. 2016 Feb 12;6:21689. doi: 10.1038/srep21689

Table 2. Simulation results for estimating β ₁ = β ₂ = β ₃ = 1 in the presence of missing data based on 200 monte carlo data sets, where n = 100 and ρ = 0.5.


		Bias	SE	SD	MSE	CR	Bias	SE	SD	MSE	CR	Bias	SE	SD	MSE	CR
q = 4	GS	0.007	0.515	0.478	0.228	0.970	−0.014	0.515	0.533	0.283	0.940	−0.037	0.519	0.496	0.246	0.970
	CC	−0.274	0.569	0.560	0.387	0.890	−0.272	0.570	0.588	0.418	0.910	−0.265	0.573	0.529	0.349	0.935
	MI-true	−0.068	0.687	0.687	0.474	0.925	0.010	0.698	0.717	0.512	0.915	−0.062	0.693	0.685	0.470	0.925
	p = 200
	MICE-RF	−0.628	0.747	0.352	0.518	0.960	−0.355	0.711	0.371	0.263	0.980	−0.510	0.725	0.349	0.382	0.975
	KNN-V	−0.930	0.708	0.545	1.160	0.800	−0.242	0.711	0.535	0.343	0.980	−0.655	0.720	0.557	0.738	0.925
	KNN-S	−0.274	0.569	0.560	0.387	0.890	−0.272	0.570	0.588	0.418	0.910	−0.265	0.573	0.529	0.349	0.935
	MICE-DURR(Lasso)	−0.853	0.548	0.407	0.892	0.720	−0.475	0.544	0.384	0.373	0.925	−0.688	0.546	0.418	0.647	0.860
	MICE-DURR(EN)	−0.852	0.548	0.412	0.894	0.695	−0.474	0.541	0.386	0.372	0.935	−0.688	0.550	0.411	0.641	0.845
	MICE-DURR(Alasso)	−0.844	0.566	0.349	0.834	0.785	−0.486	0.563	0.351	0.359	0.945	−0.701	0.570	0.374	0.631	0.870
	MICE-IURR(Lasso)	−0.094	0.668	0.690	0.483	0.930	−0.080	0.669	0.729	0.536	0.885	−0.068	0.661	0.703	0.496	0.925
	MICE-IURR(EN)	−0.092	0.653	0.701	0.498	0.935	−0.074	0.667	0.724	0.527	0.910	−0.069	0.647	0.705	0.499	0.915
	MICE-IURR(Alasso)	−0.091	0.720	0.744	0.559	0.940	−0.076	0.710	0.790	0.626	0.880	−0.123	0.719	0.761	0.592	0.920
	p = 1000
	MICE-DURR(Lasso)	−0.809	0.560	0.392	0.807	0.715	−0.534	0.558	0.445	0.482	0.905	−0.713	0.568	0.437	0.698	0.825
	MICE-DURR(EN)	−0.806	0.560	0.398	0.808	0.710	−0.530	0.560	0.458	0.490	0.920	−0.711	0.568	0.433	0.693	0.815
	MICE-DURR(Alasso)	−0.803	0.567	0.354	0.770	0.755	−0.544	0.565	0.409	0.463	0.905	−0.730	0.577	0.379	0.676	0.840
	MICE-IURR(Lasso)	0.058	0.677	0.836	0.698	0.860	−0.040	0.676	0.811	0.656	0.890	−0.197	0.677	0.752	0.601	0.905
	MICE-IURR(EN)	0.073	0.664	0.808	0.655	0.850	−0.041	0.667	0.800	0.638	0.890	−0.192	0.670	0.739	0.581	0.885
	MICE-IURR(Alasso)	−0.036	0.793	0.912	0.829	0.865	−0.209	0.783	0.852	0.765	0.885	−0.335	0.779	0.792	0.737	0.910
q = 20	GS	0.014	0.521	0.500	0.249	0.950	−0.008	0.523	0.550	0.301	0.920	−0.023	0.527	0.506	0.255	0.970
	CC	−0.352	0.525	0.536	0.410	0.855	−0.357	0.529	0.541	0.418	0.875	−0.387	0.533	0.520	0.419	0.890
