Journal article
Goodness-of-Fit Testing for Latent Class Models
Multivariate behavioral research, Vol.28(3), pp.375-389
01/01/1993
DOI: 10.1207/s15327906mbr2803_4
PMID: 26776893
Abstract
Latent class models with sparse contingency tables present problems for model comparison & selection because the distributions of goodness-of-fit indices are often unknown, which causes inaccuracies both in hypothesis testing & in model comparisons based on normal indices. Here, the extent of this problem is assessed in a simulation investigating the distributions of the likelihood ratio statistic G2, the Pearson statistic X2, & a new goodness of fit index suggested by T. R. C. Read & N. A. C. Cressie (1988). Findings reveal substantial deviations between the expectation of the chi-squared distribution & the means of the G2 & Read & Cressie distributions. In general, the mean of the distribution of a statistic was closer to the expectation of the chi-squared distribution when the average cell expectation was large, there were fewer indicator items, & the latent class measurement parameters were less extreme. It was found that the mean of the X2 distribution is generally closer to the expectation of the chi-square distribution than are the means of the other two indices examined, but the standard deviation of the X2 distribution is considerably larger than that of the other two indices & larger than the standard deviation of the chi-squared distribution. A possible solution is to forego reliance on theoretical distributions for expectations & quantiles of goodness-of-fit statistics; instead, Monte Carlo sampling can be used to arrive at an empirical central or noncentral distribution. 4 Tables, 19 References. Adapted from the source document.
Details
- Title: Subtitle
- Goodness-of-Fit Testing for Latent Class Models
- Creators
- Linda Collins - University of Southern CaliforniaPenny Fidler - University of Southern CaliforniaStuart Wugalter - University of Southern CaliforniaJeffrey Long - University of Southern California
- Resource Type
- Journal article
- Publication Details
- Multivariate behavioral research, Vol.28(3), pp.375-389
- DOI
- 10.1207/s15327906mbr2803_4
- PMID
- 26776893
- ISSN
- 0027-3171
- eISSN
- 1532-7906
- Language
- English
- Date published
- 01/01/1993
- Academic Unit
- Psychiatry; Biostatistics
- Record Identifier
- 9984280866602771
Metrics
8 Record Views