ECON 104 Lecture 18: 18
Document Summary
Intuition = when variance of x and y is large, likely to have outliers -> could affect results negatively. When you work data, it"s always best to look at dat and look at outliers. Sometimes outliers make sense and they are supposed to be there, and there are errors in data. Without this outlier, line is almost flat, x has no effect on y. Including that one data point -> tilt completely positive. Distribution is useful for constructing confidence intervals and evaluate how good/ bad it is. In our book"s appendix there are calculations - won"t be asked. If assumptions are correct, ols estimator -> unbiased estimate of slope. Variance = simple formula that tells us about what variance depends on. Distribution of beta in small samples is very complicated. Small sample can"t tell us what distirbution may be but beta 1 hat is normally distributed, this is like sample average.
