Table 1. Simulation results under sparse cases with p = 100 and n = 100 based on 50 replications.

		, q = 2				, q = 4
		CC	CS	Ridge	Lasso	CC	CS	Ridge	Lasso
Scenario1, s = 4	LR-test	−5.89	−5.88	−4.99	−10.59	−4.71	−4.55	−4.75	−8.76
	Cox-test	−8.41	−8.26	−7.32	−13.80	−6.76	−7.06	−6.95	−11.73
	Devi	66.63	45.62	−29.48	−76.92	75.34	56.30	−25.75	−60.50
	c-index	0.772	0.768	0.752	0.859	0.750	0.751	0.750	0.825
		/	0.25	74.54	7.06	/	0.28	68.81	6.59
Scenario2	LR-test	−8.88	−9.35	−7.01	−12.39	−6.38	−6.74	−6.30	−11.40
	Cox-test	−12.16	−12.35	−9.64	−14.51	−9.27	−9.94	−8.77	−14.21
	Devi	−17.25	−26.02	−43.04	−95.39	−4.63	−11.32	−36.79	−84.14
	c-index	0.828	0.833	0.790	0.879	0.785	0.790	0.770	0.864
		/	0.30	37.88	6.90	/	0.30	50.91	6.17
		, q = 5				, q = 10
		CC	CS	Ridge	Lasso	CC	CS	Ridge	Lasso
Scenario1, s = 4	LR-test	−3.88	−4.31	−4.21	−6.64	−2.28	−2.45	−2.40	−1.90
	Cox-test	−6.18	−6.19	−6.04	−9.47	−3.03	−3.03	−3.01	−2.86
	Devi	80.59	56.87	−21.44	−43.22	145.95	97.88	−9.28	−7.85
	c-index	0.725	0.722	0.722	0.790	0.659	0.656	0.652	0.649
		/	0.28	79.85	6.89	/	0.275	101.77	8.44
Scenario2	LR-test	−13.71	−13.69	−11.38	−14.52	−9.67	−9.34	−8.86	−9.65
	Cox-test	−15.18	−15.22	−14.04	−15.48	−12.68	−12.65	−11.34	−12.24
	Devi	−23.91	−34.13	−77.63	−107.14	8.563	−0.559	−55.62	−67.93
	c-index	0.886	0.885	0.862	0.889	0.843	0.835	0.822	0.838
		/	0.33	33.34	6.66	/	0.29	47.22	6.86

NOTE: For Scenario 1, each informative covariate is correlated with s non-informative covariates. For Scenario 2, the covariates for the right panel have two gene pathways and those for the left panel have one gene pathway. In each setting, q is the number of informative covariates (covariates with non-zero coefficients).