NO CHILD LEFT BEHIND Costs and Benefits The promise of providing all children with a high-quality education is a noble one. But after looking at the projected costs for 10 states to fulfill the requirements of NCLB, Mr. Mathis fears that the federal government is asking too much and giving too little. By William J. Mathis

IT IS THE cruelest illusion to promise far more than we will ever deliver. Yet throughout time reformers of all persuasions have offered Utopian visions in exchange for permission to shape the world to their view. With great fanfare about historic turning points and fervent promises to America's children, in January 2002 President Bush signed into law the No Child Left Behind (NCLB) Act, the latest reauthorization of the Elementary and Secondary Education Act (ESEA).

The rhetoric was certainly noble, and the law was sold with the guarantee that, at last, we would leave no student behind. The poor would have the same as the rich, and the strong arm of a resolute government would make it so. Public support for equality, periodic testing, highly qualified teachers, and other provisions of the law was strong.1 As shown by the 87-10 Senate vote, the law passed with substantial bipartisan support.

President Bush and Secretary of Education Rod Paige have said much about the great investments the federal government has made in education. And in strident tones, the material accompanying the passage of the law says that the public has a right to demand great returns on this investment.2

Alas, the promises are far greater than the reality. When the "historic" federal investments in education are scrutinized, the first-year increases to Title I compensatory funds amount to a mere 0.4% of total education spending. When the much-touted "flexibility" procedures that NCLB gives to local districts are examined, they allow, at best, a local district to shift around about 4.3% of its already-committed money.3 When the so-called adequate yearly progress provisions of the law are examined, independent reviewers, almost without exception, say the plans are unrealistic.4 Submerged beneath emotional appeals and rhetorical demands, hard questions about costs, the adequacy of resources, and the strength of commitments lie hidden.

The Nation's Financial Commitment

Throughout the last century, critics loudly proclaimed the nation's peril owing to the alleged poor condition of the schools. Yet results from the National Assessment of Educational Progress show that at the end of the century scores in reading and mathematics had leveled off at a 30-year high, dropouts were near all-time lows, and our nation's economic supremacy was unquestioned.5 This is hardly a picture of a "failed" system. But these facts hide the nation's true educational problems.

Much has been made of the "merely average" test scores of U.S. students in comparison with those from other countries. To be sure, U.S. scores on international examinations -- such as the Third International Mathematics and Science Study (TIMSS) and the Program of International Student Assessment (PISA) -- are at international averages in reading, math, and science.6 However, it is just as clear that the U.S. investment in K-12 education is also less than stellar. We spend the same average amount of our gross domestic product on elementary schools as other developed countries, but we fall to the bottom half in our commitment to high schools.7

The greater and more insidious danger, however, is the disparity in achievement within the United States. International test data tell us that we have the greatest inequities between our highest- and lowest-scoring students of any nation.8 In a UNICEF follow-up study, the gap between our average scorers and our low scorers gives the U.S. an abysmal ranking of 21st out of 24 industrialized nations in educational equality.9 While we are getting more productivity than we pay for, the troubling disparities in achievement reflect our disparities in funding

NCLB Costs: What We Spend Versus What We Need

In 1989 the Kentucky Supreme Court ruled that the state constitution requires the state to provide schools and students with the resources necessary to meet high state standards. Since that time, state courts in New Hampshire, Tennessee, Arkansas, Ohio, North Carolina, and New Jersey have issued similar rulings on behalf of children. In addition to court actions, the vast majority of states have enacted standards-based reforms. NCLB adopts these state standards and imposes progressively harsher penalties on schools if students do not get passing scores.

Thus figuring out how much NCLB will cost requires knowing how much it will take to ensure that all students meet the standards and pass the tests. But how do we know how much money is enough? Different methods have been used to estimate what is an adequate amount of money.

The "professional judgment" method uses panels of experts to carefully define the resources needed for each child to meet the standards. These resources are then added up to arrive at a state figure. The "successful school" technique identifies a set of high-achieving schools, examines their resource allocations and spending levels, and generalizes to other schools. The "statistical analysis" approach calculates what it takes to predict a passing score. These models are particularly useful in determining regional costs, such as what it would take to attract qualified teachers to a remote location.

