Book Review
The Cartoon Guide to Statistics
by Larry Gonick & Woollcott Smith (2000)
ISBN: 0-06-273102-5
New York: Harper Collins
This book follows the literary traditions of France and Japan in which some comic books cover serious topics and are considered
intellectual stimulants. Gonick is a well-respected illustrator of non-fiction cartoon books and has a cartoon column in Discover magazine. Smith is a research and consulting statistician at Temple University. Together they have compiled a short discussion on statistical analysis and its implications in modern academic and commercial society.
The book's 231 pages are twelve chapters followed by a bibliography and index. The first chapter offers a laconic defence of the
importance of statistics in social and academic circumstances. The second illustrates data analysis: the collecting, display, and
summary of data. The third chapter demonstrates probability and its applied application in statistical analysis. The next two chapters
explain statistical interference, probability models and the random variables. Chapter six introduces sampling design: stratification
and data clustering. Concepts such as standard error, mean, and t-distribution are clarified. Chapter seven is concerned with data
interpretation and making statistical inferences. Chapter eight discusses hypothesis testing and the value of significance testing.
Chapter nine covers population testing and statistical pitfalls to avoid. Chapter ten discusses experimental design, summarising
replication and randomisation. The next chapter introduces regression analysis: ANOVA analysis, correlation coefficient, linear
and non-linear regressions, and the interpretation of regression results. The final chapter mentions characteristics of statistical
analysis and its consequent influence on contemporary society.
This book's humour exaggerates particular perspectives at the expense of other viewpoints. The positive aspects of this book are
its historical analysis of different quantitative theorems and analyses. Readers are briefly introduced to memorable personages in
statistics. It is tempting to know more about such men as James Bernoulli
and Abraham de Moivre who explained standard normal
distribution, Chevalier de Mere (was he really a rake?),
Pierre de Fermat and Blaise Pascal who defined the laws of probability,
Rev. Thomas Bayes who developed a theorem to account for false positives,
John Tukey who invented the stem-leaf diagram,
William Gosset who developed the "students t" theorem which clarified differences between a sample and normal distribution,
Francis Galton who explained the regression to the mean,
R. A. Fisher (the father of modern statistics), and
Herman Chernoff
who devised a method of plotting more than two variables on a flat page. One cannot help wondering if there were statisticians
other than European males, but perhaps the design of the book does not allow such intellectual meandering. The brief bibliography
is also an interesting reference guide to other statistical books and software. Only one woman's material is mentioned, however,
that of Barbara Ryan, co-author of Minitab Handbook (1985).
A few irritants are the continual remarks that statistical analysis is difficult, compounded by a preadolescent sense of humour.
The latent psychological effect of the frequent references to the difficulties of understanding statistical analysis only confirms
to the novice that statistics is an obscure and tedious subject to undertake. Statistics is difficult and hard slog is needed to
comprehend and master it, but novice fears do not need to be repeatedly affirmed about a discipline that can be understood.
All the major characters, narrators, the pollsters and instructors in the book are men. The women merely observe the activities
from frame to frame. The humour is at times phallic and negates a more egalitarian purpose; such as the discussion of confidence
intervals using gender differences in salaries as an example (i.e., p. 169).
[
p. 9
]
Unfortunately, this book does not discuss contemporary issues such as Structural Equation modeling, but this could be because
it was originally published in the early 1990's. It should also be noted that a list correcting the book's statistical typos are
available on Smith's web page at http://www.sbm.temple.edu/~wksmith/corrections.pdf
[expired link].
It would be difficult to recommend this book to a beginning statistical student and some more serious statisticians could
regard the book as overly facile. Most novices do not have enough time or motivation to browse the academic development of
a particular analysis with deadlines hanging over them. It will not help the novice interpreting results or choosing which
type of analysis to employ. However, students who are interested in more surface explanations might enjoy this book.
This is not a hands-on how to do an analysis book, but a humorous digest of why such analysis is administered and what
the given results indicate. This book needs to be appreciated for exactly what is was designed for. The Cartoon Guide
to Statistics is a coffee table text designed for nerds who collect old slide rules and discarded abacuses.
Reviewed by P. L. Stribling
