Confirmatory Latent Class Analysis: Model Selection Using Bayes Factors and (Pseudo) Likelihood Ratio Statistics

Inequality constraints among class specific probabilities can be used to assign a specific meaning to the classes in a latent class model. Different models arise if different sets of constraints are used. In this paper, model selection using Bayes factors, and, (pseudo) likelihood ratio statistics evaluated using posterior predictive p-values, will be discussed. It will be illustrated that these Bayesian selection criteria do not suffer from the same flaw as maximum likelihood based selection criteria. Using a small simulation study it will be shown that, in the context of the simulation study, Bayes factors and the pseudo likelihood ratio statistic have the best proporties. The article will be concluded with an example.

[1]  G. Schwarz.  Estimating the Dimension of a Model , 1978 .

[2]  D. Rubin.  Bayesianly Justifiable and Relevant Frequency Calculations for the Applied Statistician , 1984 .

[3]  H. Akaike.  Factor analysis and AIC , 1987 .

[4]  H. Bozdogan.  Model selection and Akaike's Information Criterion (AIC): The general theory and its analytical extensions , 1987 .

[5]  B. Everitt.  A Monte Carlo Investigation of the Likelihood Ratio Test for Number of Classes in Latent Class Analysis. , 1988, Multivariate behavioral research.

[6]  F. T. Wright,et al.  Order restricted statistical inference , 1988 .

[7]  Marcel A. Croon,et al.  Latent class analysis with ordered latent classe , 1990 .

[8]  Scott L. Zeger,et al.  Generalized linear models with random e ects: a Gibbs sampling approach , 1991 .

[9]  Marcel A. Croon,et al.  Investigating Mokken scalability of dichotomous items by means of ordinal latent class analysis , 1991 .

[10]  B. Junker.  Conditional association, essential independence and monotone unidimensional Item response models , 1993 .

[11]  M. Newton.  Approximate Bayesian-inference With the Weighted Likelihood Bootstrap , 1994 .

[12]  Xiao-Li Meng,et al.  Posterior Predictive $p$-Values , 1994 .

[13]  Xiao-Li Meng,et al.  POSTERIOR PREDICTIVE ASSESSMENT OF MODEL FITNESS VIA REALIZED DISCREPANCIES , 1996 .

[14]  I. W. Molenaar,et al.  A multidimensional item response model: Constrained latent class analysis using the gibbs sampler and posterior predictive checks , 1997 .

[15]  Robert J. Mokken,et al.  Nonparametric Models for Dichotomous Responses , 1997 .

[16]  C. Mitchell Dayton,et al.  Model Selection Information Criteria for Non-Nested Latent Class Models , 1997 .

[17]  H. Hoijtink.  Constrained Latent Class Analysis Using the Gibbs Sampler and Posterior Predictive P-values: Applications to Educational Testing , 1998 .

[18]  Kazuo Anraku,et al.  An information criterion for parameters under a simple order restriction , 1999 .