# hw07a - p.3.a Obtain residuals and studentized residuals...

This preview shows page 1. Sign up to view the full content.

STA 6207 – Project 7 – Due 11/22/10 Q.1. RPD Problem 10.6 (Hint: n=4, X 1 =0,X 2 =1,X 3 =2,X 4 =9) Q.2. For the simulation examples in class compute the theoretical variance of the following estimators and compare them with the Ordinary Least Squares Estimator (see PPT slides for form of V , and note that for correlated errors, you will want to include the denominator (1- ρ 2 ) in the elements of V ): p.2.a.Heterogeneous Variances: V(Y) = σ 2 (X+0.5) 2 (Note: In simulations, σ = 1) µ ( 29 1 1 1 2 ' ' ( ) WLS X V X X V Y V Y V β σ - - - = = p.2.b. Correlated Errors: Cov(Y t , Y t-|h| ) = σ 2 ρ |h| / (1- ρ 2 ) (Note: In simulations, σ = 3, ρ = 0.5) µ ( 29 1 1 1 2 ' ' ( ) GLS X V X X V Y V Y V - - - = = Q.3. For your original LPGA sample (and best model) in Project 5, conduct a residual analysis as follows:
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: p.3.a. Obtain residuals and studentized residuals p.3.b. Plot residuals versus: fitted, and each predictor variable (Comment on each) p.3.c. Obtain a normal probability plot and test for normality (Use Shapiro-Francia version in text, or use Shapiro-Wilks if using software program with it built in). Are there any extreme outliers? p.3.d. Obtain the following Influence statistics and identify any influential golfers with respect to each: Cook’s D, DFFITS, DFBETAS j , and COVRATIO. p.3.e. Compute VIF for each of the predictors. Is there evidence of multicollinearity among the predictor variables?...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online