average, to 20 hours a week or less. The boys who followed this pattern were found to have undertaken the most months of post-secondary schooling (Mortimer and Johnson, 1998) during the years following high school. Males who worked at higher levels of intensity (more than 20 hours per week on average), as well as those who had more limited work experience (not working at all during high school or working for only short periods), had less postsecondary educational attainment. These differences were not explained by apparent selection factors. A similar pattern was observed among females, but was accounted for by earlier differences in educational motivation and performance.

There may be ethnic, racial, social class, and sex differences in these effects. Steel (1991) analyzed data from the NLSY and found that the generally positive effect of high school employment (at ages 17 and 18) on enrollment in an educational institution 2 years later is conditioned by race, sex, and the time spent working. Among white youth, being employed had a positive effect on enrollment. However, this positive relationship was mitigated by hours of work: for each additional hour worked during high school, future school enrollment dropped. This negative effect was even more pronounced among black youth. Carr and colleagues (1996), also using NLSY data, found that as hours of employment in high school increased, the likelihood of entering college and of completing college were reduced, after taking into account demographic and background variables, family differences, school aptitude, and educational expectations. They found this effect on educational attainment persisted up to a decade after high school completion. Ruhm's (1997) analysis of NLSY data found a similar effect for girls.

In general, this evidence suggests that low-intensity employment may support post-secondary educational outcomes while high-intensity employment may hinder them. These studies consistently find that as hours of work per week during high school increase, decreases are seen in the amount of future education. Although these studies statistically control for many pre-existing differences among students who work at high and low intensity, as noted above, without random assignment of students to various work patterns, it cannot be proved that high-intensity work decreases educational attainment. It is possible that unobserved differences among students are responsible for both their decisions about work intensity and education. Nevertheless, the preponderance of evidence suggests a link