Part II: Regression Analysis #1 (14 points) This part is based upon a dataset used by Hamermesh and Biddle (1994, American Economic Review, "Beauty and the labor market"). There are 1260 workers in the data. Each person was ranked by an interviewer for physical attractiveness, with the following categories used in the analysis: belavg 1 if rated as having below average looks ("homely" or "quite plain") avg I if rated as having "average" looks abyavg 1 if rated as having above average looks ("good looking" or "strikingly beautiful/handsome") The dependent variable is log wage (denoted Iwage). Below is regression output for a log-wage equation, where avg is the omitted category and several other explanatory variables are included (educ = years of education, exper = years of experience, expersq = exper squared, female = 1 if female). For this part, assume all of the finite-sample assumptions (MLR. 1 - MLR.6) hold. regr Iwage belavg abvavg educ exper expersq female Source SS df MS Number of obs = 1260 F( 6, 1253) = 117.36 Model | 160 . 094314 6 26. 6823857 Prob > F 0 . 0000 Residual 284 . 885658 1253 . 227362856 R-squared = 0 . 3598 -+- Adj R-squared = 0. 3567 Total | 444. 979972 1259 . 353439215 Root MSE = . 47683 1wage | Coef Std. Err. t P>It| [95% Conf. Interval] belavg -. 1542032 0423296 -3. 64 0 . 000 - . 2372479 - . 0711585 abvavg - . 0066465 0306562 -0. 22 0 . 828 - . 0667896 . 0534966 educ 0663221 0053094 12 . 49 0 . 000 0559058 0767384 exper 0408305 . 0044034 9.27 0. 000 0321916 0494694 expersq - . 0006301 . 0000985 -6. 40 0 . 000 - . 0008233 - . 0004368 female - . 4532832 029217 -15 . 51 0 . 000 -. 5106029 -. 3959636 cons 558981 . 0795603 7. 03 0 . 000 . 4028949 . 7150671(1 points each; 2 points total) Suppose that you use mg and abxaxg in the regression rather than helaxg and abxajgg. Circle one answer for each of these two questions: a. Which of the following quantities would change? (a) Rsquared (b) Root MSE (this is sigma-hat) (c) the intercept estimate ((1) the slope estimate on edge b. Which of the following quantities can n_ot be determined from the original output? (a) the new t-stat on bxaxg (b) the new slope estimate on bxaxg (c) the new slope estimate on mg ((1) the new t-stat on avg