A model reliability diagram (calibration plot) shows that predicted probabilities near 0.8 correspond to observed event rates of only 0.6. What does this indicate about the model?

B is correct because when predicted probabilities (0.8) are systematically higher than observed rates (0.6), the model is overconfident — it assigns too high a probability to events that occur less frequently. C (underconfidence) would be the opposite: observed rate > predicted probability. A is wrong because well-calibrated models have predicted = observed. D is wrong; discrimination and calibration are distinct properties.