Table 3 Means and standard deviations of the estimates of C for a trichotomized variable
From: Polychotomization of continuous variables in regression models based on the overall C index
true C | |||
|---|---|---|---|
n | 0.650 | 0.750 | 0.850 |
10 | 0.671 ± 0.065 | 0.759 ± 0.057 | 0.854 ± 0.039 |
25 | 0.658 ± 0.042 | 0.754 ± 0.036 | 0.851 ± 0.024 |
50 | 0.654 ± 0.030 | 0.752 ± 0.025 | 0.851 ± 0.017 |
100 | 0.652 ± 0.022 | 0.751 ± 0.017 | 0.850 ± 0.012 |
200 | 0.651 ± 0.015 | 0.751 ± 0.012 | 0.850 ± 0.008 |
500 | 0.651 ± 0.010 | 0.750 ± 0.008 | 0.850 ± 0.005 |
- Two samples of the size n from N(0, 12) and N(0,
) are generated by Monte Carlo simulation and the pair consistency probability C for the optimal trichotomized points is calculated by the parametric method. This step is iterated 10,000 times, producing means and standard deviations of C. The values of σ
d
are 1.898, 3.133 and 6.150 for the true values of C 0.650, 0.750 and 0.850, respectively.
) are generated by Monte Carlo simulation and the pair consistency probability C for the optimal trichotomized points is calculated by the parametric method. This step is iterated 10,000 times, producing means and standard deviations of C. The values of σ
d
are 1.898, 3.133 and 6.150 for the true values of C 0.650, 0.750 and 0.850, respectively.