In this lesson, we extended the idea of association in two-way contingency tables to accommodate ordinal data and linear trends. The key concepts are that ordinal data has a particular nature that we can summarize and potentially take advantage of when carrying out a significance test. We also introduced an alternative "exact" test for \(2\times2\) tables that can be used when the observed counts are too small for the usual chi-square approximation to apply.
Coming up next, we consider new ways to measure associations if more than two variables are in the picture. Specifically, we'll see how one relationship can be confounded with another so that interpretations will change, depending on whether we condition on certain variables in advance.