A Research Report from the School Dropout
Demonstration Assistance Program Evaluation
MPR Reference No.: 8014-140
HOW CAN WE HELP?
HAT WE HAVE LEARNED FROM
EVALUATIONS OF FEDERAL DROPOUT-
June 30, 1998
Submitted to: Submitted by:
U.S. Department of Education Mathematica Policy Research, Inc.
Planning and Evaluation Service P.O. Box 2393
600 Independence Avenue, S.W. Princeton, NJ 08543-2393
Room 4168, FOB-10 (609) 799-3535
Washington, DC 20202-4246
Project Officer: Project Director:
Audrey Pendleton Mark Dynarski
Dropping out of school is easy. Students who have done it say they simply stopped going to
school one day. Some said they dropped out because they thought school principals or teachers
wanted them to. Others said they dropped out because of circumstances outside their control. Either
way, they may have encountered little resistance from others around them.
As a society, we do not want students to drop out. We know that students who do probably are
not prepared for what happens to them afterward. Most dropouts will not work as much as students
who finish school and will not earn as much when they do work. Economic trends are likely to make
this worse rather than better. Dropouts are more likely to depend on public assistance, to use drugs,
to be arrested and spend time in jails or prisons. We want students to succeed in school and in adult
life. Dropping out is a signal that a young person has not succeeded in school and may not succeed
in adult life. But can dropping out be reduced or prevented?
Local school districts have long operated dropout-prevention programs, but these districts do
not often conduct large-scale evaluations to study the effectiveness of their programs. Beginning in
the late 1980s, the U.S. Department of Education conducted three large evaluations of the
effectiveness of programs designed to reduce dropping out. The programs and the evaluations were
supported by funds from the Carl Perkins Vocational Education Act (under the Cooperative
Demonstration Program) and two phases of the School Dropout Demonstration Assistance Program
(SDDAP), one operating from 1989 to 1991, the other from 1991 to 1996.
The evaluation of the second phase of SDDAP on which this summary focuses, was the longest
and largest of the three evaluations. It was conducted by Mathematica Policy Research, Inc., with
Policy Studies Associates and RMC Research. It looked into how dropout-prevention programs
operated, how programs used their funds, what kinds of students attended the programs, and whether
programs improved student outcomes. More than 20 programs around the country were part of the
evaluation, and more than 10,000 students--some who attended the programs and some who did not--
were followed for two to three years so that program effects could be measured.
The key finding from the evaluation is that most programs made almost no difference in
preventing dropping out in general. Dropping out is as hard to prevent as it is easy to do. Some
outcomes improved for some students, but no program was able to improve all the key education
outcomes such as dropping out, attendance, test scores, and grades. This finding is consistent with
findings from the other two evaluations.
On the other hand, a few programs did make a difference on some outcomes. Three programs
preparing students who had already dropped out to get the General Education Development
certificate improved GED completion rates. An alternative high school on a community college
campus reduced dropout rates. Several alternative middle schools also reduced dropout rates. And
efforts to restructure whole schools so that students are less likely to drop out improved test scores
when the focus was on changing what happened in the classroom. Educators and policymakers
continue to address the dropout issue and these results from the evaluation can shape their thinking.
If we as a society want to encourage more students to complete high school, we need to continue
trying new approaches and ideas that may work better. A starting point for a new approach is to
consider why some programs have an effect while others do not. Program successes may be rare
because of the difficulty of matching program designs with students those designs can help.
Programs that succeed simply may be the right blend of activities, approaches, and supports for their
students. Using a specific program approach, such as creating a school within a school or an
alternative school, is fundamentally a one-size-fits-all solution that is in conflict with the many
different kinds of students and the many different reasons they have for dropping out. We should
not have much confidence in a doctor who treated all patients with the same medication (“take two
aspirin”) though patients complained of different problems. We would have more confidence in a
doctor who first tried carefully to diagnose problems before prescribing treatments.
Likewise, we should not have much confidence in a dropout-prevention program that treats all
students in the same way. The program may be an effective treatment for some students and an
ineffective treatment for others. Trying to understand the particular characteristics of individual
students and using this understanding as a basis for developing interventions for struggling students,
may be a useful approach for program designers to consider.
Schools that diagnose student problems well currently may not be able to pull together the many
kinds of resources necessary to intervene successfully. Schools may need to integrate better with
other service providers to be able to help students, but we do not yet have any good examples of this
to point to. However, this may be the most promising direction for helping to reduce dropping out.
What Dropout-Prevention Programs Do
To understand dropout-prevention programs, it is useful to understand program objectives and
what programs do to meet them. It is also useful to know more about how programs allocate their
resources and the kinds of students they serve. Detailed descriptions of the programs and students
they serve are found in evaluation reports (Adelman and Rubenstein 1994; Hershey et al. 1995;
Rosenberg and Hershey 1995; Gleason and Dynarski 1994; Gleason and Dynarski 1995). Table 1
provides a brief description of programs that were part of Mathematica’s evaluation.
Most programs in the evaluation of the second SDDAP expressed their objectives simply.
Program staff wanted students to attend school more and learn more because of their experiences in
the program, which ultimately would result in students being less likely to drop out. Many programs
also wanted to build students’ self-esteem and their ability to cope with challenges and problems in
a mature way. Program staff frequently expressed the view that students had had too many negative
experiences in school and that programs should provide positive experiences. The main objective,
though, was to get students to come to school. Programs felt that they needed students to come to
school for other program services to matter at all.
Successfully completing a program meant different things, depending on the program. For
programs operating in middle schools, successful completion generally meant that students continued
on to and were better prepared for high school. Some middle school programs were trying to
accelerate their students so they could catch up with their age peers, and successfully completing the
program might mean compressing two years of middle school course work into a single school year.
Successfully completing programs for students of high school age generally meant that students
received either a high school diploma or a GED certificate. In the evaluation, these two ways of
SDDAP PROJECTS INCLUDED IN THE IMPACT EVALUATION
Program/Location Sponsor Project Approach
Middle School Projects
COMET Program CBO with school district School within a school
Twelve Together Program School district with local foundation Weekly peer discussion groups with volunteer
Chula Vista, CA counselors
Up with Literacy School district Tutoring and homework assistance, counseling
Long Beach, CA
Early Identification and Intervention Project School district General studies class for homework assistance
Rockford, IL counseling
Middle School Leadership Program CBO with school district Leadership workshop
Project ACCEL School district School within a school
Accelerated Academics Academy School district Alternative school
Griffin-Spalding Middle School Academy CBO Alternative school
High School Projects
School-Within-a-School at Wells Academy University/ School within a school
Chicago, IL school district
Corporate Academy CBO with school district Alternative school
Middle College High School School district Alternative school
JFY High School and University High School CBO Alternative school
Horizon High Schools School district Alternative school
Las Vegas, NV
Student Training and Re-Entry (STAR) Vocational school district Nine-week reentry program on vo-tech campus
Metropolitan Youth Academy CBO GED program
St. Louis, MO
Flowers with Care CBO GED program
Spruce High School and Middle School Southwest Texas State University Staff development, school-based decision-
Dallas, TX with Dallas Independent School making, health clinic and child care center
Ottawa Hills High School and Iroquois Grand Rapids Public Schools Outcomes-based curriculum, dropout-prevention
Middle School services
Grand Rapids, MI
Gratz High School and Gillespie Middle School District of Philadelphia School councils, staff development
Central High School Phoenix Union High School District Ninth-grade enclave, staff development, dropout-
Phoenix, AZ prevention services
Century High School Santa Ana School District Interdisciplinary instruction, staff development,
Santa Ana, CA dropout-prevention services
completing high school were associated with different program structures. Programs leading to
diplomas typically were set up by school districts. They were similar to regular high schools, except
that the schools often were much smaller, enrolling no more than about 400 students. Programs
leading to GEDs typically were set up by community-based organizations and were smaller than
diploma programs, enrolling no more than about 100 students. Programs leading to diplomas tended
to last longer than GED programs, especially for students who had not progressed far in school.
GED programs were more often for students who had already dropped out, and they were structured
to allow students to pace work on their own and take the GED test whenever they were ready.
