Exposés of low high school graduation rates have sparked renewed interest in the nation's dropout problem. Nearly every governor has signed a pledge to report more honest graduation data, and federal accountability policies are putting greater pressure on schools and districts to improve graduation rates than ever before.
Even without pressure from above, there are compelling moral, social, and economic reasons for districts to reduce dropout rates. First and foremost, students want to graduate: Ninety-nine percent of high school sophomores expect to earn a high school diploma, and about three in four expect to earn a bachelor's degree (Ingels et al. 2005). Far from being unreasonable, those aspirations are soundly logical. Thirty years ago most dropouts could still find jobs that paid enough to support a family, but young people who leave school today face a lifetime of economic hardship. Between 1974 and 2004, the annual earnings of families headed by a high school dropout declined by nearly one-third (Postsecondary Education OPPORTUNITY 2006).
High dropout rates beget social and economic woes for communities as well. Dropouts are far more likely to become unemployed, receive public assistance, commit crimes, and become incarcerated. At the same time, they are less likely to receive job-based health insurance and pension plans, to stay healthy and live full lives, and to vote and make other kinds of civic contributions. In fact, the average dropout pays about $60,000 less in taxes over his or her lifetime (Rouse 2005, Waldfogel et al. 2005, Muennig, 2005, Moretti 2005, and Junn 2005). Raising graduation rates would save taxpayers money, greatly expand tax revenues, boost employment, reduce crime, and improve citizenship.
This research review helps make sense of the wide range of actions a district can take to reduce dropout rates, highlighting the policies, practices, and programs that research shows are most likely to make a difference. It is organized around the major components of a comprehensive plan for raising graduation rates—the three or four things districts must do, and do well, to help significantly more teenagers complete high school.1
- Prediction: Identifying potential dropouts
- Intervention: Helping at-risk students
- Prevention: Changing the factors that schools control
- Recovery: When prevention and intervention are not enough
- Considerations for implementation
Prediction: Identifying potential dropouts
Although effective interventions can be costly, many decision makers fail to use data that can help them target scarce dollars wisely. For example, a major evaluation of federally funded dropout interventions revealed that programs frequently enroll the wrong students: "Dropout prevention programs often serve students who would not have dropped out, and do not serve students who would have dropped out" (Gleason and Dynarski 2002).
Just as in medicine, improper diagnoses can lead to wasted money and wasted lives. Reducing "false positives" is the key to saving money, because it means not spending dollars on students who would graduate on their own without additional help. Reducing "false negatives" is the key to raising graduation rates, because it means identifying more students who will drop out unless they receive additional help. Even the most effective intervention programs will fail to reduce dropout rates if they target the wrong students.
The problem is not that educators simply make uninformed guesses about who is likely to drop out. In fact, programs have long used checklists of risk factors that are generally correlated with dropping out—ways that the average dropout differs from the average graduate (Wells et al. 1989). Decades of research have yielded a huge list of such characteristics, including (Rumberger 2004, Gleason and Dynarski 2002):
- Demographic background: Students who are poor, who are members of certain minority groups, who are male, who have limited English proficiency, who have learning or emotional disabilities, who move more often, and who are overage for their grade are more likely to drop out.
- Family factors: Students who come from single parent families, have a mother who dropped out of high school, have parents who provide less oversight and support for learning, and who have older siblings who did not complete school are more likely to drop out.
- Adult responsibilities: Teenagers who take on adult roles such as becoming a parent, getting married, or holding down a job are more likely to drop out—although the last depends on gender, type of job, and number of working hours per week.
- Educational experiences: Dropouts are more likely to have struggled academically: Low grades, low test scores, Fs in English or math, falling behind in course credits, and being retained are associated with lower chances for graduation. Dropouts also are more likely to have shown signs of disengagement from school: High rates of absenteeism or truancy, poor classroom behavior, less participation in extracurricular activities, and bad relationships with teachers and peers all have been linked to lower chances for graduation.
The problem is that many of those risk factors turn out to be poor predictors of precisely which individual students actually will drop out. Exhibiting a risk factor places a student in a group whose members are, in general, more likely to drop out, but does not automatically mean that a particular student will drop out. If 40 percent of the dropouts in a district exhibit a risk factor, compared with only 10 percent of graduates, one might assume the risk factor is a good predictor because dropouts are four times more likely to exhibit it. However, that also means the majority of the district's dropouts (60 percent) do not exhibit the risk factor. A district that used such a risk factor would target more students who would graduate anyway than students who would drop out. Prediction requires more than simply knowing which personal and educational characteristics dropouts are more likely to exhibit (Jerald 2006).
Fortunately, some research studies do show that it is possible to predict dropouts with greater accuracy. Those studies follow individual students as they progress from grade to grade as members of a "cohort"—a group of students who start out in the same grade at the same time. By following cohorts, analysts can discover patterns that precede dropping out and identify the "high-yield" risk factors that are the best predictors of it.
Roderick (1993) followed a cohort of students entering fourth grade in Fall River, a small urban school district in Southeastern Massachusetts. She found that most dropouts follow predictable pathways on the way to dropping out and that educational experiences—academic performance and school engagement—are the best predictors of who will not graduate.
Moreover, Roderick discovered that the district had two very distinct sub-groups of dropouts that followed somewhat different trajectories—those who left school between seventh and ninth grades (whom she called "early dropouts"), and those who left during tenth through twelfth grades ("later dropouts"). Early dropouts could be predicted by low grades all the way back in elementary school. But later dropouts—whose fourth grade attendance and marks looked no different from many graduates'—could not be predicted until they entered middle or high school.
She also found that so-called "transition years" were a decisive turning point for many future dropouts. During the transition to middle school, academic performance and attendance declined somewhat for most students, but the deterioration was much steeper among future dropouts. The same thing happened later during the transition to high school.
More recent cohort studies carried out in Philadelphia and Chicago have confirmed and extended Roderick's findings. Researchers working with community groups in Philadelphia have discovered that they can identify about 50 percent of that city's eventual dropouts as early as sixth grade and a full 80 percent of eventual dropouts by ninth grade (Neild and Balfanz 2006). They also have uncovered "high-yield" risk factors at different points along the educational pipeline:
- Sixth graders with poor attendance (less than 80 percent), a failing mark for classroom behavior, a failing grade in math, or a failing grade in English had only a 10 percent chance of graduating within four years of entering high school and only a 20 percent chance of graduating a year late (Balfanz and Herzog 2005).
