Archived InformationThe Quality of Vocational Education, June 1998
Researchers give educational attainment a central place in models of status mobility in American society. In Blau and Duncan's (1967) classic model, for example, parental background contributes little to status when educational attainment is controlled. The effect of educational attainment on status is direct, whereas the effect of parental background is indirect. The model suggests that high-status families maintain their position by encouraging the educational attainment of their children. The children of lower status families ascend the ladder of success via educational attainment.
Vocational education usually means training for occupations requiring less than a college degree, and so the educational attainment of most vocational graduates is limited. Heyns (1974) noted that 85 percent of those in academic programs but only 15 percent of the students in other programs go on to college. This has been a cause of real concern for some educators. Compared to academic programs, vocational and other nonacademic programs seem to limit young people's opportunities for social advancement.
Although researchers agree that vocational and academic students differ in educational attainment, they disagree on what lies behind this difference. Some researchers believe that tracks shape students and that tracking is responsible for the attainment gap between academic and vocational students. Others argue that the attainment gap is the inevitable result of student self-selection into academic and vocational programs. They believe that students who select academic and vocational programs differ so much in aptitude and aspirations at the start of high school that they would differ in their attainments no matter what they experienced in high school.
In this chapter, I focus on curricular effects on educational attainment. The studies that provide the evidence on this topic are varied, and I first describe some of their features. I then turn my attention to study findings. Finally, I discuss implications of the study results.
I located studies of educational attainment in two places. My first source was a computerized search of the library data base maintained by the Educational Resources Information Clearinghouse (ERIC). I searched the full text of ERIC citations and abstracts from the years 1982 through September 1993 for the terms secondary education, vocational education, and educational attainment. I located 15 documents that included these terms in citations or abstracts. I reviewed all the documents either in full or in abstract form. My second source was the reference lists in the documents located in the ERIC search. I used these reference lists to find other relevant documents.
Through direct database searching and branching, I located a total of 8 usable studies (table 5.1). Each of the studies analyzed outcomes on a quantitative or quasi-quantitative scale of educational attainment, and almost all of the studies used data collected longitudinally from national samples of young people. Conroy and Diamond's (1976) and Hauser, Sewell, and Alwin's (1976) studies were the exceptions to the rule. Conroy and Diamond's study used cross-sectional data collected from a state-wide sample of young people in Massachusetts; the study by Hauser and his colleagues used longitudinal data collected from a representative sample of young people in the state of Wisconsin. The studies are far from uniform in all respects, however. They differ in (a) method for identifying vocational students; (b) definition of postsecondary education; (c) use of dichotomous vs. continuous measures of postsecondary enrollment; (d) length of follow-up period; and (e) handling of educational aspirations in regression analyses. Each of these characteristics can affect study outcomes.
Identification of vocational students. Campbell and Basinger (1985), Creech et al. (1977) and Hauser et al. (1976) used school records to classify students as vocational or nonvocational, but other researchers used student self-reports to make the classification. Jencks and Brown (1975) relied on self-reports from early in high school. Conroy and Diamond (1976), Grasso and Shea (1979), and Hilton (1971) used self-reports made at the end of high school or in response to questions asked in follow-up interviews. Vanfossen et al. (1987) incorporated both sets of self-reports into their classification of students as track stayers and track movers. Track stayers were students who reported being in the same curricular programs in both their sophomore and senior years in high school; track movers were those who classified themselves differently in the two years. Vanfossen and her colleagues restricted their analyses to track stayers, about 60 percent of the total sample.
Definition of postsecondary education. Most of the studies counted enrollment in a college or technical or vocational institution as postsecondary education (e.g., Campbell & Basinger, 1985; Conroy & Diamond, 1976; Creech et al., 1977; Hauser et al., 1976; Jencks & Brown, 1975; Vanfossen et al., 1987). Hilton (1971), however, measured educational attainment as number of years of education in two- and four-year colleges only. Grasso and Shea (1979) used different approaches in different analyses. For some of their analyses, they defined attainment as years of college or years of technical and vocational training. For other analyses, they defined educational attainment as years of college alone.
