IV. WHO TAKES VOCATIONAL EDUCATION?
FINDINGS FROM STUDENTS' TRANSCRIPTS

• VOCATIONAL COURSETAKING
• FACTORS EXPLAINING VOCATIONAL CONCENTRATION
• CONCLUSIONS

Clearly, there were major differences among our three schools and the students who attended them. Yet, as we noted in Sec. II, the data regarding student achievement outcomes suggest that none of the three schools appeared to make much headway with their students, in terms of either improving their overall achievement standing relative to other high school students across the nation, altering the achievement disparities among groups of students within them, or promoting the access of Latino and African American students to either college or postsecondary vocational training. The similarities among the schools' cultures regarding curriculum offerings and student assignment processes described in Sec. III begin to suggest why this is the case.

In this section and the next, we examine actual patterns of student coursetaking--patterns that reflect the decisionmaking dynamics at schools and help produce students' achievement outcomes. Here, we describe the vocational coursetaking of students at Coolidge, Washington, and McKinley and show how particular student characteristics predict different patterns of vocational coursetaking at the three schools. In Sec. V we examine students' participation in academic courses. These analyses enable us to explore the consequences of the curriculum opportunities and placement routines at the three schools and to explore the usefulness of various theories offered to explain them.

As we detail in this section and the next, the decisionmaking processes produced different placement and coursetaking patterns at each school and for groups of students within each school; these patterns resulted in a sorting of students with different background characteristics into different courses and programs. But there is considerable evidence that the schools tried to sort students according to their intellectual capacity and that these judgments appear to be informed largely by students' prior achievement. Much of the racial variation in course placements, in fact, can be "explained" by students' scores on achievement tests. But the match is not perfect, and some discrepancies relate quite clearly to race and social class. At the same time, the ability of schools to place students by either meritocratic criteria or on the basis of assumptions related to race and social class seems to have been limited. We find considerable sloppiness in both patterns, both between our schools and within them.

VOCATIONAL COURSETAKING

Like most comprehensive high schools, Coolidge, Washington, and McKinley offered a range of vocational courses, although at all three schools vocational courses constituted a relatively small percentage of the total curriculum.[27] Each school offered vocational courses in a variety of general or non-occupational subjects such as cooking, parenting, or introductory typing, as well as training in specific occupational skills. (Appendix B lists the vocational courses offered at our three schools in two broad categories: introductory or non-occupationally specific and occupationally specific.) Most vocational courses were offered as part of each school's regular curriculum, but students at Coolidge, Washington, and McKinley also had access to a wide variety of courses offered as part of a regional program. These regionally sponsored courses, generally speaking, provided more training in specific job skills than did the "regular" vocational courses. Yet, although the occupational orientation of these courses made them attractive to many students, most were offered at off-campus centers some distance from the high school, which constituted a formidable obstacle to attendance for some students.[28] However, a smaller number of courses offered under the auspices of these regional programs were held on campus.

In looking at the vocational coursetaking of students at Coolidge, Washington, and McKinley, we identify variation in the rates of participation for students among the schools, for different groups of students within the same school, and for similar groups attending different schools. We also consider how participation rates differed across the types of vocational courses offered, among courses that are held on and off campus, and among those offered as part of regional programs and those that are a part of the high school's curriculum.[29]

Most Students Take Some Vocational Education

Hoachlander and his colleagues observed that an extremely high proportion of high school students (90 to 97 percent) take at least one vocational course during their high school career (Hoachlander and Choy, 1986; Hoachlander, Brown, and Tuma, 1987). Table 4.1 indicates that this holds true for students at our three schools, despite the fact that only Washington requires that students take any vocational education courses for graduation.[30]

At Coolidge, 89.7 percent of students who attended 10th through 12th grade took at least one vocational course, all students at Washington, and 99.4 percent of those at McKinley took one or more vocational courses.

Students at all three schools participated much more heavily in non-occupationally specific vocational courses, such as home economics or introductory typing, than in those designed to teach job-specific skills.[31] This pattern is consistent with the fact that students took many fewer vocational courses held off campus at regional centers than on campus. The off-campus courses were generally much more focused on occupationally specific skills, such as airframe repair, than are most of those held on campus.[32]

Yet, we also found important between-school differences. McKinley students participated at significantly higher rates in both occupational courses and off-campus vocational courses than did students at Coolidge or Washington. More than three-quarters of all McKinley students took at least one occupational course during their high school career, whereas about half the students at the two other schools did so. Furthermore, despite discouraging actions by the school administration, almost two-thirds of the McKinley students in our sample took an off-campus vocational course, a participation rate nearly twice that of Coolidge students and nearly two and a half times the rate for Washington students.

Upon close examination, other significant between-school differences emerge with respect to the number of vocational courses and credits taken, the distribution of those courses over a student's high school career, the type of courses taken, and the demographic and academic characteristics of students in those courses.

McKinley Students Took the Most Vocational Education

Figure 4.1 shows the distribution of vocational courses taken by students at the three schools. As was also evident from Table 4.1, all Washington students, nearly all McKinley students, and 90 percent of Coolidge students took at least one vocational course during their high school career.

