Used to determine whether previously discovered associations are random or caused by real differences.