All programs in the evaluation shared two features. Programs tried to help students overcome
personal, family, and social barriers and problems that interfered with their ability to go to school
and do well there. Programs also tried to create smaller and more personal settings in which students
could feel secure and learn more effectively. The two features overlap somewhat. For example,
counselors in some programs spent part of their time helping students deal with barriers and part of
their time helping students with their studies.
Counseling Was a Primary Component of Dropout Prevention Programs
Counseling was the primary tool that programs used to help students deal with barriers and
problems affecting progress in school. Nearly all programs offered access to counselors as one of
their services, and the type of counseling offered was more geared toward intervention than that
offered by counselors in regular schools. Some programs had counselors on staff. Others linked
students with counselors if the need arose. Some programs had students meet with counselors
regularly (weekly or monthly, for example). Others had counselors on staff who met with students
as the need arose. Programs used other types of staff (such as community liaisons or mentors) who
also were able to help students, but this was much less common than using counselors.
What programs were doing in providing access to counselors highlights an important aspect of
what they could not do. Working from within schools or community organizations, counselors and
programs in general could not address large-scale community problems such as unemployment and
crime, and they could only mitigate family problems such as poverty and abuse. Research often has
shown that communities and families strongly influence academic achievement. Programs to keep
students in school that do not improve communities or family situations implicitly are assuming that
what happens to students in school contributes importantly to whether students drop out. However,
student problems that programs could address may have been minor compared to student problems
they could not address.
An alternative approach might focus less on schools and more on communities. Neighborhood
development efforts, enterprise zones, and comprehensive service initiatives such as the U.S.
Department of Labor’s “Youth Fair Chance” program (Corson et al. 1996), which target whole
neighborhoods or cities and involve a wide range of organizations within these areas, also have the
potential to reduce dropping out. Whether they do so, however, is a question that is only beginning
to be addressed.
Wehlage et al. (1989) explore the literature and provide case studies of small schools.
There’s a Program You Can Go To
Some of the schools just want to get you out. . . . For them, it’s ‘there’s a program you can go to’
and they push you to go into it so that you can leave their school.
Student, alternative high school, Boston
This math teacher told me, ‘Hey, you know what? If I were you I’d drop out of school, “cause no
matter what you do you’re not gonna pass . . . and it really depressed me because I love math.
She used to tell me every day, ‘I don’t know what you come to school for.’
Student, alternative high school, Miami
Smaller School Settings Were a Primary Component of Dropout-Prevention Programs
The other important component of dropout-prevention programs was an attempt to create
smaller school settings. Educators and researchers have long noted that students in large schools can
become alienated and uninterested to the point where they feel little attachment to school and
eventually drop out. Smaller and more personal school settings are the prescription that comes from
Some programs created smaller and more personal settings by setting aside time during a school
day where students in the program could meet in small groups to work on academic skills, or meet
with mentors. Others did so by setting up alternative schools that were smaller than schools that
students otherwise would have attended or setting up schools within schools. The alternative schools
and school within schools programs generally served no more than 400 students at a time, and some
schools had fewer than 100 students. Alternative schools and schools within schools generally also
had more teachers and staff per student than regular schools. The result was a small-school setting
where teachers and staff could give students more attention and, in theory, help them do better in
More Was Spent on Students in Dropout-Prevention Programs
Providing access to counselors and creating smaller school settings had a clear implication for
resources. Students had more spent on them as a result of being in the programs. Program budgets
and enrollments varied widely, but the median program had an annual budget of $650,000 to
$700,000 and a median student enrollment of 300, implying average spending per student of roughly
$2,200 (Adelman and Rubenstein 1994). This amount represented an increase of about a third over
what students normally would have had spent on them (Rosenberg and Hershey 1995). Programs
Everybody Knows You
Teachers know you and everybody knows you. And the teachers will say “Oh, I know you can do
better” because they know who you are.
Student, Corporate Academy, Miami, Florida, talking about her experience at the Academy
You gotta stand up, in the back of the room. There is no way I am gonna stand up to go to school.
Student, Corporate Academy, Miami, Florida, talking about crowding in her previous high school
They give you a lot of help in this program, better than in regular school.
They’re more dedicated to their work.
Students at JFY Academy, Boston, Massachusetts
devoted about three-fourths of their resources to staff, allocating the resources to teachers (31
percent), counselors (21 percent), administrators (18 percent), aides (17 percent), and others (13
The extent to which programs increased spending differed by the type of program. Programs
that added services to regular school (such as programs to build self-esteem) increased resources by
moderate amounts. Programs that reduced class sizes were associated with the largest increases in
spending on students, and programs that set up alternative schools often had the smallest increases
in spending. A program called Career Opportunities Motivated Through Educational Technology
(COMET) in Miami, Florida, operated with class sizes of 15, about half the size of regular classes
in comparable district schools. The evaluation estimated that students who participated in the
COMET program had 110 percent more spent on them than comparable students not in the program.
Students in the Metropolitan Youth Academy in St. Louis, a small program preparing students for
the GED, had about eight percent less spent on them than students in other district schools. The
Academy had higher staff-to-student ratios than regular high schools. However, it used adult-
education instructors who were paid less than teachers in regular high schools and did not offer other
services (such as libraries, sports teams, clubs, and staff professional-development opportunities)
found in regular high schools.
Most Students in the Programs Were at Risk of Dropping Out
The second phase of SDDAP supported two types of dropout prevention programs: one focused
on schools, the other on students. The first type--called “restructuring” programs--operated in
schools that had many students who were dropout-prone, and focused on changing the schools to
reduce dropping out. The second type--called “targeted” programs--operated in schools or
community organizations but on a smaller scale, identifying students who were dropout-prone or
who had dropped out and working with them to try to help them complete middle school or high
school. Targeted programs tried to get help to students who needed it, whereas restructuring
programs tried to make targeted programs unnecessary by improving schools and school outcomes
for all students.
Both program types mostly served students who were “at risk” of dropping out (Table 2).
Students in targeted middle and high school programs, and in restructuring middle and high schools,
commonly had several risk factors. These included living in a single-parent household, having
parents or a sibling who did not complete high school, being home alone more than three hours a
day, having low income or being on public assistance, or having limited English proficiency.
Overall, 74 percent of students in targeted middle school programs and 94 percent of students in
targeted high school programs had two or more risk factors, compared to 37 percent of 8th graders
and 31 percent of 10th graders nationwide (Gleason and Dynarski 1994). The proportions are similar
for students who were attending restructuring schools.
At-risk students had other characteristics that do not conform to stereotypical notions of a school
dropout (Table 2). Nearly all program participants (89 percent in the targeted middle school
programs and 84 percent in the targeted high school programs) felt they probably would graduate
from high school or were very sure they would graduate. Some had excellent attendance records and
high grades: 60 percent of students in the targeted middle school programs were absent fewer than
10 days in the previous school year. Many had high aspirations for how far they wanted to go in
school and in their careers. Seventy-five percent of students in targeted middle school programs said
they wanted to get a college degree or a postgraduate degree, compared to 66 percent of eighth
graders nationwide. Fifty-two percent indicated they expected to have jobs when they were 30 years
old that would be classified as managerial or professional (usually the highest-paying jobs),
compared to 39 percent of eighth graders nationwide.
Programs often use risk factors to identify appropriate students to serve, but programs used risk
factors in a simple way. Students with risk factors were more likely to be considered for programs,
and offsetting factors were almost never considered. Because risk factors on their own may not be
good predictors of dropping out, programs may not have the best information for determining who
they should serve.
