Table 6.
Utilising the Akaike Information criterion (AIC) and the parsimony principle to find the best four regression models (top row with ΔAIC =0)) for the four outcomes
| Outcome | Predictors (fixed effects) | Random effects | Selection criterion | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Antigen | Year | Sow | Cross reactions | No. tested | No. positive | Proportion positive | Pig ID | Test ID | Herd ID | df | AIC | ΔAIC | |
| Pig HAIT titre (all four antigen subtypes) | x | x | x | x | x | x | x | 20 | 221 425.7 | 0 | |||
| x | x | x | x | x | x | x | x | 21 | 221 424 | −1.7 | |||
| x | x | x | x | x | x | x | 20 | 221 426 | 0.3 | ||||
| x | x | x | x | x | x | x | 21 | 221 425.9 | 0.2 | ||||
| x | x | x | x | x | x | x | x | x | 22 | 221 424.7 | −1 | ||
| Pig HAIT titre H1N1pdm09 only | x | x | x | x | x | 16 | 135 555.8 | 0 | |||||
| x | x | x | x | x | x | x | 18 | 135 557.8 | 128.2 | ||||
| x | x | x | x | x | x | 17 | 135 556.1 | 0.3 | |||||
| x | x | x | x | x | x | 17 | 135 556.0 | 0.2 | |||||
| x | x | x | x | x | x | x | 17 | 135 559.1 | 3.3 | ||||
| Herd GMT (all four antigen subtypes) | x | x | x | x | x | x | x | 18 | 116 180.4 | 0 | |||
| x | x | x | x | x | x | 17 | 116 217.9 | 37.5 | |||||
| x | x | x | x | x | x | 17 | 116 230.4 | 50 | |||||
| x | x | x | x | x | 16 | 116 260.0 | 79.6 | ||||||
| x | x | x | x | x | 16 | 116 247.4 | 67 | ||||||
| x | x | x | x | x | x | 17 | 116 185.3 | 4.9 | |||||
| x | x | x | x | x | x | 16 | 116 193.2 | 12.8 | |||||
| x | x | x | x | x | 15 | 116 229.6 | 49.2 | ||||||
| Herd GMT H1N1pdm09 only | x | x | x | x | x | 14 | 59 575.76 | 0 | |||||
| x | x | x | x | x | 14 | 59 578.47 | 2.7 | ||||||
| x | x | x | x | x | 14 | 59 597.58 | 21.8 | ||||||
| x | x | x | x | 13 | 59 626.27 | 50.5 | |||||||
The four outcomes were: (1) data of pig HAIT titre for all four antigens, (2) data with only immunogen H1N1pdm09, (3) data on herd level GMT for four antigens and (4) GMT data for only H1N1pdm09.