|
| Stern, D., Finkelstein, N., Stone, J. R., III, Latting, J., & Dornsife, C. (1994). Research on School-to-Work Programs in the United States (MDS-771). Berkeley: National Center for Research in Vocational Education, University of California. |
Hundreds of thousands of high school students participate in cooperative education each year. Perhaps a hundred thousand engage in some kind of school-based enterprise. As of 1994, there are a few thousand enrolled in new youth apprenticeships (Finkelstein & Latting, forthcoming). However, these participants in various forms of school-supervised work experience are vastly outnumbered by the millions of students who hold paid jobs during the school year without any school supervision. As noted earlier, even co-op and other school-supervised work experience programs rely on students to find their own jobs. It is therefore important not to overlook the effects of these do-it-yourself arrangements for school-to-work transition.
The proportion of high school students who hold paid jobs during the school year has been increasing since the late 1940s, especially for females (Barton, 1989; Greenberger & Steinberg, 1986). The proportion declines during recessions, but the overall trend has been upward. High School and Beyond survey data showed 59% of sophomores and 76% of seniors in 1980 were in the labor force. In the National Longitudinal Survey of Youth data for 1979 to 1981, 64% of high school juniors worked at least one week during the academic year (excluding summer), as did 73% of seniors (Ruhm, 1993).
Research evidence generally indicates short-term gains in earnings and access to employment after leaving school for students who work while in school. However, the evidence also suggests that working long hours while in school may also interfere with educational attainment and thus detract from earnings and occupational status in the long run. As the number of hours per week spent working while in school increases, a tradeoff between short-term and long-term economic gains emerges.
All studies have found a positive association between amount of high school work experience and employment or earnings a few years later. The National Longitudinal Survey of the High School Class of 1972 (NLS72) data was analyzed by Meyer and Wise (1982). The National Longitudinal Survey of Youth Labor Market Experience (NLSY) has been analyzed by D'Amico (1984), Stern and Nakata (1989), Steel (1991), and Ruhm (1993). The High School and Beyond survey (HSB) was studied by Bishop et al. (1985) and by Marsh (1991). Mortimer and Finch (1986) analyzed the Youth in Transition Survey (YIT), which followed male sophomores for several years after high school. Meyer and Wise (1982) and Stern and Nakata (1989) both limited their samples to those not attending college; Ruhm's (1993) analysis found this restriction did not make much difference. Steel (1991) stratified by race and ethnicity and found positive effect significant only for whites, not African Americans or Hispanics.
There is some evidence that jobs that provide greater opportunity for students to use and develop their skills have more positive effects. Analyzing extensive NLSY 1979 data on qualitative characteristics of students' jobs, Stern and Nakata (1989) found opportunity for skill use and development was the only qualitative factor that significantly predicted subsequent employment and wages in 1980-1982. Mortimer, Ryu, Dennehy, and Lee (1992), studying a sample of 1,000 randomly chosen ninth graders from St. Paul who were followed to grade twelve, examine the relationship between work experience and the development of occupational values. They find no significant effects of hours worked or employment itself on occupational values. However, the opportunity to acquire skills at work had a substantial positive effect on development of intrinsic orientation toward work, that is, of interest in rewards embedded in the work activity itself. Similarly, Stern, Stone, Hopkins, and McMillion (1990) found in cross-sectional data that students who report greater opportunities for learning on the job also express a more positive orientation toward work in general. The absence of opportunities for skill use and development in jobs held by a sample of female students studied by Hamilton and Powers (1990) may explain why these students did not experience much occupational success in the first six months after graduation.
Selection bias has not been controlled in these studies. Students who work, who spend more time working, or who work at better jobs while in high school may also possess unobserved traits that lead to more favorable employment outcomes later. To correct for selection bias, it would be necessary to find other, "instrumental" variables that predict high school work experience but do not directly predict later employment outcomes. Such variables are hard to find. Ruhm (1993) tried using information about the students' geographic location during high school and subsequently, but these are likely to be highly correlated with each other, and the result was inconclusive. However, Ruhm did find that work experience during senior year had a positive effect on subsequent employment and earnings, while work experience during sophomore and junior years did not. This suggests that it is the senior-year work experience itself, rather than the student's predisposition to work, that accounts for the subsequent positive effects in the labor market.
Most studies, but not all, find that students who spend many hours a week in paid employment put in less time on homework, get lower grades or test scores, are more likely to drop out, or express less positive attitudes and aspirations about school. D'Amico (1984) found study time significantly lower and the likelihood of dropping out significantly higher for white females and males who spent a larger proportion of weeks working twenty hours or more. Greenberger and Steinberg's (1986) cross-sectional analysis of a small sample from Orange County, California, found lower grades for students who worked more than fifteen hours in grade ten or more than twenty hours in grade eleven. Steinberg and Dornbusch (1991), in a larger sample from Wisconsin and northern California, found lowest self-reported grades for students who worked the most hours. The same sample was analyzed by Steinberg, Fegley, and Dornbusch (1993) controlling for 1987-1988 work status (not working, working 1-19 hours/week, working 20 or more hours/week) and also controlling for the 1987-1988 lagged dependent variable in predicting self-reported grades and homework hours by work status group in 1988-1989. They found homework hours and grades were lowest in the group working 20 hours or more. Yasuda (1990) also found lowest self-reported grades in the group working most hours. Barton (1989) found 1986 National Assessment of Educational Progress (NAEP) eleventh grade scores in math, science, history, literature, and reading were lowest for students working most hours per week. Lillydahl's (1990) analysis of National Assessment of Economic Education sample found a negative coefficient on (hours/week) in predicting grades, implying lowest scores for those working most hours. Schill, McCartin, and Meyer (1985) studied high school students in the state of Washington, finding lower self-reported grades for students working more than 20 hours than for those working 1-20 hours/week. Steel (1991) found that hours/week employed in 1979 were negatively associated with weeks of enrollment in postsecondary education in 1980-1981. Marsh (1991) analyzed the HSB 1980 sophomore cohort: Those who worked during sophomore year were more likely to drop out before the spring of senior year, and among those still enrolled in 1982, hours worked in sophomore through senior years were negatively associated with the probability of going to college, standardized test scores, and attitudes/aspirations toward school. Analysis of YIT data by Owens (1992) found seniors who worked more hours were less likely to go to college.
