Question

Model 1:

You are undecided about whether to include education (educ) and gender (male) as controls. The output from a regression where you exclude education and gender is shown below: MODEL 2: affair бо + 6,relig = reg affair relig df SS MS Number of obs 601 Source F (1, 599) 10.31 Model Residual| Prob > F 0.0014 1.90494147 1.90494147 599 0.0169 110.657455 184736986 R-squared Adj R-squared 0.0153 = Total 112.562396 600 .187603993 .42981 Root MSE Interval] affair Coef Std. Err. [95% Conf. P>t relig -.0482619 0150294 -3.21 0.001 -.0777786 -.0187453 .399991 0500124 8.00 0.000 .3017701 498212 cons 4. Use the Stata output from the two models to decide which specification is better. [Hint: Determine whether education and gender are jointly significant or not.] Clearly state the null and alternative hypothesis. Do you reject the null hypothesis at a 5% significance level? Explain how you reach this а. b. conclusion. Which model do you choose? Explain why using your answers to 4a. and 4b above. C.

Answer 1

from the stata output we are going to find the significance of the variables education and gender.

A) this is a t test for the significance of the variable education

H1 20

Here the p value obtained is 0.852 which is greater than 0.05 , we do not reject the null hypothesis,hence the variable education is not significant

This is a t test for the significance of the variable male.

Ho 30 H1 B30

Here the p value obtained is 0.212 which is greater than 0.05,   we do not reject the null hypothesis,hence the variable male is not significant

But from the overall F test we see that the regression model 1 is significant.

Since the variable education and male is not significant we go with model 2.