Within the last four years, a new generation of finance studies has estimated the costs of raising all children's test scores up to a particular state's standard. While some of these studies expressly include NCLB costs, most have been based on achieving the state's own standards -- which have since been folded into each state's NCLB system. Since each state determines its own standards, has its own social and political culture, and has its own level of student needs, a great variety of outcomes exists.

Nevertheless, recent studies in different states, by different researchers, using different methods, reveal a picture of the massive costs of making sure all children pass the mandated NCLB tests.

Indiana. To enable a school to meet the "commendable" level on state tests, Indiana would have to increase its base spending from $5,468 to $7,142 per pupil -- a 31% increase -- according to an analysis by Augenblick and Myers, Inc. These estimates do not include any added costs for special education students, which range between $7,500 and $8,300 per pupil. They also do not include the cost of "hard-to-serve students," who average an additional $4,200 to $5,300 per student.10

Maryland. Calculating the costs for Maryland students to meet state standards, Augenblick and Myers, Inc., arrived at a total education cost of $12,060 per pupil for elementary schools, $9,000 for middle schools, and $9,599 for high schools. The firm further calculated that having a low-income student meet standards would require an average excess cost of $7,748 per student, or 1.7 times the base cost. The analysts used both a market-basket model and a high-achieving-school model to arrive at costs for their standards-based models. The results from both methods were similar.11

The total cost for the system for fiscal-year 2000 would have been between $7.9 and $8.8 billion. Since the expenditures for that year were $5.9 billion, the required percentage increase was between 34% and 49%. To Maryland's credit, its lawmakers boosted education spending by $1.3 billion in spring 2002.12

Montana. Montana's 2002 study was sponsored by five education organizations and assisted by the National Council of State Legislators. The analysts used a professional-judgment approach to cost out meeting NCLB requirements based on the current level of performance. They found that a base cost between $6,004 and $8,041 per pupil (depending on district type) was required, while the current base was only $4,471. Additional special-needs and remedial costs were $8,000 and $2,000 per pupil respectively. Thus base costs in Montana would increase between 34% and 80%, depending on location and level of need.13

Nebraska. The state department of education, in cooperation with various education organizations, commissioned a study of what it would take to meet current Nebraska standards under NCLB in 2002-03. Estimated costs range from $5,845 per pupil in a large K-12 district to $11,257 in a small, isolated K-12 district.

On top of this figure, at-risk and special-needs students would require an additional $1,500 to $12,000 each, depending on the level of need. "Total costs would vary, on average, from $8,103 per student in large K-12 districts to $13,525 per student in very small K-12 districts," says the report.14 Nebraska currently spends about $5,600 per pupil. Thus the state is looking at a 45% cost increase. NCLB testing and labeling have brought cries of outrage from Nebraska state and local officials. The state senate called for full federal funding of the mandate.15

New Hampshire. Mark Joyce, executive director of the New Hampshire School Administrators Association, sent his members and the citizens of the state his analysis of NCLB costs. He found that the state will receive an average of $77 of new federal money for each of the Granite State's 220,000 students, while the obligations imposed by the law will cost $575 per student. In other words, New Hampshire will receive about $17 million in new money for new obligations of $126.5 million.16

To arrive at this number, Joyce estimated a state cost for each of the elements of the law and added them together. He contends that his estimates are conservative, and he is probably right. The reason is that his analysis was confined to increased costs for local and state staff and administration. He assumed that the number of special education students would increase by 2%, but he did not include the costs of remedial programs for underachieving children. As compared to other states, this procedure results in a significant underestimate.

New York. Using a statistical technique primarily focused on regional differences in the costs of meeting standards, Professors William Duncombe of Syracuse University and Anna Lukemeyer of the University of Nevada, Las Vegas, arrived at a median statewide figure of $7,927 for extra remedial costs, on top of the regular per-pupil expenditure of $9,781. They provide several regional cost variations at different proficiency standards. Their overall regional cost adjustments add 16% to total education spending. New York's Campaign for Fiscal Equity launched a major costing study using both the successful-schools and professional-judgment models. A report of the results is expected in early 2004.17

South Carolina. To estimate the costs of getting 85% of South Carolina students to the "basic" level of the state's Palmetto tests and all students to pass the graduation tests in 2011, the 1999 base cost of $4,990 would have to be increased to $6,189 by 2005-06. This figure represents a 24% increase. However, it does not include the costs of at-risk and special education students. When figures for these populations are added in, the cost rises to $9,182 per pupil (an 84% increase), according to Augenblick and Myers, Inc., whose analysts used a professional-judgment model.18 Spending in South Carolina must nearly double --
going from $3.1 billion in 1999 to a projected $6 billion in 2006.

