You are examining the results from an experiment intending to get people with dr

Question

You are examining the results from an experiment intending to get people with drug habits back into the workforce. The subjects get randomly assigned to different number of daily hours of help for a 10 week period. There are two types of help: job skills training (skill) and drug counseling (counseling). There are 200 experimental subjects in the study. You have run the following regression log (earnings) = B_0 + B_1 skill + B_2 counseling + B_3 age + u and get an estimate of B_1 = 0.02 with se (B_1) = 0.01 and an RSS of 345.2 and an R - squared of 0.18. (a) What docs the R - squared tell you about the regression? (b) Does the job skill training haw a statistically significant effect on earnings? Conduct a one sided test at the 5 percent level. Which side did you put the test on and why? (c) How much docs one hour of job skills training affect earnings? (d) How would you rewrite the regression above to test the hypothesis that the return to job skills training is the same

Explanation / Answer

a. R square of 0.18 tells us that approximately 18% of the variation in the independent variable (log of earnings) is explained by the regression model.

b. We test :

H0: b1<=0

Ha: b1>0 as a one sided test

t stat = 0.02/0.01 = 2 >t199,0.05(1.65).

Hence we reject the null hypothesis and conclude that job skill training has a significant impact on earnings. We put b1>0 to test as we expect job skills to have a positive impact on earnings

c. One hour of job skills affects log of earnings by 0.02, hence earnings by e^0.02 - 1 = 1.0202 - 1 = 0.0202

d. We test H0 : b1-b2=0

Ha : b1-b2 not equal to zero

Here t = (b1-b2)/sqrt(se(b1)^2 + se(b2)^2)

Related Questions

Navigate

• Browse (All)
• Browse Y
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.