[This question paper contains 4 printed pages.] Your Roll No. Sr. No. of Question Paper: 1381 A Unique Paper Code : 32371402 Name of the Paper : Linear Models Name of the Course : B.Sc. (H) Statistics under CBCS Semester : IV Duration : 3.5 Hours Maximum Marks : 75 Instructions for Candidates Instructions for Candidates 1. Write your Roll No. on the top immediately on receipt of this question paper. 2. Attempt SIX questions in all. 1. Derive the analysis of variance of two-way classified data with one observation per cell under fixed effects model, Also obtain the expectation due to all different sources of variation. : (12%) 2. (a) Develop a prediction interval for the future observation yg corresponding to a specified level xp of the regressor variable x in the simple linear regression model. P.T.O.1381 2 (b) Suppose X;,¥;,Z;,i = 1,2,....n are 3n independent observations with common variance o” and expectations E(X;) = O,E(Y) = 0,E(Z) = i = 1,2,..,n. Find the BLUEs of 6;,. Compute cov(@;,0) and the residual sum of squares. Also find BLUE of 6, + 62. 7%) ._ (a) Write a short note on bias in regression estimates? Suppose the postulated model is E(Y) = By + A, x but the true model is (¥) = Bo + Bix; + Box2 . Show that both Bo and B are biased by an amount that depends the values of x’s. (b) We fit a straight line model to a set of data using the formulas b = (X'X)2X'Y and compute “Y = Xb with the usual definitions. We define H = X(X'X)~1X'. Show that SS(due to regression) = Y’HY = 7’? = PH3P. (6.6%) 4, State and prove Cochran's theorem. (12%) 5. (@) Suppose ¥ = (¥;,¥,.,Ya)'to be @ vector of n independent standard normal variates then a necessary and sufficient condition for Y’AY to be distributed as chi- square variate with k d.f.is that A is an idempotent matrix of rank k, (&) Suppose y;,(i = 1,2,...,n) is a random sample from a standard normal distribution. Show that D7, y; and D7L4 (9; — 7)? are independently distributed. (8,4) 6. Suppose we estimate the following relationship between wages and education controlling for years of experience (a quadratic form) in the workforce: Wage = B, + B,Educ + B3Exper + B,Exper? +e, Using a random sample of 54 observations. The least square estimates and the corresponding estimated covariance matrix are given by iH 4 a -0.125 0.01 2 -0.2 . 0.01 0 b,| = -[# War(b) = [. 0.125 0.01 0.04 0.12 Loon 001 0 012 0091381 3 (a) What is the covariance between B; and B, ? (b) What is a 95% confidence interval for B,? (Given t(o.05,50) = 2.01) (c) Is there a statistically significant effect of education on wages in this regression at the 5% level? Give interpretation also. (d) What is the estimated standard error for (Bz + 3 23)? (©) What is the ¢-statistic from a test that the marginal impact of years of experience on wages equals zero for someone with two years of experience? (1, 3%, 3, 3%, 1%) 2-1-1 7. (@) Suppose ¥~N,(Q,1) andlet A=4]—1 2° -1], -1 -1 2 (@ What is the distribution of YYAY and Y-A)Y 2 (ii) Are the two quadratic forms defined in part (a) independent? (iii) Are Y'AY and yrtyztys independent? (b) Write the simple linear regression model in matrix notation. Hence obtain the least squares estimators of the unknown parameters and their variances. 6,64) . (a) Ifyou fit the models (i) ¥ = Bo + Bix +e and (ii) ¥ = Bx +e fora given data, with usual assumptions, which of the above models is appropriate, justify your choice? Give three reasons for your justification. () What do you understand by analysis of covariance? State the mathematical model used in analysis of covariance for one way classification with one covariate. Explain the hypothesis to be used. (©) Explain the significance of partial F-test and sequential F-test. (5, 4,34) P.T.O.1381 4 9. Write short notes on any three of the following: (a) Orthogonal polynomials (b) Estimable functions (©) Extra sum of square principle (@) Linear hypothesis (4,4, 44) (600)