Table 1.
Simulation study results for estimating weekly PM
effects across outcome weeks: relative change in RMSE, length of 
PIs, and change in DIC and effective degrees of freedom (pD), averaged across 
simulated replicate datasets
| Exposure | Outcome | Prior | Relative | Relative 
|
|||
|---|---|---|---|---|---|---|---|
| weeks | weeks | structure |
RMSE |
PI length |
DIC |
pD |
|
| Scenario 1 | 14–26 | 27–36 | Dynamic | Reference | |||
| Independent | 2.13 (0.17) | 3.39 (0.16) | 121 (19) | 130 (12) | |||
| Exchangeable | 1.41 (0.08) | 1.53 (0.09) | 46 (11) | 36 (12) | |||
| Fixed | 0.88 (0.04) | 0.66 (0.04) |
(10) |
(3) |
|||
| Scenario 2 | 14–19 | 27–36 | Dynamic | Reference | |||
| Independent | 1.64 (0.13) | 2.74 (0.26) | 57 (15) | 69 (5) | |||
| Exchangeable | 1.14 (0.09) | 1.95 (0.23) | 38 (14) | 44 (6) | |||
| Fixed | 0.87 (0.07) | 0.63 (0.06) |
(9) |
(5) |
|||
| Scenario 3 | 14–26 | 27–31 | Dynamic | Reference | |||
| Independent | 1.74 (0.13) | 2.78 (0.28) | 70 (16) | 76 (7) | |||
| Exchangeable | 1.31 (0.16) | 1.52 (0.32) | 31 (13) | 27 (14) | |||
| Fixed | 1.62 (0.12) | 0.43 (0.06) | 106 (21) |
(6) |
|||
| Scenario 4 | 4-wk lag | 27–36 | Dynamic | Reference | |||
| Independent | 1.34 (0.09) | 2.75 (0.30) | 23 (9) | 46 (5) | |||
| Exchangeable | 1.32 (0.09) | 2.63 (0.38) | 21 (8) | 45 (5) | |||
| Fixed | 1.13 (0.11) | 0.41 (0.08) |
(9) |
(4) |
|||
| Scenario 5 | 4-wk lag | 31–36 | Dynamic | Reference | |||
| Independent | 1.25 (0.09) | 2.85 (0.45) | 17 (10) | 44 (6) | |||
| Exchangeable | 1.24 (0.13) | 2.68 (0.61) | 15 (7) | 40 (9) | |||
| Fixed | 0.93 (0.08) | 0.45 (0.08) |
(6) |
(5) |
|||
The standard deviations across replicate datasets are given in parentheses.