* Use BWGHT2.dta
regress lbwght npvis npvissq
summarize npvis if npvis > 21
regress lbwght npvis npvissq mage magesq
summarize mage if mage > 30
* Getting the R^2 and SST from the bwght equation
regress bwght npvis npvissq mage magesq
* R^2_bwght = 0.0192, SST = 593759796
* Now get the predict values of birthweight from the lbwght equation
regress lbwght npvis npvissq mage magesq
predict lbwght_hat 
generate bwght_tilda = exp(lbwght_hat + 0.0411052/2)
generate SSR_y_tilda = 1764*(bwght - bwght_tilda)^2
summarize SSR_y_tilda
* SSR_y_tilda = 5.85e+08
* R^2_y_tilda = 1 - SSR_y_tilda/SSt = 1 - 5.85e+8/593759796 = 0.0147531
*  Since the bwght equation had a higher R^2_bwght than the corresponding
* R^2_y_tilda implied from the lbwght equation, we go with the bwght specification.