Skip to main content

Table 2 Mean coefficient estimates (and mean standard errors) from 1000 simulations

From: Incorporating sampling weights into robust estimation of Cox proportional hazards regression model, with illustration in the Multi-Ethnic Study of Atherosclerosis

  

Normal

5% 2x mean

5% 3x mean

10% 2x mean

Population

Cox PL

0.223 (0.007)

0.141 (0.005)

0.081 (0.003)

0.118 (0.005)

(unweighted)

PH-robust

0.224 (0.021)

0.130 (0.015)

0.074 (0.009)

0.106 (0.013)

 

outlier-robust90

0.224 (0.015)

0.204 (0.012)

0.175 (0.009)

0.185 (0.011)

 

outlier-robust95

0.224 (0.014)

0.200 (0.011)

0.163 (0.008)

0.179 (0.009)

Sample

Cox PL

0.225 (0.011)

0.149 (0.015)

0.084 (0.009)

0.123 (0.012)

(weighted)

PH-robust

0.233 (0.024)

0.150 (0.022)

0.084 (0.014)

0.121 (0.020)

 

outlier-robust90

0.224 (0.021)

0.192 (0.015)

0.152 (0.010)

0.171 (0.013)

 

outlier-robust95

0.224 (0.025)

0.189 (0.017)

0.144 (0.010)

0.167 (0.014)

  1. The true value of the coefficient β is log(1.25)=0.223. Two versions of the outlier-robust method are included: one using truncation parameter M=0.90 (outlier-robust90) and the other using M=0.95 (outlier-robust95)