A restricted latent class model with polytomous attributes and respondent-level covariates

Table 3 Results of second simulation study

\(N\)	\(J\)	\(K\)	\(L\)	\(\rho\)	\(\gamma\)	\(\eta\)	\(R\)	\(\lambda\)	\(\alpha _n\)
3000	45	4	3	0.000	0.044	0.015	0.030	0.053	0.917
3000	70	5	2	0.000		0.009	0.026	0.051	0.976
3000	70	5	3	0.000	0.027	0.013	0.024	0.047	0.957
3000	45	4	3	0.250	0.058	0.024	0.053	0.080	0.846
3000	70	5	2	0.250		0.007	0.022	0.046	0.985
3000	70	5	3	0.250	0.056	0.024	0.053	0.064	0.843
3000	45	4	3	0.500	0.115	0.061	0.163	0.127	0.655
3000	70	5	2	0.500		0.017	0.067	0.076	0.914
3000	70	5	3	0.500	0.081	0.035	0.100	0.084	0.791
5000	45	4	3	0.000	0.040	0.014	0.026	0.050	0.907
5000	70	5	2	0.000		0.008	0.022	0.049	0.972
5000	70	5	3	0.000	0.025	0.011	0.020	0.044	0.946
5000	45	4	3	0.250	0.050	0.022	0.049	0.073	0.857
5000	70	5	2	0.250		0.005	0.015	0.041	0.993
5000	70	5	3	0.250	0.062	0.023	0.053	0.059	0.819
5000	45	4	3	0.500	0.121	0.064	0.167	0.126	0.620
5000	70	5	2	0.500		0.018	0.063	0.088	0.906
5000	70	5	3	0.500	0.084	0.036	0.099	0.087	0.756

Values displayed for all columns except \(\alpha _n\) are the average, taken over all elements of the parameter, of the mean absolute error of estimation of each element over all replications. The values displayed for \(\alpha _n\) are the percentage of draws for which the vector \(\alpha _n\) was drawn correctly