Archived InformationThe Quality of Vocational Education, June 1998
Although the proportion of young people who finish high school has increased substantially during the past half century, many people think that today's dropout rate is still far too high. One reason for their concern is the total number of high school dropouts. Although 85 percent of young adults have been completing high school recently, compared to less than 40 percent in 1940 (Rumberger, 1987), still nearly 500,000 students drop out annually (Peng & Takai, 1983). The staggering number of dropouts places a great burden on social welfare programs for housing, health care, and employment, and the over-representation of minorities among the dropouts is a sad reminder that this country still has a long way to go to achieve equity among all its citizens.
Proponents of vocational education have long believed that vocational programs can help prevent high school dropout. Mertens, Seitz, and Cox (1982) cite an early instance of concern about high school dropouts:
"By 1823, two years after the opening of the first publicly supported high school in America, 76 of the entering class of 176 had dropped out. The School Committee of the City of Boston, in response to this high attrition, recommended that the most useful and practical subjects should be offered in the first year . . . The use of occupationally relevant instruction to prevent high school dropout had begun. Unfortunately, how well this instructional change was implemented, or if the change helped to keep students in school, was never documented. Things have not changed a great deal in the past 150 years. Relevant education is repeatedly urged as a way to reach and retain poorly performing youth in school, but the utility of this approach has yet to be clearly demonstrated" (1982, p. 1).
Nor have things changed much in the past decade. In their study of dropouts, Mertens et al. (1982) found statistical support for the proposition that vocational programs keep students in school, but the size of the effect was too small to be of any practical significance. Catterall and Stern (1986) also judged their findings on dropouts from vocational programs to be inconclusive. Findings seemed to vary with the method used to identify vocational students and with the control variables used in an analysis. Statistics cited by the General Accounting Office (1986) and the National Center for Education Statistics (1985) seem to show no dropout-preventing effects for vocational education, but reviews by Lotto (1986) and Weber (1988) suggest that vocational programs may help keep young people in schools.
The whole area of curricular effects on high school completion is clearly worthy of more careful study. My purpose in this chapter is to give the existing research the attention it deserves. My plan is first to examine overall dropout rates by curricular program. I then examine studies that use statistical or experimental control to compare dropout probabilities for vocational and nonvocational students who are similar in background, aptitudes, and other factors.
I located dropout studies 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 dropouts. I located 99 documents that included these terms in citations or abstracts. I reviewed all the documents either in full or in abstract form, but found that relatively few described studies that were suitable for review in this chapter. 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 data-base searching and branching, I located a total of 8 usable studies (table 3.1). Each of the studies was a longitudinal, quantitative study that examined dropout data in a national or regional survey. Although the studies were thus uniform in several important respects, they also differed from one another in some key features. The most important of these were the ways in which they (a) identified vocational students, (b) formed comparison groups, and (c) analyzed the data.
Identification of vocational students. Most of the researchers relied on student self-categorization to identify vocational students. Peng and Takai (1983) used self-categorizations that students made during their sophomore year in high school. Coombs and Cooley (1968) relied on categorizations that students made in both 9th-grade and 12th-grade. Weber (1988) used 12th-grade self-categorizations, and Grasso and Shea (1979) relied on student self-reports made during follow-up surveys.
Other researchers identified vocational students from the courses that they took in high school. Catterall and Stern (1986) classified students on the basis of course work before the junior year and then examined dropout rates during junior and senior years. Mertens et al. (1982) noted the number of vocational courses that students took in grades 9, 10, and 11, and they tried to determine whether the number of such courses taken in a given year predicted dropping out during the succeeding year. Wagner (1991) used a similar approach. Perlmutter (1982) classified students as vocational or nonvocational at the beginning of high school from school records.
Comparison groups. The simplest solution to the problem of comparisons is to list the dropout rates for academic, general, and vocational programs. Peng and Takai (1982) and Weber (1988) used this approach, and Catterall and Stern (1986) contrasted vocational students with all other students in the population. Other researchers compared dropout rates for vocational students to dropout rates of other restricted populations. Coombs and Cooley (1968), for example, compared dropout rates for vocational students to dropout rates for other students who did not go on to college. Grasso and Shea (1979) concluded that the best comparison was between dropout rates for vocational students and dropout rates for students in general high school programs. Mertens et al. (1982) compared the dropout rate for vocational students to the dropout rate for other high-risk students. Perlmutter (1982) formed a comparison group that matched her vocational students in aptitude and elementary school origins. Wagner (1991) examined dropout rates for vocational and other students in the special education population, a population that contains very few academic students.