Yet, at the other end of the coursetaking spectrum, significant differences emerge. A majority of McKinley students (69.1 percent) took five or more vocational courses. Yet less than half of the students at the other two schools (49.1 percent at Washington and 40.8 percent at Coolidge) took five or more courses.[33]

The numbers at Washington High are not perfectly comparable with those at Coolidge or McKinley, since Washington required students to take two vocational courses. Moreover, among those courses meeting Washington's vocational requirement (and included in Table 4.1 and Fig. 4.1) were rigorous computer science courses that the school also classified as mathematics or science courses.

NOTE: The distribution between schools are significantly different at the .01 level.

Fig. 4.1--Distribution of Vocational Courses Taken, by School

Figure 4.2 depicts the vocational coursetaking patterns of students at the three schools when we subtract the two required courses from the number of courses taken by Washington students.[34]

We must be cautious about these comparisons as well, since it is likely that at least some of Washington's students would have taken additional courses, even without the requirement. Nevertheless, this figure graphically displays the difference in the vocational coursetaking pattern at low-income, all-minority McKinley, and at the two schools enrolling large numbers of white and middle-class students.

McKinley Students Sought Job Training; Coolidge and Washington Students "Dabbled"

The connection between school demographics and vocational coursetaking becomes even clearer when we examine differences in the types of vocational courses students at the three schools took. Table 4.2 displays student coursetaking patterns by vocational course type.[35] McKinley students took significantly more occupational courses than their Washington and

NOTE: The distribution between schools are significantly different at the .01 level.

Fig. 4.2--Distribution of Vocational Courses Taken, by School
(Less Two Required Courses for Each Washington Student)

Coolidge counterparts, and they took most of those courses through their affiliated regional centers. Figure 4.3 graphically displays McKinley's much higher rate of participation in occupational courses.[36]

Many regional courses were held off campus and during regular school hours, factors that presented a significant obstacle for students' participation at each of the schools. McKinley's principal strongly opposed student attendance at these classes, both because he believed high school should prepare students for college rather than for jobs and because he feared that allowing students to leave campus during the day would make McKinley more vulnerable to gang activity. These constraints made attendance at regional program classes even more difficult for low-income, minority McKinley students, and thus makes their relatively high rate of attendance even more significant. Further, participation in off-campus courses divides our two schools with significant white and middle-class populations. Coolidge with its far more racially and socioeconomically diverse student body evidenced greater regional vocational coursetaking than did more homogeneous, middle-class white and Asian Washington. And, it is probably safe to speculate that the difference between the two schools would be even larger without Washington's practical arts requirement.

Table 4.2 also shows significant school differences in the types of occupational courses students took. Washington and Coolidge students gravitated overwhelmingly toward business courses; those at McKinley were somewhat more likely to take trade courses than business courses. Nearly twice the percentage of McKinley students took trade-related occupational courses than at the other two schools.

Fig. 4.3--Distribution of Occupational Courses Taken, by School

These patterns does not hold, however, when we examine participation in non-occupationally specific vocational courses. Substantial percentages of students at all three schools took non-occupational or introductory courses in consumer education/home economics, business, and industrial arts. Here, Coolidge differs from the other schools, in that a larger percentage of students took industrial arts and a smaller percentage took business classes.[37]

In addition to its vocational course offerings, each case study school offered credit to students for "work experience," i.e., employment outside of school. Although students could get high school credit for a variety of jobs, most held minimum-wage, service positions that offered neither skill training nor advancement possibilities. As a result, the school administrators and counselors with whom we spoke offered generally negative assessments of the value of work experience programs in preparing students for the workforce. Students at McKinley participated much more heavily in their school work experience program than did Washington and Coolidge students, with almost no participation at Coolidge. However, at Coolidge, similar credit was given to a substantial percentage of students who performed routine clerical services at the school as "aides."

Washington Students "Got It Out of the Way" Early

The patterns we observed in the type of vocational courses that students at our three schools took are consistent with differences in the timing of vocational coursetaking. Table 4.3 displays vocational participation.

At Washington, the timing of vocational participation may be driven by that school's two-semester requirement. Most students either took those courses early, during their freshman year, or waited until they were close to graduation. More than half of the students in our sample took at least one vocational course during their freshman year, more than twice the rate of 9th grade participation at McKinley, and significantly higher than at Coolidge as well. As we noted above, 93 percent of these students attended Washington during their freshman year and from 10th through 12th grade. Another sizeable chunk of students in the Washington sample (close to 35 percent) took vocational courses in their senior year. Moreover, the mean number of courses Washington students took during their freshman and senior years is the same (1.6 courses), close to the two-semester requirement.

The coursetaking patterns of students at Coolidge and McKinley, however, are quite different from those at Washington. At Coolidge, no clear trend in the timing of vocational coursetaking emerges; students took a vocational course during their sophomore year only somewhat less frequently than during the other three years. At McKinley, the rate of vocational participation was lowest during 9th grade and highest during 10th and 11th grades, dropping off somewhat during the 12th grade.

