Skip to main content
. Author manuscript; available in PMC: 2014 Jul 10.
Published in final edited form as: Addict Behav. 2013 Oct 17;39(1):308–315. doi: 10.1016/j.addbeh.2013.10.017

Table 2.

Summary of Model Fit for Unconditional Latent Profile Models (N = 583)

Model tested −2LL df AIC BIC SSA-BIC p-value for LMR LRT Entropy % of individuals in smallest class
1-class −6034.016 8 12084.032 12118.978 12093.581 - - -
2-classes −5494.685 13 11015.371 11072.157 11030.887 < .0001 .982 12.521
3-classes −5211.895 18 10459.790 10538.418 10481.274 .0007 .986 11.149
4-classes −5081.848 23 10209.696 10310.164 10237.148 .0212 .989 5.146
5-classes −4982.670 28 10021.341 10143.650 10054.760 .1976 .988 3.431
6-classes* −4876.198 33 9818.397 9962.547 9857.784 .1663 .984 1.372

Note. −2LL = −2 log likelihood; AIC = Akaike information criteria; BIC = Bayesian information criteria; SSA-BIC = sample size adjusted BIC; LMR LRT = Lo-Mendell-Rubin Likelihood Ratio Test.

*

standard errors for model parameters not trustworthy due to a non-positive definite first-order product matrix.