Also consider that while the study used the same passengers multiple times, the study cannot control which drivers showed up. A 5% difference could be explained by the driving style of the drivers (running yellow lights instead of stopping, average speed, etc). Perhaps drivers drove more conservatively with women…