Question

There are important applications in which due to known scientific constraints, the Problem 5 of 5 regression line must also go through origin (i.e. the intercept must be zero). In other words, the model should read Y Bai,i 1,2,.,n This model is often called the regression through the origin model. Assuming that e's are independent with distribution N(0, o2) (a) Show that the least squares estimator of the slope is ΣL Υ B = Σ (b) Show that B in part (a) is also an unbiased estimator for B (a) (c) Obtain Var Is this higher than the variance of the original linear regression with non-zero inter- cept?

Answer 1

madel is Aai he ad- p + E n Jesnt a) he eghare entirmator miimimizing fle) - 1 26 ' .(a- riny Juge bath e'des hove we l1,.-pX.Jo Dyperenhahng wt p heve Ir-AX) E-x.) - 2 Eypating to X(T,-AX) have > 2 E A) Spx+E For urbiaredoens E En: E1EX) 2- + ElA) 2. ElR) entimator oon unbianed

c)Vorl)- Very Ver ly)+ avyy 11 fadeperdert 62 have with iatercept madel We Vor lAy 2 have lomparing and 2 we

Vor lA) -Vor lA) - 2 Xi o 70 Var lp) Vor lA)<Var ()