How useful are approximations to mean and variance of the index of dissimilarity?