Clustering of multiple lifestyle behaviors among migrant, left-behind and local adolescents in China: a cross-sectional study

Table 2 Index of latent classes of multiple lifestyle behaviors: Survey on child and adolescent health development in China (Year 2014–2015) (N = 1364)

Class	k	G²/LL	df	χ²	AIC	BIC	aBIC	Entropy	LMR	BLRT	Probabilities
1	10	898.58	1025	2283.87	14,279.64	14,331.84	14,300.08				1.000
2	21	659.88	976	1187.43	13,863.59	13,973.22	13,906.51	0.616	< 0.0001	< 0.0001	0.644/0.356
3 ^a.	32	595.86	969	1125.13	13,786.73	13,953.78	13,852.13	0.688	< 0.01	< 0.0001	0.078/0.301/0.621
4	43	530.44	958	1086.72	13,742.33	13,966.81	13,830.21	0.588	0.1079	< 0.0001	0.072/0.223/0.500/0.206
5	54	484.87	947	926.48	13,717.77	13,999.67	13,828.13	0.635	0.3604	< 0.0001	0.153/0.096/0.112/0.609/0.030

Note.^a. Bold indicates the final choice for the number of clusters. AIC Akaike information criterion, BIC Bayesian information criterion, aBIC adjusted Bayesian information criterion. Entropy indicates classification quality. LMR = P value of likelihood ratio test. BLRT = P value of bootstrap likelihood ratio test

ISSN: 1471-2458