Table 4. Binary logistic regression models predicting substantial damage to Navy (USN) and Air Force (USAF) within the United States.
This model represents the best as evaluated by the lowest Akaike’s Information Criterion (AIC) value (see S8 Table for other candidate models). McFadden’s r2 value is 0.23. Predictor variables include flyway (Central, Mississippi, Pacific, Atlantic), airframe (cargo, rotorcraft, stealth, fighter), avian log body mass, military branch, and the airframe × avian log body mass interaction.
| Confidence intervals | ||||
|---|---|---|---|---|
| Parameter* | Coefficient | SE | 2.5% | 97.5% |
| Intercept | -8.54 | 0.21 | -8.96 | -8.13 |
| Central | -0.18 | 0.09 | -0.37 | 0.01 |
| Mississippi | -0.31 | 0.10 | -0.51 | -0.10 |
| Pacific | 0.19 | 0.11 | -0.02 | 0.40 |
| Fighter | 1.94 | 0.29 | 1.37 | 2.50 |
| Rotorcraft | 1.42 | 1.02 | -0.83 | 3.26 |
| Stealth | -1.47 | 1.05 | -3.77 | 0.39 |
| Log mass | 0.98 | 0.03 | 0.91 | 1.04 |
| USN | -1.73 | 0.25 | -2.25 | -1.28 |
| Fighter × log mass | -0.26 | 0.05 | -0.35 | -0.17 |
| Rotorcraft × log mass | -0.32 | 0.18 | -0.67 | 0.06 |
| Stealth × log mass | 0.32 | 0.16 | 0.03 | 0.66 |
*Reference categories include Atlantic flyway, cargo airframe, USAF branch, and the interaction term cargo× log mass.