Hi. I'm having trouble with two different questions in linear models. Any help would be greatly appriciated.
1. In a regular linear regression model I have to find the least squares estimate for beta under the null hypotheses of the form H_0 : C*beta = a I know that the way to do this is by Lagrange multipliers, defining a function g(beta, lambda) = ||y - X*beta||^2 + lambda(C*beta - a) and minimizing with respect to beta and lambda. The problem is, after taking the partial derivative with respect to lambda I get the completely miningless result of C*beta - a = 0. I looks like a stupid error which I can not find. 2. In a one-way ANOVA model with random effect y_ij = myu + alpha_i + e_ij I have to find the maximum likelyhood estimators for model parameters myu, alpha_i and sigma2 and also test the hypotheses H_0 : alpha_1 = ... = alpha_m = 0 I know how to do this with fixed effects, but I get confused by the random effects implications. Any help would be greatly appriciated. Thanks. . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
