Table 2.
Model Fit Statistics for Determining the Optimal Number of Classes
| Model | BIC | LMR LRT (p) | Class assignment probability (class size) | Entropy |
|---|---|---|---|---|
| 1 class | 18,140.65 | – | – | – |
| 2 classes | 17,518.68 | 650.62 (0.001) | 0.95 (59%) / 0.93(41%) | 0.79 |
| 3 classes | 17,356.07 | 201.57 (0.001) | 0.89(46%) / 0.91(38%) / 0.90(16%) | 0.79 |
| 4 classes | 17,355.18 | 43.48 (0.23) | 0.92(38%) / 0.69(12%) / 0.85(36%) / 0.93(14%) | 0.76 |
| 5 classes | 17,330.23 | 66.99 (0.43) | 0.91(33%) / 0.90(9%) / 0.80(16%) / 0.85(11%) / 0.84(31%) | 0.79 |
BIC Bayesian Information Criteria, LMR LRT Lo-Mendell-Rubin adjusted likelihood ratio test. Bold type indicates the selected best fit model