Crain, R. L., Allen, A., Thaler, R., Sullivan, D., Zellman, G., Little, J. W., and Quigley, D. D. (1999). The Effects of Academic Career Magnet Education on High Schools and Their Graduates (MDS-779). Berkeley: National Center for Research in Vocational Education, University of California.

CHAPTER  5
Career Magnet Schools:
Effects on Student Behavior And Perceived Parental Support:
Part One

• Methodology
• Model I: Predicting Institutional Effects
• Model II: Perceived Parental Willingness to Sacrifice Financially for College
• Discussion
• Conclusions
• References

      A large body of research on school effects indicates that school characteristics matter in student outcomes (e.g., Good & Brophy, 1986; Reynolds, 1992; Stringfield & Herman, 1995). Further, those characteristics that matter most are often alterable ones such as an orderly school climate, high teacher expectations for student performance, strong principal leadership, and a conscious effort to create a school environment that incorporates these elements (Stringfield & Herman, 1995). The goal of this paper was to explain what it is about attending a career magnet program that contributes to the better outcomes described in this volume.

      Bryk, Lee, and Holland (1993) analyzed the reasons that disadvantaged family background is less predictive of educational outcomes in Catholic high schools than in public ones. They found that shared organizational beliefs, including an academic core curriculum for all students, an emphasis on academic pursuits, extensive extracurricular involvement and religious activities, and the importance of character development, combined with much higher percentages of staff who teach in these schools because they want to produce improved outcomes, account for this phenomenon.

      While career magnets obviously do not offer the religious activities of Catholic schools, in many respects these schools are similar. Nearly all of the students (lottery winners and others) clearly want to be there; there is an academic core curriculum for all students so that career concerns do not replace academic ones. While traditional extracurricular activities are lacking in career magnets for want of time, the internships may serve some of the same functions. They may foster the building of personal relationships between student and employer, and student and advisor, as well as provide the reinforcement of one of the school's key organizing beliefs: the value of work and the possibility of a meaningful future career. Beliefs in the possibility of meaningful work and career success have, in turn, been associated with decreased involvement in high-risk behaviors such as alcohol and drug use, truancy, and unprotected sex and teenage childbearing (e.g., Dryfoos, 1990; Jessor & Jessor, 1977; Levitt, Selman, & Richmond, 1991; Miller, Carol, Parkoff, & Peterson, 1992). The career magnet internship program also helps to reinforce the lessons of the classroom and a key shared belief among faculty and students: the importance and legitimacy of workplace socialization, including on-time behavior, appropriate attire, and personal responsibility.

      In addition, like Catholic schools, the awareness of the family and individual sacrifice required for attendance (in Catholic schools, tuition payments; in career magnets, time required for often long commutes that may preclude paid employment), and the understanding that poor performance may lead to expulsion contribute to an environment of shared effort and improved outcomes.

      Research also finds that a young person's family is an important force in shaping behavior and achievement motivation (Middleton & Loughead, 1993). Family background and processes are the basis from which career planning and decisionmaking evolve. Parents provide daily models of cultural standards, attitudes, support, and expectation and, in many ways, determine self-esteem, interpersonal skills, and role models for work (De Ridder, 1990).

      At the same time, as young people mature and individuate, their behavior evokes parental responses that affect educational and career motivation and success. Students' behavior, motivation, and academic achievement influence the level of parental support and expectations for their child. Understanding these behaviors and motivational interactions between parent and child and their relation to young people's decisions about school and work is one of the key aims of these analyses.

      By interviewing a sample of recent career magnet and comprehensive high school graduates who won and lost the career magnet lottery respectively, we hoped to collect data that would help us understand the institutional and family effects of career magnet programs.