Risk Factors Did Not Accurately Predict Dropping Out
The coexistence of risk factors with factors associated with high academic achievement suggests
that dropping out may not be easy to predict using risk factors. Clear evidence that risk factors may
not be good predictors of dropping out comes from a comparison of risk factors and whether students
were dropouts two or three years later (Gleason and Dynarski 1998). If risk factors were good
indicators of dropping out, knowing a student’s risk factors would make a precise prediction of
whether the student will be a dropout in the future possible. Follow-up data collected as part of
Mathematica’s evaluation allow the quality of the risk factors as predictors of dropping out to be
THE DIVERSE NATURE OF AT-RISK STUDENTS
Targeted Targeted Restructuring Restructuring
Middle School High School Middle School High School
Programs Programs Programs Programs
Male 49 55 53 52
Female 51 45 47 48
Black 39 41 42 36
White 18 32 15 20
Hispanic 32 17 37 39
Other 11 9 6 6
Single-Parent Family 42 55 35 35
Family Receives Public Assistance 35 31 24 14
Limited English Proficiency 8 4 16 20
Overage for Grade Level 47 78 32 31
Low Grades 21 54 11 15
Disciplinary Problems 54 73 51 39
External Locus of Control 49 44 45 38
Has Own Children 0 14 1 3
At Least Two Risk Factors 74 94 62 57
How Far Student Would Like to Get in School
High school or less 10 14 10 9
Vocational school or some college 15 43 16 20
Four-year college or graduate degree 75 43 74 71
Student Certainty of Graduating from High School
Very sure 58 48 64 76
Probably 31 36 32 20
Probably not 8 13 3 3
Surely not 3 3 1 1
School Attendance and Performance
Days Absent During Baseline Year
10 or less 60 27 63 46
11 to 20 25 25 20 41
More than 20 15 48 17 13
Percentile Score on Standardized Test
Reading 36 39 34 38
Math 37 38 40 38
SOURCE: Gleason and Dynarski 1994; 1995.
OTE: All numbers are in percents.
checked. The key finding from the analysis is that commonly used risk factors are weak predictors
of dropping out. The analysis showed that many students with numerous risk factors stayed in
school and many with no evident risk factors dropped out. The risk factor that was best able to
predict whether middle school students were dropouts--high absenteeism--correctly identified
dropouts only 16 percent of the time. Using risk factors was better than using no information at all,
but it was not as good as their widespread use might indicate.
The result that risk factors do not predict dropping out has troubling implications for the
effectiveness of dropout-prevention programs. It is difficult to believe that programs can make a
dent in the dropout problem if they cannot accurately predict who will be a dropout in the first place.
Even if dropouts could be identified accurately from risk factors, the wide variation in characteristics
of students programs serve means that programs are working with a range of problems and issues
that may require different approaches.
Can Old Dogs Learn New Tricks?
The context of the dropout-prevention programs affected their ability to operate effectively and
adapt if necessary. The programs generally operated within schools or school districts or were
collaborations between schools or school districts and other organizations. The programs often
brought new and different elements into the local educational landscape: new services, different
teaching styles, changes in how schools were organized, and an emphasis on trying to help students
who had not succeeded in other schools. However, few institutions in America are more stable and
slow to change than schools. The question is whether dropout-prevention programs, which were
relatively small compared to their host schools or school districts, could be implemented and be
accepted in a setting long noted for its resistance to change.
The evaluation observed that dropout-prevention programs were implemented more smoothly
when they were not trying to affect how regular schools worked (Hershey et al. 1995). Dropout-
prevention programs that tried to affect the way regular schools worked did so at their own peril.
Districts and schools often expressed support for reducing the dropout rate and often were strong
advocates of dropout-prevention programs. When dropout-prevention programs tried to change the
status quo, however, district and school support evaporated quickly. Dropout-prevention programs
could survive and flourish by accommodating school districts, but programs should not expect
districts to adapt to accommodate them.
Experiences of three types of dropout-prevention programs illustrate the kinds of issues that
arise when new elements come together with established settings and procedures. These three types
of programs are (1) partnerships between community-based organizations and school districts,
(2) alternative schools, and (3) school restructuring initiatives.
Partnerships between a community-based organization and one or several school districts were
a common type of dropout-prevention program, in part because the U.S. Department of Education
promoted these kinds of partnerships. Often, the community-based organization provided the
impetus for the program, and its partner district provided teachers or access to students in schools.
Regardless of the respective roles in the partnership, the success of the partnership depended heavily
Alternative Schools Are Threats
Creators of alternative schools have to realize this is a threat to the public school system because
the implication is that ‘you have failed and we have come in to pick up the pieces.’
Alternative schools are inherently suspect because they’re different.
Alternative school administrator
It’s almost impossible for an alternative school to be all the things that people perceive.
High school principal
on actions of districts and schools, which in turn depended heavily on larger forces (such as budgets
and changes in leadership) affecting districts.
A long list of examples illustrates the fragility of district-community-based organization
partnerships. Three programs that were partnerships between community-based organizations and
school districts relied on districts to contribute teachers. Each was affected at one time or another
by district decisions to reduce or withdraw the teachers. Another program operated by a community-
based organization began as a leadership course taught during the regular school day. After one year
of operation, the district told the organization that students would be attending an alternative middle
school and the course would no longer be offered. The community-based organization was not
consulted about the change but had to conform to it. Another community-based organization had
to alter its plans to offer a values-enrichment course for at-risk students when most schools in its two
partner districts declined to offer the course. As the examples suggest, the relationship between
school districts and community-based organizations rarely was a relationship between equal partners.
The Challenge of Being Different
Alternative schools were another common type of dropout-prevention program in the evaluation.
By their nature, alternative schools tried to change student experiences, not by changing existing
schools, but by offering students a new kind of school (Hershey et al. 1995; Feister and Rubenstein
1994). Alternative schools struggled to coexist with regular schools, however. Alternative schools
ran into problems recruiting students, which led districts to question the need for the schools. And
alternative schools struggled with the difficult challenge of creating an academic setting in which
students who had not succeeded in other settings would do better.
Recruiting troubles may seem an odd problem for alternative schools to have. How could small
schools for at-risk students operating in urban districts with many at-risk students have trouble
recruiting enough students? The answer is that simply opening the doors of an alternative school
is not enough to attract students. They have to know what the school is all about and have to want
to come. The fact that some schools had to work hard to recruit students suggests that schools had
difficulty spreading the message about what they were all about or that students may have had
reservations about the schools.
Spreading the message about an alternative school’s mission is no small challenge. It was
certainly not safe for alternative schools to assume students and parents were eager to experiment
with the schools. Alternative middle schools were new and untested, and students and parents may
have been reluctant to be test cases. Alternative high schools in general have a longer history than
alternative middle schools, but many districts in the evaluation had operated alternative high schools
only for a short time. The schools did not have a public profile that attracted students: they had to
contend with a public image that “bad kids” went there. The schools had not developed a network
of referral sources. They had limited resources for advertising, and districts put pressure on them
to not appear to be taking students away from regular schools. Schools were better able to attract
students when the schools were close to where students lived, were flexible in who they would
accept, and had extracurricular activities that made them more like regular school (Hershey et al.
1995). Students wanted their alternative-school experiences to be different from regular schools, but
not too different.
Districts might have been able to help alternative schools in their recruiting efforts, but most did
not. The lack of action is perhaps predictable. Alternative schools for at-risk students symbolize
the fact that some students fail in regular schools. Few organizations or individuals hold symbols
of their failure in high regard. Also, regular schools viewed alternative schools as competing for
resources and students, which made it difficult for them to support the schools.
Some alternative schools made the recruiting task more difficult by restricting the kinds of
students they would accept. The schools believed they needed to be restrictive to avoid becoming
a dumping ground where regular schools could send their worst students. For example, one
alternative school interviewed prospective students to assess whether they were motivated enough
to succeed in the school. Another alternative school put incoming students on a two-week
probationary period, during which they had to demonstrate exemplary behavior and attendance or
be sent back to the regular schools.
In restricting the kinds of students they wanted to serve, alternative schools were trying to have
it both ways. On the one hand, the schools professed their mission to be that of serving students who
had not succeeded in regular schools. On the other hand, they wanted to serve students who could
succeed in alternative settings, students with potential to succeed that had not yet been tapped.
Recognizing the first objective, regular schools sent all kinds of students to alternative schools
including “the worst misfits you can imagine,” in the words of one alternative-school teacher.