The mean numbers of courses taken by 10th, 11th, and 12th graders at McKinley (1.5, 2.0, and 2.3, respectively) are the highest in our sample. During these three years, then, a greater proportion of McKinley students (except for the equivalent proportion of Washington seniors) took vocational courses, and those who did took more than vocational coursetakers at Coolidge or Washington. Moreover, the 11th and 12th grade means at McKinley are significantly higher than are the means for 9th grade Washington students, even though the rate of participation was extremely high for these 9th graders.

The greater vocational coursetaking at McKinley, however, should not mask a general pattern at all three schools--the increasing mean number of courses taken in grades 10, 11, and 12. This increase at all three schools may indicate that as graduation approaches many students may view college entrance as increasingly unrealistic, because of either academic or financial deficiencies, and turn to vocational training courses as an alternative. And, it would not be surprising that such a shift would occur more often at a school enrolling low-income, minority students.

Asian Students Take the Least, Latino and White Students Take More

The transcripts also reveal significant differences at each school in the nature and extent of vocational coursetaking by students with different demographic characteristics and from different academic tracks. Table 4.4 displays the mean number of vocational courses taken by students according to their sex, race, and academic-track participation.

Gender differences in a number of vocational courses taken were significant only at Washington High. However, we found significant racial and ethnic group differences at both Coolidge and Washington.

At these two schools, Asian students took the fewest vocational courses and whites and Latinos took the most.[38]

But College-Bound Students Take the Least

Table 4.4 also compares the number of vocational courses taken by students at each school who we defined to be in a college-preparatory curriculum--those who were enrolled in Algebra 2 or "college-prep" English during their junior year--with that taken by students in a non-college-prep course of study. Not surprisingly, college-bound students at all three schools took significantly fewer vocational courses than did non-college-bound students. However, Washington's practical arts requirement and the ability of advanced computer science courses to meet that requirement make us cautious about between-school comparisons, since these factors probably both increase vocational coursetaking and alter its character for many Washington students. With these required courses subtracted from Washington's mean, we again find a pattern related to the student composition of the schools: As the percentage of white and affluent students at a school increases, participation in vocational education by college-prep students decreases. This finding suggests a modification to our speculation above that McKinley students turn toward vocational education as college attendance becomes more remote. This is still likely to be the case, even for those enrolled in college-preparatory programs, but it may also be that low-income, minority students who intend to go to college have less confidence than their more-advantaged peers in their chances of getting through postsecondary schooling without some episodes of full-time work, and therefore may place greater value on vocational training.

Generally speaking, college-prep students at all schools participate less frequently in occupationally oriented courses, particularly those offered as a part of regional occupational programs (see Tables 4.5 and 4.6).

In contrast to this overall pattern, however, participation in occupational business courses is roughly comparable for both groups of students at all schools, and business courses generally attract more college-preparatory students than any other types of vocational courses. Again, with the above cautions about between-school comparisons in mind, we find that participation in non-occupational courses generally and non-occupational business courses in particular is quite comparable for both groups at Washington and McKinley but significantly lower for college-prep students at Coolidge. In fact, Coolidge stands out as the only school where overall participation in vocational courses generally, and in non-occupational and on-campus courses in particular, is significantly lower for college-prep students. Coolidge may be more typical of schools with large, middle-class, white populations than is Washington with its practical arts requirement, which eliminates any distinction between college-bound and non-college-bound students.

Table 4.7 compares the vocational coursetaking of students in the top 10 percent of their class with those who are not. Here too, Coolidge is the only school where the participation of the most academically talented students is significantly different from other students overall, and in non-occupational and on-campus vocational courses.

Yet the best Coolidge students participate at rates only slightly lower than other students in consumer/home economics courses, whereas differences between these groups is two to four times as great at Washington and McKinley.

At McKinley, students in the top 10 percent participate with other students at comparable rates in a number of areas including regional occupational courses, occupational and non-occupational business courses, work experience, and off-campus courses.

Taken together, these rates of participation of students from various academic and vocational tracks reveal some interesting patterns. The differences in the rate of vocational coursetaking are most often significant at Coolidge, with its large middle-class contingent where there is no vocational requirement. At Washington, the two-semester requirement appears to have mitigated strong differences in coursetaking behavior such as we observe at Coolidge. Such differences are also muted at McKinley, but for different reasons. At this lower-income, minority school, all students participate in vocational education at significantly higher rates than at Coolidge or Washington (see Tables 4.1 and 4.2). As our field study suggests, this may result both from the widespread view that vocational courses are more appropriate for low-income, minority students and from the school's tradition of offering a far larger number of vocational courses than the other schools. These factors, possibly in combination with students' lower confidence about college, may overwhelm or carry equal weight with McKinley administration's emphasis on college preparation.

