regress rob, regression model with robust std. errors

[Iryna Ulmer]

Dear community!

I have a model with the violation the assumption about homoscedasticity. So first i run the normal model and then i run the regression with robust std. errors to handle with heteroscedasticity.
My question is: what exactly STATA do by calculation of these new robust std. errors?

Thanks for help in advance

[Carlo Lazzaro]

Iryna:
just type -help robust- and access all the explanations you need.

[Phil Bromiley]

The pdf documentation often has additional discussion not present in the help from the command line (although the help from the command line usually mentions where the pdf documentation has the additional discussion).

You can find an index by subject by clicking on Index, then Subject Index, then the appropriate letter. Under robust, it mentions regress with robust errors. In the documentation for regress, there is a section on robust standard errors. And, like most Stata documented procedures, the back of the entry on regress has the details of calculations [see Robust calculation for regress, page 2256 in the Stata 15 documentation - appears in all Stata documentation but probably on different page numbers.]

[Iryna Ulmer]

Thank you for all your answers.

It will a little bit difficult for me to understand the explanation by the helf command or by the stata user guide.

Can somebody explain, how exactly will be the robusted std. errors calculated? in which kind of way they will be greater as by the normal regression modell (OLS) and maybe some easy example?

It is very important for my bachelor thesis presentation

Thanks a lot for all the help

[Carlo Lazzaro]

Iryna:
the Stata .pdf manual devotes a paragraph to that issue (see 20.22 Obtaining robust variance estimates).
Moreover, the topic you're interested in is covered in any decent textbook of econometrics.
As an aside, if -robust- is very important for your bachelor thesis, it is worth devotinbg some of your time to get familiar with its features, that cannot be made easier than necessary.