Recognizing the second objective, alternative schools sometimes sent “misfits” back to regular
schools. The fact that alternative schools turned students away as not being motivated enough or not
being able to behave well enough furthered the view among principals and district administrators
that alternative schools were receiving preferential treatment. Certainly it was asking a lot of a
regular-school principal or teacher to refer students to an alternative school, only to have the school
send back students it preferred not to serve. After all, regular schools themselves could not be
selective. Alternative schools may have gained more control over who they served by being
selective, but it cost them much-needed support from regular schools.
Teaching the Hardest To Teach
More than other dropout-prevention programs, alternative schools had to address the issue of
how to help dropouts or potential dropouts succeed in the classroom. Students who do well
academically in regular schools typically do not attend alternative schools. Alternative schools had
to succeed with the other students, those with low ability, those with ability who were alienated in
regular schools, and those who were unable or unwilling to learn in regular schools. Many students
were attached to school by a thread that could easily break if they were pressured academically or
were in some way dissatisfied with their experiences in alternative schools.
What did alternative schools do differently to work with these students? Their small sizes and
added services may have helped at-risk students. In addition, the types of curriculum and instruction
observed in the evaluation provide a clue as to what alternative schools did differently. The ways
schools worked with students in the classroom can be grouped into four approaches:
1. Some schools taught the usual material in the usual way, though typically in smaller
2. Some schools picked up the pace of standard curricula and pushed students to learn
more and faster so that those who were behind grade level could accelerate their
progress through school.
3. Some schools developed challenging curricula, often by creating thematic and
interdisciplinary units that required students to use knowledge from several subject areas
to address real-world issues.
4. Some schools used competency-based curricula that allowed students to work
independently and progress as fast as their own efforts would allow.
A natural question is whether one approach is better than others. The evaluation’s answer is that
the choice of teachers was more important than the choice of curriculum. The evaluation observed
inspired and creative teachers engaging students intellectually with traditional material. More often,
the evaluation observed classes with interesting subject material in which almost no teaching
occurred, and classes where traditional lecture styles were used and students were not engaged.
Students said that, when teachers pushed them to learn while caring about them as individuals, they
wanted to work harder and succeed. They did not appreciate having teachers spoon-feed material
to them as a way to help them succeed academically.
Teachers Make a Difference
[This teacher is] always ready to help you. When I first came to this school, I didn’t like her, “til
I realized that the only thing she was trying to do was help me.
Student, Accelerated Academics Academy, Flint, Michigan
The teachers stay on you . . . they’ll keep staying on you until you get your goals.
Student, Project ACCEL, Newark, New Jersey
[The teachers] believe you can do the work, and you don’t want to let them down.
Student, Up With Literacy program, Long Beach, California
There’s this teacher over [the regular high school] . . . you can put anything down and he’ll give
you a check mark for it. He doesn’t check it. He just gives you a mark and says, ‘OK, you did
your work.” How you gonna learn from that? You ain’t gonna learn nothing.
Student, JFY Academy, Boston, Massachusetts
Alternative schools succeeded in creating smaller and more personal settings where students felt
comfortable and more connected with teachers and other staff. But they also struggled to recruit
students and engage them academically. Can alternative schools become a more permanent part of
the landscape in school districts? The answer depends partly on their effectiveness at keeping
students in school. As described below, it is questionable whether the alternative schools in the
evaluation achieved their primary objective of keeping students in school. An issue for later
consideration is whether better options are available for helping students who need help.
Change Needs to Come From Within
The clearest examples of how difficult it can be to change schools came from the restructuring
initiatives that were a major part of the second-phase SDDAP. Restructuring initiatives were the
largest dropout-prevention grants in the program, with individual grants averaging $1 million a year
in the first four years of the grants. Grants of this size were not large compared to annual budgets
of average urban school districts, but the funds represented large amounts for particular schools
As the label suggests, the major objective of restructuring efforts was to change the structure
of schools so that more students would stay in school. However, the label applied loosely to a
variety of school reforms and services attempted by districts that received restructuring grants.
Reforms, for example, included school-based governance, school councils, outcomes-based
education, block scheduling, minischools, and interdisciplinary instruction. Services more directly
related to dropout prevention, such as counseling and attendance monitoring, also were significant
components in the restructuring initiatives. The term “restructuring” is used here in this general
sense. Other reports provide more detail about activities in restructuring schools (Hershey et al.
1995; Rubenstein 1995).
The federal investment in restructuring was intended mostly to promote change in schools so
that fewer students would drop out. The evaluation did not observe much change, however, or even
signs of it beginning. Restructuring schools found it easier to add dropout-prevention services than
to change teaching and learning. Some initiatives managed to change teaching and learning to a
degree, but the changes were fragile and easily undone if district leadership changed or local political
contexts shifted. The evaluation found that the investment was more likely to be associated with
change in districts where restructuring had been under way before the grant began and had a broad
base of support among teachers, principals, and the outside community. In districts where this
support had not yet been built, the investment mostly yielded short-term services that could be
dropped easily when funding ended.
Unlike other types of dropout-prevention efforts, restructuring required changing schools as
organizations. Successful programs to change schools as organizations, such as James Comer’s
School Development Program and Henry Levin’s Accelerated Schools program, are built around
three elements. The programs first strive for consensus that change is needed (why change?), pull
people together and develop a process or plan for change that people believe in (how can we make
change happen?), and provide leadership and resources throughout the effort (do we have support
Some or all of the elements were missing for nearly all the restructuring initiatives. The first
element--building a consensus that change was needed--is perhaps the most crucial but was the one
most frequently absent in restructuring initiatives. School staff clearly understood that their schools
needed to change, that their school dropout rates were too high and attendance rates and test scores
were too low. But the evaluation almost never observed a consensus for change and, in particular,
agreement that restructuring was the way to change. Without consensus, there was little
The lack of consensus was related to a lack of reflection and debate about the source of the
problems. Understanding the dropout problem (or a school performance problem) would mean
asking hard questions: Why are students dropping out? What are the barriers students face in
school? How can the barriers be removed or mitigated? How much are schools responsible?
Parents? Communities? What can teachers do differently that would help? But the evaluation
found little evidence that school staff came together to identify sources of a problem and commit to
addressing them. Some restructuring activities observed in the evaluation were consistent with a
view that large schools needed to be made more personal. For example, some schools were
reorganized into smaller units, or students were scheduled together in groups. Adding counselors,
student advocates, and mentors is also consistent with making large schools more personal, by giving
students more access to adults who can help them and be good role models. However, these
activities did not emerge from schools’ critical examinations of how they operated and affected
students, but as promising ideas that had been adopted with success elsewhere.
More often, diagnoses stopped at symptoms without trying to understand causes. For example,
teachers and principals often said that poor attendance was a serious problem and schools needed
to do something to improve attendance. The response often was to implement more attendance
monitoring, through automated telephone systems or more efforts by staff to contact parents or
students. But why were students not coming to school? Were they bored and in need of more
challenging and engaging academic work? Were they unable to succeed because their school was
too challenging? Were they working because their families needed the money? Were they at home
taking care of sick family members or siblings? Were all of these true but for different students?
Accurately diagnosing the attendance problem could point to much different solutions.
Automated telephone systems that called the homes of students who were absent and played a
recorded message would be a solution for only a few of the causes. Restructuring initiatives that
installed automated telephone systems may have believed they were trying to improve attendance,
but such systems work only if parents get the messages and are able to do something about it. The
evaluation saw no evidence that these kinds of assumptions were checked to see if they were valid.
The second element of effective change--working together to develop a plan for change
consistent with the diagnosed problems--was not evident in restructuring initiatives. More often, the
people who developed the plans were from outside the schools that were supposed to carry them out.
District administrators or grant writers often played key roles in developing restructuring plans,
usually with little or no input from principals or teachers of schools that were to take part. Teachers
were simply told that their school needed to change in some ways, which almost guaranteed that
teachers were not committed to change.
Change would have been easier to promote if teachers or principals believed that how they did
things contributed to the dropout problem. The evaluation rarely found this to be the case. The
evaluation commonly observed instruction in restructuring schools that was repetitive, boring, and
even sometimes demeaning, but teachers and principals either were oblivious to this or ignored it.