FACTORS EXPLAINING VOCATIONAL CONCENTRATION

What factors determine or predict whether students take a little vocational education or a lot when other factors are taken into account? Do these factors vary by school? To investigate these questions, we classified the students at the three schools into two groups, "concentrators" and "non-concentrators," and examined the factors that determine whether a student is a vocational concentrator.[39] We defined concentrators as those students who took six or vocational courses at their high schools.[40] In addition, to account for the practical arts requirement at Washington, we have alternatively defined concentrators at that school as those students who took six or more courses beyond the two-course practical arts requirement.

Who Concentrates in Vocational Education?

Table 4.8 reiterates the differences across the three schools in the fraction of students who are vocational "concentrators." Using our definition, McKinley had the most students "concentrating" in vocational education (57 percent) and Washington, after accounting for the practical arts requirement, had the least (16 percent). Girls at Washington were less likely to be vocational concentrators than were boys, however, the size of this difference diminishes significantly once the practical arts requirement is accounted for. At Coolidge and McKinley, there is no significant difference in the fraction of girls and boys who are vocational concentrators.

Racial Differences. We found significant differences in the percentage of students who are concentrators by race and ethnicity at Washington and Coolidge. At both schools, Asians are far less likely than other groups of students to be vocational concentrators.[41] At Washington, white students were most likely to concentrate. This contrasts with Coolidge, where Latinos were the most likely to take a concentration of vocational classes. At diverse Coolidge, Latinos were more likely than whites, whites more likely than African Americans, and African Americans more likely than Asians to be vocational concentrators (38 percent, 32 percent, and 27 percent, respectively). The difference in Latino participation between Washington and Coolidge may be explained, in part, by differences in the socioeconomic status of Latinos at the two schools. One clue to the difference for Latinos at the two schools comes from our interviews. Several Washington respondents noted that the school draws from a relatively affluent neighborhood; less than 1 percent of the student body qualifies for Aid to Families with Dependent Children (AFDC). One Washington teacher told us that because of the cost of housing, "Latinos here are not like in East L.A . . . [they] are just like whites."

In contrast, as noted above, Coolidge High School draws from an economically diverse population with a large cohort of Latino immigrants. As we described in Sec. III, we encountered a widely held belief among Coolidge respondents that many of the Latino students at that school have poor basic skills (including severe English language deficiencies), poor motivation, and limited future expectations.[42]

In contrast, there is no significant difference in the fraction of African American and Latino students at McKinley who are vocational concentrators.[43]

Achievement Level Differences. The relationship between achievement test scores and vocational participation is shown in Figs. 4.4 and 4.5. The distribution of 10th grade math and reading achievement scores, measured by the 5th percentile, mean, and 95th percentile, is shown for vocational non-concentrators and concentrators.[44] As our field study would lead us to believe, at all three schools, average math and reading scores are significantly lower for vocational concentrators than for non-concentrators.[45] However, the range of scores as measured by the 5th and 95th percentiles shows that vocational concentrators are not exclusively low-achieving students, nor are vocational non-concentrators only high-achieving students.

Fig. 4.4--Distribution of Math Scores for Vocational
Non-Concentrators and Concentrators, by School

Fig. 4.5--Distribution of Reading Scores for Vocational
Non-Concentrators and Concentrators, by School

For all three schools, the 5th percentile in math achievement scores is higher for non-concentrators than for concentrators, whereas the 5th percentile in reading scores is very similar for the two groups. At the other end of the achievement spectrum, we find students with high math and reading scores in both the vocational concentrator and non-concentrator groups. Apparently, vocational concentrators, although they have lower math and reading achievement scores on average, are students with a wide range of ability as measured by their 10th grade reading and math scores. Likewise, there are students with low math ability and with the lowest reading ability who do not concentrate in vocational education.

Track Level Differences. In addition to these achievement differences in the rates of participation across different groups of students, there are important differences as well in the type of vocational courses that students in different academic tracks took. Table 4.9 compares the type of vocational courses that vocational concentrators took with those taken by non-concentrators.

Not only did more concentrators take occupational courses at each school than did non-concentrators, as expected, but the difference in the two group's rates of participation in occupational and off-campus courses is statistically significant at all three schools. At Coolidge and McKinley, but not at Washington (probably because of the two-course requirement), there are also significant overall differences in the participation of concentrators and non-concentrators in non-occupational and on-campus vocational courses.

Across nearly all occupational and non-occupational course types, concentrators participated at significantly higher rates than did non-concentrators. The notable exception is non-occupational business courses at Washington and McKinley, where concentrators and non-concentrators subscribed at nearly equal rates.

Explaining Who Concentrates in Vocational Education

The preceding analyses highlighted several differences within and between the three schools in the rates at which various groups of students were vocational concentrators. In what follows, we investigate the role of a number of student characteristics in the probability of being a vocational concentrator. For example, is the likelihood that a student takes a large number of vocational courses related to his or her academic performance as measured by achievement scores in math and reading? If we control for differences in test scores, are there still differences in the probability of being a vocational concentrator for students of different races and ethnic groups? Do differences exist across schools in the likelihood of being a vocational concentrator once we control for differences in student characteristics?

