Hello everyone,

I have done LPA analysis and because of continuing AIC and BIC values, I have decided 4-class model based on ease to interpret and parsimony of the model. However, the reviewer came back with the below comment.

"Probably you will not lose too many people if you use a model with 4 segments but then identify and discard people who don't fit any segment well."

From my understanding, everyone has to be assigned to one of the identified profile class based on the posterior probability. Not sure how to response and what the reviewer referring too - I mean I thought this is about classification error but not sure how to address this issue.

Please share your experience and knowledge?

any idea appreciated.