The address of my own Website, mentioned above, is http://www.uvm.edu/~dhowell/
StatPages/StatHomePage.html (capitalization in this address is critical) and I encourage
users to explore what is there.
This edition shares with its predecessors two underlying themes that are more or less inde-
pendent of the statistical hypothesis tests that make up the main content of the book.- The first theme is the importance of looking at the data before jumping in with a hypothe-
sis test. With this in mind, I discuss, in detail, plotting data, looking for outliers, and
checking assumptions. (Graphical displays are used extensively.) I try to do this with each
data set as soon as I present it, even though the data set may be intended as an example of
a sophisticated statistical technique. As examples, see pages 330–332 and 517–519. - The second theme is the importance of the relationship between the statistical test to be
employed and the theoretical questions being posed by the experiment. To emphasize
this relationship, I use real examples in an attempt to make the student understand the
purpose behind the experiment and the predictions made by the theory. For this reason I
sometimes use one major example as the focus for an entire section, or even a whole
chapter. For example, interesting data on the moon illusion from a well-known study by
Kaufman and Rock (1962) are used in several forms of the t test (pages 190), and most
of Chapter 12 is organized around an important study of morphine addiction by Siegel
(1975). Chapter 17 on log-linear models, which has been extensively revised in the
edition, is built around Pugh‘s study of the “blame-the-victim” strategy in prosecutions
for rape. Each of these examples should have direct relevance for students. The increased
emphasis on effect sizes in this edition helps to drive home that point that one must think
carefully about one’s data and research questions.
Although no one would be likely to call this book controversial, I have felt it important
to express opinions on a number of controversial issues. After all, the controversies within
statistics are part of what makes it an interesting discipline. For example, I have argued that
the underlying measurement scale is not as important as some have suggested, and I have
argued for a particular way of treating analyses of variance with unequal group sizes
(unless there is a compelling reason to do otherwise). I do not expect every instructor to
agree with me, and in fact I hope that some will not. This offers the opportunity to give
students opposing views and help them to understand the issues. It seems to me that it is
unfair and frustrating to the student to present several different multiple comparison proce-
dures (which I do), and then to walk away and leave that student with no recommendation
about which procedure is best for his or her problem.
There is a Solutions Manual for the students, with extensive worked solutions to odd-
numbered exercises that can be downloaded from the Web at the book’s Web site—
http://www.uvm.edu/~dhowell/methods/. In addition, a separate Instructor’s Manual with
worked out solutions to all problems is available from the publisher.
Acknowledgments
I would like to thank the following reviewers who read the manuscript and provided
valuable feedback: Angus MacDonald, University of Minnesota; William Smith, California
State University – Fullerton; Carl Scott, University of St. Thomas – Houston; Jamison
Fargo, Utah State University; Susan Cashin, University of Wisconsin-Milwaukee; and Karl
Wuensch, East Carolina University, who has provided valuable guidance over many
editions. In previous editions, I received helpful comments and suggestions from Kenneth
J. Berry, Colorado State University; Tim Bockes, Nazareth College; Richard Lehman,
Franklin and Marshall College; Tim Robinson, Virginia Tech; Paul R. Shirley, UniversityPreface xix