

orate the thirtieth anniversary of the first published

guidance on the interpretation of numerical results. One

papers on selection and ranking procedures by publish-

reviewer described it as "an excellent book for those

ing some of the papers presented at the Conference on

who wish to know how to use the techniques of regres-

Statistical Ranking and Selection held at Santa Barbara

sion and analysis of variance without grasping the

in December 1984.

underlying mathematics." Another considered it to be

In Part I, the work of Robert Bechhofer, Shanti

"a highly competent, lucid, and relatively compact vol-

Gupta, and Milton Sobel, on the indifference-zone

ume"; however, he questioned the use of complicated

approach, Gupta's formulation in subset-selection

symbolisms, subscripts, and multiple summations, and

theory, and the parallel between the selection of the

thought that "perhaps most will find it more useful as

best amongst k populations and the "central tendency"

a reference than as a first introduction."

problem, is discussed in papers by these authors and

In their second edition the authors have adhered to

papers about these authors. Part II contains some one

their previous mathematical level and to their slightly

dozen research and review articles, mostly by authors

dry literary style. Besides a summary, references, and

whose work is less well known and sometimes contro-

problems for the student, each chapter now contains a

versial, on topics such as inverse sampling, multinomial

section entitled "With the Computer" which discusses

selection, indices of diversity, nonparametric selection,

widely used computer programs in relation to the sta-

multiple comparisons associated with ranking and se-

tistical content of the chapter. The illustrative examples

lection, selection with preliminary testing, profit/sales

have been obtained by simulations based on research

selection of firms, computer software, and related

problems in the literature; these, like the student prob-

topics.

lems, cover many fields of interest.

Chapters 1-4 give a brief introduction/review.

Chapters 5-9 describe fixed-effects models, whilst

SINGH, D. and CHAUDHARY, F. S. Theory and

Chapter 10 deals with infinite variable effects and mixed

Analysis of Sample Survey Designs. Halsted Press,

models. Chapter 11 contains new material on repeated

Wiley, New York, 1986, 380pp. $26.95/f23.50. ISBN

measures designs. Linear regression and multiple regres-

0-470-20266-1.

sion (including polynomial regression) are studied in

Chapters 12-14. Covariance analysis is covered in This is a manual for those involved in project planning

Chapter 15. The final Chapter, 16, examines various and data analysis, and a textbook for graduate students

techniques for screening data before or during analysis, in areas such as business management and agricultural

a topic which the authors consider to be so vital as to sciences. It combines Singh's previous collaborative

warrant presentation in a single chapter positioned as Hand Book on Sample Surveys with Chaudhary's lec-

conspicuously as possible. ture notes for advanced undergraduates and graduate

This content downloaded from 129.96.252.188 on Mon, 21 Mar 2016 05:41:15 UTC
All use subject to JSTOR Terms and Conditions
 Book Reviews 319

students. The authors have attempted to give a system- examine a broad spectrum of procedures, such as the

atic exposition of existing theory up to and including estimation of quantiles, proportions, and means,

modem developments important in survey work, to skewed distributions and goodness-of-fit tests, estimat-

extend the application of the theory to field data, and ing the mean and variance from censored data, outlier

to provide references to relevant recent literature. detection, and control charts. A listing of a lengthy

A major appeal of the book is the combination of FORTRAN 77 program is given for implementing many

attractive price with good format, clear type, and illus- of the trend and seasonality techniques discussed in

trative examples using actual data from recent surveys Chapters 16 and 17. The emphasis in the book is on

in India. estimation rather than hypothesis testing. Nevertheless

certain standard nonparametric tests are described (with

very little formal basis) in the final chapter.

ABBOTT, L. A., BISBY, F. A., and ROGERS, D. J.

Taxonomic Analysis in Biology. Columbia Univer-

sity Press, New York, 1985, 336pp. $40.00 (cloth),

NADDEO, A. (ed.). Italian Contributions to the Meth-

$16.50 (paperback). ISBN 0-231-04926-9 (cloth),

odology of Statistics. Societa Italiana di Statistica,

0-231-04927-7 (paperback).

