From the preceding discussion, the apparent conclusion—and the one accepted in this
book—is that the underlying measurement scale is not crucial in our choice of statistical
techniques. Obviously, a certain amount of common sense is required in interpreting the
results of these statistical manipulations. Only a fool would conclude that a painting that
was judged as excellent by one person and contemptible by another ought therefore to be
classified as mediocre.
1.4 Using Computers
When I wrote the first edition of this book twenty-five years ago, most statistical analyses
were done on desktop or hand calculators, and textbooks were written accordingly. Meth-
ods have changed, however, and most calculations are now done by computers.
This book attempts to deal with the increased availability of computers by incorporat-
ing them into the discussion. The level of computer involvement increases substantially as
the book proceeds and as computations become more laborious. For the simpler proce-
dures, the calculational formulae are important in defining the concept. For example, the
formula for a standard deviation or a t test defines and makes meaningful what a standard
deviation or a t test actually is. In those cases, hand calculation is emphasized even though
examples of computer solutions are also given. Later in the book, when we discuss multi-
ple regression or log-linear models, for example, the formulae become less informative.
The formula for deriving regression coefficients with five predictors, or the formula for es-
timating expected frequencies in a complex log-linear model, would not reasonably be ex-
pected to add to your understanding of such statistics. In those situations, we will rely
almost exclusively on computer solutions.
At present, many statistical software packages are available to the typical researcher or
student conducting statistical analyses. The most important large statistical packages,
which will carry out nearly every analysis that you will need in conjunction with this book,
are Minitab®, SAS®, and SPSS™, and S-Plus. These are highly reliable and relatively easy-
to-use packages, and one or more of them is generally available in any college or univer-
sity computer center. Many examples of their use are scattered throughout this book. Each
has its own set of supporters (my preference may become obvious as we go along), but they
are all excellent. Choosing among them hinges on subtle differences.
In speaking about statistical packages, we should mention the widely available spread-
sheets such as Excel. These programs are capable of performing a number of statistical cal-
culations, and they produce reasonably good graphics as well as being an excellent way of
carrying out hand calculations. They force you to go about your calculations logically,
while retaining all intermediate steps for later examination. Statisticians often rightly criti-
cize such programs for the accuracy of their results with very large samples or with sam-
ples of unusual data, but they are extremely useful for small to medium-sized problems.
Recent extensions that have been written for them have greatly increased the accuracy of
results. Programs like Excel also have the advantage that most people have one or more of
them installed on their personal computers.
1.5 The Plan of the Book
Our original example, the examination of the effects of a program of stress management on
self-esteem, offers an opportunity to illustrate the book’s organization. In the process of
running the study, we will be collecting data on many variables. One of the first things we
will do with these data is to plot them, to look at the distribution for each variable, to
Section 1.5 The Plan of the Book 9
