Lab Session Bwght - It\'s a lab assignment that counted as homework. PDF

Preview

Summary

It's a lab assignment that counted as homework. ...

Description

Exercise 2 (Stata) 1. Open the data set “BWGHT.DTA” (Stata format) 2. Produce descriptive statistics for cigs, bwght, fatheduc Useful commands: sum (for continuous variables), tab (for categorical variables) Variable | Obs Mean Std. Dev. Min Max -------------+--------------------------------------------------------cigs | 1,388 2.087176 5.972688 0 50 bwght | 1,388 118.6996 20.35396 23 271 fatheduc | 1,192 13.18624 2.745985 1 18

3. Regression analysis using OLS (use bwght as Y) a. Start with one control variable: cigs. Interpret the estimated slope coefficient. With every additional cigarette smoked during pregnancy, the baby’s weight is expected to drop by .5137721 ounces. This is significant at the 1% level because of the p-value being less than 0.00. b. Then multiple controls: cigs,faminc,parity,male,white. Interpret the estimated slope coefficients and coefficient of determination. With every additional cigarette smoked during pregnancy, the baby’s weight is expected to drop by .4899816 ounces. It’s significant because the p-value is below 0.00, which means it is significant at the 1% confidence level. The coefficient of determination is 0.0526 means that 5.26% of the variation of bwght around its mean can be explained around our regression equation. c. Add fatheduc to the above set of controls. What happens to the number of observation used in this regression? When adding the variable fatheduc the number of observations decreases from 1,388 to 1,192, meaning that there are missing values from the fatheduc variable. Since there is substantial difference in number of observations you cannot compare these results with the regressions done before. 4. Results table: look for important information such as coefficients, t-test and overall significance test results.

Some other important coefficients in the table are significant are the parity, male, white, and _cons. Faminc has a coefficient of 0.0365999 which means that the baby’s weight will increase 0.0365999 ounces for every unit increase in Faminc and it has a p-value of 0.313, which means it is statistically insignificant. Parity has a coefficient of 1.976294, which means that the baby’s weight increases 1.976294 oz for every unit increase in the parity variable and it is statistically significant because the p-value is 0.003. Male has a coefficient of 3.81355 and this means that if the baby is a male it’ll be 3.81355 oz heavier than if it were a female and it is statistically significant at the 5% level with a p-value of 0.001. White has a value of 4.770994, which means that the baby will be 4.770994 oz heavier if it is white compared to if it is born a different race and it is statistically significant because it has a p-value of 0.003. Fatheduc has a coefficient of 0.2569012 which means that the baby’s weight will increase 0.2569012oz with every unit increase in its father’s education. Another significant value is the coefficient of determination, which is 0.0534, meaning that 5.34% of the variation of bwght around its mean can be explained around our regression equation. 5. Case selections: e.g., reg bwght cigs faminc if cigs>0

6. Missing values: identify and exclude. Command: reg bwght cigs faminc if fatheduc!=.

7. Try alternative combinations of control variables as well as different measures of the dependent variable. What is your final choice of the dependent variable (i.e., bwght, lbwght, or bwgthlbs) and control variables?

8. Write down your version of the estimated regression equation and interpret each slope coefficient. See if you can make a conclusion regarding the effect of smoking on birth outcomes. Bwght = 119 - .5839cig

9. Based on your estimated regression equation, calculate the residuals for the 2nd and 36th observations. Command after running the regression: predict residual, r The 2nd = .8513018 The 36th = -.4925316...