Educational Goods: Values, Evidence, and Decision-Making 9780226514208

Citation preview

Educational Goods

Educational Goods Values, Evidence, and Decision Making

h a r r y b r i g h o u s e , h e l e n f. l a d d , s u s a n na l oe b, a da m s w i f t the uni versit y of chicago press

chicago and london

The University of Chicago Press, Chicago 60637 The University of Chicago Press, Ltd., London © 2018 by The University of Chicago All rights reserved. No part of this book may be used or reproduced in any manner whatsoever without written permission, except in the case of brief quotations in critical articles and reviews. For more information, contact the University of Chicago Press, 1427 E. 60th St., Chicago, IL 60637. Published 2018 Printed in the United States of America 27 26 25 24 23 22 21 20 19 18

2 3 4 5

isbn-13: 978- 0-226-51403-1 (cloth) isbn-13: 978- 0-226-51417- 8 (paper) isbn-13: 978- 0-226-51420- 8 (e-book) doi: 10.7208/chicago/9780226514208.001.0001 Library of Congress Cataloging-in-Publication Data Names: Brighouse, Harry, author. | Ladd, Helen F., author. | Loeb, Susanna, author. | Swift, Adam, 1961– author. Title: Educational goods : values, evidence, and decision making / Harry Brighouse, Helen F. Ladd, Susanna Loeb, Adam Swift. Description: Chicago ; London : The University of Chicago Press, 2018. | Includes bibliographical references and index. Identifiers: lccn 2017009046 | isbn 9780226514031 (cloth : alk. paper) | isbn  9780226514178 (pbk. : alk. paper) | isbn 9780226514208 (e-book) Subjects: lcsh: Education—Decision making. | Education—Aims and objectives. | Education and state—Decision making. | Education—Philosophy. | Values. Classification: lcc lb2806 .b734 2018 | ddc 371.2—dc34 lc record available at https://lccn.loc.gov/2017009046 This paper meets the requirements of ansi/niso z39.48-1992 (Permanence of Paper).

Contents List of Figures vii Preface and Acknowledgments ix Introduction 1 PA RT

1

Values

chapter 1.

Educational Goods

chapter 2.

Distributive Values and Independent Values 30

chapter 3.

Achievement as an Educational Good 44

PA RT

2

19

Decision making

chapter 4.

Combining Values and Evidence 73

chapter 5.

School Finance 85

chapter 6.

School Accountability

chapter 7.

School Autonomy and Parental Choice

Conclusion

155

Notes

163

References Index

187

173

108 132

Figures 1.1 1.2 2.1 3.1 3.2 3.3 3.4 3.5 3.6 3.7 5.1

5.2

Educational goods and flourishing / 21 Educational goods and capacities / 23 Independent values / 37 Distribution of student-level achievement as measured by test scores / 49 Comparison of groups with different levels of average achievement / 50 Comparison of groups with different amounts of variation / 50 Depiction of two possible thresholds of adequacy / 51 Achievement gap between two groups / 57 Trends in average reading scores for thirteen-year- old black and white students / 60 Program of International Student Assessment (PISA) reading scores (2009) / 64 Spending on elementary and secondary education across select Organization of Economic Co- operation and Development (OECD) countries (2012) / 86 Current per-pupil expenditures for public elementary and secondary education in the United States (fiscal year 2013) / 87

Preface and Acknowledgments

W

e fi rst met at a series of workshops funded by the Spencer Foundation and held at the Institute of Advanced Study, Princeton, in 2009– 2010. The workshops were motivated by a concern that we already shared: educational social scientists often proceed with their empirical work without sufficient attention to normative considerations while philosophers of education often conduct their normative work without sufficient attention to empirical reality. The result is that the former sometimes fail to focus on interesting, decision-relevant, empirical questions while the latter sometimes miss interesting, decision-relevant, normative questions. The workshops brought together researchers and scholars from history, English, philosophy, sociology, psychology, economics, public policy, and political science with the aim of fostering an interdisciplinary dialogue and understanding. The intention was to produce a new kind of research and to enhance the ability of the participants to work more deeply across disciplinary boundaries in the future. Some of the results can be found in Education, Justice, and Democracy (University of Chicago Press, 2013), edited by Danielle Allen and Rob Reich. The four of us chose to continue working together after the workshops were fi nished. Our ambition was to produce something informed by serious engagement across the disciplinary boundaries of philosophy and the educational social sciences in a way that would be useful for social scientists, philosophers, and reflective decision makers. This ambition exposed us to the logistical difficulties inherent in a project involving four collaborators located in four disciplinary homes (education, public policy, politics, and philosophy), in four time zones, on two continents, and all with numerous other commitments. We held several prolonged face-to-face meetings in which we discussed preliminary ideas,

x

Preface and Acknowledgments

and after writing commenced, we discussed drafts and then revisions of drafts during numerous Skype and phone meetings involving various combinations—and, usually all—of us. While we can still identify the initiating authors of some parts, every chapter and every passage (if not quite every sentence) reflects the contributions of us all. Many others, however, have also contributed, and we can acknowledge some of them here. We are grateful to Danielle Allen and Rob Reich in particular for the matchmaking that brought us together and, eventually, produced this book, continuing the interdisciplinary enterprise they nurtured. Thanks also to the other participants: Sigal Ben-Porath, Angel Harris, Andrew Jewett, Tony Laden, Patrick McGuinn, Seth Moglen, Melissa Roderick, Richard Rothstein, Anna Marie Smith, Carola SuárezOrozco, Marcelo M. Suárez- Orozco, and Gregory M. Walton. We are also grateful to the Spencer Foundation both for supporting the original workshop that brought us together and for funding the meetings during which we worked on this project. Harry Brighouse is also grateful to the Carnegie Corporation of New York for a scholarship that allowed him, some years ago, to start thinking about educational reform. And, of course, we thank our home institutions—Duke University, Stanford University, the University of Warwick, and the University of Wisconsin– Madison—for the supportive environments they continue to give us and the license they grant faculty to pursue knowledge. We owe much to audiences at the American Educational Research Association (AERA), the Association of Public Policy and Management (APPAM), Trent University, the University of Amsterdam Center on Inequality Studies (AMCIS), Northwestern University, the Sanford School of Public Policy at Duke University, and the School of Government at the Victoria University of Wellington. We are grateful also to Robert Bifulco, Jonathan Boston, Tim Brighouse, Matthew Clayton, Randall Curren, Jennifer Jennings, Michael McPherson, Kate Norlock, Karen Robertson, and Ken Shores for their individual comments. We are grateful to Stephanie Sheintul for compiling the index. We also thank our editor, Elizabeth Branch Dyson, who has been patient and encouraging throughout, and three readers for the University of Chicago Press, whose comments on our fi rst draft, together with Elizabeth’s, did much to improve the fi nal product. The conceptual framework for the book has previously been published as “Educational Goods and Values: A Framework for DecisionMakers,” Theory and Research in Education 14, no. 1 (2016): 3– 26.

Introduction

S

ocieties have to make decisions about what children should learn, how to teach them, how to divide teaching between family and school, and how many resources to allocate to the education of different children. Education is the process that transforms newborn infants into adults with characters and temperaments, with skills and abilities, with beliefs and attitudes. Decisions about education influence the character of the whole society. They shape the way people become individuals, their capacities to live flourishing lives, and how those capacities are distributed across the population. Policy makers at top levels of government, teachers in individual classrooms, and parents all know that their decisions matter. Because these decisions affect such fundamental concerns—how well people’s lives go and whose lives go better than others—they are morally significant. Aware of that significance, most education decision makers want to make morally responsible decisions. But what is a morally responsible decision? However keen they may be to do the right thing, decision makers are often not well equipped with the conceptual resources they need. There is a growing body of academic work on how decision makers in fact gather and interpret research evidence and some on how they should do so.1 But very little is available on how decision makers, particularly those in education, should combine their interpretations of the evidence with considerations about values when deciding what, actually, to do. In this book we offer a clear and manageable framework for combining value considerations with empirical data to make judgments about how well specific policies are likely to realize valued outcomes. We do not recommend particular policies. What we offer instead is a method for assessing and evaluating policy options. This method draws on our joint expertise as philosophers and social scientists. Philosophy supplies resources for thinking better about

2

Introduction

the goals or values that decisions are, or should be, trying to achieve, while social science research yields information about how likely the various options are to achieve them. 2 We have three main aims. The fi rst is to broaden the language for talking about the goals of education policy. To that end, we have coined the terms educational goods and childhood goods. As we explain fully in chapter 1, we use the term educational goods to refer to the knowledge, skills, attitudes, and dispositions that children develop both for their own benefit and for the benefit of others. These goods are varied, and include much more than the cognitive skills that are often associated with student achievement. We offer a way of thinking about the educational goods that educators should seek to promote by describing capacities of individuals that lead to flourishing. These are capacities such as the capacity to function in the labor market, to be a good democratic citizen, to develop healthy personal relationships, and to treat others with respect and dignity. The word goods signifies that they are valued in the sense that they help individuals to flourish and to contribute to the flourishing of others either in the present or in the future. The adjective refers to the fact that these goods emerge from an educational process, one that is broadly conceived to include the contribution not only of schools but also of families and communities. We supplement the concept of educational goods with that of childhood goods, which are the good experiences enjoyed by children during their childhood years. Childhood is a distinct phase of life with its own special joys that are no longer available, or much harder to achieve, as adults. Identifying the valued outcomes of the educational process and acknowledging the value of childhood experiences is the fi rst goal of the book. The second aim is to add precision and clarity to the discussion of the distributive values that are essential for good decision making about education. A decision maker will need to consider not only the total amount of educational goods but also how those goods are distributed across individuals or groups of individuals. Clear thinking about distributive considerations helps one avoid the commonly used but vague and imprecise terms social justice and equity, which often mean different things to different people. To this end, we spell out three distributive values: equality, adequacy, and benefiting the less advantaged. Some policies may promote all three. Others may promote one or two while harm-

Introduction

3

ing the third. Armed with these three values, the decision maker is in a position to make conscious trade- offs among them. Our third aim is to provide a framework for individual decision makers that enables them to combine values and evidence in the evaluation of policy options. For reasons we explain in chapter 4, we focus on decisions related to schools. To be sure, the decision makers we have in mind—whether they be governors, legislators, school board members, school superintendents, school principals, teachers, voters, or parents— are unlikely to be in a position to make such decisions on their own. Instead, decisions will be arrived at through a collective decision-making process. That may be a public political procedure involving elected officials, or it may simply be the process by which teachers or administrators within an individual school devise school policy. This book does not address the collective aspects of decision-making processes. Instead, our goal is to help those who contribute to the collective process think in an organized way about the values at stake in any decision and what the evidence has to say about the potential for realizing those values so as to make the trade- offs inherent in policy choices clearer. Although the main audience for this book is decision makers in the field of education policy, the framework also has implications for philosophers and social scientists who want to contribute to the policy-making process. Our experience is that philosophers tend to think at a general level about what is important but often do not devote much attention to what is, or could be, measured or attend sufficiently to complicated issues of institutional and policy design. Meanwhile, social science research is not always well oriented to providing the information needed given the values that are actually at stake. Both groups can benefit from the approach we present here.

The Role of Values and Evidence The evaluation of policy involves interactions between values and evidence. One can assess policies simply in terms of how likely they are to achieve, or how well they have in fact achieved, their stated goals. We can investigate their effects, or their likely effects, both intended and unintended, in a value-neutral, technical way. But evaluating policies involves more. It requires judging whether policies are good or bad.

4

Introduction

By values we mean whatever is good or right about an action or state of affairs, whatever it is that makes them valuable. If, for example, you think it would be good for the achievement gap between black and white students to be smaller than it is at present, you must think something makes a society with smaller achievement gaps better than one with bigger gaps. Perhaps more than one thing makes it better. Maybe it would give its children more equal opportunities and be more productive and its members would be less likely to experience racial prejudice. Often there is more than one value that accounts for why some states of affairs are better than others. The important point is that at least one value judgment must be involved. To claim that A is better than B is to claim that A is more valuable than B, which necessarily involves judgments about values. Because there are many values, they often confl ict, and one needs to make trade- offs among them. Perhaps the best-known example is the clash between liberty and equality. In the educational context, that arises mainly as a tension between parents’ freedom to devote their resources to their children’s education and the concern that children born into different families should enjoy something approaching equal opportunities in life. Other confl icts may be less dramatic, but judgments about how to balance the competing claims of different values are implicit in every decision to allocate scarce resources. Some readers may well think that smaller achievement gaps between racial groups, better provision for those with learning disabilities, and the recruitment of higher- quality teachers would all be good. Each of them would help to realize one or more values. But because in practice they compete with one another for resources, any judgment about how to allocate those resources requires one to make trade- offs between the different values. Thorough evaluation of policy proposals depends not only on weighing the different values in play but also on assessing the empirical evidence about how each policy option would affect the values. The careful analysis of values requires philosophical reflection, but decision makers cannot make good policy without equally careful analysis of how society actually works, which is the job of the social scientist. Sometimes this distinction is formulated in terms of a contrast between the “normative” and the “positive.” The normative is concerned with questions of value: what is valuable or desirable. The positive is concerned with the description, explanation and prediction of what is and what could be. Policy decisions inevitably involve both.

Introduction

5

Detailed and thorough investigation of relevant evidence, with all its complications, is an essential prerequisite for good policy decisions. Social scientists are trained to understand institutional frameworks, to use appropriate quantitative and qualitative methodologies to answer welldefi ned questions, and to construct explanatory theories based on the careful accumulation of empirical evidence. They have learned about what inferences are and are not warranted on the basis of what data, and they know when research fi ndings corroborate or falsify particular claims. Given the nature of this training, social scientists may well be skeptical about the possibility of making judgments about values in ways that provide the kind of objectivity they associate with their own research fi ndings. But value judgments are inescapable. Anybody who thinks one thing is better than something else is implicitly relying on some judgments about what makes things better or worse. Those are value judgments. Social scientists’ choices about what to study—academic achievement, civic participation, occupational choice—typically reflect their views about what is valuable. Inequalities can be bigger or smaller, test scores can go up or down, and different school regimes can differently accommodate parents’ views about how their children should be educated. Without some standard of evaluation, nobody can judge any of those alternatives to be better or worse than the other, and nobody can view research fi ndings as yielding even the most tentative implication for policy. Those judgments and views about policy will be better if they are based on criteria or standards that have been subject to careful analysis. This book is based on the assumption that individuals can make better policy choices if they are systematic in their consideration both of the relevant empirical evidence and of the relevant moral values. That may sound daunting or unmanageable. We aim to show how it can be done. Sometimes it may be wise for a decision maker deliberately to conceal the complicated trade- offs between values that are always implicit in policy judgments. Rarely are there only winners and no losers from any proposed reform, however desirable that reform may be given all the relevant considerations. So those keen to “sell” a reform may well fi nd it strategic to talk only about the good bits or to come up with feel-good rhetoric that highlights the upside while concealing the downside. We understand that in some contexts precise and explicit public acknowledgment of trade- offs is not always wise. That said, reasons for saying vague, imprecise, or confused things are

6

Introduction

not reasons for holding vague, loose or confused beliefs. When it comes to thinking—including thinking about decisions—clarity, precision, and truth have to be better than the alternatives. Even where it is strategic not to be too precise in the positions offered to a particular audience, one can still be clear about one’s own beliefs, about what values one expects such a strategy to realize, and why one endorses those values. The pragmatic and strategic pursuit of carefully thought-through policy aims is one thing. Flailing around without a moral compass, or with an inadequate sense of the complex terrain one is negotiating, is another.

