![[Table of Contents]](/icons/up.gif){kind=icon}
Archived Information
The Quality of Vocational Education, June 1998Our underlying conceptual model suggested that academic courses contribute to work outcomes by improving cognitive skills, which in turn add to labor market success. In addition to considering overall effects of course work, it is appropriate to examine the separate.
Since the 1980s, a variety of studies have provided consistent evidence that students who enroll in more academic courses score higher on tests of cognitive achievement. Studies of math enrollment in the early 1980s, using data from NLS-72 (Welch, Anderson, and Harris, 1982), HSP (Walberg and Shanahan, 1983), and the National Assessment of Educational Progress (Schmidt, 1983) showed substantial achievement benefits for each additional math course. These estimates, it turned out, were inflated by the absence of control for prior math achievement. Still, later studies that took prior achievement into account sustained the earlier conclusions: Students who enrolled in more math courses in high school gained more knowledge and skills (Alexander and Pallas, 1984; Moore and Smith, 1985; Jones, et al., 1986; Gamoran, 1987a; Sebring, 1987; Ekstrom, Goertz, and Rock, 1988; Lee and Bryk, 1988; Meyer, 1992). Several studies observed that higher level courses, (i.e. geometry, algebra 2, trigonometry, calculus) contributed more than lower-level math course, but after correcting for measurement error in an achievement test, Meyer (1992) showed that pre-algebra and algebra1 also bring substantial benefits. A number of studies also provided evidence of curriculum effects on achievement in science, English, social studies, and /or foreign languages (Welch, Anderson, and Harris, 1982; Walberg and Shanahan, 1983; Alexander and Pallas, 1984; Jones et al., 1986; Gamoran, 1987a; Sebring, 1987). Generally, the strongest effects have been observed n math. Gamoran (1987a) noted that whereas differences in the number of English courses did not affect reading or vocabulary achievement (presumably because students vary little on this measure), enrolling in an honors English course contributed to achievement. He interpreted this finding to indicate that high-level English courses contribute more to achievement than lower-level courses.
Overall, cognitive skills contribute substantially to job success, but most of this effect runs through educational attainment. Still, net of schooling, research with several national and regional surveys has shown consistent effects of general cognitive ability on wages and earnings. Crouse (1979) reported analyses of five large surveys, of which four showed positive and significant effects on earnings, with gains as high as fourteen percent for a standard deviation increase in test scores among men aged 35 to 59 with the same amount of schooling. Generally, the benefits of cognitive skills for earnings increase with age. More recent analyses have continued to exhibit notable ties between ability and wages or earnings (e.g., Blackburn and Neumark, 1993; Cameron and Heckman, 1993). A few exceptions can be found (Griffin and Alexander, 1978; Hollenbeck, 1993), but the bulk of the research clearly indicates a positive connection between test scores and long-term work outcomes, even among persons with the same amount of schooling.
Having examined numerous types of ability tests, research in personnel psychology concluded that general cognitive ability is the strongest predictor of job performance (e.g., Hunter and Hunter, 1984). This is consistent with research on the SCANS tests and the work force, which indicates that foundation skills of basic literacy and numeracy, communication, problem-solving, and an ability to work responsibly and with others, matter more for job performance than competencies more oriented to the workplace (Cappelli and Rogovsky, 1993a, 1993b). Studies of test scores and earnings have not examined as wide a wide range of tests, mainly relying on quantitative, verbal, and general cognitive skills tests. Taubman and Wales (1975) found that math ability was linked to earnings, but verbal ability, coordination, and spacial perception were not. Two recent studies observed that for males, quantitative skills are more closely tied than verbal scores to both employment (Rivera-Batiz, 1992) and wages (Murnane, Willett, and Levy, 1993). For females, higher verbal and math scores appeared to enhance employment opportunities (Rivera-Batiz, 1992). Yet another research team found that for males, vocabulary achievement mattered more than math scores for employment, whereas math was more salient for females (Kang and Bishop, 1986). Inconsistencies in the relative salience of math and verbal skills across studies may indicate that an underlying set of general cognitive skills, with which scores in all these realms are correlated, contributes to employment outcomes. General cognitive ability may still be enhanced by academic course work, as indicated by studies of course taking and achievement.
