Book | Education

Class and Schools: Using Social, Economic, and Educational Reform to Close the Black–White Achievement Gap

Class and Schools: Using Social, Economic, and Educational Reform to Close the Black–White Achievement Gap

Click here to order

Table of Contents | Prefaces | Introduction | About the author | Press release | Purchase this book

Table of contents

Lawrence Mishel, Economic Policy Institute
Arthur E. Levine, Teachers College


Chapter 1: Social class, student achievement, and the black_white achievement gap
The legacy of the Coleman report
Some common misunderstandings about the gap
Genetic influences
Social class differences in childrearing
Cultural influences on achievement, and black underachievement
Health differences and school performance
Housing and student mobility
Social class differences between blacks and whites with similar incomes
Does culture or social class explain the black_white achievement gap?
Summer and after-school learning

Chapter 2: Schools that ‘beat the demographic odds’
The success of some poor children doesn’t mean that poverty doesn’t matter
Dr. William Sanders and the Tennessee value-added assessment system
The Heritage Foundation’s ‘no excuses’ schools
The Education Trust’s ‘high-flying’ schools
’90/90/90′ schools, and Boston’s Mather School
Pentagon schools
Rafe Esquith, KIPP, and affirmative action programs like AVID

Chapter 3: Standardized testing and cognitive skills
Standardized tests’ imperfect description of the gap
Defining proficiency
Alignment of tests, standards, and instruction
The inaccuracy of tests that hold schools accountable for closing the gap

Chapter 4: The social class gap in non-cognitive skills
The goals of education, including non-cognitive goals
The anti-social score gap
Affirmative action’s evidence of leadership: Bowen-Bok and the ‘four percenters’
Persistence in school, self-confidence, and adult earnings
Complementing school curricula with civil rights enforcement
Testing integrity, personality, and employability
Civic and democratic participation
Perry Preschool, Head Start, and Project STAR
Comparing school and social reform to improve cognitive and non-cognitive skills

Chapter 5: Reforms that could help narrow the gap
School integration, and Sen. Moynihan’s call for making choices
Income inequality
Stable housing
School-community clinics
Early childhood education
After-school programs
Summer programs
The dangers of false expectations, and adequacy suits
Teacher morale


Appendix. What employers say about graduates


In this book, Richard Rothstein asks us to view the black-white and low- to middle-income achievement gaps with a wider lens. His revealing and persuasive analysis of how social class shapes learning outcomes forces us to look at the differences in learning styles and readiness across students as they enter school for the first time. Further, he prods us to consider the influence of income, health, safety and other gaps affecting students as they proceed through school. Even the racial and income gaps facing adults play a role, particularly as students look to their elders for signs of a payoff to education and sometimes find the evidence lacking. Consequently, according to Rothstein, addressing the achievement gap requires no less than a significant transformation of social and labor policy along with extensive school reform.

Such an analysis provides little room for easy answers and leaves few institutions off the hook. A few inspiring, dedicated teachers will not do the trick. Nor will higher expectations, in isolation, yield big payoffs for those left behind. In fact, school reform itself must be supplemented by a comprehensive compensatory program in the early years of school, along with after-school, summer, and pre-kindergarten programs. Holding schools accountable may be part of the answer, but what schools can do, even when they are at their best, will solve just part of the problem.

One can hope and expect that Rothstein’s analysis will catalyze a broader discussion of how to enhance learning in our society and, even more hopefully, inspire a broader commitment to addressing the gross inequalities that pervade American life. One can also hope that researchers will pursue empirical examination of many of Rothstein’s conjectures so that our understanding of education and the learning process deepens further.

The Economic Policy Institute is proud of its long association with Richard Rothstein, who has been an EPI research associate for over a dozen years, and pleased to publish this pathbreaking work along with Teachers College at Columbia University.

— Lawrence Mishel
President, Economic Policy Institute

Two important things happened at Teachers College in 2004. First, the college completed a two-year strategic planning exercise and determined that we would focus our resources — people, programs, and dollars — on what we believed to be the most urgent issue facing education — educational inequity, which is popularly called “the achievement gap” — the chasm in access, expectation, and outcomes in education between low-income and high-income populations; whites/Asians and blacks/Hispanics; urban and suburban residents, etc.