Methodology

      We chose to focus our analyses on two institutions that strongly influence young people's academic progress and outcomes: school and family. We were, of course, especially interested in identifying the effects of career magnet enrollment on students' academic and job-related outcomes, and on measures of emotional well-being. We also wanted to look at any impacts of career magnet attendance on family outcomes. To do this, we developed two models of institutional effects: one focused on the predictors of graduating from a career magnet or comprehensive high school; the other on an important family outcome--the students' perception of parental support for college attendance.[9] In the first model, significant predictors of career magnet graduation define key institutional effects of career magnet programs. In the second, significant predictors of perceived parental support for college clarify a complex process through which students communicate to parents their commitment to hard work and education, and parents in return signal a willingness to sacrifice financially to help a motivated young person pursue her or his goals.

      Semistructured interviews were conducted with teachers and administrators at four career magnet schools about their experiences with career magnets and the lottery. Where a teacher's tenure had predated the lottery, she or he was asked to talk as well about perceived differences in student body, teaching style, and her or his own sense of efficacy before and after the lottery began. For the most part, school-site interviewees were selected on the basis of willingness to cooperate and free periods that meshed with project staff schedules.

      The multivariate analyses used two estimation methods: (1) ordinary least squares (OLS) and (2) logistical regression (LOGIT). The latter method is preferred when the dependent variable is dichotomous, as is the case for both of our outcome measures because a linear specification may misrepresent the underlying functional form of the relationship between the dependent and independent variables. The coefficients of OLS regression are easier to interpret, however, as they indicate a percentage change in the value of the dependent variable due to a one percent increase in the independent variable. To benefit from the advantages of each method, the OLS regression results must closely replicate those of the LOGIT regression in terms of fit, error, and significance.

      In this analysis, the coefficient estimates from the OLS and the LOGIT have similar levels of overall significance and have similar levels of significance on individual independent variables, which allows this dual method of statistical analysis. Therefore, both models were estimated and used for interpretation.[11]

      Since we were looking for and hoped to find differences between the two groups in outcomes, it was important to document at the outset the different ways in which career magnet and comprehensive high school graduates had experienced high school. A most significant and expected difference between the two groups was the substantially higher level of career-related coursework that the career magnet students had taken. Career magnet students averaged 13.4 hours/credits of career-related coursework during high school; comprehensive high school graduates averaged 5.2 hours/credits. This large difference was not surprising given the focus of the career magnet programs.

      We also found that career magnet students were significantly more likely to feel that their school was safe and desirable. For example, career magnet students were more than four times as likely as comprehensive high school students to rate their high school's reputation as "good to excellent." To some extent, these feelings may demonstrate cognitive dissonance at work: career magnet students had to travel farther and longer to reach their schools than comprehensive high school students did. They may also demonstrate the effects of selectivity: career magnet students wanted to be there, and they were twice as likely as comprehensive graduates to indicate that they would choose the same high school again. In addition, since we interviewed only graduates, there may have been students who did not perceive sufficient value added to stay at the career magnet schools. Indeed, attrition rates are substantially higher in career magnet programs. More optimistically, these positive assessments may reflect the shared values and increased success opportunities that characterize special schools.

      Interestingly, we did not find some of the differences that we expected between the two groups on variables that are thought to differentiate career magnet and special schools from others. For example, we expected that career magnet students would report significantly more out-of-class and informal contacts with teachers and counselors. This did not prove to be the case. Indeed, we found that comprehensive high school students were significantly more likely to report having spent time during high school talking with a counselor about a career or future work; they were also significantly more likely to have gotten to know an adult on a personal level in high school.

      At the same time, we found that career magnet students were significantly less likely to engage in a variety of behaviors that are associated with reduced school performance. Career magnet graduates were significantly less likely than comprehensive high school graduates to have ever been in a fight during or since high school; to have ever smoked, used drugs, drunk alcohol at least weekly; or ever became pregnant or made someone pregnant. In sum, 41% of career magnet graduates reported no risk behaviors, while only 19% of comprehensive high school graduates fell into the "no reported risk behaviors" category. Indeed, the reduced incidence of these high-risk behaviors constituted the biggest differences between career magnet and comprehensive graduates.