- Eighth graders with poor attendance (less than 80 percent) or a failing grade in math, or a failing grade in English had less than a 25 percent chance of graduating within eight years of entering high school (Neild and Balfanz 2006).
- Among entering freshmen who had exhibited no eighth grade risk factors, those who had very poor ninth grade attendance (less than 70 percent), who earned fewer than two credits during ninth grade, or who did not earn promotion to tenth grade had only a one-in-four chance of earning a diploma within eight years (Neild and Balfanz 2006).
- Based on similar cohort studies, the Chicago Consortium on School Research combined two highly predictive ninth grade risk factors to create an "On-Track Indicator" for high school freshmen. A student is considered on-track at the end of ninth grade if he or she has accumulated enough course credits to earn promotion to tenth grade while receiving no more than one F (based on semester marks) in core academic subjects. The indicator is 85 percent successful in predicting which members of the freshmen class will not graduate on time, and nearly as good at predicting who will not graduate within five years (Allensworth and Easton 2005).2 (See Table 1.)
Table 1: Examples of Highly Predictive Risk Factors for Dropping Out from District Cohort Studies
|Type of Risk Factor
||Fall River, Mass.
- Earning an "F" in English or math during 6th or 8th grade
- Failing courses and falling behind in credits in 9th grade
- Failing to earn a promotion in 9th grade
- Low grade point average in 9th grade
- Failing grades in 9th grade
- Low credits earned during 9th grade
- Falling "off track" during 9th grade; i.e., either receiving more than one semester F in core academic courses or not earning enough credits to be promoted to 10th grade
- Very low grades or attendance in 4th grade
- Significant decline in grades from 5th to 6th grade
- Significant decline in grade point average from 8th to 9th grade
- Being retained in any grade during K-8 or in high school
- Low attendance (80% or lower) during 6th or 8th or 9th grade
- Receiving a failing classroom behavior mark during 6th grade
| Low attendance during 9th grade
||Significant drop in attendance beginning in 6th grade
|Source: adapted from Jerald, 2006
Taken together, these studies offer five essential lessons that local districts should consider before they invest in intervention or prevention programs:
First, educational experiences are the best predictors of dropping out—better than race, poverty, age, gender, and personal circumstances
Although educators often believe dropping out to be driven by personal and family circumstances unrelated to schooling (Roderick 2006), most dropouts exhibit highly predictive educational warning signs. That finding supports results from surveys. For example, a federal survey revealed that dropouts are twice as likely to cite school-related reasons than family- or work-related reasons for leaving school (Berktold et al. 1998), something that held true for all demographic subgroups (Jordan et al. 1999).3(See Table 2: Why teenagers drop out).
Table 2: Why Teenagers Drop Out
Percentage of spring 2002 high school sophomores who had left school without completing a four-year program as of spring 2004, by reason for leaving school
|Reason for Leaving School
|Missed too many school days
|Thought it would be easier to get a GED
|Getting poor grades/failing school
|Did not like school
|Could not keep up with schoolwork
|Got a job
|Thought could not complete course requirements
|Could not get along with teachers
|Could not work at same time
|Had to support family
|Did not feel belonged there
|Could not get along with other students
|Was suspended from school
|Had to care for a member of family
|Became father/mother of a baby
|Had changed schools and did not like the new one
|Thought would fail competency test
|Did not feel safe
|Was expelled from school
|Got married/planned to get married
|1Percentage of female respondents only. The reason could only be selected by female respondents.
Note: This indicator shows the percentage of high school students in the spring of their sophomore year who, in the spring two years later, were not in school and had not graduated with a regular diploma or certificate of attendance. The 1 percent of sophomores who left school and earned a General Education Development (GED) certificate or other form of equivalency certificate as of the spring two years later are counted as having left school without a regular diploma or certificate of attendance.
|Source: U.S. Department of Education, National Center for Education Statistics, Condition of Education Table 27-3, from Education Longitudinal Study of 2002 (ELS: 2002/04), "First Follow-Up, Student Survey, 2004," previously unpublished tabulation (January 2006).
Finn (1989, 1993) argued that it would be a good thing if educational risk factors turned out to be better predictors because they are "alterable," as opposed to "status" risk factors that educators have little or no control over, such as poverty, gender, race, and family background. As we discuss below, one intervention program that heeded Finn's long-ago advice to monitor and address early educational warning signs has achieved marked success keeping more students in school.
Second, for most dropouts, leaving school is not a sudden or surprising event. Indeed, the vast majority (80 to 85 percent) follows observable patterns through the education pipeline, exhibiting very clear signs of educational difficulty and disengagement well before tenth grade and often prior to high school. That means schools and districts can identify most potential dropouts early enough to intervene.
Third, transition years are critical gateways on the road to graduation, and many eventual dropouts first display warning signs during the year they enter middle or high school. That is not surprising: In addition to having to negotiate a new and often larger institutional setting, students often find that coursework is more intellectually demanding while teachers are less supportive, peer relationships are more complicated, and they have more autonomy with less supervision (Roderick and Camburn 1999, Neild et al. 2001). Problems develop early, and reliable predictors of dropping out—such as declining grades or attendance—can be observed very early in the year.
Students with a history of disengagement and academic difficulty are more likely to run into problems when they transition to middle or high school, but they are not alone. About one in four students who entered Chicago high schools with high eighth grade test scores (in the top quarter) fell off track during ninth grade, and only about one-third of those students recovered to graduate on time (Allensworth and Easton 2005). Likewise, nearly one-third of Philadelphia dropouts exhibited no warning signs in eighth grade but "hit the wall" when they transitioned to high school (Neild and Balfanz 2006).
Fourth, it appears that academics and engagement both matter for predicting who is in danger of not graduating—a question about which there has been much recent confusion. Last year a non-representative survey of dropouts found that most had received passing grades but were simply bored and unmotivated by school (Bridgeland et al. 2006), prompting a spate of national news stories suggesting that academic failure plays no significant role in the dropout problem. Balfanz and Legters (2006) countered that such findings conflict with evidence from cohort studies in places like Philadelphia and Chicago—where most dropouts leave school behind in credits after failing academic courses.