We are convinced this problem is to education the equivalent of AIDS or cancer in health care. It is a scourge that robs children of their futures. Today’s information economy demands more education and higher levels of skills and knowledge for employment than ever before in history. Children on the lower end of the achievement gap without adequate skills, knowledge, and education have little chance for economic well-being in this country. When a quality education is denied to children at birth because of their parents’ skin color or income, it is not only bad social policy, it is immoral.

Teachers College chose to focus on educational equity as an affirmation of our historic mission of serving urban and disadvantaged populations. We believed, too, that our faculty, embracing three major fields — health, psychology, and education — had the capacity to study the issue comprehensively because it involves not only schools, but also families, communities, and social services such as health care.

The second important happening is that Richard Rothstein, an eminent scholar and former education columnist for The New York Times, came to Teachers College as the Sachs Lecturer. He gave three extraordinary lectures on the achievement gap. Rich with data, they comprehensively outlined the causes of the gap, identified the fallacies and misperceptions regarding the gap, redefined the meaning of the achievement gap, and offered proposals to narrow the gap. The lectures were mesmerizing, drawing a crowd even during a blizzard. They became the talk of the college, providing content not only for classes, but also for sidebar conversations at meetings on administrative matters and lunch discussions around tables in the cafeteria.

Given TC’s new focus, we asked Richard Rothstein if we could publish an extended version of his lectures. We thought his work both groundbreaking and fundamental to understanding educational equity. Accordingly, we wanted it to be our inaugural publication on the subject.

Four things make Richard Rothstein’s book unique.

First, recent years have brought an avalanche of publications on the achievement gap, characterized much too often by simplistic sound bites and ideological blinders. In contrast, Richard Rothstein’s book, drawing on a wealth of previous research, provides the most intelligent, compelling, and comprehensive analysis of the causes of the achievement gap I have ever read. He demonstrates that the problem cannot be attributed simply to variations in the quality of the schools attended by children of different races and economic classes. He shows that schools alone cannot provide a remedy. Along the way, the author presents evidence debunking the popular myth that there are super-schools capable of eliminating the gap. Presenting study after study, Rothstein highlights the far greater impact of health care, nutrition, parents, home, and community.

The author makes a second invaluable contribution in this volume. He enlarges upon the traditional conception of the achievement gap, which has almost universally focused on disparities in cognitive or academic achievement. Rothstein makes the case that non-cognitive outcomes — attitudes and behaviors — must be incorporated as well. He presents data to show both that these affective outcomes are important to employers, and that an achievement gap exists in this area as well.

Third, Richard Rothstein makes a series of policy recommendations on how to narrow the gap, focusing on education, health, housing, and income differentials. He proposes earned income tax credits, policies to stabilize family housing, school-community health clinics, early childhood education, after-school programs, and summer programs.

Fourth, based on his recommendations, the author does something very unusual. He matches his ideas with dollars, determining the cost of significantly reducing the achievement gap.

In sum, Richard Rothstein has written a unique and powerful volume that needs to be read by scholars, policy makers, and practitioners who have the capacity to shape tomorrow.

I am enormously grateful to Larry Mishel and the Economic Policy Institute for making this volume possible. He joined the college in publishing this book. He produced the volume in only a few months, warp speed in academic publishing, and disseminated it to the leading policy makers and practitioners in this country.

— Arthur E. Levine
President, Teachers College, Columbia University


The 50th anniversary of the Supreme Court’s school desegregation order has intensified public awareness of the persistent gap in academic achievement between black and white students. The black-white gap is partly the difference between the achievement of all lower-class and middle-class students, but there is an additional gap between black and white students even when the blacks and whites come from families with similar incomes.

The American public and its political leaders, along with professional educators, have frequently vowed to close these gaps. Americans believe in the ideal of equal opportunity and also believe that the best way to ensure that opportunity is to enable all children, regardless of their parents’ stations, to leave school with skills that position them to compete fairly and productively in the nation’s democratic governance and occupational structure. The fact that children’s skills can so clearly be predicted by their race and family economic status is a direct challenge to our democratic ideals.

Policy makers almost universally conclude that these existing and persistent achievement gaps must be the result of wrongly designed school policies — either expectations that are too low, teachers who are insufficiently qualified, curricula that are badly designed, classes that are too large, school climates that are too undisciplined, leadership that is too unfocused, or a combination of these.

Americans have come to the conclusion that the achievement gap is the fault of “failing schools” because it makes no common sense that it could be otherwise. After all, how much money a family has, or the color of a child’s skin, should not influence how well that child learns to read. If teachers know how to teach reading, or math, or any other subject, and if schools emphasize the importance of these tasks and permit no distractions, children should be able to learn these subjects whatever their family income or skin color.

