This preview shows page 1. Sign up to view the full content.
Unformatted text preview: University of Waterloo A few sample questions from the final exam, STAT331
1. True or False? (a) Correlation is always positive. True ; False (b) Akaike Information Criteria (AIC) tends to suggest a model with more parameters, compared with a model proposed by Bayesian Information Criteria (BIC). True ; False (c) The Durbin-Watson test is always conclusive. True ; False (d) If you have too many runs in the runs test, it indicates positive serial correlation. True ; False (e) Can the adjusted R2 be negative. True ; False 2. (a) Which of assumptions we do not need for the GaussMarkov theorem i. ii. iii. iv. homoscedasticity uncorrelatedness normality (T) heavy tailness (T) (b) You have a multiple linear regression model with p predictors and intercept, estimated using n observations. How many degrees of freedom are needed to estimate a variance of residuals? i. n - p ii. n - p - 1 (T) iii. n - 1 (c) What is the distribution of sample variance of residuals? i. 2 (T) ii. F iii. t (d) Can SSE be larger than SST? i. always ii. never (T) iii. sometimes 1 ...
View Full Document
This note was uploaded on 06/21/2009 for the course STAT 331 taught by Professor Yuliagel during the Fall '08 term at Waterloo.
- Fall '08