      The substantially lower incidence of a wide range of at-risk behaviors might well be due to the higher attrition rates in career magnet programs, discussed earlier. They may also reflect the impact of the institutional setting on career magnet students. An academic core curriculum for all students, shared beliefs in the importance of work, and the legitimacy of workplace socialization led to the enforcement of many behaviors such as punctuality, appropriate attire, and personal responsibility that are incompatible with high-risk behavior. The teaching of career skills may have led as well to a sense that work and a career could be attained, beliefs that are also incompatible with taking high risks.

      Better behavior and skills acquisition appeared to pay off. Career magnet graduates indicated a starting hourly wage that was one dollar higher than that for comprehensive students, with the former reporting a starting hourly wage of $7.27, and the latter a starting hourly wage of $6.28. Current hourly wage also varied in the same way for the 61 interviewees who were currently working. Reported wage rate for the career magnet graduates was $8 an hour, while comprehensive high school graduates reported an hourly wage of $7.01.

      These findings too could reflect differential attrition in the two groups of students. At the same time, there were indications that the high school experiences of career magnet and comprehensive high school students were rather different; these differences also could have affected student behavior and outcomes. These results suggested that we might be successful in modeling and understanding institutional effects.

      As noted above, our analyses were directed toward two models. Variables included in each model were selected on the basis of several criteria. For the first model, we selected variables that proxied five constructs that theorists (e.g., Bryk et al., 1993) suggest are important in explaining institutional effects. These constructs included self-efficacy, career identity, institutional characteristics, student risk behaviors, and parental and family characteristics. To get closer to measuring these constructs, we collapsed some of the variables, and created scales from others. Prior to this construction process, we eliminated variables with "n"s below 100, as we believed that we needed at least that number of observations to reliably use the item. (In most instances, when the n was under 100, it was considerably under this number, averaging 60 to 80 because the item was not asked of all interviewees.) This left us with five "groups" of variables representing predetermined theoretical constructs.

      Then, for each of the five groups, we separately ran regression analyses predicting career magnet or comprehensive high school program graduation to understand the interactions and individual impacts of variables within each construct. Variables that predicted at p < .05 were also tested for proper model specification. (Tests for multicolinearity and heteroscedasticity were conducted before variables were included in the final equation.) Based on the hypotheses tested concerning self-efficacy, career identity, institutional characteristics, and parental influences, a model was constructed to predict high school placement.

      For the second model, a similar approach was used. We chose variables that the literature suggested might be important predictors of perceived parental willingness to sacrifice for their child's college education. In this category, we included a range of student behaviors that might signal seriousness of purpose. These included self-reports of truancy, fighting in high school, other risk behaviors, performance in high school, occupational specificity, and self-efficacy. We selected these variables on the assumption that a major factor in a parent's willingness to support a child's college attendance is that child's level of commitment to education, as evidenced by good academic performance and the absence of behaviors that suggest lack of commitment (keeping in mind that parents' own educational level and financial status were randomly distributed across the two groups). Within each of the five construct groups, various hypotheses were tested to identify both potential predictors and influence among variables. Regressions were run to identify those variables significant at the p < .05 level. Then, the potentially significant (statistically and theoretically) variables were analyzed further in terms of their ability to explain the student's perception of his or her parent's willingness to sacrifice for college. We selected for inclusion in the model those variables that contributed most to explaining parents' willingness.

      Both models were tested for multicolinearity and influential data points affecting the fit of the regression model.[12] The tests indicated that the models were specified correctly and did not contain colinear independent variables or influential data points.

Model I: Predicting Institutional Effects

      Our analyses revealed that four variables together explain 24.5% of the variance in school program type. Table 5.2 reports the results from the OLS regression on school program type. These variables include the student's perception of parental willingness to sacrifice financially for college, the student's propensity to engage in risk behaviors during high school, the student's possible influential friendships during high school, and student's confidence in ability during high school English classes most or all of the time. The logistic regression predicts that a student who perceives that his or her parents do not think that college is important enough to give up other things for, who reports having engaged in at least two high-risk activities in high school (e.g., smoking, drug use, fighting), whose closest friend in high school had no ideas about a career, but who felt confident most or all of the time in his or her high school English class has a 32.2% probability of having graduated from a career magnet program, while the same student had an 89.1% probability of having graduated from a comprehensive high school.

