CENTERS > Chief Justice Earl Warren Institute on Race, Ethnicity and Diversity > Research >

The Limitations of the No Child Left Behind Act as a Strategy for
Improving High School Graduation Rates

Russell W. Rumberger
University of California, Santa Barbara

Revised

June 2006

Improving the nation’s high school graduation rate has remained an elusive goal for many years.  In 1990, the nation's governors and the President of the United States adopted six national education goals, including increasing the high school graduation rate to 90 percent and eliminating the gap in high school graduation rates between minority and non-minority students by the year 2000 (U.S. Department of Education, 1990).  But these goals were voluntary and the best estimates suggest little improvement to the nation’s high school graduation rate.[1] 

Beginning in the 1990s, many states implemented performance-based accountability systems that mandated improvements in student and school performance.  Although in most accountability systems tests scores were the sole indicator of performance, increasingly states included dropout and graduation requirements.  In many cases, the mandates were accompanied by rewards and sanctions that carried high stakes for both students and schools.  For students, sanctions included passing a high school exit exam in order to receive a diploma.  For schools, sanctions included state takeover or even dissolution. 

The federal No Child Left Behind (NCLB) Act of 2001 requires that states receiving Title I funds institute an assessment system for all the public schools and districts in the state, whether or not they receive Title I funds, and an accountability system for the schools and districts that do receive Title I funds.  Although states are allowed to set content and performance standards for students, NCLB mandates a timeline for schools and districts to reach proficiency on those performance standards and imposes sanctions on all Title I schools that do not meet them.  NCLB also requires that public high schools demonstrate improvement in both student achievement (test scores) and graduation rates. 

Yet NCLB and performance-based accountability systems alone are unlikely to improve the nation’s high school graduation rate.  First, schools can only address some of the factors that contribute to students dropping out of school.  Second, performance-based accountability systems in general, and NCLB in particular, do not provide the right mix of incentives, resources, and support to alter the will and capacity of schools and the people in them to improve.  Finally, the current accountability provisions of NCLB for improving graduation rates are weak.  After discussing the limitations of performance-based accountability systems and NCLB for improving high school graduation rates, this paper suggests ways to address these shortcomings.  

The Contribution of High Schools to Dropout Rates

One of the most important and fundamental issues in educational research and policy is the extent to which schools contribute to student achievement.  The reason is that many educational policies, including NCLB, are predicated on the belief that differences in student achievement are largely due to differences in schools, so improving schools—particularly low-performing schools—will raise overall student achievement and reduce differences in achievement between high-performing advantaged students and low-performing disadvantaged students.  

While researchers and policymakers from all political persuasions agree that schools contribute to student achievement, the extent of that contribution is much less clear.  Famed sociologist James Coleman sparked widespread debate with the publication of his report in 1966 by concluding that schools had relatively little impact on student achievement compared to the background of the students who attend them:  “family background differences account for much more variation than do school differences” (1966, reprinted in Coleman, 1990, p. 124).  More specifically, Coleman found that schools only accounted for 5 percent to 38 percent of the variation in achievement among different grade levels, ethnic groups, and regions of the country (Coleman, 1990, p. 77).  Although Coleman’s analysis was subjected to considerable scrutiny and debate, more recent research studies using more refined statistical techniques continue to find that no more than a third of the variation in student achievement lies between different schools (Raudenbush & Bryk, 2002; Rumberger & Palardy, 2004).[2]  This has important implications for policies designed to improve student achievement that focus on schools as the sole arena for change, a topic I revisit below (Coleman, 1990; Rothstein, 2004). 

Not only did Coleman find that differences in schools account for relatively modest differences in student achievement, but he also found that “the social composition of the student body is more highly related to achievement, independent of the student’s own social background, than is any school factor, ” including school facilities and attributes of teachers (Coleman, 1990, p. 119).  One recent study confirmed that the social class background of the student body had almost as large—and sometimes larger—an effect on student learning as students’ own social class background (Rumberger & Palardy, 2005a).[3]  These findings, too, have implications for school reform strategies.

