Please answer this question (1) and (2). Thanks.

We consider the following regression: YKX = (X1, . .., X10)} = 1 + X1+, where & ~ N(0, 1). All means of X1, ..., X10 are zero. We fit the linear regression of Y X, and call this "ml" through the least squares. Let _1 represent 3 after dropping the estimate of the intercept, and compute xb = XTB_1. Next fit a regression of Y|X' and call it "m2". Suppose we have the following new observations of Xnewl = (0, 1, . .., 1), Xnew2 = (0, 2, ..., 2), and Xnew3 = (0, 3, ..., 3). Using these new observations, construct X = XT new1B-1, Xb new2 = XTew28_1, and Xb new3 = XT new38-1. It is supposed that one constructs 95% prediction confidence intervals (PCI) for ml and m2 with {Xnew1; Xnew2; Xnew3} and {Xnewli Xnew2; Xnew3}, respectively. (1) What do you expect, if you compare 95%PCIs from ml for { Xnew1; Xnew2; Xnew3}. (2) What do you expect, if you compare 95%PCIs between those from ml and those from m2