In the United States the number of traffic deaths has remained relatively constant at about 41,000 per year for the last decade. In the early years of the 20th century, few people were killed in U.S. traffic crashes because there were few motorized vehicles (Figure 2). As car ownership increased rapidly, so did traffic deaths, peaking in 1972 at 54,589. In nations that are less thoroughly motorized, for example in the world’s most populous nation, China, the number of fatalities per year continues to increase. But in all countries, the number of deaths per registered vehicle has declined over time. Since the late 1930s, the U.S. rate has declined by about 2.7 percent per year, or by half every 25 years. If the 1935 rate were to apply to the present U.S. vehicle population, annual U.S. traffic fatalities would exceed a quarter of a million. Traffic fatalities continue to increase in China, but the rate per vehicle is declining even faster—by 10 percent per year, or by half in 7 years. Arguably, a higher rate of decrease is to be expected when initial fatality rates are higher.

Although traffic fatality rates have declined dramatically in the United States since their peak, U.S. traffic has slipped from being the safest in the world to 13th place as other motorized nations have made far greater strides toward improved safety. Why this has happened should particularly interest less-motorized nations looking for models for their future. In comparing trends among different nations, I would like to shed light on the factors that influence traffic safety and the relative importance of those factors. Among the factors to consider are vehicle engineering—a factor that has dominated U.S. attention—and driver behavior. These are issues I studied extensively in a 33-year research career with General Motors Corporation and have continued to study.

Measuring Traffic Safety

Traffic safety can be measured in a number of ways. One measure is “fatalities per thousand registered vehicles.” Another measure, “fatalities per same distance of vehicle travel,” tends to decline faster because vehicles tend to travel greater distances per year as roads and vehicles improve. The fatality rate per distance driven may have intuitive appeal, but it is estimated in few countries, and then with much uncertainty, so the rest of this article will refer to the per-vehicle fatality rate as the fatality rate. This rate varies by a factor of more than 100 among countries (Figure 3). No relation is found between vehicle age and safety in highly motorized countries, and the countries with high fatality rates use vehicles manufactured in countries with low rates. This indicates that traffic safety involves much more than automotive engineering.

In all countries, male traffic fatalities outnumber female fatalities by about a factor of two. The age-dependent variation in fatalities per year shown in Figure 4 is typical of all countries. Traffic safety is largely a problem of young male drivers. There is also a problem with older drivers but the data help keep it in perspective.

Complex combinations of variables affect the risk of traffic death and account for the striking variation in rates observed in different countries. Countermeasures aiming to reduce deaths can either reduce the chance of harm when crashes take place (crashworthiness) or reduce the probability that crashes happen in the first place (crash avoidance). These two concepts can impinge on each other—a perception of lower risk in a crash can reduce driver caution. If a crash-avoiding measure reduces crash involvement by 5 percent and a crashworthiness measure reduces fatalities by 5 percent, the crash-avoidance measure prevents more harm; crash avoidance prevents both deaths and injuries, but those not killed in a more crashworthy vehicle are still likely to suffer severe injuries.

Factors determining traffic safety can be classified broadly in two groups—those related to driver behavior and those related to engineering, whether of roads or motor vehicles. Modern medicine also affects fatality rates but lies outside the scope of this article. As is the case with longevity across populations and nations, it is difficult to quantify the contributions that affect traffic safety, but an attention to trends can be revealing.

Engineering

In the earliest days of motorization, mechanical failure was a common cause of crashes. The first well-documented fatal crash was described in the March 1899 issue of the British journal The Autocar. An engineering factor, the collapse of a wheel, proved critical. However, the coroner advised the jury at an inquest that “the car appeared to be going at too rapid a pace to be safe, either for the occupants themselves or the public.” More than a hundred years ago, the central importance of the tension between engineering and the use of its products was identified.

Roadways exercise a large influence on traffic safety. On rural two-lane roads, vehicles traveling in opposite directions pass each other only a meter or so apart. Even if drivers obey speed limits, the combined speeds of their vehicles may still far exceed 150 kilometers per hour. At such relative speeds, a head-on crash, caused perhaps by improper overtaking or loss of control on curves, will likely be fatal. On freeways, where there is physical separation between traffic traveling in opposite directions, the only vehicles close to each other are traveling in the same direction at similar speeds. Fixed objects, such as trees, are far removed from the road. Replacing intersections with under- or over-passes largely eliminates side impact. The roadway-engineering improvements typified by the differences between rural two-lane roads and freeways constitute one of the most effective engineering countermeasures. The risk of fatalities on U.S. rural interstate freeways is 55 percent lower than the average risk on all other rural roads. The lowest rate in Figure 6 is 85 percent below the highest rate.