To what extent do schools contribute to dropout and graduation rates?  This question is more difficult to answer than a similar question concerning student achievement, but there is little reason to suggest that it is much different than with student achievement.  Research from one recent study found that more than half of the differences in dropout rates among a sample of 912 high schools from across the United States were due to differences in the background characteristics of students who attended the schools rather than differences in the characteristics of schools themselves (Rumberger & Palardy, 2005b).  Yet even controlling for differences in student background characteristics, there remain substantial differences in dropout rates of U.S. high schools:  a student with average background characteristics is less than half as likely to drop out from a high-performing high school than from a low-performing high school.[4]  At the same time, an analysis of school dropout rates reveals that the majority of high schools in the United States have modest dropout rates and relatively few have what could be considered very high dropout rates.[5] 

Existing research suggest the potential and limitations of schools as a vehicle for improving the nation’s high school graduation rate.  Even if all the existing differences between high schools in the U.S. were eliminated, substantial differences in dropout rates among students would remain.  Yet improving the nation’s high schools can substantially improve students’ prospects for graduating

Effective Strategies for Reducing Dropout Rates in High Schools

There are two basic approaches for reducing dropout rates in high school.  The first approach focuses on programmatic interventions that either provide supplemental academic and social supports to students within an existing school program, or provide an alternative school program in an existing school (school within a school) or in a separate facility (alternative school).  This approach attempts to alter the values, attitudes, and behaviors of targeted, at-risk students without attempting to alter the characteristics of the larger families, schools, and communities that may contribute to those individual characteristics.  The second approach focuses on systemic interventions that attempt to fundamentally transform the way families, schools, and communities serve all youth.

Both approaches have strengths and weaknesses.  Programmatic interventions are easier to develop, evaluate, and disseminate widely than systemic interventions, but they can be costly and impact relatively few students.  Despite the large number of dropout prevention programs in the U.S., and the relative ease in evaluating them, few dropout intervention programs have been evaluated using scientifically rigorous, random-assignment evaluations.  For example, the General Accounting Office[6] surveyed more than 1,000 dropout programs in the fall of 1986, yet it found only 20 rigorous evaluations of the 479 programs that responded to the survey (U.S. General Accounting Office, 1987).  A more recent review of more than 100 federally-funded dropout prevention programs found that only 30 programs had rigorous evaluations, and only three significantly reduced high school dropout rates (Dynarski, 2004).  Although most dropout prevention programs focus on students at risk of dropping out in middle and high school, a number of preschool programs designed to improve the cognitive and social development of disadvantaged children have been shown to significantly reduce high school dropout rates (Barnett, 1995).

Systemic interventions have the potential to reduce dropout rates for a much larger number of students by improving some of the environmental factors in families, schools, and communities that contribute to dropout behavior.  This was the position taken by the National Research Council Panel on High-Risk Youth (1993) who argued:

The primary institutions that serve youth—health, schools, employment, training—are crucial and we must begin with helping them respond more effectively to contemporary adolescent needs.  Effective responses will involve pushing the boundaries of these systems, encouraging collaborations between them and reducing the number of adolescents whose specialized problems cannot be met through primary institutions (p.193). 

Systemic solutions may be particularly appropriate in schools and communities where the concentration of students at risk of dropping out is high. 

Although the promise of systemic solutions to the dropout problem is great, the reality is not.  The reason is simply that systemic changes are extremely difficult to achieve because they involve making fundamental changes in the way institutions work individually and within the larger system of which they are a part.  In addition, systemic interventions are difficult to evaluate with rigorous, scientific methods involving random assignment (Cook, 2005). 

To date, most systemic intervention strategies in education have relied on externally developed comprehensive school reform (CSR) models that transform the entire operation of schools through a series of coherent and reinforcing strategies.[7]  These models have been adopted by hundreds of schools in the United States with financial support from the federal government through the Comprehensive School Reform Program and NCLB (Desimone, 2002).  However, there is little evidence that CSR models have been effective in reducing high school dropout rates.  A recent evaluation of 29 of the most widely adopted CSR models found that most models focus on elementary schools, and those that focus on high schools have been evaluated primarily in terms of their effects on student achievement (test scores) and not dropout rates (Borman, Hewes, Overman, & Brown, 2003). 

There is evidence that reforming high schools to reduce dropout rates may be particularly challenging.  In their study of 207 urban high schools that were attempting major school reform programs based on the effective schools literature, Louis and Miles (1990) found widespread improvement in a number of areas such as student behavior, and student and staff morale.  But even among programs that had implemented these changes for several years and enjoyed improvements in student achievement, improvement in dropout rates were “rarely achieved no matter how long a program had been in operation” (Louis & Miles, 1990, p. 49).  Another review of five school restructuring efforts supported by large, multi-million dollar grants from the federal dropout prevention program found that none of these restructured schools significantly reduced dropout rates in relation to comparable schools (Dynarski & Gleason, 1998).[8] 

Despite the paucity of rigorous scientific evidence establishing the effectiveness of programmatic or systemic intervention strategies in reducing dropout rates, there is sufficient evidence on proven and promising strategies to identify the common features of effective programs and effective schools that reduce dropout rates.

