Table 2 : Empirical performance of estimators of the regression coefficient for the binary 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 (Ei)  EM Algorithm

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

Exponential Regression, No Censoring

1.5 2/3 1  0.368  0.098  93.6  9.259  0.100  83.2  7.831  0.107  84.6  -0.479  0.118  93.8
  2 1  1.782  0.232  93.6  23.635  0.193  75.4  28.068  0.152  48.2  -9.314  0.387  94.0
  2 4  0.429  0.095  94.6  7.897  0.096  87.6  10.019  0.092  83.0  -0.369  0.111  95.6
2 2/3 1  -0.378  0.089  96.0  16.652  0.099  58.2  13.829  0.103  71.2  0.233  0.109  96.0
  2 1  3.526  0.210  95.4  35.715  0.207  51.4  43.804  0.150  14.2  -8.262  0.318  95.0
  2 4  0.059  0.094  97.0  13.361  0.095  73.4  15.706  0.092  62.2  -0.419  0.107  96.4

Exponential Regression, 25% Administrative Censoring

1.5 2/3 1  0.860  0.105  95.4  8.994  0.109  86.0  8.005  0.113  90.2  -0.744  0.129  95.8
  2 1  -1.927  0.301  94.8  24.461  0.216  80.2  29.121  0.173  56.8  -1.338  0.570  95.8
  2 4  -0.007  0.116  94.6  7.933  0.112  91.4  10.840  0.105  86.4  0.123  0.137  95.8
2 2/3 1  -0.298  0.102  95.6  14.478  0.108  71.0  12.947  0.111  76.2  0.375  0.126  95.8
  2 1  -0.176  0.312  94.2  40.487  0.235  53.4  48.858  0.175  21.0  4.514  0.747  94.8
  2 4  -0.734  0.122  95.8  13.723  0.120  80.0  17.789  0.110  66.6  0.299  0.144  97.0

Cox Regression, No Censoring

1.5 2/3 1  -0.211  0.098  94.2  -9.342  0.100  82.8  -8.001  0.106  85.0  -0.232  0.122  95.4
  2 1  -1.543  0.237  93.6  -23.878  0.192  74.0  -28.380  0.152  48.0  -10.127  0.442  94.0
  2 4  -0.039  0.100  94.8  -7.951  0.098  87.8  -10.149  0.095  82.4  0.245  0.121  94.4
2 2/3 1  0.524  0.094  96.0  -17.621  0.102  56.0  -14.948  0.106  68.0  0.484  0.122  95.6
  2 1  -2.889  0.217  95.0  -36.702  0.202  51.0  -44.949  0.146  12.6  -7.136  0.338  96.2
  2 4  0.213  0.100  97.4  -15.107  0.094  70.2  -17.508  0.091  58.2  -0.051  0.118  97.0

Cox Regression, 25% Administrative Censoring

1.5 2/3 1  -0.834  0.106  95.4  -9.046  0.109  86.0  -8.071  0.113  90.2  -0.725  0.130  95.2
  2 1  1.889  0.301  94.6  -24.511  0.215  80.0  -29.169  0.172  56.6  -1.915  0.585  95.8
  2 4  0.003  0.116  94.6  -7.967  0.112  90.8  -10.865  0.106  86.4  0.174  0.139  96.0
2 2/3 1  0.298  0.103  96.2  -14.852  0.108  71.0  -13.351  0.111  76.0  0.404  0.129  95.4
  2 1  0.204  0.312  94.2  -40.506  0.235  53.6  -48.860  0.175  21.2  1.285  0.637  93.6
  2 4  0.688  0.123  95.8  -13.997  0.120  80.0  -18.047  0.110  66.4  0.204  0.147  96.4

EBIAS is the empirical bias ×100.

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