MI (nearest neighbour)
Mass imputation where the non-probability sample serves as the donor pool for k-nearest-neighbour imputation onto the probability sample.
nonprobsvy version: 0.3.0 | R: 4.6.0 | run: 2026-05-24 08:08:12 | commit: 4b2ba9a
| script | k | var_method | rep_type | num_boot | alpha | n_reps | bias | rmse | mc_se | mean_se | coverage | ci_width |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| mi_nn | 1 | analytic | NA | NA | 0.05 | 500 | 0.024 | 0.104 | 0.005 | 0.068 | 0.792 | 0.265 |
| mi_nn | 5 | analytic | NA | NA | 0.05 | 500 | 0.038 | 0.092 | 0.004 | 0.052 | 0.708 | 0.203 |
| mi_nn_boot | 1 | bootstrap | subbootstrap | 50 | 0.05 | 100 | 0.015 | 0.094 | 0.009 | 0.055 | 0.740 | 0.214 |
| mi_nn_boot | 5 | bootstrap | subbootstrap | 50 | 0.05 | 100 | 0.029 | 0.088 | 0.008 | 0.050 | 0.720 | 0.194 |
Notes
- DGP:
default. - The analytical SE for NN-imputation does not account for the additional variability introduced by the donor selection step. Expect under-coverage; bootstrap variance is the recommended remedy.
- Sweeps
k = 1andk = 5.