. Author manuscript; available in PMC: 2024 Jul 6.

Published in final edited form as: Dev Psychopathol. 2017 Mar 14;29(4):1333–1351. doi: 10.1017/S095457941700030X

Table 2.

Comparison of fit among latent class growth analysis solutions for externalizing behavior

Fit Statistic	One Class	Two Class	Three Class	Four Class	Five Class	Six Class^a

1. BIC	122.62	−10.40	−52.12	−67.59	−72.53	−59.13
Adj. BIC	113.11	−29.42	−80.65	−105.63	−120.07	−116.18
2. Entropy	1.0	.88	.80	.81	.83	.85
3. In classes	100%	11%, 89%	4%, 26%, 71%	4%, 10%, 16%, 71%	1%, 3%, 11%, 17%, 68%	1%, 1%, 3%, 11%, 17%, 67%
4. Average probability of trajectory class membership	100%	89%, 98%	84%, 92%, 100%	79%, 83%, 92%, 99%	77%, 82%, 87%, 93%, 100%	77%, 82%, 91%, 91%, 93%, 96%
5. LMR-LRT	NA	140.86, p = .163	54.80, p = .064	30.06, p = .013	26.32, p = .094	2.84, p = .284
6. BLRT	NA	149.44 (3), p < .001	58.14 (3), p < .001	31.89 (3), p < .001	27.93 (3), p < .001	3.02 (3), p = .227

Note: BIC, Bayesian information criterion; Adj. BIC, sample-size adjusted BIC; LMR-LRT, Lo–Mendell–Rubin adjusted likelihood ratio test; BLRT, bootstrap likelihood ratio test (difference in number of parameters).

The best log likelihood was not replicated, so the six-class solution may not be trustworthy.