No corrected R-squared when I weight by pweight!

[Danny Gecerguen]

Hello all,

Within my bachelor thesis I am performing several multiple regression analyses, which I would also like to compare using the corrected R-squared value.

I am using a dataset which already contains a representative weight: In agreement with my supervisor I use the probability weighting (pweight).

The problem now is that I do not get the corrected R-squared value output by Stata for this case. I am a bit puzzled by this, since this value is just an adjustment of R-squared, which should work independently of the weighting?

I have now calculated the corrected values manually, that is not the problem. Rather, I would need a good explanation why I had to do this calculation myself - preferably also with corresponding evidence from the literature.

Maybe someone can help me and explain the problem.

Thanks in advance!

Addition: The uncorrected R-squared is given to me normally

[Rich Goldstein]

see #'s 3 and 5 in https://www.statalist.org › forums › forum › general-stata-discussion › general › 1164515-how-to-get-adjusted-r-square-in-survey-regression

[Dirk Enzmann]

I found two posts related to your question: https://www.statalist.org/forums/for...eneral/1164515 and https://www.stata.com/statalist/arch.../msg00682.html. Perhaps Richard Williams can join in?

Cross-posting with #2

[Richard Williams]

Others have a better memory for my old posts than I do! I agree with my 2015 self -- it is rare that adjusted R^2 is useful, and, I suspect, even rarer for it to be useful with a pweighted sample.