An F-test is a statistical technique designed to measure the likelihood of two samples having come from two normal distributions with different variances. With an F test we can contrast a null hypothesis (with equal variances) with an alternative hypothesis (where variances are believed to be different).