[Niels Henrik Bruun]

Dear Hillel Alpert and Nerea Becerra
I can see I did promise too much in #27 and #30. I'm sorry!
The only way to handle clogit, mixed models and svy is by using -ici- and generate the needed variables as described in the -ic- help.
I have no plan in the near future of extending -ic- command.

Kind regards
nhb

[Roger Keller Celeste]

Hi,
Congrats on the command. I noticed it worked with counting variables as the outcome, but originally those measures (synergy index, etc) were developed for binary outcomes. Were there any adjustments made in the calculations to accommodate that type of variable? There is a paper by VaderWeele (IJE, 2007) about it, but it deals with continuous predictors.

Cheers,

[Niels Henrik Bruun]

Sorry for not answering before.
And sorry for not being precise.
The ic commands work for Poisson regressions with binary outcome.
With vce set to robust, the confidence intervals are "correct".
And this Poisson regression were/are more stable than the binomial regression.

[Roger Keller Celeste]

Sorry for not answering before.
And sorry for not being precise.
The ic commands work for Poisson regressions with binary outcome.
With vce set to robust, the confidence intervals are "correct".
And this Poisson regression were/are more stable than the binomial regression.

Thanks a lot