How to calculate leverage values in robust regression

[Thng]

Hi,

I am running a regression analysis that clusters standard errors (at the classroom level due to nested data). I am checking the regression assumptions, but ran into problems when trying to calculate the leverage values. I ran the command <predict h, hat> and <predict h, leverage>, but Stata returned an error message stating "Option hat not allowed after robust estimation."

Does anyone know why this is the case? Is it appropriate to calculate hat values in my case, and how do I go about doing it?

Thanks!

[Roman Mostazir]

Thng, please do consider registering yourself with your original First and last name, this is the preference of this forum. Use the "Contact us" tab at the bottom right corner of this page and request admin for changing the name.

On the point, if the idea is to only check the leverage, I would check the leverage with normal OLS regression without robust standard error. Estimating robust standard error does not affect the parameter estimate, it will change the S.E of parameter estimates and the test of it. Therefore, checking the leverage with normal reg should suffice. Or if your data has lots of noise, try robust reression with "rreg" command.

[Carlo Lazzaro]

Thng:
as an add-on to Roman's sound advice, you may be interested in taking a look at: http://www.statalist.org/forums/foru...reg-vce-robust