Method of analysis. Reports by Catterall and Stern (1986), Coombs and Cooley (1968), Peng and Takai (1983), Perlmutter (1982), and Weber (1988) list high school dropout rates by curricular program in simple tables. Some of the tables present two-way classifications of students, in which one basis of classification is curricular program and the other is high school completion vs. non-completion, but other tables provide three-way classifications. Other reports list results from more sophisticated statistical analyses. Grasso and Shea (1979), Mertens et al. (1982), and Wagner (1991) provide regression equations that relate high school completion vs. non-completion to curricular program and other student characteristics.
Vocational programs do not seem at first glance to encourage students to complete high school. Peng and Takai (1983) analyzed data from a nationally representative sample of approximately30,000 students who were part of the sophomore cohort of the High School and Beyond (HSB) survey. They classified the students into curricular groups based on their sophomore-year answers to a question about curricular program. They also classified students as high school completers and dropouts on the basis of the answers on a follow-up survey conducted two years after the base survey. Vocational students appeared to have the highest dropout rate:
Vocational curriculum - 15 percent
General curriculum - 13 percent
Academic curriculum - 4 percent
Weber (1988) showed, however, that the dropout picture changes with the method that is used to classify student programs. He first noted that most high school sophomores have only limited experience with vocational courses, and he suggested therefore that it was inappropriate to use sophomore-year statements to classify students into curricular tracks, as Peng and Takai had done. Weber reasoned that the answers that high school sophomores give to questions about their curricular programs might reflect their aspirations rather than their experiences. He therefore classified HSB students into academic, vocational, and general programs based on their senior-year answers on a 1982 follow-up questionnaire. For a sample of the students in the HSB database, he calculated the correlation coefficient between number of courses taken and program identification, and he found that classification by answers to the 1982 question was more consistent with actual course taking than was classification by answer to the 1980 question. The correlation between the 1982 answer and course taking was .37; between 1980 answer and course taking, it was .24.
What is more, Weber noted that dropout findings change when students are classified into tracks by their senior-year rather than sophomore-year answers. When track classifications were based on student self-reports provided later in high school, the vocational group had a lower dropout rate than did the general group:
Vocational curriculum - 16 percent
General curriculum - 21 percent
Academic curriculum - 3 percent
Coombs and Cooley's (1968) report supports Weber's analysis. The data used by Coombs and Cooley came from Project Talent, a survey of more than 440,000 students attending public and private schools throughout the country. Project Talent researchers tested the students as 9th graders in 1960 and then sent a follow-up questionnaire to the students in 1964. Coombs and Cooley examined records of all the known dropouts in the original 9th-grade sample, and they also examined records for a random subsample of 25 percent of the male and 20 percent of the female graduates who did not continue their education beyond high school.
Coombs and Cooley presented their findings in an idiosyncratic way. They reported the percentage of dropouts who had reported in the 9th grade that they were either enrolled or expecting to enroll in general, college-preparatory, and vocational curricula, and they also reported the percentage of high school completers who had reported in the 9th grade that they expected to be in these curricular programs. In addition, they reported the percentage of the dropouts and completers who answered on the one-year follow-up questionnaire that they had been in the general, college-preparatory, and vocational programs.
