Table 1.

Fit Criteria for Discrete and Continuous Liability Models in the National Epidemiologic Survey on Alcohol and Related Conditions

	Criterion
Model	k*	ln(L)	BIC
Entire sample (N = 43 093)
Latent class models
2 classes	13	−46306.66	46376.02
3 classes	20	−46023.09	46129.80
4 classes	27	−46006.01	46150.07
5 classes	34	−45996.14	46177.55
6 classes	41	−45983.80	46202.56
Latent trait models
2 values	13	−46306.66	46376.02
3 values	14	−46097.12	46171.82
4 values	15	−46026.45	46106.48
5 values	16	−46021.74	46107.11
6 values	17	−46011.51	46102.21
Normal	12	−46032.62	46096.65
Females (n = 24 575)
Latent class models
2 classes	13	−21302.73	21368.44
3 classes	20	−21174.89	21275.99
4 classes	27	−21163.74	21300.22
5 classes	34	−21155.15	21327.02
6 classes	41	−21151.86	21359.10
Latent trait models
2 values	13	−21302.73	21368.44
3 values	14	−21210.25	21281.01
4 values	15	−21177.75	21253.57
5 values	16	−21176.42	21257.29
6 values	17	−21175.52	21261.45
Normal	12	−21185.26	21245.92
Males (n = 18 518)
Latent class models
2 classes	13	−23558.19	23622.06
3 classes	20	−23405.57	23503.83
4 classes	27	−23387.54	23520.20
5 classes	34	−23382.78	23549.83
6 classes	41	−23373.16	23574.60
Latent trait models
2 values	13	−23558.19	23622.06
3 values	14	−23439.44	23508.23
4 values	15	−23408.12	23481.82
5 values	16	−23395.67	23474.28
6 values	17	−23395.80	23479.33
Normal	12	−23407.60	23466.56

Abbreviations: BIC, Bayesian information criterion; ln(L), log likelihood.

^*k indicates number of parameters.