What is the normality assumption? What is this model called?
It assumes the error terms are normally distributed. The model is knwon as the classical normal linear regression model.
How justified is the normality assumption.
Not so bad when applied judicously. See central limit theorum
What are the applications of the normality assumption?
Gives us the distributions of parameters, and some sweet properties
What is the distribution of beta1 hat in terms of the variance?
What is the distribution of beta2 hat in terms of the variance?
What are the properties of the CNLRM?
- Minimum variance
- Consistancy (estmates converge to parameters with increasing n) - beta1 hat normally distributed
- beta2 hat normally distributed
- (n-2) (sample-variance)/(population-variance) has a chi-square distribution
- beta1 & beta2 hat both distributed independantly of sample variance
-beta1 & beta2 hat have minimum variance in their the class of estimators, linear or not. Is BUE
What is BUE?
Best unbiased estimator