Contingency Table

An application used to test the relationship between two variables. For example, a financial research firm decides to investigate whether there is a relationship between reports that are completed in less than or equal to 3 days and those that make accurate predictions. The hypothesis is:

• H0 there is no relationship between less than or equal to 3-day completion and accurate predictions.
• H1 there is a relationship between less than or equal to 3-day completion and accurate predictions.

	Accurate	Not Accurate	Total
Equal or Less Than 3	53	23	76
Greater Than 3	78	42	120
Total	131	65	196

The table allows us to glance at the proportion of reports that are less than or equal to 3 days and greather than 3 days versus accuracy and non-accuracy; however, the proportions are not equal so the statistical significance of the difference between the variables can be tested with formulas such as the Chi Square (or ‘Goodness of Fit’) test, G-test or Fisher’s exact test.