This is perhaps explained by an organizational structure that provides few rewards or incentives for
outstanding teaching while penalizing deviations from traditional norms of schooling (covering the
prescribed content; keeping students reasonably quiet in class). Whatever the case, few principals
and teachers were eager to change what they were doing even as students were dropping out in
response to it.
The third element of successful change--providing leadership support--also proved difficult for
districts that received restructuring grants. Teachers and principals may not have viewed
restructuring as a solution to a self-identified problem, but strong and steady leadership could
possibly have pushed restructuring ahead against apathy or resistance. However, nearly all districts
that received restructuring grants experienced turnover of top leaders (superintendents, project
directors, principals) during the grant period. New leaders generally were less enthusiastic about
restructuring than those they replaced. New leaders had their own agendas, their own ideas for
change. Because the restructuring plan was imposed on schools, there was no constituency within
schools that wanted to keep it going when leadership changed. Support for restructuring may have
been low at the outset of the effort, and the course of events did nothing to raise it.
Changing evolved practices and relationships in schools can take a long time and be a risky
venture, so failures are likely along with successes. What stands out as the most significant obstacle
is that the same people did not write and carry out the grants. Enthusiasm for restructuring on the
part of grant writers does not mean enthusiasm for restructuring on the part of teachers and
principals, whose activities, roles, and relationships potentially are being changed by restructuring.
Pushing grant money into a school to support change when the school has not charted its own course
of change is like pushing on a string.
The Crucial Question of Program Impacts
To what extent did programs successfully reduce or prevent dropping out? For example, for a
student who is struggling in a regular district high school and is at risk of dropping out, how likely
is it that participating in one of these programs will encourage the student to stay in school to get a
The evaluation could not observe what would have actually happened to program participants
if they had not come to the program. So, for targeted programs, it took the next best approach and
used an experimental design to measure program impacts. It compared what happened to program
participants to what happened to students who were statistically equivalent to program participants.
These students were eligible for the dropout prevention programs but were denied entry to the
programs as part of the evaluation. Experiences of equivalent students are a proxy for what would
have happened to program participants if they had not been able to enter the program. Random
assignment ensures the statistical equivalence of the two groups of students by randomly assigning
students into a treatment group (whose members are allowed to enter the program) and a control
group (whose members are denied access to the program). For each group, the evaluation followed
their academic experiences for a two-to three-year period following random assignment. This type
of experimental design is widely viewed as the best way to measure program impacts.
Two features of the experimental design used for the evaluation affect how results are
interpreted. First, students in the treatment group did not necessarily enter or stay in the dropout
prevention program being studied. Some students may have moved or may have lost interest in the
program. This problem was exacerbated in a few sites where random assignment took place in the
spring for a program beginning in the fall. By using the entire treatment group and control group,
instead of just program participants, the evaluation measured the impact of access to dropout
prevention programs, which may underestimate the impact of participation in dropout prevention
Also, students in the control groups were able to receive other dropout-prevention services
available to them. They could attend regular high school or any other dropout prevention program
other than the program in the evaluation. Thus, the measure of program impacts reveals how the
program affects students relative to other programs in the area. A finding of “no impacts” does not
imply that a program did not help students. It means the program helped students about the same
as other available programs. In a service-rich environment, a program with high-quality dropout
prevention services might have no measurable impact on student outcomes. In an area with few
dropout prevention services, even a lackluster program might have statistically significant impacts.
These programs included the Albuquerque, New Mexico, Middle School Leadership Program,
Measuring the effects of school restructuring on dropping out and other student outcomes
required a different approach. The evaluation could not randomly assign students to treatment and
control groups, because every student in a restructured school was influenced by “the treatment.”
Instead, the evaluation used a comparison group methodology, selecting a sample of students in
schools with similar student characteristics to the schools involved in the restructuring effort. The
evaluation then measured student, teacher, and parent outcomes over time in both restructuring and
comparison schools and compared trends in these outcomes. Improvements over time in these
student, teacher, and parent outcomes in restructuring schools that were not found in comparison
schools were evidence of positive restructuring impacts.
The comparison group methodology has a known weakness that affects how results are
interpreted. Although the evaluation selected comparison schools that were similar to restructuring
schools in observable characteristics, schools can differ in other ways not easily seen. For example,
the culture of a school--whether it has a respected or a lax principal, or high or low academic
standards--may not be captured by observed characteristics. And even if restructuring and
comparison schools were similar at the outset of the evaluation, they could change over time in
important ways having nothing to do with school restructuring. A school’s principal may retire or
move to another school, or a school may come under pressure from the district to improve its
performance. These kinds of changes could be unrelated to the restructuring effort but important
enough to influence outcomes such as teachers’ attitudes or student attendance. Most of the effort
in interpreting the results from the restructuring impact analysis was in trying to understand the
influence other events may have had on the results.
Alternative Middle Schools Have Promise
Communities and school districts that take the term “dropout prevention” most literally set up
programs for students in middle school (or earlier grades). To truly prevent dropping out, the
thinking goes, schools or programs need to intervene in students’ lives before they have left school
or are so far behind academically and alienated from school emotionally that dropping out is just a
matter of time. Middle school programs tried to step in and help students become more engaged and
productive in school before they went too far off track.
A key issue for middle school dropout-prevention programs is how best to serve students.
Should programs leave the basic structure of school intact but give students supplemental help? Or
should they fundamentally change students’ school experiences by having them attend a different
kind of school or a school within a school? The more intensive approach may have a greater chance
of helping students, but it risks stigmatizing and further alienating them.
The modest evidence of middle school program impacts favors the more intensive intervention
approach, however, among the eight middle school dropout prevention programs in our evaluation,
half provided low-intensity supplemental services such as tutoring or occasional classes to promote
self-esteem or leadership. These supplemental programs had almost no impacts on student
the Chula Vista, California, Twelve Together Program, the Long Beach, California, Up With
Literacy Program, and the Rockford, Illinois, Early Identification and Intervention Program.
Table 3 shows average student outcome levels among treatment and control group students
across the supplemental middle school and alternative middle school programs in the evaluation.
Since data were not always available from every site, the table also shows the number of sites on
which the treatment and control averages for a particular outcome are based. The table also shows
the number of sites for which impacts were statistically significant at the 10 percent level.
outcomes. None of the programs affected the dropout rate, and average student grades, test scores,
and attendance were similar among treatment and control group students (Table 3). Supplemental
programs are relatively straightforward to implement, but they do not appear to keep students in
school or improve their attendance or academic performance.
Four middle school programs in the evaluation took a more intensive approach to serving at-risk
students. Two of these programs--the Griffin-Spaulding Middle School Academy near Atlanta,
Georgia and the Accelerated Academics Academy in Flint, Michigan--were alternative middle
schools with facilities that were physically separate from the regular district middle schools. The
other two programs--Project COMET in Miami, Florida, and Project ACCEL in Newark, New
Jersey--were located within regular schools but separated students from other students within the
school for much of the day. These four programs typically taught students in smaller classrooms
than regular middle school students and provided more intensive counseling services. Three of the
four programs primarily served students who were overage for their grade level, and these programs
attempted to accelerate students’ academic progress to allow them to “catch up” with their age peers.
The alternative middle school programs in our evaluation successfully kept kids in school and
accelerated their academic 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 (Table 3). These programs seemed to keep students in school longer.
On the other hand, alternative middle schools do not seem to help students learn more in school.
Alternative middle schools in the evaluation had no impacts on grades or test scores, and they had
impacts on attendance in the wrong direction (treatment group students were absent more often than
control group students). Although students were promoted at a faster rate than students in regular
middle schools, student learning did not seem to improve in these programs.
The effects of alternative middle schools were concentrated primarily in the Atlanta and Flint
programs (see box). These programs show that intensive intervention can keep students in school
longer or even accelerate their progress in school. Perhaps these sixth- through eighth-grade years
represent a critical juncture in students’ school careers, where attention and positive feedback may
send them in one direction, while lack of attention or negative school experiences may send them
in another. Evidence from Atlanta and Flint suggests that something positive happened for their
students that kept them in school. But, on the sobering side, the programs’ lack of effects on
attendance or academic performance outcomes suggests that participants are not learning more than
students in regular schools.
