2.4.2. GFWO Nest Success
We created logistic regression models in RStudio version 1.15.2, (R Core
Team 2013) with the package car (Fox &
Weisberg,
2019) using recorded cavity metrics to predict GFWO nest success. We
considered variance inflation factors (VIFs) >5 as
indicators of multicollinearity between variables and z-scaled all
continuous variables to account for varying units of measurement
(O’Brien, 2007). To create candidate models, we used the MuMInpackage (Barton, 2020) in R to generate a model selection table (Burnham
& Anderson, 2002; Field, Miles, & Field, 2012), and evaluated model
fit using AIC adjusted for small sample sizes (AICc) (Burnham &
Anderson, 2002). Models that had ≥10% of the weight of the top model
were considered candidate models for model averaging (Burnham &
Anderson, 2004; Mazerolle, 2006). Using the R package AICcmodavg(Mazerolle, 2020) we estimated the parameter coefficients through model
averaging and determined which parameters were significant using P≤ 0.05 and corresponding confidence intervals.