Decisions about Education Policy Education policy choices are often posed in terms of their effects on student achievement. One advantage of that focus is that achievement can be measured, albeit imperfectly, by test scores. Measures allow social science researchers to use quantitative empirical methods to examine the effects of programs on achievement—often taking into consideration the details of how the programs were implemented. The findings that emerge from such studies may be useful to policy makers, but they are not conclusive. Even if policy makers are interested only in how a policy affects achievement and the research evidence is clear and persuasive, they may need to weigh predicted changes in one dimension, such as higher overall achievement, against changes in another, such as a widening achievement gap. Thus, they need to compare the value of higher levels of achievement with the disvalue of larger gaps. They may also need to clarify what it is about the distribution of achievement that matters. Should they care primarily about gaps between groups or about the low achievement of the lower-performing group? Moreover, student achievement is rarely the only consideration at stake. What about other outcomes, such as how students view each other? It matters whether children learn to regard one another as equals rather than as superiors or inferiors. What about the experience of childhood itself? Some worry about sacrificing people’s childhoods on the altar of improving their test scores. How should one incorporate such concerns into the decision-making process? What language can one use to talk about these other considerations? How would the inclusion of these other values in policy discussions affect the research endeavor? Values that are not in themselves educational can also be important

Introduction

7

factors in educational decisions. For example, parents often have views about their children’s schooling—which may include views about the importance of academic achievement. Respect for parents’ preferences about their children’s education suggests that those preferences should play some role in key decisions. But how can the policy maker coherently show that respect when different parents have different views, and how can she do that when there is even disagreement about the extent to which parents should be able to influence their children’s schooling? In practice, decision makers may not know what to do, or may make a decision that is not the best one given what they care about. That could be because they do not know how to think beyond measurable achievement in considering outcomes or because the research fi ndings are incomplete, inconclusive, or not directed at the goals they are most concerned with. It could also be because they are not clear in their own minds about what outcomes they value, what distributive principles should guide their decisions, and how to make trade- offs among them. In this book we aim to help them make better decisions. The book is divided into two parts. In part 1, we set out the values that we believe are most salient to educational decision makers and discuss how test scores (and gaps in test scores) might relate to educational goods (and gaps in educational goods) understood in the much richer way we propose. In part 2, we propose a method for combining values and evidence and explore in some detail the application of that method to three major arenas of educational policy. Here we briefly consider a few introductory examples to illustrate some elements of our approach. The fi rst and shortest example illustrates trade- offs among educational goods. The second highlights the importance of careful attention to specific distributive values and the role of evidence in determining the trade- offs among them. The third and most complicated example involves all three components: trade- offs among educational goods, distributive considerations, and the use of evidence. Pressure on Schools to Raise Student Test Scores In chapter 6 we describe how policies to hold schools accountable, especially in the United States, focus attention on student achievement, typically as measured by test scores in math and reading. Here we focus on the limitations of current language for evaluating that approach. Of course most people understand that schools have many goals. But

8

Introduction

how can decision makers engage in discussions about which of the many things that schools could be doing are important and why? The term educational goods can help them by explicitly linking the outcomes of the education system to outcomes that are valuable both for those being educated and for others. Some people may be concerned that the pressure to raise student achievement in math and reading that comes with accountability will induce schools to reallocate scarce resources in favor of those subjects and away from others, such as social studies, art, or physical education, with adverse consequences for student learning in the shortchanged subjects. Whether or not the policy has this effect is an empirical question that, in principle, could be answered by social scientists. Even more important, however, is for policy makers to consider how and why they should be concerned about less attention to these other subjects. Most people are likely to regard student achievement in core subjects as important, both for future learning and for a child’s subsequent success in the labor market. But it is not as clear how one should think about the relative importance of student learning in other subjects for which there is less evidence and less consideration of their implications for longer-run outcomes. Without a concept such as educational goods, it may be tempting to focus most of the school’s attention on the clearly beneficial—and measurable (albeit imperfectly)—achievement of students in math and reading. A downside of this narrow focus is that it forces the supporters of other subjects to try to justify attention to them on the ground that they contribute to students’ measured achievement in one or both of the core subjects. For example, some might argue that social studies improves reading skills by introducing students to new and interesting materials to read. Art teachers might try to make the case for art on the ground that it encourages student to be spatially analytical, which could potentially spill over to the skills needed for math. Arguments of this type are not likely to be very successful. Additional resources devoted directly to math or reading will inevitably be more productive in improving knowledge of those subjects than additional resources devoted to the other subjects, except in some unusual cases. More importantly, the approach ignores how the other subjects might be valuable for students in ways that have no bearing on measured achievement in math or reading. In contrast, the concept of educational goods incorporates a broader range of valuable outcomes of the education process. As we spell out in

Introduction

9

chapter 1, these goods include the knowledge, skills, attitudes, and dispositions that enable individuals not only to participate productively in the labor market but also to be citizens actively involved in the democratic process. In addition, they include the capacity to be autonomous, to develop strong personal relationships, to achieve personal fulfi llment, and to treat others with dignity and respect. Decision makers will have different views about the relative importance of these capacities. Some will think that some of the educational goods are more appropriately developed by families than by schools. Nonetheless, decisions about policies that affect schools will undoubtedly be better if decision makers pay attention to a broad range of educational goods and then make explicit, evidence-informed trade- offs among them. In sum, the concept of educational goods draws attention to a broad range of educational outcomes, challenging education decision makers to prioritize among them and use them to establish which evidence is relevant to the choices they have to make. Boston Public Schools’ Home-Based Choice Plan To introduce distributive trade- offs, we turn now to a real case. The Boston Public Schools district allocates children to schools using a parental choice program designed to give parents a range of options close to their homes. All families with children entering kindergarten are given a “basket” of ten to eighteen schools to which they can apply with the final allocation determined by a district-wide lottery. Although the baskets must contain schools meeting various minimal criteria of measured quality, the system is designed so that families in more affluent neighborhoods within the city have more high- quality options available to them than families in less affluent neighborhoods. In this way, the system privileges already advantaged families. 3 Why might the choice system be set up this way? One explanation is that district policy makers may be concerned that parents will move out of, or never enter, the district, or that they will move their child to a private school, if they fear that their children will be assigned to a bad public school. Moreover, they may understand that such responses are easier for middle- or high-income families than for low-income families. Catering to the interests of families with resources is in fact a common practice within urban districts, which often make decisions about school catchment area boundaries in ways designed not to alienate more

10

Introduction

affluent families. A district with two schools in two contiguous neighborhoods, one with an 80 percent poverty rate, the other with a 20 percent poverty rate, could, if it chose, cross assign the students so that 50 percent of the children in both schools were from poor households, but many districts in that situation will not pursue that strategy. To evaluate the Boston plan, the policy maker needs to identify the values at stake. One such value is equality. For reasons we spell out in chapter 2, equality needs careful interpretation. For now, the reader can understand it as requiring that each child, regardless of her background, has an equal chance of attending a good school. Although the Boston policy clearly violates the value of equality, it might serve a different distributive value, namely benefiting the less advantaged. Specifically, Boston policy makers may be concerned that if the share of middle- class families within the district declines, the consequences would be detrimental to the low-income students remaining in the system. The loss of middle- class families might lower property values (an important determinant of overall funding for schools) and deprive the district of the political capital that more affluent parents can invest in both supporting higher tax rates and lobbying state officials for additional funds for the city. It also deprives other children of the beneficial peer effects of being in a more socioeconomically diverse district. If more affluent families can be enticed to keep their children in district schools, and if their continued presence enhances the educational experiences of other students, policy makers have a reason to cater to those parents. That reason is grounded not in a concern for equality but in a commitment to the value of benefiting the less advantaged. This example illustrates how being explicit about values helps to specify what social scientific evidence is really relevant. Decision makers guided by a concern for benefiting the less advantaged in this context would need evidence about two matters. First, how will the decision to cater to the interests of more affluent families affect the proportion of those families entering and remaining in the district? Second, will the continued presence of those families in fact bring educational benefits to the remaining lower-income children? We have suggested several mechanisms by which their ongoing presence might have that effect: peer effects in schools, the maintenance of property values (and thus district revenues), and the political capital the district has thanks to middle- class parents lobbying to increase resources and improve educational provision. These effects may not be meaningful in magnitude. They may

Introduction

11

not even be positive. The effect on property values may be small, and middle- class parents may use their lobbying power in ways that simply transfer existing resources to their children (e.g., by lobbying to have their own children put in the classes of the better teachers at the expense of other children). Being explicit about what values matter helps decision makers understand what is at stake and helps researchers orient themselves to provide the relevant evidence. Proficiency Grouping of Students Finally, we consider a perennial debate about the allocation of children to classrooms within schools. Should they be grouped by “ability,” or should they be taught in “mixed-ability” classrooms? Grouping by ability or proficiency is often referred to as ability grouping or tracking in the United States and as streaming in other countries. We shall use the term proficiency grouping, which, we think, most accurately denotes what its advocates really support. Such grouping might apply to individual classrooms, to sequences of courses through school, or to whole schools. Discussions of proficiency grouping typically start with its impact on student achievement. Advocates claim that when students are grouped by ability, teachers will be in a better position to raise the achievement of all groups than they will when classrooms are more heterogeneous. In heterogeneous classes, teachers may teach to the middle-achieving students as a way of reaching the most students and in the process fail to meet the needs of both low and high performers. Further, either because of their own inclinations and abilities or because of pressure from parents or policy makers, teachers may devote disproportionate attention to high-ability students, doing an even greater disservice to the lowachieving students. In classes grouped by proficiency, by contrast, teachers will be in a better position to raise achievement across the board because they can gear their lessons to the skills of each of the more homogeneous groups. If so, proficiency grouping may both raise overall achievement and generate the greatest benefit to the low-achieving students, who are especially vulnerable in the heterogeneous classroom. Opponents of proficiency grouping typically challenge the contention that low-achieving students will benefit. A common counterargument is that schools are likely to consign low-achieving students to classrooms with the least skilled teachers, producing detrimental effects on achievement that will be further compounded by negative peer effects.

12

Introduction

As is common, the existing empirical studies do not all come to the same conclusion, and some are more rigorous and focused than others. The effects of proficiency grouping are likely to depend on the context. On balance, the weight of the evidence drawn from studies of British and US schools suggests that, when compared with an assignment plan with no proficiency grouping and no special interventions, proficiency grouping does not raise average achievement by much, if at all, but it does increase gaps between high-achieving and low-achieving students.4 Typically, this widening of gaps is the result both of increased achievement of high-ability groups and of reduced achievement of low-ability groups. Recent cross national research—both descriptive studies and more careful causal studies examining the relationship between the differentiation of a country’s school system and inequality in student achievement 5— also fi nds that proficiency grouping exacerbates inequalities with little or no impact on overall average achievement. Studies of a policy aimed at reducing ability grouping in math by having almost all eighth-grade students take algebra in mixed-ability classrooms have found somewhat different effects depending on the state. In North Carolina, for example, the effects were negative for the lower-achieving students who advanced into algebra in eighth grade but slightly positive for the higher-achieving students who would have taken algebra even without the change.6 In California, the effects were negative for both groups.7 So while on average proficiency grouping may not improve student achievement, especially in elementary school, decision makers will need to interpret the general evidence in light of their own contexts. The decision maker, furthermore, may well be concerned not only about averages and gaps in student achievement but also about other educational goods and indeed about other considerations that go beyond the educational success of the students. When considering the allocation of students to classrooms, for example, one might be concerned about the effects on their development of the capacity for students to treat other students with dignity and respect. Opponents of proficiency grouping sometimes claim that lower-achieving students will be stigmatized by the decision to group them together. Both they and the other students will know they have been designated as “dumb,” which may adversely affect their inclination to exert effort and may undermine all students’ capacity to respect one another as equals. Whether proficiency grouping has these effects is an empirical question on which there is not much evidence, but we shall explain in chapter 1 why this is a well-motivated con-

Introduction

13

sideration: why the capacity to treat others as equals is an educational good. Another relevant issue concerns children’s interest in developing the dispositions and skills needed for democratic competence. Opponents of proficiency grouping sometimes worry that separating groups by proficiency will make it difficult for members of both groups to develop those attributes. This concern may be particularly significant if ability groups correspond closely to other categories, such as students’ race or socioeconomic background. Where they are raised in communities segregated by race and socioeconomic class, students may need to mix across demographic lines in school and in the classroom in order to learn about one another’s practices, beliefs, and communities so that they can take each other’s interests appropriately into account when contributing to the democratic process later in their lives. Finally, children spend a large amount of their waking life in school. Even if proficiency grouping had no effect on their achievement or on the development of other competences, simply being in an environment that values achievement and in which one feels, or knows, one has been labeled “dumb,” either by teachers or peers, may have damaging effects on one’s daily lived experience. That effect would be a cost to what we label, in chapter 2, “childhood goods.” Because the social science research has not paid much attention to values other than achievement, good measures of democratic competence or of the tendency to treat others as equals are not available. Research on what causes and impedes their development is very limited. Although measures of school climate have been developed, they are typically not closely related to children’s daily lived experience. Without better measures and sustained social scientific attention, decision makers have to judge as best they can in light of whatever evidence they can gather. Perhaps proficiency grouping reduces the childhood goods available to members of the lower group by affecting their self- esteem and making their school days a miserable experience. Or perhaps it allows both groups to enjoy more childhood goods: the lower-proficiency children may enjoy learning more if they are not in classrooms where they have to compete (typically unsuccessfully) with higher-proficiency children, and the latter may enjoy schooling more if their progress is not slowed down by the presence of less able classmates. Overall, the best solution when it comes to assigning students to schools, tracks, classrooms, or groups within classrooms will depend on several factors. Different policy makers will place different weight

14

Introduction

on the various values, and they may also interpret the evidence in different ways. Those who agree about how to trade off the various values at stake may disagree about which policies are most likely to achieve their preferred balance of values. Those who agree on the likely effects of different policies may disagree about what policy should be trying to achieve. And in particular contexts, the available resources will narrow the range of feasible options, thus affecting what the best choice is. Still, a few clear lessons and trade- offs emerge from this example. One is the importance of paying attention to educational goods that go beyond the narrow good of student achievement. Just because these other educational goods are not readily measured does not mean they can be ignored. Indeed, the very fact that they are often so hard to measure—whereas student achievement is easily quantified by test scores—makes it especially important to ensure that they receive their proper attention in policy discussions. The weight one attaches to these values is likely to depend in part on how closely ability groups correspond to other categories, such as students’ race or socioeconomic background. In chapter 3, we help the reader identify which groups are particularly salient for the purposes of education policy making. Another lesson is the importance of being clear about one’s distributive values. Perhaps, in some circumstances, grouping by ability improves the level of achievement of those at the bottom but also increases the gap between them and those at the top of the distribution. How much weight should be placed on the value of promoting equality if that has to be traded off against the value of helping the worse off? Or under what conditions might one support policies that raise the achievement levels of more able students but do nothing to improve the achievement of the less able?

