Skip to main content
. Author manuscript; available in PMC: 2010 Sep 21.
Published in final edited form as: Biometrics. 2010 Sep;66(3):845–854. doi: 10.1111/j.1541-0420.2009.01322.x

Table 3.

Ninety-five percent confidence coverage comparing complete-case analysis (CCA), single missing indicator (SMI), modeled missing indicator (MMI), and multiple imputation (MI) for 1:1 design, number of case–control sets = 400, exp(βX ) = 2, exp(βZ) = 1.42, pr(X = 1) = 0.5. Based on 1000 trials.

In exp(β̂X) In exp(β̂Z)
Missing
pr(M = 1)-Confounding type CCA SMI MMI CCA SMI MMI
50%-no MCAR 96 97 96 95 95 96
MAR(Z) 96 97 97 95 95 95
MAR(D) 95 96 96 95 96 96
NI(X) 97 96 96 97 95 97
NI(X,Z) 95 96 96 95 95 96
50%-strong MCAR 96 - 94 97 - 97
MAR(Z) 95 - 93 95 - 97
MAR(D) 96 - 95 95 - 95
NI(X) 94 - 93 95 - 94
NI(X,Z) 94 - 94 94 - 96
20%-no MCAR 95 97 97 96 95 96
MAR(Z) 96 97 97 96 95 96
MAR(D) 96 96 96 97 96 95
NI(X) 96 96 96 96 95 95
NI(X,Z) 95 96 96 95 95 95
20%-strong MCAR 93 - 93 95 - 95
MAR(Z) 94 - 94 95 - 95
MAR(D) 94 - 94 95 - 94
NI(X) 95 - 94 93 - 95
NI(X,Z) 95 - 94 94 - 95

MCAR, missing at random unconditionally; MAR(D), missing at random conditional on D; MAR(Z), missing at random conditional on Z; NI(X), missing not at random conditional on X; NI(X,Z), missing not at random conditional on X,Z.