![Home](http://my.ilstu.edu/~oakman/dataanalysis/_derived/home_cmp_copy-of-blank010_hbtn.gif)
![Up](http://my.ilstu.edu/~oakman/dataanalysis/_derived/up_cmp_copy-of-blank010_hbtn.gif)
![Small](http://my.ilstu.edu/~oakman/dataanalysis/_derived/small_expected_frequencies.htm_cmp_copy-of-blank010_hbtn_p.gif)
![Large](http://my.ilstu.edu/~oakman/dataanalysis/_derived/large_expected_frequencies1.htm_cmp_copy-of-blank010_hbtn.gif)
Collapse Categories for x2
One solution for low expected frequencies is to collapse some of the categories. For example, if there are 4 categories of one variable, combining some of them to form only two categories may increase the expected value enough to perform a x2 test.