I am trying to recreate the results in this table. The results have been obtained by difference in difference estimation. I can obtain values from all columns except for column 5 and 6.

Column 5 says that it is obtained by subtracting column 1 from column 2, and this is simple to do for the means, however I don't know of a way to add OLS regression standard errors of 2 groups together to give a combined standard error, and I don't know of any other way to estimate the combined standard errors.

Does anyone know if/how this can be done. Or does anyone have any advice on alternative methods for obtaining these values.

Thank you for any help.


tables i need to recreate


Let $X_1$ be column 1, $X_2$ be column 2, $X_5$ be column 5.


$$X_5 = X_2 - X_1$$ $$Var(X_5) = Var(X_1) + Var(X_2) -2Cov(X_1,X_2) =[SE(X_1)]^2 + [SE(X_2)]^2 -2Cov(X_1,X_2)$$

And $SE(X_5)=\sqrt{Var(X_5)}$.

So you need to get $Cov(X_1,X_2)$ in order to get $SE(X_5)$. Or you can prove that $X_1$ and $X_2$ are independent such that $Cov(X_1,X_2) =0$.


Your Answer

By clicking "Post Your Answer", you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.