When a vehicle crashes, its size is the attribute of the vehicle that most affects the occupants’ injury risk. Larger vehicles offer more protection in crashes mainly because they are heavier. Their larger dimensions also provide increased protection. Relations between injury risk and vehicle weight are among the most reliably established findings in traffic-safety research. When a car crashes into another that is twice as heavy, the driver in the lighter car is 12 times as likely to die as the driver in the heavier car. But such large weight disparities are rare. The most common weight disparity in a U.S. two-car crash is for one car to be 20 percent heavier than the other. In that case, the driver of the lighter car is twice as likely to die as the driver of the heavier car.

Conservation of linear momentum explains the effect of mass. If two identical cars traveling at 50 kilometers per hour crash head-on into each other, then each car will undergo a speed change of 50 km/h during the crash, assuming an inelastic crash in which both vehicles come to rest in contact with each other at the location of the crash. If one of the cars is 5 percent heavier, the heavier car undergoes a speed change of 48.8 km/h compared with 51.2 km/h for the lighter car; after the crash the pair of cars will move at 1.2 km/h in the direction of the lighter car. Fatality risk increases steeply with increasing speed change; from empirical knowledge, the driver in the heavier car is expected to have a 17 percent lower risk of dying than the driver in the lighter car. Simply adding on a passenger can easily add 5 percent to the weight of a car. A recent study of fatal head-on crashes showed that accompanied drivers were 14 percent less likely to die than drivers traveling alone.

If cars of the same mass crash into each other, each will undergo identical speed changes. However, the risk decreases as the common mass of the vehicles increases (Figure 7, right). This effect is not because of the mass but because heavier vehicles are generally larger. The large size provides more crushable space and therefore more time for the occupants to decelerate. Deceleration forces are what cause injuries and death. Additional size and weight provide additional protection to occupants in essentially all types of crashes, including the proverbial crash into a brick wall (Figure 7, top).

Driver Behavior

Many studies have identified driver error as a factor that contributes to more than 95 percent of traffic crashes. Such findings have generated suggestions that the first priority for increasing safety should be to train drivers to be more skillful and knowledgeable, but this suggestion does not distinguish between two different concepts: driver performance, or what the driver is capable of doing, and driver behavior, what the driver in fact does. Improved driver performance is not generally associated with lower crash risk; paradoxically, greater driver performance is associated in some cases with higher crash risk. Racing-car drivers have higher on-the-road crash rates than average drivers. Everywhere young male drivers have the highest crash rates. Yet young people are the group with the best visual acuity, swiftest reaction times and fastest cognitive processing skills, and males tend to be more knowledgeable about and interested in vehicles and driving.

Higher levels of skill and knowledge are often used to take greater risks. Similarly, many safety technologies do not always lead to greater safety; vehicles equipped with antilock brakes are actually more likely to be in rollover crashes and fatal crashes than vehicles not equipped with this effective and impressive safety technology. Driver behavior—what the driver chooses to do—is the factor that dominates traffic safety.

The average driver has a crash about once per decade, usually causing minor property damage. (The corresponding rate for fatal crashes is about one per 4,000 years.) Drivers often dismiss their own crashes as unpredictable and unpreventable bad luck, or the other driver’s fault. Such interpretations provide one of many reasons why safety professionals avoid using the word accident, a word the mass media also increasingly avoid. A more appropriate interpretation of crashing once per decade is that average driving produces average crash risk.

Airline pilots do not learn to avoid crashes by experiencing them. Instead, they follow procedures reflecting others’ accumulated knowledge. For road vehicles, traffic law attempts to play a parallel role. Even though drivers routinely violate such laws, changes in driver behavior brought about by legislation have led to large reductions in casualties (Figure 8).

Speed limits generally constrain drivers to travel slower than they would choose, thereby reducing risk to them and other road users. Travel speed rather than the speed limit, as such, influences safety. Crash risk increases approximately in proportion to travel speed, injury risk in proportion to travel speed squared, and fatality risk in proportion to travel speed to the fourth power. When speed limits on the U.S. rural interstate system were reduced in 1974 from 70 miles per hour to 55 mph following the October 1973 Arab oil embargo, average travel speed dropped from 63.4 mph to 57.6 mph. This would predict a fatality risk decrease of 32 percent, remarkably close to the 34 percent decline observed.