Cleup, Padova, 1987, 657pp. L.50,OOO.

The authors use the word taxonomy to denote the

This review, in English, of Italian contributions to

process of classification, rather than the result-they do

statistics in this century has been written by a number

not see a clear distinction between the terms taxonomy

of leading Italian statisticians under the auspices of the

and systematics. In describing both computer-aided and

Societa Italiana di Statistica, to celebrate the centenary

traditional methods, they walk a tightrope between the

of the foundation of the International Statistical Insti-

numerical school "who have believed that automatic

tute and the centenary of the first public meeting of the

analysis by computer ensures against the taxonomist's

ISI in Rome in 1887. The idea underlying the volume

preconceived biases regarding the proper organization

"was the fact that Italian statistical works are being less

of the taxa and therefore leads to objective classifica-

and less known abroad. The main reason being that

tions that can be produced rapidly and in standardized

Italian is not very well understood by most foreign

form," and those who "resist this view vigorously and

statisticians. In fact, it is not uncommon to find, in

scoff at the idea that a trained taxonomist can be

foreign statistical literature, the rediscovery of concepts

replaced by a computer." Their philosophy "is essen-

and statistical indices, which were born in Italy and

tially phenetic," and they emphasise that the characters

have for a long time been part of the cultural back-

that are useful for making classifications or keys need

ground of every Italian statistician."

to be carefully refined from the raw data. They consider

The work contains 19 chapters, some divided into

to be illusory the idea that there will ever be a single

several sections, dealing with various aspects of statis-

ideal classification for any group, and stress that the

tical methodology. Topics of interest to biometricians

special purpose of any classification should be clearly

include: dissimilarity; the treatment of qualitative char-

stated.

acters; multidimensional data analysis; and inference

This comprehensive and unique book is divided into

on linear and nonlinear models. Many little-known

four parts. Standard traditional methods are described

references to the Italian literature are given. No serious

and illustrated in Part I, whilst Part II considers formal

attempt has been made to relate the Italian contribution

aspects of data structures and mathematical models.

to work in other countries.

Part III examines in detail computer-assisted taxonomic

analysis; character analysis, agglomerative cluster analy-

sis, optimization cluster analysis, ordination, diagrams

U.S. DEPARTMENT OF COMMERCE. State and

of variation pattern, identification aids, and the princi-

Metropolitan Area Data Book. Bureau of the Census,

pal cladistic methods all receive attention, and many

Washington, D.C., 1986, 750pp. $28.00.

specific examples are given. Data base management,

taxonomic data bases, and goals for the future are Fifty federal and private agencies have contributed to

discussed in Part IV. this massive numerical thesaurus of information at state

level on topics as diverse as population and housing,

incomes and earnings, state rates and revenues, major

GILBERT, R. 0. Statistical Methods for Environmen-

crops and minerals, retail trade, energy prices, labor

tal Pollution Monitoring. Van Nostrand Reinhold Co.,

force, federal programs, leading manufacturing indus-

New York, 1987, 320pp. $49.95. ISBN 0-442-23050-8.

try, education, and election results. We quote, "Last

This is a textbook coveling standard statistical material published in 1982, the popular compendium offers

for nonstatisticians such as environmental scientists and 1,873 new and updated items for states, many with

hydrologists. Their interests are well catered for in the earlier data for comparison, 298 facts on each metro-

examples and exercises. The general layout is good, and politan area, and 86 items on metro area central cities."

answers to the exercises are provided. The accompanying literature indicates that this infor-

The first three chapters are introductory. Chapters mation is also available on computer tape and on

4-10 deal with field sampling designs. Chapters 1 1-18 diskettes.

This content downloaded from 129.96.252.188 on Mon, 21 Mar 2016 05:41:15 UTC
All use subject to JSTOR Terms and Conditions