IMPACTS OF MIDDLE SCHOOL DROPOUT PREVENTION PROGRAMS
Average Average Number of Sites
Treatment Control Group Number with Significant
Group Mean Mean of Sites Impacts
Dropout Rate (Percentage)
End of Follow-Up Year 2 7.8 7.0 4 0
End of Follow-Up Year 3 11.5 15.0 4 0
During Follow-Up Year 2 10.5 10.0 4 0
During Follow-Up Year 3 14.3 14.3 4 0
Follow-Up Year 2 69.5 68.3 4 1(+)
Follow-Up Year 3 67.5 67.0 4 0
Reading Test Score (percentile)
Follow-Up Year 2 36.0 35.5 2 0
Follow-Up Year 3 37.0 34.0 1 0
Alternative Middle School Programs
End of Follow-Up Year 2 4.7 9.3 3 1(-)
End of Follow-Up Year 3 9.0 18.0 2 1(-)
Highest Grade Completed
End of Follow-Up Year 2 7.9 7.4 3 3(+)
End of Follow-Up Year 3 8.6 8.1 2 2(+)
During Follow-Up Year 2 18.3 15.3 4 3(+)
During Follow-Up Year 3 18.0 17.0 2 0
Follow-Up Year 2 65.0 66.3 3 0
Follow-Up Year 3 62.0 64.0 2 0
Reading Test Score (Percentile)
Follow-Up Year 2 16.3 16.7 3 0
Follow-Up Year 3 28.0 31.0 1 0
OURCE: Dynarski et al. (1997)
Plus and minus signs indicate whether impacts were positive or negative.
The four sites operating alternative high schools were Boston (JFY High School and University
High), Las Vegas (Horizon High Schools), Miami (Corporate Academy), and Seattle (Middle
College High School). The school within a school approach was used in Chicago (Wells Academy).
Alternative Middle Schools - Some Good News
The alternative middle schools in Atlanta and Flint had similar impacts on student outcomes. The
schools reduced dropping out and accelerated students’ progress in school. Neither program
positively affected student achievement.
Dropout Rate (Percent) 6 14 2 11
Highest Grade Completed 8.6* 7.9 8.5* 7.8
Math Grade 59 63 67 66
Reading Test Score (Percentile) -- -- 12 12
OTE: All outcomes measured at the end of the second follow-up year, except for highest grade completed, which
is measured at the end of the third follow-up year in Flint.
* Significantly different from the control group at the ten percent level, two-tailed test.
OURCE: Dynarski et al. (1998a).
For Some High School Students, Short Roads May Be Better
Dropout-prevention programs for students of high school age face a different dilemma than
those for middle school students. The choice between supplemental and intensive intervention is
moot, since their students are older and have either already dropped out or are on the verge of doing
so. They need intensive intervention. The dilemma for these programs involves the appropriate
educational goal for the program. Should their students take the long road and earn a high school
diploma? Or should they take the short road and earn a GED certificate?
The modest impact evidence favors the short road. Five of the high school dropout-prevention
programs in the evaluation offered high school diplomas. Four were structured as alternative high
schools and one was a school within a school. None of these programs significantly lowered
dropout rates (Table 4), though more of their students received diplomas than GED certificates. In
contrast, participants in the three GED programs were more likely to earn their GED certificates and
even somewhat more likely to complete their diplomas than control group students (this result arises
IMPACTS OF HIGH SCHOOL DROPOUT-PREVENTION PROGRAMS
Treatment Control Group Number of with Significant
Group Mean Mean Sites Impacts
Number of Sites
Alternative High School Programs
End of follow-up year 2 35 30 5 1(+)
End of follow-up year 3 39 40 3 0
High school diploma 21 15 4 0
GED 13 19 4 1(-)
Either 33 34 4 0
End of follow-up year 2 56 58 3 0
End of follow-up year 3 57 60 3 0
High school diploma 9 3 3 0
GED 30 20 3 0
Either 39 24 3 1(+)
OURCE: Dynarski et al. (1998a).
OTE: For alternative high schools, completion rates refer to the second follow-up year for two programs and the
third follow-up year for two programs. For GED programs, completion rates refer to the third follow-up
Helping Motivated Students Get Diplomas
Impacts of Seattle’s Middle College High School
Seattle’s Middle College High School is an alternative high school set up on a community college
campus. The program served dropouts or students on the verge of dropping out of regular high
schools. It screened students when they first applied to the school to ensure that they were
motivated to succeed. The school had success helping low-risk students get diplomas. It also was
able to reduce dropping out for high-risk students.
Low-Risk Students High-Risk Students
Dropout Rate 33 33 27* 42
Completion Rate 53 56 59 58
High school diploma 33 24 27 25
GED 20 32 32 33
Still In High School or GED Program 13 11 13 0
OTE: All outcomes measured at the end of the third follow-up year. Percentages may not add to 100
because of rounding.
*Significantly different from the control group level at the ten percent level, two-tailed test.
OURCE: Dynarski et al. (1998a)
because students who start in GED programs can leave the program and go to other programs or back
to high school). The total effect is that GED programs improved the overall high school completion
rate from 24 percent to 39 percent, a relative increase of over 60 percent.
One positive finding for alternative high schools is that they influenced whether students earned
a diploma or a GED. Four of the five alternative high school programs served students who were
old enough to graduate during the follow-up period. In these programs, more students earned high
school diplomas and fewer earned GED certificates. The differences were not statistically significant
in any of the four sites, but the pattern is consistent across sites. Control group students were less
likely to earn a high school degree and more likely to earn a GED. Alternative high schools affected
how students completed school but not the overall completion rate.
A closer look at Seattle’s Middle College High School provides insight about how alternative
high schools can affect different students. Middle College High School had higher high school
completion rates and lower GED completion rates (see box) for students whose characteristics
suggested that they were least likely to drop out (termed “low risk” students in the box, though most
were at some risk of dropping out). The school also reduced dropping out for high-risk students.
Serving the Hardest-to-Serve Students
Impacts of the St. Louis Metropolitan Youth Academy
St. Louis’s Metropolitan Youth Academy is a GED program for highly at-risk students. Nearly
all of the students served were dropouts and had, on average, the most risk factors of any program
in the evaluation. The program was more successful at helping students earn GEDs, though many
of its students were able to complete their GED.
Treatment Group Control Group
Dropout Rate 62 64
Completion Rate 26* 16
High school diploma 5 5
GED 21 11
Still in High School or GED Program 12 10
OTE: All outcomes measured at the end of the second follow-up year.
* Significantly different from the control group at the ten percent level, two-tailed test.
OURCE: Dynarski et al. (1998a)
One key feature of Middle College High School is that it made a serious effort to serve students
who were motivated to succeed in school by having staff and current students interview prospective
students. The positive impacts of the school suggest that alternative high schools can be successful
when they serve students who want to succeed. Other features of Middle College High School, such
as its dedicated staff and location on a college campus, no doubt also played a role in its success, but
serving students who wanted to succeed seems crucial.
Overall, alternative high schools did not reduce dropout rates much, and a natural question is
whether GED programs could be more successful. Two programs in the evaluation--the Queens,
New York Flowers with Care Program and the St. Louis, Missouri, Metropolitan Youth Academy--
were designed to help students prepare for the GED. A third program--the Student Training and Re-
entry Program in Tulsa, Oklahoma--was a transition program for high school dropouts to help them
determine and achieve an appropriate educational goal (which usually turned out to be a GED
Among the three GED programs, the Metropolitan Youth Academy in St. Louis had the largest
impacts (see box), with 26 percent of treatment group students earning a GED certificate or a high
school diploma within two years, compared to 16 percent of control group students. This is a
substantial effect, and it is especially notable since the academy served students who were more at
risk than any other program in the evaluation. The results show that it is possible to help the hardest-
Typically, the evaluation collected data from one restructuring and one comparison school at
both the high school and middle school levels in each site.
