problematic observations

Please attach all the code from R and the output you get. Justify you findings at every step.

Run an OLS regression with price as your dependent variable (using the first 16

regressors).

Include your summary output and the standardized residual-leverage plot. Do you see

any

“problematic” observations? What would that mean? Manually standardize the residuals

and calculate how many standard deviations each is from the residual mean. Which

observations have residuals that are more than 4 standard deviations from the mean?

Describe those observations and whether or not you see implausible values.

After removing any observations for which you DID find implausible values in (1) run the

regression again. Use a Q-Q plot and a Jarque-Bera test to determine whether

residuals appear to be normally distributed. Use a RESET test to determine whether a

linear functional form appears to be appropriate.

Which variables do you think might have nonlinear relationships with price? Why?

Which

variables might require interaction terms? Why?

Use a series of F tests to determine the best polynomial structure to use and whether or

not

to include the interaction terms that you suggest in (3). Run an OLS regression with

your new, nonlinear model and include your summary output. Run the Jarque-Bera and

RESET tests again, does the result change? Do you see evidence of misspecification?