Texas. While Texas saw large increases in the percentage of students passing the state test, these tests were at an eighth-grade, basic-skills level.19 Even with the low standards, statistical modeling of NCLB costs on earlier data would require an increase in state aid of 101% or 6.9 billion new state dollars. Assuming the local contribution remained the same, this is about a 35% increase in spending. For comparison purposes, the Bush Administration has proposed a $1-billion increase for the nation as a whole for fiscal year 2004.20 In Texas, the largest increases are needed in the districts with very low-wealth populations and in the very large-city districts. A new test is being implemented in Texas. Obviously, if the standards are raised, the remedial costs will go up proportionately.

Vermont. In my own study of Vermont's situation, I counted the number of students below state standards. Depending on the test and grade level, Vermont scores between 22 and 32 percentile points above national norms, and this advantage over the nation is increasing. Nevertheless, because Vermont has extremely high standards, 46.5% of the students "fail" one of the tests. I assumed that one-fourth of these students would be able to reach the standards within existing resources. Using estimates from adequacy-cost studies and looking at the number of students affected by poverty and those with moderate needs, I arrived at additional remediation costs for the state of $149.5 million. Testing costs and lost instructional time added $8.7 million to that figure, for a total of $158.2 million in new costs. However, the state receives only $51.6 million in all titles of ESEA combined.21 Vermont would add 15.5% to its total school costs for remediation and testing alone.

Wisconsin. Using parameters supplied by the Institute for Wisconsin's Future, Whitney Allgood and Richard Rothstein found that adequate funding in Wisconsin would be $11,231 per pupil, averaged across all pupils in the state. For high-risk pupils, the cost would be $27,879 per pupil -- more than 2.5 times the cost of previous estimates. In arriving at this figure, the authors demonstrated that overcoming the effects of poverty requires interventions beyond the traditional school. Thus they included community clinics, before- and after-school programs, early childhood intervention, and summer school programs.22

Simply teaching children will have little effect if they return to bad neighborhoods, single-parent homes, foster care, inadequate health care, and a general lack of support. The authors marshaled convincing evidence that expecting students to reach high standards without essential support systems in place overestimates the ability of schools to cure social ills.23

A follow-up study in 2002 by the Institute for Wisconsin's Future determined that $11,121 per pupil in school spending was needed, but the current level was only $8,241. The difference represents a 35% increase for Wisconsin spending.24

Estimating the National Costs of NCLB

These cost studies from 10 states are all based on bringing the state's children up to an academic standard. As we have seen, they vary considerably in methods, assumptions, and procedures, and they use a variety of analytic techniques. All are recent. Yet, for all their diversity, a number of unambiguous findings emerge.

• Providing a "standards-based" NCLB education for all children will require massive new investments in education spending. Seven of the 10 studies show increases in base cost that are greater than 24%, and of these, six were between 30% and 46%; two were in the 15% range; one did not directly address the base cost.
• Traditional estimates of the costs of remedial instruction, such as Title I or state-funded programs, are clearly underestimates at both the state and federal levels.25 Eight of the 10 studies found the real additional costs to be approximately 100% higher -- that is, double the cost of regular instruction. The other two studies did not address or break out these costs.
• The federal government boasts that it is fully paying Title I NCLB costs with the additional $1 billion planned for fiscal year 2004. As indicated by the New Hampshire study, it is not likely that such a small appropriation will pay for the added bureaucracy, testing requirements, qualified-teacher costs, paraprofessional tests, and other mandates in the law.
• Perversely, states with high standards, such as New York, Michigan, and Vermont, will have the highest remedial needs and costs while those with low standards will have the lowest costs.26

Public spending on K-12 education was $422.7 billion in 2001-02.27 If we use a broad -- yet easily justified and extremely conservative -- estimate of 20% added costs for the nation as a whole, that translates into a national increase of about $84.5 billion. An estimate of 35% additional costs yields a national increase of $148 billion.