Features of Effective Dropout Prevention Programs

Although dropout prevention programs differ widely in the age and characteristics of the students they serve, the services they provide, and the way they are structured, there appear to be several common features among effective programs (Dynarski & Gleason, 1998; Wehlage, Rutter, Smith, Lesko, & Fernandez, 1989).  These common features include:

a focus on meeting both the academic and non-academic needs of students;

·        a non-threatening environment (accepting students as they are);

·        a caring and committed staff who accept personal responsibility for student success;

·        a culture that encourages staff risk taking, self governance, and professional collegiality;

·        a structure that provides for a low student-teacher ratio and a small class size to promote student engagement.

These features focus on developing the commitment and competencies of both the students and the staff who work with them, and on providing an organizational structure to nurture and support this development.

Features of Effective Schools

Since the publication of the Coleman report in 1966, countless studies have been undertaken to identify the characteristics of effective schools.  These studies have ranged from large-scale statistical studies of national datasets (as Coleman did) to small-scale case studies of particular schools.[9]  Although a number of specific features have been identified in these studies, they essentially address two basic features of schools: school inputs, or the material conditions of schools, which include school resources and school structure, and the characteristics of students and teachers; and school processes, which include the attitudes, behaviors, and practices of students, teachers, and administrators as well as the policies, practices, and culture (or climate) of the school as a whole.   

There is considerable disagreement on the requisite inputs needed to develop and support effective schools.  That is, what resources, teacher characteristics, student characteristics, and structural features are necessary to create an effective school?  For example, there is ongoing debate on whether school resources make a difference.  A major review of 187 studies that examined the effects of instructional expenditures on student achievement concludes:  “There is no strong or systematic relationship between school expenditures and student performance” (Hanushek, 1986, p. 47).  Other reviewers conclude, however, that school resources can make a difference (Hedges, Laine, & Greenwald, 1994).  There are similar debates about the extent to which other school inputs matter:  (1) teacher characteristics, such as credentials and training, (Wayne & Youngs, 2003); (2) racial or socioeconomic composition of students in the school (Kahlenberg, 2001; Orfield & Lee, 2005); (3) certain structural features of schools, such as school size (Luyten, 1994; National Research Council, Committee on Increasing High School Students' Engagement and Motivation to Learn, 2004); and (4) organizational control, such as private, public, charter, and magnet schools (Bettinger, 2005; Bryk, Lee, & Holland, 1993; Coleman, Hoffer, & Kilgore, 1982; Gamoran, 1996).

One reason for the lack of consistent findings on the relationship between school inputs and student outcomes is that school inputs may be necessary to create effective schools, but not sufficient.  What is equally important is how the inputs are allocated and used, which, in turn, depends on what goes on within schools.

Consequently, many researchers have sought to identify the policies, practices, and climate within effective schools.  Although there is considerable agreement that certain discernable features characterize effective schools, there is less agreement on the exact features and their relative importance.  Edmonds (1979) identified five characteristics of effective schools:  strong administrative leadership, high expectations for children’s achievement, an orderly atmosphere conducive to learning, an emphasis on basic-skill acquisition, and frequent monitoring of pupil progress.  Purkey and Smith (1983) identified two sets of variables to characterize effective schools:  nine organizational-structural variables (school site management, instructional leadership, staff stability, curriculum articulation and organization, schoolwide staff development, parent involvement and support, schoolwide recognition of academic success, maximized learning time, and district support) and four process variables that describe a school’s climate or culture (collaborative planning and collegial relationships, sense of community, clear goals and high expectations, and order and discipline).  Teddlie and Reynolds (2000) identified nine processes of effective schools:  effective leadership, effective teaching, developing and maintaining a pervasive focus on learning, producing a positive school culture, creating high (and appropriate) expectations for all, emphasizing student responsibilities and rights, monitoring progress at all levels, developing staff skills at the school site, and involving parents in productive and appropriate ways. 