The positive results for the Metropolitan Youth Academy need to be considered with caution,
for two reasons. First, although the program improved school completion, the overall level of school
completion for its students was low (about 38 percent after two follow-up years). Put another way,
well over half of its students were high school dropouts two years later. Second, the educational goal
of programs like the Metropolitan Youth Academy remains in question. Researchers continue to
debate the benefits of the GED, with some arguing that GED completers are not much better off than
dropouts. The jury is still out on the benefits of getting a GED.
Taken as a whole, our findings on the effects of high school dropout-prevention programs reveal
no magic bullets. The limited success of the eight programs we studied suggests that we do not
know how to promote educational success among most high school age students who have dropped
out or are on the verge of doing so. Several of the programs we studied offer promising approaches,
however. For motivated students, the middle college model used in Seattle may improve their
chances of finishing high school. For students who are already dropouts and who have substantial
other problems hindering their educations, GED programs can help them receive a GED. The results
cannot be compared directly--we cannot say middle college schools are better or worse than GED
programs--because the types of students served were different. The most that can be said is that
high-risk students are probably better off taking the short road, and motivated students are probably
better off taking the long road. But putting high-risk students on the long road may do little for
Can Restructured Schools Reduce Dropping Out?
As described earlier, the largest federal grants went to efforts to restructure schools. These
efforts included reforms of school governance, changes in classroom instruction and curricula, and
services for students. The evaluation collected and analyzed data on student, teacher, and parent
outcomes in restructuring and comparison schools in four sites that began school restructuring efforts
in the 1991-1992 school year: Dallas, Texas; Grand Rapids, Michigan; Phoenix, Arizona; and Santa
Generally, impacts of restructuring on student outcomes were negligible, though there were
several exceptions that we talk about below. Few differences in student outcomes between
restructuring and comparison schools emerge in these sites, at either the middle school or high
school level. Where absenteeism or test scores are higher or lower in a restructuring school relative
to a comparison school, the difference typically is the same difference evident in the baseline year
(1991-1992), before restructuring had gotten off the ground. At the high school level, for example,
the average number of days absent in 1994-1995 is higher in comparison schools than in
restructuring schools, but that was also true in 1991-1992.
In middle schools, the dropout rate in 1994-1995 (for students who were sixth graders in 1991-
1992) was higher for restructuring schools than comparison schools. In high schools, the average
dropout rate was higher in comparison schools. However, this difference was caused almost entirely
by a large difference in Grand Rapids (where the 1994-1995 dropout rate was 22 percent in the
comparison high school and 12 percent in the restructuring high school). The lower dropout rate in
the Grand Rapids restructuring high school could not be attributed to the positive effects of school
restructuring, however, because further analysis suggested other sources of the difference (Dynarski
et al. 1998). Restructuring efforts in Grand Rapids were slow to develop, and the comparison school
went through a significant leadership change that appeared to affect student and teacher outcomes.
The results in Grand Rapids probably arose because of problems at the comparison school instead
Restructuring did appear to affect student outcomes in the Dallas middle school that took part
in the restructuring effort. Between 1991-1992 and 1993-1994, the school’s reading test scores
improved substantially while the comparison school’s reading test scores stayed the same (Dynarski
et al. 1998). The score gains may be related to the “accelerated schools” model adopted by the
school, but the timing of the model’s implementation suggests other factors also were affecting
School restructuring efforts were more likely to influence teacher outcomes than student
outcomes. For example, middle school teacher perceptions about the climate of their school were
more positive between 1992-1993 and 1994-1995 in restructuring schools and more negative in
comparison schools (Table 5). High school teacher perceptions showed little difference, except in
Phoenix, where teachers in restructuring schools were more positive about their school’s climate.
The evaluation found that restructuring did not influence parent perceptions of the climate of
their children’s schools. However, parents had very positive perceptions of schools overall.
Restructuring was unlikely to cause parents to feel better about schools with which they were already
The nature of the restructuring initiatives in schools where student or teacher outcomes
improved provides a clue about the type of initiative with the most promise. Schools were more
likely to show improved student or teacher outcomes when their restructuring efforts focused on
improving curriculum and instruction through staff development workshops, summer training
sessions, and classroom implementation. In some schools, the restructuring approach placed more
emphasis on providing student services related to dropping out. It is not surprising that teacher
outcomes failed to improve in schools where restructuring consisted mostly of providing more
student services. However, student outcomes also did not improve in these schools. In the end,
restructuring has more promise when it focuses on changing what happens in classrooms rather than
on providing services for students.
Learning from the Dropout-Prevention Demonstrations: How Do We Get There from Here?
At the 1989 Education Summit in Charlottesville, Virginia, the nation’s governors worked with
President Bush to identify goals for education. One goal was to increase the high school completion
rate to 90 percent by the year 2000. In 1996, the rate was 86 percent, where it had been since 1990.
The evaluation’s results do not offer much direction for how to achieve this goal with reasonable
IMPACTS OF SCHOOL RESTRUCTURING INITIATIVES
Middle Schools High Schools
Students in Students in Students in Students in
Restructuring Comparison Restructuring Comparison
Schools Schools Schools Schools
End of 1994-1995 9 8 14 16
Percentage of Days Absent
During 1991-1992 8 8 13 14
During 1994-1995 23 22 17 19
Reading Test Score (Percentile)
1991-1992 28 29 36 37
1993-1994 28 32 24 26
Math Test Score (Percentile)
1991-1992 35 40 36 38
1993-1994 27 31 29 32
Teachers’ School Climate Index
End of 1992-1993 46 51 54 54
End of 1994-1995 54 45 54 54
Parents’ School Climate Index
End of 1992-1993 75 75 74 71
Ed of 1994-1995 79 80 76 73
OURCE: Dynarski et al. (1998b).
To get from the 5 percent dropout rate to the 86 percent high school completion rate, we assume
95 percent of 16-year-olds continue on in school, 95 percent of those who continue on as 17-year-
olds complete another year of school, and 95 percent of those who make it through another year as
18-year-olds complete school.
To see this, consider how many students programs would need to keep in school for the high
school completion rate to rise by four percentage points and meet the goal. The current annual
dropout rate of five percent, if applied to 16-, 17-, and 18-year-old students, yields a high school
completion rate of 86 percent. A lower annual dropout rate of 3.5 percent, if applied to 16-, 17-,
and 18-year-old students, yields a high school completion rate of 90 percent, which is the target rate.
So in terms of the dropout rate, the goal is to go from the current 5 percent rate to a 3.5 percent rate.
Going from a dropout rate of 5 percent to 3.5 percent means a reduction in the number of dropouts
of about 145,000 a year as of 1996. Could the few dropout-prevention programs that showed some
success be replicated and reduce the number of dropouts this much?
The answer is probably not. The results in Table 5 show that the three GED programs reduced
the dropout rate from 60 percent to 57 percent. Each program served about 100 students at a time.
So 3 more students did not drop out as a result of each program. We would need more than 48,000
programs such as the three GED programs to reduce the dropout rate by the target amount of 145,000
students a year. The programs had average grants of about $300,000 a year, so the cost of the
expanded effort would be about $14.5 billion. Alternative high schools had smaller effects on
dropout rates and so would cost even more if replicated on a wider scale to achieve the goal.
Alternative middle school programs had larger effects on dropout rates, but it is not yet clear whether
their effects will carry over to high school. The goal of increasing the high school completion rate
from 86 percent to 90 percent seems reasonable. Given what we know about impacts of dropout-
prevention programs and their costs, however, it does not look like we can get there from here.
The evaluation provided other important lessons. The pattern of evidence discussed above
points to some ways dropout-prevention programs could be more effective. Alternative middle
school programs that created smaller class sizes, more personalized settings, and more focused
teaching and learning, faced more difficult implementation challenges but had some effects,
including higher rates of grade promotion and lower dropout rates. Alternative middle school
programs did not result in more learning according to outcomes such as standardized test scores, so
the evidence for their effectiveness is mixed. However, the results suggest that more intensive
interventions for middle school students may be needed to reduce dropping out and improve other
Programs oriented to get students GED certificates had the largest impacts, though their results
were not large in an absolute sense. For example, the St. Louis program raised GED completion
rates but still had about a 60 percent dropout rate. However, the program served students who were
among the hardest to serve of any in the evaluation. The program’s results show that some success
is possible with the hardest to serve but that not many of the hardest to serve had success.
