Scalar c95=invnorm(.95) display c95



This assignment is due by Wednesday 10/17. For the problems in this section, use the following data set: g:\eco\evenwe\eco311\wooldridge data\FERTIL1.dta.This data set contains information on the number of children born to a woman and was used in a study to exam the effect of parental education on number of children. Key variables are the number of children born to the mother (kids), years of education received by the woman (educ), her age (age), and a dummy variable indicating whether she is African-American (black).Output all the regressions you estimate below into a single word file using esttab, An illustration of how this is done is provided below:reg kids age educ/*estimate the regression*/eststo spec1/*store the results as spec1*/reg kids age educ blackeststo spec2esttab spec1 spec2 using m:\eco311\ch3_homework.rtf, replace /*save table of regr results*/You may be required to provide output this way in a future quiz/exam – so be prepared. You may also be asked to use the techniques below for other questions on future quizzes.Estimate a simple regression of kids on educ.What is the coefficient on educ? What does it imply about the effect of education on number of children?How many more or less children would a woman with 16 years of schooling be predicted to have as compared to a woman with 12 years?Compute a 90% confidence interval for the estimated effect of education on kids. (Note: you can use the following commands to get the critical value at the.95 level in stata. We want the area to the left of the critical value to be .95.scalar c95=invnorm(.95) display c95Can you reject the null hypothesis that education has zero effect on the number of children at the .05 level? .01 level? Explain.Estimate a regression of kids on educ, age, and black.What is your interpretation of the coefficient on age? Is it statistically significant at the .05 level? .01 level?What is your interpretation of the coefficient on the black dummy variable? Is it statistically significant at the .05 level? .01 level? [Hint: refer to the p-values to draw your conclusions.]What is the predicted number of children for a 40 year old white woman with 16 years of education? Note: if you want to create a scalar in Stata using a combination of coefficients you can execute the following commands:scalar yhat=_b[_cons]+2*_b[_meduc] + …._b[x] retrieves the coefficient from the most recently executed regression for the variable x. If you wanted to create a variable instead of a scalar, you could use the following type of command:gen yhat = _b[_const] + _b[meduc]*meduc + ….If the woman in (c) has 4 children, what is the prediction error (residual) for this woman?[you can use the scalar command in Stata to generate the result and keep it in your log file].What happened to the R-squared when age and black were added as control variables to the regression? Will R-squared always move in this direction when you add control variables?Re-estimate the regression in (2) after replacing educ with educ_dev where educ_dev is defined as education measured as the deviation from its mean. How does this change your regression results? (i.e. do any coefficients, standard errors, or t-stats change?) Explain.Note: you can create a variable containing the mean of a variable using the “egen” (extended generate) command as follows:egen educ_mn=mean(educ) /*create a variable containing mean of education*/gen educ_dev=educ-educ_mn /*create a variable containing dev from mean*/Re-estimate the regression in (2) after replacing age in years with age measured in decades (i.e. age/10). How does this change your regression results? (i.e. do any coefficients, standard errors, or t-stats change?) Explain.Re-estimate the regression in (2) but add the variable farm (dummy indicating whether the woman lived on a farm at age 16).How does living on a farm at age 16 affect the number of children?Is the effect of living on a farm statistically significant at the .05 level?How does the coefficient on educ change when farm is added to the regression? Based on what you know about omitted variables bias, what two conditions must be necessary to cause the coefficient on educ to move in the direction you observe?As discussed in class, after estimating your regression, you can generate predicted values for the dependent variable (kids) and the estimated residuals.reg y xpredict yhat, xb predict uhat, residualUsing the regression model in (2), compute the variance of the predicted residual. How does this compare to the “root mean squared error” reported at the top of the regression output?What is the mean of uhat? Why should you have expected this?Compute the correlation between uhat and each of the control variables using the corr command in Stata. . What are these correlations? Why should you have expected this?How does the average value of yhat compare to the mean of kids? Why should you have expected this? ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download