To address these questions, we conducted a logistic analysis predicting the probability that a student would be a vocational concentrator. Logistic models were estimated separately by school and with students pooled across all three schools. In both cases, the probability of being a vocational concentrator was modeled as a function of the student's gender, race/ethnicity, and achievement scores.[46] In addition, since the sample includes students who entered their respective schools in the 10th grade as well as those who entered in the 9th grade, we included a variable indicating that a student was at his or her respective school for four years. In this way, we control for the possibility that, since the four-year students had one more year to take vocational education, they may be more likely than three-year students to be vocational concentrators. In addition, we included our measure of SES in the model estimated for Coolidge, and an indicator for foreign-born students in the models estimated for Washington and McKinley.[47]

Using the logistic analysis, we present the predicted probability that a student with various characteristics will be a vocational concentrator in Tables 4.10 to 4.12.[48] Table 4.10, for example, shows the estimated probability that a "representative student," characterized by gender and race or ethnicity, will become a vocational concentrator at the three schools. In this case, the "representative student" is one who attended his or her school for four years and has math and reading scores equal to the average for his or her respective school.

Race Matters a Lot. The estimated probabilities in Table 4.10 show that, even after controlling for a student's achievement scores, significant differences remain in the likelihood that students of different racial or ethnic backgrounds will be vocational concentrators at Washington and Coolidge, but not at McKinley. For example, the probability that a representative boy at Coolidge is a vocational concentrator is 8 percent for Asians, 21 percent for African Americans, 28 percent for Latinos, and 38 percent for whites. The low probability estimate for Asian students at Coolidge is highly significant.

In contrast, "representative" Asian and Latino students at Washington are about equally likely to be vocational concentrators, but comparable white students have a much higher estimated probability of being a concentrator. This pattern occurs with and without adjusting for Washington's practical arts requirement.[49]

Gender Differences Are Not Strong. The differences for boys and girls shown in Table 4.10 are not as striking. Although the estimated probabilities for girls are less than for boys at Washington and Coolidge, and the reverse at McKinley, the differences are statistically significant only at Washington, when there is no adjustment for the practical arts requirement. In all other cases, the likelihood of being a vocational concentrator is not associated with a student's gender.

The estimated probabilities in Table 4.10 also allow comparisons between the three schools for students with similar characteristics. When we make no adjustment for Washington's requirement, the same ranking applies for six of the eight different groups of students in Table 4.10: McKinley students have the highest probability of becoming a vocational concentrators and Coolidge students have the lowest probability. The two exceptions are Latino boys and girls, who have a higher probability of being vocational concentrators at Coolidge than at Washington.[50] Again, this may be explained by the greater social-class differences between the cohort of Latino students at Washington and those at Coolidge. If we adjust for the requirement at Washington, the ranking between Washington and Coolidge reverses. Thus, the higher probabilities for students at Washington can be explained by the existence of the requirement at Washington and the absence of one at Coolidge.

As Test Scores Rise the Probability of "Concentrating" Declines. The previous comparisons were for students of different gender and races, holding student achievement measures constant within schools. Tables 4.11 and 4.12 present the estimated probabilities for boys by race or ethnicity when test scores vary.[51] First, Table 4.11 compares the probability of being a vocational concentrator for students with test scores at the same relative point in the test score distribution within each school, namely, the 25th percentile, the 50th percentile (median), and the 75th percentile. Table 4.12 compares students across schools with the same absolute test scores, specifically, with national percentile scores equal to 30, 50, and 80.

Both tables show the same overall pattern: Students with higher test scores are less likely to be vocational concentrators. Table 4.11 shows that the probability of a white male at Coolidge being a vocational concentrator decreases from 58 percent to 19 percent as his test scores increase from the bottom fourth of the class (25th percentile) to the top fourth (75th percentile). The negative relationship between test scores and the probability of being a vocational concentrator is highly significant and holds for both math and reading scores at all three schools.[52] Note, however, that the probability of being a vocational concentrator for students at the top fourth of their class or with test scores at the 75th percentile is still greater than zero. Thus, student ability as measured by achievement scores is a good predictor of the likelihood of taking a substantial number of vocational courses, but vocational concentrators are not exclusively students with low ability.

The same within-school differences between students of different races and ethnic groups that we observed in Table 4.10 appear when we compare students at each of these ranks. Again, white students at Washington have a higher probability than comparable achievers at the school who are Asian or Latino. Asian students at Coolidge stand out with a very low estimated probability relative to their comparably achieving schoolmates from different racial and ethnic groups. In contrast, however, the probabilities are similar for African American and Latino students at McKinley.

Even more interesting differences appear when we compare across the three schools. Again, McKinley students, regardless of their relative standing at their school, always have a much higher probability of being a concentrator than students at the same point in the test score distribution at the other schools. For example, a Latino male at McKinley with math and reading scores in the 75th percentile will be a vocational concentrator with 44 percent probability, but the probabilities are only 13 percent and 11 percent for his counterparts with the same relative test score standing at Coolidge and Washington (without adjusting for the practical arts requirement). These large differences occur as well for students at the low end of the distribution; the probabilities for a Latino male with test scores in the 25th percentile range from 68 percent at McKinley to 29 percent at Washington (without the adjustment).