The truth seems to be that academic performance and school engagement matter equally, and that they are often, but not always, intertwined. Finn (1989, 1993) argued that disengagement in the form of absences, misbehavior, and poor class participation can lead to failing grades. Simply put, students who do not "participate" enough in school—show up, pay attention, and follow the rules—are more likely to fail their classes. On the other hand, academic failure—caused either by low skills or low effort—can cause students to feel alienated from school, leading to even greater withdrawal and lack of participation over time.
Finally, it is dangerous to assume that districts can guess precisely which measures of academic performance and educational engagement will turn out to be the best predictors. "Behavior marks" given by middle school teachers in Philadelphia were much better than suspensions at predicting which sixth graders would eventually drop out of high school. In both Philadelphia and Chicago, more "subjective" measures of academic performance like classroom grades turned out to be better predictors than "objective" measures like test scores. And, while low attendance shows up as a strong predictor in virtually every study, the precise threshold for defining "low" can vary across districts or even across grade levels within the same district.
Although other districts can adopt the predictors identified in Chicago and Philadelphia, Jerald (2006) recommends that local education leaders strongly consider conducting their own cohort analyses to discover the precise "high-yield" predictors in their own school systems. Fortunately, conducting a cohort study is neither overly complicated nor prohibitively expensive. Districts can simply backtrack through student records, collecting information on cohorts who have already graduated—a process that took the Philadelphia researchers only a few months even though the district lacked an electronic student record system at the time.
Interventions: Helping at-risk students
So far research has told us much more about identifying potential dropouts than about how to keep them in school (Christenson and Thurlow 2004). Nevertheless, several targeted intervention programs have demonstrated very promising results, and important lessons are beginning to emerge. (See the sidebar called "Consumer Beware" to understand why some evaluation methods tell us much more than others.)
The most important lesson to date comes from large-scale evaluations of intervention programs funded by the federal government during the 1990s: Low-intensity programs that provide occasional tutoring, counseling, or activities to boost self-esteem—the norm in most districts—do almost nothing to keep students in school. On the other hand, some high-intensity interventions can significantly reduce dropout rates (Dynarski and Gleason 2004).
For example, an evaluation of the federal School Dropout Demonstration Assistance Program used experimental methods to assess the impact of eight local programs targeted toward middle school students. Four programs that provided low-intensity supplemental services—such as tutoring, counseling, or workshops to enhance self-esteem or leadership skills—had no impact on dropout rates (Dynarski and Gleason 1998).
The other four programs all provided more intensive services, such as smaller classes, very intensive counseling, and accelerated instruction aimed at helping overage students catch up with their peers. Two were designed as schools-within-schools and two were alternative middle schools with their own campuses. Both of the alternative middle schools—one in Atlanta and other in Flint, Michigan—dramatically reduced dropout rates and accelerated students' progress: "Compared with control group students, treatment group students admitted to these programs were half as likely to drop out and completed an average of half a grade more of school" (Dynarski and Gleason 1998).
Results from another federal initiative during the 1990s confirm that intensity matters but suggest that services need not be delivered on separate campuses in order to make a difference. In 1990, the U.S. Department of Education's Office of Special Education Programs funded three five-year projects to develop interventions for middle school students with learning or emotional abilities who were at high risk of dropping out. Collectively known as the "ABC Projects," all three interventions were designed not only to provide direct services to students but also to improve connections between homes, schools, and communities (Thornton 1995). Two of the projects, ALAS and Check & Connect, yielded some of the best results seen for any intervention to date.
Achievement for Latinos through Academic Success (ALAS)
ALAS was implemented from 1990 to 1995 in a Los Angeles County junior high school serving a predominantly low-income Latino population. The program employed counselors who provided a set of comprehensive and coordinated supports to students and parents (Larson and Rumberger 1995, Gándara et al. 1998).
Student attendance was monitored on a period-by-period basis every day, and parents were contacted daily about truancy or cut classes. ALAS counselors communicated a personal interest that students attend regularly, taking initiative to help families overcome obstacles that stood in the way, and expecting students to make up missed time. They also helped teachers establish a system of regular feedback to parents and students about behavior, class work, and homework—on a monthly, weekly, or even daily basis as needed.
Students received ten weeks of instruction in a ten-step problem-solving strategy created by one of ALAS's developers, along with two years of follow-up coaching. During those sessions, counselors discussed teachers' feedback and coached students in how to use the strategy to think through problems related to attendance, behavior, and academic progress. They also followed up with teachers to keep them informed about how students and parents had decided to address problems.
In addition, counselors provided parents with direct instruction and modeling on how to participate in their child's schooling and manage adolescent behavior, as well as helping them to connect with a wide range of community programs and social services. And they provided students with frequent positive reinforcements and group bonding activities, striving to help them feel more connected to school by showing them that caring adults were taking an interest in them.
ALAS was evaluated using an experimental study that randomly assigned students to treatment and control groups. By the end of ninth grade—the final year of the program—participating students were only half as likely to have logged excessive absences or received failing grades, less likely to have fallen behind in credits (28 percent versus 47 percent), and less likely to have dropped out of school (2 percent versus 17 percent). Students also were less likely to have dropped out by tenth grade and by eleventh grade, but differences between the control and treatment groups were negligible by senior year—suggesting that interventions must be sustained to keep highly at-risk students in school (Larson and Rumberger 1995, Gándara et al. 1998, What Works Clearinghouse 2006a).
ALAS has not been implemented beyond the initial demonstration project. However, Check & Connect—an ABC project that used similar strategies to garner impressive results—has been replicated in urban and suburban settings and adapted for implementation in elementary schools and high schools.
Check & Connect
Like ALAS, Check & Connect is a highly targeted, individualized intervention focused primarily on boosting students' engagement in school. The program's design was heavily influenced by early proponents of the theory that dropping out is a long-term process of educational withdrawal preceded by observable, "alterable" warning signs of academic difficulty and disengagement (Finn 1989, 1993)—a theory supported by findings from recent cohort studies as described above. As the program guide explains,
Alterable risk factors [. . .] are more powerful predictors of students' exit from school than status variables [like race, poverty, or family structure]. More importantly, alterable risk factors are by definition amenable to intervention by educators, youth advocates, parents, and students. Some of the most powerful predictors are student behaviors that are directly indicative of school withdrawal, such as chronic absenteeism, problem behavior, or course failures (Evelo et al. 1996).
