Table 4.
Logistic regression model statistics: goodness-of-fit, prediction statistics, likelihood ratio test and multimodel inference statistics
| Model 1 | Model 2 | Model 3 | Model 4 | |
|---|---|---|---|---|
| Full model logistic regression | ||||
| AIC | 1,257 | 1,102 | 1,203 | 1,078 |
| Hosmer–Lemeshow test | ||||
| χ2 | 7.19 | 11.57 | 4.09 | 14.19 |
| P | 0.52 | 0.17 | 0.85 | 0.08 |
| Prediction statistics | ||||
| Threshold of presence | 0.20 | 0.20 | 0.20 | 0.20 |
| Highest kappa value | 0.19 | 0.39 | 0.29 | 0.42 |
| Sensitivity | 0.32 | 0.54 | 0.49 | 0.54 |
| Specificity | 0.90 | 0.91 | 0.88 | 0.92 |
| Positive predictive value | 0.25 | 0.40 | 0.31 | 0.43 |
| Negative predictive value | 0.92 | 0.95 | 0.94 | 0.95 |
| Likelihood ratio test | ||||
| Residual deviance | 1,175 | 1,018 | 1,119 | 992 |
| Residual df | 2,022 | 2,021 | 2,021 | 2,020 |
| Difference vs. Model 1 (χ2) | − | 157 | 56 | 183 |
| P | − | < 0.01 | < 0.01 | < 0.01 |
| Multimodel inference | ||||
| Lowest AIC | − | − | 1,177 | 1,042 |
| Highest Akaike weight | − | − | 0.11 | 0.14 |
| Number of generations | − | − | 680 | 640 |
| Weighted models | ||||
| Threshold of presence | − | − | 0.20 | 0.30 |
| Prediction statistics | ||||
| Highest kappa value | − | − | 0.28 | 0.42 |
| Sensitivity | − | − | 0.47 | 0.49 |
| Specificity | − | − | 0.88 | 0.94 |
| Positive predictive value | − | − | 0.31 | 0.46 |
| Negative predictive value | − | − | 0.94 | 0.94 |
df = degrees of freedom.