This pattern is partly explained by the higher average achievement at Coolidge and Washington than at McKinley: Students at the top quarter of their class at McKinley are likely to be much lower achieving than their counterparts at Coolidge and Washington. Likewise, students who ranked in the bottom quarter of their class at Washington or Coolidge would be much higher achieving than students with that relative standing at McKinley. Thus, for students with the same standing in their class, the higher overall probability of being a vocational concentrator at McKinley is likely to be somewhat balanced by their lower levels of achievement.

Table 4.12 explores this explanation more precisely by comparing the probabilities of being a vocational concentrator for students with test scores in the same national percentile ranking. The between-school comparisons show the importance of the variation in student achievement across the three schools. Although students at McKinley still have higher probabilities when their test scores equal 30, 50, or 80 than students with the same scores at the other two schools, the differences are not as large as those found in Table 4.11. A Latino male at McKinley with test scores equal to 80 will be a vocational concentrator with 29 percent probability, compared to 15 percent at Washington without the adjustment, 13 percent at Coolidge, and less than 2 percent at Washington when the required courses are discounted.[53] Students with very low test scores exhibit similar patterns. If a Latino male has test scores equal to 30, he will be a vocational concentrator with a 66 percent probability at McKinley, 60 percent at Coolidge, 47 percent at Washington without the adjustment and 9 percent when the required courses are discounted.

It is important to note that achievement alone does not come close to explaining all of the race-linked coursetaking differences between the three schools. Whites with comparable achievement are more likely to take a concentration of vocational courses at Washington than at Coolidge, with the biggest differences among high-achieving whites, but once the required courses are subtracted their chances are greater at Coolidge. African Americans at McKinley are far more likely to concentrate than are their equal scoring peers at Coolidge, with the size of the differences in their prospects at the two schools increasing as their test scores go up. We can speculate from our data, although the practical arts requirement at Washington makes it difficult, that schools become more vocationally oriented overall as their populations of minority and low-income students increase. (This speculation is consistent with the national data cited in Sec. I.) As a consequence, we would expect to see more students take large numbers of vocational courses at low-income, minority schools, regardless of their test scores.

Other Factors: SES and Country of Birth. Factors other than race, gender, and test scores may also play a role in determining the likelihood that a student concentrates in vocational education. In particular, a student's socioeconomic status (SES) and country of birth may affect his or her coursetaking behavior. Data limitations precluded us from examining the role of these two variables at each of the three schools. However, using the SES data for Coolidge and the data on country of birth for Washington and McKinley, we can examine the importance of these factors individually at these schools.

Table D.3 contains the estimates, including SES in the logistic model, for predicting the probability that a student at Coolidge is a vocational concentrator. The results show that low-SES students are more likely than middle- or high-SES students of the same racial groups or with comparable test scores to be vocational concentrators. Middle-SES students are more likely to be vocational concentrators than to high-SES students, but the difference is not statistically significant. These findings mesh with what our respondents at Coolidge told us about differences among racial groups and the probabilities of vocational concentration at Coolidge that we described above. African American students were portrayed as a middle-class group and considerably more affluent than the Latino group as a whole and than many of the whites. In addition, since the SES data were based on counselors' assessments of a student's family income, the data were reported as missing when the counselor did not know the student well enough to make an estimate. This happened for about 10 percent of the sample. Consequently, an indicator that SES was missing was included in the logistic model. The resulting positive and significant coefficient indicates that these students were also more likely than high-SES students to be vocational concentrators. Thus, students from families with lower income or students for whom counselors are not able to evaluate their family's income appear to be more likely to take a large number of vocational courses.

Using the information on country of birth, the logistic models for Washington and McKinley were estimated with an indicator that a student was foreign-born. The estimates in Table D.3 for Washington, with and without the adjustment for the practical arts requirement, and for McKinley show that there is no significant relationship between the probability that a student is a vocational concentrator and being born in a foreign country.

CONCLUSIONS

Our data on vocational coursetaking at the three schools are consistent with national patterns: Although most students take some vocational education, low-income students and disadvantaged minority students take more courses, and particularly more occupationally oriented courses, than do whites and middle-class minority students. These differences appear both between and within schools. For example, African American boys at McKinley were more than twice as likely (and girls four times as likely) as their African American peers at Coolidge to concentrate in vocational education. And the least advantaged group within our socioeconomically diverse school (Latinos at Coolidge) was far more likely to take a concentration of vocational courses. Moreover, the least advantaged students (both between and within schools) were more likely to take courses related to the trades, and more advantaged students leaned toward courses in business.

One explanation for these patterns may be that schools permit students at all points of the achievement continuum to choose whether or not to take large numbers of vocational courses, and that low-SES and minority students, even those planning on college, place more value on attaining trade-specific skills than do more advantaged students, and gravitate toward these courses. Certainly, this was the perception of some of the school faculty. Moreover, such an explanation would be consistent with the fact that achievement criteria do not fully explain vocational coursetaking, since we find vocational concentrators across a very wide range of achievement at all three schools.