Effects on earnings and wages among young workers. The claim that higher test scores improve labor market outcomes has one very important caveat: Effects on wages and earnings are weak or even non-existent immediately after high school, for those who do not pursue further schooling. Bishop (1985, 1988, 1989, 1993) has shown in a series of reviews and analyses that few benefits of cognitive ability are evident for young men and women. Most estimates for 19-year-olds are close to zero (Hause, 1975; Meyer, 1982; Bishop, 1985, 1989; Kang and Bishop, 1986). Effects for persons in their early twenties are small, but generally non-zero. Taggart, Sum, and Berlin (1987) provide a useful metric for understanding this effect: among 19- to 23-year-olds, one more grade-level equivalent in basic skills was worth 3.6 percent in earnings, or $185 for the year in 1981, whereas one additional grade of schooling completed was worth $715, and a high school diploma was worth $927, net of attainment. Although one study found zero or even negative effects for males as far as ten years out of high school (Bishop, 1993), most research has observed small to moderate positive effects on wages and/or earnings, for men as well as women in their mid- to late twenties (Crouse, 1979; Meyer, 1982; Meyer and Wise, 1982; Cameron and Heckman, 1993; Murnane, Willett, and Levy, 1993). Estimates in these studies of returns to wages or earnings from an increase of one standard deviation in test scores ranged from 3 percent to about 8 percent for males, and from 4 percent to over 15 percent for females.
Studies that allow for changes over the early career in the effects of cognitive skills have found no growth in the first year or two, but increasing benefits thereafter. In their classic study of educational self-selection, Willis and Rosen (1979) found that for men who did not attend college, ability tests had no impact on earnings at labor market entry, but the effects of reading ability became increasingly important over a 20 year period. Examining HSB seniors, Kang (1984) reported no initial effects and no growth in the impact of test scores during the first 21 months after high school. In contrast, Meyer (1982) observed that whereas effects of ability on wages were trivial for NLS-72 participants in 1973 (one year after high school), the benefits had increased and were statistically significant by 1979, six years later. Murnane, Willett and Levy (1993) analyzed math skills in both of these data sets, and found similar results: two years after high school, the benefits of high test scores were nil for males and small for females, but six years after high school the benefits were more substantial.
Murnane, Willett, and Levy (1993) also documented increases in the payoff to academic skills for the more recent cohort (1980 high school seniors) as compared with the earlier cohort (1972 seniors). Whereas an increase of about two standard deviations on the math test was worth $0.46 per hour for men and $1.15 per hour for women 6 years out of high school in 1978, the same test-score increase amounted to hourly wage benefits of $1.15 for men and $1.42 for women at the same career stage in 1986 (all in 1988 dollars). These results were obtained with years of schooling held constant. The authors reported no interactions of test scores with educational attainment, consistent with Crouse's (1979) earlier results with several large surveys (but see Hause, 1975, and Blackburn and Neumark, 1993, for evidence of possible schooling-by-ability interactions).
Murnane, Willett, and Levy's (1993) findings may indicate that academic skills are becoming more important in the labor market, as claimed by a number of recent reports (Commission on the Skills of the American Workforce, 1990; Secretary's Commission on Achieving Necessary Skills, 1991, 1992). Although much research has addressed changes in the returns to education (see Levy and Murnane, 1992, for a review), the work of Murnane, Willett, and Levy is the first to consider historical changes in the impact of cognitive skills using equivalent test instruments and identical statistical models; this despite the fact that Crouse (1979) had decried the lack of historical analyses of ability and earnings more than a decade ago.
