Table 3.

MSE, bias and 90% coverage probabilities for estimating conditional regression coefficients from the simulation study

Design	Model	Results for the following parameters:
		β1 = 0	β2 = 0.5	β3 = 0	β4 = 0.5
(a) MSE (× 100)
1, PS—weak	PS	2.00	2.50	0.78	1.32
	Gaussian	1.09	3.55	0.47	2.97
	Independence	1.88	2.83	0.76	1.77
2, PS—strong	PS	11.92	12.19	0.87	1.69
	Gaussian	1.18	15.76	0.35	12.52
	Independence	2.21	11.40	0.56	11.29
3, Gaussian	PS	1.89	38.10	1.29	43.93
	Gaussian	0.59	0.64	0.41	0.47
	Independence	0.98	1.80	0.72	1.81
4, PS—higher censoring	PS	2.63	3.73	1.10	2.50
	Gaussian	0.96	3.52	0.66	2.31
	Independence	1.70	3.08	1.05	1.59
(b) Bias (× 100)
1, PS—weak	PS	1.48	5.77	0.08	4.49
	Gaussian	1.33	−15.89	0.40	−15.86
	Independence	1.87	−9.89	0.91	−9.32
2, PS—strong	PS	−1.26	13.15	0.19	7.59
	Gaussian	−0.71	−32.9	0.56	−34.69
	Independence	−0.06	−30.17	1.11	−32.36
3, Gaussian	PS	1.09	28.77	−0.15	29.63
	Gaussian	0.53	−1.18	−0.11	−1.49
	Independence	0.01	8.07	−0.44	8.39
4, PS—higher censoring	PS	0.46	8.59	0.28	9.63
	Gaussian	0.02	−13.95	0.91	12.28
	Independence	0.28	−8.12	0.62	−6.35
(c) Coverage probability of 90% intervals (× 100)
1, PS—weak	PS	88	90	90	87
	Gaussian	69	33	89	19
	Independence	62	51	79	61
2, PS—strong	PS	90	86	92	85
	Gaussian	68	4	92	0
	Independence	58	11	91	1
3, Gaussian	PS	87	63	88	60
	Gaussian	85	85	91	86
	Independence	74	62	81	64
4, PS—higher censoring	PS	91	90	91	81
	Gaussian	84	47	81	49
	Independence	67	60	74	70