A second explanation is also supported by our field study--that schools match students to those courses where they are seen as most likely to succeed, given their motivation and prior achievement. This explanation is consistent with the view expressed by many at the schools--that course placements should "accommodate" students' abilities and motivation.

To the extent that vocational education is seen as more appropriate for lower- than for higher-achieving students, each school seemed to use meritocratic criteria for sorting students into courses to some degree. Within all three schools, concentrated vocational education coursetaking was largely, but not entirely, reserved for the least academically able students in the school, as measured by their scores on standardized achievement tests. On average, as achievement scores decreased, the likelihood of taking a concentration of vocational courses increased. The clustering of low-income and minority students in vocational programs, then, is a function of racial and ethnic group differences in average achievement test scores.

The darker side of this second explanation, however, is the picture that many faculty at the school painted of vocational education, particularly trade-oriented courses: a "dumping ground" where schools place students who are not expected to be successful in academic programs. Among vocational offerings, only business courses appear to escape this image. At all three of our schools, business courses stand out as being equally subscribed to by college-bound and non-college-bound students, and by concentrators and non-concentrators. This explanation is particularly troublesome, given the perceptions of many faculty that students' membership in various racial or "cultural" groups affects their suitability for academic courses. In particular, the perception that Latino students bring with them disadvantages such as mobile or unsupportive families and low academic motivation suggests that minority and low-income students may be more often the object of low faculty expectations. It is possible that some of this reasoning lies behind the fact that students at our disadvantaged minority school were most likely to be concentrators in vocational education, even those whose enrollment in college-preparatory programs indicated that they intended to go to college, and even those with test scores equivalent to white and middle-class non-concentrators at the other schools.

Patterns between and within the schools argue against choice, achievement screening, or racial stereotyping as a single explanation for vocational coursetaking. First, there are proportionately more vocational course "slots" at low-income, minority McKinley than at the other schools, so that even students in the the top 25 percent of their class have a greater probability of concentrating in vocational courses there than their counterparts at the more advantaged schools. More important, differences in the number of slots do not correspond neatly to differences in overall achievement levels at the schools. Although the lower-achieving school had the greatest vocational participation, and the higher-achieving school had the least (beyond required "practical" courses that included advanced computer courses), these differences were not proportionate. The result was that equally high-achieving students at the disadvantaged, minority school were considerably more likely to take a concentration of vocational courses than were their peers at the two more advantaged schools.

Second, race, ethnicity, and social class, independent of achievement, are related to the variation in vocational participation within schools as well as between them. Students with comparable achievement but different background characteristics had considerably different probabilities of taking large numbers of vocational courses at the two least vocational schools--those enrolling significant percentages of white students. At Coolidge, the racially and socioeconomically diverse school, the probability of concentrating was highest for whites, followed by Latinos and African Americans, and the lowest for Asians. But across these groups at Coolidge, low-income students were significantly more likely to be vocational concentrators than were middle- or upper-income students. At more homogeneously affluent Washington, whites were more likely than Asians to concentrate in vocational programs.

Thus, schools with larger concentrations of minorities and low-income students seemed to be disproportionately vocational, with the result that students of all backgrounds attending those schools were more likely to be vocational concentrators than their peers with comparable achievement scores who attended schools with larger numbers of white and middle-class students. And within schools as well, race and social class seem to influence vocational course participation over and above achievement. These factors suggest that decisions about how much vocational education to offer at a school and decisions about which students should take these courses are influenced by more than test scores. It is likely that students' own choices and the choices made for them by counselors and teachers play an important role as well.

[27]At Coolidge, 8 percent of the total course sections offered were vocational, 9 percent at Washington, and 15 percent at McKinley. See the discussion of the curriculum offerings at these schools in Sec. I and in Selvin et al. (1990).

[28]In addition to the problem of distance, some regional courses were also offered during early morning or evening hours.

[29]The sample for the analysis presented in this section and in Sec. V is the 10th through 12th grade cohort at Washington, Coolidge, and McKinley. These students were present from their sophomore through senior years. (Of this cohort, 85 percent of the Coolidge students were also enrolled in the 9th grade at that school, 93 percent at Washington, and 81 percent of the McKinley students.) Consequently, as noted in Sec. II, we are focusing on those students who are the least mobile and who have not dropped out of school before their senior year. Because of the differences in attrition and mobility across the three schools that we observed earlier, this analysis is biased toward making the schools and students' coursetaking appear more similar than they actually are.

[30]Washington requires that all students take at least two practical or vocational courses.

[31]See Appendix B for the occupational/non-occupational typology used in this analysis.

[32]The heavier participation in non-occupationally specific courses may also reflect the cutbacks that vocational programs at each school experienced in recent years. Our case study respondents told us that many advanced courses--those most likely to teach specific job skills--were eliminated. Fewer occupational courses relative to non-occupational courses are available at Coolidge, Washington, and McKinley (see Selvin et al., 1990, for additional detail).

[33]See Table C.1. The distribution of vocational credits taken follows a similar pattern. See Fig. C.1 and Table C.2.