Other researchers have looked beyond these discernable features of the effective schools and tried to identify the requisite building blocks necessary to create effective schools and effectively utilize school inputs.  These building blocks have to do with the characteristics of both individuals within the school and the school as a whole.  For example, Cohen, Raudenbush, and Ball (2003) argue that conventional school resources—such as teachers’ formal qualifications, books, facilities, and time—only offer the capacity to improve teaching and learning. But to do so requires the teachers’ personal resources, which they define as their will, skills, and knowledge.  Similarly, Newmann (1983) argues that new organizational structures, such as site-based management or team teaching, may do little if teachers and administrators do not have the requisite commitment and competencies.  Finally, Bryk and Schneider (2002) argue that a requisite building block for school improvement is a social resource known as relational trust, which they define as the social relationships at work in a school:  “we view relational trust as creating the fertile social ground for core technical resources (such as standards, assessments, and new curricula) to take root and develop into something of value” (p. 135).  In their simplest form, the two essential building blocks of school reform are will and capacity that characterize both individuals and schools as institutions.

One fundamental question is whether will and capacity constitute prerequisites for developing effective schools, or whether schools—with the proper incentives, resources, and support—can develop the requisite will and capacity.  Some research suggests they are prerequisite.  For example, many comprehensive school reform models require that teachers display the necessary will to reform by first voting on whether to adopt the model (Borman et al., 2003).  A comprehensive review of efforts to scale up comprehensive school-reform models found that a lack of teacher capacity led to weak implementation of the reforms (Berends, Bodilly, & Kirby, 2002).  Bryk and Schneider (2002) argue that trustworthy relationships are easier to build in schools where both students and teachers have chosen to attend, thereby assuring a shared commitment (p. 142).  This literature suggests that it may be necessary to recruit and select teachers with at least some pre-existing level of will and capacity before undertaking school reform.[10]  Yet other research suggests that, in some instances, “belief or commitment can follow mandated or coerced involvement at both the individual and system level” (McLaughlin, 1990, p. 13).[11]  In addition, certain organizational features, such as smaller school size and shared decision making, may be necessary to develop and support teachers’ will and capacity (Newmann, 1993). 

The Challenge of Reforming High Schools

The challenge of reforming high schools to reduce dropout rates is formidable.  The challenge is less formidable in the relatively large proportion of high schools where students at risk of dropping out of school comprise a small proportion of the student population.  In those schools, programmatic interventions should be sufficient to provide the additional academic and social supports that at-risk students need to stay in school.  The challenge is much more formidable in the relatively small proportion of high schools where a large proportion of students drop out, many of which are large, urban high schools.  In these schools, systemic interventions that address the needs of the entire student body will be required to transform the entire school.

What role can state and federal policies play, including NCLB, in bringing about such reforms?  There is extensive research literature examining the implementation and impact of past federal and state policies designed to bring about widespread reform of public schools.  This literature provides a very sobering assessment of the prospect for reforming schools:

Perhaps the overarching, obvious conclusion running through empirical research policy implementation is that it is incredibly hard to make something happen, most especially across layers of government and institutions.  It’s incredibility hard not just because social problems tend to be thorny.  It’s hard to make something happen primarily because policymakers can’t mandate what matters.  We have learned that policy success depends on two broad factors: local capacity and will.  Capacity, admittedly a difficult issue is something that policy can address…But will, or the attitudes, motivation, and beliefs that underlie an implementer’s response to a policy’s goals or strategies, is less amenable to policy intervention (McLaughlin, 1987, p. 172).

A recent evaluation of the ten-year effort of the New American Schools (NAS) to develop, implement, and scale up research-based comprehensive school reform models reached a similar conclusion: 

The causal chain of events leading to strong implementation and outcomes have proven to be more complex than originally considered by NAS and one that remained largely outside of its control and influence…in keeping with the literature on implementation indicating the complexity of the change process (Berends et al., 2002, p. 147).

Not only is reform difficult, it is also uneven, varying widely across schools and districts (McLaughlin, 1990). 

The difficulty and unevenness of reform efforts can be traced to a number of factors (Desimone, 2002):

1.      Specificity (locus of development, professional development information and materials, and monitoring);

2.      Consistency with other school reform efforts and with state and district policy;

3.      Authority through social norms and through institutional and individual support;

4.      Power (rewards and sanctions);

5.      Stability of students, teachers, and administrators, and of the policy environment;

6.      Local politics (unions, lack of capacity, inadequate resources).

Specificity is most related to the nature of the reform itself and has the most impact on the fidelity of implementation. Power has the most impact on short-term effects. Authority, consistency, stability, and local context have more impact on long-term effects, yet have less to do with the nature of the particular reform than with local contexts and conditions that are largely outside the control of the reform itself.  In short, even with sufficient incentives and support, local conditions and contexts remain formidable obstacles that must be addressed if widespread reforms are to be successful (Berends et al., 2002; Desimone, 2002).