Each student is assigned a "monitor" who acts as a cross between a case manager, mentor, problem solver, coach, and advocate (Lehr et al. 2004a). The position was modeled after the factor most commonly identified by research on resilient children—the presence of a caring, concerned adult (Sinclair et al. 2003). Monitors agree to work flexible hours so they can be "on call" outside of the normal workday all twelve months of the year, and they are employed by the district rather than by a specific school so they can stick with students who transfer. According to the program's developers, continuity is key:
The monitor's primary goal is to promote regular school participation and to keep education the salient issue. The monitor's message is that a caring adult wants the student to learn, do the work, attend class regularly, be on time, express frustration constructively, stay in school and succeed. Maintaining a focus on students' educational progress also serves to keep interventions tightly focused on those factors most amenable to change (Sinclair et al. 2003).
As the name suggests, the monitors oversee two main program components. The "check" component consists of routine monitoring of alterable risk factors. Monitors keep close tabs on each student's academic performance and school engagement on a near daily basis—tracking a common set of indicators related to absenteeism, tardies, skipping class, suspensions, detentions, behavior referrals, course failures, credit accrual, and progress toward meeting other graduation requirements. They keep written records of each indicator on a "monthly monitoring sheet" for each participant, where they can also note how problems are being addressed via intensive interventions.
The "connect" component consists of interventions and supports to address increases in risk factors identified by the checking process. In order to maximize the use of finite resources, monitors provide a form of "basic intervention" to all participants and individualized "intensive interventions" as necessary (Lehr et al. 2004b).
The basic intervention focuses on problem-solving. Monitors meet at least monthly with all participants (weekly at the elementary level) to coach them in a formal five-step problem-solving strategy. During the meeting, monitors review the monitoring sheet with students and help them use the five steps to brainstorm solutions to indicators of academic difficultly or disengagement (for example, arriving at school late too often). They also might discuss students' concerns, model problem-solving using hypothetical scenarios, and reinforce the importance of staying in school by sharing information about the long-term economic benefits of a high school diploma.
Intensive interventions are more frequent and tailored to each student's individual needs
When students exhibit behavior that signals increased risk—as recorded on the monthly monitoring sheet—monitors act immediately to address the problem and reconnect students with school. Intensive interventions generally fall into three categories: Academic supports (e.g., homework assistance or tutoring), targeted problem-solving (e.g., conflict resolution, helping students overcome barriers to waking up and getting to school on time, negotiating alternatives to out-of-school suspensions), and helping students explore recreational or community service opportunities.4
Check & Connect has been evaluated in several experimental studies. One tracked a group of ninety-four students with learning or emotional disabilities—a high percentage of whom were low-income African American males from single-parent families—from seventh through ninth grades (Sinclair et al. 1998). All students participated in Check & Connect during seventh and eighth grade. Prior to ninth grade, researchers randomly assigned students either to a treatment group whose members continued to participate in the program or to a control group whose members did not. At the end of ninth grade, students in the treatment group were significantly less likely to have dropped out (9 percent versus 30 percent), to have experienced excessive absences (15 percent versus 36 percent), and to have fallen off track to graduate within five years (32 percent versus 71 percent).
A more recent study followed 144 students from ninth grade through twelfth grade in seven urban, Midwestern high schools (Sinclair et al. 2005). The majority were male, African-American, low-income students from single parent families who had been diagnosed with behavioral or emotional disabilities. By the end of twelfth grade, students in the treatment group were only slightly more likely to have graduated from high school (30 percent versus 29 percent). However, students in Check & Connect were significantly less likely to have dropped out altogether (39 percent versus 58 percent, which translates into a one-third reduction in the dropout rate). Moreover, among a subgroup tracked for a fifth year, Check & Connect participants dropped out at less than half the rate of their peers in the control group (42 percent versus 94 percent).
Taken together, those findings suggest that an intensive, sustained intervention like Check & Connect can cut dropout rates by as much as half, but it will take additional efforts to help highly at-risk students actually complete high school in a timely manner. (See the sidebar called "Understanding Outcomes" for more information on the difference between low dropout rates and high graduation rates and why districts must address both outcomes.) Therefore, the program's developers believe that Check & Connect ideally should be coupled with schoolwide reforms (such as those discussed in the next section) designed to help students stay on track to graduate (Sinclair et al. 2003, Sinclair et al. 2005). (See Chart 1: Impact of drop out prevention programs.)
Although the elementary-level version of Check & Connect has not been evaluated using experimental methods, researchers tracked 147 participants for two years to assess whether they became more engaged in school (Lehr et al. 2004a). At the end of two years, the number of participants with fewer than two tardies per month had more than doubled, from 42 percent to 86 percent. Likewise, the number of students absent less than two days per month had more than doubled, from 17 percent to 40 percent.
Several other targeted interventions have demonstrated results based on controlled studies, even if not as rigorously scientific.
The Coca-Cola Valued Youth Program (VYP). Developed by the San Antonio-based Intercultural Development and Research Association in 1984, the Valued Youth Program has since been implemented in more than 250 schools in twenty-five U.S. cities (Montecel et al. 2004). The VYP seeks to curb dropout rates by enlisting at-risk secondary students in a cross-age tutoring program. Student tutors are paid a minimum-wage stipend to work with elementary-age children and are honored for their efforts at functions where they receive tokens of appreciation. Tutoring is meant to build self-esteem, develop independence and problem-solving skills, and encourage at risk students to attend school (Intercultural Development Research Association 2004).
The tutoring component also functions as a way to leverage other kinds of direct support for academic performance and educational engagement. In order to serve as tutors, participants must agree to attend a weekly class that bolsters their own basic academic skills in addition to teaching them tutoring strategies. Moreover, program coordinators closely monitor daily attendance and immediately follow up with phone calls or home visits when student tutors miss school. If absences develop into truancy, coordinators provide additional counseling and can help link families with social services. Finally, they meet with each participant's parents three times per year.
The program was evaluated in four San Antonio-area middle schools using a "quasi-experimental" design in which outcomes for VYP tutors were compared with outcomes for a group of randomly-selected comparison students (Fashola and Slavin 1998). After two years, only one VYP participant out of 101 (1 percent) had dropped out of school compared with eleven of the ninety-three comparison group students (12 percent). Although that 1989 study remains the only rigorous evaluation of VYP, the program's sponsor says that just 2 percent of participants have dropped out of school since its inception (Lehr et al. 2004b).
