Confidence intervals
Population mean Assumption: sample from normal distribution
1: Population variance 2 is known
Given X 1, X 2, …, X N – independent, normally distributed, what is the distribution of ?
Confidence interval 95% of data
With probability 0.95
Equivalently
2: Population variance 2 is not known Use estimate:
Not true. Intuition tells us that the interval should be wider
E.G., N=7 95% CI looks like this
Student t distribution With N degrees of freedom (d.f.)
Gamma function L. Euler, XXVIII century
Normal samples and t distribution X 1, X 2, …X N – independent normal Has t distribution with N-1 d.f.
Linear regression and t - distribution U – normally distributed disturbance x i - controlled variable, fixed values
Assumption
Leads to
Residual mean square – estimates variance of normal disturbance U
Follow t distribution with n-2 degrees of freedom
Non normal samples For large n, estimates can still be used due to Central Limit Thorem