It takes only a few simple computations to calculate more familiar dropout rates for vocational and other students from their figures. Dropout rates by program when programs are determined from students' 9th-grade self-categorizations are:
Students in vocational curriculum - 21 percent
Non-college-bound in other curricula - 15 percent
When a student's program is determined from the response on a follow-up questionnaire filled out after the student left school, dropout rates by program are:
Students in vocational curriculum - 11 percent
Non-college-bound in other curricula - 24 percent
Dropout rates from vocational programs thus seem to be high when curricular programs are determined from self-reports made early in high school; dropout rates seem lower when curricular programs are determined from self-reports made at the end of high school
Catterall and Stern's (1986) study provides additional important data on the relationship of curricular program to dropout rate. These researchers also used HSB data to determine the effects of vocational programs on dropout rates, but they categorized students as vocational and nonvocational from the pattern of courses the students had taken during their freshman and sophomore years in high school (Catterall & Stern, 1986, table I, p. 79). The researchers also classified students by the students' own estimates of the likelihood that they would complete high school. I calculated from their classification table the following dropout rates for vocational participants and non-participants:
Vocational participants - 10 percent
Non-participants - 12 percent
In a separate analysis, Catterall and Stern (1986, table II, p. 81) classified students as vocational concentrators or non-concentrators based on the number of vocational courses they took in a single area before their junior year. They also classified students by the type of school that they attended: Some students were in schools with many dropouts, other students were in schools with few dropouts. From their classification table, I calculated the overall dropout rates for vocational concentrators and non-concentrators:
Vocational concentrators - 5 percent
Non-concentrators - 12 percent
The results are clear. If we identify vocational students from the courses they take, we will find a definite relation between vocational education and high school dropout. Students who pursue a vocational program are less likely than other students to drop out of high school.
Table 3.2 lists results from the reports of Catterall and Stern (1986), Coombs and Cooley (1968), Peng and Takai (1983), and Weber (1988). The table shows that the conclusions we draw about dropout rates for vocational students depend on how we identify vocational students. It apparently matters a great deal whether we let vocational students identify themselves or whether we identify vocational students on the basis of the courses they take. It also matters a great deal whether we use student self-identifications made early or late in high school. Student self-categorizations made early in high school seem to reflect aspirations more than experience, and so they may not be very useful in gauging actual program effects. Student self-categorizations made later in high school are more likely to reflect the experiences that students have in vocational programs, and so they may be more useful. It appears, however, that the courses that students take provide the best indicator of their contact with vocational education. If that is so, the dropout rate for students taking vocational courses is lower than the dropout rate for other high school students.
Although these simple descriptive analyses suggest that vocational programs help keep students in school, the analyses are not conclusive. They do not take into account other factors that influence student decisions to drop out of or to stay in school. We know that factors such as socioeconomic status, school aptitude, race, and gender play a role in dropout decisions, and we also know that vocational students and other students are not equal in these factors. If vocational students were equivalent to other students in background characteristics, would they be more or less likely to drop out? What would analyses with better experimental or statistical controls show?
Perlmutter (1982) made an effort to control for factors such as academic ability and school background through the experimental design of her study. She collected survey data from three matched groups of students on their entry into high school in New York City in 1976. The three members of each triplet differed in high school programs, but came from the same intermediate or junior high school, studied at the same level in mathematics, and had similar reading scores. One member of each triplet was admitted into a vocational high school, one member was denied admission, and one member did not apply for admission to a vocational high school. Perlmutter studied the school records of the three groups of students during the three-year period after their entry into high school.
She found a small difference in dropout rates for students in the three matched groups. During the five terms of the study, 16 of the 99 students in the vocational high school dropped out, as opposed to 18 of the 99 students who did not apply for entry and 24 of the 99 who applied and were not admitted. The effect of vocational education on high school dropout, however, may actually be larger than these figures suggest. Perlmutter also found that many of those who did not apply for admission and many of those who were not admitted to vocational schools ultimately took vocational courses in high school. Of the 178 in the three groups who took vocational courses or attended vocational school, 21 (or 12 percent) dropped out. Of the 119 who took no vocational courses, 37 (or 31 percent) dropped out. The difference in dropout rates is a significant one.
Three investigations used regression analysis to examine effects of vocational programs on dropout rates. Grasso and Shea (1979) applied regression techniques to data collected for the National Longitudinal Surveys of Labor Market Experience (NLSLME); Mertens et al. (1982) used regression with data from the New Youth Cohort of the National Longitudinal Survey of Labor Force Behavior (NLS-Youth); and Wagner (1991) used regression with data from the National Longitudinal Transition Study of Special Education Students (NLTS). Their analyses were meant to overcome some of the problems of simple descriptive analyses. With regression analysis, it is possible to specify the effects of curricular programs on students who are comparable in relevant antecedent factors.
