Book Reviews 1657

around questions rather than addressing them (sample titles: “When unlike would be like”. “The
hypothesis of the comparable”, “.-\ stratification of hypothesis”). Yet those who are not put off
by the pedantic style and the need for a broad and deep background in philosophy can gain the
benefit of creative thought not readily available elsewhere. At least Riedl’s graphics often conve)
ideas readily when his prose is obscure.
This book is one of the few that combines biological and psychological inquiries without reducing
important questions to trivial or reductionistic shadows. Riedl mentions that his book is in the
tradition of Freud, Jung. Piaget. Vollmer. Brunswick, Chomsky. Popper and Lorenz. Readers fa-
miliar with physics will see parallel inquiries in other authors. Prigogene. for example. Riedl is
unique in beginning with the major philosophical and epistemological questions, where others begin
with natural science and draw. secondarily, philosophical and epistemological implications.
Although this book is valuable and constitutes a step forward, I could not have found it so useful
without having read several other authors. Primary among these are Lumsden and Wilson whose
book Genes, Mind, and Culture: A Coevolutionary Process (1981) has a preliminary chapter that
addresses epistemological questions similar to Riedl’s in a more readable, more biological. more
data-based. but less philosophical manner. Neither Lumsden and Wilson (in this book) nor Riedl
offer offensively reductionistic sociobiology. Another book to read before Riedl, Hofstadter’s
Giitiel. Escher and Bach (1980). esamines how we know what is knowable and the criteria for truth
from the perspectives of the computer scientist. biologist, chemist. mathematician. and artist. It is
written so as to give experience (yes. really) with concepts discussed. if readers don’t let themselves
be put off by the discussion of well-formed strings or by the length of the book. A third book.
Gregory Bateson’s Mirtd and Aiarltre (1979). addresses preliminary forms of Riedl’s questions, while
Mary Catherine Bateson’s Our Oivn Metaphor (1972) continues that discussion at a sophisticated
process-oriented level. James IMiller’s monumental Living Sytems (1978) is a brilliant tour de force
of functional adaptive analogical processes (one of Riedl’s issues) in biological, psychological. and
social systems. Other general systems theorists provide good background reading. too. With these
as background, Riedl’s ideas became far more interesting.
Some investigators working in cognitive psychology now are addressing mechanisms for humans
knowing what Riedl knows about truth and adaptation. These include students of relativistic post-
formal thought. e.g. Sinnott. in Commons er al. Beyond Formal Oprraths (1984). Add to this
ideas like those in Churchman’s Tlze Design of Inquiring Systems (1971). Bastick’s data-based
synthetic theories [f~zrrrition: How We Think and Act (1982)]. and the newer M.ork of Pribram. and
a reader could have a solid grounding in the scholarship and science that makes Riedl’s ideas come
alive. So, Riedl’s work could be either an interesting beginning or a synthesis endpoint in one’s
exploration of the evolutionary elements of reason.

Ps_u%ology Department JXN D. SINNOTT

To\l.soll State Uni\,ersit?
Baltimore and Gerontology Research Cente,
National Institute on Aging. IVIH
Francis Scott Key Medical Center
Baltimore, MD 21204

Statistics for Experimenters, An Introduction to Design, Data Analysis, and Model Building.
G. E. P. Box, W. G. Hunter and J. S. Hunter, John Wiley and Sons, New York, NY.
1978.

Since it first made its appearance. this book has been popular among statisticians and widely used
as a textbook for courses in analysis of variance. design of experiments. and regression. The topics
covered in it might be described as “central” to the classical field of statistics and the application
of statistics in scientific fields such as chemistry and biology. From the point of view of an es-
1658 Book Reviews

perimenter who needs to apply statistical theory to his own work in regard to data analysis or design
of experiments, the topics covered in this book are essential. It is well written, of high quality, and
covers the topics very fully. However. it does not really introduce the reader to enough probability
distribution theory to be read by someone who has no previous background in probability and
statistics. The book by Box. Hunter and Hunter should be used as a second book, after first reading
a book (or taking a course) on elementary probability and statistics.
In Part I. Chaps. l-5. the ideas of significance tests and confidence intervals are introduced.
Topics relating to the comparison of two experimental groups, e.g. treatment versus control group.
are emphasized. Part II. Chaps. 6-8. deals with comparisons of more than two groups, the full
analysis of variance. Part III, Chaps. 9-13. goes into more detail on the analysis of variance covering
factorial designs of various kinds and their applications. Part IV, Chaps. 14-18, presents the various
topics of regression analysis including a chapter on time series.
This book is well suited to self-learning as well as being a good textbook for a course. The
exposition contains plenty of illustrative examples and there are many exercises and problems. The
reader might be interested in comparing this book with others, also well written, on similar topics,
such as Probability and Statistics for Engineers and Scientists by R. E. Walpole and R. H. Myers,
or Applied Linear Statistical Methods by D. F. Morrison.

Illinois Institute of Technology BARBARA HELLER

Chicago, IL 60616

Statistical Computation. J. H. Maindonald, John Wiley and Sons, New York, NY. 1984.

This provides an excellent introduction to the computational techniques which are currently being
used for some of the most commonly utilized statistical procedures. A professional statistician, or
student of statistics, who is not expert in numerical methods but who uses statistical packages for
computing would find it well worth his while to study. read, or reference this high quality text in
order to gain an intelligent insight into the proper use of packages. For those whose computing
goes beyond packages, this book is invaluable.
Maindonald’s book also appears as if it would be a good textbook for a course in numerical
methods and analysis which is specifically designed for statistics students. The exposition is well
organized, clearly stated, and contains many illustrative examples. There is a good set of references
and each chapter ends with a set of exercises.
The first four chapters are devoted to linear regression and correlation topics along with their
consequent matrix manipulations. Chapter 5 deals with analysis of variance and Chap. 6 with mul-
tivariate techniques. Chapter 7 covers various nonlinear topics such as maximum likelihood equa-
tions, nonlinear least squares and log linear models. Chapter 8 deals with splines. robust least
squares, and certain topics in time series analysis. Chapter 9 give, an introduction to pseadorandom
number generation and the last chapter, Chap. 10. provides information on some commonly used
computer packages.
Most of the chapters are written in two parts. computational and theoretical. The computational
part is explained from the point of view of a reader who is learning the material for the first time,
and is presented in adequate detail with examples. However, the theoretical part. dealing with
statistical theory itself, is sparsely stated, providing a good review for a graduate student in statistics,
but probably unintelligible to one whose background in statistics is not strong. In addition. the
reader must be comfortable with matrix notation and have some background in linear algebra.

Illinois Institute of Technology BARBARAHELLER

Chicago, IL 60616