For comparison purposes, the current federal Title I appropriation is $11.3 billion, and the Administration's budget request of $12.3 billion is below the authorized amount of $18 billion in NCLB.28 President Bush said in his weekly radio address of 4 January 2003 that the additional $1 billion was "more than enough money" and that "we are insisting that schools use that money wisely."29

Who Has to Pay the NCLB Bill?

Legal scholars have opined that the federal government cannot be sued to force adequate funding of the law. In fact, Secretary of Education Richard Riley in a letter dated 19 January 2001 said that states have the responsibility of providing educational resources to meet new standards: "Indeed, raising standards without closing resource gaps may have the perverse effect of exacerbating achievement gaps and of setting up many children for failure."30

The alternative for states is to reject the federal money and, along with it, the mandates. However, if states take the money and require local districts to meet state standards, then these same local districts can legally demand that the state provide adequate money to meet these standards. Local districts can cite a growing number of financial adequacy studies to support their case in the courts.

With the National Governors' Association estimating that states face a total fiscal-year 2003 deficit of $58 billion, state governments will be hard pressed to fund an additional $84.5 billion -- to say nothing of $148 billion. In many states, budgets are being balanced in part by cutting education dollars. Tax cuts, a sluggish economy, and the cost of war in Iraq all suggest that significant fiscal help will not be forthcoming from the federal government.

Few reasonable people argue against the idea that all children must be well educated or that extra services must be made available for our most needy. In fact, it has been the dream of many educators throughout our nation's history. However, if funding remains inadequate, then at best the law will represent the attenuated efforts of an overpromising government, which will leave behind our poorest and most needy children.

Promised Benefits: Accountability

The primary promised benefit of NCLB is that 95% of all student groups will reach their state test standards by 2014. Obviously, we don't know if that goal can or will be reached. But if the system is not adequately funded, then reaping that benefit is a remote and forlorn hope.

Assessing the possible benefits of NCLB requires answering two questions -- one technical, the other about values. The technical question is whether the system can ever work. At the heart of the plan is the requirement that each subgroup of students in each school improve test scores in equal yearly increments. That is, they are required to make "adequate yearly progress" (AYP). The values question is whether the goals of the system, narrowly conceived as improved test scores, are the right goals for public education in a democratic society.

Promised Benefits: 'Adequate Yearly Progress'

There is simply no body of accepted scientific knowledge that says that all students and all subgroups of students can reach meaningful high standards, at the required AYP pace, given the levels of funding and the lack of social, economic, and family assets of many of our children. It is also doubtful that the "machinery" will work. Indeed, there is scant evidence that the AYP train can even get out of the station.

Test score reliability. The centerpiece of NCLB is the requirement that test scores must improve annually. Before NCLB became law, Thomas Kane and Douglas Staiger demonstrated that 70% of the year-to-year change in test scores for grade levels or schools is simply random variation.31 Differences in the student body from one year to the next, combined with the statistical error in the tests themselves, make it impossible to know whether the tests are measuring real gains (or losses) or whether the changes are merely random noise.

Similarly, Walt Haney examined the scores of all Massachusetts schools. He found that those that received a medallion for large gains in one year saw those gains disappear the following year.32 In Florida, the same pattern emerged, with 69% of the schools that posted gains in the first cycle of testing falling back in the next cycle.33 In Maine, Jaekyung Lee found the same phenomenon and noted that the random fluctuation, not surprisingly, increased as the size of the school decreased.34

The problems become far more difficult as the number of subgroups increases. A school with a diverse population (and many subgroups) has many more opportunities to fail. Thus the diverse school, which faces greater challenges, is penalized.

