Table 2.
Average number of imputations outside the unit interval for the linear model (LM), linear model with log transformation (LMlog) and linear quantile regression model (QR), along with average proportions of missing values for each simulated model. Averages are computed over 1000 replications.
| Sample size | Imputation method | Proportion (%) of missing values | ||
|---|---|---|---|---|
| LM | LMlog | QR | ||
| Model 1 | ||||
| 100 | 22.9 | 1.6 | 18.5 | 16.9 |
| 300 | 68.1 | 4.9 | 55.4 | 17.2 |
| 1000 | 228.1 | 15.6 | 182.4 | 17.4 |
| Model 2 | ||||
| 100 | 26.3 | 9.6 | 10.7 | 17.3 |
| 300 | 75.9 | 27.1 | 24.8 | 17.5 |
| 1000 | 247.5 | 88.0 | 74.7 | 17.3 |
| Model 3 | ||||
| 100 | 20.8 | 28.0 | 10.9 | 22.4 |
| 300 | 55.8 | 82.0 | 17.5 | 22.3 |
| 1000 | 178.4 | 271.2 | 35.1 | 22.3 |