Grasso and Shea's (1979) regression equations predicted likelihood of high school dropout from scholastic aptitude, socioeconomic origin, area of residence, and most recent high school curriculum. Their equations therefore provide a basis for comparing the effects of vocational and other programs on dropout rates when all other influences are held constant. Grasso and Shea especially emphasize comparisons of vocational vs. general programs. They report that students in vocational and general programs are similar enough in background characteristics for comparisons of the two groups to be meaningful. Grasso and Shea report that results from comparisons of academic vs. general students are far more difficult to interpret because the two groups are so different in background characteristics.
Grasso and Shea calculated regression weights separately by gender and race and by commercial and other vocational programs. They found that, compared to a general program, a commercial program reduced dropout by 6 percent for white men, but increased dropouts by 7 percent for black men; other vocational programs reduced dropout by 1 percent for white men and by 7 percent for black men. Commercial programs reduced dropout by 9 percent for white women and by 3 percent for black women; other vocational programs reduced dropout by 8 percent for white women and by 4 percent for black women. The overall weighted effect of vocational education in Grasso and Shea's regression analysis was to reduce dropout by 5 percent.
Wagner (1988) based her regression analysis on data from the National Longitudinal Transition Study of Special Education Students (NLTS), a study of a nationally representative sample of more than 8000 students aged 13 to 21 who were in special education in the 1985-86 school year. NLTS data came from telephone interviews with parents in 1987 and also from school records and a survey of educators in the schools attended by the students. Wagner restricted her sample to students with disabilities attending regular secondary schools.
She carried out three regression analyses relating vocational education to school performance. Dependent variables in the analyses were (a) the number of days absent from school, (b) failure in a course, and (c) dropout from school. Each of the analyses included a dichotomous variable measuring whether students had been enrolled in occupationally oriented vocational education in their most recent school year. Wagner hypothesized that students who had been enrolled in vocational education would be more likely to attend school regularly, thereby having fewer days absent, and would be less likely to fail courses or to drop out of school.She found that students taking occupationally oriented vocational education were significantly less likely to drop out of school when other factors were controlled. The likelihood of dropping out of school was 2.7 percentage points lower for vocational students. Vocational students were also about 3 percentage points less likely than others to have failed a course, and they were also absent from school less often. Wagner noted that these effects on course failure and absences had to be taken into account when interpreting her results.
Wagner pointed out that her regression equations included absenteeism and course failure as predictor variables. The 2.7 percent difference in dropout rates was for vocational and nonvocational students who were similar in absenteeism and course failure. The reduction would be greater for comparable students in vocational and other programs whose absentee and failure rates were typical for their programs. Wagner estimated that a student with a typical background would have an 8 percent chance of dropping out in a vocational education program and a 15 percent likelihood of dropping out (as well as a greater chance of failure and more absences) in a nonvocational program. Wagner concluded: "Hence, vocational education appears to offer the potential for a significant benefit to students with disabilities in terms of their school performance and school completion." (p. 24).
Mertens et al. (1982) also used regression analysis to determine whether curricular programs influence dropout rate. Their study employed data from the NLS-Youth sample, a nationally representative sample of approximately 13,000 young people who were 14 to 21 years old in 1978. Mertens and her colleagues analyzed data from 1979 and 1980 interviews as well as from high school transcripts. The researchers first divided the sample into two groups: a group with a high probability of dropping out of high school and a group with a lower probability. They used the group with the high probability of dropping out in their regression analyses. The dependent variable in the analyses was probability of dropout; independent variables came from five categories, representing both high school experiences and individual, family, contextual, and school characteristics.
Mertens and her colleagues reported that the more vocational courses students took, the less likely they were to drop out of school in a succeeding year. They found a significant relationship between dropping out and the number of courses taken in a preceding year in grades 10 and 12, but not in grade 11. The size of the coefficients, however, was very small in all years. The effect of obtaining one vocational credit in grade 9 was to reduce the probability of dropping out of grade 10 by one tenth of a percentage point. One vocational credit in grade 11 was associated with a .02 percent reduction in the probability of dropping out of grade 12.
There is a methodological problem in this analysis, however, that should not be ignored. Like Wagner, Mertens and her colleagues used both background factors and outcome variables as predictors in their regression equations. Unlike Wagner, however, Mertens and her colleagues did not estimate the effects of vocational education from background factors alone. Their regression coefficients describe the effect of curriculum on students who are identical in both background factors and such outcomes as absence from school, GPA, aspiration for further education, and so on. The regression coefficients do not therefore indicate the importance of curriculum for students who are similar in background characteristics only . They do not therefore serve as an index of the importance of curriculum.
