6. (8 pts) Assume that all six of the Classical Linear Model assumptions hold for our baseline model: 1 . 6. We can really write the true model as a linear function of the parameters and error term . Our sample was selected at random from the population . There's some variation in the X51.- values and no predictor can be perfectly expressed as a linear combination of the others The values of Xi}- and 8f are not correlated for any 1' . The error terms are homoskedastic (formally, VarCa'i IX i1: \"-Xik) = 0'2 for all observations) The error terms are independent and normally distributed (the errors are iid N (O, 02)) Which of these assumptions support our claim that the estimates of the coefcients in the baseline model are unbiased? We have a pretty large sample size do these assumptions still matter for unbiasedness? 7. (8 pts) Which graph in your handout supports our claim that assumption 6 holds? We have a pretty large sample size does the normality assumption still matter when we're doing t tests and F tests? \fGraph 2: Scatter of birthweight and smoking 250 200 O O 150 Birthweight in ounces oo O O 00 0 O O D OOO CO Q0 00 0 O o O OO CO GOOD OID 0 00 0 100 O 8 O 0 0 00 0 10 20 30 40 Average cigarettes per day while pregnant\fBirthweight in ounces 8 100 200 250 O OO O OO OO O O OD O Graph 4: Scatter of birthweight and mother's education 10 Mother's years of education O O O 15 oo 0 00 0 O O Q CD COMEDIDILE OOO OD\fGraph 7: Residuals (e-hat) from Regression 1 against Smoking 150 100 DO OO 8 0 0 Residuals O O O O O O 00 0 O O O 0DO O O 0 00 -100 0 10 20 30 40 Average daily cigarettes during pregnancy