the National Educational Longitudinal Survey (NELS), High School and Beyond, and the National Youth Survey (see Appendix A for more information about these surveys). Studies that use data from surveys of regional samples and cross-sectional studies (with data from one point in time) are also discussed when they complement the national longitudinal studies and when they cover information not available in the national studies. Examples of cross-sectional studies are Monitoring the Future, which surveys a national sample of high school seniors annually, and first-year data from the National Longitudinal Study of Adolescent Health.1 Regional studies include the Youth Development Study, which follows a random sample of students who were in the 9th grade in 1988 in St. Paul, Minnesota, and a study by Steinberg and colleagues (1993), which followed a sample of students from nine high schools in Wisconsin and northern California for 1 year. A possible shortcoming in regional studies is that conclusions based on these samples may not be generalizable to the national population. (For a good description of other possible methodological shortcomings in many of the studies discussed below, see Ruhm, 1997:Table 1.)
Research on the effects of working on adolescents has focused on a variety of outcomes, including education, vocation, relationships, personal development, and problem behaviors. Although these studies look at a wide variety of consequences, they tend to treat work rather unidimensionally. Most of the studies examine the effects of work intensity, generally measured by average hours of work per week during the school year. Only a few studies try to take into account variations in work intensity over time. No studies on the consequences of work have considered actual work schedules, such as the hours worked on school days versus nonschool days or start and stop times on school days. These types of details are more difficult to collect, given the variability of young people's work schedules, and they are not available in the datasets used to study work outcomes.