Skip to main content
. Author manuscript; available in PMC: 2014 May 28.
Published in final edited form as: Fathering. 2012 Jul 26;10(2):213–235. doi: 10.3149/fth.1002.213

Table 4.

Model Fit Estimates for Different Latent Class Analysis Models

1 Profile 2 Profile 3 Profile 4 Profile
BICa 1991.23 2013.83 2042.65 2070.61
ABICb 1975.37 1978.94 1988.72 1997.65
G2 29.15 16.84 10.76 3.82
Lo-Mendell-Rubind 10.96 5.91 6.75
 df - 6 6 6
 p-value - 0.55 0.04 0.44
Entropy - 0.34 0.88 0.79

Note.

a

BIC means Bayesian information criterion.

b

ABIC means adjusted Bayesian information criterion, adjusted for sample size.

c

G2 refers to the likelihood ratio chi-square estimate.

d

Lo-Mendell-Rubin Adjusted LRT test is a nested model test, using bootstrapped methodology to replicate a normal distribution, comparing an adjusted LRT model solution with k number of profiles against a model solution with k-1 profiles. A significant p-value indicates the k profile solution improves fit as compared to the k-1 solution.