The Structure of This Book In part 1 we explain more fully the values at stake in educational decision making. In chapters 1 and 2 we expand on what we mean by educational goods, provide a detailed exposition of our various distributive principles, and clarify additional values, not in themselves educational, that policy makers must also take into account in many education decisions. In chapter 3, the discussion becomes more concrete as we focus on student achievement and consider the conditions under which differen-

Introduction

15

tial achievement levels of different groups are likely to be relevant to the values we have identified. In part 2 we take the reader into decision making itself. In chapter 4 we outline a method for combining values and evidence, and in the following three chapters we illustrate its application to specific policy areas. In chapter 5 we look at school fi nance, in chapter 6 at measures for holding schools accountable, and in chapter 7 at policies that seek to increase both school autonomy and parental choice. The book ends with a short concluding chapter.

Part 1 Values

chapter one

Educational Goods

M

any values are relevant to decisions about education. We divide them into three groups: educational goods, distributive values, and independent values. We discuss educational goods in this chapter and distributive values, independent values, and the relationship between the three in chapter 2.1 Educators have aims—they want to instill knowledge, skills, dispositions, and attitudes in the children they teach. These attributes are what we call educational goods. They inhere in the people who have been educated, contributing to their own prospects for flourishing and to their ability to contribute to the flourishing of others. Many aspects of children’s upbringing create educational goods. How parents talk to, discipline, and socialize their children are as relevant to the development of educational goods as are experiences in day care, school, and other formal settings outside the family. In addition, the knowledge, skills, attitudes, and dispositions that children develop are influenced by peers and by their interactions with cultural content through the media and the internet. The process of developing educational goods begins before children enter formal schooling and continues long after they leave it. Most people continue to acquire knowledge and skills, and their attitudes and dispositions evolve throughout their lives. But we focus in this chapter, and in this book, on the processes producing educational goods before adulthood because that is when educational goods are produced most rapidly and because deficits in childhood are difficult to eliminate in adulthood. Moreover, public policies, primarily in the form of schooling, have great leverage on the production of educational goods at this stage in people’s lives. Most industrialized societies have taken responsibility for

20

Chapter 1

the development of educational goods in children, creating large-scale, heavily resourced institutions—namely, schools—for that purpose. We further focus on decision making linked directly to schooling even though children also develop educational goods at home, at the playground, and in early childhood educational settings. Similarly, health policies, tax policies, and housing policies can all affect children’s educational development. Although the division of policy sectors is artificial, decision makers are bound to focus on the values that are most readily realized by the levers at their disposal. Schools are the natural focus because they are designed specifically to produce educational goods in children. We characterize educational goods as knowledge, skills, attitudes, and dispositions. Knowledge, in this context, and to simplify greatly, involves both understanding and warranted true belief, such as knowing the names of the US presidents, knowing the branches of the US government, understanding how an automobile engine works or how a law is passed, or knowing that the square of a right triangle’s hypotenuse is equal to the sum of the squares of its other two sides. Skills involve being able to do things, from analyzing data and identifying errors in reasoning to planning and cooking a meal or negotiating a compromise. Attitudes are conscious bases for motivation that normally (though not always) result in action when triggered by external stimuli; one might have an attitude of respect for people who are manifestly kind. Dispositions are the inclinations, often unconscious and sometimes even irrational, to deploy whatever skills and knowledge one has in particular situations. Courage, for example, is a disposition to act in particular ways when confronted with danger. Dispositions and attitudes are usually congruent, but they need not be: somebody might consciously believe that they should exercise regularly and yet, contrary to their attitude, when faced with the choice of stairs or an elevator still choose the elevator. As we shall make clear in the next section, educators should usually be aiming to instill both dispositions and the corresponding attitudes. 2 The use of the term goods in the phrase “educational goods” may suggest concrete or material commodities, but we are using it just in the sense in which they are opposite to bads. They are positive in that they contribute to valuable outcomes for the individual possessing them or for others in either the present or the future. Cognitive skills and socio-

Educational Goods

21

emotional capacities are educational goods because they generate value in the current period for those who are being educated and contribute to their future income and health and hence to their overall well-being. They also benefit others, whose lives go better through the actions of those being educated. Attitudes and dispositions that enable and incline individuals to participate responsibly in the democratic process may sometimes benefit the individuals themselves and may at other times benefit only other members of their polity. They are educational goods in both cases. The fundamental value that underlies our discussion of educational goods is human flourishing. Educational goods help people’s lives go well—and what matters, ultimately, is the creation and distribution of opportunities for people to flourish. We focus on opportunities for flourishing rather than flourishing itself because the most that educational goods can do is equip people with what they need for their lives to go well, including the capacity to make good choices. Whether people do in fact choose well is a further question. Luck—serious injury or illness, for example—is also bound to play a role in determining the extent of people’s flourishing however well equipped they are and however well they choose. Figure 1.1 describes this relationship. As a guiding principle, “produce opportunities for human flourishing” is not, by itself, particularly helpful. It does not describe flourishing in enough detail to identify human qualities that are likely to enhance it. In order to act on the principle, one needs to know what constitutes human well-being, because otherwise one cannot have a sense of what knowledge, skills, dispositions, and attitudes to aim for. Although there

Figure 1.1. Educational goods and flourishing.

22

Chapter 1

is widespread agreement on some elements of flourishing, there is no consensus theory. 3 Rather than attempting to defend a particular view, we rely on reasonably (though not entirely) uncontroversial assumptions about some of the constituents and prerequisites of a flourishing life. Disagreement will persist, but our approach is useful even for those who disagree with our choice of the particular constituents of flourishing. It lays out a method for moving from theories of flourishing to determining which educational goods to pursue and, ultimately, which education policies and practices to endorse. The method could be applied to other conceptualizations of flourishing. The value of any given set of knowledge and skills depends on context. In the United States today, literacy is more or less essential for generating income in the labor market, but it was much less important in the eighteenth century. Physical strength and coordination are less valuable today than they were then, and technological change has reduced their value even since the 1970s. Some capacities, of course, such as the capacity to defer gratification or the cognitive capacities that psychologists call “executive function” (planning for the future, attention, working memory, connecting past experiences to present situations), may be essential for some reasonably high level of well-being regardless of context. So decision makers have to supplement the directive “promote flourishing” with a set of intermediate educational aims, which are the specific educational goods—knowledge, skills, dispositions, and attitudes—they should be trying to create in their particular context. Constructing a comprehensive list of the specific knowledge, skills, dispositions, and attitudes needed to enable people to flourish, and to contribute to the flourishing of others, would be an unmanageable task. The list would be extremely long, and the precise items on it would vary across contexts. But we can identify, at a general level, six capacities that everyone should have in modern societies and which, when deployed effectively in appropriate circumstances, will tend to support the flourishing of both the educated person and others: the capacities for economic productivity, personal autonomy, democratic competence, healthy personal relationships, treating others as equals, and personal fulfi llment. These capacities—and the dispositions to act on them in the right circumstances—should guide decision makers in determining what specific educational goods to foster. Figure 1.2 shows this relationship between educational goods and the capacities that contribute to flourishing.4

Educational Goods

23

Figure 1.2. Educational goods and capacities.

Capacity for Economic Productivity In market economies, unless they have extremely wealthy parents or some other source of guaranteed income, people’s flourishing depends on their ability to participate effectively in the economy. Some will not need to work for an income to meet their needs, but we cannot identify most of them in advance, so a sensible policy will aim to equip all children with this ability. Even those with independent sources of income usually benefit from the kinds of capability that labor markets reward. Developing individuals’ economic productivity—for example through enhancing their cognitive skills—is also in the interest of the broader society. The increased economic capability of the educated person increases the aggregate stock of human capital that society can harness to the benefit of all. Of course, this capacity is only beneficial if it is deployed, so alongside the capacity, educators should inculcate a disposition to work. As with other dispositions, the educator must exercise and encourage moderation. For most people a flourishing life will be one in which the disposition to work is balanced by other dispositions (e.g., to engage in leisure activities or to devote oneself to friends and family). 5

Capacity for Personal Autonomy Children benefit from the ability to make and act on well-informed and well-thought- out judgments about both how to live and what to do in their everyday lives. For human beings to flourish, they need to engage

24

Chapter 1

in activities and relationships that reflect their sense of who they are and what matters to them. So, for example, some people may flourish within the constraints laid down by the religious strictures of their parents, but others may be stunted by those same requirements. Knowledge of other religious views, and of nonreligious views, supports flourishing by giving the individual the opportunity to choose alternatives or aspects of them.6 Even with knowledge of the alternatives, the self-knowledge, habits of mind, and strength of character to make the appropriate alternative choice are also needed. The same logic applies to choice of occupation. Some children find themselves under very heavy parental pressure to pursue a particular occupational path. Nonautonomous people will follow the path chosen by their parents because they do not know about alternatives or lack the personal qualities to choose them. Autonomous people, by contrast, will have sufficient knowledge of the relevant variables, and sufficient self-knowledge and fortitude, to make the parental pressure an appropriate influence on their choice. Whether ultimately they choose for or against will depend on their own independent judgment of the fit between the occupation and their interests. However, autonomy does not only contribute to flourishing via its significance for major life decisions. In their everyday lives, people make and act on judgments about what to do that are not fully determined by their most fundamental commitments—like what to eat, what leisure activities to engage in, and who to talk to and about what. These choices, too, will typically contribute more to their flourishing if they have a reasonable range of valuable options and the capacity to make and act on their own judgments.

Capacity for Democratic Competence In a democratic society, citizens benefit from the ability to use their political institutions both to press their own interests and to give due weight to the legitimate interests of others.7 Educating children to have the personal attributes that enable and incline them to become effective and morally decent participants in social life and political processes benefits both them and others. The knowledge and skills needed for democratic competence are various and depend on context. A basic understanding of the history and structures of a society’s political institutions is usu-

Educational Goods

25

ally valuable, as is a basic ability and disposition to bring reason and evidence to bear on claims and arguments made by others. Institutions vary considerably in the informational demands they place on citizens, and in the deliberative resources they provide. The US electoral system, for example, with its numerous levels of government and frequent elections, places high demands on citizens, especially in those states where candidates for most elections may not register their party affi liation on the ballot paper. Political advertising gives citizens very limited help in their deliberations. Democratic systems with fewer and less frequent elections and more controls over political advertising may make it easier for citizens to participate in an informed and meaningful way, and thus will require less in terms of knowledge, skills, dispositions, and attitudes. Many policy issues are hard for citizens to evaluate because they lack a good understanding of the way the institutions work and of the possible side effects of any proposed reform. We are not advancing, here, a particular theory of what constitutes democratically competent behavior. For some theorists obedience to the law suffices; for others actual engagement in the political process is required, while some hold that competent behavior might sometimes involve challenging and breaking the law even in a democracy. Exactly what the capacity requires depends on settling these issues. But on any account, being able to engage is required, and acquiring the capacity for democratic competence is important for a flourishing society.8

Capacity for Healthy Personal Relationships Recent empirical literature confi rms the commonsense view that successful personal relationships are at the center of a happy life. The same is probably true of a flourishing life. For most of us, flourishing requires a variety of relationships, including lasting and intimate relationships with others. People derive meaning from their relationships with their spouses, their parents and children, their close friends, and even from looser ties with acquaintances in their neighborhoods and at work. Successful personal relationships require certain attributes—emotional openness, kindness, a willingness to take risks with one’s feelings, trust—that do not develop automatically but are in large part responses to one’s environment. We can hope that families will provide the kind

26

Chapter 1

of environment in which children will develop these qualities, but not all will, and even if they do this process can be supplemented and reinforced by other institutions, including schools.

Capacity to Treat Others as Equals Equal respect for the basic dignity of persons underlies the idea that everybody has the same basic human rights regardless of their sex, race, religion, or nationality, and it grounds norms against discrimination in hiring, promotion, and government provision. Regarding others as equals does not require that we care about strangers as much as we do about our family members or ourselves. Nor does it rule out judgments that people are unequal with respect to attributes such as strength, intelligence, or virtue. It means simply that we think of all people as fundamentally equal in moral status. That attitude and the accompanying dispositions are important for flourishing. Racism, for example, does not have to be legally enforced in order to be damaging. Even without legal discrimination, black Americans continue to be disadvantaged, due not only to the continuing material effects of legal discrimination but also to their treatment by others who, often unconsciously, assume superiority. The experience of slights grounded in assumptions of racial superiority—as with gender, sexuality, or physical or mental abilities— undermines the self-respect and self- confidence of the slighted, making it harder for them to flourish. The impact is worse if the slighted themselves share the attitude that they are inferior, or, while not sharing it, are nonetheless disposed to accept the slights as their due. Developing and, crucially, exercising the capacity to treat other people as moral equals is important also for one to strike the right balance between pursuing one’s own flourishing and discharging one’s obligation to contribute to the flourishing of others.