[34]Figure C.2 presents the distribution of vocational credits after the subtraction of 10 credits among the total credits taken by each Washington student.

[35]See Appendix B on the development of the typology used in this table and accompanying discussion.

[36]This figure displays the actual number of courses taken by students at each school. The distribution of vocational credits taken in occupational subjects mirrors course distributions. See Fig. C.3.

[37]At Coolidge, these patterns are probably affected by the fact that the school houses the business courses offered through its regional occupational program, and students from two other schools come to Coolidge to take these courses. Consequently, more business courses are offered at Coolidge than might be the case if the school did not provide this regional service. However, the fact that largely middle-class Coolidge would house a regional program in business, rather than in the trades, is, in itself, consistent with our data about Washington and other studies showing a preference for business-oriented vocational education among middle-class students.

[38]Differences in the means are consistent with distributional patterns we observed. See Tables C.3 and C.4.

[39]Alternatively, we could have chosen to analyze those factors that determine the number of vocational courses or credits taken. A regression analysis, similar to the logistic analysis presented below, in which we modeled the factors that determine the number of vocational courses or credits, led to similar conclusions. Because of space limitations, we present only the results from the logistic analysis.

[40]The cutoff at six or more courses is somewhat arbitrary. However, the findings reported in this section are similar when we define the cutoff as five or more courses, or seven or more courses. Essentially, the between-school differences and the within-school differences are not affected by the point in the distribution we use to divide the students into non-concentrators and concentrators.

[41]Table 4.8 also shows that when we subtracted the two required semesters from each Washington student's record, the relative participation of Asians and Latinos shifted. Because of the large discrepancy in the two scores for Latinos and this shift, it is difficult to make conclusions about the proportion of concentrators in this small Latino segment of the student body (5 percent of the total), or how their participation compares to that of Asians.

[42]Note that these perceptions are not borne out by students' achievement test scores. At both schools, Latinos scored considerably lower than did whites. Moreover, Washington and Coolidge Latinos' scores were quite similar (see Table 4.5).

[43]We will examine the vocational and academic coursetaking patterns of these students we have defined as "vocational concentrators" in more detail below and in Sec. V.

[44]We omitted the scores of those students who fell in the top and bottom five percentiles to avoid distorting the range with "outliers"--a single student whose score may differ dramatically in one direction or the other from his or her classmates.

[45]The data for Figs. 4.4 and 4.5, shown in Table C.7 in Appendix C, document that the mean math and reading scores are significantly different for vocational concentrators than for non-concentrators. Table C.7 also shows the mean, 5th percentile, and 95th percentile for 8th grade math and reading scores at Washington and Coolidge. The pattern is similar to that shown in the figures based on 10th grade achievement scores.

[46]Since 8th grade achievement scores were not available for McKinley, we used 10th grade achievement scores in math and reading so that we could make comparisons across the three schools. The results are similar for Washington and Coolidge when 8th grade achievement scores are used instead.

[47]The pooled model included a separate intercept term (or dummy variable) for each school.

[48]The estimated coefficients from the logistic analysis are presented in Appendix D. The results in Tables 4.10 to 4.12 are based on separate logistic models for each school. This allows the effect of student characteristics on the probability of becoming a vocational concentrator to vary by school and is therefore appropriate for within-school comparisons. These results can also be used to make between-school comparisons. Alternatively, a pooled model with dummy variables can be used for between-school comparisons. The estimated probabilities from the pooled model lead to similar conclusions in most cases. Exceptions will be noted in the discussion.

[49]Here, too, differences in Latino participation may be related to the social-class differences between Coolidge and Washington Latinos. In addition, as we discussed above, Washington has a relatively smaller percentage of Latino students than Coolidge or McKinley. Race or ethnicity appear to play a lesser role in vocational coursetaking at McKinley, as there is no significant difference in the probability that representative African American and Latino students will be vocational concentrators. Neither, however, is there any evidence of socioeconomic differences between these two groups at the school.

[50]When the probabilities are estimated using the pooled model, there is a consistent ranking from highest probability to lowest of McKinley-Washington-Coolidge when there is no adjustment for Washington's practical arts requirement, and McKinley-Coolidge-Washington when there is an adjustment. In the first case, the estimated coefficients do not show a significant difference between Washington and Coolidge, whereas the coefficient for McKinley is highly significant. In the second case, the three schools are significantly different from one another on the basis of estimated coefficients on the school dummy variables.

[51]These estimates are shown for boys only. The differences between boys and girls will be the same as those reflected in Table 4.10. In general, girls and boys of the same race or ethnicity with test scores at the same point in the distribution exhibit no significant difference between the estimated probabilities. The exception is Washington, when there is no adjustment for the practical arts requirement.

[52]One exception is Washington, where the estimated coefficient on the reading score is negative but insignificant (see Table D.3).

[53]We suspect that, given these percentages, although more than 2 percent of the Latino students would be likely to take a concentrated vocational program even if Washington had not required students to take two courses, there would still be a larger percentage of Latino concentrators at Coolidge than at Washington.