Sarah Wu posted on Thursday, March 05, 2015 - 12:11 pm
hi dr. muthens,
i'm trying to run an LPA with twin data. i used the CLUSTER variable to account for the non-independence of the data. i'm curious if there is 1) anything inherently not okay about running an LPA with twin data, and 2) whether it impacts the interpretability of the TECH11 output (namely, the likelihood ratio tests)? i'm finding the results to be quite different when i account for the twins than when i don't. something i did not see with the EFA.