IIHS research bibliography

A respecification of the regression model applied to additional data results in an estimate of about 45,000 lives preserved by regulations during 1975-78, similar to my previous estimates, and 105,000 lives during 1975-1982. The model allows for introduction of new regulations after 1968 and increased compliance. This result is similar to estimates obtained by several investigators using a variety of methods.

The effectiveness of federal automobile safety standards was examined using detailed data on 236,000 vehicles in fatal crashes in the United States during 1975-1978. Controlling statistically for type of regulation, types of vehicles, and ages of vehicles, the federal motor vehicle safety standards were associated with substantial reductions in car occupant death per 100 million vehicle miles travelled, and some reductions in fatal collisions of the federally regulated vehicles with pedestrians, motorcyclists, and bicyclists. Some 37,000 fewer deaths occurred in 1975-1978 than would have been expected without the federal standards.

Conflicting results have emerged from studies of the effect of governmental regulations on motor vehicle related deaths. A number of researchers have found that seat belts—first required by state laws—and designs chosen by manufacturers to meet Federal Motor Vehicle Safety Standards substantially reduced deaths to vehicle occupants in crashes. However, Sam Peltzman has concluded that the effect of regulation on occupant fatalities are offset by increased nonoccupant deaths. He bases his analysis on trends in published fatality data in relation to economic and other variables. He hypothesizes that drivers have a “demand for risk” that is a function of “driving intensity,” which is affected by a desire to increase earnings and by anticipated cost of crash involvement. According to Peltzman, drivers respond to increased occupant crash protection by increased “risky driving” to have more time to increase earnings, thus increasing the overall crash rate and injury to pedestrians, motorcyclists, and bicyclists.The assumptions, model, and data set forth by Pelztman will be examined here, and his conclusions will be found unwarranted. Data which he does not consider support the conclusion that motor vehicle safety standards are effective in reducing fatalities and that his theory of driver behavior is not supported empirically.

Car occupant and pedestrian fatalities per 100,000 cars registered in Maryland were calculated for the calendar years 1972–1975. Cars sold prior to state or federal regulation (models prior to 1964) had an average yearly occupant fatality rate of 44. Cars with front outboard lap seat belts required by state law (1964–1967 models), as well as some crash protection installed in relation to standards required of cars sold to the government (mainly in 1967 models), averaged 35 occupant deaths per 100,000 registered cars, 20% less than pre-1964 cars. For federally regulated (post-1967) cars, occupant deaths averaged 27 per 100,000 cars yearly, 23% less than 1964–1967 models and 39% less than pre-1964 models. The differences were unrelated to driver age. No consistent differences in the rates at which they killed pedestrians were found among unregulated, state regulated or subsequent federally regulated cars. Thus, there is no evidence to support Peltzman's contention that increased occupant protection resulted in increased hazard to pedestrians.

Federal motor vehicle safety standards must be specified in minimum performance and not design terms. Consequently, different individual designs and approaches can be and frequently are chosen by the various manufacturers to satisfy the minimum performance requirements of a federal motor vehicle safety standard. Using example derived from present standards, basic rules for comprehensively evaluating these safety performance standards are developed. The first stage of an evaluation of a federal motor vehicle safety standard should be a statistical comparison indicating whether or not the performance requirements of the standard, however implemented, are reducing the target damage to people and other associated societal losses. The second stage should be a more clinical and in-depth measurement of the performance, in reducing the target losses, of the various design alternatives chosen by the manufacturers to meet the standard under scrutiny, to determine whether some designs have more payoff than others. There can be no justification for weakening or revoking an existing standard unless there has been a full, comprehensive two-stage evaluation.

Throughout the history of public health progress, including progress in the field of crash protection for motor vehicle occupants, passive approaches have been associated with great and lasting effectiveness in terms of lives saved and injuries reduced. Active approaches, on the other hand, have been associated with far lower levels of payoff in the prevention of death and injury when humans come into contact with the potential hazards in their environments.

The concepts of “cost-benefit” and “cost-effectiveness” are increasingly cropping up in debate over present and future motor vehicle standards. Often they are introduced to attack, or justify, a particular standard. Just as often, they are misunderstood and misused.