Dr. Mannering’s study of accidents in Washington State from 1992 to 1997, a period during which air bags and antilock brakes became prevalent, showed that counter to what would seem logical, there was no major reduction in accidents even with the spread of two seemingly effective safety features. “It just shows how hard it is to determine what causes an accident,” he said. “The one thing you might conclude — it’s called an offset hypothesis — is that people have an acceptance of a level of safety. If they felt safer because of the air bags and brakes, then maybe they drove faster or switched the radio on and off more to, so to speak, compensate. “What seems like an exact science — determining who or what causes accidents — is actually quite difficult,” Dr. Mannering said. Insurance companies, carmakers, inventors, safety advocates and clearly drivers themselves all have an interest in learning about what might reduce the number of accidents or at least make them less severe. Yet there is surprisingly little data to help them.