CHAPTER 12
REGRESSION AND CORRELATION
In previous chapters, one variable was usually studied at a time. An exception to
this was Chapter 11, in which the odds ratio and the chi-square test were introduced
for analysis of two categorical variables. In Chapter 11 we discussed analysis of
categorical data from a single sample, from two or more samples, and from matched
samples. In this chapter a similar format is followed, but here the data are assumed
to be continuous (either ratio or interval, as described in Section 5.4.3).
First, we discuss the relationship between two variables when both variables are
measured from a single sample. This is by far the most common use of the techniques
described in this chapter. For example, we might take a sample of fifth graders and
measure their height and weight. Or with a sample of adult males we might measure
their age, height, weight, systolic blood pressure, and diastolic blood pressure. From
this set of five variables, comparisons could be made two variables at a time. We also
briefly mention the fixed-X case, where the values of X are fixed in advance.
Usually, two variables are studied together in the general hope of determining
whether there is some underlying relation between them, and if so, what kind of
relationship it is. Sometimes, on the other hand, two variables are studied in the hope
of being able to use one of them to predict the other. For example, we might have
two methods of measuring a constituent of blood, one inexpensive to perform and the
Basic Statistics: A Primer for the Biomedical Sciences, Fourth Edition.
By Olive Jean Dunn and Virginia A. Clark
Copyright @ 2009 John Wiley & Sons, Inc.165