Likewise, many rural and small schools do not have enough students in a grade level or a population subgroup to draw valid conclusions. While some states call for a minimum of 30 or 50 students in a subgroup, new analyses are finding that these cell sizes are still too small to validly measure AYP gains. For example, modeling in Vermont shows that 170 students per grade level are needed for the scores to be stable.35

Validity. Most states claim that their testing systems are "aligned" with their state curriculum standards. Generally, this means that they are not grossly incompatible. It does not mean that they are a faithful, accurate, and balanced representation of the state's standards for instruction. Since most tests are geared toward reading and mathematics, social studies and science get short shrift. Even within math and reading, the ability of any test to sample validly an ever-expanding knowledge base is suspect. The reading and math wars demonstrate that even these basic areas are subject to great controversy. Moreover, different tests give different results for the same students -- even when they are supposedly measuring knowledge of the same subject matter.36

Testing companies, state agencies, and local districts all have their own incentives to keep the time, amount, and expense of testing to a minimum. The result, unfortunately, is a tradeoff. Thus it is doubtful that any state accountability test can be considered a valid and representative sampling of the state's curriculum expectations for an educated youth.

Improvement of learning through tests. Each year, the NCLB system progressively increases sanctions against schools that do not meet annual growth targets. Ultimately, the state could take over the school, change its management, or disband it altogether. The assumption is that the fear of these penalties will drive schools to even higher levels of performance.

Leaving aside whether schools have the resources and whether students have the social capital to reach the high levels sought by NCLB, it is questionable whether punitive incentive systems work. (B. F. Skinner disproved negative reinforcement systems 45 years ago.) In looking at 18 states with high-stakes testing systems, Audrey Amrein and David Berliner considered the scores on the high-stakes tests along with the scores on other tests. If all scores went up, they concluded that learning was taking place. If only the high-stakes scores went up, they concluded that test preparation and curriculum narrowing were taking place. They found that scores on the other tests were not related to scores on high-stakes tests. Thus the basic assumption that high-stakes systems lead to improved learning must be suspect at least.37

Texas is cited as a state in which the increase in the percentage of students meeting the standards was paralleled by increases in the state's scores on the National Assessment of Educational Progress. However, the low level of the state's tests and the very different trend lines of the state and NAEP tests call this conclusion into question. More troubling still is that the increase in test scores was not accompanied by increases in outcomes of high value, such as increased high school completion or college attendance.38

Unintended Consequences

The assumption of the NCLB system is that the test results represent what an educated person should know and be able to do. Few would say that an educated person only has high test scores. Most would say good scores are desirable but not sufficient to define an educated person. Most would say that schools must also produce good citizens, strong family members, contributors to society, and people engaged in democratic governance. None of these characteristics are measured by or deemed of importance in the federal accountability system. Along the way, a number of unintended consequences also appear likely.

Curriculum narrowing. As noted above, statewide achievement tests do not measure the vast expanse of curriculum set forth by states and school districts. Tests tend to measure those things that are easy to measure, in an efficient and economical way. This means that the focus is on lower-order thinking skills, with a light smattering of higher-order skills, such as writing a short essay.39 Schools and teachers, faced with ever-increasing demands to avoid the "failing school" label, will logically focus on the curriculum content that is most likely to improve test scores. Leaving aside the fact that these tests provide little useful instructional feedback, the inevitable results will be that the nation's curriculum will be narrowed and the level of expectations will be lowered.40

Failing schools. While the federal government has recently announced that the "failing" label for schools should be replaced with the more politically acceptable term "in need of improvement," the negative moniker sticks in the minds of the people and of the media. Regardless of the solid record of achievement test scores and the good graduation rates of the nation's schools, public school critics have been successful in painting schools as "failing," and they have made the most of the cooperation of the media, which have natural incentives to report negative news.

A plethora of estimates have been put forth regarding the number of schools across the U.S. that will turn out to be failing under NCLB. The Center for Assessment says 75%, North Carolina estimates 60%, Vermont calculated 80% over three years, and Louisiana reports 85% -- even though two-thirds of their schools show improved scores.41

Furthermore, as Lowell Rose has pointed out with regard to Indiana, "A failing label will be assigned frequently, based on the crushing impact of poverty." Students with large and diverse populations will find it most difficult to show progress while schools with a breakout group in special education will find it impossible.42 Black students showed a 94% failure rate, while Hispanics registered a 68% failure rate. Students who received free and reduced-price lunches showed a 56% failure rate.

