Two of the measured nest parameters, nest height and Disturbance Index, exhibited univariate P -values < 0.25 when comparing successful and failed nests ( Table 4 ) and were therefore included in a final model. The final logistic regression model, controlling for year and area, found the Disturbance Index to be a significant predictor of nest success, with nest height as a potential predictor (disturbance: χ2 2 = 5.57, P = 0.02, nest height: χ2 1 = 3.41, P = 0.09). We found significantly less direct disturbance in territories with successful nests. Nests which were higher in nesting trees showed a trend to be more successful than nests lower in trees, with successful nest height averaging 14.0 ± 3.6 m ( n = 66) and failed nest height averaging 12.1 ± 4.0 m ( n = 107).