However, evaluations of another widely known program demonstrate the value of systematically replicating and evaluating dropout interventions, offering a cautionary tale districts should consider before deciding to invest large sums of money in one particular program or another.
Quantum Opportunities Program (QOP). Originally funded by the Ford Foundation as a four-year demonstration project in five cities from 1989 to 1993, QOP provides interventions through community-based organizations rather than school districts. Students are assigned a counselor who coordinates services and sticks with participants no matter what, acting as a mentor, role model, big brother or sister, disciplinarian, broker, and problem solver (Eisenhower Foundation 2006). The counselor is expected to build strong relationships with participants in order to help them achieve a wide range of goals—stay in school, keep up academically, stay out of trouble, behave responsibly, respect property and persons, complete high school, and go on to be successful adults.
The program also aims for students to spend 250 hours per year in each of three different kinds of activities hosted by the community-based sponsors—personal development, community service, and education (the last of which might include assessment of academic needs, one-on-one tutoring, and computer-assisted learning). Participants are paid a stipend of about $1.25 for each hour they spend in program-related activities (other than mentoring and recreation).
QOP is very unusual for having been systematically replicated and evaluated across multiple sites using large-scale, experimental evaluations (Stern and Wing 2004). The first study, which examined outcomes in the original five sites between 1989 and 1993, uncovered seemingly spectacular results (Hahn 1994). At the end of four years, participants in the treatment group were much less likely to have dropped out than students in the control group (23 percent versus 50 percent) and much more likely to have graduated (63 percent compared with 42 percent).
Those findings excited a great deal of enthusiasm from potential funders: The U.S. Department of Labor paid for a five-year QOP replication project in five additional cities and the Ford Foundation funded two sites. However, an experimental study of those QOP programs from 1995 and 2001 yielded much less impressive results (Schirm et al. 2003). Compared with their peers in the control group, QOP participants were only somewhat more likely to have earned a diploma (46 percent versus 40 percent), and they were only somewhat more likely to have earned a diploma or GED or to still be enrolled in school or a GED program (79 percent versus 72 percent).
One obstacle might have been the voluntary nature of some aspects of the program: Participants averaged only 174 hours per year on QOP activities, well short of the 750-hour target, and participation declined significantly each year students were in the program (Maxfield et al. 2003). Also, members of a debriefing session pointed out that the replication sites targeted students based on different risk factors: While the original sites enrolled students based on economic disadvantage regardless of their educational experiences, the replication sites targeted students based on poor academic performance—a stronger predictor of dropping out and one that some sites were not ready to handle (Eisenhower Foundation 2005).
The interventions discussed above are by no means the only formal programs available. Indeed, literally dozens of dropout interventions are being implemented in school districts all over the United States. However, very few of them have been evaluated using rigorously controlled studies, and many offer little solid evidence to back up their claims.
Several resources can provide information on additional evidence-based interventions
Last fall the federally funded What Works Clearinghouse began to list dropout programs backed by studies that meet its criteria for scientific research. (However, see the "Consumer Beware" sidebar.) So far it has listed one program, Check & Connect, as having "positive effects" on dropout rates and several programs as having "potentially positive effects" on staying in school (including ALAS and Career Academies, described below). Lehr et al. (2004b) provide profiles of eleven evidence-based intervention and prevention programs, including five not discussed in this paper.5
Prevention: Changing the factors that schools control
As long as twenty years ago, some researchers became concerned that educators too often viewed dropping out as a problem unrelated to schools, a social phenomenon they could do nothing about (Whelage and Rutter 1986). Researcher Melissa Roderick observed that tendency first-hand during her tenure as director of planning for the Chicago Public Schools (Roderick et al. 2004): "Educators argued vehemently that differences in the dropout rate across high schools were simply a reflection of differences in the students they served, and were not a result of any actual differences in the quality of a school's programs, teachers, or administrators."
But research has now challenged that assumption. Institutional factors matter at least as much as individual characteristics, and some schools have much greater "holding power" than others. For example, Allensworth and Easton (2005) found that dropout rates varied widely across Chicago high schools—even after they controlled for a host of individual risk factors, including race, gender, prior academic achievement, family socioeconomic status, and whether students are overage when they enter ninth grade.
Holding power seems to have a lot to do with whether schools ameliorate or exacerbate the stress of transition years. Roderick and Camburn (1999) found that rates of ninth grade course failure and recovery from first semester failure varied widely across Chicago high schools—above and beyond what would be expected based on individual risk factors. Their analysis found that only 30 percent of the overall variation in ninth grade course failure across high schools could be explained by differences in their "intake," that is, the characteristics of entering freshmen.
Other researchers have begun to piece together exactly how schools affect graduation rates. Interestingly, just as with individual risk factors, certain school characteristics that are "alterable," such as curriculum and teacher-student relationships, turn out to have a much bigger impact on school completion than other factors that educators would be unable to change, such as the demographic makeup of the student body and whether a school is public or private (Lee and Burkam 2003).
Generally speaking, the school characteristics that boost "holding power" fall into two broad categories—supportive environments and academic challenge. More specifically, researchers have found that students who attend high schools that have enrollments lower than 1,500, better interpersonal relationships among students and adults, teachers who are more supportive of students, and a more focused, academically rigorous curriculum tend to drop out at lower rates (DeLuca and Rosenbaum 2000, Croninger and Lee 2001, Lee and Burkam 2003).
The positive impact of attending a school with a more supportive environment is especially big
Croninger and Lee (2001) found that, other things being equal, high schools whose teachers are highly supportive of students manage to cut the probability of dropping out nearly in half. The finding held equally true for students at low, medium, and high risk of dropping out.6 On the other hand, academic challenge also seems to play a big role, which might surprise many observers in- and outside of schools who believe there is a zero-sum tradeoff between higher academic rigor and higher graduation rates (Roderick et al. 2004).
Lee and Burkam (2003) found that high schools offering a more focused and rigorous curriculum—composed of mainly academic courses with very few remedial or non-academic courses—have significantly higher graduation rates, other things being equal. In fact, for every two additional math courses that high schools offered below the level of Algebra I, students experienced more than a 30 percent increase in the odds of dropping out. "This finding flies in the face of those who say that high schools must offer a large number of undemanding courses to keep uncommitted students in school," the researchers concluded.
