The amount of between-group variability must be significantly greater than the amount of within-group variability to reject the null hypothesis.
