Consider a normal random sample from population N (u, o2), where 2 is assumed to be known. Consider the following hypotheses: Ort > Irt = 1: 1Hsa a or = 1: 0H Let the rejection region be { : < po zao/n}, then the type-Il error is B = P(X > Ho zao//n| = 41). 1. For a fixed sample size n, when a decreases, B will .(Type in "O" if you think it will decrease, and "1" if you think it will increase.) 2. For a fixed sample size n, when B increases, a will . (Type in "O" if you think it will decrease, and "1" if you think it will increase.) 3. Let o = 0.12, i = Ho 0.03, a minimum number of samples are needed if we want to ensure that a < 0.025, 8 < 0.025. (Hint: round UP to the nearest integer.)