![[Table of Contents]](/icons/up.gif){kind=icon}
Archived Information
The Quality of Vocational Education, June 1998Before turning to the evidence, I will address theoretical, practical, and statistical issues that must be borne in mind while weighing the evidence. What reasons are there to expect a linkage between academic preparation at the secondary level and success on the job market for noncollege youth? What problems are likely to arise in the attempt to estimate the strength of this linkage? Considering these questions sets the stage for reviewing the evidence.
Research in this area has mainly been motivated by a narrow policy agenda: In light of criticisms of vocational education, and in response to calls by national commissions for increased academic course work, we need to know how academic courses are related to success in the labor market. While these concerns are obviously important, placing them in a broader theoretical context will aid substantially in the interpretation of results. By anticipating the findings that may be expected, and considering how such findings may differ across studies depending on model and measurement differences, we will be in a better position to assess the evidence. Although studies in this literature have given more attention to the methodological complexities of estimating curriculum effects, it is also important to review statistical issues in preparation for interpreting evidence from varied analyses.
Straightforward application of human capital theory leads to the prediction that academic preparation is beneficial in the competition for jobs and wages after high school (e.g., Schultz, 1963; Becker, 1964). Academic study reinforces essential skills such as basic literacy and numeracy, which may help persons find and keep jobs and perform well in them. Academic courses also nurture general cognitive abilities, such as the capacity to learn, to think critically, and to work out a solution to a problem. These abilities may then aid performance in the workplace. Indeed, a large literature in personnel psychology indicates that cognitive ability is the strongest predictor of job performance in a wide range of occupations (see Hunter and Hunter, 1984, for a thorough review). Personal investment in the development of cognitive skills thus may pay off in the labor market.
It is well known that more years of schooling results in better jobs and higher pay for individuals (e.g., Blau and Duncan, 1967; Welch, 1974; Featherman and Hauser, 1978). Human capital theory interprets this evidence as an indication that capacities developed through schooling are utilized at work (e.g., Wise, 1975; Psacharopolous, 1987). One can make the same argument for academic course work in high school: these experiences provide new capacities and reinforce skills learned earlier, allowing individuals to find better jobs, to be more productive, and to earn higher pay. Recent support for this formulation comes from the Secretary's Commission on Achieving Necessary Skills (1991, 1992), which has reported that foundation skills such as basic reading and math abilities, communication and thinking skills, and the ability to work with others are important components of successful job performance (see also Cappelli and Rogovsky, 1993a, 1993b).
A number of complications raise questions for the application of human capital theory in this context. First, even if skills learned in school aid job performance in general, one may question whether skills learned in academic courses in high school are relevant for the types of jobs available to persons who obtain no more than a high school education. Studies of employers' needs indicate that entry-level jobs typically require eighth-grade reading and/or math skills (U.S. General Accounting Office, 1990; Levin, 1993; Rosenbaum and Binder, 1994.) In what, then, would students gain by enrolling in serious academic courses in high school?
One response to this question is that studies of academic course taking and achievement show gains in basic as well as advanced cognitive skills as a consequence of increased academic work (Jones et al., 1986; Gamoran, 1987a). Indeed, several studies show greater benefits to basic skills from college-preparatory courses than from remedial courses (Kerckhoff, 1986; Gamoran, 1987a, 1987b; Meyer, 1992; Kifer, in press). This finding may be less surprising if one bears in mind that many students fail to master what are considered eighth-grade skills while they are in junior high school (National Assessment of Educational Progress, 1990a, 1990b, 1993). The math curriculum in particular tends to be highly repetitive in American school systems; one international study called it a "spiral curriculum," referring to the large amount of time devoted to reviewing previous material before new concepts are introduced (McKnight et al., 1987). If these findings and interpretations are correct, then students who plan to cease their schooling after high school may still acquire useful skills in courses offered in an academic program in high school.
It may be that skills acquired in early high school courses are more relevant for work opportunities and performance than are skills offered in more advanced courses. In that case, the relation between academic courses and job success may be non-linear. That is, one may expect some payoff for increased academic study, but the benefits of additional work may diminish as more courses are added and the curriculum moves beyond what is applicable to entry-level jobs. The question thus becomes one of identifying the point at which such returns begin to diminish, or cease entirely.
Even if the benefits of additional academic studies are limited when one first enters the labor market, greater benefits may accrue as time passes. Higher-order cognitive abilities, such as problem-solving and critical thinking skills, and a heightened capacity for learning, may have limited relevance for entry-level jobs, but may lead to promotion and added productivity in more advanced positions in the workplace. Even more basic skills, such as those required to write a coherent memo or read an operations manual, may be rewarded not at entry level, but through promotion. Thus, one may predict that the benefits of academic courses in high school among those who do not attend college are likely to increase with work experience. More academic preparation in high school may provide a foundation for advancement at the workplace over time.