Capacity for Personal Fulfi llment Healthy personal relationships are important for flourishing, but so too are complex and satisfying labor and projects that engage one’s physical, aesthetic, intellectual, and spiritual faculties. People fi nd great sat-

Educational Goods

27

isfaction in music, literature, and the arts; games and sports; mathematics and science; and religious practice. In these and other activities, they exercise and develop their talents and meet challenges. A great deal of paid work is dreary, or carried out in the context of stressful status hierarchies, and people in such jobs have limited opportunities to flourish at work. School is a place in which children’s horizons can be broadened. They can be exposed to—and can develop enthusiasms for and competence in—activities that they would never have encountered through familial and communal networks and that sometimes suit them better than any they would have encountered in those ways. The capacity to fi nd joy and fulfi llment from experiences and activities is at the heart of a flourishing life.9

Conclusion These six capacities all contribute to flourishing lives. Although they overlap in some cases, they differ one from another, and in some circumstances the decision maker may need to trade them off against each other. Perhaps, for example, the conditions needed to foster children’s capacity for personal fulfi llment compete with those needed to maximize their economic productivity. If so, decision makers need to judge which of the capacities is more important in that context. Debates of the kind we discussed in the introduction about tracking students into different classrooms or schools on the basis of measured ability sometimes invoke beliefs about the relative importance of these different capacities. Advocates of tracking may claim that it better produces the skills and knowledge conducive to economic success while opponents worry about its impact on the capacity to treat others as moral equals, especially in contexts where measured ability correlates with other characteristics such as race or socioeconomic status. Although the approach we have suggested here focuses on the promotion and distribution of flourishing, it could readily be supplemented by nonconsequentialist considerations. Some educational goods may be important for meeting moral claims that matter independently of flourishing, and some of those claims may act as constraints on the pursuit of flourishing and its valuable distribution. For example, the capacity for personal autonomy typically contributes to flourishing, but

28

Chapter 1

some theorists regard people as having a claim to autonomy for other reasons: it might be part of respecting other people that we enable and allow them to make their own choices about how to lead their own lives even if those choices end up making them worse off than if we had made the choices for them. Furthermore, some of the independent values we shall outline in chapter 2—which, we say, can reasonably be balanced against the production of educational goods—may be important independently of their contribution to flourishing. The most obvious example here is respect for democratic processes, which some regard as owed to people by virtue of their moral status as citizens, not because it makes their lives go better. Meeting such claims might involve limiting the pursuit of flourishing or ruling out particular ways of distributing it. The fact that we present here a consequentialist framework for educational decision making that gives pride of place to flourishing does not mean that we regard these other considerations as irrelevant or misconceived. It is important to remember, too, the role that institutions other than schools play in the production (and sometimes in frustrating the production) of educational goods. Children are subject to influences from the home, the neighborhood, the media, their peers, and the economy. Children whose home life is dysfunctional may need particular experiences outside the family in order to develop the capacity for healthy emotional relationships. It might be more difficult to foster the capacity for economic participation in a child whose community experiences very high unemployment than in a child who expects her efforts to be rewarded by college and lucrative employment. A political culture short of examples of high-profi le reasonable deliberation among politicians who disagree might frustrate the development of democratic competence. We are offering an account of what educational goods should be developed in children, not saying that schools should be left to, or even could, fulfi l this task on their own. A complete theory of flourishing would show how to weigh its different constituent elements against one another. One would then be able to judge, for example, whether somebody whose capacity for economic productivity is very well developed and possesses the other capacities, but only to some threshold level, has greater prospects for overall flourishing than somebody who has very high levels of those other capacities but is not so economically productive. We do not have such a theory. Like almost everybody else, we lack a reliable method for rendering the dif-

Educational Goods

29

ferent constituent elements of flourishing commensurable.10 Still, decision makers can make some comparative judgments even without a finetuned conversion measure. Indeed, they must make those comparative judgments, just as we must all make them in deciding what to do in our own lives and how to treat others well.

chapter two

Distributive Values and Independent Values

E

ducational goods are valuable because they enable individuals who have them to flourish and to contribute to the flourishing of others. Creating educational goods is the central purpose of the schooling system. But the education system also inevitably influences how educational goods are distributed. In fact, directly or indirectly, much public debate about schooling addresses the distribution of educational goods. Efforts to “close the achievement gap” focus on one measure of educational goods—performance on standardized tests—and aim to reduce the difference in average test scores between groups, or differences in the rates at which members of different demographic groups reach proficiency or similar levels of achievement. Advocates of more funding to schools with particular characteristics presumably hope that funding will improve the learning—that is, the level of educational goods—of the students, or some of the students, in those particular schools. Educational goods are not created in a vacuum. Policy choices that affect the level of educational goods and their distribution across students often also affect the realization of other values. Eliminating expensive private schools might lead to a more equitable distribution of educational goods, but it would also reduce parents’ ability to make decisions that affect their own children. Strong punishments for poor performance could lead to increased achievement levels for students while reducing the pleasure they get from their experiences in school. Scholars commonly talk about values such as “social justice” or “equity.” Sometimes they refer to the “common good” or “democracy.” But they usually use these terms rather loosely. Careful identification of

Distributive Values and Independent Values

31

the more specific, distinct values subsumed under such general labels can help us understand better which considerations are at stake in any attempt to determine whether one option is better or worse than another. We start this chapter by outlining three different distributive values and giving reasons for caring about each of them. After that, we look at what we call independent values. These are values that can come into confl ict with the production of educational goods, or with efforts to influence their distribution, and which sometimes demand that tradeoffs be made.

Distributive Values Distributive values typically have two components: (i) a distributive rule and (ii) an object of distribution to which that rule applies. We propose, without much argument, three distributive values: adequacy of educational goods (adequacy), equality of educational goods (equality), and the distribution of educational goods that most benefits those with the worst prospects for flourishing (benefiting the less advantaged).1 Note that these distributive values refer to educational goods as a whole, not to each educational good considered separately. Individuals can have an equal chance of flourishing with very different distributions of educational goods. Some may have a high capacity for economic productivity but a relatively low capacity for autonomy. Others may have high capacity for personal fulfi llment, perhaps from music or sports, but less capacity for healthy personal relations. Part of a society’s strength resides in the diversity of the basket of educational goods held by its members. Nonetheless, while allowing for diversity in content, many people value some distributions of the potential for flourishing that each basket creates more than they value other distributions. In practice, decision makers cannot directly distribute educational goods or prospects for flourishing. Federal, state, and district-level decision makers determine how funding is distributed and how it may be used. They regulate schools by creating incentives, constructing and implementing accountability systems, and imposing licensing requirements. Decision makers at the school level choose how to allocate students to teachers, which teachers to hire, and what kind of instructional leadership to provide. Classroom teachers decide how to allocate their time, energy, and attention within the classroom and to what end. When

32

Chapter 2

doing these things, they are typically aiming at (though often failing to bring about) one or more of these distributive values. Adequacy Adequacy has been an appealing principle in the context of school finance litigation in the United States primarily because several state constitutions can be interpreted as requiring that every child receive an adequate education. But provision can only be judged adequate relative to some specified goal. Among the goals philosophical advocates of adequacy have specified are “earning a living wage,” “functioning as a democratic citizen,” and “being able to participate as an equal in social and political life”. 2 Each of those goals refers to capacities that we have described separately but have presented as contributions to the more ultimate goal of flourishing. So we understand adequate educational provision in terms of the level required for some acceptable (adequate) level of educational goods that are, in turn, adequate for flourishing in adulthood. Public institutions, in other words, should ensure that everyone has the educational goods adequate to enable them to have a reasonable chance of attaining some threshold level of overall flourishing as adults. 3 Some advocates of adequacy principles claim that no other distributive principle is needed: as long as everyone has enough educational goods, inequalities between individuals do not matter. Others fi nd this stance unsatisfying. Imagine that everyone is adequately well educated (understanding adequacy however you prefer). Now suppose that new resources are available for educational purposes and that, however they are distributed, everybody’s education remains adequate. Adequacy gives us no guidance as to how to distribute the new resources. But it seems intuitive that some ways of distributing them are better than others. Equality One reason to introduce equality as a second distributive principle is to supplement adequacy. Educational goods enrich the lives of those who have them, enabling them to live emotionally healthier, more fulfi lling lives and to contribute more to the projects they care about. Giving children more equal opportunities for their lives to go well would require a more equal distribution of educational goods. Surveys suggest widespread public support for equal opportunity, which is hardly surprising.

Distributive Values and Independent Values

33

Because children are morally equal, equality might seem like a sensible default.4 Another reason to consider equality is that some educational goods are, in part, what economists call positional goods. That is, their worth, to those who have them, depends partly on how much they have relative to others. When competing for jobs and others positions that are allocated on the basis of one’s holdings of educational goods, what matters is not one’s own absolute level but one’s level relative to the levels of one’s competitors. In some contexts, inequalities do no harm to those who have less, but positional goods are different: the fact that some have more than others reduces the competitive chances of those who have less. 5 Endorsing a more equal distribution of educational goods does not involve a commitment to egalitarianism with respect to flourishing itself. As we have explained, educational goods provide people with opportunities; how wealthy they are or how well their lives go may properly depend on their choices as well as luck. Still, the idea of unequal outcomes being acceptable where they arise from equal opportunities is difficult to accept when we are thinking about choices made in childhood. Children, being children, are not fundamentally responsible for their choices, so one might argue that the level of educational goods, or flourishing, enjoyed by adults should not depend on the choices they made as children. Of course, schooling policy is limited in what it can do to equalize educational goods. Background inequalities of education, wealth, income, and other kinds of advantage inevitably influence how well children respond to educational offerings. It is hard to imagine a society that tolerates extensive background inequalities being willing to distribute public educational resources unequally enough to counteract those effects fully. Although a commitment to equalizing educational goods does not imply a commitment to other kinds of equalization in theory, in practice greater equality of educational goods might depend on greater equality of background conditions. The inequalities in educational goods resulting from differences in individual capacities to develop educational goods, even among children within the same family, are also difficult to compensate for fully. When the only way to achieve an equal distribution is to make some people worse off without making others better off, then equality recommends leveling down. Leveling down often seems unjustified, especially when it affects the distribution without actually benefiting anyone, for example, if the government tried to reduce the happiness level of all to

34

Chapter 2

that of the unhappiest person for the sake of equality. It is easy to see why equalizing educational goods might require leveling down. In a society with substantial social and economic inequalities, some children will enjoy very little private investment, and there may be a limit to how well public resources can compensate. That society would have to limit support for advantaged children so that they could not attain more educational goods than could be achieved by poorer children. Similarly, some children are born with severe cognitive impairments. Requiring cognitive outcomes for these children to be equal to those of unimpaired children would currently require reducing the capacities of the unimpaired, as we do not yet have the technology to bring those with low aptitudes up. Even within the group of children considered unimpaired, the range in cognitive abilities is large. Equalizing educational goods would require extremely unequal investments in children. It would involve refraining from fostering—and perhaps even reducing (e.g., through physical or psychological mistreatment)—the capabilities of talented children. Equalizing educational goods could be very costly for total productive capacity. Pursuing equality by any form of leveling down decreases the human capital available to society, reducing material resources and the prospects for the development and affordable production of lifeimproving technologies. Investing in the development of highly talented people can pay off for others through their enhanced productivity, which can redound to the benefit of all. Suppose there is indeed a trade- off between equality in the distribution of educational goods on the one hand and the total amount of those goods produced on the other. It may seem perverse to favor equality in those cases where an unequal distribution of educational goods would benefit those who have least. Benefi ting the Less Advantaged Concerns of this kind motivate a third principle, which demands distributions that serve, over time, to benefit the less advantaged.6 Of course, as with adequacy, it is very hard to identify with any precision exactly which particular distributions satisfy this rule, partly because, given the complex and dynamic interaction between education and changes in the economy, it is difficult to predict what the effects of a particular distribution of education right now will be on the long-term flourishing of the students. There is likely to be disagreement about how much inequality with respect to which capacities is required or permitted by

Distributive Values and Independent Values

35

the principle. Still, the general idea that inequalities should be licensed where they are needed to raise those at the bottom is intuitively appealing and is likely to play a role in educational decision making as in other policy areas. It is important here to defi ne the less advantaged and to be clear about the significance of educational goods for their overall opportunities for flourishing. It is one thing to distribute educational goods (or the resources that produce them) in such a way as to increase the educational goods possessed by the worse off members of society. It is another thing to distribute educational goods (or relevant resources) in ways that do most for their overall prospects for flourishing. This distinction may be particularly salient in the case of the cognitively impaired, whose opportunities for flourishing may not be greatly influenced by their possession of educational goods beyond some threshold. To make their lives go better over their life course, we might sometimes plausibly subordinate their level of educational goods to whatever distribution of educational goods will produce the technological, medical, or social developments most conducive to their longer-term well-being. So, for example, for a given unit of resource, it might sometimes be better for the seriously cognitively impaired person that resources be devoted to better training for future caregivers, who will support them in adulthood, than to improving their own education. In some circumstances, prioritizing benefit to the less advantaged can be particularly urgent. In a society like the United States, people in the bottom 20– 25 percent of the occupational structure usually have jobs that are insecure, badly paid, dull, lacking in meaning, and without career prospects. The low and often intermittent pay, the need to change jobs frequently, and a regulatory regime that is generous to landlords mean that workers have insecure housing situations, usually in neighborhoods that are dangerous and unhealthy. They have poor access to primary health care and often suffer from de facto social exclusion. Educating students from this kind of social background better, thereby improving their prospects for escaping their situation and for coping with it if it cannot be transcended, is a matter of moral urgency. We ended the previous chapter with an observation about the plurality of educational goods and the need for trade- offs between them. The same applies here. Decision makers sometimes face trade- offs between these different distributive values—sacrificing equality for the sake of benefiting the less advantaged, for example, or choosing to ensure that

36

Chapter 2

as many children as possible achieve adequacy rather than helping those with the lowest prospects for flourishing. So policy makers must not only weigh different educational goods against one another, they must also try to get the right balance between the various principles that apply to their distribution. This point is perhaps particularly relevant to discussions of equality, which is sometimes rejected on the simple ground that full or complete equality (of anything) is a very implausible goal. Equality need not be all or nothing: one could value a move toward a less unequal distribution of educational goods without endorsing strict equality. Similarly, one could aim to move more students over a threshold of adequacy without pursuing that goal for all.

Independent Values Educational goods and their distribution are key values, but they are clearly not the only factors that decision makers need to take into account. Educational goods make an important contribution to flourishing, but so do many other goods that compete with them for resources. Even where it is not a matter of allocating resources, the amount and distribution of educational goods that can be achieved must be balanced against other values. Consider, for example, attempts to move toward greater equality of educational goods. Some policies or practices aimed at increasing equality can be accomplished without excessive cost in terms of other values. For example, public education—which almost certainly mitigates the inequalities that would be produced by an unregulated privately funded school system—is widely regarded as acceptable. Completely equalizing educational goods, by contrast, would probably require abolishing the family and reducing the capacities of those with greater native abilities. Why should society refrain from taking those measures? Not because equalizing educational goods is undesirable but because these measures confl ict with other important values. We call these “independent values simply to indicate that they are neither educational goods nor valuable distributions of educational goods. These also contribute to flourishing but not via the production of educational goods. The full set would be unmanageable, but the independent values most relevant to educational decisions can usefully be reduced

Distributive Values and Independent Values

37

Figure 2.1. Independent values.

to five: childhood goods, parents’ interests, respect for democratic processes, freedom of residence and of occupation, and other goods. Educational decision makers can systematically assess the merits of the options that they are considering by explicitly considering the implications of those options for these five values alongside the level and distribution of educational goods. Childhood Goods The values we described in chapter 1 largely (though not entirely) concern how the child is supposed to turn out as an adult, and spending on education is rightly regarded as investment in the future. Any psychologically realistic theory of flourishing will recognize the formative and developmental importance of childhood. Childhood experiences have profound effects on what a person is like, and therefore what their life is like, in adulthood. But children are more than just adults in formation. Childhood is itself a significant part of a person’s life, and the quality of a childhood is intrinsically important independently of its consequences for the quality of the adulthood that follows.7 Some goods may be available only in childhood. Purposeless play, naive curiosity, unreserved joy, and carefreeness are the most obvious examples. More controversial additions might include innocence of adult

38

Chapter 2

sexuality and unawareness of certain horrors such as violent death and mortal illness. These goods may well contribute to or even be essential for healthy development, but children have an interest in experiencing them regardless of their developmental effects.8 They are good things in themselves, part of a good childhood and a good life. Among equally effective ways of prompting learning, that which least compromises these childhood goods is better. Even in cases where a child would reap some benefit as an adult from educational experiences that undermine the special goods of childhood, it still might be better to forego the adult benefit. Other goods experienced in childhood are not special to that stage of life—for example, friendship, physical pleasures and the enjoyment of the process of learning. Missing out on some of these in childhood, as at any other stage of life, can more easily be justified by subsequent benefits. We often deprive children of some goods in the moment for the sake of education that will enable them to get more goods over their lives as a whole. But childhood is itself part of a life, and the flourishing enjoyed during it should not be discounted. Deficits in childhood matter even when they are instrumental for benefits later. In some circumstances, for example, frequent and rigorous testing might be part of the most effective method for improving achievement, but it might make some children very anxious at the time. Even if we were confident that a rigorous testing regime were the best strategy for improving children’s performance and would have no lasting effects on their emotional or psychological development, we might choose to deprive them of some of the educational goods that regime could yield for the sake of not making their school days unduly miserable. Parents’ Interests Children are normally raised in families, by parents who invest a great deal of time, energy, and emotion in the well-being of their children. Parents care about how their children are educated and typically feel responsible for seeing that their children’s development goes well. Children often benefit from their parents’ authority over them and from the considerable discretion their parents have over how they are raised. This authority and discretion may also be good for parents if it allows them to maintain close relationships with their children, raise their children as members of a particular faith, or do their best, by their own lights, for their children.