      Controlling for the other variables in the OLS regression model, we find that a student who perceives that his or her parents think college is important enough to make some financial sacrifices for is 27.4% more likely to have graduated from a career magnet than those who do not think their parents are willing to make such sacrifices. Reported risk behaviors in high school--including smoking, taking drugs, or fighting in school--were 15.8% less likely to occur at a career magnet high school. Having a closest friend during high school who had ideas about a career was 17.1% more likely among career magnet graduates. Finally, a student who felt confident in high school English classes most or all the time rather than none of the time or sometimes was 16.8% less likely to have graduated from a career magnet program.

      These variables suggest that career magnet programs exercise important effects on enrollees. Conceptually, the finding that career magnet graduates were significantly more likely to have a closest friend who had career ideas is most significant. One's closest friend in high school is an important index and transmitter of social norms and values. Having a best friend with career ideas suggests that career magnet students are more likely to be exposed to an environment in which career thinking and career planning are normative. It is unconventional to present a regression equation with the independent variable "school type" used as if it were a dependent variable, but it allows us to demonstrate that the career magnets have an effect on each outcome even when the other outcomes are controlled. (A more traditional regression analysis confirms that these results are significant, including "closest friend . . . .")

      This, in turn, is likely to facilitate one's own career thinking. Our life history interviews suggest that career magnet programs facilitate the establishment of friendships that are more likely to be based on shared career concerns because they bring students into a new environment peopled by students who are likely to have shared career interests, and do so at a time when friendships are more likely to form around more mature issues than those friendships formed in early adolescence at the beginning of junior high school. Our finding that career magnet graduates are significantly more likely than comprehensive high school graduates to report that their closest friend in high school came from school rather than from the neighborhood supports the notion that new friendships form and matter. In addition, attending a school in which half of the students are admitted on merit no doubt improves the overall quality of the student body, increasing the chances that school norms and individual friendships will encourage effort and success.

      The model also reveals that career magnets may operate to minimize those behaviors that are inimical to school and career success. An extensive literature on risk-taking behaviors (e.g., Elliott & Morse, 1989; Hardy et al., 1997; Weinstein, 1980) tells us that students who smoke, drink, and act out in other ways are less likely to be successful in school and in life. Our model suggests that the school environment, no doubt aided by differential attrition, can act to reduce these behaviors. Strict rules in at least some of the career magnet programs represented in our study and the heavy focus on the future in all of them may have created an environment inhospitable to the development or manifestation of risk-taking in a population no more or less inclined to take risks than their lottery-losing counterparts. These forces may well have impelled risk-takers or potential risk-takers to leave school. In addition, a student body in which half of the students are selected on their merits contributes to norms that discourage acting out.

      Other factors operating in career magnets may also reduce the inclination to engage in risky behaviors. More hands-on work, the promise of a marketable skill, and less "school-like" activities may contribute to students' reduced likelihood of acting out. Career magnets may also increase students' options, as they are intended to do. We learned, for example, that some career magnet graduates were using the skills that they learned in high school, particularly secretarial skills, to earn money to support themselves through college. Indeed, our data show that career magnet graduates had completed more college credits (p = .068) at the time of their interviews and were significantly more likely to have declared a college major.

      The confidence about English variable is consistent with several other instances where lottery losers who graduated from comprehensive high schools described themselves as more confident, less worried, and more prepared for the world of work. The most parsimonious explanation for these findings is that classwork in comprehensive high schools may be easier. After all, comprehensive high schools lack the carefully and stringently chosen students who comprise half of the student body in career magnet schools, whose students must be diverse in terms of test scores, but typically have good grades and attendance. Combined with a small percentage of good readers chosen at random in the lottery, a plurality of career magnet students are high achievers. No doubt they raise both teacher expectations and the level of instruction in career magnet programs.

