
Message/Author 


Hello, I conducted a LPA in Mplus and am choosing between a threeclass solution with diagonal classinvariant structure and a twoclass solution with a diagonal classvarying structure. The twoclass solution is preferred by model fit statistics (BIC, Consistent AIC, approximate correct model probability, etc.), although the threeclass solution is more theoretically compelling. I believe that I should select the twoclass solution given better model fit; however, entropy is somewhat low (.69) which results in negative BCH weights when I attempt latent class regression. In this circumstance, is it defensible to select the threeclass model even though its relative fit indices are poorer, given that it has higher entropy (.89) and does not run into difficulty estimating BCH weights? Alternately, is there another method of LCR that you would recommend to sidestep the negative BCH weight problem? Thanks so much for your help! 


Sometimes using more classes than say BIC suggests can be motivated substantively I think  for instance if one class gets split into two more meaningful ones. You may want to survey the opinions on SEMNET for this. 

Back to top 