	MI-true	−0.077	0.689	0.593	0.356	0.965	−0.047	0.688	0.675	0.456	0.920	−0.092	0.692	0.628	0.401	0.965
	p = 200
	MICE-RF	−0.390	0.717	0.427	0.333	0.990	−0.286	0.708	0.466	0.298	0.985	−0.412	0.721	0.382	0.315	0.995
	KNN-V	−0.450	0.685	0.530	0.482	0.950	−0.214	0.697	0.561	0.359	0.970	−0.441	0.703	0.469	0.413	0.980
	KNN-S	−0.352	0.525	0.536	0.410	0.855	−0.357	0.529	0.541	0.418	0.875	−0.387	0.533	0.520	0.419	0.890
	MICE-DURR(Lasso)	−0.491	0.631	0.451	0.444	0.955	−0.329	0.633	0.486	0.343	0.970	−0.532	0.644	0.421	0.459	0.975
	MICE-DURR(EN)	−0.489	0.631	0.455	0.445	0.920	−0.340	0.627	0.486	0.351	0.970	−0.518	0.641	0.401	0.428	0.970
	MICE-DURR(Alasso)	−0.481	0.655	0.425	0.412	0.940	−0.348	0.653	0.452	0.325	0.975	−0.529	0.668	0.377	0.422	0.960
	MICE-IURR(Lasso)	−0.022	0.678	0.729	0.530	0.905	−0.024	0.684	0.752	0.563	0.870	−0.057	0.683	0.689	0.475	0.905
	MICE-IURR(EN)	−0.023	0.666	0.695	0.481	0.920	−0.022	0.670	0.730	0.531	0.855	−0.036	0.675	0.674	0.453	0.925
	MICE-IURR(Alasso)	−0.144	0.731	0.721	0.539	0.910	−0.103	0.719	0.764	0.591	0.895	−0.181	0.720	0.718	0.546	0.955
	p = 1000
	MICE-DURR(Lasso)	−0.436	0.640	0.415	0.361	0.978	−0.399	0.628	0.490	0.396	0.956	−0.458	0.640	0.393	0.363	0.989
	MICE-DURR(EN)	−0.459	0.631	0.437	0.399	0.978	−0.375	0.626	0.492	0.380	0.967	−0.460	0.636	0.407	0.375	1.000
	MICE-DURR(Alasso)	−0.450	0.651	0.419	0.376	0.978	−0.387	0.642	0.482	0.380	0.978	−0.487	0.655	0.384	0.382	0.989
	MICE-IURR(Lasso)	0.012	0.682	0.750	0.556	0.900	0.004	0.660	0.712	0.502	0.956	−0.079	0.687	0.630	0.398	0.933
	MICE-IURR(EN)	0.020	0.672	0.738	0.539	0.878	0.018	0.662	0.725	0.521	0.911	−0.086	0.653	0.660	0.438	0.933
	MICE-IURR(Alasso)	−0.204	0.766	0.767	0.624	0.878	−0.135	0.745	0.778	0.617	0.889	−0.318	0.737	0.666	0.540	0.911

Bias, mean bias; SE, mean standard error; SD, Monte Carlo standard deviation; MSE, mean square error; CR, coverage rate of 95% confidence interval; GS, gold standard; CC, complete-case; KNN-V, KNN by nearest variables; KNN-S, KNN by nearest subjects; MICE-DURR, MICE through direct use of regularized regressions; MICE-IURR, MICE through indirect use of regularized regressions; EN, elastic net; Alasso, adaptive lasso.

Table 2. Simulation results for estimating β 1 = β 2 = β 3 = 1 in the presence of missing data based on 200 monte carlo data sets, where n = 100 and ρ = 0.5.

Table 2. Simulation results for estimating β ₁ = β ₂ = β ₃ = 1 in the presence of missing data based on 200 monte carlo data sets, where n = 100 and ρ = 0.5.