      Finally, perceived parental support for college significantly distinguished career magnet and comprehensive students, with career magnet students more likely to report that their parents valued their college attendance and would be willing to sacrifice financially for it. There are several possible reasons for this effect. First, continuing to attend a career magnet program requires evident dedication and sacrifice, as discussed above. Long travel times in many cases require that students get up earlier and spend more hours in pursuit of an education than is the case for students who attend comprehensive high schools in the neighborhood. This commitment may be communicated to parents and increase their inclination to support the student's ambitions. Alternatively, the student who does all this may believe that he or she deserves parental sacrifice; this may predispose such students to report greater parental commitment. Other evidences of student commitment may also incline parents to be more supportive. As discussed above, we know that career magnet students are more likely to have a closest friend who has thought about careers. A fairly intense relationship with a serious peer may incline parents to believe that their child, too, is serious about getting ahead.

      These speculations could take us only so far. We felt that perceived parental support for college was important enough to pursue in more detail since family values and expectations are a major influence in the educational choices that students make (e.g., Baker & Stevenson, 1986; Duncan, 1994). Consequently, we attempted to predict students' perceptions of their parents' willingness to sacrifice financially to help them attend college.

Model II: Perceived Parental Willingness to Sacrifice Financially for College

      This model assumes that students' perceptions of parental willingness to sacrifice financially to help them attend college is a product of family ability to pay and the anticipated payoff of that investment. Since SES was randomized in this sample, we focus on the latter concept here. There is considerable data that indicate that lower-income parents value education as much as higher-status ones (Lareau, 1989; Lightfoot, 1978), but their own lack of education impedes the transmission of this value. Indeed, Lankard (1995) found that among Hispanic families in their study, parents valued education more highly than their children. At the same time, college is costly, and lack of money impedes plans for many. Certainly, if there are multiple children, parents are more likely to invest in the education of those children who appear motivated to succeed. The interaction of many individual behavioral or motivational variables within a family is a significant factor in studying family influence on career development (Middleton & Loughead, 1993). For these reasons, we included in the model a number of student behaviors that parents are most likely to be able to discern. For example, we think that parents are more likely to learn that their child has gotten into a fight in school or is drinking heavily than if their child is using drugs or smoking. We include an indicator of how much the student drank during their senior year: almost daily to two or three times a week, or occasionally to never. We include an indicator of whether the student ever talked to his or her parents about financial help to go to college. Based on what we found in Model I, we included institutional type in the model to test both the explanatory strength of institutional effects on perceived parental support predilections and to determine the strength of other factors on perceived parent behaviors controlling for institutional effects. Other variables in the model focus on a measure of the student's own career self-efficacy.

      Our analyses revealed that the four variables together explain 21.3% of the variance in students' perceptions of parental willingness to sacrifice financially for college. Table 5.3 reports the results from the OLS regression. These include type of school attended, student's level of confidence that he or she will be in their desired career within the next six to ten years, student-reported drinking behavior in the senior year of high school, and level of college-related planfulness. This latter variable was scored as planful if the student indicated that he or she had talked to his or her parents about financial assistance in college. A number of other variables that seemed related (e.g., frequent truancy, reported fighting in high school, other risk behaviors such as smoking or taking drugs, other planful behaviors such as taking the SAT/ACT or visiting a college campus) were tried in the model but did not contribute to explaining perceived parental willingness to financially sacrifice for college. Nor did a higher level of occupational specificity, reported happiness, or reported level of control over life appear to matter in predicting perceived parental support for college.

      Logistic regression predicts that students who graduated from a career magnet, who are very confident that they will be in their desired career within the next six to ten years, who drank less in high school, and who discussed college finances at home are more likely to perceive that their parents are willing to sacrifice to send them to college.

      Controlling for the other variables in the OLS regression model, we find that a student who graduated from a career magnet is 30% more likely than a comprehensive high school graduate to perceive that his or her parents are willing to sacrifice financially for college. Students who report that they are very confident that they will be in their desired career within the next six to ten years are 19% more likely than students with little confidence in this area to perceive that their parents are willing to make financial sacrifices to support college attendance. Those who reported drinking infrequently in their senior year of high school were 16.4% more likely to perceive that their parents would be willing to sacrifice financially for college. Those who reported drinking daily as high school seniors were 32.8% less likely than those who reported drinking less than two or three times a week to perceive such parental support. Finally, those who discussed college at home were 22.5% more likely to perceive that their parents were willing to sacrifice so that they could go.

