Fit indices, entropy and average posterior probabilities across models with different numbers of subgroups
| No. of subgroups | BIC | LMR-LRT p value | BLRT p value | Entropy | Posterior probabilities |
|---|---|---|---|---|---|
| 2 | 29007.29 | <0.0001 | <0.0001 | 0.899 | Subgroup 1 (n=474): 0.98 Subgroup 2 (n=223): 0.96 |
| 3 | 28299.85 | 0.0003 | <0.0001 | 0.887 | Subgroup 1 (n=330): 0.97 Subgroup 2 (n=257): 0.92 Subgroup 3 (n=110): 0.96 |
| 4 | 28058.11 | 0.01 | <0.0001 | 0.868 | Subgroup 1 (n=280): 0.95 Subgroup 2 (n=236): 0.90 Subgroup 3 (n=136): 0.89 Subgroup 4 (n=45): 0.97 |
| 5 | 27969.60 | 0.12 | <0.0001 | 0.817 | Subgroup 1 (n=195): 0.91 Subgroup 2 (n=184): 0.83 Subgroup 3 (n=176): 0.87 Subgroup 4 (n=109): 0.88 Subgroup 5 (n=33): 0.97 |
The lowest value of BIC indicates the best fitting model. Significant p values (p<0.05) for the LMR-LRT and BLRT indicate that a model with a k number of latent trajectory subgroups has a better fit than a model with a k−1 number of subgroups. Entropy and posterior probability values close to 1 indicate good classification.
BIC, Bayesian information criterion; BLRT, bootstrap likelihood ratio test; LMR-LRT, Vuong–Lo–Mendell–Rubin likelihood ratio test.