Dichotomous vs. continuous criteria of educational attainment. Some researchers (e.g., Campbell & Basinger, 1985; Conroy & Diamond, 1970; Creech et al., 1977) used a dichotomous measure of educational attainment. They classified students into the two categories of (a) those who continued their education beyond high school and (b) those who did not. Other researchers used a continuous measure of attainment. Jencks and Brown (1975), for example, used total years of schooling as their dependent measure. Vanfossen et al. (1987) used a five-category variable: (a) non-student; (b) part- or full-time student at other institution; (c) part- or full-time student at a two-year college; (d) part- or full-time student at a public or four-year college; (e) part- or full-time student at a private four-year college. Hilton (1971) used a similar five-category variable in some analyses, but he used a ten-category variable for other analyses. Grasso and Shea (1979) used different approaches in different analyses. They used a dichotomous dependent variable in some analyses; for other analyses their dependent variables was number of years of schooling.
Length of follow-up period. Creech et al. (1977) determined educational attainment from follow-up data collected about 18 months after students completed (or were scheduled to complete) high school. Vanfossen et al. (1987) analyzed data from a two-year follow-up. Hilton's (1971) data came from one- and three-year follow-ups. Campbell and Basinger (1985) collected follow-up data one through seven years after students left high school. Conroy and Diamond (1976), Grasso and Shea (1979), Hauser et al. (1976), and Jencks and Brown (1975) determined educational attainment five or more years after students were scheduled to graduate from high school.
Educational aspirations. Campbell and Basinger (1985), Creech et al. (1977), Jencks and Brown (1975), and Vanfossen et al. (1987) included variables in their regression equations that represented the high school educational aspirations of the students. Their regression equations are therefore designed to provide estimates of the importance of curricular programs for students with the same educational aspirations. Hauser et al. (1976) and Hilton (1975) did not include variables representing educational aspirations in their regression equations. Their equations provide estimates of the importance of curricular programs for students who are similar in aptitude and background but who differ in educational aspirations at the start of high school. Grasso and Shea (1979) included high school educational aspirations as a variable in some of their equations but they did not include aspirations as a variable in other equations.
I classified studies into three types for the sake of exposition. Studies of the first type provide simple descriptive information. They report the proportion of students in each program who continue their education or training beyond high school. The second type of study uses regression analysis to determine the proportion of students in each program who would continue their education if the students in the programs were comparable in aptitude and background. The third type of study also uses regression analysis. The dependent variable in these regression analyses is the highest year of schooling that students complete.
Descriptive studies. Four studies provide simple descriptive statistics on the educational attainments of students in different curricular tracks (table 5.2). Two of the studies (Creech et al., 1977; Hilton, 1971) examined likelihood of enrollment in two- and four-year colleges. Three of the studies (Campbell & Basinger, 1985; Conroy & Diamond, 1976; and Creech et al., 1977) examined the likelihood that the students would pursue either postsecondary education or some other form of postsecondary training.
College enrollment. Creech and his colleagues based their report on follow-up data collected 18 months after students in the Class of 1972 sample had graduated from high school. They found that 69 percent of the academic students, 27 percent of the general students, and 14 percent of the vocational students entered two- or four-year colleges. Hilton (1971) reported simple descriptive statistics from a one-year follow-up of the Academic Growth Study sample. He found that 76 percent of the students in academic programs, 30 percent of the students in general programs, and 26 percent of the students in vocational programs enrolled in two- or four-year colleges. Thus, both studies found that vocational students were far less likely than academic students and only slightly less likely than general students to enroll in college after high school graduation.
Postsecondary education or training. Analyzing data from the New Youth Cohort of the National Longitudinal Surveys of Labor Market Experience (NLS-Youth) and the Class of 1972, Campbell and Basinger (1985) reported that 92 percent of academic students, 60 percent of general students, and 61 percent of vocational students pursued some form of postsecondary education. In their analysis of data from the 18-month follow-up of the Class of 1972, Creech et al. (1977) found that 81 percent of the academic students, 40 percent of the general students and 29 percent of the vocational students enrolled in a college or postsecondary training institution. Finally, Conroy and Diamond (1976) found that 81 percent of the nonvocational students and 48 percent of the vocational students in the Massachusetts Transition project had enrolled in some type of postsecondary institution after high school. Thus, using a broader definition of postsecondary education, these investigators found a smaller gap between academic and vocational students in educational attainment. They also found almost no gap between vocational and general students in attainment.
Conclusions from descriptive studies. Although the proportion continuing on to college varies from one data set to another, the similarities in the data are more striking than the differences. In each of the data sets, the vast majority of academic students and a minority of general and vocational students enroll in college programs. On the average about 75 percent of academic students, 30 percent of general students, and 20 percent of vocational students enroll in two- or four-year colleges, and about 85 percent of academic students, 50 percent of general students, and 45 percent of vocational students pursue some form of postsecondary education or training. The difference between academic and nonacademic programs in postsecondary enrollments is large; the difference between general and vocational programs is small.
