Presentation to the National Highway Traffic Safety Administration
November 10, 2004
Clarence M. Ditlow, Executive Director
Center for Auto Safety
The Center for Auto Safety has generally advocated that crashworthiness and occupant restraint are the most critical areas in which the National Highway Traffic Safety Administration should regulate and issue consumer information. Nevertheless, we agree that new developments in electronics, computer systems and sensors offer major opportunities for crash avoidance.
We believe, however, that NHTSA is ill-prepared to embark on new crash avoidance initiatives. The reason is its serious lack of crash data on which to base decisions about the most effective strategies for crash avoidance; for early detection of road safety problems; and for rapid evaluation of new concepts, technologies and designs.
To be more specific, the accident data systems of the agency are all nearly twenty years old and have not been subject to serious review or upgrading. They have been systematically compromised by funding cuts. More specifically, the Fatality Analysis Reporting System (FARS) continues to rely almost entirely on police accident reports, the National Accident Sampling System is a mere shadow of its original conception, the General Estimates System is of little serious value, and very few Special Investigations are currently conducted.
There is both a lack of quality (details of each crash) and lack of quantity (number of crashes) in the databases. Even FARS, which is supposed to be a census of vehicle deaths, is far from that and misses over 2,000 vehicle deaths recorded by CDC through death certificates. Whole areas such as fire deaths in vehicles not in motion and child deaths off public roads or in non-moving vehicles are missed. One cannot adopt policies and countermeasures for hazards that are missed by data gatherers.
The consequence of these shortcomings is that we have a very dim picture of what factors are contributing to crashes and crash injuries. We receive belated, if any warnings about special problems such as overly aggressive air bags, the Firestone tires on Ford Explorers, and serious differences in the performance of various side air bag systems. In effect, as a nation, we are driving blind. As a nation, we adopted policies to move children in child seats out of harmâ€™s way from frontal airbags without understanding the extent to which we were placing them in harmâ€™s way of front seats collapsing on them in rear impacts. In effect, as a nation, we a driving blind.
A particular example of the need for substantially upgraded and expanded crash data systems to support crash avoidance strategies came with anti-lock brakes. Despite major claims that these systems offered a major advance in safety, when they were evaluated in the mid-1990s, they were found to provide no net benefit. In particular, they increased rollover casualties at a time when rollovers were becoming a major problem with the extensive use of light trucks as passenger vehicles. Yet, the lack of sufficient crash data kept us from understanding why this supposed safety feature was having this effect. Later, we learned that the agencyâ€™s "stomp and steer" advice may actually have exacerbated the rollover problem.
In the Insurance Institute for Highway Safetyâ€™s recent evaluation of electronic stability systems, it found that while the European systems apparently reduced crash involvement, the General Motors (Cadillac) system may have had the opposite effect. Again, we have insufficient data to understand these phenomena, particularly in vehicles sold in modest numbers such as those typically equipped with new, expensive technologies. The reluctance of the agency to do make and model analysis in areas of emerging technologies handicaps the agencyâ€™s ability to develop rules and policies that advance the technology winners.
Let me be specific about what improvements are necessary to support a major initiative in crash avoidance (or in any area of safety):
Substantially increased funding is critical. The funding for crash data collection and analysis should be at minimum $75 million per year â€“ well under a penny for each time a motor vehicle is filed with fuel. The increased funding for this activity should come from the Highway Trust Fund.
FARS should be enhanced by taking a sample of FARS cases â€“ perhaps one out of eight â€“ and conducting a NASS-level investigation into them. Since these are the most serious crashes, the enhanced data from them would be invaluable. These investigations could be conducted by the nearest NASS team.
NASS should be increased to its original design size: 75 teams with 2 to 4 investigators per team conducting investigations into more than 18,000 crashes annually. If this were done, the current restrictions on the types of crashes investigated could be lifted and we could learn more about pedestrian and heavy truck casualties.
The GES system should be replaced by a system along the lines proposed by Hans Joschke in a study for NHTSA that would use data on selected aspects of crashes that are particularly well reported in particular states. We suspect that the quality of police-reported data has improved in the last twenty years, making this an even more attractive approach to obtaining survey data on a large number of non-fatal crashes.
NHTSA should engage in research into the reliability and potential faults of advanced electronic systems that are becoming common in new motor vehicles. We have found that numerous defects and faults in these systems have led to both expensive repairs and, more importantly, compromises in safety. Examples include the thick film ignition defect in Fords about a decade ago, sudden acceleration caused by defective cruise control electronics, and more recently failures of Volvo throttle body electronic control modules. The University of Maryland Computer Aided Life Cycle Engineering, Electronic Product and Systems Center has worked on these issues and could be engaged to provide substantial guidance and insights into problems with these advanced systems.
We think that new technologies and designs offer major opportunities for reducing the occurrence or severity of crashes. However, without the underlying support of adequate crash data, analysis, and related research, it will be a search in the dark with the public playing the role of test pilots to find successes and failures.