·         Developing content standards that specify what all students are expected to know and do.

·         Developing assessments to measure whether students are meeting those standards.

·         Specifying performance standards that students and schools are expected to reach and a timetable for reaching them.

·         Providing a series of rewards and sanctions—high stakes—for both students and schools to meet the performance standards.

What is the theory of action underlying performance-based accountability systems?  According to Elmore (2004a):

Performance-based accountability systems operate on the theory that measuring performance, when coupled with rewards and sanctions…will cause schools and individuals who work in them, including students, teachers, and administrators, to work harder and perform at higher levels (p. 277).

Elmore argues that this theory, too, is “underspecified” (p. 278) because it remains unknown how schools and the individuals in them are supposed to respond to these rewards and sanctions in order to produce the desired outcomes.  He goes on to argue that the way schools respond is largely determined by the degree of the alignment between the requirements of the external accountability system and the school’s internal accountability mechanisms that reflect the coherence and agreement around individual responsibility and collective expectations for teaching practice and student learning (Elmore, 2004, Chapter 4).  In addition to an underspecified theory of action, performance-based accountability systems suffer from a number of other technical shortcomings, including a lack of valid, reliable, and accurate assessments of student and school performance (Furhman & Elmore, 2004).

Performance-based accountability systems rely on high-stakes incentives as the primary mechanism for reforming schools and improving student achievement, rather than providing resources and support to increase capacity.  This assumes that the primary impediment to improved student outcomes is a lack of will rather than a lack of capacity, an assumption that some observers question:

Is it plausible to assume that educators actually know how to substantially improve student performance, but that they are for some obscure reason withholding this knowledge because they have been insufficiently motivated or rewarded by existing incentive structures (Elmore, 2004a, p. 280)?

While past research on policy implementation clearly recognizes the importance of incentives (Hanushek & Jorgenson, 1996), it also clearly demonstrates that students, educators, and schools require resources and support to improve their capacity (Elmore, 2004b; McLaughlin, 1987, 1990).  Moreover, improving capacity may also lead to improvements in will (Guskey, 1989).[12]

The Shortcomings of NCLB

NCLB suffers from many of the same shortcomings as all performance-based accountability systems.  NCLB also suffers from a number of other general shortcomings as a strategy for improving schools and specific shortcomings with respect to high school graduation rates.

One of the general shortcomings of NCLB is one that afflicts all federal policy initiatives in education:  the fragmented educational system in the United States.  The constitutional authority for education in the United States resides in states that, in turn, have delegated substantial authority to the more than 15,000 local education agencies.  This fragmentation of authority and power makes it difficult for the federal government, which provides about eight percent of the funding for public education, to affect widespread and meaningful change in educational practice at the local level (Cohen & Spillane, 2002). 

NCLB suffers from this problem.  It is a complex law that utilizes the full array of policy instruments described earlier (see Table 2).  Many of the key provisions of this law specifies ambitious federal policy goals, but allow states considerable authority in how they respond to these provisions.  For example, while NLCB sets a federal policy goal that all students in the United States should be proficient in key academic subjects by the year 2014, states have the authority to set content and achievement standards.  Similarly, NCLB mandates that schools receiving Title I funds show adequate yearly progress (AYP) toward reaching the performance target for all identifiable subgroups of students (e.g., racial and ethnic minority, poor, English language learner, and disabled students), but states have the authority to determine how AYP is calculated and the size of the identifiable subgroups.  As a result, there is considerable unevenness among states in how many schools are meeting AYP goals and for which students (Sunderman, 2006).[13]  There are also inconsistencies and conflicts within NCLB as a federal accountability system and state accountability systems that have different policy goals and performance measures.[14]