Regression analyses with dichotomous outcome variables. The students who pursue vocational and nonvocational programs differ in aptitude and other characteristics on entry into high school, and vocational and nonvocational students also follow different curricula during the high school years. Each of these factors could lead to different educational outcomes for vocational and nonvocational programs. Simple descriptive analyses like those described above do not help us decide whether student characteristics or program characteristics determine outcomes. Regression analysis with longitudinal data promise to provide a sounder basis for conclusions.
One type of regression analysis examines a dichotomous measure of educational attainment. Students who continue their education form one category; students who do not continue in school form the other category. The goal of the regression analyst is to determine whether curricular track is related to category membership when other influences on educational attainment are held constant. Three sets of investigators carried out this type of analysis (table 5.3). Grasso and Shea (1979) focused on enrollment in two- or four-year colleges. Campbell and Basinger (1985), Creech et al. (1977), and Grasso and Shea (1979) looked at curricular tracks and enrollment in any type of postsecondary institution, including training institutes.
College enrollment. Grasso and Shea (1979) examined the influence on college enrollment of student aptitude, background, and curricular track, but they did not study the role that college aspirations played. They found that the likelihood of an average student enrolling in a two- or four-year college was 62 percent if the student followed an academic program in high school; the likelihood of the same student going to college was 21 percent if the student followed a vocational program and 32 percent if the student followed a general program. Thus, Grasso and Shea found that students from academic and vocational programs differed in likelihood of college enrollment even when they were similar in aptitude and background. Similar students from general and vocational programs, however, differed only slightly in likelihood of enrollment in college.
Postsecondary education or training. Campbell and Basinger (1985) reported that the effects of curricular track are small when family background, college aspirations, and 10th grade test scores are controlled. They found that vocational graduates are 8 percent less likely than similar academic program graduates and 3 percent less likely than similar general program graduates to enroll in postsecondary educational institutions. For a student who is average in background, aspirations, and aptitude, the likelihood of enrolling in a college or training institute would be 73 percent if the student graduated from a high school academic program, 68 percent if the student graduated from a general program, and 63 percent if the student graduated from a vocational program.
Creech et al. (1977) found that curricular track had a somewhat stronger effect. They reported that the probability of postsecondary enrollment depends on high school curricular program. They found that students in academic programs who are average in scholastic aptitude, socioeconomic background, and educational aspirations have a 75 percent chance of continuing their education beyond high school; similar students in nonacademic programs have a 44 percent chance of continuing.
Grasso and Shea's (1979) results also suggest that curricular programs have an effect on educational attainment. One of their analyses examined the likelihood of a student's enrollment in either a college or postsecondary vocational institute. The analysis controlled for scholastic aptitude and socioeconomic origins but not for educational aspirations. They found that the likelihood of an average student enrolling in either a college or a postsecondary training institute was 81 percent if the student pursued an academic program in high school. The likelihood of postsecondary enrollment for the same student was 62 percent if the student pursued a vocational program in high school.
Conclusions from analyses with dichotomous outcome variables. It is clear that the educational attainments of academic and vocational students are different. All studies show that academic students are more likely than vocational students to enroll in college. It is also clear that regression analyses that hold constant student aptitude, background, and aspirations do not change the picture much. The regression analyses show that academic students are more likely to enroll in postsecondary institutions than are vocational students who appear to be comparable in aptitude, background, and aspirations.
Such results seem to show that curricular tracks shape educational attainment, but these regression comparisons are methodologically flawed. It is important to note that students selecting academic and vocational programs differ in many characteristics that influence their decisions about program enrollment. For regression results to be accurate, researchers would have to measure all these characteristics fully and reliably. Available surveys simply do not meet this requirement. They either ignore student educational aspirations, or they use crude and unreliable measures of student aspirations. It is a good idea therefore to treat with skepticism results from regression comparisons of academic and vocational students.
In addition, comparisons of educational attainments of academic and vocational students miss the point. After all, students usually enroll in college-prep programs because they intend to go to college; students often enroll in vocational programs because they do not intend to go to college. Why compare the two programs on whether they channel students into college? The criterion seems appropriate for college-prep programs but inappropriate for vocational programs. Asking whether vocational programs channel vocational students into college is, in effect, asking whether the programs route students to places where they do not choose to go. Grasso and Shea (1979) are among those who believe that it is inappropriate therefore to compare academic and vocational programs on their effects on college enrollment. They believe that it is more appropriate to compare programs on common goals.