      These variables suggest that behaviors on the part of the student while in high school such as commitment to schoolwork and career are associated with the student's perception that his or her parents support college attendance and are willing to sacrifice financially to help the student attend. The model reveals that the most important of the four variables is attendance at a career magnet. This may be because this is by far the most visible of the variables; it may also be that career magnet attendance, being associated with the other variables discussed in Model I, conveys seriousness of purpose to parents.

      The behavior and college-related planfulness bear the expected relationship to perceived parental commitment to the student's education. Students who convey career orientation and confidence, stay out of trouble, and show evidence of their own commitment to attend college seem to pull (or perceive that they have pulled) such commitment from their parents. As discussed in Model I, career magnets appear to create for many students an environment in which behaviors that foster life success are more likely to occur. Model II suggests that career magnet attendance also may have a salutary effect on parents that is communicated to their children.

Discussion

      Our data indicate that students who attend career magnet programs benefit from their involvement in them. The data also point to several ways in which these programs may achieve this effect. Career magnet attendance seems to reduce the likelihood of engaging in the sorts of risk-taking behaviors that mitigate against school and life success. Career magnet attendees report that they engage in fewer high risk behaviors overall, and their lower level of drinking in particular is associated with perceptions of greater parental willingness to sacrifice financially for college.

      Career magnet students are also significantly more likely to describe their closest friend as being career oriented. This fact suggests that career magnet students may live in a world that is more supportive of career thinking and willingness to demonstrate concerns for the future. This future orientation is also expressed in their greater involvement in college planning activities, including discussions with their parents about college attendance. Career magnet students also seem to have a somewhat more realistic view of their own capabilities.

      What explains these effects? Our data do not allow us to determine precisely what accounts for the institutional effects that we observed. We can offer a few ideas, however. First, there is the commitment notion. Continuing involvement in a career magnet requires continuing commitment, as evidenced by long travel times and consequently longer hours devoted to schooling. Second, there is the social norm effect, about which we do have some data. If more career-committed students attend career magnets, they are more likely to create and transmit norms that support a career commitment. Indeed, our data show that one of the more powerful effects of career magnets is the greater likelihood that one's closest friend has thought about careers. The fact that career magnet graduates are significantly more likely to report that their friends come from school rather than the neighborhood supports the idea that career magnet schools work in part because by their nature, they foster new friendship groups and new group norms. These new friendship groups may matter because they contribute to an environment that supports hard work and achievement and eschews high-risk behaviors. The dramatically lower levels of a range of high-risk behaviors among the career magnet graduates suggest that the school setting suppresses these behaviors, which, in turn, increases the likelihood of positive academic and social outcomes.

      Third, career magnets may generate more interest and effort because they represent more desirable environments for learning. Career magnet graduates were twice as likely as comprehensive school graduates to tell us that they would choose the same high school again, and they are four times as likely as comprehensive graduates to rate their high school's reputation as "good to excellent." Certainly, our interviewees told us that they felt far safer in career magnet schools; we heard informally that many students apply to career magnet schools primarily because of their reputations as safer learning environments.

      Our informal interviews with teachers in some of the career magnet schools support this notion of a more facilitative learning environment. Some told us that they prefer teaching in career magnet schools because they feel that they have some leverage over students that is absent in comprehensive high schools: Students who misbehave can be sent back to their home school. In addition, the safer school environment acts as a unifying factor; since everyone values a safer environment and benefits from it, faculty and administrators feel more empowered to enforce rules and regulations that promote and ensure safety. This fits with our finding that career magnet students are significantly less likely to report cutting classes. In fact, we learned in one career magnet school that the absence of free periods during the day enables faculty and administrators to immediately identify students who are cutting class. Finally, faculty told us that the shared academic career focus of the school helps them to develop curriculum materials that are more relevant to students. Consequently, students become more involved in the work. An economics teacher told us, for example, that his students relate well to the concepts he introduces because so many of them are participating in the world of work and can relate class material to their on-the-job experiences. The shared focus on careers and the shared value of working and earning a living also allows teachers to make demands such as public speaking in the service of becoming more marketable. In comprehensive high schools, students might well object to such a demand as there is no strong reason for it. Several teachers told us that a whole school of students who are studying related careers (e.g., health services) helps to create a community of committed students. All of these factors serve to increase teacher morale in career magnet schools, which contributes as well to a more positive school climate.

