Skip to main content
. 2025 Jun 26;25:2175. doi: 10.1186/s12889-025-23539-w

Table 2.

Comparison of GMMs with different numbers of latent trajectory classes

2 classes 3 classes 4 classes1 5 classes1
Entropy 0.874 0.782 0.802 0.821
LCPP 0.865 0.833 0.757 0.739
BIC 56592.488 56456.164 56377.519 56316.237
AIC 56495.632 56332.404 56226.855 56138.668
VLMR p-value < 0.001 0.009 0.245 0.674
Class sizes

219 (14%)

1,386 (86%)

192 (12%)

1,251 (78%)

162 (10%)

63 (4%)

159 (10%)

160 (10%)

1,223 (76%)

73 (5%)

150 (9%)

1,207 (75%)

144 (9%)

31 (2%)

Note. LCPP = lowest average latent class posterior probability. BIC = Bayesian Information Criterion. AIC = Akaike Information Criterion. VLMR = Vuong-Lo-Mendell-Rubin Likelihood Ratio Test. 1 For the GMMs with 4 and 5 latent trajectory classes, the best loglikelihood values could not be replicated