METHODS
Our data came from the Monitoring the Future project, which has conducted annual surveys of nationally representative samples of American high school seniors since 1975. Survey procedures are described in detail elsewhere.17 Nationally representative samples of about 17 000 12th-grade students, located in about 135 schools, were selected each year through a multistage scientific sampling procedure. Confidential, self-completed questionnaires were administered during school hours, usually in a regularly scheduled class period, by professional interviewers employed by the University of Michigan. The questions on driving or riding after drinking or using drugs were included on only 1 of 6 forms, distributed in a random sequence within the classroom, so responses to these questions came from a random one sixth of the total sample of students. Questions on driving or riding in a motor vehicle after use of marijuana or other illicit drugs were added to the study in 2001, so we analyzed data from 2001 to 2011. Student response rates averaged 82% (range = 79%–85%); the great majority of nonresponse was attributable to absenteeism