At the same time, other studies lend support to the idea that curricula should be engaging and relevant to students' interests or career plans. For example, a team of Johns Hopkins University researchers found that career and technical education (CTE) can boost graduation rates for some students, especially in combination with rigorous academic courses (Plank et al. 2005). The study found that the ideal ratio appears to be one part career or technical coursework to two parts academic coursework.
The emerging consensus among researchers seems to be that high school curricula should be both challenging and engaging, and that schools should do a better job providing students with the academic support they need to master challenging material. Indeed, "remedial" or "general" courses are often both unchallenging and unengaging. As one Philadelphia teenager summed up when asked why students drop out, "A lot of the teachers give you one worksheet a day and after that, you're sitting there the rest of the day. Students want to go to school and learn things" (Snyder 2003).
Interestingly, Dynarski (2004) observed that intensive dropout interventions that were more effective at curbing dropout rates tried to challenge students academically rather than making classes easier for them: "The programs recognized that students needed a measure of academic challenge, that even students with undistinguished academic records could respond to teachers pushing them to learn, especially when learning somehow was connected to their personal experiences."
While we still do not know enough about exactly how to put all of those lessons into practice, research has revealed that, under the right circumstances, taking action can make a difference:
Ninth grade supports
As discussed above, students often have difficulty making a successful transition to high school, and many eventual dropouts experience sharp declines in grades and engagement after they enter ninth grade. High schools tend to be larger, more bureaucratic institutions that are more academically and socially demanding than students are used to. Research has found that restructuring ninth grade to create a more supportive experience can help curb dropout rates.
The School Transitional Environmental Program (STEP) assigns at-risk students to homerooms in which all their classmates are program participants. Students from the same homeroom take all of their core academic courses together in one part of the school. Homeroom teachers take on the roll of guidance counselors, administrators, and instructors, helping students handle the educational and personal challenges of ninth grade. The program was evaluated using several quasi-experimental studies that tracked STEP participants and a randomly selected comparison group over time. After five years STEP participants were much less likely to have dropped out compared with those in the control group (21 percent versus 43 percent) (American Youth Policy Forum 1998).
Kerr and Legters (2004) found that Maryland high schools grouping ninth graders into interdisciplinary teaching teams had significantly lower dropout rates, all else being equal.7 High schools that used small learning communities (or schools-within-schools or "freshman academies") in ninth grade also had lower dropout rates—especially if they combined that with other reforms, such as extra academic support. Between 1994 and 2000, Maryland high schools making widespread and sustained use of interdisciplinary teams experienced a 50 percent decline in dropout rates, and those using small learning communities cut their dropout rate by nearly two-thirds.
A number of comprehensive schoolwide reform models combine those two strategies with other supports. Two models that have been rigorously evaluated by the research firm MDRC have shown promise for reducing dropout rates:
One of the oldest high school reform models, career academies can take the form of small, stand-alone high schools or—in the version aimed most specifically at curbing dropout rates—of small learning communities contained within larger high schools. Career Academies combine academic coursework with career and technical coursework, and they give students the chance to work with local employers.
MDRC evaluated the model's impact on dropout rates using an unusually sophisticated and rigorous experimental study (Kemple and Snipes 2000). Researchers followed 1,764 students who applied to a Career Academy, 959 of whom were accepted for admission and 805 of whom were not. The admissions were based on a random lottery, ensuring that the treatment group and the control group were alike demographically, academically, and even motivationally. In addition, the researchers tracked a subgroup at very high risk of dropping out based on signs of poor academic performance or educational engagement.
By the end of their expected senior year, high-risk members of the Academy group were less likely to have dropped out than high-risk members of the control group (21 percent versus 32 percent, which translates into a one-third reduction in the dropout rate). (See Chart 2: Targeted interventions can reduce drop out rates of high-risk students.) Researchers also examined which components of the Career Academy design helped explain the positive impact. The clear winner was greater support. Students who received especially high levels of support from teachers and peers during ninth or tenth grade were much less likely to exhibit chronic absenteeism or drop out of high school, and career academies that did a poor job enhancing support actually increased dropout rates for some students.
However, the Career Academies had no differential impact on long-term completion rates. When researchers checked in on the cohort four years later, high-risk members of the Academy group were no more likely to have earned a diploma than members of the control group (65 percent versus 64 percent). Taken together, those findings imply that Career Academies have stronger "holding power" than regular high schools, but that their lower dropout rates do not necessarily translate into higher long-term completion rates. (See the sidebar called "Understanding Outcomes" for an explanation of the differences between dropout rates and completion rates.)
Chart 2: Targeted interventions can reduce drop out rates of high-risk students
Talent Development high schools
Developed and supported by Johns Hopkins University, the Talent Development model was designed to improve achievement and graduation rates in high-poverty urban high schools where many students enter ninth grade one or more years behind grade level in math and reading. Talent Development reorganizes schools into several small learning communities to create a more supportive learning environment with better relationships among teachers and students. Ninth grade is restructured as a self-contained school-within-a-school called the Success Academy, where groups of students share the same teachers in interdisciplinary teams. After ninth grade, students can choose from among several differently themed Career Academies that blend career and technical coursework with a rigorous college-prep curriculum.8
The model also incorporates intensive academic supports. During the first semester, ninth graders take three courses designed to enable them to overcome poor preparation and succeed academically—Strategic Reading, Transition to Advanced Mathematics (TAM), and a Freshman Seminar that develops personal and educational survival skills. Unlike traditional "remedial" classes, Strategic Reading and TAM use specially developed curricula designed to accelerate learning and get students caught up academically. Ninth graders then take regular college-prep courses like Algebra I during the second semester.
In order to get students caught up and on track in one year, Talent Development uses block scheduling to double the amount of time ninth graders spend in math and language arts. Ninety-minute math and reading classes during the first semester allow teachers to fit in a full year's worth of catch-up instruction, and "double-dosing" on algebra and English during the second semester allows students to complete a full year's worth of credit-bearing coursework. In this way, students do not fall off track to graduate. To prevent students from becoming bored in ninety-minute classes, Talent Development developed curricula that incorporate highly motivational materials and activities designed to appeal to teenagers.
