Further, a couple of studies have documented a relationship between the amount of employment adolescents undertake, and their subsequent levels of school engagement in national samples. For instance, in a national sample of adolescents who were examined over a two-year period, Mihalic and Elliott (1997) reported that adolescents who worked during both years of the study reported spending less time on schoolwork than those working only one of the years, who in turn spend less time on schoolwork than those working neither of the years. This relationship was identified even after controlling for important factors that might distinguish students who choose to work more years, such as their involvement in school activities, their families' economic circumstances, and their race, age, and gender. A second study found that the greater the number of hours adolescents are employed during their sophomore, junior, and senior years of high school, the less time they report spending on homework (Marsh, 1991a). Yet, it is important to note that not all longitudinal studies examining the relationship between adolescents' employment and school engagement found a negative relationship between these two variables. In fact, Steinberg and Avenevoli (1998) find little evidence of a link between the number of hours an adolescent works and his or her satisfaction with and feelings about school in a ethnically-diverse sample of adolescents from California and Wisconsin, once pre-existing levels of engagement are accounted for.