Table 1 : Empirical performance of estimators of the regression coefficient for the continuous covariate under parametric and
semi-parametric Cox regression models with and without adminstrative censoring; sample size m=500, number of simulations nsim=500.


       True Concentration()  Sample Count ()  Observed Count (Ci)  EM Algorithm

exp(β1) μ φ EBIAS ESE  ECP% EBIAS ESE  ECP% EBIAS ESE  ECP% EBIAS ESE  ECP%

Exponential Regression, No Censoring

1.1 2/3 1  -0.024  0.610  95.6  1.122  0.639  50.0  -0.689  0.832  80.8  0.161  0.780  95.4
  2 1  0.008  0.588  95.2  2.278  0.594  2.0  0.975  0.748  63.8  -0.030  0.809  94.6
  2 4  -0.037  0.295  95.6  0.922  0.359  18.0  -0.962  0.462  25.6  0.032  0.389  96.4
1.25 2/3 1  -0.057  0.610  96.0  3.500  0.968  0.8  -0.331  1.248  73.6  0.429  1.066  93.0
  2 1  -0.013  0.609  93.8  6.011  0.865  0  3.213  1.114  4.4  -0.090  1.028  95.0
  2 4  -0.028  0.303  95.4  3.073  0.689  0  -0.951  0.929  31.2  0.026  0.592  94.0

Exponential Regression, 25% Administrative Censoring

1.1 2/3 1  -0.032  0.642  95.4  1.111  0.665  52.8  -0.706  0.858  80.0  0.166  0.800  95.8
  2 1  -0.022  0.670  93.6  2.340  0.669  4.2  1.064  0.832  64.6  0.009  0.956  95.2
  2 4  -0.050  0.319  94.8  0.957  0.382  20.2  -0.908  0.492  38.4  0.045  0.422  95.6
1.25 2/3 1  -0.074  0.633  96.0  3.551  1.017  1.2  -0.256  1.303  76.0  0.469  1.132  92.8
  2 1  -0.018  0.751  94.0  6.793  0.965  0  4.301  1.193  1.4  -0.130  1.250  94.4
  2 4  -0.039  0.331  94.4  3.265  0.752  0  -0.659  1.020  44.2  0.032  0.659  93.6

Cox Regression, No Censoring

1.1 2/3 1  0.016  0.694  96.8  -1.306  0.689  46.6  0.425  0.877  90.2  0.117  0.877  96.0
  2 1  0.033  0.674  95.2  -2.668  0.631  1.2  -1.511  0.774  42.4  0.004  1.026  95.4
  2 4  0.069  0.461  95.2  -1.451  0.473  10.0  0.198  0.586  87.8  -0.021  0.594  96.0
1.25 2/3 1  0.040  1.078  94.8  -5.388  1.126  0  -2.248  1.395  44.0  0.129  1.698  95.8
  2 1  0.026  1.027  93.6  -9.185  0.872  0  -7.310  1.015  0  -0.426  1.969  93.8
  2 4  0.094  0.929  94.8  -7.334  0.798  0  -4.778  0.983  0.4  -0.584  1.463  92.4

Cox Regression, 25% Administrative Censoring

1.1 2/3 1  0.018  0.703  96.6  -1.253  0.7  50.6  0.503  0.887  89.0  0.121  0.883  96.6
  2 1  0.035  0.714  95.2  -2.527  0.677  3.0  -1.317  0.832  55.6  0.033  1.070  95.0
  2 4  0.074  0.464  95.2  -1.399  0.475  12.8  0.271  0.588  87.6  -0.016  0.596  96.8
1.25 2/3 1  0.035  1.096  94.0  -5.288  1.148  0.2  -2.104  1.421  48.6  0.124  1.700  94.8
  2 1  0.004  1.071  94.6  -8.907  0.915  0  -6.957  1.066  0  -0.376  1.988  93.8
  2 4  0.097  0.943  94.2  -7.246  0.829  0  -4.661  1.032  0.4  -0.562  1.496  92.8

† EBIAS is the empirical bias ×100, ESE is the empirical standard error ×100.

Cotton et al.Journal of Medical Statistics and Informatics  2014 2:1DOI : 10.7243/2053-7662-2-1