Table 1.
Results for the first simulation scenario using 1000 Monte Carlo data sets . E{Y* (gopt)} = 8.12. Ê(ĝopt) shows the Monte Carlo average and standard deviation of estimated values of the true E(ĝopt) using (1). E(ĝopt) shows the Monte Carlo average and standard deviation of values E{Y*(ĝopt)} obtained using 106 Monte Carlo simulations for each data set. PS correct and PS incorrect indicate whether the specified propensity score model is correct or not.
| n=200 | n=500 | n=1000 | ||||
|---|---|---|---|---|---|---|
|
| ||||||
| Estimator | Ê(ĝopt) | Ê(ĝopt) | Ê(ĝopt) | Ê(ĝopt) | Ê(ĝopt) | Ê(ĝopt) |
| PS correct | ||||||
|
| ||||||
| RG | 7.56 (0.35) | 7.49 (0.08) | 7.56 (0.21) | 7.53 (0.05) | 7.55 (0.16) | 7.54 (0.04) |
| Zhang et al. | 7.83 (0.37) | 7.53 (0.08) | 7.76 (0.22) | 7.59 (0.06) | 7.73 (0.17) | 7.62 (0.04) |
| ĈAIPWE | 8.08 (0.37) | 8.07 (0.06) | 8.10 (0.22) | 8.11 (0.02) | 8.11 (0.16) | 8.12 (0.01) |
| ĈIPWE | 7.18 (0.53) | 7.02 (0.42) | 7.66 (0.40) | 7.57 (0.38) | 7.93 (0.26) | 7.90 (0.22) |
| Ĉreg | 7.51 (0.36) | 7.43 (0.18) | 7.52 (0.23) | 7.49 (0.10) | 7.52 (0.17) | 7.50 (0.08) |
|
| ||||||
| PS incorrect | ||||||
|
| ||||||
| RG | 7.50 (0.30) | 7.49 (0.08) | 7.50 (0.18) | 7.53 (0.05) | 7.50 (0.13) | 7.54 (0.04) |
| Zhang et al. | 7.68 (0.30) | 7.52 (0.09) | 7.62 (0.19) | 7.57 (0.06) | 7.59 (0.14) | 7.59 (0.05) |
| ĈAIPWE | 7.95 (0.31) | 8.08 (0.04) | 7.98 (0.19) | 8.11 (0.02) | 7.99 (0.14) | 8.12 (0.01) |
| ĈIPWE | 6.98 (0.34) | 6.93 (0.25) | 7.08 (0.22) | 7.04 (0.16) | 7.12 (0.16) | 7.08 (0.12) |
| Ĉreg | 7.44 (0.31) | 7.43 (0.18) | 7.45 (0.20) | 7.49 (0.10) | 7.44 (0.14) | 7.50 (0.08) |