predicted probabilities and caterpillar (ie residual) plots after bayes: melogit

[Ariel Beccia]

Hello,

I am trying to obtain a caterpillar plot of group-level residuals and group-level predicted probabilities generated from a multilevel logistic model. For context, I have individuals at level 1 nested within groups at level 2, and I am using the "bayes: melogit".

bayes, prior({outcome:_cons}, flat): melogit outcome || group:, cov(un) 
bayes, showreffects({U0[1/16]}) 

egen pickone = tag(group)
generate pred_prob = invlogit({outcome:_cons} + {U0[group]})  
list group pred_prob if pickone==1

However, I keep getting an error that says the syntax is invalid for the generate pred_prob command. Also, I am unsure how to go about obtaining a caterpillar plot using the bayes command; normally I use "serrbar" if just using xtmelogit for example. Any help would be appreciated!

Thank you!!

[Ariel Beccia]

In case this was confusing, by caterpillar plots, I mean a plot of the group random effects. Thanks!

[Andrew Musau]

These are not stored in matrix e(b) as regular regression coefficients, so you should look at how the matrix storing the results is named from

ereturn list

In future, provide a reproducible example using dataex. The following example illustrates how you can generate your predictions (Stata 16), but I cannot help with graphing as I do not understand what you are after.

webuse bangladesh, clear
bayes, prior({ c_use :_cons}, flat): melogit c_use || urban :, cov(un)
bayes, showreffects({U0[1/2]})
mat list e(mean)
gen pred_prob=.
levelsof urban, local(urban)
foreach gr of local urban{
   replace pred_prob = invlogit(e(mean)[1,"c_use:_cons"]+ e(mean)[1,"U0[urban]:`gr'"]) if urban==`gr'
}

[Ariel Beccia]

thank you very much!