pdf file of assignment
[updated question 3 & 4 as of 2007-11-16 10:52
| solution R code
- You have two ways to compare models:
- Log likelihood ratio test statistic for comparing a restricted model ($H_0$) to a unrestricted model ($H_A$). Using R, you can extract the maximum log likelihood using the logLik function on an lm or glm object.
- F test by partitioning sums of squared residuals. Using R, you can test two (or more) models, say m.1 and m.2, using anova(m.1,m.2).