Table A1.
Goodness-of-fit statistics for multilevel latent class models.
| LL | BIC | BIC-G | Parameters | Class.Err. | ||
|---|---|---|---|---|---|---|
| M1 | 1 Class | −10,188.17 | 20,431.07 | 7 | 0.00 | |
| M2 | 2 Class | −9520.81 | 19,158.91 | 15 | 0.10 | |
| M3 | 3 Class | −9359.63 | 18,899.10 | 23 | 0.12 | |
| M4 | 4 Class | −9263.64 | 18,769.66 | 31 | 0.14 | |
| M5 | 5 Class | −9226.75 | 18,758.43 | 39 | 0.22 | |
| M6 | 6 Class | −9213.63 | 18,794.75 | 47 | 0.25 | |
| M7 | 5 Class + 1 Gclass + r | −8247.22 | 17,112.10 | 16,995.61 | 79 | 0.14 |
| M8 | 5 Class + 2 Gclass + r | −8221.21 | 17,099.16 | 16,975.30 | 84 | 0.14 |
| M9 | 5 Class + 3 Gclass + r | −8215.97 | 17,127.79 | 16,996.55 | 89 | 0.14 |
Note: BIC = Bayesian information criterion; BIC-G = group-based Bayesian information criterion; Class.Err. = mean of the proportion of classification error for the latent class; Gclass = respondent level cluster; r = random parameter.