In the full logistic regression model with all four predictors, age (odds ratio [OR] = 1.022, 95% confidence interval [CI] [0.993, 1.052]) and gender (OR = 0.746, 95% CI [0.392, 1.420]) were highly nonsignificant, p > .14; whereas the AMPS Motor (OR = 0.351; 95% CI [0.212, 0.579]) and Process scores (OR = 0.013, 95% CI [0.005, 0.038]) were highly significant, p < .001. When compared with the full model, the reduced model with the AMPS Motor and Process scores as the only predictors had a very similar AUC (0.922 vs. 0.924), 2(1) = 0.56, p = .45 (Figure 3). This means that age and gender contributed almost no information to prediction in addition to AMPS Motor and Process scores. The reduced model also suggested the prediction that a person would fail or receive driving restrictions when the predicted probability of that person was above .345. In other words, the logistic regression model predicts the probability that a person will fail or receive driving restrictions on the basis of their AMPS Motor and Process scoresthe basis of the ROC curve. This yielded a sensitivity of 84.6% and a specificity of 88.8%