A contingency table shows the frequency of various combinations of values of two or more attributes (the cross-attribute frequencies). For example, suppose we are interested in how skill level (coded as novice/expert) affects error rate (low/medium/high) for an application. We gather some novice and expert participants and measuere theor eror rate. We then count how many times we see people in each skill level an count how many perform at each error level, and lay these out in a table.
| error rate | ||||
|---|---|---|---|---|
| low | medium | high | TOT | |
| novice | 2 | 17 | 21 | 20 |
| expert | 14 | 6 | 0 | 40 |
| TOT | 16 | 23 | 21 | 60 |
Note: (1) row and column totals are very commonly added as they are here; and (2) you can start to see a pattern, where experts are performing better – this is often confirmed using a chi-squared test.
Used in Chap. 13: page 150
Also used in hcistats2e: Chap. 4: page 58
Used in glossary entries: chi-squared test, cross-attribute frequencies
