Reading and Comprehension Questions for Chapter 12 1. Any regression model that is linear in the unknown parameters is a linear regression model. True False True 2. The least squares estimator of the model parameters in multiple linear regression is βˆ = (XX)-1 Xy . True False True 3. If a multiple linear regression model with three regressors is fit to a sample of 20 observations and the residual or error sum of squares is 32, the estimate of the variance of the model errors is a. 16.0 b. 2.0 c. 4.0 d. None of the above. Answer – b. The estimate of the error variance is ˆ 2 SS E /(n p) 32 /(20 4) 2.0 4. When using the method of least squares to estimate the parameters in multiple linear regression, we assume that the model errors are normally and independently distributed with mean zero and constant variance. True False False – the normality assumption is not required for parameter estimation, but it is required for hypothesis tests and confidence intervals. 5. The test for significance of regression in multiple regression involves testing the hypotheses H 0 : 1 2 ... k 0 versus H1 : at least one j 0 . True False True 6. The ANOVA is used to test for significance of regression in multiple regression. True False True 7. The R2 statistic can decrease when a new regressor variable is added to a multiple linear regression model. True False False - The R2 statistic can never decrease with the addition of a new regressor variable to the model. 8. If SST 100, SSE 15, n 20, and p 2 the adjusted R2 statistic is a. 0.8875 b. 0.8324 c. 0.8525 d. None of the above. 2 1 Answer – b. RAdj SS E /(n p) 15 /(20 3) 1 0.8324 . SST /(n 1) 100 /(20 1) 9. The adjusted R2 statistic can decrease when a new regressor variable is added to a multiple linear regression model. True False True 10. The test statistic for testing the contribution of an individual regressor variable to the multiple linear regression model is ˆ j T0 se( ˆ ) j True False True 11. When testing the contribution of an individual regressor variable to the model if we find that the null hypothesis H 0 : j 0 cannot be rejected, we usually should a. Remove the variable from the model b. Do nothing. c. Add a quadratic term in xj to the model. Answer – a. Removing a nonsignificant regressor may improve the fit of the model to the data. 12. The extra sum of squares method is used to test hypotheses about a subset of parameters in the multiple regression model. True False True 13. A 95% confidence interval on the mean response at a specified point in the regressor variable space is 34 Y |x 36 . The length of this interval is constant for all points in the regressor variable space so long as the confidence level doesn’t change. True False False – The standard deviation of the predicted response depends on the point and that this standard deviation determines the length of the CI. 14. Standardized residuals have been scaled so that they have unit standard deviation. True False False – Studentized residuals have been scaled so that they have unit standard deviation. 15. Cook’s distance is a measure of influence on the regression model for the individual observation in a sample. True False True 16. A polynomial regression model is a nonlinear regression model. True False False – any polynomial regression model is a linear regression model because it is a linear function of the unknown parameters. 17. Regression models can only be used with continuous regressor variables. True False False – Indicator variables can be used to represent categorical regressors. 18. All possible regressions can be use to find the subset regression model that minimizes the error or residual mean square. True False True 19. The C p statistic is a measure of the bias remaining in a subset regression model that has p parameters because the correct regressors are not in the current model. True False True 20. If there are eight candidate regressors then the number of possible regression models that need to be considered if we are using all possible regressions is: a. 128 b. 64 c. 96 d. 16. Answer – b. The number of equations is 28 = 64. 21. The PRESS statistic is a measure of how well a regression model will predict new observations. True False True 22. Large values of the PRESS statistic indicate that the regression model will be a good predictor of new observations. True False False – small values of the PRESS statistic indicate that the regression model will be a good predictor of new observations. 23. Stepwise regression is a procedure that will find the subset regression model that minimizes the residual mean square. True False False – while stepwise regression usually finds a good model, there is not guarantee that it finds a model that satisfies any specific optimality criterion. 24. Forward selection is a variation of stepwise regression that enters variables into the model one-at-a-time until no further variables can be added that produce a significant increase in the regression sum of squares. True False True 25. Multicollinearity is a condition where there are strong near-linear dependencies among the regressor variables. True False True