Comparing general and vocational students makes more sense. General and vocational students are similar in academic aptitude and background, and the two groups do not differ radically in their goals. Survey results also show that the two groups are similar in likelihood of enrollment in college and postsecondary training institutions. About 30 percent of those from general programs and about 20 percent of those from vocational programs enroll in college. About 50 percent of those from general programs and 45 percent of those from vocational programs pursue some form of postsecondary education or training. Holding constant aptitude, background, and aspirations in regression analyses does not change the picture much. With background, aptitude, and aspirations controlled, students from general programs are slightly more likely than vocational students to go on to college, but the two groups do not differ at all in likelihood of pursuing some form of postsecondary education or training.
Regression analysis with continuous outcome variables. Another type of regression analysis used a continuous measure of educational attainment. The outcome variable in these analyses was number of years of postsecondary education or an equivalent measure. Five research studies used such continuous measures (table 5.4). Two of these studies (Grasso & Shea, 1979; Hilton, 1971) investigated attainment in two- or four-year colleges. The other three studies (Hauser et al., 1976; Jencks & Brown, 1975; Vanfossen et al., 1987) investigated attainment in two- or four-year colleges or vocational institutes.
College enrollment. Grasso and Shea (1979) found that students who are average in socioeconomic background and scholastic aptitude would finish 14.0 years of schooling if they were graduates of an academic program in high school, 12.6 years if they were graduates of a general program, and 12.3 years if they were graduates of a vocational program. The average number of years of schooling for all students in the total sample was 13.13, and the total standard deviation was 2.06. The z-score equivalents for number of years of schooling are therefore 0.41 for the academic program, -0.24 for the general program, and -0.40 for the vocational program. The difference in years of schooling for students in academic vs. vocational programs is equivalent to 0.81 standard deviations. The difference between students in general vs. vocational programs is equal to 0.16 standard deviations.
Cohen (1977) has characterized differences of 0.2 standard deviations as small, differences of 0.5 standard deviations as moderate, and differences of 0.8 standard deviations as large. By Cohen's guidelines, placement in an academic rather than vocational program will have a large effect on years of schooling; placement in a general rather than vocational program will have only a small or trivial effect on years of schooling.
Like Grasso and Shea, Hilton (1971) analyzed educational attainment without taking educational aspirations into account. He categorized educational attainment on a five-point scale in his one-year follow-up and on a ten-point scale in a five-year follow-up. In his one-year follow-up study, he found that students who were average in socioeconomic background and aptitude differed by 0.40 standard deviations in educational attainments if they were in academic and nonacademic programs in high schools. Results from Hilton's five-year follow-up were remarkably similar. Hilton's study suggests that curricular programs have a moderate effect on educational attainment.
Postsecondary education or training. Hauser et al. (1976) analyzed data from the Wisconsin class of 1957. Like Grasso and Shea (1979) and Hilton (1971), they did not include educational aspirations in their prediction equations but instead focused on the effects of academic aptitude and socioeconomic background. They found that the average number of years of postsecondary schooling was 13.7 for average students who were in an academic program in high school, and 13.1 for comparable students in a nonacademic program. The average z-score on a scale of educational attainment was 0.16 for those in the academic program and -0.17 for those in the nonacademic program. The difference of 0.33 standard deviations would ordinarily be considered a small to moderate difference.
Jencks and Brown (1975) used five-year follow-up data collected as a part of Project Talent in their analysis. They also distinguished between only two curricular tracks, academic and nonacademic. Although this limits the usefulness of their results, their study is an excellent one in other respects. Jencks and Brown's predictor variables included measures of background, aptitude and achievement, grades, and educational plans, and they measured educational attainment as years of postsecondary schooling completed. They found that students who are average in background, aptitude, grades, and educational plans would complete 14.4 years of schooling if they followed an academic program in high school, and 14.0 years of schooling if they followed a nonacademic program. The z-score equivalents are 0.13 for academic students and -0.13 for nonacademic students. The difference of 0.26 standard deviations is a small difference.
Vanfossen et al. (1987) used HSB data to estimate curricular effects on educational attainment. They examined effects of academic, general, and vocational programs separately, but they restricted the sample for their analysis to stayers, students who did not change curricular programs between their sophomore and senior years. In addition to curricular program, predictors in their regression equations included measures of scholastic aptitude, socioeconomic background, aspirations, and so on. They found that average students would be 0.28 standard deviations above the mean on a scale of educational attainment if they followed academic programs in high school. They would score 0.17 standard deviations below the mean if they were in general programs and 0.27 standard deviations below the mean if they were in vocational programs. The difference in educational attainment between comparable students in academic and nonacademic programs is therefore 0.48 standard deviations.