Effects of test scores on employment. While test scores bear little immediate impact on wages, that is not the case for employment opportunities. Indeed, basic skills may have their strongest impact at labor market entry. Cameron and Heckman (1993) reported larger effects on annual hours worked for men at age 25 than at age 28. Similarly, Meyer and Wise (1983) found that the benefits of high test scores for men's employment were greater in the first three years out of high school than in the subsequent two years. Both of these studies are exemplary, in that they provided statistical adjustments for selection into the labor force (as opposed to schooling). Also, Meyer and Wise used a statistical model that takes into account the lower and upper truncation on weeks worked in a year. The trend of larger effects on employment at labor market entry may pertain more to men than to women: Meyer (1982) also observed this pattern for men, but for women he found that test scores exerted stronger effects at age 25 than at age 19. Neither Meyer and Wise (1982) nor Cameron and Heckman (1993) included women in their analyses. Kang (1984) observed significant and steady positive effects of test scores on hours worked during the first 21 months out of high school, for both males and females.
Qualitative studies of the employment process support the conclusion from survey research that stronger basic skills enable high school graduates to find jobs, even if they do not obtain higher wages. Testimony cited by the William T. Grant Foundation (1988b) indicates that employers are reluctant to hire persons with weak skills. One business executive explained: "We do not employ people who do not have the basic skills because they are not able to benefit from the training we offer, and they are dangerous to themselves and to others in the workplace" (p. 93). In a more systematic investigation, twenty out of thirty employers interviewed by Rosenbaum and Binder (1994) stated that a lack of basic skills makes some high school graduates unsuitable for entry-level jobs in their firms. Employers who were not concerned with basic skills were generally those who anticipated large turnover or little advancement among entry-level employees.
No study has specifically measured the indirect effects of academic programs or courses on labor market outcomes via growth in academic skills. By juxtaposing studies of course work and achievement on the one hand, and studies of achievement and job success on the other, we can make inferences about the strength of this causal connection. These inferences are highly speculative, because they come from separate analyses using somewhat different model specifications. They should be regarded as extremely tentative.
The four panels of table 5 list (a) estimates of the achievement effects of additional academic courses, adapted from Gamoran (1987a, tables 2, 4, and 5); (b) labor market effects of achievement, as reported by Kang and Bishop (1986, table 6) from the same data set; (c) contributions of academic courses - at any level, and at advanced levels - to labor market outcomes via achievement, computed by multiplying the effects in panel (a) by those in panel (b); and (d) for comparison, direct effects of courses on labor market outcomes, net of achievement, from Kang and Bishop (1986, table 5). The computations in panel (c) show, first, no indirect effects on wages, a result of the fact that cognitive skills have no impact on the wages of 19-year-olds, as explained earlier. Second, effects of academic courses without regard to level are also close to zero for months worked and annual earnings. Third, for advanced academic courses, effects on months worked are consistently positive, though still very small. For example, an advanced academic math course would raise months employed by .033, or about one day (30 x .033 = .99), and an honors English course would increase months employed by almost that much. A student who enrolled in two additional years of advanced math and science and honors English, as well as two more years of regular social studies, could expect to work an additional .182 months, or about five and a half days [2 x (.033 + .009 + .002 + .011 + .007 + .029 = .182; and .182 x 30 = 5.46], in the first 21 months after leaving high school. This estimate is highly speculative, and probably reflects a non-linear relation, in which students with more academic training who do not go to college are slightly more likely to work than not to work.
Both Meyer (1992) and Murnane, Willett, and Levy (1993) corrected HSB tests for measurement error using an errors-in-variables approach. With that correction, Meyer found that lower-level math courses such as pre-algebra and algebra 1 (but not basic or general math) contribute more than two points, or about .30 standard deviation, to the HSB math test. Combining this finding with estimates of test score effects on wages from Murnane, Willett, and Levy (1993), one observes that these courses would add close to nothing to wages immediately after high school (because of the weak impact of test scores at that stage), but as much as $0.21 per hour for males and $0.24 per hour for females six years after high school (that is, .30 standard deviation of the HSB age-24 effects, evaluated at the means for males and females). Again, these conclusions are highly speculative, but they are reasonable given available theory and evidence.
[Interpreting the Evidence]