That combination of intensive social and academic supports seems to be paying off. In Philadelphia, a group of neighborhood high schools replicating Talent Development have seen substantial gains in attendance, academic credits earned, and promotion rates for several cohorts of ninth graders (Kemple et al. 2005). In contrast, a group of matched comparison high schools showed little improvement in outcomes over the same period. The researchers are still measuring the model's impact on dropout and completion rates, but so far the impact seems positive.
At the same time, even with such intensive supports, more than one-third of freshmen still fell off track and did not earn promoting to tenth grade. Such results have led Talent Development officials to believe that districts (especially those in big cities) must also focus attention on better preparing students academically before they reach ninth grade (Balfanz and Legters 2004).9
Better preparation for high school. Research suggests that improving the instruction students get during kindergarten through eighth grade would prevent problems during transition years and have a positive impact on graduation rates (Finn et al. 2005, Prevatt and Kelly 2003).
Roderick (2006) contends that Chicago provides a good case study that shows improving basic skills prior to high school can help improve graduation rates. Eighth grade achievement began rising during the 1990s due to a host of reform efforts being implemented by the Chicago Public Schools, and high school dropout rates began declining. Although part of that reduction in dropout rates can be explained by other factors, most can be explained by improvements in the academic skills of students leaving middle schools (Allensworth 2004).
However, several cautions are in order:
One policy meant to ensure students are prepared before they enter high school might have negative effects on graduation rates—that is, the policy to end "social promotion" by holding back students who do not pass standardized tests. Being retained at any point prior to high school and becoming overage for grade places students at greater risk of dropping out (Roderick 1994). Based on an analysis of the "net effects"—both positive and negative—of Chicago's social promotion policy, Allensworth (2004) concluded, "For students retained under high-stakes testing, the adverse affects on dropping out from retention outweighed the beneficial effect of rising achievement."
Moreover, one well-known program that attempts to raise graduation rates by improving early academic preparation has met with mixed success. Project GRAD seeks to improve "feeder patterns" by providing professional development and strong curricula in elementary and middle schools that send students to Project GRAD high schools. In the program's flagship high school, Project GRAD had a positive impact on the proportion of students completing a core academic curriculum within four years. However, several replication sites did not see positive impacts (Snipes et al. 2006).
Finally, while it might be tempting to hope that raising math and reading achievement in K–8 will cause the dropout problem to disappear, the truth is more complicated. Findings from cohort studies described above suggest that sending better-prepared students to high schools will help solve some of the problem, but by no means all of it. Engagement matters too, and even high-achieving students can experience rocky transition years. Allensworth notes that in Chicago, ninth grade absences are twenty times more predictive of eventual graduation than eighth grade test scores (Education Week 2006).
Recovery: When prevention and intervention are not enough
Effective intervention and prevention strategies can significantly reduce the number of students who develop risk factors and drop out. However, in larger districts with very high proportions of at-risk students and very low graduation rates, those measures might not be enough. Even the best interventions and preventions take time to establish, and so far none has proven to be 100 percent effective. Recovery programs can provide a second-chance "backstop" for students who drop out despite schools' best efforts. Most dropouts deeply regret their decision to leave school (Bridgeland 2006), and many later attempt to earn a diploma or GED (Berktold et al. 1998).
Moreover, not all dropouts can be identified based on early educational warning signs. So far, researchers have been unable to reliably predict about 15 to 20 percent of eventual dropouts in cities like Chicago and Philadelphia. Those students exhibit no early warning signs and tend to drop out in the upper grades; many have earned the majority of credits needed for a diploma. The researchers speculate that many members of this group might be "life event" dropouts who leave because of premature transitions to adulthood, such as work or child care responsibilities, and find it difficult to attend school full-time (Roderick 2006, Allensworth and Easton 2005, Neild and Balfanz 2006).
The first step in considering whether and how to implement recovery programs is to analyze data to determine the size and characteristics of the young adult population who could benefit. In general, students who need recovery help will fall into two sub-groups with different needs:
Dropouts who did not fall severely behind in credits before leaving high school: For example, about 15 percent of the twenty-six, 301 teenagers who dropped out of Chicago Public Schools over the course of the 2001, 2002, and 2003 school years, left after making it to eleventh grade having earned the majority credits necessary for a diploma (Roderick 2006). Those older teenagers would need flexible full-time or part-time alternatives to traditional, comprehensive high schools.
Teenagers who have become overage for grade and have fallen severely behind in credits: For example, an analysis in New York City uncovered 138,000 overage and under-credited young people between the ages of sixteen and twenty-one, about half of whom had dropped out and half of whom were still enrolled in school (New York City Office of Multiple Pathways to Graduation 2006). Such students need more intensive programs that can help them expeditiously earn credits toward a diploma.
Proponents of dropout-recovery strategies recommend that districts create "multiple pathways to graduation" for such students. The New York City Department of Education recently created an Office of Multiple Pathways (OMP) to oversee an expanding portfolio of second-chance options, including Young Adult Borough Centers for teenagers who have at least half the credits necessary to graduate and Transfer High Schools, small, academically rigorous schools designed for teenagers who are overage and under-credited.
So far, little to no formal research has been conducted on recovery programs like those just mentioned. An analysis conducted by New York City's OMP found that Transfer High Schools have a graduation rate of 56 percent, compared with a district wide rate of 19 percent for overage, under-credited youth in regular high schools (Office of Multiple Patheways 2006). The district hopes to maintain such outcomes as it expands the number of Transfer High Schools over the next few years.
However, some research studies on alternative schools elsewhere have shown mixed results. The research firm Mathematica conducted an experimental study of two alternative schools—one in California and one in Kansas—using random assignment of applicants (Dynarski and Wood 1997). At the end of four years, one school showed a positive impact on graduation rates for the treatment group compared with the control group (17 percent versus 11 percent), but the other school did not. The researchers concluded that alternative schools can help more students graduate at higher rates, but will not automatically do so.
Implementation: Things to consider
While the dropout problem is daunting, research conducted over the past decade should dispel the notion that schools are powerless to do anything about dropping out and encourage optimism that districts can help many more young people stay in school. We now know it is possible to identify most dropouts by ninth grade and many well before that. We have examples of interventions that can significantly reduce dropout rates even for groups of students at very high risk of dropping out. And we have seen how educators in even the toughest urban neighborhoods can ease the transition to high school and keep more students on track to graduate.
