Table 4.
Model fit comparisons for two-class/one-factor FMM-3 with different covariate effects
| Model | LL | No. of parameters | BIC | SABIC | Entropy | |
|---|---|---|---|---|---|---|
| 1 | No covariates1 | −2481.07 | 29 | 5142.70 | 5050.65 | 0.54 |
| 2 | Model 1 + sex on latent class | − 2434.62 | 30 | 5056.04 | 4960.82 | 0.76 |
| 3 | Model 1 + sex on observed variables | − 2416.15 | 38 | 5068.90 | 4948.29 | 0.44 |
| 4 | Model 2 + age on latent class | −2434.22 | 31 | 5061.47 | 4963.07 | 0.74 |
| 5 | Model 2 + age on observed variables | − 2403.71 | 31 | 5000.44 | 4902.05 | 0.79 |
Notes. BIC Bayesian Information Criterion, BLRT parametric bootstrap likelihood ratio test; LL =
Log-Likelihood, FFM factor mixture model, SABIC sample size adjusted BIC
1 corresponds with the two-class/one-factor FMM-3 from Table 1