ABSTRACT
So far we have looked at statistical tests which determine whether the distribution of two or more variables differs significantly. However, we are often interested in the extent to which two or more variables are associated, the direction of that association and the likelihood of that association occurring by chance. For example, we may wish to know whether management students are more likely to have tutorials than other social science students and what the strength and statistical probability of any such association is. Or, we may want to find out whether higher pre-test marks are related to higher post-test marks and how strong and statistically significant such a relationship is. In addition, we may wish to discover whether an association between two variables, or the lack of one, is the result of their relationship with one or more other variables. For instance, if it was found that students with higher marks had tutorials, then this association may result from the fact that students who were more interested in their course were more likely both to obtain higher marks and to attend tutorials. To discover the answers to such questions, we would need to carry out the appropriate test of association, the selection of which partly depends on whether the data are categorical, ordinal or interval/ratio. In this chapter we will deal with tests which are suitable for categorical and ordinal data.