This common-sense perspective, however, is misleading and dangerous. It ignores how social class characteristics in a stratified society like ours may actually influence learning in school. It confuses social class, a concept that Americans have historically been loath to consider, with two of its characteristics, income and, in the United States, race. For it is true that low income and skin color themselves don’t influence academic achievement, but the collection of characteristics that define social class differences inevitably influences that achievement.

Recognizing social class and its impact on learning

This book tries to explain how social class differences are likely to affect the academic performance of children. For example, parents of different social classes often have different styles of childrearing, different ways of disciplining their children, different ways of communicating expectations, and even different ways of reading to their children. These differences do not express themselves consistently or in the case of every family; rather, they influence the average tendencies of families from different social classes.

That there would be personality and childrearing differences, on average, between families in different social classes makes sense when you think about it: if upper-middle-class parents have jobs where they are expected to collaborate with fellow employees, create new solutions to problems, or wonder how to improve their contributions, they are more likely to talk to their children in ways that differ from the ways of lower-class parents whose own jobs simply require them to follow instructions without question. Children who are raised by parents who are professionals will, on average, have more inquisitive attitudes toward the material presented by their teachers than will children who are raised by working-class parents. As a result, no matter how competent the tea

cher, the academic achievement of lower-class children will, on average, almost inevitably be less than that of middle-class children. The probability of this reduced achievement increases as the characteristics of lower-social-class families accumulate.

Many social and economic manifestations of social class also have important implications for learning. Health differences are among them. Lower-class children, on average, have poorer vision than middle-class children, partly because of prenatal conditions, partly because of how their eyes are trained as infants. They have poorer oral hygiene, more lead poisoning, more asthma, poorer nutrition, less adequate pediatric care, more exposure to smoke, and a host of other problems. As will be discussed in this book, each of these well-documented social class differences is likely to have a palpable effect on academic achievement, and, combined, the influence of all of these differences is probably huge.

The growing unaffordability of adequate housing for low-income families is another social class characteristic that has a demonstrable effect on average achievement. Children whose families have difficulty finding stable housing are more likely to be mobile, and student mobility is an important cause of low student achievement. It is hard to imagine how teachers, no matter how well trained, could be as effective for children who move in and out of their classrooms as teachers can be for children whose attendance is regular. Differences in wealth between parents of different social classes are also likely to be important determinants of student achievement, but these differences are usually overlooked because most analysts focus only on annual income to indicate disadvantage. This practice makes it hard to understand, for example, why black students, on average, score lower than white students whose family incomes are the same. It is easier to understand this pattern when we recognize that children can have similar family incomes but be ranked differently in the social class structure, even in economic terms: black families with low income in any year are likely to have been poor for longer than white families with similar income in that year. White families are likely to own far more assets that support their children’s achievement than are black families at the same current income level.

Throughout this book, the term “lower class” is used to describe the families of children whose achievement will, on average, be predictably lower than the achievement of middle-class children. American sociologists once were comfortable with this term, but it has fallen out of fashion. Instead we tend to use euphemisms like “disadvantaged” students, “at-risk” students, “inner-city” students, or students of “low-socioeconomic status.” None of these terms, however, capture the central characteristic of lower-class families: a collection of occupational, psychological, personality, health, and economic traits that interact, predicting performance not only in schools but in other institutions as well that, on average, differs from the performance of families from higher social classes.

The critique in this book tries to show that much of the difference between the average performance of black and white children can probably be traced to differences in their social class characteristics. But there are also cultural characteristics that likely contribute a bit to the black-white achievement gap. These cultural characteristics may have identifiable origins in social and economic conditions — for example, black students may value education less than white students because a discriminatory labor market has not historically rewarded black workers for their education — but values can persist independently and outlast the economic circumstances that gave rise to them.

