| Abstract: | The exact multinomial test described in this article can be used to test the significance of variations in the numbers of observations distributed into 2 or more mutually-exclusive categories. When there are only 2 categories the test reduces to the binomial test. The test is valid for samples of any size but it quickly becomes prohibitively difficult to apply as the total number of categories increases. A comparison with the chi-square test shows how seriously the latter may be in error when the number of observations is small. (PsycINFO Database Record (c) 2010 APA, all rights reserved) |
