Table 4.

Parameter estimates for the full Add Health sample (N = 6, 504) by different computation strategies.

Computation Strategy		Latent Classes
		Class 1		Class 2		Class 3		Class 4
gibbs() sampling	${\widehat{\mathit{\pi }}}_j$	0.481*	(0.016)†	0.088	(0.010)	0.139	(0.014)	0.293	(0.019)
	${\widehat{p}}_{jk}$
	Lied to parents	0.272	(0.013)	0.919	(0.020)	0.728	(0.021)	0.728	(0.019)
	Publicly loud/rowdy/unruly	0.170	(0.015)	0.968	(0.019)	0.561	(0.032)	0.775	(0.021)
	Damaged property	0.006	(0.003)	0.818	(0.046)	0.258	(0.028)	0.256	(0.016)
	Shoplifting	0.021	(0.005)	0.875	(0.021)	0.949	(0.035)	0.050	(0.024)
	Stolen something worth < $50	0.008	(0.003)	0.862	(0.024)	0.706	(0.043)	0.047	(0.014)
	Taken part in group fight	0.041	(0.006)	0.627	(0.037)	0.201	(0.025)	0.325	(0.017)
SAS PROC LCA	${\widehat{\mathit{\pi }}}_j$	0.479	(0.016)	0.086	(0.011)	0.140	(0.016)	0.295	(0.019)
	Lied to parents	0.270	(0.012)	0.922	(0.020)	0.731	(0.021)	0.726	(0.019)
	Publicly loud/rowdy/unruly	0.168	(0.014)	0.974	(0.023)	0.561	(0.033)	0.774	(0.022)
	Damaged property	0.005	(0.020)	0.822	(0.048)	0.260	(0.030)	0.253	(0.015)
	Shoplifting	0.020	(0.005)	0.879	(0.021)	0.964	(0.038)	0.043	(0.028)
	Stolen something worth < $50	0.008	(0.002)	0.862	(0.025)	0.695	(0.044)	0.052	(0.014)
	Taken part in group fight	0.040	(0.006)	0.631	(0.038)	0.202	(0.027)	0.324	(0.016)
poLCA() in R	${\widehat{\mathit{\pi }}}_j$	0.297	(0.019)	0.084	(0.010)	0.140	(0.015)	0.479	(0.016)
	Lied to parents	0.726	(0.018)	0.924	(0.020)	0.734	(0.020)	0.270	(0.012)
	Publicly loud/rowdy/unruly	0.773	(0.021)	0.978	(0.023)	0.565	(0.030)	0.167	(0.014)
	Damaged property	0.254	(0.015)	0.829	(0.045)	0.263	(0.028)	0.005	(0.004)
	Shoplifting	0.043	(0.028)	0.880	(0.021)	0.972	(0.042)	0.020	(0.005)
	Stolen something worth < $50	0.055	(0.015)	0.864	(0.025)	0.697	(0.045)	0.008	(0.002)
	Taken part in group fight	0.324	(0.016)	0.635	(0.036)	0.205	(0.025)	0.040	(0.006)

^*: Original parameter estimates without reordering the labels.

^†: posterior standard deviation.