Recall from Chapter 3 that there are three broad categories of variables, namely:
-
Categorical variables: Data such as the industry of a firm or religion of a person, that indicates only
membership of an observation in a certain category, with no other mathematical meaning
to the data (no preference or order of the categories, no possibility of mathematical
operations on the data).
-
Ordinal data: Data such as rankings of consumer car preferences. The data has only an order, but
no sense of the distances between the data points.
-
Continuous data: Data on a scale running from low to high on a relatively mathematically fine scale,
such as prices or perhaps answers to a survey question where there are enough scale
answers. Recall that single questions on Likert or semantic differential-type scales
are considered debatable especially when they have five answer points (some believe
this to be ordinal data). Some researchers believe that 1 to 7 answer scale data has
more chance of beginning to be continuous.
The point of this section is that all types of data can be associated, but the statistical methodology involved depends
on the types of data in the association.
In this chapter, we start by looking at the more simple tests of statistical association.
We will focus largely on bivariate associations of data here, where associations between only two variables are analyzed.
The following sections discuss each of these techniques. Note, also, that if you intend
to test whether one of the variables may cause the other, your options expand. Such
techniques are introduced later in the book.