Distributive Values and Independent Values

39

There is considerable dispute about how much say parents should have in their children’s education. Parents’ wishes may confl ict with the development of children’s educational goods, such as the capacity for economic productivity or healthy relationships, or with distributive values such as equality. For example, there is an obvious confl ict between educational equality and the interest of wealthy parents in conferring advantage on their children, which has recently received considerable attention because of increasing income inequality, especially at the top of the income distribution.9 Different readers will doubtless weigh the different goods and distributive values differently. Our purpose is not to offer any judgments about their relative importance but simply to point out that parents’ interests matter and that the decision maker cannot avoid making a judgment about their weight and the extent to which, and ways in which, respecting them constrains or enhances the promotion of educational goods and their valuable distribution. Respect for Democratic Processes Some things that decision makers would like to do are not feasible, perhaps because the technologies required are not reliable or are too expensive or because producing the desired outcomes is impossible given certain institutional features they have no power to change. But sometimes, even when decision makers could technically do what they wanted, they are constrained by respect for democratic processes. For example, they may judge that important distributive values could be realized by redistributing funds away from schools in wealthy neighborhoods to those serving students from the poorest families. If they believe that the voters who understood the policy would oppose it, then they might feel constrained by respect for democracy. Even if instituted, in the long run this redistributive policy might not be secure because voters may remove decision makers from their positions of authority. Just as there is a value to individuals having control over their own lives and choices independently of the quality of their decisions, so there may be a value to people exercising control of their shared environment and the rules (including the rules about education) that apply to them collectively even if some of the decisions are suboptimal. For many decisions, being made through democratic processes realizes a value that they lack if they are made in other ways. Sometimes better decisions

40

Chapter 2

about the production and distribution of educational goods should be rejected because those decisions could only be implemented by circumventing democratic procedures. Freedom of Residence and of Occupation Some freedoms not directly or obviously connected with education can be in tension with what decision makers might seek to do to produce more educational goods and distribute them better. Some think that freedom per se is valuable irrespective of its particular content, but we focus on two specific freedoms that are both independently valuable and particularly pertinent to the design of school systems: residential location and occupation. One way to lessen educational inequalities might be to restrict people’s freedom of residential choice in order to reduce neighborhood, and therefore school, segregation. Alternatively, schools might be more effective if authorities could simply draft adults with special talents to become teachers or school administrators in particular schools—just as countries sometimes conscript people into the military—rather than relying on incentives to influence the occupational choices of individuals. That would restrict people’s freedom of occupational choice. Decision makers must take into account adults’ interests in these and other freedoms when pursuing the values concerning children’s education, judging which are important enough to serve as legitimate constraints on the promotion and/or valuable distribution of educational goods. As with other values, we cannot offer a precise interpretation of exactly what freedom of occupation and residence amount to. We assume that, whatever the actual value of freedom of residence and of occupation, most contemporary liberal democracies will not contemplate eliminating the housing market or conscripting teaching labor. They might, however, reasonably consider other measures that impose costs on residential choice, such as imposing higher tax rates on properties in wealthy residential areas than on properties in socioeconomically mixed neighborhoods. Similarly, though they may not conscript, they might seek to induce talented people to become teachers by raising wage rates in the early part of their career or reducing the costs of entering the profession with signing bonuses and debt forgiveness. As with the other values, of course, readers will disagree about exactly what freedoms should be included and how much weight they

Distributive Values and Independent Values

41

should have relative to educational values. We remind readers that we are simply offering a framework—a kind of skeletal structure—for formulating and assessing the moral and political considerations that are inevitably involved in judgments about what to do in educational contexts. We are not providing an exhaustive account of all possibly relevant values. We have merely highlighted those that are most frequently salient for decisions about education. Other Goods Educational goods are very important, especially when understood in the broad way proposed here. But no society would want to devote all its resources to the production or valuable distribution of educational goods. To spend resources on education is to invest in children’s future flourishing by producing educational goods. But other goods also contribute to both current and future flourishing. Decisions about how much to spend on education affect what is available for investment in other government and private activities—health, transportation, housing, environmental protection, and so on. Here, too, it will be a question of balancing values and weighing the contributions that different forms of resource allocation can be expected to make to overall flourishing and its distribution, both current and future.

Freedom and Markets Readers may be concerned that we have left out something—or perhaps two things—from our list of distributive and independent values. One of the key changes over the past forty years, especially in Anglophone countries, has been the deployment of markets in many areas of service delivery within the public sector—and notably in schooling. Enthusiasm for markets is often connected to a valuing of freedom or liberty. This is most striking in the work of Milton Friedman, who advocated a much greater role for markets in education based on general considerations that ground support for markets in a political philosophy that gives pride of place to freedom.10 It is worth noting, then, why neither free markets nor freedom (as such) appear in our list of values. Markets are not, and do not represent, a distinctive value. The decision to open up schooling to markets,

42

Chapter 2

whether through introducing intradistrict school choice schemes or charter schools or vouchers or, most radically, by withdrawing public support and regulating schools only in the same way as other businesses, is an institutional choice. Markets constitute a policy decision, not a value, so they do not belong in this chapter. Chapter 7 considers one proposal for increasing the role of market mechanisms in schooling—the proposal that we think is the most feasible in wealthy democracies in the current era. If we thought more radical proposals were feasible we might have considered them. But increasing the role of markets, as Friedman himself makes entirely clear, is a policy choice that needs to be justified and justified in terms of values. That is, markets are not valuable in themselves but are institutions that realize values. Although enthusiasm for markets is connected with valuing freedom in public discourse, the connection is complex. Among theorists who promote markets in education, one of the most radical—because he advocates removing the government from provision, funding, and regulation—is James Tooley. His argument, however, does not appeal to freedom. Rather, he invokes one of the distributive values that we have already surveyed—adequacy—which he believes would be more secure if education were provided entirely by the free market. This book is not the place to go into his argument in any detail, partly because, in our view, his proposal is not feasible, and it depends on the unsupported empirical conjecture that unregulated open markets would be better for the less advantaged students than governments currently are.11 Markets are not valuable, or valued, in themselves but are mechanisms for realizing values. By contrast, freedom is widely regarded as a value. For reasons that would take us too far from the practical task of this book, we think this is also a mistake. We do not see freedom per se as the right value to be concerned with when thinking about policy decisions. Instead, we think of specific freedoms, those to which people have particularly strong claims, as relevant to those decisions; for example, freedom of expression, freedom of religion, and freedom of association. So in our list of independent values we have identified those freedoms that are most plausibly and commonly regarded as weighing in the balance against the production and distribution of educational goods. These are the freedoms associated with family life (which we have labeled parents’ interests), free choice of occupation, and the freedom to choose where to reside. Readers will disagree about how weighty those freedoms are compared with the other values. Some may disagree with

Distributive Values and Independent Values

43

us philosophically about the value of freedom per se. We hope such readers can nevertheless think through the issues we discuss using the procedure we recommend in chapter 4 but weighting the values as they see fit.

Conclusion The previous chapter identified the goods that educational systems— and schools in particular—should seek to produce. But producing educational goods is not the only thing that matters. On the one hand, some distributions of those goods—or of opportunities to possess them—are better than others. On the other hand, there are other, independent values that properly constrain both the production and distribution of educational goods. Our aim in this chapter has thus been twofold. First, by distinguishing three values that are often run together under a vague term such as equity or justice, we have provided the resources needed for clear and careful thinking about the distributive issues raised by education policy. Second, we have enumerated a manageable list of the other, independent values that responsible decision makers will want to keep in mind as they make their decisions. In the next chapter we illustrate the framework we have set out so far by focusing on the example of student achievement as an educational good.

chapter three

Achievement as an Educational Good

I

n chapter 1, we defi ned the term educational goods as the knowledge, skills, attitudes, and dispositions that enable individuals to lead flourishing lives and to contribute to the flourishing of others. In practice, however, both policy makers and researchers often focus on the far narrower concept of student achievement. By achievement we refer to the cognitive knowledge and skills purportedly measured by the various standardized tests generally used in the United States (and elsewhere) to gauge student outcomes. Policy makers and researchers concentrate on achievement in part because they believe it is particularly valuable and in part because it is easier to measure, albeit imperfectly, than other valued outcomes. Because cognitive outcomes are only one element in the set of educational goods, however, and the typical tests do not even measure the full range of cognitive skills, the reader may wonder why we devote a whole chapter to this topic. We do so for several reasons. One is that it provides a vehicle for showcasing the role of social science evidence in thinking about educational goods. Another is that the distributions of quantitative measures such as test scores enable us to clarify the abstract distributive principles that we introduced in chapter 2 and to illustrate some of the normative issues and trade- offs that arise when considering achievement levels and gaps. Finally, it allows us to highlight both the strengths and limitations of achievement as a proxy for the broader set of educational goods that are the focus of the book. We begin with a brief discussion of achievement as an educational good. We then discuss the level and distribution of test scores—as a measure of achievement—across individual students. This approach allows us to illustrate concretely some of the values and principles we intro-

Achievement as an Educational Good

45

duced in chapter 2. We also note how consideration of the distribution of some educational goods rather than the full range could lead to incorrect inferences. Finally we analyze achievement gaps between groups of students with particular attention to the question of which gaps are normatively salient.

Achievement as an Educational Good The most obvious link between student achievement as measured by test scores and subsequent flourishing is through the labor market. According to a conventional economic model, high-achieving students bring greater knowledge and skills, at least on one dimension, to the labor market than do lower-achieving students and thus may be more productive in their jobs. Employers reward these attributes with high wages and good opportunities for learning on the job and career advancement. Low skills in turn relegate a person to the low-wage labor market, which, in addition to low wages, often brings with it greater difficulty fi nding a job and limited opportunities for advancement. This economic model is supported by the fi nding of a clear relationship between the test scores of male teenagers on a modified version of the Armed Forces Qualifying Test (AFQT) and the subsequent wages of those same males, both when they were twenty-six to twenty- eight and in their late thirties. The AFQT is considered a fairly broad and accurate representation of “achievement and learned skills” or “cognitive skills.” Compared with their lower-scoring counterparts, workers with higher test scores as teenagers end up in jobs with higher wages in their twenties, and the wage gap is even larger when they are in their thirties.1 A similar relationship emerges for test scores of the type used in many US states and school districts. In a recent study of one large US city, test scores of students in grades three through eight (based on administrative records) were matched to their subsequent earnings (based on tax records) when they were in their twenties. Conditional on the previous year’s test scores and other student and control variables, the authors found that a difference of one standard deviation (a common measure of variation described below) in test scores was associated with an average difference of 11.6 percent in earnings. 2 Various earlier studies show the same patterns. Higher measured achievement predicts higher future earnings.

46

Chapter 3

The possession of cognitive skills and knowledge makes it possible for individuals to flourish in other ways than higher income as well, both while they are in school and as adults. Such skills enrich a person’s life by opening doors to science, literature, arts and music, and, importantly, by creating opportunities for higher education and the acquisition of degrees and other accomplishments. People with more years of schooling typically receive nonpecuniary benefits connected with their jobs, such as higher job satisfaction, higher occupational prestige, and lower levels of unemployment. They also receive a variety of benefits outside the labor market. As adults they have better health, are less likely to get divorced or separated, have increased trust in others, are better parents, and are less likely to be arrested and to smoke. In addition, they are likely to enjoy more leisure and to have healthier families. 3 Cognitive skills and years of schooling benefit not only the individuals who have them but also others. Everyone gains from having a skilled labor force that is productive and innovative. One study found that for every 10 percent increase in the percent of college graduates in a city, the wages increased by 5– 7 percent.4 Further, those with high cognitive skills can generate the technological innovations that are needed for productivity growth. People with more education also exhibit higher voter turnout and civic participation, thereby contributing to the strength of our democracy. 5 One must, however, be careful in attributing causation to observed statistical relationships among cognitive skills of the type that are measured by test scores, years of schooling, and other outcomes. It may be that other types of knowledge, skills, and dispositions—that is, other educational goods—are an important part of the story. The US experience with the General Educational Development program (commonly referred to as the GED) is illustrative. The GED test was initially designed to enable returning World War II veterans to claim a high school degree without having to go back to high school. It is now used primarily as a way for high school dropouts to show they have mastered the same cognitive skills as those who graduate from high school. With that credential, GED holders would then gain access to the jobs and higher education opportunities available to those with high school degrees. The Nobel Prize–winning economist James Heckman, however, has demonstrated the fallacy of the approach. The problem is that the GED measures only the cognitive skills that can be captured by tests and fails to measure what he calls the noncognitive skills that are required not only

Achievement as an Educational Good

47

for young people to complete high school but also to succeed in the labor market or in subsequent educational endeavors.6 Although left unspecified by Heckman, the term noncognitive skills is usually understood to include characteristics such as persistence, motivation, and self- control, some of which may in fact include a cognitive component and some of which may be more like dispositions and attitudes than skills. Other research suggests that these noncognitive skills interact with cognitive skills, making it difficult to separate out their specific contributions to subsequent flourishing. The potential for cognitive skills, especially those typically measured by standardized tests, to contribute to flourishing in the ways described above may depend on an individual possessing other educational goods as well.7 While cognitive achievement undoubtedly contributes to the development of some of the capacities we identified in chapter 1, it is more important for some than for others. Recall that these competencies include the capacity for economic productivity, for personal autonomy, for democratic competence, to regard others as equals, for healthy personal relationships, and for personal fulfi lment. Cognitive skills are particularly relevant for economic productivity, personal autonomy, democratic competence, and some aspects of personal fulfi lment. They are less central, however, for an individual to develop strong personal relationships or to develop the capacity for treating others as equals. If cognitive achievement is just one of many educational goods that contribute to flourishing either by themselves or in complicated ways, why does it get so much attention in policy discussions? One reason is the broad agreement that such achievement is valuable, that a certain level is needed to function as an adult in the current context, and that schools in particular should aim to develop these attributes in their students. Other educational goods and their associated capacities are less widely recognized or are often not regarded as a matter for schools. A second reason is that, compared with other educational goods, achievement is far easier to measure. That does not mean, however, that the tests used to measure it do a good job of accounting for the full set of cognitive knowledge and skills that contribute to flourishing. For one thing, the tests used for elementary and middle school students often focus on math and reading alone, which represent a small subset of valuable knowledge and skills. Moreover, to keep costs at a manageable level, many states rely heavily on electronically gradable multiple- choice tests. These tend to focus on basic skills rather than broader conceptual, analytical, and

48

Chapter 3

problem-solving skills that may be of greater relevance for flourishing. As we shall see in chapter 6, not only may these tests measure the wrong thing but relying on them can create incentives for teachers not to teach valuable skills simply because they are not tested. Despite these limitations, student achievement still serves as a useful vehicle for illustrating some of the values we have set out in previous chapters. For that purpose, the discussion that follows simply assumes that test results provide a good measure of each student’s knowledge and skills in subjects relevant for subsequent flourishing. Even on that assumption, problems can arise if only this one type of educational good is taken into account when making decisions about schooling and education policy more generally.

