Omnibus test

From Wikipedia, the free encyclopedia

Omnibus tests are a kind of statistical tests. They test the null hypothesis that in a set of data there is no significant difference at all. One example is the F-test in the analysis of variance. A significant omnibus test is as a prerequisite for multiple testing. If an omnibus tests maintains the null hypothesis that in a set of means there is no significant difference one should not conduct pairwise tests of significance. By applying this rule one can reduce false positive errors due to multiple comparisons.