Probit vs Logit Asis Option and Standard Errors

[Matt Baugh]

Hello,

When I use the probit command and the asis option the coefficients asis includes (that would normally be ejected) come with standard errors and t stats.

When I use the logit command and the asis option the same coefficients do not have standard errors / t stats.

Does anyone know why logit does not report the SEs? Is there anyway to get them?

Thanks so much for your thoughts!

[Carlo Lazzaro]

Matt:
everything would be easier if you posted what you typed and what Stata gave you back, along with an example/excerpt of your data, that yiou can share with -dataex-. Thanks.

[Richard Williams]

The documentation says asis may introduce numerical instability. Maybe things are even more unstable with logit. I don't know exactly why anyone would want to use this option anyway, but maybe there is a reason.

[Matt Baugh]

Matt:
everything would be easier if you posted what you typed and what Stata gave you back, along with an example/excerpt of your data, that yiou can share with -dataex-. Thanks.

I'm not sure the legality of sharing (even a subset of) my data. I'd prefer to err on the side of caution.

But I've attached my Stata output which has the relevant commands. I'm interested in the coefficient/SE on restatement_nr.

I know that in my data there are no instances where dismissal_cf = 1 and restatement_nr = 1, which is why the variable would be ejected without the asis option. My question is mostly why does asis with probit report standard errors and asis with logit does not? And is there anyway to get SEs out of logit with asis? The output shows now errors or other warnings as far as I can tell to tip me off to a problem.

[Carlo Lazzaro]

Matt:
like Richard, I'm a bit surprised that you need to invoke -asis- option (which might be a part of the problem or the culprit).
That said, why not considering a panel data estimator (see -xtlogit- and -xtprobit-) given the structure of your dataset?