CHAPTER 10
CATEG 0 R I CAL DATA: P RO PO RT I0 N S
In previous chapters, the data considered were in the form of continuous (interval
or ratio) measurements; for each individual in a sample, some characteristic was
measured: height, weight, blood cholesterol, gain in weight, temperature, or duration
of illness. This chapter and Chapter 11 are both concerned with a somewhat different
type of data: data consisting merely of counts. For example, the number of survey
respondents who are male or female may be counted. Here the variable is gender
and there are two categories: male and female. The proportion of males would be
the count of the number of males divided by the total number of males and females.
Some variables, such as race or religion, are commonly classified into more than two
categories.
According to Stevens’ system for classifying data, the data would be nominal.
In this chapter we give methods of handling categorical data for populations whose
individuals fall into just two categories. For example, for young patients who un-
derwent an operation for cleft palate, the two outcomes of the operation might be no
complications (success), and one or more complications (failure). This type of data
is essentially “yes” or “no,” success or failure. Categorical data with more than two
categories are discussed in Section 1 1.4.
Basic Sratistics: A Primerfor the Biomedical Sciences, Fourth Edirion.
By Olive Jean Dunn and Virginia A. Clark
Copyright @ 2009 John Wiley & Sons, Inc.125