Chebyshev's inequality can be used to bound the tails of random variables that are sums of random variables that are not independent, provided that pairs of random variables are not too positively correlated with each other. Let random variable X = _" Xi be the sum of indicator variables X1, ..., Xn for events Al, ..., An. Suppose that E[Xi] = P(A;) = p for each i and that for every i # j, we have E[X;X;] = P[Ain A; ]