It is worth noting that Bishop (1988) has given a different account of the findings of Mertens and her colleagues. Bishop reported that completion of one vocational education course in the 9th grade reduces the dropout rate in the 10th grade from 9 percent to 6 percent. In addition, completion of one vocational education course in the 9th, 10th, and 11th grades lowers the dropout rate in the 12th grade from 20 percent to 14 percent. Bishop reported that the completion of a single vocational education course in grades 9 through 11 would therefore raise the high school graduation rate from 64 percent to 70 percent. Completion of two courses would raise the graduation rate to 76 percent. Bishop cites Mertens et al. (1982) as the source for his statistics, but he does not point out that his statistics are strikingly different from those given by Mertens and her colleagues. Nor does he explain how he derived his figures from the Mertens et al. (1982) report. My own conclusion is that at present neither the figures cited by Mertens and her colleagues nor those cited by Bishop can be used as estimates of program effects on dropout rates in the NLS-Youth sample. Both sets of statistics raise more questions than they answer. Grasso and Shea, Perlmutter, and Wagner provide more useful findings on vocational education and high school dropout (table 3.3).
The evidence shows that vocational programs help keep students in high school. Without vocational programs, more at-risk students would drop out of school each year than currently do. With vocational education, more students complete high school. The evidence for these propositions comes from both simple descriptive classifications of students and regression analyses. Results of the simple descriptive analyses are suggestive. Results of the regression analyses are more nearly conclusive.
The most compelling of the descriptive analyses is Catterall and Stern's (1986). These investigators found a dropout rate of 10 percent for HSB students who took vocational courses before their junior year, as opposed to a dropout rate of 12 percent for those who did not take vocational courses. Making the reasonable assumptions that about 40 percent of those who did not take vocational courses were in academic programs and that the dropout rate from academic programs was about 4 percent, we can deduce that the dropout rate for nonacademic students who did not take vocational courses was 20 percent. Catterall and Stern also found a dropout rate of 5 percent for students who took one year of vocational courses in a single area before their junior year, as opposed to a dropout rate of 12 percent for those who took fewer courses in a single area. Making reasonable assumptions about the dropout rates in academic programs and about the number of the non-concentrators who were in academic programs, we are again left with the deduction that the dropout rate for nonacademic students who did not concentrate in a vocational area was about 20 percent. These findings suggest that students who are not going on to college are far more likely to complete school if they become involved in vocational programs. Although Catterall and Stern restricted their analyses to HSB students who lived in California, there is little reason to think that their results apply exclusively to one geographical region. Although other factors might contribute to these differences in dropout rates, the most likely explanation for the low dropout rate from vocational programs is that a vocational curriculum appeals to students who are not college-bound. Lotto (1986) has pointed out that vocational courses are usually among those that students like best in high school. It seems likely therefore that high-risk students stick with vocational education because they find it interesting, relevant, and rewarding. They drop out of general programs because they find them less interesting, relevant, and rewarding.
Regression analyses firm up these conclusions. It is unfortunate indeed that no researchers have used regression analysis with the HSB data set to determine the effects of vocational programs on high school dropout and that the regression analyses of NLS-Youth data have produced unclear and confusing results. Fortunately, however, researchers have carried out adequate regression analyses of NLTS and NLS-LME data. Their studies show a difference in dropout rates of 6 percent for non-college bound youngsters in vocational and general programs when pre-existing characteristics are controlled. Perlmutter's (1982) study suggests that in high-dropout areas such as New York City, the dropout-preventing effect of vocational education may be even greater.
A difference of 6 percent in dropout rates for non-college bound youngsters in vocational and other programs may not seem large, but it is nonetheless an important effect. About 450,000 students in nonacademic programs drop out of high school each year. If vocational education were not an option for high school students, the number of dropouts would undoubtedly be higher. The dropout rate for youngsters currently in vocational programs might go up by 6 percent (from 8 percent to 14 percent) if these youngsters had to pursue other programs in high school. The total number of dropout from those not in college-prep programs would increase from 450,000 to 500,000. An effect of this magnitude would be too large for policy makers to ignore.