Another complication is that different occupations may vary in the extent to which academic preparation pays off. For example, the possibility of finding work and advancing in a white collar occupation may be enhanced by additional academic course work in English, while job success in a manual occupation could be much less sensitive to such academic training. Similarly, academic math courses may add to one's prospects of advancement in some fields, but not others. Recent technological advances have probably resulted in some convergence of required cognitive capacities; for example, extensive computerization of factories calls for increased math and reading skills among workers (National Commission on Excellence in Education, 1983; Secretary's Commission on Achieving Necessary Skills, 1991). Still, it would be useful to consider distinct payoffs for different fields of work from taking varied configurations of academic courses.
A final difficulty with predicting an impact of academic preparation on labor market outcomes for non-college youth is that effects that did or did not hold in the past, may differ in the future. The Commission on the Skills of the American Workforce (1990) observed that the traditional workplace required only rudimentary literacy and numeracy skills among entry-level workers. Until now, this situation has characterize the vast majority of positions open to workers who lack college-level education. However, the Commission argued that the "high-performance" workplace of the future would make substantially greater demands on workers at all levels, requiring greater capacities for higher-order thinking, communicating, and problem-solving. These claims were echoed by the Secretary's Commission on Achieving Necessary Skills (1992, p. 5-6):
"A high-performance workplace demands workers who have a solid foundation in the traditional basic academic skills, in the thinking skills necessary to put knowledge to work, and in the personal characteristics that make a worker confident, trustworthy, and responsible....High-performance workplaces also require the ability to manage resources, to work amicably and productively with others, to acquire and use information, to understand and master complex systems, and to work comfortably with a variety of technologies."
So far, only about 5 percent of workplaces are "high-performance," but members of these commissions expect that figure to increase over time. Unfortunately, research to date can only tell us about patterns that have occurred until now, and we will need to speculate about how the situation may change in the future.
Aside from these practical complications, there are at least two theoretical challenges to the human capital formulation which raise doubts about the likely impact of increased academic work on labor market outcomes for non-college-bound youth. Adherents of "screening theory" argue that skills and knowledge acquired through schooling are not the reason more educated persons get better jobs. Instead, education leads to occupational success because employers use it as a signal of a candidates' general ability, and as a sign of perseverance and motivation (Arrow, 1973; Collins, 1979). Following this reasoning, one would not expect more academic work to pay off in the labor market, at a given level of schooling. Employers rarely read transcripts (Bishop, 1985; Rosenbaum, 1989), so they do not know anything about a candidate's record other than whether she/he received a diploma. According to this view, there would be no mechanism by which additional academic courses could pay off for someone who does not obtain further schooling.
Questions for this perspective are raised by studies of the hiring process, which show that employers often assess not only candidates' certification, but their competencies (Bills, 1988, 1992; Rosenbaum and Binder, 1994). A high school diploma often is not considered sufficient indication that a candidate holds the necessary skills. Indeed, today's business executives have complained that high schools are not providing sufficient cognitive or social preparation for work (William T. Grant Foundation, 1988b; O'Neil, 1992). Employers do not inspect high school grades (Bishop, 1989; Rosenbaum, 1989), and they are unlikely to do so because grading standards vary across schools and teachers. However, employers are interested in how well job applicants have mastered the kinds of skills they expect students to have acquired in high school. Since the diploma is not a guarantee, they may rely on interviews, recommendations, and/or tests to provide additional information (Bills, 1988; Rosenbaum and Binder, 1994). Thus, despite the screening hypothesis, there is reason to believe that what students learn in high school may help in the workplace. Consistent with this argument, Wise (1975) showed that not only do ability and achievement improve the productivity of college graduates, but college experiences contribute directly to these enhanced abilities.
Another theoretical objection comes from social reproduction arguments, such as that proposed by Bowles and Gintis (1976). In their view, what is learned in school has great relevance for the workplace. However, it is not cognitive but non-cognitive capacities that matter, according to this perspective. Bowles and Gintis argued that non-college programs in high school prepare students with personality characteristics such as conformity, obedience, and punctuality, which allow non-college-bound students to fit into the capitalist workplace. Instead of aiding labor market success, additional academic training might hinder students on the job market, because it would not foster the attitudes desired by employers.
