Thanks, Jeff - that did it!

Extracting the parameter indices was a bit tricky for more complex models, so here’s what I came up with, in case it might be helpful for others:

-Run the model with “profile.int=FALSE”. Here, I’ve named this model “m1”.

-“unique(m1$results$real)” will return the unique, real parameter estimates, but the “get.real()” function returns the par.index field you’ll need, so join these tables:

- Code: Select all
`merge(x=unique(m1$results$real), y=get.real(m1,"Psi",se=TRUE), by="row.names")`

Note that the “unique()” function may not return some parameters if they were estimated to be the same value as another parameter (eg. both were at a boundary, or had the same covariate value in a covariate model), so you may have to double-check that all parameters of interest are included.

-Repeat this join for each parameter type, as needed (in my case: S, p, and Psi).

-Re-run the model, specifying the vector of par.index values for profile CIs: eg. “profile.int=c(1,2,3)”.

Be patient as the model re-runs!