If we have two group means to compare, we use a z or t test as we have seen, assuming we can meet the assumptions of a parametric test. If we have three or more group means to compare, we would have to do multiple z or t tests, but this is just not feasible, and multiple tests would increase our probability of type 1 error. So we use ANOVA. ANOVA uses the F ratio (a ratio of Variances) to inform us on any differences among group means.