Using conditional independence for parsimonious model-based Gaussian clustering
Statistics and Computing
| Hi-index | 0.03 |
The paper is aimed to investigate the performance of information criteria in selecting latent class analysis models which are often used in research of phenotype identification. Six information criteria and a sample size adjustment (Psychometrika 52 (1987) 333) are compared under various sample sizes and model dimensionalities. The simulation design is particularly meaningful for phenotypic research in practice. Results show that improvements by the sample size adjustment are considerable. In addition, the sample size and model dimensionality effects are found to be influential in the simulation study.