Fisher's Exact Test
When you open the file (see hyperlink at the bottom of this page), the following Excel file will emerge.
To run the program, place your data frequencies in the cells. For instance, if you had the following data
, you would be able to insert the frequencies in the cells as follows:
If any cell contains a zero, you are done. If not get a piece of paper and keep track of the little red number at the bottom right corner of the spreadsheet. So far, this number indicates the probability of finding this arrangement of frequencies at random (p = .36, rounded).
To find the probability of such an occurrence of frequencies or ones even more extreme (provided the column and row totals are the same), we make a change in the cell frequencies.
|look at the diagonal frequencies and find the diagonal pair that is smallest. In our example, the smallest diagonal frequencies are indicated by the circled frequencies .|
|Subtract one from each of these smallest diagonal frequencies and add one to the other diagonal frequencies.|
When this step is taken, the spreadsheet becomes:
Add the probability value (p = .13 rounded) to the previous probability score.
Since no cell contains a zero, the process is repeated. Subtract 1 from each of the diagonal frequencies that are smallest. Add 1 to each of the diagonal frequencies that are largest.
Since a cell contains a zero, the process is finished. In this case, the probability of .01 rounded is added to the previous probability values of .36 and .13, which equals .50. Because this value is above a probability of .05 (the typical decision rule set), most researchers would not reject the null hypothesis and, thus no claim of a statistically significant relationship would be made.
NOTE: You must have Excel installed on your computer to use this spreadsheet.
This program was prepared with Microsoft Excel 2003.*
Download Excel file to compute a Fisher's Exact Test for a 2 x 2 table
* Microsoft Excel is a registered trademark of the Microsoft Corporation