Conclusions from analyses with continuous outcome variables. The overall picture is similar to the one that emerged from studies that used dichotomous outcome variables. Like those studies, these show that students from academic and nonacademic programs complete different amounts of schooling. Controlling for factors such as ability, background, and aspirations of the students does not equalize attainment outcomes. The typical student would complete about 14 years of schooling if enrolled in an academic program and about 12.5 years if enrolled in a nonacademic program. The attainment gap would be somewhat less if postsecondary vocational training were counted as postsecondary schooling, but even with this broader criterion of educational attainment, there would still be an attainment gap between academic and nonacademic students.
It is impossible to attribute this attainment gap to curricular programs, however. As I have already pointed out, academic and nonacademic students differ greatly in their educational goals at the start of high school, and it is extremely unlikely that their goals are measured reliably or adequately in survey studies. Without good measures of educational aspirations, regression analyses cannot control adequately for existing differences in goals. The results of these regression comparisons of academic and nonacademic students should therefore be regarded with extreme suspicion.
It is more relevant to compare the educational attainments of students from general and vocational programs. Students who choose these two types of programs are similar in aptitude and background, and they do not differ radically in their educational aspirations. Regression results also suggest that the two groups are similar in educational attainments. Average students would complete about three months of college if they graduated from a vocational program in high school and about five months if they graduated from a general program.
The educational attainments of academic and vocational students are clearly different. About 75 percent of students from academic programs and about 20 percent of students from vocational programs enter college. The average student from a high school academic program completes about two years of college; the average student from a vocational program completes only a few months. The difference between academic and vocational students is therefore substantial, and regression results suggest that even if similar students enrolled in academic and vocational programs, they would still differ in how much schooling they complete.
These regression comparisons are inadequate, however, and their results are misleading. It is important to note that the groups being compared differ profoundly in educational aspirations at the start of high school. Students usually follow college-prep programs because they intend to go to college; students often follow vocational programs because they do not intend to go to college. The goals of the two groups seem almost by definition to be non-overlapping. Statistical manipulation can never equate the aspirations of groups with such fundamentally different goals, and so regression comparisons of educational attainments of academic and nonacademic students must be treated with skepticism.
Comparing postsecondary attainments of academic and vocational students also seems to be beside the point. Why compare the number of college enrollments from college-prep programs and from programs not designed to prepare students for college? It seems the equivalent of comparing medical school enrollments from pre-med and pre-law programs, or graduate school enrollments from four-year colleges and postsecondary training institutes. The criterion fits one of the programs being compared but not the other. It makes better sense to compare educational programs by examining their achievement of shared objectives.
Regression comparisons of vocational and general students are less problematic. The educational aspirations of students in vocational and general programs are not totally different, and the students who select the two programs are similar in many respects. Regression results also show that the educational attainments of students who complete vocational and general programs are similar. About 20 percent of vocational students and about 30 percent of general students enroll in college after high school. The average student from a vocational program completes several months of college, and the average student from a general program completes only a few months more. Regression results suggest that these difference in college attendance cannot be explained entirely by differences in background, aptitude, and aspirations of vocational and general students. If the same students took vocational and general programs in high school, they would still differ slightly in likelihood of college enrollment.The results change, however, when we include all of postsecondary education in the picture. General and vocational students differ only slightly in likelihood of pursuing some form of postsecondary education or training. About 50 percent of general students and 45 percent of vocational students pursue postsecondary education of some form. Regression results suggest that even this small difference is probably due to group differences in background, aspirations, or aptitude. If the same students took vocational and general programs in high school, they would not differ at all in their likelihood of involvement in some form of postsecondary education.
Participation in vocational programs thus seems to have a small effect on college enrollment but no effect on the likelihood that a student will pursue some form of postsecondary education or training. General students are slightly more likely than comparable vocational students to enroll in college, but they are no more likely than vocational students to seek some form of postsecondary education or training. With a greater emphasis on vocational education in high school, we would expect a slight decline in the number of students going on to college but no decline in the number of students involved in postsecondary education. Some policy makers might find the decline in college attendance unsettling, but other would not find it troubling. They might argue that vocational programs are not meant to increase the likelihood of college enrollment and that their success should be measured by other standards.