The GED credential is not the same as a high school diploma--employment rates and earnings
of GED recipients are higher than earnings of dropouts, but not by much. However, Agodini and
Dynarski (1998) show that the small increase in earnings attributable to the GED is enough to induce
some students to choose a GED over a high school diploma, especially those who are behind grade
level or who can work while preparing for the GED. Being behind grade level means students take
longer to complete high school and pay greater costs by not being able to work during that time.
Instead of trying to direct these students back toward the high school diploma, it may make more
sense to link the GED with postsecondary education or training in a way that overcomes the stigma
that employers attach to the GED. Employers are more likely to focus on prospective employees’
most recent training as evidence of their ability to perform well. GED recipients who go on to
acquire useful skills at the postsecondary level will be better prepared for a labor market that appears
to be putting an ever-increasing premium on skills honed through formal education.
Restructuring schools to reduce dropping out seems like an attractive way to approach the
dropout problem. If restructuring succeeded, students would not need the kinds of interventions
represented by the Seattle and St. Louis programs. The evaluation’s results show that this is a big
“if.” A few restructuring efforts improved school climate, according to teachers, and student test
scores rose in some middle schools. Dropping out was not affected, though restructuring efforts may
not have operated long enough to have an effect on this problem.
The real issue for restructuring is whether schools want to do it. The evaluation mostly observed
the resistance to restructuring and its temporary nature. Changes in leadership invariably
undermined support for restructuring. School staff themselves, who may have been able to continue
efforts to restructure on their own, were not enthusiastic supporters. When restructuring had an
effect, it came when the effort focused on changing what happened in classrooms. If schools want
to change what happens in the classroom--how teaching and learning occurs--these changes may
affect school climate and learning. Simply adding services--a health clinic, a child care center,
attendance monitoring--altered little and affected little. The trade-off is clear. Services were
straightforward to implement and supported by school staff because they met the needs of students.
But the services did not yield changes in any measured outcomes. Classroom changes took more
effort to implement, but they changed some outcomes in a few schools.
Future dropout-prevention efforts could build on the lessons noted here by promoting alternative
middle schools, alternative high schools for students with motivation or academic potential, GED
programs for older students, and restructuring built on classroom change. Another path, perhaps a
more challenging one, is to go back to the drawing board and ask why students drop out and what
can be done about it. Do we really know why students are dropping out and how to stop it? Schools
and community organizations can do only so much with students. Is it enough? Students who drop
out are difficult to identify in advance. Do we know who to serve? Can a particular program be the
right approach for more than a few of those it serves? Are we only addressing symptoms, without
knowing what the real problems are?
If the answer is that we do not really know why students drop out, setting up dropout-prevention
programs will almost surely yield weak results. Simply pumping resources into schools to create
smaller schools, reduce class sizes, and provide support services will not generate results if large
schools, large classes, and a lack of support services were not the real problem in the first place.
Millions of dollars would be spent to reduce dropping out only slightly, an undesired outcome in an
era of performance standards and pressure for more accountable government and schools.
Perhaps a useful way to think about dropout prevention is to view attending school as going to
a job and prospective dropouts as employees who are not happy with their jobs or who are not
performing well in them. Some employees who are not doing well in a job are coached and
encouraged to do better, others quit, and others are fired. A company concerned about its high rate
of people leaving might adopt a policy of meeting with employees one on one and asking straight
questions: Why are you unhappy here? Why don’t you like your job? What can we do to make you
like your job more? Would you like a different job here instead? This approach is based on the view
that each employee brings something valuable to the company and that the company wants to
understand how it can best use the assets an employee brings to it. A change of job, a change of
venue, or a change of personnel mix might be the answer for an employee to begin working more
Schools may work with their actual employees--teachers and staff--using the above approach,
but do they do this with students? When students are doing poorly or are on the verge of dropping
out, or when they want to return to school after dropping out, do school staff talk with them about
problems they are having and set up a plan to mitigate these problems? Is adherence to the plan
monitored, and are results fed back to the staff and students so they can make adjustments? Do
schools have access to services that can help families and students deal with personal problems? If
a student feels he or she is not learning much or is having difficulty with a particular teacher, is
someone in the school thinking about how to get the student more engaged in learning? If a student
has disagreements or arguments in school with teachers or other students, does someone in the
school intervene to mediate? Are parents actively working with teachers and school staff to solve
problems and address issues affecting their children?
If answers to these questions are “sometimes,” “maybe,” or “no,” there is room for an
individualized approach to dropout prevention. A school has limited ability to make most of its
students happy (or at least willing to attend), and perhaps it is expecting too much of schools to think
they can. But this approach is worth a try. Focusing on ways to identify students who are having
difficulty and actively working to address the sources of the difficulty may be a more effective use
of prevention resources than setting up a program to catch students after they fail.
How can we help? We can help by asking what it would take for schools to keep their students
and get rid of their dropout-prevention programs. This is an ambitious goal, but the benefits of
striving for it are huge. Perhaps preventing dropping out should be only one objective within the
larger goal of creating schools where all students are given the assistance and opportunities they need
to learn and develop into successful adults.
Adelman, N., and M. Rubenstein. “The National Evaluation of the School Dropout Demonstration
Assistance Program, 1991 and 1992 Grantees (Draft).” Washington, DC: Policy Studies
Agodini, R., and M. Dynarski. “Understanding the Trend Toward Alternative Certification for High
School Graduates.” Princeton, NJ: Mathematica Policy Research, 1998.
Corson, W., M. Dynarski, J. Haimson, and L. Rosenberg. “A Positive Force: The First Two Years
of Youth Fair Chance.” Princeton, NJ: Mathematica Policy Research, 1996.
Dynarski, M., P. Gleason, A. Rangarajan, and R. Wood. “Impacts of Dropout Prevention
Programs.” Princeton, NJ: Mathematica Policy Research, 1998a.
Dynarski, M., P. Gleason, A. Rangarajan, and R. Wood. “Impacts of School Restructuring
Initiatives.” Princeton, NJ: Mathematica Policy Research, 1998b.
Feister, and M. Rubenstein. “Building Support for Alternative Schools: Lessons from the SDDAP.
Washington, DC: Policy Studies Associates, 1994.
Gleason, P., and M. Dynarski. “Do We Know Whom to Serve? Issues in Using Risk Factors to
Identify Dropouts.” Princeton, NJ: Mathematica Policy Research, 1998.
Gleason, P., and M. Dynarski. “Falling Behind: Characteristics of Students in Federally Funded
Dropout Prevention Programs. Part Two: Restructuring Projects.” Princeton, NJ: Mathematica
Policy Research, 1995.
Gleason, P., and M. Dynarski. “Falling Behind: Characteristics of Students in Federally Funded
Dropout Prevention Programs. Part One: Targeted Projects.” Princeton, NJ: Mathematica
Policy Research, 1994.
Hayward, B., and G. Talmadge. “Strategies for Keeping Kids in School: Evaluation of Dropout
Prevention and Reentry Projects in Vocational Education.” Raleigh, NC: Research Triangle
Institute, June 1995.
Hershey, A., N., Adelman, and S. Murray. “Helping Kids Succeed: Implementation of the School
Dropout Demonstration Assistance Program.” Princeton, NJ: Mathematica Policy Research,
Rosenberg, L., and A. Hershey. “Costs of Dropout Prevention Programs.” Princeton, NJ:
Mathematica Policy Research, 1995.
Rossi, R. “Evaluation of Projects Funded Under the School Dropout Demonstration Assistance
Program.” Palo Alto, CA: American Institutes of Research, March 1994.
Rubenstein, M. “Professional Development for School Restructuring: Lessons from the SDDAP.”
Washington, DC: Policy Studies Associates, 1995.
Wehlage, G., R. Rutter, G. Smith, N. Lesko, and R. Fernandez. “Reducing the Risk: Schools as
Communities of Support.” New York: Falmer Press, 1989.