Table 1.
Simulation results for parameter estimation (ϑ̂) based on 200 Monte Carlo data sets, where ϑ = 1. PL-AFT, partly linear AFT model with r knots; SK-AFT, stratified AFT estimator with K strata; AFT, standard linear AFT model with both Xi and Zi modeled linearly; and AFT-φ, AFT model with true φ plugged in. Range of SEs, the range of SEs for the corresponding performance measure in each column. NA, SE of a performance measure can not be computed for SD. All numbers are multiplied by 1000.
| φ(X) = 2X | φ(X) = 2X2 | |||||
|---|---|---|---|---|---|---|
| Bias | SD | MSE | Bias | SD | MSE | |
| n = 50 | ||||||
| PL-AFT (r = 2) | -12 | 159 | 25 | -2 | 166 | 28 |
| PL-AFT (r = 4) | -10 | 159 | 25 | -1 | 168 | 28 |
| S5-AFT | 95 | 288 | 92 | -65 | 436 | 195 |
| S10-AFT | 28 | 223 | 50 | -43 | 299 | 91 |
| S25-AFT | 31 | 303 | 93 | -38 | 381 | 146 |
| AFT | -4 | 153 | 23 | 21 | 1214 | 1475 |
| AFT-φ | -7 | 154 | 24 | -5 | 158 | 25 |
| n = 100 | ||||||
| PL-AFT (r = 2) | -9 | 113 | 13 | -2 | 115 | 13 |
| PL-AFT (r = 4) | -9 | 113 | 13 | -1 | 115 | 13 |
| S10-AFT | 44 | 163 | 29 | -23 | 210 | 45 |
| S25-AFT | 1 | 157 | 25 | -9 | 185 | 34 |
| S50-AFT | -7 | 193 | 37 | 8 | 209 | 44 |
| AFT | -8 | 113 | 13 | 71 | 755 | 575 |
| AFT-φ | -9 | 113 | 13 | -2 | 111 | 12 |
|
| ||||||
| Range of SEs | 8-21 | NA | 1-12 | 8-86 | NA | 1-209 |