Stata handout #4 (interpreting stats & regression output) PDF

Preview

Summary

very helpful in understanding some specifics content. If you miss some lectures, this will be very instrumental to you to understand Stata....

Description

ECON 113

Stata Handout 4 Interpreting Stata Regression Output and Relationships Between Sample Statistics and OLS Estimates

Of the statistics produced by Stata’s regress command, those we’ve discussed so far are highlighted: . use wages.dta, clear . regress wage educ

“Sum of Squares”

Source | SS df MS -------------+---------------------------------SS model Model | 54070.9621 1 54070.9621 Residual | 156590.357 795 196.969002 SS residual -------------+---------------------------------Total | 210661.319 796 264.649898 SS total

Number of obs F(1, 795) Prob > F R-squared Adj R-squared Root MSE

= = = = = =

797𝒏 274.52 0.0000 0.2567R2 0.2557 14.035

Standard error (= 𝐯𝐚𝐫)

√ -----------------------------------------------------------------------------wage | Coef. Std. Err. t P>|t| [95% Conf. Interval] Dependent variable (𝒚): -------------+---------------------------------------------------------------Explanatory variables (𝒙): educ | 3.045239 .183797 16.57 0.000 2.684455 3.406024 2.677142 -6.65 0.000 -23.06528 -12.55507 Constant term (intercept) _cons | -17.81018 -----------------------------------------------------------------------------coefficient

You should also be familiar with the statistics produced by the following commands: . sum wage educ Variable | Obs Mean Std. Dev. Min Max -------------+--------------------------------------------------------wage | 797 25.77457 16.26806 .25 99 educ | 797 14.31242 2.706475 0 20 . corr wage educ (obs=797) | wage educ -------------+-----------------wage | 1.0000 educ | 0.5066 1.0000 . corr wage educ, cov (obs=797) | wage educ -------------+-----------------wage | 264.65 educ | 22.3064 7.32501

Note that if we run the “reverse regression” (educ on wage) we get: . regress educ wage Source | SS df MS -------------+---------------------------------Model | 1496.582 1 1496.582 Residual | 4334.12502 795 5.45172959 -------------+---------------------------------Total | 5830.70703 796 7.32500883

Number of obs F(1, 795) Prob > F R-squared Adj R-squared Root MSE

= = = = = =

797 274.52 0.0000 0.2567 0.2557 2.3349

-----------------------------------------------------------------------------educ | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------wage | .0842865 .0050872 16.57 0.000 .0743006 .0942723 _cons | 12.13997 .1550244 78.31 0.000 11.83567 12.44428 ------------------------------------------------------------------------------

ECON 113

Stata Handout 4

Questions: 𝟎, 𝜷  𝟏 , 𝒓𝒙𝒚 , 𝒔𝒙𝒚 , 𝒔𝒙𝟐 , 𝒔𝟐𝒚, 𝒔𝒙 , 𝒔𝒚 ,  𝒙 ,  1. In the output above, label: 𝜷 𝒚  𝟏 were missing from the regression table, how else could you calculate it? 2. If the value of 𝜷 There are at least three ways you can use the other statistics in the four tables! 3. If the value of 𝑹𝟐 were missing from the regression table, how else could you calculate it? There are at least three ways you can use the other statistics in the four tables! 4. In the “reverse regression” of educ on wage, which statistics stay the same as in the regression of wage on educ? 5. Can you calculate the slope coefficient on wage from the reverse regression using the other reported statistics (i.e., without seeing the table produced by regress educ wage)?...