Table 3.

Average clustering results over 100 data sets of size $N=100$ and $N=1000$, simulated from a latent class model with two classes, obtained through sparse latent class models (SFM) with $K=10$ and $K=20$ and DPM for three different priors on the precision parameters $e_0$ and $\alpha$ as well as using EM estimation as implemented in the R package poLCA (Linzer et al. 2011)

Prior	Method		$N=100$				$N=1000$
			$\text{E}(p.p.|\mathbf{y})$	${\widehat{K}}_{+}$	ari	err	$\text{E}(p.p.|\mathbf{y})$	${\widehat{K}}_{+}$	ari	err
$\alpha \sim \mathcal{G}(1,20)$	SFM	$K=10$	0.009	1.94	0.44	0.18	0.010	2.05	0.54	0.13
		$K=20$	0.005	1.92	0.43	0.18	0.005	2.02	0.54	0.13
	DPM		0.092	1.99	0.44	0.18	0.110	2.29	0.53	0.14
$\alpha \sim \mathcal{G}(1,2)$	SFM	$K=10$	0.064	2.29	0.46	0.17	0.068	2.23	0.53	0.14
		$K=20$	0.035	2.38	0.45	0.17	0.032	2.24	0.53	0.14
	DPM		0.599	2.44	0.45	0.17	0.670	2.62	0.52	0.15
$\alpha \sim \mathcal{G}(2,1)$	SFM	$K=10$	0.189	3.56	0.45	0.19	0.163	2.97	0.52	0.15
		$K=20$	0.086	3.34	0.45	0.19	0.072	3.28	0.51	0.16
	DPM		1.517	3.50	0.44	0.19	1.360	3.72	0.49	0.17
	poLCA			1.37	0.18	0.35		2.00	0.54	0.13

The reported values are averages of the posterior expectation $\text{E}(p.p.|\mathbf{y})$ of the precision parameter $e_0$ (SFM) and $\alpha$ (DPM), the estimated number of clusters ${\widehat{K}}_{+}$, the adjusted Rand index (ari) and the error rate (err)