One of the bars to our understanding of the achievement gap is that most Americans, even well-educated ones, are inexpert in discussions of statistical distributions. The achievement gap is a phenomenon of averages, a difference between the average achievement of lower-class children and the average achievement of middle-class children. In human affairs, every average characteristic is a composite of many widely disparate characteristics. For example, we know that lead poisoning has a demonstrable effect on young children’s I.Q. scores. Children with high exposure to lead, from fumes or from ingesting paint or dust, have I.Q. scores, on average, that are several points lower than the I.Q. scores of children who are not so exposed. But this does not mean that every child with lead poisoning has a lower I.Q. Some children with high lead levels in their blood have higher I.Q. scores than typical children with no lead exposure. When researchers say that lead poisoning seems to affect academic performance, they do not mean that every lead-exposed child performs more poorly. But the high performance of a few lead-exposed children does not disprove the conclusion that lead exposure is likely to harm academic achievement.

This reasoning applies to each of the social class characteristics that are discussed in this book as well as to the many others that, for lack of space or the author’s ignorance, are not discussed. In each case, social class differences in social or economic circumstance likely cause differences in the average academic performance of children from different social classes, but, in each case, some children with lower-class characteristics perform better than typical middle-class children.

Good teachers, high expectations, standards, accountability, and inspiration are not enough

As is argued in this book, the influence of social class characteristics is probably so powerful that schools cannot overcome it, no matter how well trained are their teachers and no matter how well designed are their instructional programs and climates. But saying that a social class achievement gap should be expected is not to make a logical statement. The fact that social class differences are associated with, and probably cause, a big gap in academic performance does not mean that, in theory, excellent schools could not offset these differences. Indeed, there are many claims today, made by policy makers and educators, that higher standards, better teachers, more accountability, better discipline, or other effective practices can close the achievement gap.

The most prominent of these claims has been made by a conservative policy institute (the Heritage Foundation), by a liberal advocacy group (the Education Trust), by economists and statisticians who claim to have shown that better teachers do in fact close the gap, by prominent educators, and by social critics. Many (although not all) of the instructional practices promoted by these commentators are well designed, and these practices probably do succeed in delivering better educations to some lower-class children. But a careful examination of each claim that a particular school or practice has closed the race or social class achievement gap shows that the claim is unfounded.

In some cases, the claim fails because it rests on the misinterpretation of test scores; in other cases, the claim fails because the successful schools identified have selective student bodies. Remember that the achievement gap is a phenomenon of averages — it compares the average achievement of lower- and middle-class students. In both social classes, some students perform well above or below the average performance of their social class peers. If schools can select (or attract) a disproportionate share of lower-class students whose performance is above average for their social class, those schools can appear to be quite successful. Many of them are excellent schools and should be commended. But their successes provide no evidence that their instructional approaches would close the achievement gap for students who are average for their social class groups.

The limitations of the current testing regime

Whether efforts to close the social class achievement gap are in-school or socioeconomic reforms, it is difficult to know precisely how much any intervention will narrow the gap. We can’t estimate the effect of various policies partly because we don’t really know how big the achievement gap is overall, or how big it is in particular schools or school systems.

This lack of knowledge about the merits of any particular intervention will be surprising to many readers because so much attention is devoted these days to standardized test scores. It has been widely reported that, on average, if white students typically score at around the 50th percentile of achievement on a standardized math or reading test, black students typically score at around the 23rd percentile. (In more technical statistical terms, black students score, on average, between 0.5 and 1.0 standard deviations below white students.)

But contrary to conventional belief, this may not be a good measure of the gap. Because of the high stakes attached to standardized tests in recent years, schools and teachers are under enormous pressure to raise students’ test scores. The more pressure there has been, the less reliable these scores have become. Partly, the tests themselves don’t really measure the gap in the achievement of high standards we expect from students because high standards (for example, the production of good writing and the development of research skills and analysis) are expensive to test, and public officials are reluctant to spend the money. Instead, schools use inexpensive standardized tests that mostly, though not entirely, assess more basic skills. Gaps that show up on tests of basic skills may be quite different from the gaps that would show up on tests of higher standards of learning. And it is not the case that a hierarchy of skills are gained sequentially by students. Truly narrowing the achievement gap would not require children to learn “the basics” first. Lower-class children cannot produce typical middle-class academic achievement unless they learn basic and more advanced skills simultaneously, with each reinforcing the other. This is, in fact, how middle-class children who come to school ready to learn acquire both basic and advanced skills.

