vcetype 'cluster' not allowed

[Claudio Schiener]

Dear all,

I have a weird problem runing my do-file.

If I run my dofile from the beginning it works till a point where I want to estimate a correlated random effects model.After my command:

xtlogit inlf i.heart2 c.age c.agesq c.num_kids c.net_hh_inc i.year c.heart2bar c.agebar c.agesqbar c.num_kidsbar c.net_hh_incbar if final==1, re vce(cluster pid)

I get the error message: "vcetype 'cluster' not allowed". So I would expect that I cannot use clustered standart errors in this regression.

Now the wired thing is that if I start the dofile agian with the problematic comand (without any changes), the command perfectly works and I dont get any error message and the model is estimated. Here is the Output

.....
. *Coloumn 3 of table 5: CRE with logit
. xtlogit inlf i.heart2 c.age c.agesq c.num_kids c.net_hh_inc i.year c.heart2bar c.agebar c.agesqbar c.num_kidsbar c.net_hh_incbar if final==1, 
> re vce(cluster pid)
vcetype 'cluster' not allowed
r(198);

end of do-file

r(198);

. do "C:\Users\cschiene\AppData\Local\Temp\STD4114_000000.tmp"

. *Coloumn 3 of table 5: CRE with logit
. xtlogit inlf i.heart2 c.age c.agesq c.num_kids c.net_hh_inc i.year c.heart2bar c.agebar c.agesqbar c.num_kidsbar c.net_hh_incbar if final==1, 
> re vce(cluster pid)

Fitting comparison model:

Iteration 0:   log pseudolikelihood = -24591.609  
Iteration 1:   log pseudolikelihood = -20657.731  
Iteration 2:   log pseudolikelihood = -20305.965  
Iteration 3:   log pseudolikelihood = -20301.557  
Iteration 4:   log pseudolikelihood = -20301.554  
Iteration 5:   log pseudolikelihood = -20301.554  

....

Any idea why this happens or should I contact the technical support?

Thanks for your help!

[daniel klein]

Check your version statements. Earlier releases of Stata (before 13 or 14, cannot remember right now) did not accept the vce(cluster ...) option for xtlogit.

Best
Daniel

[Claudio Schiener]

Thank you very much! Problem solved