Drunk driving is a major safety problem, accounting for as much as half of all traffic fatalities. Reducing the availability of alcohol has been shown to reduce traffic deaths. When all U.S. states increased the minimum age to purchase or consume alcohol to 21 years, from earlier limits of 18 to 20 years in the various states, the affected group of drivers experienced a 13 percent reduction in fatal-crash involvement. Police use of random breath testing to enforce drunk driving laws has also proven effective; the Australian state of New South Wales tests about a third of all drivers each year, many of them more than once. Road fatalities declined by 19 percent after the implementation of this policy.

Driver behavior is crucial in the use of the most effective occupant protection device, the familiar lap and shoulder safety belt. Belt use reduces a driver’s risk of death in a crash by 42 percent. All U.S. states except New Hampshire, and most countries, require that seat belts be worn by law. But even countries that have achieved wearing rates exceeding 90 percent do not obtain reductions in driver fatalities as high as 90 percent of the 42 percent effectiveness of the belts. This is because of selective recruitment—the high-risk drivers most in need of protection from belts are the very ones least likely to wear them. The best-evaluated wearing law is that for Britain, where fatality rates for drivers and left front passengers declined by about 20 percent after a belt-wearing law was instituted and wearing rates rose from about 40 percent to more than 90 percent.

The effects of changes in law can be estimated by comparing safety measures before and after the law takes effect. Even larger changes in safety trends may occur for other reasons but may not be detected because they take place gradually over an extended time. Social norms can be quite an important influence. Drivers tend to behave in ways they think their peers would approve of. Various influences, such as media messages on safe driving, may have contributed to an evolving social norm in which irresponsible driving is less acceptable, just as smoking has become increasingly unacceptable in the last several decades. The claim that irresponsible driving is more widespread, as suggested by the term “road rage,” is not supported by any evidence.

Nevertheless, drivers use vehicles for purposes that go beyond transportation, including competition, sense of power and control or, more generally, hedonistic objectives. Speed and acceleration produce pleasurable sensations and excitement even when no specific destination lies ahead and there is no point in haste. The mix of motives evolves in a more utilitarian direction as drivers mature—likely one reason why crash risk is so much lower for 40-year-olds than for 20-year-olds.

Crash risk relates to factors at the very core of human personality. The gender and age profiles of those involved in severe single-vehicle crashes and of those involved in crimes unrelated to traffic offenses (say, burglary) show remarkable similarities, suggesting that the activities share some common sources. No one would suggest that 40-year-olds commit fewer burglaries than 20-year-olds solely because the 40-year-olds have learned how not to commit burglaries! This should invite a parallel caution against interpreting lower crash rates for 40-year-old drivers compared with those for 20-year-old drivers to mean that the 40-year-olds have simply learned how not to crash. Experience may contribute to increasing driver safety, but it seems clear that involvement in severe crashes and arrests for offenses unrelated to driving both reflect deep human characteristics.

A schematic representation of the relative importance of different factors to traffic safety shows the dominant importance of driver behavior (Figure 10). Keep in mind that the various factors interact with each other and that the boundaries are not as clear as in this judgmental representation.

U.S. Safety Trends

Although the factors affecting driver behavior are deep-rooted, it is possible for countries to formulate effective approaches to increasing safety, consistent with the relative contribution of the different factors. The U.S. continues to have many advantages, among them the safest vehicles and good roads. But while the U.S. had the lowest fatality rate in the world before the mid-1960s, by 2000 it had sunk into 13th place, behind Norway, Iceland, Sweden, Britain, Switzerland, Japan, the Netherlands, Australia, Germany, Canada, Finland and Luxembourg—nations with disparate population densities and traffic patterns—and it is still sinking.