The high stakes recently attached to standardized tests have given teachers incentives to revise the priorities of their instruction, especially for lower-class children, so that they devote greater time to drill on basic skills and less time to other, equally important (but untested) learning areas in which achievement gaps also appear. In a drive to raise test scores in math and reading, the curriculum has moved away not only from more advanced mathematical and literary skills, but from social studies, literature, art, music, physical education, and other important topics where test scores do not result in judgments of school quality. We don’t know how large the race or social class achievement gaps are in these subjects, but there is no reason to believe that gaps in one domain are the same as the gaps in others, or that the relationships between gaps in different domains are consistent at different ages and on different tests. For example, education researchers normally expect that gaps in reading will be greater than gaps in math, probably because social class differences in parental support play a bigger role for reading than for math. Parents typically read to their very young children, and middle-class parents do so more and in more intellectually stimulating ways, but few parents do math problems with their young children. Yet, on at least one test of entering kindergartners, race and social class gaps in math exceed those in reading.

Appreciating the importance of non-cognitive skills

We also don’t know how large are the social class gaps in non-cognitive skills — character traits like perseverance, self-confidence, self-discipline, punctuality, communication skills, social responsibility, and the ability to work with others and resolve conflicts. These are important goals of public education; in some respects, they may be more important than academic outcomes.

Employers, for example, consistently report that workers have more serious shortcomings in these non-cognitive areas than in academic proficiency. Econometric studies show that non-cognitive skills are a stronger predictor of future earnings than are test scores. In public opinion surveys, Americans consistently say they want schools to produce good citizens and socially responsible adults first, and high academic proficiency second. Yet we do a poor job, actually no job at all, in assessing whether schools are generating such non-cognitive outcomes. And so we also do a poor job of assessing whether schools are successfully narrowing the social class gap in these traits, or whether social and economic reform here, too, would be necessary to narrow these gaps.

There is some evidence that the non-cognitive social class gaps should be a cause for concern. For very young children, measures of anti-social behavior mirror the academic test score gaps. Children of lower social classes exhibit more anti-social behavior than children of higher social classes, both in early childhood and in adolescence. It would be reasonable to expect that the same social and economic inequalities that likely produce academic test score gaps produce differences in non-cognitive traits as well.

In some areas, however, it seems that non-cognitive gaps may be smaller than cognitive ones. In particular, analyses of some higher education affirmative action programs find that, when minority students with lower test scores than white students are admitted to colleges, the lower-scoring minority students may exhibit more leadership, devote more serious attention to their studies, and go on to make greater community contributions. This evidence reinforces the importance of measuring such non-cognitive student characteristics, something that few elementary or secondary schools attempt. Until we begin to measure these traits, we will have no insight into how great are the non-cognitive gaps between lower- and middle-class students.

Moving forward

Three tracks should be pursued vigorously and simultaneously if we are to make significant progress in narrowing the achievement gap. First is school improvement efforts that raise the quality of instruction in elementary and secondary schools. Second is expanding the definition of schooling to include crucial out-of-school hours in which families and communities now are the sole influences. This means implementing comprehensive early childhood, after-school, and summer programs. And third are social and economic policies that enable children to attend school more equally ready to learn. These policies include health services for lower-class children and their families, stable housing for working families with children, and the narrowing of growing income inequalities in American society.

Many of the curricular and school organizational reforms promoted by education critics have merit and should be intensified. Repairing and upgrading the scandalously decrepit school facilities that serve some lower-class children, raising salaries to permit the recruitment of more qualified teachers for lower-class children, reducing class sizes for lower-class children (particularly in the early grades), insisting on higher academic standards that emphasize creativity and reasoning as well as basic skills, holding schools accountable for fairly measured performance, having a well-focused and disciplined school climate, doing more to encourage lower-class children to intensify their own ambitions — all of these policies, and others, can play a role in narrowing the achievement gap. These reforms are extensively covered in other books and in public discussions of education and are not dwelt upon in this book. Instead, the focus here is the greater importance of reforming social and economic institutions if we truly want children to emerge from school with equal potential.

Readers should not misinterpret this emphasis as implying that better schools are not important, or that school improvement will not make a contribution to narrowing the achievement gap. Better school practices can probably narrow the gap. School reform, however, is not enough. In seeking to close the achievement gap for low-income and minority students, policy makers focus inordinate attention on the improvement of instruction because they apparently believe that social class differences are immutable and that only schools can improve the destinies of lower-class children.

