Table 1.
Number of Classes | df | AIC | BIC | adj-BIC | Entropy | Bootstrapped LRT, p-value | Significant Change in Fit? |
---|---|---|---|---|---|---|---|
One | 22 | 15344.67 | 15427.99 | 15358.20 | N/A | N/A | N/A |
Two | 34 | 14766.9 | 14895.66 | 14787.81 | 0.9 | 601.77, p < 0.001 | Yes |
Three | 46 | 14382.06 | 14556.25 | 14410.35 | 0.94 | 408.84, p < 0.001 | Yes |
Four | 58 | 14035.45 | 14255.09 | 14071.11 | 0.96 | 370.61, p < 0.001 | Yes |
Five | Could not be specified | ||||||
Six | Could not be specified |
Note: N=326. A model was deemed to have better fit if there were significant decreases in AIC, BIC, and adj-BIC values, a significant LRT, and increases in entropy values in comparison with the prior class solution. Df = Degrees of freedom; AIC = Akaike information criteria; BIC = Bayesian information criteria, adj-BIC = sample-size-adjusted Bayesian information criteria; LRT = parametric bootstrapped likelihood test.