STATA Not giving results

[Oliva Safari]

Friends,

I have this on table;
I fitted a random slope model. But i am very surprised that STATA project took around two DAYS and Yet nothing is shown. ie. No results coming up. It is just STATA keeps running a command, non stop with just this part " Refining starting values" and sometimes i am forced to break the process.
I have tried to copy and past command and the output displayed over the course of two days while keeps running.
Does any one have idea on this? Would appreciate a lots.

. xtmelogit mc i.age_cat wealth* || v101:wealth*, cov(unstructured)||v101:, or


Refining starting values:

--Break--
r(1);

[Daniel Shin]

There could be several factors. It could be your dataset (obs, # of variables, etc), it could be the machine, it could be the complexity of the model you're fitting (number of parameters, etc), or combination of these factors. I've had issues with xtmelogit being computationally difficult and had to rethink the modeling approach.

[Oliva Safari]

Thank you so much Daniel for a reply.
In the end you used what modelling approach, i would opt that.

Thanks so much, i real appreciate your reply

[Leonardo Guizzetti]

Knowing nothing of your data (none was shown) or substantive question (nothing was described), your model is mis-specified.

you have specified your model with -xtmelogit- as:

xtmelogit mc i.age_cat wealth* || v101:wealth*, cov(unstructured)||v101:, or

Note that the unless your are running a very old version of Stata, the modern command is -melogit- (or -meqrlogit- if -melogit- fails).

You appear to want a repeated-measures model with an unstructured covariance among your wealth random slopes and intercept, but have then nested -v101- below that, which eliminates those random slopes. Try this instead:

melogit mc i.age_cat wealth* || v101:wealth*, cov(unstructured), or