Achievement and Values Some children achieve at substantially higher levels than others. This variation reflects several distinct but interrelated factors, such as differences in children’s family background and access to educational opportunities, differences in characteristics such as intelligence or talent, and differences in preferences or motivation. Choices made by educational decision makers can affect the distribution of achievement. We consider these distributions, and changes to them, in order to provide concrete examples of educational goods and of the distributive values that we outlined in chapters 1 and 2, respectively. We use the bell-shaped distribution in figure 3.1 to portray the variation in achievement across students. Along the horizontal axis is student achievement, which in practice would be measured by test scores. The area under the curve between any two achievement levels represents the  percentage of students whose achievement falls in that range, with the area under the full curve representing 100 percent of the students. To keep things simple, we have drawn the curve as symmetric, which means that the average (or mean) achievement level is at the center of the distribution as marked on the figure by Ā. The bell-shaped distribution implies that many students achieve at levels close to the average and that fewer students have achievement levels in either of the tails of the distribution. In practice, the distribution need not be symmetric. If the relevant population of students included a disproportionate number of disadvantaged

Achievement as an Educational Good

49

students who achieve at lower levels than their more advantaged counterparts, for example, the area under the left tail would be larger, and the average would be farther to the left. Similarly, if the relevant population included a disproportionate number of highly motivated students who achieve at high levels, the area under the right tail would be larger, and the average would be farther to the right. For the current illustrative discussion, the precise shape of the distribution is not important. The other distributions—in figures 3.2, 3.3, and 3.4—highlight three concepts that are useful for normative discussions about student achievement. Figure 3.2 illustrates a higher average achievement level with no difference in how achievement is distributed around the average. In that figure, the whole distribution is simply farther to the right. Figure  3.3 illustrates the concept of variation. With more students having very low or very high levels of achievement, the variation is clearly larger in the dashed distribution than in the original distribution. For bellshaped curves, empirical researchers often use the concept of a standard deviation—which is a statistical measure of the average difference between the achievement of each student and the mean achievement—to talk about this variation. A useful rule of thumb is that for a bell-shaped distribution, 68 percent of the observations fall within one standard deviation of the average (with 34 percent falling within one standard deviation below and 34 percent within one standard deviation above the

Figure 3.1. Distribution of student-level achievement as measured by test scores. Ā  denotes average achievement.

50

Chapter 3

Figure 3.2. Comparison of groups with different levels of average achievement. The solid line depicts higher levels of achievement with no change in the distribution of achievement around the average.

Figure 3.3. Comparison of groups with different amounts of variation. The dashed line depicts a distribution with greater variation around the average.

average). Thus, the standard deviation would be larger for the dashed distribution in figure 3.3 than for the solid one. The advantage of a specific measure of this form is that it allows policy makers and researchers to use numbers to compare the variation across distributions and, as we discuss below, to put the magnitudes of gaps into perspective. Finally, Figure 3.4 illustrates the concept of an adequate level of achievement. If the adequate level of achievement were deemed to be A L (where L signi-

Achievement as an Educational Good

51

Figure 3.4. Depiction of two possible thresholds of adequacy. A L denotes a low threshold. A H denotes a high level.

fies low), we would see that well over half of all students are achieving at an adequate level or higher, or, stated differently, that a relatively small proportion are achieving at levels that are inadequate. If the adequate level of achievement were deemed to be A H (where H signifies high) in the figure, however, well over half the students would be achieving at inadequately low levels. With these figures in mind, we can now turn to the normative significance of achievement and how it is distributed across students. Because achievement is an example of an educational good, and hence is something valuable, all else being equal a higher average level of achievement as shown in figure 3.2 is better than a lower level. Some cautionary notes are in order. Gains in average achievement are not necessarily better overall. To the extent that higher average achievement is attained only by increases in student achievement at the top of the distribution, with no gains—or even losses—at the bottom (rather than by a rightward shift in the whole distribution as shown in fig. 3.2), policy makers would need to trade off the benefits of having more of this educational good against the costs in terms of one or more distributive values. When increases in student achievement come at the expense of other educational goods, policy makers face another trade- off. If, for example, the policy mechanisms used to raise student achievement adversely affect other dispositions and attitudes that also contribute to future flourishing, policy makers will have to weigh these losses against the increase in achievement.

52

Chapter 3

Finally, trade- offs with other, independent values may arise; childhood goods and parents’ interests are especially relevant. While higher achievement for young children is desirable, so also are goods such as play and naive curiosity. Concern for healthy childhood development and happiness within childhood may sometimes argue against an excessive focus on achievement. A trade- off with parents’ interests arises whenever policy interventions designed to promote higher achievement interfere with parents’ views about how their children should be raised. Concern for parents’ interests means that that any potential gains in achievement, though no less important in themselves, should be weighed against the cost to those interests. Distributive Principles In chapter 2 we identified three main distributive principles—adequacy, equality, and benefiting the less advantaged—and emphasized that they can apply to the distribution of either educational goods or flourishing. If we apply them to student achievement, we must keep in mind that this is just one educational good that contributes to flourishing, and sometimes it competes with others. Adequacy Recall that adequacy requires that all students receive at least an adequate level of educational goods with no attention to their distribution above the threshold. We suggested that adequacy should be understood in terms of what students will need to flourish as adults. So, ideally, adequacy would be specified in terms of the full bundle of educational goods, not simply in terms of student achievement as measured by test scores. In the 1989 Rose school fi nance case that addressed adequacy, the Kentucky court articulated a broad set of educational goods, as have several other state courts since then. The Kentucky court defi ned an adequate education as one that fosters seven learning goals: oral and written communication skills, knowledge of economic social and political systems, understanding of governmental processes, self-knowledge, grounding in the arts, and academic or vocational skills. The intention was for each child to emerge from school ready to function in a complex and changing world.8 An individual may have adequate levels of educational goods with quite low levels of cognitive skills if they have other educational goods that compensate, or they could have high test scores but

Achievement as an Educational Good

53

inadequate goods for flourishing because they have such low levels of other educational goods. Whatever our conception of adequacy, raising the achievement  of those at the bottom of the achievement distribution up to the level deemed adequate, with no change in what happens at the top of the distribution, would concomitantly make the overall distribution of achievement more equal. How much increasing equality of achievement increases equality of flourishing will partly depend on the extent to which achievement is a positional good, that is, one for which the benefits depend not on how much of the good someone has but rather on his or her position in its distribution. Thus, while the pursuit of adequacy defi ned in terms of achievement may lead to a more equal distribution of the capacity to enjoy literature or solve math problems, it may do nothing to equalize the capacity of people to obtain good jobs if such jobs are limited. In that case, despite the higher skills of those at the bottom of the distribution, the good jobs will continue to go to people who are relatively more skilled. Setting aside the positional aspects of achievement, there are two main ways to meet an achievement-based adequacy standard. One is to move the whole achievement distribution so far to the right that all children attain at least the adequate level. The appeal of that approach is that it would simultaneously achieve two normative goals—adequacy and more educational goods for everyone—but the costs in the form of substantial additional investments in schooling, and hence in other consumption goods foregone, are likely to be very high. A modified version of this approach would be to intervene in ways that increase achievement at the bottom of the distribution but increase it even more at the top. In this case, the movement toward an adequacy standard brings with it greater inequality, thereby generating a trade- off between those two distributive principles. An alternative, more cost- effective strategy—though possibly politically difficult to implement—would be to target resources and attention on the bottom of the distribution with the goal of raising the achievement of all the low-achieving students to the specified level. That strategy has the advantage of allowing policy makers to craft policies and programs specifically designed to address the learning needs of the lowachieving children and is less costly in that it requires no additional investments for children already above the threshold. If the budget for edu-

54

Chapter 3

cational goods was fi xed, however, that strategy would require a transfer of resources from those above to those below the standard and possibly a reduction in the total amount of educational goods produced. Equality Suppose that instead of, or perhaps in addition to, valuing adequacy, policy makers aim to promote an equal distribution of educational goods across individuals. Equality in the distribution of those goods need not require equality in the distribution of achievement, because any inequalities in that particular educational good could potentially be offset by inequalities in others. Indeed, in principle, observing achievement differences across individuals might not tell us much about the distribution of educational goods overall. In fact, the distinct but interrelated factors and processes that give rise to achievement differences—such as inequalities of family wealth, stress, levels of parental employment, quality of housing, neighborhoods, school and teacher quality, and health—often simultaneously give rise to similar inequalities in other educational goods. Moreover, inequalities in achievement themselves often lead directly to inequalities in these other goods. That said, we note that in some situations, a heavy emphasis on equalizing cognitive achievement might interfere with the equalization of educational goods overall. Even with a focus on achievement alone, it may not be clear who should be equal to whom because of disagreement about the normative significance of the various determinants of student achievement. One determinant is variation in the natural talents that children are born with. Many people regard any differences in student achievement associated with variation of this type as entirely acceptable. Others might see inequalities caused even by natural variation as undesirable from the point of view of equality but view them as justified overall because of other values. Differences in natural ability are substantial, so measures to equalize are likely to be very expensive or involve reducing the achievement levels of higher-achieving students. In that case, the cost in terms of the total amount of educational goods produced, or even in terms of the overall flourishing of the less advantaged, could outweigh the goal of trying to eliminate inequalities due to differences in natural talents.9 Those who do not object to achievement inequalities due to natural differences tend nonetheless to worry about the extent to which children’s development and exercise of their innate capacities may be a func-

Achievement as an Educational Good

55

tion of other characteristics, such as their race, gender, or family’s socioeconomic position. We discuss the factors that may make some gaps more normatively salient than others later. Here we use the achievement gap between children from low-income and higher-income families, which is widely regarded as normatively troubling, to illustrate the policy implications of a particular specification of the equality principle. Students from higher-income families achieve, on average, at higher levels than their counterparts from less economically advantaged families. The reasons for this pattern are well understood: children from higher-income families tend to be exposed to more books, a wider vocabulary, and generally to cognitively richer environments from a very early age. They are less likely to come to school with untreated health problems. Their parents are better able to work the school system to their children’s advantage through their choice of school and teacher assignment and can provide a richer array of after-school and summer experiences. A variety of policy interventions could potentially reduce inequalities of this type. To the extent that current disparities in schooling are detrimental to economically disadvantaged students, one might require at a minimum that all children have access to equal quality schools and teachers. In that case, at least schools would not exacerbate the variation in achievement associated with home environments. In fact, though, full equality would require far more: schools and classrooms serving lowperforming students from disadvantaged backgrounds would have to be of higher quality in order to compensate for the learning challenges that such children typically bring to school. In particular, such schools might well need more—and higher- quality—teachers as well as more nurses and counselors and other adults who could connect needy children and their families to social and health services that middle- class families take for granted. Full equality may also require a variety of public interventions outside the traditional school sector for children from disadvantaged families, including, for example, early childhood and preschool programs as well as after-school and summer enrichment programs. The full equalization of educational goods might further involve restrictions on how parents contribute to the educational goods of their children (e.g., by sending their children to private schools or spending money for enrichment activities) and other public interventions into family life. Thus, equality might demand extensive intervention into the family lives of both high-

56

Chapter 3

achieving and low-achieving students, some of which would undoubtedly interfere with what we have called the independent value of parental interests. That said, the difficulty of achieving full equality is not an argument against efforts to move in the direction of less inequality. Benefiting the Less Advantaged The third distributive value avoids the need for a threshold as required by the adequacy standard by asserting that decisions should increase the prospects for flourishing of those who are worse off. Their prospects should be increased quite generally without regard to any particular specification of what counts as “adequate.” It also avoids the problem that, strictly understood, equality is concerned only with reducing gaps between people and not at all with their absolute level. Taken on its own, equalizing educational goods could imply “leveling down”—making some people worse off even if none are made better off. Benefiting the less advantaged, on the other hand, calls only for bringing benefit to those who have less. In chapter 2 we advocated this principle with respect to flourishing rather than to educational goods, much less to a specific educational good. Suppose for a moment, though, that we could regard a person’s educational achievement as a decent proxy for her flourishing as an adult. The principle would then assess a policy intervention by considering how likely it is to shift the left tail of the achievement distribution to the right even if it also reduces the total amount of achievement or increases the gap between those with lower and higher achievement levels (or both). As we pointed out in chapter 2, it may sometimes be that over time, the worse off people in the society will be better off if the society is more productive, and that greater productivity requires technological innovations that will be created only by those with high levels of cognitive ability. In that case, resources might sensibly be invested in the highly able even if that increases inequality in achievement levels or reduces the test scores of the less advantaged. It might be possible to raise achievement at the bottom end of the distribution in the short term by putting the marginal resources there, but taking a longer-term perspective, the value of benefiting the less advantaged would in that case direct policy makers to allocate resources at the top. It is, of course, an entirely empirical question whether this rationale holds at any particular time. It will depend, for example, on whether the social and economic structures effectively diffuse the benefits of increased productivity to those at the bottom.

Achievement as an Educational Good

57

Achievement Gaps Achievement gaps between some groups of students—such as black and white students—receive a tremendous amount of attention in US policy debates. Depending on how the gaps are measured, efforts to reduce them could be consistent with the distributive goals of either equality or adequacy. In some cases, however, what is really at stake may be the low achievement level of members of the lower-performing group rather than the size of the gap between the average achievement of the two groups. Here the relevant distributive value may be adequacy or benefiting the less advantaged. The most straightforward measure of an achievement gap is the difference between the average performance of one group of children, call them group A, and the average performance of another group, call them group B. Figure 3.5 illustrates a hypothetical gap between group A and group B. Even if the members of group B outperform the members of group A on average—thereby generating an achievement gap that favors group B—some members of group A may well do better than some members of group B because the distributions of the two groups are likely to overlap. So the focus on group averages can be misleading in so far as it attributes to individuals the average characteristics of the group. Normatively speaking it is the well-being of individuals that matter—

Figure 3.5. Achievement gap between two groups. The gap is the difference between Ā B and Ā A.

