Confidence interval for predicted probabilities of the mlogit

[arma ayat]

Hi,
I want to calculate confidence interval for predicted probabilities of the mlogit. For example, after below command I want to find CI for predicted probabilities of each cases. How can I figure out?

predict pmlogit1 pmlogit2 pmlogit3 , pr

for example, https://www.stata.com/support/faqs/s...nce-intervals/ said for logistic these commands:

 . webuse lbw, clear
 . logistic low age lwt i.race smoke, coef  . predict phat   
 . predict xb,xb   
 . predict error, stdp
 . generate lb = xb - invnormal(0.975)*error
 . generate ub = xb + invnormal(0.975)*error
 . generate plb = invlogit(lb)  
. generate pub = invlogit(ub)

[Hong Il Yoo]

Hi Arma, you may already know of -margins- that allows you to compute marginal effects of regressors and associated CIs. You can also use -margins- to compute predicted probabilities and associated CIs. For example, in Stata 16, if you type -margins- without any option, you'll get the mean predicted probabilities in your estimation sample along with the 95% CIs.

[arma ayat]

Thank you for your answer.
previously I tried to use -margins- for my problem, but I cannot find the option of margins that compute predicted probabilities and associated CIs for each case of dataset. like below that compute predicted probabilities for each case:

 predict pmlogit1 pmlogit2 pmlogit3 , pr

Do you know how can handle this by -margins-? My Stata version is 15.0.

[Hong Il Yoo]

Hi Arma, in Stata 15, if you type -margins- without any option, you'll get the mean predicted probabilities in your estimation sample along with the 95% CIs. If you're interested in predicted probabilities evaluated at a particular combination of individual characteristics, you can add -at(.)- option to -margins-.