To address elderly driver crashes at stop sign intersections, we developed a driver evaluation system using DMS-based head movement analysis and driving scores. The system, integrated with a driving simulator, generated diagnostic reports to provide feedback on driving behavior. Among eight elderly drivers, driving scores increased by 9% on average after feedback, with a 40% longer checking time and reduced vehicle speed during confirmation. However, not all drivers showed similar improvements; drivers with reduced cognitive function or overconfidence demonstrated limited progress. These findings highlight the potential for short-term behavioral change and the need for tailored feedback strategies for individual drivers.