any of a class of statistical procedures, such as analysis of variance or multiple regression, that rely on the assumption that the calculated statistic—the F ratio—follows the F distribution when the null hypothesis is true. F tests are tests of hypotheses about population variances or of whether regression coefficients are zero.