Tim Seifert posted on Wednesday, May 30, 2007 - 3:59 am
Hello, I have run an LCA using SAS Proc LCA and the resultws I get are exactly what they should be. I run the LCA in MPlus and the solution is considerably different than the SAS solution. What might explain the differences and how might I get convergence?
You should send the SAS output and the input, data, output, and your license number to firstname.lastname@example.org so we can see how the SAS and Mplus models differ. If you have the same model and the same estimator, you should get the same results.
I ran the analysis in MPlus, LCAP and Proc LCA. The MPlus and LCAP solutions were similar. All three had similar loglikelihood functions. But the BIC for SAS was different than that for MPlus and LCAP. I asked the question of the author of Proc LCAP and the solution seems to be a difference in the way the BIC is calculated. They use a different equation.