Table 3.

Bias and Power for Normal Data MAR

			Null		Alternative
Methods	Estimate	Missing	Type I error (1 sided) (95% C.I.)	Type I error (2 sided) (95% C.I.)	Power (1 sided) (95% C.I.)	Power (2 sided) (95% C.I.)
Trapezoidal		Logistic	0.000 (0.000, 0.004)	0.872 (0.850, 0.892)
ICCPW	Truth	Logistic	0.048 (0.036, 0.063)	0.053 (0.040, 0.069)	0.917 (0.898, 0.933)	0.865 (0.842, 0.886)
ICCPW	Truth	U-shape	0.045 (0.033, 0.060)	0.045 (0.033, 0.059)	0.898 (0.878, 0.916)	0.848 (0.824, 0.870)
ICCPW	Logistic	U-shape	0.045 (0.033, 0.060)	0.040 (0.029, 0.054)	0.940 (0.923, 0.954)	0.893 (0.872, 0.911)
ICCPW	Truth	Threshold	0.025 (0.016, 0.037)	0.037 (0.026, 0.050)	0.762 (0.734, 0.788)	0.620 (0.589, 0.650)
ICCPW	Logistic	Threshold	0.006 (0.002, 0.013)	0.009 (0.004, 0.017)	0.320 (0.291, 0.350)	0.169 (0.146, 0.194)