Subsequent evidence has tended to contradict the reproductionist view. Although non-cognitive skills are related to employment opportunities, the personality characteristics emphasized by Bowles and Gintis do not seem central. Employers appear less concerned with obedience to authority, and more interested in reliability and the ability to get along with co-workers (Commission on the Skills of the American Workforce, 1990). A recent review indicated that employee job performance is positively associated with responsibility, self-initiative, and consistency, rather than with compliance (Cappelli, 1992). An important earlier study found that traits such as "leadership" and "executive ability" exerted significant effects on earnings, whereas factors such as "cooperativeness" and "dependability" were less salient (Mueser, 1979). Another analysis demonstrated that the connection of education to earnings is more closely associated with cognitive than non-cognitive characteristics (Olneck and Bills, 1980). At a minimum, one may conclude that there is enough doubt about the causal chain elaborated by Bowles and Gintis (non-college programs in high school emphasize conformity, which is rewarded on the job) to make it worthwhile to investigate the hypothesis that additional academic courses contribute positively to job success.
These theoretical and practical considerations support a prediction that academic preparation in high school adds to labor market success for youths who do not attend college. However, there are at least four important qualifications to this prediction: (1) the effects are likely to be weakest, and possibly non-linear, at entry level; (2) benefits of academic preparation may increase over time, as knowledge and capacities are utilized in more advanced positions; (3) effects may differ across occupational fields; and (4) effects may differ in the future compared to the past, but information is available only about the past.
Estimates of the impact of curriculum on labor market outcomes have mainly relied on a standard human capital equation, with employment, wages, or earnings predicted as a function of schooling, including curricular differences within levels of schooling, and a variety of background conditions (Mincer, 1974). Typically, this equation has been estimated using ordinary least squares (OLS) regression. Beyond these broad similarities, studies of curriculum effects on work outcomes vary methodologically in important ways. These differences are partly a function of data set differences, but also result from differential attention to model complexities which cannot be addressed in the simple OLS framework. These complexities must be kept in mind as the evidence is presented.
Measurement issues. Exposure to academic courses has been indicated in two different ways: as enrollment in particular academic courses, in contrast to not taking the courses; and as membership in an academic track, that is, a college-preparatory program in high school, in contrast to general and/or vocational programs. At first glance, it would appear the first approach is more useful than the second, given our interest in academic course work. Measuring the number of academic courses taken fits the research question. Often, separate indicators have been used for different school subjects, allowing even more precise measurement of curriculum effects.
A disadvantage to this approach is that it does not distinguish among types or levels of courses within academic subjects. This is especially troubling for English and social studies, because in these subjects there is little variation in the number of courses students take, but substantial variation in the levels of courses (Gamoran, 1987a, 1989). For example, although students in all curricular tracks average between three and four years of English, students in an academic program are much more likely to take an honors English class (Gamoran, 1987a). Variation in exposure to math and science is reflected to a much greater degree in the number of courses taken in these subjects, but differences between advanced, regular, and remedial courses are also important. Among the major academic subjects, foreign language study is probably the only one for which variation among students can be almost completely captured by measuring the numbers of courses taken without regard to level.
By using track designations to indicate exposure to academic courses, researchers have addressed the ambiguity about levels of courses within subjects to some extent. On average, academic subject courses in the college track are pitched at a more demanding level than those taken by general- and vocational-track students (Oakes, 1985; Gamoran, 1987a; Page, 1991; Gamoran et al., 1992). Also, surveys in the 1980s indicated that college-track students took two or three more academic courses during their four years of high school than general-track students (Daymont and Rumberger, 1982; Sebring, 1987). Hence, examining differences in labor market outcomes for students from different tracks amounts to assessing the impact of taking two or three additional courses, with these and other courses taken at a higher academic level, on average.
Disadvantages to this measurement strategy include (a) reliance for the most part on student reports of track positions, which often fail to reflect variation in course taking (Campbell, Orth, and Seitz, 1981; Gamoran and Berends, 1987; Berends, 1992); (b) failure to disentangle effects of course taking from other conditions that may be more closely linked to track position than particular course enrollment, such as future plans (Berends, 1992); and (c) possible changes over time in the meaning and content of high school curricular programs, making it difficult to compare studies over time (Moore and Davenport, 1988). Thus, there are advantages and disadvantages to both measurement strategies, and it is clearly worth considering results obtained with each.
Another measurement issue concerns the incorporation of an indicator of cognitive skills in the analysis of curriculum and job success. Because students who are more able prior to high school take more academic courses, and because ability is correlated with job success, it is important to control for cognitive ability prior to high school in assessing the effects of academic courses taken in high school on work outcomes after high school.
