Statistic (estimated using a 16% hold-out test sample)	Main effects logistic regression model	Final classification tree model	Final random forest model
Accuracy (i.e., percentage correctly classified)	69.8%	78.3%	78.5%
Sensitivity (i.e., true positive rate)	49.6%	57.5%	60.3%
Specificity (i.e., true negative rate)	83.2%	92.1%	90.6%
Balanced accuracy (mean of sensitivity and specificity)	66.4%	74.8%	75.5%
Area under the ROC curve	74.2%	81.9%	83.8%