Skip to main content

Table 7 Simulation results for scenario G with the correct logistic model (25%missing)

From: Comparison of methods for handling covariate missingness in propensity score estimation with a binary exposure

Missingness MechanismMethodn = 500n = 1000n = 5000
Bias (SD)SERMSEBias (SD)SERMSEBias (SD)SERMSE
 complete−0.002 (0.045)0.0700.0450.000 (0.030)0.0490.0300.001 (0.014)0.0220.014
MCARSI + pe + pu−0.002 (0.053)0.0700.053−0.001 (0.035)0.0490.0350.000 (0.016)0.0220.016
SI + pe−0.003 (0.052)0.0700.0520.000 (0.035)0.0490.0350.000 (0.016)0.0220.016
TMI−0.013 (0.062)0.0760.063−0.009 (0.041)0.0530.042−0.008 (0.018)0.0230.020
MI−0.002 (0.048)0.0710.048−0.001 (0.033)0.0490.0330.000 (0.015)0.0220.014
MIMP−0.002 (0.048)0.0710.049−0.001 (0.033)0.0500.0330.000 (0.015)0.0220.014
MAR1SI + pe + pu0.009 (0.052)0.0700.0530.009 (0.035)0.0490.0360.010 (0.016)0.0220.019
SI + pe0.008 (0.051)0.0700.0520.010 (0.034)0.0490.0350.011 (0.016)0.0220.019
TMI0.019 (0.057)0.0730.0600.021 (0.039)0.0510.0440.023 (0.018)0.0230.029
MI0.008 (0.047)0.0700.0480.009 (0.031)0.0490.0330.010 (0.015)0.0220.017
MIMP0.008 (0.048)0.0710.0480.009 (0.032)0.0490.0340.009 (0.015)0.0220.017
MAR2SI + pe + pu−0.001 (0.052)0.0700.0520.003 (0.036)0.0490.0360.002 (0.016)0.0220.016
SI + pe0.000 (0.052)0.0700.0520.003 (0.035)0.0490.0350.002 (0.016)0.0220.016
TMI−0.019 (0.064)0.0770.067−0.015 (0.043)0.0540.046−0.015 (0.019)0.0240.024
MI0.000 (0.047)0.0710.0480.003 (0.032)0.0490.0330.002 (0.015)0.0220.014
MIMP0.001 (0.047)0.0710.0460.004 (0.033)0.0500.0330.005 (0.015)0.0220.015
MAR sinisterSI + pe + pu0.003 (0.050)0.0700.0500.002 (0.035)0.0490.0350.000 (0.016)0.0220.016
SI + pe−0.003 (0.052)0.0700.0520.002 (0.035)0.0490.0350.000 (0.016)0.0220.016
TMI0.002 (0.061)0.0760.0610.003 (0.042)0.0520.042−0.004 (0.018)0.0230.018
MI0.001 (0.047)0.0710.0480.002 (0.033)0.0490.0320.000 (0.015)0.0220.015
MIMP0.001 (0.047)0.0710.0470.002 (0.033)0.0490.0330.000 (0.015)0.0220.014
  1. Note. Complete: logistic regression with complete data before introducing missingness; SI + pe + pu single imputation + prediction error + parameter uncertainty; SI + pe single imputation + prediction error; TMI treatment mean imputation; MI multiple imputation (m = 20); MIMP multiple imputation missingness pattern (m = 20); SD standard deviation; SE standard error; RMSE root mean squared error