Hi everyone, i have a problem with a mixed logit model with multiple imputated data. I have an interaction term with two dummy variables- female & married-. I want to explore other combinations, like male##married or female##notmarried. I think i have two possibilities: change the reference group or get the margins of the other combinations. I tried both, change the reference group in the same ecuation obviously has a collinearity problem. So i runned separate models for every combination. I read that margins in mi data isnt eclass, but i dont know how to perform. My code is below.

Code:
mi estimate, cmdok eform: meglm expect_trab i.ses_nino dad female##ecivil contrato i.oficio edad salud i.nedu act_fisica depresion vivienda sit_laboral ||folio:, fam(bernoulli)
estimates store m1 
 margins female#ecivil, asbalanced post last estimates not found
i also found this but i still dont know what to do. It says that theres no way to perfom margins with mi estimate. Maybe 4 years laters someone discovered how to solve this.
https://www.statalist.org/forums/for...th-mi-estimate

please help