

Between level latent class analysis 

Message/Author 


(1) I wish to specify a betweenlevel latent class variable in the context of a multilevel model  but am wondering what might be the minimum number of cluster units (level 2) for this analysis to work? I have approx. 37 level 2 units(neighborhoods)  but am thinking that this might be too few because in a singlelevel latent class analysis, one usually wants/needs several hundered or more individuals. Is this the same for a level 2 factor. (2) Also, is it possible to also have a level 1 latent class factor in the same model? Thanks, Scott 


1) Yes, you are right. Unless the class differences are very, very large. 2) I assume that when you say "latent class factor" you mean a latent class variable. If so, the answer is yes  see examples 3 and 6 in the Version 4.2 Addendum. 

Sean Mullen posted on Thursday, March 18, 2010  2:53 pm



Is the above mentioned example still available somewhere? I'm having a hard time with Henry & Muthen (in press) and would love to see another example. 


No. 

Back to top 

