CMP models-problems with notation and other issues

[Rafael Garcia]

Ok, I am trying to run this code for a cmp model

svy (etc.)

cmp setup

svy: cmp (hindex08 = fem92 white92 wpart92 raeduc92 cendiv92 hindex92 c.age92##c.age92 nonsmk92 c.bmi92##c.bmi92 incpvr92) ///
(hindex08 = ins92 fem92 white92 wpart92 raeduc92 cendiv92 hindex92 c.age92##c.age92 nonsmk92 c.bmi92##c.bmi92 incpvr92) ///
(ins92 = fem92 white92 wpart92 raeduc92 cendiv92 hindex92 c.age92##c.age92 nonsmk92 c.bmi92##c.bmi92 incpvr92 risk92) , ///
ind("cond(hindex08, $cmp_cont, $cmp_left)" ins92 $cmp_probit) nonrtolerance

which runs perfectly on other laptop, but no way on mine. Only taking off the ///, and then I would get a totally different output from hers. Any ideas?

Besides the syntax problems, a greater one. As probably some of the readers might spot right away, I am trying to model censored data(hindex08-tobit?), suspecting there might be some sample selection problems (heckman model?), and suspecting the main regressor to be endogenous (ins92-iv technique?). So I am trying to setup the equivalent of a tobit-heckman-iv, and as I said, it ran on another laptop, yet not sure where, from all the output, where I can see the effects of my main regressor. Any ideas with this cmp analysis? Thanks in advance for any help.

[Konrad Zdeb]

as I said, it ran on another laptop (...)

So is my understanding correct that it's not running on some other laptop? As Stata, same as any other software should behave in nondeterministic (unless forced to behave pseudorandomnly) manner it has to work. I would suggest that that you post sample data and elaborate on the code you are trying to get (are you getting an error message?). I must admit that I find your message slightly cryptic, but maybe it's because I'm not an expert on the subject. On a more generic matter, if you are running code on different computers and is working on one the issue may be related to limits (if you have different versions of Stata) and different configurations (SSC packages, etc.), but that's just a wild guess.

Please also note that the forum requires users to use their full names. It also appears that you posted the same post twice, you may consider deleting one.

[Rafael Garcia]

Thanks for the remarks and suggestions. I am very willing to post the complete code (commands and output) but not sure what do you mean by posting sample data. A description of the variables? Some descripite statistics? Thanks in advance for the help. I really appreciate it.