58

Chapter 3

their membership in groups is relevant only in so far as the group characteristics affect them as individuals.10 Group characteristics may influence an individual’s well-being in different ways. Knowing that black students tend to achieve at lower levels on average than white students directs policy makers toward the causal factors that influence the achievement and subsequent flourishing of different individuals and possibly toward policy interventions to address those factors. Perhaps, for example, it is the fact that black students are typically economically worse off than white students that explains—or at least helps to explain—the blackwhite achievement gap. But being perceived as, or conceiving of oneself as, a member of a group defi ned in, say, racial terms may itself affect an individual’s achievement if it affects others’ expectations or one’s expectations of oneself. Group-based stereotypes can also affect individual flourishing in more subtle ways: the very fact that black students generally tend to achieve at lower levels than white students can adversely affect the lives even of those black students who perform well. Any comparisons of achievement between the two groups need to be measured on the same scale. In the United States, for example, scale scores in math and reading on the National Assessment of Educational Progress (NAEP), commonly referred to as the nation’s report card, are reported every two years for representative samples of fourth and eighth graders in every state (and now in twenty large school districts). These permit comparisons across groups of students at one point in time and over time. When test scores are not measured or reported on the same scale, the typical approach is to make the scores comparable by standardizing the units of each one to have an average of zero and a standard deviation of one. If both distributions have the bell shape as shown in the figure, a difference of one standard deviation in the means indicates that the achievement level of the typical person in group A is below that of about 84 percent of the members of group B (50 percent plus 34 percent of the entire distribution). An alternative approach to measuring achievement gaps is to focus on differences across groups in the percentage of students who meet some threshold, often referred to as proficiency. On this measure, it is not relevant to what extent some children exceed and others fall short of the proficiency standard. Although potentially appealing in the way it combines equality and adequacy considerations, this approach is problematic because the gap between groups is heavily dependent on the

Achievement as an Educational Good

59

level of the proficiency standard. The higher the standard, the larger the gaps are likely to be. If the standard is set sufficiently low, then the gap may disappear. This means that policy makers can manipulate the magnitude of the gap by changing the proficiency standard. Nonetheless, measuring gaps in this way is common practice in the United States, where one often hears statements such as X percent of one group of students meet the proficiency standard in contrast to only Y percent of another group. Not all achievement gaps between groups of students matter. A gap between fourth and fi fth grade students, for example, would have no normative significance. Indeed, in this case, provided achievement was measured on the same scale for both groups, the absence of an achievement gap would be disturbing since we expect children to perform at higher levels as they mature and progress through school. Gaps between children of different ages are not problematic because they do not have implications for the distribution of overall flourishing. Similarly, as we noted earlier, if group B were composed of children with IQs above 90 and group A of children with IQs below 90, many would not consider gaps in the observed performance between the two groups normatively problematic. Here the gaps do indeed have implications for inequalities in flourishing but, on the assumption that IQ accurately measures children’s natural ability, some regard such gaps as unobjectionable. Quite different are achievement gaps between groups of children defi ned by their race, gender, their parents’ socioeconomic position, and possibly other characteristics as well. These gaps are generally regarded as troubling because they suggest that children’s prospects for flourishing are influenced by factors that should be irrelevant. The underlying assumption is that, all else being equal, the achievement distributions should be the same for these different groups (blacks and whites, boys and girls, high income and low income). The gaps may indicate distributive failures—in terms of adequacy, equality, or benefiting the less advantaged—but they may also affect how members of different groups relate to one another and so have implications for people’s capacity to treat one another as equals. History is relevant here not because a legacy of mistreatment calls for restitution but because of its impact on social relationships here and now.

60

Chapter 3

The Black-White Achievement Gap One of the most vexing policy issues for US policy makers has been the persistence of large achievement gaps between black and white children. Figure 3.6 shows the average test scores of thirteen-year- old black and white students and the gap between the two groups on the NAEP in reading.11 Average scale scores are on the vertical axis. The lower line shows the scores for black students for each year that a particular test was administered, the upper line the test scores for white students. Reading scores for black thirteen-year- olds increased by 25 scale points during the full period (from 222 in 1971 to 247 in 2012), which is far more than the 9-point gain for white students (whose scores increased from 261 to 270). As a result, the black-white achievement gap in reading fell from 39 to 23 points during the period. Much of the narrowing of the gap occurred during the 1970s and 1980s, however, and, after a period of widening may only recently have begun to narrow again. The pattern is quite similar for math achievement (not shown).12 The recent trends indicate that the gaps continue to be large and persistent, but the early trends indicate that they can be narrowed.

Figure 3.6. Trends in average reading scores for thirteen-year- old black and white students. (NAEP scale scores). Source: U.S. Department of Education, Institute of Educational Statistics, National Assessment of Educational Progress Long Term Reading. Black refers to African Americans; scale scores range from 0 to 500; starting in 2004, the assessment format was changed.

Achievement as an Educational Good

61

Some of the reasons for such gaps are clear. They include the long history in the United States of slavery, policies of racial separation, and segregated schooling, with schools for black children typically being far inferior to those for white children. Despite the reduction in school segregation in the late 1960s and early 1970s that eventually followed the Supreme Court’s 1954 Brown v. Board of Education ruling, black students’ schooling is often inferior to that of white students. Moreover, the achievement of black children is impeded by the disproportionately low income levels of their families, many of which are headed by single parents. One obvious normative concern is that racial achievement gaps lead to differences in wages and hence to inequality in the potential for members of the two groups to flourish economically. Interestingly, writing about the black-white achievement gap in the early 1970s, the well-known sociologist Christopher Jencks suggested that closing the black-white achievement gap at that time would not do much to close the wage gap, which would be maintained instead by racial discrimination. The argument for reducing the black-white achievement gap thus appealed mainly to the value of equalizing other capacities that contribute to flourishing, such as democratic competence, healthy personal relationships, and personal fulfi llment. Those other considerations still have force, but, later, in the early 1990s, Jencks suggested that, while racial discrimination in the labor market undoubtedly still existed, reducing the achievement gap could in fact make an important contribution to reducing the corresponding wage gap.13 That conclusion is consistent with more recent evidence. Based on the same longitudinal analysis of AFQT scores referred to earlier in this chapter, researchers have found that differences in the test performance of blacks and whites accounted for about three- quarters of the wage gap between those groups when they were in their twenties, and about twothirds when they were in their 30s.14 The rest of the black-white wage gap is attributable to a variety of other factors, including employer discrimination. The data are still consistent with the presence of discriminatory behavior in that black workers typically receive lower wages than their white counterparts with the same cognitive ability, and applicants with otherwise identical resumes receive fewer call backs if their names are more typical of blacks than of whites.15 The observation that so much of the black-white wage gap appears to be attributable to differences in cognitive achievement between black and white workers, however, high-

62

Chapter 3

lights the importance of the achievement gap. In particular, it implies that reducing or eliminating that gap while students are in school could go a long way to reducing corresponding wage differentials once they are in the labor market. The two other distributive principles, adequacy and benefiting the less advantaged group, are also relevant. Regardless of the magnitude of the gap, adequacy considerations would argue for policy interventions to raise achievement levels for black students if they are currently achieving at levels deemed inadequate. If there are gaps in the proportion of black and white students achieving the chosen adequacy threshold, then these are the gaps the adequacy principle addresses. The value of benefiting the less advantaged would also justify efforts to raise the achievement of black students in so far as black students are likely to be less advantaged than white students. If one’s absolute level of achievement translated directly into one’s absolute level of flourishing, then such efforts would be justified even if the measures taken made no difference to the size of the achievement gap—or even if it made the gap wider. The issue is complicated, however, by the fact that achievement is partly a positional good. The achievement levels of black students could improve, but if those of whites remain, on average, superior, then black students may continue to suffer from competitive disadvantage in the labor market or fi nd it hard to function effectively in the political arena. The positional aspect of achievement yields one reason to care about the achievement gap between black and white students rather than focusing entirely on the level and trends in achievement of black students. Another reason is that race has a particular salience, and historical significance, that makes this specific group-based inequality in achievement particularly problematic. It is relatively easy to identify students as black or white, and the history of race relations in the United States, particularly as that has affected schooling, gives inequalities between races distinctive significance. Together, these considerations bring into focus one of the competencies we described in chapter 1, namely, the capacity to treat others as equals. The racial achievement gap brings with it the serious risk that white students will view black students, and black students will regard themselves, as inferior simply on the basis of their color. Eliminating or reducing the achievement gap between black and white students will leave differences between individual children, of course, but such differences do not raise the same concerns as those raised by systematic differences across groups defi ned by their race.

Achievement as an Educational Good

63

Notice that the size of the gap may matter in different and complicated ways depending on the value at stake. From the point of view of equality, for example, the size of the inequality might seem directly related to the extent of the normative problem, with bigger gaps between group averages straightforwardly, and proportionately, worse than smaller ones. It is important to remember, however, that the relationship between achievement and well-being is not linear. If achievement has a strong positional aspect, then even small gaps in achievement may lead to big inequalities in the metric that is of ultimate significance. The size of the gap is likely to bear also on the capacity to treat others as equals. The bigger the gap, the more likely members of a group are to perceive themselves and others as superior or inferior. This is partly because smaller gaps between group averages imply more overlap between members (recall fig. 3.5), so that the groups will be less well defi ned as higher and lower achieving. Achievement Gaps by Family Socioeconomic Status In the US context, a substantial body of research has demonstrated that children from disadvantaged households perform less well in school on average than those from more advantaged households. This empirical relationship shows up in studies using observations at the level of the individual student, school, district, state, and country and with different measures of family socioeconomic status (SES): income-related measures such as family income or poverty; education level of the parents, particularly of the mother; and, in some contexts, parents’ occupation type or employment status. Studies based on US administrative data often measure SES quite crudely, using eligibility for free and reducedprice lunch, for example, as a proxy for low family income. Research using longitudinal surveys usually includes richer measures of family background. Regardless of the measures used and the sophistication of the methods, similar patterns emerge. Achievement gaps between disadvantaged and more advantaged children are not unique to the United States. They exist even in countries such as Finland, Canada, and South Korea, whose children on average perform extremely well on international tests—and far better on average than children in the United States. Figure 3.7 illustrates this. On the horizontal axis is a measure of a family’s economic, social, and cultural status (ESCS), a measure that was designed by the Organi-

64

Chapter 3

Figure 3.7. Program of International Student Assessment (PISA) reading scores (2009) by economic, social, and cultural status percentile for fourteen countries. This graph was initially published in Ladd (2012).

zation of Economic Co- operation and Development (OECD) to be comparable across countries. Thus, a child who is from a low ESCS family in one country is comparable to a child who is from a low ESCS family in another country. On the vertical axis are the Program of International Student Assessment (PISA) test scores of fi fteen-year- olds grouped by ESCS category for the United States and the thirteen countries whose students outperformed the United States.16 The fi rst point to emerge from the figure is the upward gradient of test scores across categories of family disadvantage. Average test scores for students in the fi fth percentile across all the countries are about 350, far below the average of about 660 for students in the ninety-fi fth percentile, and the test scores rise monotonically both overall and within each country. The second point is that a close look at the children in the bottom 5 percent of the ESCS distribution shows that some countries do better than the United States with their disadvantaged students. Finland and South Korea appear to have the most success with their very low ESCS students. This relative success reflects factors such as strong institutional commitments to education and equal opportunity and far fewer disad-

Achievement as an Educational Good

65

vantaged children growing up in communities where disadvantage is highly concentrated. But even Finland and South Korea have substantial differences between students from low and high ESCS families.17 The performance of US students follows the same pattern as the other thirteen countries (see the bars at the far right in each set). Notably, however, US students in families with ESCS below the median perform particularly badly relative to their low ESCS peers in other countries, while US students from more advantaged backgrounds perform reasonably well by international standards. That is, the largest shortfalls in performance among US students are concentrated among those with relatively low ESCS. This suggests there is room for the United States to do better by its disadvantaged students. The importance of doing so is reinforced by the observation that the United States has a far greater proportion of disadvantaged students than many other developed countries. For example, the percentage of students living in low ESCS families (defi ned by the OECD as those more than one standard deviation below the mean) in the United States is more than 2.5 times that in Finland and Canada and 50 percent more than in the Netherlands. The same pattern is even starker if we shift the focus away from the OECD’s absolute measure of disadvantage to a country-specific measure of poverty, such as the proportion of students who live in households with income less than 50 percent of the country’s median income. By this measure, more than one in five children in the United States live in poverty, far more than the one in twenty-five in Finland, one in nine in the Netherlands, and one in seven in Canada. Many of the normative questions raised by these gaps between children of unequal socioeconomic status are similar to those raised by the black-white achievement gap. One difference, however, is that there is not the same history of group-based assumptions about superiority and inferiority, and membership of different groups is less visible and salient in everyday life. This means that the gaps do not have the same significance for the educational good of treating others as equals. Gaps in achievement between children raised in families of unequal socioeconomic status still offend against a principle of distributive equality, but their implications for social relationships may be different. Partly for this reason, normative attention is more likely to focus on the level and trends in achievement of the most disadvantaged students rather than on the existence or magnitude of the gap between students from disadvantaged and advantaged homes.

66

Chapter 3

Suppose that policies could be designed to increase the achievement of economically disadvantaged children with no adverse effects on that of more advantaged students. Not only would there be more achievement overall—which could be expected to lead to an increase in overall productivity and, ultimately, flourishing—but the immediate gain would accrue to those at the bottom of the distribution. Still, individuals’ achievement gains do not translate directly into improvements in their overall flourishing both because achievement is only one kind of educational good, which may confl ict with others, and because of its positional aspect. Those considerations aside, both a concern for the total amount of achievement (and subsequent flourishing) and a concern to benefit the less advantaged would support such a policy, as would adequacy if the improvement at the bottom took children to the adequacy threshold. The situation is less clear if the policies necessary to raise achievement of the disadvantaged students come at the cost of lower achievement for more advantaged students. That might be the case, for example, if the policies operate primarily through the schools and within a fi xed resource constraint. Additional resources for disadvantaged students— whether in the form of higher- quality teachers, more teachers, more support staff at the school level, or more attention by the teacher at the classroom level—would then mean fewer resources for the more advantaged students. Even in this case, both the distributive principles of adequacy and benefiting the less advantaged could justify the action. With respect to the latter principle, however, policy makers would also need to consider how the decline in resources for the more advantaged students would affect the achievement, and ultimately the flourishing, of disadvantaged students in both the short and long run. For example, if higherincome families responded to the shift of resources by taking their children out of public schools, those families and perhaps later their own grown children as well may reduce their political support for the public school system to the detriment of all children remaining in the public schools, including the disadvantaged. Readers committed to an equality principle would want to go further and equalize achievement levels across groups of students defi ned by their socioeconomic disadvantage. For them, gaps of the type depicted in figure 3.5 matter rather than simply the achievement levels of the less advantaged students. The difficulty in this case is that equalization across groups could require bringing down achievement at the top end of the distribution. That, in turn, would have negative effects on eco-

Achievement as an Educational Good

67

nomic productivity, particularly if we assume that the higher-achieving students from advantaged families are poised to become particularly productive and innovative contributors. To the extent that a more productive economy would lead not only to greater overall flourishing but also to an improvement in the flourishing of the least advantaged—an outcome that would depend on a variety of other decisions about how the extra product is distributed—the principle of benefiting the less advantaged would support permitting the inequalities.

