Skip to main content
. Author manuscript; available in PMC: 2022 Jan 27.
Published in final edited form as: Adv Child Dev Behav. 2016 Dec 28;52:105–152. doi: 10.1016/bs.acdb.2016.10.003

Table 1.

Model Fit Statistics for Latent Class Analysis of Neighborhood Types

L 2 % Reduction in L2 BIC (L2) AIC (L2) Entropy R2
2-Class 31,236.26 12.89 13,274.44 26,612.26 0.87
3-Class 28,237.79   9.60 10,384.73 23,641.79 0.89
4-Class 26,487.35   6.20    8743.05 21,919.35 0.91
5-Class 24,978.07   5.70    7342.54 20,438.07 0.93
6-Class 23,977.94   4.00    6451.17 19,465.94 0.92
7-Class 23,048.01   3.88    5630.01 18,564.01 0.92
8-Class 22,288.84   3.29    4979.61 17,832.84 0.92
9-Class 21,707.28   2.61    4506.81 17,279.28 0.92
10-Classa 21,208.99   2.30    4117.23 16,808.99 0.93
11-Class 20,747.10   2.18    3764.17 16,375.10 0.93
12-Class 20,314.18   2.09    3440.01 15,970.18 0.93
13-Class 20,033.83   1.38    3268.43 15,717.83 0.94
14-Class 19,669.99   1.82    3013.35 15,381.99 0.93
15-Class 19,392.19   1.41    2844.32 15,132.19 0.93
a

A −2LL bootstrapping test was also used to assess the fit of the 10-class solution; compared to a 9-class solution, the test statistic (−2LL difference = 498.29, p = 0.0000) indicates the 10-class solution provides a better fit than a solution with fewer classes.

Notes: L2 is the likelihood-ratio Chi-square statistic, which indicates the amount of the association among the variables that remains unexplained after estimating the model; the lower the value, the better the fit of the model to the data.

Source: National Longitudinal Study of Adolescent to Adult Health, Wave I (n = 2366 census tracts).