Schools labeled as "failing" will not receive their label because they have failed. Rather, schools will be branded because they are in poor or diverse neighborhoods, because they are small and rural, because they are underfunded, and because the AYP system cannot tell the difference between a learning gain and random noise.

Dropouts. While it is still too early to determine whether students will drop out of school as a result of the NCLB requirements, an examination of the national longitudinal database shows that students subjected to eighth-grade promotion examinations are more likely to drop out by 10th grade.43 Anecdotal evidence suggests that some students are encouraged (or provided subtle incentives) to drop out. This is consistent with the "uncertainty principle" mechanisms set forth by Amrein and Berliner. Simply put, the more intense the negative consequences held over a system in an effort to get high results, the more likely the system is to game the rules to show better results.

Conclusion: Sure Costs, Uncertain Benefits

The No Child Left Behind law claims noble aims and sets unyielding expectations for schools. Yet there is a troubling difference between the language of the federal government and its actions. While asking for the highest educational achievement scores in the world, we ignore that we are, at best, mediocre in the commitment of our substantial wealth to our schools. When it comes to equality for all our children, the U.S. is among the least equitable nations in the world.

It is in this context that NCLB has promised equality for all. Yet in the 10 states profiled in this analysis, the costs for making these promises a reality are far from being met. Seven of the 10 states require new base investments in education of at least 24%. The federal Administration has asked for an increase of $1 billion in Title I, but we need at least $84.5 billion if we are to make a realistic effort to leave no child behind. The states, currently wallowing in deficits totaling $58 billion, will be legally forced to take on these added burdens, but they lack the capability. With war pushed to the front burner and another tax cut planned, there is little reason to believe that federal commitment in the form of federal dollars will follow federal rhetoric.

If we were willing to fund our educational obligations to the poor and the needy, the social benefits would be enormous. But funding alone will do little to stop an unworkable AYP system from randomly assigning punitive sanctions to our schools. The system does not recognize that a hungry child with a poor, single parent and a violent home may not be focused on phonics each morning. The system does not ensure adequate money for an underfunded school. It gives no promise that children will not have to go to a dilapidated school. The system makes no distinction between a school with well-educated parents and generous resources and an impoverished school. Both schools are held to the same standard.

The program is likely to increase the number of dropouts, narrow the curriculum, and label a great many schools as failing -- even as NAEP reading and mathematics scores are at very high levels. The effect will not simply be to punish schools and children for failing when they never had a chance. The effect will be that our society accepts a meaner vision of what it means to be educated in America. The effect will be to take money from those schools and those communities that need it most and transfer it to "successful" schools. Ultimately, the effect will be to shift the purpose of schools away from education for a democracy and away from the provision of equal opportunities for all children.

If we are to work seriously to attain the goal of educating all children, there are a number of requirements that must be met.

• Funding for education, prevention, and remediation must be adequate. This will require major new investments -- particularly in poor, rural, and inner-city environments. We must undertake this effort not because it is the law but because it is what we should do.
• Ultimately, adequate funding is not a matter of fiscal capacity as much as it is a matter of political will. Obtaining adequate funding will require a level of political involvement not traditionally seen from educators.
• States and districts must conduct their own cost/benefit analyses. Even though schools desperately need the small Title I sums they receive, districts and states should reject Faustian bargains that fail to compensate them adequately for the obligations they take on. The inevitable result is that the district agrees to being publicly branded a failure when it never had a reasonable chance for success.
• States and districts must work with federal officials -- elected and appointed -- for the repeal or massive revision of the NCLB law so that it provides a workable accountability system. This system must include comprehensive and democratic conceptions of educational goals rather than a narrow reliance on tests. It must also measure gains according to where students are coming from as much as where they are going.
• Finally, educators must embrace accountability. We must work to ensure that no school provides substandard, inadequate, or inequitable educational programs. We must do so not because it is politically expedient but because it is what we owe the children, our society, and ourselves.

PDK Home | Site Map
Kappan Professional Journal
Last updated 25 April 2003
URL: http://www.pdkintl.org/kappan/k0305mat.htm
Copyright 2003 Phi Delta Kappa International