Another shortcoming of NCLB is the mix of policy instruments it employs.  NCLB draws on all five policy instruments in order to achieve its ambitious policy goals, but the mix of instruments is problematic.  In particular, NCLB is heavy on mandates and short on inducements and capacity building.  It has been criticized because it fails to provide sufficient financial support for its mandates, such as annual testing, resulting in a claim of it being an “unfunded mandate” (Kim & Sunderman, 2004).[15]  More fundamental is the heavy reliance on mandates – which evoke short-term responses and less reliance on capacity building – which rely on long-term investment to improve capacity.  The immediate need to meet the short-term mandates of the NLCB performance goals undermines the ability to provide the investment needed to improve the long-term capacity of individuals and institutions to improve schools (Elmore, 2004b).  For example, under NCLB, failure of schools to meet AYP goals for two years invokes the provision that allows students to transfer to another public school in the district and failure to meet goals for three years invokes a provision that low-income students be offered supplemental educational services.  Both actions drain resources and effort away to improving the long-term capacity of low-performing schools.  Moreover, the capacity-building provisions of NCLB are focused on improving the individual capacity of teachers through professional development, not on improving the organizational capacity of schools so they can provide ongoing, collaborative professional development that will yield sustained improvements in teaching and learning (Elmore, 2004b, Chapter 3). 

In addition to the suffering from the fundamental shortcomings of all performance-based accountability systems, NCLB suffers from a number of shortcomings as a policy for reducing high school dropout rates.  These shortcomings have to do with the weak accountability provisions concerning dropout rates (Orfield, Losen, Wald, & Swanson, 2004):

1.      Whereas NCLB mandates a common performance goal—100 percent proficiency on state exams by the year 2014—NCLB allows states to set their own performance goals with respect to dropout rates.  As a result, some states have set high performance goals and others have set low performance goals. 

2.      Whereas NCLB gives states little leeway to document yearly progress toward meeting the performance goals for student achievement, NCLB allows states incredible leeway in meeting performance goals for graduation rates.  As a result, some districts could take up to 300 years to reach their performance goals for graduation rates. 

3.      Whereas NCLB requires that schools, districts, and states report and demonstrate performance of seven sub-groups of students with meeting achievement goals, NCLB does not require any breakdowns with respect to meeting graduation goals. 

Strong accountability for test score performance coupled with weak accountability for graduation rate performance provides perverse incentives for districts.  One way to raise test score performance is to remove or discharge low-performing students (Bowditch, 1993; Riehl, 1999; Rumberger & Larson, 1998).  For example, both the New York City and Houston school systems have been investigated over the past years for “pushing out” their low performing students into alternative, non-diploma programs or classifying them as transfers so they would also not be counted as dropouts (Howard, 2004; Lewin & Medina, 2003).   

Recommendations for Strengthening NCLB

Could NCLB be strengthened in such a way as to foster meaningful reform in high schools that would help reduce dropout rates?  Yes, but the task would be formidable.  It would involve not just strengthening the provisions of the law that focus specifically on high school graduation, but also altering some fundamental provisions of the law designed to improve public schools more generally.

Changes in the law involving high school graduation rates would be relatively easy.  One recommended change is to strengthen the accountability provisions for high school graduation rates to make them consistent with those for student achievement provisions so that schools have more incentive to address this aspect of student performance.  These changes should include:

1.      Setting national performance goals for school dropout rates.

2.      Establishing more rigid annual progress goals.

3.      Requiring performance goals for the same demographic subgroups required for student achievement.

Of course, as suggested earlier, simply altering incentives alone will do little to improve graduation rates without also improving the capacity of schools to respond.  This would require a substantial investment in both programmatic and systematic reforms.  More resources should be provided to fund proven dropout prevention and recovery programs in schools where programmatic reforms may be sufficient to address the dropout problem.  The federal government’s What Works Clearinghouse, which was established by the U.S. Department of Education in 2002 to provide scientific evidence on what works in education, will soon release its evaluation of scientific evidence on effective dropout prevention programs.[16]  Together, these actions would help schools to focus on the problem of school dropout and to provide some resources for them to undertake useful, programmatic changes.

But to improve high school graduation rates in the schools where the problem is endemic will require major revisions to NCLB to address some of the underlying flaws in this approach to high school reform.  Elmore identifies five design principles for accountability systems that would bring about meaningful and sustained reform in schools (2004a, pp. 290-294):

1.      Individual and collective stakes should be based on defensible, empirically based theories about what it is possible to accomplish on measured performance within a given period of time.

2.      Stakes should be based on valid, reliable, and accurate information about student and school performance.[17]

3.      Students should not be accountable for learning content they have not been taught.

4.      Schools should be accountable for the value they add to student learning, not the effects of prior instruction; school systems should be accountable for the cumulative learning of students over their career in the system.

