Robustness Check with stata

[Sabrina Schmidt]

Hi all,

I need to check the robustness of my model. I read that it is possible to check the robustness by specify the regression. I added additional controls or other measures of my independent variables. However, I want to do a regression which excludes missing observations of the independent variables.

I tried this command : xtreg var1 var2 var3 var4 dummy if var2=!(missing) & if var3=!(missing) & if var4=!(missing), fe vce (robust).

This command did not work. I do not even think if it´s the right approach. Hope you can help me.

Cheers
Linda

[Nick Cox]

As for "did not work", I can only guess that

if var2=!(missing) & if var3=!(missing) & if var4=!(missing)

was problematic as it's not legal code for indicating non-missings. But none of that is needed as if any one of those variables is missing, the corresponding observation will be omitted from the regression any way.

This looks legal to me

xtreg var1 var2 var3 var4 dummy, fe vce (robust).

but I wouldn't say that it provides robustness checks, just, it is to be hoped, more honest standard errors.

[Carlo Lazzaro]

Sabrina:
as an aside to Nick's helpul advice, I would also wonder to myself: "Robust with respect to what?".
:

[Sabrina Schmidt]

Thanks for help!

I thought that I can check the robustness of my base model "xtreg var1 var2 var3 dummy, fe vce (robust)". But as Nick said, the regression analysis already excludes missing values.