A simple comparison of fatalities from 1979 to 2000 shows that while U.S. deaths declined by 18 percent, fatalities in Canada, Britain and Australia declined by 50 percent, 46 percent and 48 percent respectively. In the same 21-year period the number of vehicles increased in all four countries, so that the U.S. fatality rate declined by 45.6 percent. Even though this might seem impressive, it corresponds to a compound reduction of 2.9 percent per year, compared to the nation’s average reduction of 3.2 percent per year from 1900 to 1978. While the fatality rate declined by 45.6 percent in the U.S., it declined by 64.7 percent in Canada, 65.5 percent in Britain and 69.4 percent in Australia. The reasonable interpretation is that if the U.S. had adopted approaches similar to those in the other countries, about 15,000 fewer Americans would have died in the single year 2000. If the U.S. rate had changed each year by as much as rates did in these other nations, since 1979 more than 150,000 American lives would have been saved.

In the United States, an obsessive focus on vehicle factors that are marginally relevant to safety has excluded appropriate emphasis on more relevant factors. Prominent product-liability trials that focus on the injuries of relatively few people dominate public attention. News reports covering these trials rarely mention the larger context—that 41,000 Americans die annually in traffic crashes and a few million others are injured. A population that thinks the marginally relevant is important and that the important is marginally relevant is unlikely to make decisions that promote safety.

The U.S. public directs a great deal of attention to the safety of vehicles; much notice goes to crash-test ratings and to product-liability trials and product recalls, which focus attention on allegations of design and manufacturing defects. No other nation devotes so much discussion to vehicle factors, although many have lower fatality rates. It seems to me that the vast litigation industry in the United States has played an important role in creating and sustaining this misplaced focus. Traffic crashes are a source of vast riches to the legal industry in this country. The legal system also happens to drive the political process in the United States. In other countries, legislators come from varied backgrounds, but in the United States, the political structure is dominated by lawyers.

The enormous influence of the legal system eclipses technical knowledge. This influence has grown dramatically since the publication in 1966 of Ralph Nader’s Unsafe at Any Speed. This book spearheaded a new fixation on vehicle factors, yet if the vehicle that was supposed to be unsafe at any speed had been free of the deficiencies the book alleged, the number of traffic fatalities in 1966 would have been reduced from 50,894 to perhaps 50,870, a difference of less than 0.05 percent. In 1966, the United States had the lowest traffic fatality rate in the world; in the years since then, despite an obsessive focus on vehicle safety or arguably because of it, the U.S. ranking in the world has steadily slipped.

The focus on vehicle factors—factors over which they have no control—has encouraged American drivers to regard safety as something out of their control. The American focus on airbags is one example of this. These were mandated by a lawyer-led safety agency that claimed safety benefits far in excess of published technical estimates and ignored technical information documenting their harmful effects. Before a drug can be prescribed in the United States, it must meet two basic standards—efficacy and safety. Airbags were not shown to meet either, but they were not merely offered to the public, their installation was required. Now, the United States is the only country in the world in which it is illegal to purchase a new car without a device that is known to increase the net harm to women (Dalmotas et al. 1996) and to increase fatality risk in children (Glass et al. 2000). The focus on airbags contributed to postponing mandatory belt-wearing laws, a delay that produced thousands of additional deaths. More fundamentally, the focus on airbags helped mislead U.S. road users into the belief that safety would be achieved without action on their part.

This focus lingers on in the futile pursuit of a safe or smart airbag and the misplaced attention given to this pursuit. After a crash has already commenced, there is so little time for an airbag to inflate that it must inflate at a speed high enough to injure anyone who happens to be in the deployment space. Making airbags less violent will reduce the injuries they cause but will subtract from their already modest net contribution to safety. Effective occupant protection requires that the occupant be restrained before the crash commences.

Nations with safer records than the United States have stricter alcohol laws, which they enforce more stringently; higher belt-wearing rates; and many driver-focused policies. As one example of the difference in attitudes, the United States has an active market in radar detectors whose only purpose is to facilitate illegal speeding, a device that is prohibited in other countries.

This discussion should not be misinterpreted to suggest that vehicle factors are not important. The number of deaths and injuries is so great that any intervention that reduces risk by one percent will prevent more than one death per day in the United States. Many vehicle improvements have had larger effects than this, and additional, but small, improvements in vehicle structure and protection devices are possible through research. Also, many effective policies have been, and are being, developed in the United States. Mandatory safety-belt laws are now in place, speed-limit enforcement is making important contributions to safety, and much progress has been made in counteracting drunk driving. Less-motorized nations have a long way to go before reaching U.S. safety levels.

