Selection of the number of classes in latent class analysis. The BIC and the CAIC are criterion for model selection. Good models (e.g., the true model or nearly true models) have relatively small BIC and CAIC values.
No. Classes
BIC
CAIC
2
868.0
926.0
3
895.3
983.3
4
932.4
1050.4
5
1004.8
1152.8
6
1091.4
1269.4
Significant data in bold face. BIC: Bayesian Information Criterion; CAIC: Consistent Akaike Information Criterion.