      Another effect of the academic career focus is that students tend to take multiple classes from the same teacher. According to the teachers we interviewed, this allows them to get to know students better and inclines them to feel more responsible for students' performance and future plans.

      Our finding that career magnet students are significantly more likely to believe that their parents value college and would be willing to sacrifice financially to enable them to attend suggests that career magnets may multiply their effectiveness by impacting on parents as well. The more pro-social behaviors of career magnet students, their greater commitment to school, and their increased likelihood of taking a series of steps to prepare for college all seem to incline parents to support college or at least to create that impression in their offspring. Since school and family are major factors in young people's lives, the fact that career magnet programs seem to exercise a positive effect on students directly as well as indirectly through their parents is particularly impressive.

Conclusions

      Our data indicate that career magnet programs promote positive outcomes, and do so by increasing positive student behaviors and decreasing negative ones. Our best guess is that they achieve these effects by creating a school culture that supports hard work, dedication, and continuity of purpose. A safer learning environment allows this culture to flourish. A shared commitment to a general career area, combined with a commitment to learn about the world of work, clearly leads to a culture whose diverse elements combine formally and informally to promote career discussion, career and college planning, and realism about the future.

      From a policy perspective, it is important to note that at least some of these effects are achieved because students who attend career magnets are all in some sense "winners" and, thus, feel grateful to be there. Continued attendance requires continued extra commitment. There is some evidence that the possibility of expulsion adds to the sense that these programs are opportunities that should be appreciated and not squandered. In addition, a student body enriched by those who entered the school on their merits contributes to the salutary effects of career magnet programs.

      Another advantage that accrues to the career magnet experience is the sense of shared values that goes with it. As Bryk et al. (1993) note, a sense of shared values allows teachers to make more demands of students, a key to successful teaching and learning.

      Can career magnet effects occur in other settings? We cannot answer that question with our data; however, the data do suggest some of the key factors that promote positive outcomes among career magnet graduates. Policymakers and administrators would be wise to consider ways to encourage these effects in other contexts.

References

Baker, D., & Stevenson, D. (1986). Mothers' strategies for children's school achievement: Managing the transition to high school. Sociology of Education, 29,156-166.

Bryk, A., Lee, V., & Holland, P. (1993). Catholic schools and the common good.Cambridge, MA: Harvard University Press.

Chartrand, J., & Camp, C. (1991). Advances in the measurement of career development constructs: A 20-year review. Journal of Vocational Behavior, 39,1-39.

De Ridder, L. (1990). The impact of parents and parenting on career development.ERIC, Position Paper No. 120.

Dryfoos, J. (1990). The determinants of early childbearing and other risk behaviors: Framework for a prevention strategy. In L. Lipsitt & L. Mitnick (Eds.), Self-regulatory behavior and risk-taking: Causes and consequences (pp. 23-41).Norwood, NJ: Ablex.

Duncan, G. (1994). Families and neighbors as sources of disadvantage in the schooling decisions of white and black adolescents. American Journal of Education, 103,20-53.

Elliott, D., & Morse, B. (1989). Delinquency and drug use as risk factors in teenage sexual activity. Youth and Society, 21,32-60.

Good, T., & Brophy, J. (1986). School effects. In M. C. Wittrock (Ed.), Handbook of research on teaching (3rd ed.) (pp. 570-602). New York: Macmillan.

Hardy, J., Shapiro, S., Mellits, E., Skinner, E., Astore, N., Ensminger, M., LeVeist, T., Baumgardner, R., & Starfield, B. (1997). Self-sufficiency at ages 27 to 33 years: Factors present between birth and 18 years that predict educational attainment among children born to inner-city families. Pediatrics, 99(1), 80-87.