Because of data set limitations, several studies have controlled for ability at the end of high school in assessing the impact of course taking. While this approach accounts for the differential enrollment of more and less able students in academic courses, it raises other problems. The human capital formulation described above indicates that growth in cognitive capacities is a major mechanism through which course taking effects are likely to operate. If this view is correct, then the benefits of academic course work occur by raising levels of cognitive skills, and estimates of course work effects should diminish when skills subsequent to course taking are included in the model. Indeed, assuming reliable tests, there is little reason to expect an effect of academic courses once test scores subsequent to course taking are controlled; by controlling for test scores at the end of high school, the effects of academic course taking are probably missed. As we assess the results of different studies, it is essential to bear in mind whether and how cognitive skills were incorporated into the analyses.
Selection issues. Even with controls for cognitive ability, OLS regression may not yield unbiased estimates of course work effects, because of unmeasured differences among students who take different amounts of academic courses. First, the tests used to control for ability are not perfectly reliable, so there may be aspects of ability that are not tapped by the tests which nonetheless affect both course taking and job success. Second, there may be other differences among individuals that lead to both more academic courses and better work outcomes, such as perseverance and motivation. Both of these sources of bias could lead to overestimates of positive effects of academic course work on labor market outcomes.
At the same time, other factors may lead to biases in the opposite direction. A large majority of students who take a heavy load of academic courses go on to postsecondary schooling (Alexander and Cook, 1982; Vanfossen, Jones, and Spade, 1987). Students who take many academic courses, but do not enter college, may have characteristics that make them less likely to do well on the job market. They may have taken many courses, but performed very poorly. They may be prone to daydreaming. They may have especially unrealistic views about their own capabilities, or they may have drifted through high school without a clear sense of direction (Powell, Farrar, and Cohen, 1985). In the absence of evidence, all this is speculation, but it is clearly reason enough for interpreting estimates with great caution. With work success as the outcome, selection problems are even more complex than in OLS analyses of tracking and achievement (e.g., Gamoran, 1987a; Natriello, Pallas, and Alexander, 1989): In research on tracking, the direction of the bias is well understood, even when there is uncertainty about its strength (Gamoran and Mare, 1989). In the present context, even the direction of selection bias is unknown.
Another way of looking at this problem begins with the decision by many analysts to restrict their samples to youths who do not attend college, as a way of focusing on outcomes for that group. It is likely that the students who benefited most from their academic courses did proceed to college, and are thus not included in the analyses. Students remaining in the sample who took more than the average number of academic courses may be negatively selected for gains from these courses. They may have profited even less than students who took fewer academic courses would have, had they enrolled. Hence, predictions about potential benefits of increasing academic course-taking among non-college-bound students may well be flawed by selection problems.
Analyses of course enrollment generally assume that course taking effects subsequent outcomes, whether at work or school. In reality, however, course enrollment and subsequent experiences are probably jointly determined, at least to some degree. That is, decisions about what courses to take may occur in combination with decisions about one's likely activities after high school (Gamoran and Mare, 1989). On the one hand, this may inflate estimates of academic course-taking effects: Those who foresee a high payoff to such courses are more likely to enroll, while those who do not expect to benefit refrain, and in that case OLS estimates of the benefits of academic course work would be biased upward, because effects are only observed for those who expect to gain. On the other hand, joint determination of course-taking and postsecondary outcomes may lead to underestimates of effects on labor market outcomes, under a different scenario: Students who are highly motivated towards immediate earnings after high school may take the minimum number of academic courses during high school, yet be successful in pursuing a job. In this case, the benefits of academic courses for the average student may be obscured by the labor market success of these earnings-oriented students.
A final selection issue arises because studies differ in the way they treat respondents who are unemployed. By excluding them or by counting their wages as zero—the two most common approaches—researchers may distort estimates of curriculum effects on wages. This is because factors that lead to employment (versus unemployment) may differ from conditions that lead to higher or lower wages. A few studies have used econometric techniques to consider labor force participation and wages simultaneously, and these should be given substantial weight in our review.
Other specification issues. Theoretical considerations led us to anticipate possible non-linear effects of academic course-work on job success. This can be incorporated into an OLS framework, although most studies do not do so. We also predicted that effects may increase for individuals over time. This may be addressed by comparing studies that consider effects at different career points. Also, a few studies report effects for individuals at more than a one-time point.
Most analyses provide estimates of the effects of academic programs or courses, with other conditions held constant. Yet if academic course work is to increase, one may ask, what would it replace? Thus, it is worth noting not only the effects of academic courses, but effects of the alternatives—for example, vocational courses. It is also possible, however, that academic courses could increase modestly among non-college-bound students without sacrificing courses in other areas. This is evident because non-college students currently take fewer courses and acquire fewer credits overall, compared with college-preparatory students (Daymont and Rumberger, 1984). The most thorough way of addressing this issue is to evaluate the impact of combinations of courses, and a few studies have taken this approach.
[Direct Evidence on the Impact of Academic Courses]