3. One advantage of using a minimal sufficient statistic is that unbiased estimator will have smaller variance, as this problem will show. Suppose that T is sufficient and T is minimal sufficient, U is an unbiased estimator of 0, and define U = E(U|T) and U2 = E(U|T2). (a) Show that U = E(U1|T2). (b) Now use the conditional variance formula to show that Var(U2) Var(U1).