Display R2 with esttab command

[Sophie Klein]

Dear Stata listers,

in my fixed effects model linear Regression I want to provide r2 within values in the Output table.
I´m using the esttab command which currently Looks like this:

xtreg $ylist $xlist i.Year, fe vce(robust)
eststo: quietly xtreg $ylist $xlist i.Year fe vce(robust)
estadd local fixed "yes", replace

xtreg $ylist $xlist age*tenure i.Year, fe vce(robust)
eststo: quietly xtreg $ylist $xlist age*tenure i.Year, fe vce(robust)
estadd local fixed "yes", replace

esttab se(2) b(2) star(* .01 ** .005 *** .01) s(fixed N, Label("Year fixed effects")) drop (*Year)

The Output provides the Models in one table and Displays everything I wanted correctly: two Point decimals for coefficients and Standard Errors, significance Levels as indicated by stars at the coefficients and a bottom line stating "fixed effects" yes for every model.

Now, I´d like to add some Information about r2. Most preferably I want to disclose r2 withing because that is the r2 being of interest for the fixed effects model.
However, I can not figure out a way how to do that. At most I manage to get Overall r2.

Is it not possible at all to Display the values for r2 within in Connection with the esttab command? If that is the case, could anyone please provide some help?

Thank you very much,
Sophie

[Sophie Klein]

Sorry I made a mistake with the star Labels, I meant:
star(* 0.1 ** 0.05 *** 0.01)

[Andrew Musau]

stats(r2_w)

[Sophie Klein]

Hi Andrew,

thanks for your Reply. Unfortunately, however, it is not possible to put both commands "stats" and "s(fixed N, Label("year fixed effects")).
When I try to put both commands, Stata remarks:

"Option stats() not allowed" if I put the s command first and
"Option s() not allowed" if I put the stats command first resprectively.

Alternatively, is there adifferent way to Display the bottom row with "fixed effects = yes" ?

Thank you,
Sohpie

[Andrew Musau]

stats() and s() are one and the same

esttab, se(2) b(2) star(* 0.1 ** 0.05 *** 0.01)  stats(fixed N r2_w, label("Year fixed effects")) drop (*Year)

So, no commas between displayed statistics (if displaying multiple stats)


Also, you do not need to accept default names such as "r2_w" because these may not be immediately apparent to the reader. Therefore, you can name a scalar and feed the value of the stat of interest to this scalar, e.g.,

estadd scalar Within_R2= e(r2_w)
esttab, se(2) b(2) star(* 0.1 ** 0.05 *** 0.01)  stats(fixed N Within_R2, label("Year fixed effects")) drop (*Year)

[Sophie Klein]

Thank you so much Andrew!

[Tingting Tan]

Hello,

I am having trouble with the code displaying pseudo square in logistic regression using esttab. I tried scalars(r2) and stats(r2) but neither worked.

eststo r1:  logit stem_eh ${covsZ} ,
                fitstat
                estadd scalar r2 = r(r2_mz)
    eststo r2:  logit stem_eh ${covsZ} ${covsA} ${covsB} ,
                fitstat
                estadd scalar r2 = r(r2_mz)
    eststo r3:  logit stem_eh ${covsZ} ${covsA} ${covsB} ${covsC} ${covsD},
                fitstat
                estadd scalar r2 = r(r2_mz)

esttab r1 r2 r3 using "logit1.csv", ///
        b(%9.2fc) se(%9.2fc) scalars(r2)  ///
        replace label compress nogaps

Thanks,
Tingting

[Andrew Musau]

After running the estimation command, type

ereturn list

to see how the statistic is named. The pseudo-R2 is "e(r2_p)".