On the other hand, Greenberger and Steinberg's (1986) longitudinal analysis failed to confirm that working long hours leads to lower grades. Mortimer, Shanahan, and Ryu's (1991) analysis of a sample from St. Paul, Minnesota, found no significant relationship between grades and working more than 15 hours/week. Hotchkiss (1986) found no significant relationship between hours of work and grades among a sample of students from Columbus, Ohio.
A study by Goldstein (1991) suggests that working students' grades may overstate their actual performance because teachers lower their expectations for these students. Eighty-eight percent of the teachers he interviewed felt that outside jobs had a negative impact on students' classroom performance. However, 60% admitted they had changed the way they taught because of students' outside employment, 47% said they had lowered their expectations, and 28% admitted to lowering their standards for grading. Statistical analysis found no significant relationship between working and grades in this sample from one high school.
Most studies, but not all, find a positive association between working a moderate number of hours/week and school outcomes. D'Amico (1984) found the proportion of weeks spent working 1-20 hours/week was associated with lower (i.e., better) class rank for white males and with lower probability of dropping out for white males and females. Barton (1989) found the highest NAEP scores among students who worked less than 6 or 11 to 15 hours per week. Lillydahl's regression contained a positive coefficient on the linear hours/week term, implying that students who work a moderate amount (13.5 hours, to be exact) get the best grades. Schill et al. (1985) found self-reported grades were higher for students who worked 1-20 hours than for those who were not working at all. Steel (1991) discovered that being employed in 1979 was positively associated with white students' enrollment in postsecondary education in 1980-1981. Steinberg and Dornbusch (1991) found that students working 1-10 hours/week had slightly better grades than those who were not employed. Charner and Fraser's (1987) review concluded that there seems to be a curvilinear relationship between hours/week and grades, with a "`magical cutoff" at 20 hours/week.
However, some studies have found a negative association between working a moderate number of hours per week and grades or other measures of school performance. Barton (1989) discovered NAEP scores were lowest among African American eleventh graders who worked 1-6 hours/week; and among all students, time spent on homework declines monotonically with employed hours/week. Greenberger and Steinberg (1986), in their longitudinal analysis, found that students reduced their homework time after they found jobs. Steinberg et al. (1993) found a similar result: Among students who were not employed in 1987-1988, those who were employed 1-19 hours/week in 1988-1989 spent significantly less time on homework than those who were still not employed in the latter year (in an ANCOVA controlling for time spent on homework in the earlier year).
Although most studies of economic outcomes have not controlled for selection bias, some studies of educational outcomes have attempted to do so in various ways. Steinberg et al. (1993) used students' prior work status as a stratifying variable and controlled statistically for the lagged dependent variable but still found significant negative associations between employed hours per week and school behavior. Hotchkiss (1986) controlled for lagged dependent variables and found no significant effect of working. Lillydahl (1990) used two-stage least-squares, the first stage of which included a tobit equation to predict work hours per week and a probit equation to predict whether the student planned to attend college; in the second stage she found significant linear and quadratic terms in work hours per week predicting grades. Mortimer, Shanahan, and Ryu (1991) developed equations to predict grades and homework time using a Heckman term estimated on the basis of background variables and the timing of first employment in addition to the lagged dependent variables and other control variables; in these equations, work hours per week ceases to be a significant predictor.
A positive relation between working while in college and earnings a few years after college was observed in data from the NLS Young Men cohort (San, 1986; Stephenson 1981, 1982). The relation between working and grades has variously been found to be null (Ehrenberg & Sherman, 1987, analysis of NLS72; Bella & Huba, 1982, single institution); positive (Augenblick & Van De Water and Associates, 1987, survey in state of Washington; Hammes & Haller, 1983, single institution); and negative (Hay, Evans, & Lindsay, 1970, single institution, but men in jobs related to field of study had significantly better grades than men in unrelated jobs). In contrast, there is a consistent negative relation between working and persistence in school, as estimated by Kohen, Nestel, and Karmas (1978); Augenblick et al. (1987); Ehrenberg and Sherman (1987, but students working on campus are more likely to persist through graduation).
These results, as far as they go, are fairly consistent with findings on high school students. Unlike the high school studies, research at the college level has not focused on work hours per week.
|
| Stern, D., Finkelstein, N., Stone, J. R., III, Latting, J., & Dornsife, C. (1994). Research on School-to-Work Programs in the United States (MDS-771). Berkeley: National Center for Research in Vocational Education, University of California. |