Table 2.
Bias (bold font), coverage (italics font) and MSE (normal font) for simulation scenario 2 comparing the non parametric Pohar Perme estimator, a conditional model (without covariates), regression standardization under proportional hazards (PH), regression standardization under non-proportional hazards (Non PH) and a marginal model. Bias is expressed as a difference in probabilities
| Years from diagnosis) | |||
|---|---|---|---|
| Method | 1 | 5 | 10 |
| Pohar Perme | 0.0001 | -0.0000 | 0.0005 |
| 94.6 | 95.1 | 95.0 | |
| 180.679 | 416.913 | 936.228 | |
| Conditional Model | 0.0092 | 0.0274 | 0.0434 |
| 87.8 | 61.6 | 34.9 | |
| 235.224 | 1052.450 | 2264.524 | |
| Regression standardization (PH) | 0.0002 | -0.0000 | 0.0008 |
| 88.9 | 96.1 | 94.7 | |
| 154.728 | 313.687 | 395.280 | |
| Regression standardization (Non PH) ∗ | -0.0001 | 0.0006 | 0.0022 |
| 88.6 | 96.1 | 96.7 | |
| 158.038 | 370.164 | 630.970 | |
| Marginal model | -0.0003 | 0.0011 | 0.0046 |
| 95.5 | 95.0 | 93.5 | |
| 151.551 | 374.557 | 888.256 | |
| Relative % increase in precision + | |||
| Regression standardization (PH) | 16.8 | 32.9 | 137.1 |
| Regression standardization (Non PH) ∗ | 14.3 | 12.7 | 49.4 |
| Marginal model | 19.3 | 11.7 | 8.0 |
bias, Coverage, MSE
∗ 18.3% of models did not converge
+ compared to Pohar Perme
PH - proportional hazards
Non PH - non proportional hazards