But that optimism should be tempered with caution when it comes to crafting specific strategies for tackling the problem. Districts should keep the following lessons in mind as they craft comprehensive plans to raise graduation rates.
Experimental studies have provided solid evidence that early interventions must be more intensive than the programs typically implemented by schools and districts. Moreover, the interventions that work best incorporate multiple strategies rather than relying on one kind of support (for example, tutoring or mentoring). The best-proven interventions tend to have many of the following elements (Anderson et al. 2004, Lehr et al. 2004b, Dynarksi 2004):
- Highly personalized supports and services. Strong relationships with adult counselors who pay a great deal of attention to students.
- Systematic strategies to monitor and address "alterable" risk factors. Formal coaching in specific problem-solving strategies.
- Substantial communication with and support for parents.
- Connections between schools, families, and community services—while managing to keep the primary focus on educational progress.
Moreover, some effective interventions address low academic skills through highly structured "catch-up" instruction (Dynarski and Gleason 1998, Fashola and Slavin 1998). Finally, the research strongly suggests that even the best available early interventions cannot inoculate students against academic problems and disengagement as they move into and through high school, but rather must be sustained until students graduate.
The "relationships" aspect of intensive intervention should not be underestimated. In a study of elementary and middle school participants in Check & Connect, Anderson et al. (2004) found that stronger relationships between monitors and students produced bigger improvements in attendance and other forms of engagement, everything else being equal. The researchers note that, "Other than Check & Connect, we are aware of only a handful of relationship-based interventions [. . .]. It should also be noted that while there are a number of mentoring programs targeted to youth, many of these are community- or peer-based and/or do not have structured intervention components with a focus on educational outcomes."
This is a key point: Unlike many programs that use counselors or mentors merely to provide role models or enhance self-esteem, Check & Connect uses them to leverage a more intensive set of targeted interventions and supports. Monitors not only "bond" with students but also take a relentlessly "data-driven" approach to tracking their school performance and engagement. This allows monitors to intervene rapidly at the very first sign of trouble rather than waiting for students to come to them after they land in serious hot water, as well as to tailor support to students' specific needs. It also allows them to keep everyone focused on student's educational progress toward graduation.
As Larson and Rumberger (1995) conclude, for students at high risk of dropping out, "interventions must be intensive, comprehensive, coordinated, and sustained. Anything less is naïve and will sow only marginal results."
Of course, coordinated, comprehensive, and intensive interventions also are usually more expensive. A program like ALAS would cost about $1,185 per student per year in 2005 dollars (What Works Clearinghouse 2006). Check & Connect costs about $1,600 per student per year (Social Programs that Work 2006). However, that does not mean intensive interventions "aren't cost-effective." In fact, while low-intensity interventions that provide some mentoring, homework help, or occasional counseling are cheaper and easier to implement, they might not be cost-effective at all in the long run because they do not reduce dropout rates.
While conserving scarce resources is always important, districts should try to avoid an all-in-one-basket approach to addressing the dropout problem. Just as in dentistry, interventions are often necessary to repair teeth and fill cavities. But prevention in the form of good brushing and flossing, a healthy diet, fluorinated water, and regular check-ups can help prevent problems from developing in the first place. Keeping more teeth in your mouth to a ripe old age requires both prevention and intervention, and the same is true when it comes to keeping more students in school and on track to graduate.
A district's preventive measures should pay particular attention to transition years—especially ninth grade—and target high schools with weak holding power for comprehensive schoolwide reforms. Those reforms should balance rigorous and relevant coursework with high levels of social and academic support. Districts should work to boost academic achievement to better prepare students for high school, but at the same time keep in mind that better K-8 achievement is no panacea.
Finally, districts should take a data-driven approach to tackling the problem. Investing in data up front will garner a much bigger "bang for the buck" from their intervention, prevention, and recovery efforts. By no longer targeting the wrong students with the wrong solutions, districts will avoid wasting dollars. By targeting the right students with the right interventions and the right schools for preventive reforms, districts will sow bigger increases in their high school graduation rates. And by improving graduation rates districts will reap a wide range of long-term social and economic benefits for their communities—saving taxpayers money, expanding tax revenues, boosting employment, reducing crime, and improving citizenship.
1While people often use the terms "intervention" and "prevention" interchangeably, in this paper we use the term "interventions" to refer to services and programs targeted toward individual students at high risk of dropping out and "prevention" to refer to reforms that fix the ways schools exacerbate risk factors or cause students to develop them in the first place.
2The Chicago Public Schools adopted the indicator and now publicly reports each school's overall on-track rate in its annual High School Scorecards.
3That is not to argue that educational factors are the only "ultimate causes" of dropping out, which is a very different question from how educators can predict and intervene in the process (Neild & Balfanz, 2006). Students with certain social and demographic characteristics and from certain family backgrounds certainly are more likely to drop out, but before they do, the vast majority of them develop very clear education-related warning signs.
4Although specific interventions are based on the judgment of monitors working with students and parents, the Check & Connect manual provides suggested thresholds for considering a student at high risk on each indicator the program tracks (e.g., three or more absences per month) as well as ideas for solving problems related to each risk factor (Evelo et al. 1996).
5Interpersonal Relations/Personal Growth Class, Preventing School Dropout Beginning in Elementary Grades (a behavior modification program), Project COFFEE, Support Center for Adolescent Mothers (Family Growth Center), and Teen Outreach Program (TOP).
6Those findings would not surprise ALAS counselors or Check & Connect monitors, who found that schools can be inflexible and unsupportive, forcing program personnel to act as advocates on behalf of families and students. For example, administrators often preferred to assign out-of-school suspensions for misbehavior even though such forms of discipline tend to depress academic performance and engagement. Founders of both programs have argued that targeted interventions must be coupled with schoolwide reforms to create more supportive environments in order to substantially raise graduation rates (Larson and Rumberger 1995, Sinclair et al. 1998).
7Such teams are made up of teachers across different subjects who share a common group of students.
8Students who have special needs or who transfer in mid-year can take advantage of an after-school Twilight Academy.
9Johns Hopkins has developed and is currently testing a Talent Development Middle School model.
This document was written by Craig Jerald, president of Break the Curve Consulting, located in Washington, D.C. Jerald was previously a principal partner at the Education Trust, an advocacy and research organization, and a senior editor at Education Week.
Posted: April 5, 2007
©2007 Center for Public Education