Table 2.
Results of simulation study comparing 3 different models: Binomial mixture (BM), independent ascertainment (IA), and spatial ascertainment (SA). For predictions at locations without both S1 and S2 data, Pois-NA assumes no surveillance data and Pois-S1 assumes only S1 data. Root mean square errors (RMSE), 95% empirical coverage probabilities (CVG) and averaged posterior standard deviation (SD) are reported.
| Location | Measure | BM | IA model | SA model | ||
|---|---|---|---|---|---|---|
| Pois-NA | Pois-S1 | Pois-NA | Pois-S1 | |||
| S1-S2 | RMSE | 1.55 | 1.53 | 1.52 | ||
| linked | CVG | 0.95 | 0.99 | 0.99 | ||
| sites | SD | 1.51 | 1.49 | 1.49 | ||
| S1-only | RMSE | 7.95 | 10.84 | 7.91 | 8.34 | 6.46 |
| sites | CVG | 1.00 | 0.70 | 0.75 | 0.84 | 0.87 |
| SD | 14.45 | 4.64 | 3.98 | 5.26 | 4.29 | |
| Total | RMSE | 39.90 | 56.06 | 43.35 | 41.78 | 34.95 |
| sites | CVG | 1.0 | 0.92 | 0.81 | 0.99 | 0.92 |
| SD | 65.99 | 56.06 | 28.73 | 53.66 | 31.05 | |