5.      Reciprocity of accountability and capacity—for each increment in performance I require of you, I have an equal and reciprocal responsibility to provide you with the capacity to produce that performance.

Adopting these principles would require time, research, resources and most importantly, a political will to reexamine the premises underlying this reform strategy.  In essence, it would require a complete rethinking of virtually all the current provisions of NCLB.  While it is beyond the scope of this paper to comment on all of them, below are a few of the provisions that need to be revised:

1.      Performance goals.  The current performance goal of all students reaching proficiency by 2013-14 is arbitrary and unrealistic, especially for disadvantaged students who have the lowest achievement levels and therefore the most ground to make up.  To achieve this goal, schools enrolling such students have to achieve higher annual performance targets (known as “annual measurable objectives’ or AMO) and, as a result, are more likely to be labeled as failing even whey they show the same level of improvement as schools enrolling advantaged students (Kim & Sunderman, 2005).  Performance targets should be altered so that all schools are accountable for the same annual improvement no matter what the population of students they enroll.

2.      Performance measures.  Current school performance measures are based on mean proficiency levels of students, which are “scientifically indefensible” because they largely reflect differences in the students when they first enter school  (Raudenbush, 2004, p. 35).  Performance measures should accommodate differences in the background characteristics of the students who enter schools by employing value-added measures that reflect the improvement in student learning over time (Meyer, 1996; Seltzer, Choi,& Thum, 2003).  Performance measures should also account for known annual variations in performance (Kane & Staiger, 2003; Linn & Haug, 2002).

3.      Sanctions.  Currently, NCLB sanctions those who fail to make meet AMO for two years.  This timetable is unrealistic given the extensive research that demonstrates how long it takes for teachers and schools to make fundamental and lasting changes in their capacity to improve student achievement (Berends, Bodilly, & Kirby, 2002; Elmore, 2004b; Guskey, 1989).  The timetable for sanctions should be reevaluated coupled with revisions in performance goals so that schools that demonstrate reasonable improvement or schools engaged in proven comprehensive school reform efforts are given more time to improve.

4.      Capacity building.  NCLB relies more on pressure than support to improve schools.  This balance should be changed.  Moreover, the types of support provided needs modification.  Currently, support comes largely in the form of categorical funds for disadvantaged students and limited support for school improvement and teacher professional development.  Instead, more support should focus on capacity building, not simply for educators, but also for schools based on the idea that many schools lack the institutional capacity to improve student performance (Elmore, 2004b).  To improve the capacity of schools requires not simply more resources and information on effective practices, but a system of technical assistance delivered through intermediate institutions.  Such a system would need to focus on implementing long-term, systemic reforms in schools rather than simply providing short-term, programmatic assistance, as was often the case in past efforts (Haslam & Turnbull, 1996).

Conclusions

Improving the nation’s high school graduation rate remains a formidable challenge.  Performance-based accountability systems, such as NCLB, attempt to improve graduation rates by improving public schools.  However, these systems lack a sound and tested theory of action as to how students and schools are supposed to improve.  In particular, they rely on high-stakes sanctions designed to motivate students, educators, and schools to improve without providing sufficient resources and supports to develop the capacity to improve.  The research literature suggests that such a strategy is unlikely to yield sustained and meaningful improvements in teaching and learning, or to improve graduation rates.  For this strategy to be effective will require a better balance of incentives, resources, and support based on valid, comprehensive measures of student and school performance and a tested theory of action.  In other words, it will require a complete rethinking of NCLB and state-level performance-based accountability systems

Yet even with substantial revisions, school-based strategies are limited because schools alone cannot be the sole arena for the improvement of student performance.  Existing research provides substantial evidence that most of the differences in students’ achievement are due to differences in their families and communities, not the schools they attend.  Therefore, to improve student outcomes and close the achievement gap will require addressing the pervasive inequalities found in family and community resources (Armor, 2003; Rothstein, 2004).  And it will require social investment in early childhood (preschool), where research demonstrates that early investment lowers the cost of later investment, such as high-school dropout prevention (Heckman, 2005).

Revised version of a paper presented at the roundtable, “The No Child Left Behind Act: How Does It Affect High School Reform?” held in Washington, D.C., October 14, 2005.  I would like to thank Dan Losen, Gail Sunderman, Jill Morningstar, and Lorraine McDonnell for their helpful comments on earlier versions of this paper.