Jessor, R., & Jessor, S. (1977). Problem behavior and psychosocial development: A longitudinal study of youth.New York: Academic Press.

Lankard, B. A. (1995). Family role in career development. ERIC Digest, No. 164.Columbus, OH: ERIC Clearinghouse on Adult, Career and Vocational Education. Available on-line at <ericacre.org>.

Lareau, A. (1989). Home advantage.Philadelphia: Falmer Press.

Levitt, M., Selman, R., & Richmond, J. (1991). The psychosocial foundations of early adolescents' high-risk behavior: Implications for research and practice. Journal of Research on Adolescence, 1(4), 349-378.

Lightfoot, S. (1978). Worlds apart: Relationships between families and schools.New York: Basic Books.

Middleton, E., & Loughead, T. (1993). Parental influence on career development: An integrative framework for adolescent career counseling. Journal of Career Development, 19(3), 161-173.

Miller, B., Card, J., Paikoff, R., & Peterson, J. (Eds.). (1992). Preventing adolescent pregnancy.New York: Sage.

Reynolds, D. (1992). School effectiveness and school improvement. In J. Bashi & Z. Sass (Eds.), School effectiveness and improvement: Selected proceedings of the Third International Congress for School Effectiveness (pp. 67-87). Jerusalem, Israel: Magnes Press.

Stringfield, S., & Herman, R. (1995). School effectiveness and school improvement: Recent U.S. advances. In B. Creemers (Ed.), International developments in theory and practice of school effectiveness and school improvement. Leeuwarden, the Netherlands: RION.

Weinstein, N. D. (1980). Unrealistic optimism about future life events. Journal of Personality and Social Psychology, 39,800-820.

[9] We were somewhat stymied in this effort because of the nature of the interview process and the difficulties inherent in measuring career identity (e.g., Chartrand & Camp, 1991). Many questions were designed to elicit descriptive information only. For example, demographic and family characteristics questions would serve as checks on the randomized nature of the student assignment process; we did not expect to find differences on these variables between the career magnet and comprehensive high school groups. The interview form was set up so that most interviewees were eligible to answer only a fraction of the 440 questions on the interview form. Consequently, many of the most interesting questions were answered by only a small percentage of interviewees--too few to use many items in our models. This forced us to limit our modeling effort to those questions that were asked of all interviewees, a far smaller question pool. For this reason, some of the variables that we include in our models are in some cases proxies for other measures for which we had too few responses.

[10] This latter measure is not an indicator of financial status, but the student's assessment of how important college attendance is to his or her parents. The options, which include "college is a waste of time and money"; "college is a good idea if you could get financial support"; "college is important, but not enough for the family to give up other things"; and "college is important enough for the family to give up other things," were used to assess the extent to which some of the student's own behaviors affected his or her perceptions of parental support for college. No one chose the first option, "college is a waste of time and money." The second and third options, "college is a good idea if you could get financial support," and "college is important, but not enough for the family to give up other things," were combined, resulting in a dichotomous variable. The two options described parents who were perceived by their child as willing or not willing to make financial sacrifices to facilitate college attendance.

[11] In the discussion of results, the OLS regression estimates are interpreted and the LOGIT regression estimates are used to calculate the probability that typical students identified by the characteristics represented in the models by the independent variables were either graduates from comprehensive high school or career magnet programs (Model I) or perceived that their parents valued their college attendance sufficiently to make financial sacrifices to facilitate it (Model II).

[12] Variance inflation factors were used to test for multicolinearity. First and second moment specification tests were used to test model fit. To check for influential data points, we conducted a test of a normalized change in the OLS estimate of the ith value of the dependent variable resulting from omitting the ith observation when calculating the OLS coefficient estimates.

Crain, R. L., Allen, A., Thaler, R., Sullivan, D., Zellman, G., Little, J. W., and Quigley, D. D. (1999). The Effects of Academic Career Magnet Education on High Schools and Their Graduates (MDS-779). Berkeley: National Center for Research in Vocational Education, University of California.