Outreg2: add Chi2 statistic for random-effects logit model

[Florian Schmidt]

Hi,

I'm relatively new at using Stata and want to create some output tables for my Master's thesis using outreg2. I figured out a lot via googling and some guides on the internet, but there are two things I just cannot get done:

1) I want outreg2 to include the Chi2-statistic from my xtlogit regression below the observations in the output table. I tried the option "addstat", but it didn't work.

My command is: "outreg2 using test.doc, replace ctitle (xxx) label". Now I would like to add the Chi2-statistic also...

2) I want to create a correlation table with pairwise correlations of all variables used. Additionally, the first two columns after the variable should show the standard deviation and the mean of the variable. I cannot find any information about creating correlation tables with outreg2...

Can I do these things with outreg2?

Thanks for your help!
Florian

[Oded Mcdossi]

As for your first question:

webuse union, clear
xtlogit union age grade i.not_smsa south year
outreg2 using myfile, adds(Likelihood-ratio test of rho=0, e(chi2_c), Prob, `=chiprob(1, e(chi2_c))*0.5') replace
seeout using "myfile.txt"

[Florian Schmidt]

Thank you for your fast reply!

However, now I have the following output below my Observations and ID.

But the values from the xtlogit were as follows:

Wald chi2(1) = 2.31
Log likelihood = -86.71818
Prob > chi2 = 0.1286

[Oded Mcdossi]

Yes this is different test from below the table. If you want the Wald chi2 from above, then:

outreg2 using myfile, adds("Wald chi2 (`=e(df_m)') ", e(chi2), Prob > chi2,  e(p)) replace

[Florian Schmidt]

Wow, great! Thank you so much Oded!

[Florian Schmidt]

Hi again.

Unfortunately, I realized that I have to use the Likelihood Ratio test for evaluating model fit instead of the Wald test. For my two panel models using xtlogit this works fine with above option:

adds(Likelihood-ratio test of rho=0, e(chi2_c), Prob, `=chiprob(1, e(chi2_c))*0.5')

However, I also have a non-panel model where I just use the logit function. The regression gives me the Likelihood Ratio test, but I cannot produce an output with outreg2 using above formula. Stata always tells me:

. outreg2 using Acquisition_Integration_final.doc, label ctitle (Absorption) adds(Likelihood-ratio test of rho=0, e(chi2_c), Prob, `=chiprob(1, e(> chi2_c))*0.5')

check eret list for the existence of e(chi2_c)
invalid syntax
r(198);

Can someone please help me with the option for a LR test in the logit function?

Thanks a lot!

[Oded Mcdossi]

In this case the code in #2 would work just fine:

outreg2 using myfile, adds("LR chi2(`=e(df_m)') ", e(chi2), Prob > chi2, e(p)) replace

You can access all these results that Stata save after estimation of models by the ereturn list command

[Florian Schmidt]

Thank you again, it works. Strange, because I used the code from #2 before and it didn't work... Always gives me back "Invalid syntax". But great, thanks!

[Hend She]

Hi, I have same problem when I try to add Wald test of exogeneity: χ 2 to the bottom of my regression table

outreg2 [probit ivprobit iv2sls] using myfile.doc, label adds("Wald chi2 (`=e(df_m)') ", e(chi2), Prob > chi2) replace

I get "check eret list for the existence of e(chi2)" too

[Santosh Dash]

Hi,

I want to export certain coefficients from a matrix in your example #2. In that example, I want to retrieve and export rho statistic.

  
 webuse union, clear xtlogit union age grade i.not_smsa south year  mat t = r(table) outreg2 using myfile, adds(Likelihood-ratio test of rho=0, e(chi2_c), rho, t[1,10])) replace 

I am getting the following typo:

too few ')' or ']'
r(132);

Oded Mcdossi Could you kindly help?

Thank you

[Oded Mcdossi]

It is easier to use e(rho) instead,

outreg2 using myfile, adds(Likelihood-ratio test of rho=0, e(chi2_c), rho, e(rho)) replace

but if you want to go with the matrix way you need to refer to it directly

outreg2 using myfile, adds(Likelihood-ratio test of rho=0, e(chi2_c), rho, `=t[1,10]') replace

[Santosh Dash]

Thank you Oded Mcdossi for your kind help.