Table 2.
Empirical coverage of 95% confidence intervals for the treatment group mean after parametric multiple imputation
| ICC | Cluster size | Clusters | NORM | NORM | NORM |
|---|---|---|---|---|---|
| ρ | m | k | FE | RE | IGN |
| 0.03 | 4 | 100 | 99.4 | 96.2 | 95.3 |
| 8 | 10 | 99.6 | 98.0 | 97.4 | |
| 20 | 99.2 | 96.3 | 95.6 | ||
| 40 | 99.6 | 96.7 | 96.0 | ||
| 40 | 10 | 99.2 | 97.7 | 94.6 | |
| 20 | 98.5 | 95.8 | 91.8 | ||
| 40 | 98.4 | 95.1 | 91.6 | ||
| 400 | 4 | 97.8 | 97.5 | 90.6 | |
| 0.08 | 4 | 100 | 99.3 | 95.6 | 94.8 |
| 8 | 10 | 99.4 | 97.4 | 96.0 | |
| 20 | 98.8 | 95.5 | 93.7 | ||
| 40 | 99.3 | 96.1 | 94.7 | ||
| 40 | 10 | 98.3 | 96.2 | 90.9 | |
| 20 | 97.6 | 94.9 | 86.9 | ||
| 40 | 97.3 | 95.2 | 88.3 | ||
| 400 | 4 | 96.8 | 96.5 | 86.7 |
Bolded values indicate coverage rates below 95%.
Italicized values indicate coverage rates above 95%.
Missing data were generated completely at random (MCAR) with a response rate of 60%, and data were imputed using an imputation model that correctly specified the covariate effects.