Latent Profile Analysis in R

1
votes

I would like to know if anyone does know a possibility to conduct a latent profile analysis within R. This kind of SEM-model utilizing continuous manifest variables to identify a latent categorial variable can be done within MPLUS (see here for an example), but I did not find any comparable approaches within lavaan or any other R-package (although I am not sure if openMX can do it).

Questions:

1.) Does anyone has a suggestion for doing this as a SEM within R?

2.) Would be any classification algorithm like clustering or ordinal regression also appropriate to do the job?

Thanks!

2
r

2 Answers

4
votes

Have you tried the mclustModel in the mclust library? 

library(foreign)

library(mclust)

idd<- read.dta(data)

idd_ss <-subset(idd,select=c(variable1, variable2, variabel3, variable4, variable5))

iddBIC<- mclustBIC(idd_ss)

cl <- mclustModel(idd_ss,iddBIC)

cl
0
votes

The best you can try is poLCA which is a package outputting polytomous (categorical) latent classes.