The Future

As the total number of vehicles in the world continues to grow, the number of traffic deaths is likely to follow a trend similar to those already observed in individual motorized countries. World fatalities are expected to increase beyond the current level of more than a million per year but eventually to reach a peak and then decline. We cannot know how high fatalities will climb before peaking, when the peak will occur and the rate of decline from the peak, but these phenomena will profoundly affect many millions of human lives.

Less-motorized countries, and many already motorized countries, can learn valuable lessons from the most successful motorized countries. A number of them, including Canada, Britain, Australia and Sweden, have reduced their annual fatalities to less than half of their peak values. (U.S. fatalities are still within 24 percent of their all-time high.) Each country has unique features, so it is not feasible to provide one universal blueprint for all. It is best for each country to implement a balanced mix of policies from among those that have been successful elsewhere, and to avoid undue emphasis on factors that are of marginal relevance.

An obsessive focus on vehicular factors would be particularly ill-advised in less-motorized countries. Improving vehicular factors is expensive—the cost of the airbags on U.S. roads exceeds the gross national product of many countries. Also, in less-motorized countries traffic victims are more likely to be vulnerable road users like pedestrians and bicyclists rather than vehicle occupants. A successful set of policies will address, with appropriate focus, factors that relate to vehicles, roadways, road-user performance and, most important of all, road-user behavior.

Bibliography

Aldman, B. 1974. Possible effects of air bag inflation on a standing child. Proceedings of the 18th Annual Conference of the American Association for Automotive Medicine.

Dalmotas, D. J., J. Hurley, A. German and K. Digges. 1996. Air bag deployment crashes in Canada. Paper 96-S1O-05, 15th Enhanced Safety of Vehicles Conference, Melbourne, Australia, 13–17 May.

Ernst, E., E. Bruhning, K. P. Glaeser and M. Schmidt. 1991. Compatibility problems of small and large passenger cars in head on collisions. Paper presented to the 13th International Technical Conference on Experimental Safety Vehicles, Paris, 4-11 November.

Evans, L. 1991. Traffic Safety and the Driver. New York: Van Nostrand Reinhold.

Evans, L. 1993. Medical accidents: No such thing? British Medical Journal 307:1438–1439.

Evans, L. 1994. Small cars, big cars: What is the safety difference? Chance—New Directions for Statistics and Computing. 7:9–16.

Evans, L. 1996. The dominant role of driver behavior in traffic safety. American Journal of Public Health 86:784–785.

Evans, L. 1999. Antilock brake systems and risk of different types of crashes in traffic. Crash Prevention and Injury Control 1: 5–23.

Evans, L. 1999. Transportation Safety. In Handbook of Transportation Science, ed. R. W. Hall. Norwell, Mass.: Kluwer Academic Publishers, pp. 63–108.

Evans, L. 2001. Causal influence of car mass and size on driver fatality risk. American Journal of Public Health 91:1076–81.

Evans, L., and P. Wasielewski. 1987. Serious or fatal driver injury rate versus car mass in head-on crashes between cars of similar mass. Accident Analysis and Prevention 19:119–131.

Glass, R. J., M. Segui-Gomez and J. D. Graham. 2000. Child passenger safety: Decisions about seating location, airbag exposure and restraint use. Risk Analysis 20:521–527.

Hertzman, C. 2001. Health and human society. American Scientist 89:538–545.

Kahane, C. J. 1996. Fatality reduction by air bags: Analysis of accident data through early 1996, NHTSA technical report HS 808 470, U.S. Department of Transportation, Washington, D.C.

Mackay, M. 1985. Seat belt use under voluntary and mandatory conditions and its effects on casualties. In Human Behavior and Traffic Safety, ed. L. Evans and R. C. Schwing. New York, NY: Plenum Press, pp. 259–278.

Patrick, L. M., and G. W. Nyquist. 1972. Airbag effects on the out-of-position child. SAE paper 720442. Warrendale, Pa.: Society of Automotive Engineers.

Leonard Evans received a doctorate in physics from the University of Oxford, England. He spent a 33-year research career with General Motors Corporation. In 2000, he formed Science Serving Society to continue research and other professional activities. He is a member of the National Academy of Engineering and a Fellow of the Society of Automotive Engineers, among other honors. He is currently a Sigma Xi Distinguished Lecturer addressing traffic-safety topics. Address: 973 Satterlee Road, Bloomfield Hills, MI 48304. Internet: http://www.scienceservingsociety.com