Model Fit and Comparison in Finite Mixture Models: A Review and a Novel Approach

One of the greatest challenges in application finite mixture models is model comparison. A variety statistical fit indices exist, including information criteria, approximate likelihood ratio tests, and resampling techniques; however, none these describe amount improvement when a latent class added to model. We review statistics propose novel approach, increment percentage per parameter ( LIPpp ), targeting relative Simulation work based on two previous simulation studies highlighted potential for identify correct number classes, provide context magnitude fit. conclude with recommendations future research directions.