years the focus on reducing motor vehicle injury was on improving drivers' skills (Waller, 1994). Although the vehicles offered for sale in the 1950s were safer in many respects than those of earlier decades, the technology for protecting occupants in a crash had developed still further and offered potential gains in safety that were not being put into production (DeHaven, 1942; Stapp, 1957).
In recent years, an increasingly sophisticated and comprehensive approach to the motor vehicle injury problem has developed that addresses safety issues for the driver, occupant, vehicle, and highway system (CDC, 1994). The conceptualization of this approach has been based largely on the work of William Haddon, who developed models for the systematic exploration of countermeasures to reduce or prevent injuries involving components of the causal sequence leading to injury—pre-crash, crash, and post-crash events (see Chapter 1 for a fuller description of the Haddon matrix; Haddon [1972, 1980]). When applied to motor vehicle safety, this approach involves addressing issues concerning the host (driver and passenger), agent (vehicle), and environment (roads and highways). The effort to reduce motor vehicle injuries has been both sustained and multi-pronged, involving surveillance systems; regulatory measures; behavioral, biomedical, and engineering research; state and local programs; and public support. This comprehensive, data-driven approach has been supported by stable federal funding and involves efforts at the federal, state, local, and private-sector levels.
National data systems have been instrumental in allowing trends in motor vehicle injuries to be tracked and injury patterns identified and in assessing the outcomes of prevention interventions. Data systems include the National Vital Statistics System, with information on deaths in the United States; the Fatality Analysis Reporting System (FARS), an in-depth collection of data on all fatal motor vehicle crashes on public roads; and the National Automotive Sampling System, which consists of the Crashworthiness Data System that collects detailed data through crash investigations and hospital injury data and the General Estimates System, a national probability sample of police-reported crashes. The federal government has also made major efforts to improve the quality and consistency of state data systems and has fostered linkages of data systems (e.g., police crash reports and hospital data). These data systems have been used both to identify research questions of importance and to address such questions with high-quality data.
Surveillance data were essential for the enactment and assessment of legislation establishing the minimum age for the purchase of alcoholic beverages. In the early 1970s, about half of the states lowered their minimum age for alcohol purchase, from 21 to 18 in most cases. Beginning in 1976, states reversed course and began to raise the minimum age, a trend that continued into the early 1980s. Multistate research based on FARS and vital statistics data showed the beneficial ef-