Multiple overlap integrals are necessary in order to compute the second moment estimators of chi-squared and covariance (strictly, the second moment estimator of the covariance may be computed by expanding the covariance as the sum of powers of the 
's, but this ignores much of the symmetry that leads to simplifications.) Multiple overlap convolution integrations are rather tricky, and much development needs to be done in order to present the results, amounting to more than three pages just to describe the new notation needed, and another four to present the integration theorems. The interested reader is encouraged to look in other work by this author where the necessary multiple overlap convolution integrals are computed, see reference [95], theorems 22 through 27.