Table 3.

Simulation results for scenario 2 where a binary variable X₁ has missing values.

			N = 250				N = 500				N = 1000
Type	K	Method	Bias	SD	rMSE	Com	Bias	SD	rMSE	Com	Bias	SD	rMSE	Com
–	1	CD	0.013	0.348	0.349	0.0	0.004	0.239	0.239	0.0	0.004	0.166	0.166	0.0
		CC	0.402	0.491	0.635	0.0	0.408	0.347	0.535	0.0	0.407	0.239	0.472	0.0
U	5	iMI	0.077	0.444	0.451	0.0	0.046	0.324	0.327	0.0	0.023	0.227	0.229	0.0
		avgmMI	0.052	0.585	0.587	2.0	0.059	0.441	0.445	2.0	0.056	0.293	0.298	2.0
		cslMI	0.014	0.472	0.472	3.0	0.005	0.332	0.332	3.0	0.004	0.226	0.226	3.0
		siMI	0.014	0.468	0.468	10.7	0.005	0.331	0.331	10.2	0.003	0.232	0.232	9.7
	10	iMI	0.180	0.404	0.442	0.0	0.105	0.309	0.326	0.0	0.051	0.222	0.228	0.0
		avgmMI	0.115	0.624	0.635	2.0	0.128	0.502	0.518	2.0	0.119	0.333	0.354	2.0
		cslMI	0.014	0.470	0.470	3.0	0.005	0.331	0.331	3.0	0.003	0.230	0.231	3.0
		siMI	0.014	0.469	0.469	10.7	0.005	0.329	0.329	10.2	0.003	0.229	0.229	9.7
E	5	iMI	0.056	0.447	0.451	0.0	0.037	0.324	0.327	0.0	0.019	0.229	0.230	0.0
		avgmMI	0.108	0.524	0.535	2.0	0.047	0.348	0.351	2.0	0.022	0.233	0.234	2.0
		cslMI	0.016	0.533	0.533	3.0	0.005	0.344	0.344	3.0	0.004	0.233	0.233	3.0
		siMI	0.014	0.469	0.469	10.7	0.004	0.330	0.330	10.2	0.002	0.229	0.229	9.7
	10	iMI	0.169	0.413	0.446	0.0	0.081	0.319	0.329	0.0	0.041	0.226	0.230	0.0
		avgmMI	0.226	0.653	0.691	2.0	0.125	0.386	0.406	2.0	0.054	0.241	0.247	2.0
		cslMI	0.030	0.792	0.793	3.0	0.007	0.446	0.446	3.0	0.003	0.259	0.259	3.0
		siMI	0.015	0.467	0.467	10.7	0.004	0.330	0.330	10.2	0.003	0.228	0.228	9.7

Reported are Bias, average bias; SD, Monte Carlo standard deviation; rMSE, root mean squared error; Com, number of communications. N is the total number of samples. See Table 1 for local sample sizes. Results are based on 1000 Monte Carlo datasets.