Table 4.
Expected Number of Limitations for Full Sample FIML Analysis and Sensitivity Scenarios (N=9914)
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| Full Sample Estimation –Unadjusted FIML | Scenario with– Dropouts in Good Health | Scenario with– Dropouts in Poor Health | Scenario with– Dropouts who Die in Poor Health | |
| Round 2 | ||||
| Mean Predicted Limitations | 1.86 (2.15) | 1.88 (0.98) | 1.86 (1.72) | 1.88 (1.94) |
| Skewness | 1.78 | 1.71 | 2.56 | 2.04 |
| Round 6 | ||||
| Mean Predicted Limitations | 2.45 (2.13) | 1.91 (1.75) | 4.09 (3.73) | 2.73 (2.30) |
| Skewness | 1.45 | 1.59 | 1.05 | 1.02 |
| Round 10 | ||||
| Mean Predicted Limitations | 3.36 (1.68) | 1.84 (2.49) | 6.53 (4.57) | 4.09 (2.94) |
| Skewness | 1.20 | 1.61 | 0.11 | 0.46 |
| Average Across Rounds | ||||
| Mean Predicted Limitations | 2.52 (1.99) | 1.89 (1.59) | 4.14 (3.09) | 2.84 (2.18) |
| Skewness | 1.40 | 1.32 | 0.73 | (0.89) |
|
Absolute Difference Between FIML and Simulation in Predicted Limitations Across Rounds | ||||
| Mean | – | 1.57 (1.35) | 2.33 (2.43) | 1.05 (1.24) |
| Median | – | 0.75 | 1.29 | 0.23 |
| 90th Percentile | – | 3.31 | 7.73 | 2.96 |
| Skewness | – | 2.02 | 1.14 | 1.48 |
Standard deviation in parentheses
Estimates are based on predicted values from LCGA estimates.