ABSTRACT
Theorem A. 1.1 (Chebyshev's other inequality) Let T be a p-dimensional random variable. Suppose the functions #i,#2 : W — >• 5R1 satisfy
for all ti,*2 *ft the support of the distribution ofT, then E(9l(T)92(T)} > E[9l(T)]E[gz(T)}
provided the expectations exist, i.e., the covariance between Q\ (T) an is non-negative.