This is a peculiarly American belief — that schools can be virtually the only instrument of social reform — but it is not based on evidence about the relative effectiveness of economic, social, and educational improvement efforts. While many social class characteristics are impervious to short-term change, many can be easily affected by public policies that narrow the social and economic gaps between lower- and middle-class children. These policies can probably have a more powerful impact on student achievement (and, in some cases, at less cost) than an exclusive focus on school reform, but we cannot say so for sure because social scientists and educators have devoted no effort to studying the relative costs and benefits of non-school and school reforms. For example, some data presented in this book suggest that establishing an optometric clinic in a school to improve the vision of low-income children would probably have a bigger impact on their test scores than spending the same money on instructional improvement. We can’t be certain if this is the case, however, because there have been no experiments to test the relative benefits of these alternative strategies.

Proposals to increase the access of lower-class families to stable housing should also be evaluated for their educational impact, as should proposals to improve all facets of the health of lower-class children, not their vision alone.

Incomes have become more unequally distributed in the United States in the last generation, and this inequality contributes to the academic achievement gap. Proposals for a higher minimum wage or earned income tax credit, designed to offset some of this inequality, should be considered educational policies as well as economic ones, for they would likely result in higher academic performance from children whose families are more secure.

Although conventional opinion is that “failing” schools contribute mightily to the achievement gap, evidence indicates that schools already do a great deal to combat it. Most of the social class difference in average academic potential exists by the time children are three years old. This difference is exacerbated during the years that children spend in school, but during these years the growth in the gap occurs mostly in the after-school hours and during the summertime, when children are not actually in classrooms.

So in addition to school improvement and broader reforms to narrow the social and economic inequalities that produce gaps in student achievement, investments should also be made to expand the definition of schooling to cover those crucial out-of-school hours. Because the gap is already huge at three years of age, the most important focus of this investment should probably be early childhood programs. The quality of these programs is as important as the existence of the programs themselves. To narrow the gap, early childhood care, beginning for infants and toddlers, should be provided by adults who can provide the kind of intellectual environment that is typically experienced by middle-class infants and toddlers. This goal probably requires professional care givers and low child-adult ratios.

Providing after-school and summer experiences to lower-class children that are similar to those middle-class children take for granted would also likely be an essential part of narrowing the achievement gap. But these experiences can’t comprise just after-school or summer remedial programs where lower-class children get added drill in math and reading. Certainly, remedial instruction should be part of an adequate after-school and summer program, but only a part. The advantage that middle-class children gain after school and in the summer likely comes mostly from the self-confidence they acquire and the awareness they develop of the world outside their homes and immediate communities, from organized athletics, dance, drama, museum visits, recreational reading, and other activities that develop their inquisitiveness, creativity, self-discipline, and organizational skills. After-school and summer programs can be expected to have a chance to narrow the achievement gap only by attempting to duplicate such experiences.

Provision of health care services to lower-class children and their families is also needed to narrow the achievement gap. Some health care services are relatively inexpensive, like a school vision clinic. Dental clinics likewise can be provided at costs comparable to what schools typically spend on less-effective reforms. A full array of health services, however, will cost more, but can’t likely be avoided if there is a true intent to raise the achievement of lower-class children. Some of these costs, however, are not new; they can be recouped by school clinics with reimbursements from other underutilized government programs, like Medicaid.

For nearly half a century, the association of social and economic disadvantage with a student achievement gap has been well known to economists, sociologists, and educators. Most, however, have avoided the obvious implication of this understanding — raising the achievement of lower-class children requires amelioration of the social and economic conditions of their lives, not just school reform. Perhaps this small volume can spur a reconsideration of this needlessly neglected opportunity.

About the author

Richard Rothstein is a research associate of the Economic Policy Institute and a visiting lecturer at Teachers College, Columbia University. From 1999 to 2002 he was the national education columnist for The New York Times; he is now a senior correspondent for The American Prospect. Mr. Rothstein’s recent publications include The Way We Were? Myths and Realities of America’s Student Achievement (Century Foundation Press 1998), All Else Equal: Are Public and Private Schools Different? (with Luis Benveniste and Martin Carnoy; RoutledgeFalmer 2003), and Where’s the Money Going? Changes in the Level and Composition of Education Spending (EPI 1995, 1997).

This research was funded by The William and Flora Hewlett Foundation. We thank them for their support. The William and Flora Hewlett Foundation has been making grants since 1966 to help solve social and environmental problems at home and around the world. The Foundation concentrates its resources on activities in education, environment, performing arts, population, conflict resolution, and U.S.-Latin American relations. In addition, the Hewlett Foundation has initiatives supporting neighborhood improvement, philanthropy, and global affairs.