Conclusion In this chapter we have begun to illustrate the approach presented in the previous two chapters by addressing the normative issues at stake when considering achievement levels and gaps. One needs to have in mind the full range of educational goods and to be clear about how particular gaps—gaps between particular groups and with respect to particular indicators—relate to that full range of goods. Care is also needed when thinking about the different distributive values relevant to the assessment of gaps and to the measures that might be taken to address them. It is important also to take heed of a range of independent values—things that matter but are not themselves educational goods—such as childhood goods and parents’ interests. There are several reasons why policy makers and social science researchers tend to focus so much attention on cognitive skills of the kind indicated by test scores as measures of achievement. One is that, as we have documented, those skills demonstrably contribute to individual flourishing, both through the labor market and more directly, by making it possible for individuals to live richer lives—partly by opening up opportunities for higher education and the acquisition of further accomplishments. Another reason is that the current state of technology enables policy makers to measure student achievement, or at least some aspects of it, using tests or other forms of assessment that can readily be implemented on a broad scale, something that is far more difficult for other educational goods. Finally, student achievement is directly susceptible to schooling, which is the major policy lever for influencing the production of educational goods. Indeed, many people believe that the main purpose of formal schooling is to raise achievement defi ned in this narrow way.

68

Chapter 3

While understandable, such an approach carries real dangers. Heavy emphasis on cognitive skills may undermine focus on other abilities and dispositions that also contribute to flourishing. These other skills may be important in their own right, as we highlighted earlier in the context of the GED program, or because of their contribution to a student’s ability to develop cognitive skills. Further, too narrow a focus may lead to misleading conclusions about the effectiveness of policy interventions. Studies of the Perry Preschool Project illustrate the potential problem. This famous program, initiated in Ypsilanti Michigan in 1962, continues to receive attention in public discussion about early childhood education largely because it was a high- quality model program carefully designed as an experiment: poor families were fi rst recruited into a study before being randomly assigned either to the treatment group of children, who had access to the program, or to the control group with no access. The fi fty- eight treated children in the Perry Preschool Project attended a half- day preschool every weekday and had weekly ninety-minute home visits for eight months of the year for two years. Both groups received intensive follow-up over time. Although the treatment children performed significantly better on cognitive tests for a year or two relative to the children in the control group, the test score effects faded out by third grade. However, the program clearly did something positive for the children because the treated group exhibited greater flourishing as adults many years later. Specifically, as of age forty, the treated group in the Perry Preschool had higher lifetime earnings, greater educational attainment, and far less involvement with crime than the control group. To judge the program only by its impact on cognitive skills would be to miss its fostering of capacities important for flourishing.18 Because standardized tests are imperfect instruments for measuring the capacities needed for flourishing, it may not always be desirable to administer tests of cognitive achievement to all students even if it is technically possible. Testing is costly, and collecting achievement measures may mean forgoing other educational resources. Budgetary constraints may also lead to narrower tests than would be appropriate given the ultimate goal of flourishing. Even within the realm of cognitive skills, some are more readily testable than others. If the tests only reflect the most easily measured dimensions, then they may bias the conclusions drawn from the scores even more than a well- designed and implemented exam. Moreover, there are real costs to the use of test scores when they are used for high-stakes decisions about schools or teachers. They can create

Achievement as an Educational Good

69

incentives for teachers to narrow the curriculum and game the system through teaching to the test and other mechanisms, including, in some well- documented cases, cheating. We have emphasized potential trade- offs between increasing achievement overall, raising the achievement of those at the bottom of the distribution, and closing achievement gaps. To properly assess the weight of those different considerations, one must always have an eye on the fact that student achievement is not the thing that really matters. This conclusion would be true even if achievement were measured in a richer and fuller way. What matters is people’s overall flourishing. It is in terms of their contribution to that, and to its distribution, that achievement levels and gaps must be evaluated. We have indicated some ways in which the conversion of achievement into flourishing makes these judgments complex. The positional aspects of achievement that arise in the context of some gaps between groups are relevant here, as are the implications for the capacity to treat others as equals. We have not offered any judgments about how to balance the various considerations that would arise in any evaluation of or policy response to achievement gaps. Indeed, in this chapter we have not been concerned with policy or decisions at all. The aim has been simply to illustrate in more concrete fashion some of the values that we have identified and show how they relate to a particularly salient concern for education decision makers. In contrast, in the chapters that make up part 2 we engage directly with policy questions and how to decide them. In chapter 4 we outline a method for integrating values and evidence and discuss some difficulties in interpreting evidence. In the subsequent chapters we illustrate the method by applying it to three fundamental issues that must be addressed in the design of any school system. In chapter  5 we look at the funding of schools; in chapter 6 we consider decisions about who should be accountable to whom and for what; and in chapter 7 we address the distribution of authority among the various actors in the system with particular attention to parental choice of schools and school autonomy. We hope that the three examples will convince the reader that the method and normative concepts are applicable to a wide range of educational decisions.

part 2 Decision Making

chapter four

Combining Values and Evidence

I

n this second part of the book we explore three major issues in education policy—school fi nancing, school accountability, and school autonomy and parental choice. We do not offer policy prescriptions or overall judgments about whether particular policies are good or bad. Our aim, instead, is to illustrate a particular method for combining value judgments with empirical evidence to arrive at morally responsible decisions.1 In chapters 1 and 2 we identified the values likely to be salient for a wide range of decisions about education. The four-step method we outline here would also be appropriate for decision making in other areas of policy and practice—such as those concerning health, international aid, welfare, or transportation—although one would fi rst have to identify a parallel set of relevant values. Of course, judgments will always need to be made about how to allocate resources among different policy areas and about how policies in different areas—such as health and education—might interact. The policy issues discussed in the coming chapters all concern schools. Yet, as noted in chapter 1, other factors—such as the family, the neighborhood, the media, and religious institutions—also develop educational goods. We focus on schools because over the past two hundred years they have taken on increasing responsibility for children’s development. Although parents retain considerable authority over their children’s upbringing for the fi rst three to five years, in most developed economies children are required to attend school for at least nine years, often more. For the majority of children, their school years exceed the compulsory period: many attend preschools, and they often spend additional years in secondary school and various forms of postsecondary

74

Chapter Four

schooling. Thus schools are important contributors to children’s development, and they are the primary policy lever that governments use to affect children’s educational opportunities. Government decision makers can typically have considerable direct influence over what happens in schools but less over other domains (such as the family) that also have powerful effects. This means that they cannot fully determine educational opportunities for students. They have to take many of the parameters affecting the creation and distribution of educational goods as fi xed—and in some cases as challenges to be addressed by schooling. What happens in a school turns partly on what happens outside it. Children come to school less or more ready to learn depending on their other experiences, and various factors in the home environment of lowincome children may interact with schooling to impede the development of educational goods. 2 Such children face challenges such as limited access to preventative medical and dental care for health problems that leave them short of adequate sleep or exercise, unable to see what the teacher is writing on the board, or cause them to miss school altogether. They are more likely than other children to have been born with low birth weights and to have been exposed to lead or other contaminants that may diminish their cognitive ability or lead to behavioral challenges that make learning more difficult. Moreover, family poverty can lead to unstable living environments that may force them to switch schools more often and so lose continuity of instruction. But that is not all. As Rothstein says, Poor children are, in general, not read to aloud as often or exposed to complex language and large vocabularies. Their parents have low-wage jobs and are more frequently laid off, causing family stress and more arbitrary discipline. The neighborhoods through which these children walk to school and in which they play have more crime and drugs and fewer adult role models with professional careers. Such children are more often in single-parent families and so get less adult attention. They have fewer cross- country trips, visits to museums and zoos, music or dance lessons, and organized sports leagues to develop their ambition, cultural awareness, and self- confidence. 3

Duncan and Murnane, in their study of programs that have shown some success in improving educational outcomes for poor children, include the New Hope program in Milwaukee in the 1990s. New Hope provided

Combining Values and Evidence

75

an earnings supplement for families in which a parent was working at least thirty hours a week as well as subsidies for child care and health insurance. Without any intervention in the schools, after two years boys in those families ranked 33 points higher in SAT-type tests than untreated families, and girls ranked 12 points higher. Subsequently, a similar program, the Minnesota Family Investment Program, had comparable results.4 The implication of Rothstein’s comment and the New Hope evidence is that a complete analysis of decision making about education would include consideration of policy around health care, housing, labor-market regulation, the criminal justice system, welfare, even marriage. 5 Still, children spend a substantial amount of their waking time in school. Indeed, simply by adjusting the amount of time children spend in school— for example, by allowing or requiring longer school days or more days per year—policy makers can change the relative contribution of family, neighborhood, and school to a child’s education. In this way, and many others, they can influence the creation and distribution of educational goods.

Four Steps We want the framework presented in this chapter to be relevant for decision makers at all levels, from central government through state, provincial, county, or district levels right down to the school and classroom. Obviously, what options are available at each level is affected by decisions made at the other levels. Central or state governments may set standards and create budgets, for example, that constrain what districts and schools can do. The capacity in schools similarly limits a district’s options: it takes time to build new capacities, and in the short term a district may not be able to pursue an initiative that personnel are not yet competent to carry out. But wherever decision makers are located, they will need an understanding of the valued outcomes at stake and the relevant evidence if they are to make good choices among policy options. Yet even with a clear understanding of the implications of options available to them, decision makers are often unable to implement the choice they would most prefer. Decisions about schools are generally made as part of complicated policy-making processes that involve compromise between agents possessing different degrees of influence,

76

Chapter Four

making different judgments, and, sometimes, having different interests at stake. Even within schools, decisions are often the outcome of a collective process. As noted in the introduction, in this book we do not address the complexities raised by the collective nature of much decision making about education. We aim only to help readers gain a clear understanding of how they as individuals can coherently bring together values and evidence in making their own contribution to whatever decisionmaking processes they are involved in. We summarize our basic model of how decision makers can combine values and evidence in the following schema: 1. Identify the main values in play 2. Identify the key decisions relevant to those values 3. Assess the options in light of the values and evidence 4. Establish what is the best policy overall in the circumstances

The initial headings may suggest that the procedure is a linear sequence—do 1 before moving on to 2, and so on—but our fuller description of them will make clear that in practice they interact. One should think of them as distinct elements in the process rather than a sequence to be followed. Now we outline each of these steps in more detail. 1. Identify the Main Values in Play We suggested earlier that some scholars and decision makers may be reluctant to think seriously about values because they are skeptical about the possibility of objectivity or rigor. Others may avoid this part of the process because it can quickly seem so complicated that it is just not worth the effort. In chapters 1 and 2, however, we attempted systematically to set out the aims of education and identified a clear and manageable list of the values most relevant to the assessment of education policy. We began with educational goods, which are the capacities in individuals that enable them to live flourishing lives and to contribute to the flourishing of others. Educational goods can be categorized into various capacities: to be economically productive, to make autonomous judgments about how to live, to contribute well to the democratic process, to form rewarding personal relationships, to treat others as equals, and to obtain personal fulfi llment. While these educational goods inhere in

Combining Values and Evidence

77

an individual, they increase the likelihood of flourishing for both the individual and others. We are all better off if children are raised to be law-abiding, tolerant, and productive fellow citizens capable of fulfi lling personal relationships and of relating to each other on terms of equal respect. It would be convenient if a single set of policies tended to produce the maximum amount of each of these different capacities, both for those  being educated and for those with whom they will interact. In practice decision makers must make trade- offs. Different policies will promote different aspects of well-being to different extents and, no less importantly, different policies will differently affect the well-being of different people. So it is not only the various educational goods that are “in play.” Different distributive values—equality, adequacy, and concern for the worse off—may also be at stake. Finally, decision makers must consider the other values that, though not directly concerned with the production or distribution of educational goods, act as constraints on their production or distribution. These include considerations such as the goods intrinsic to childhood, parents’ interest in making choices about their children’s education, respect for democratic processes, freedom of residence and of occupational choice, and other goods that consume resources such as housing, entertainment, and health care. We deliberately use the loose phrase “in play” because this fi rst step involves a variety of different elements. At its most abstract, the task is to identify the range of values relevant to education. That is what we did in chapters 1 and 2. But decision makers need more than a general list. They need a sense of which of those values in particular the policy proposals should be directed toward. They may be worried about low educational standards overall, about the low achievement of those at the bottom, or about the gap between them and those at the top. They may be concerned that schools are focusing on the wrong things or that parents do not have enough (or have too much) influence over their children’s schooling. Policy makers should, of course, try to keep all values in mind and be sensitive to the full range of trade- offs between them. In practice their analysis is likely to be prompted by a sense that one or two things in particular are the ones that really matter or the ones where action is most urgent. Perhaps that sense will be rather vague initially. Careful thinking about the full range of values may even reveal that the initial concern was not what really matters after all. Nonetheless, one

78

Chapter Four

has to start with the feeling that things should be better and a sense of where action is particularly needed. 2. Identify the Key Decisions Relevant to Those Values Suppose one starts by being troubled by the standard of basic numeracy and literacy among the lowest achievers. The analysis of educational goods we have offered in chapter 1 should help one understand more precisely the nature of the concern. It will almost certainly have something to do with the way numeracy and literacy skills relate to people’s well-being as adults and their capacity to contribute to the flourishing of others. It will probably also reflect the way those skills are distributed across the population (and the consequent distribution of flourishing). Whatever one judges to be really worrisome, the next task is to identify some key decisions that look most relevant to the values at stake. Which policies plausibly address whatever now seems to be the problem? As we have said, our method is intended to help decision makers at all levels, from those devising national policies that will affect every school in the country to those in charge of individual schools and even classrooms. They might be trying to improve the numeracy and literacy skills of the lowest-achieving children in the school or class. They might be thinking about more systemic attempts to raise standards at the bottom by changing the way children are allocated to different schools or by reforming the way schools are fi nanced. Which decisions are relevant to the values clearly depends on what exactly the decision maker is trying to achieve and what decision-making options are actually available. In the next few chapters, we look at large-scale policy arenas, where what is at stake is the way schooling is organized and fi nanced at the macrolevel. But there are important local—within- district, within-school, and even within- classroom—decisions to which our method also applies. While some decision makers will be constructing an education platform at the national level, others will be choosing among different curricula or working out how best to spend time with their class or with individual children. All decision makers, whether in education or elsewhere, and however close to or distant from the impact of their decision, identify a space for action, either implicitly or explicitly. Our procedure demands explicitness: one must figure out what range of actions are feasible and choose

Combining Values and Evidence

79

from within that set.6 But of course one will be interested specifically in those actions, among the feasible set, that are plausibly going to make things better. The aim at this stage, then, is to identify actions that are plausible in two senses: fi rst, it must be plausible that their implementation would contribute to the values in play, and second, it must be plausible that they could actually be implemented. The fi rst kind of plausibility requires a preliminary judgment that the actions would, if put into practice, contribute to the values identified in step 1. This necessarily involves predictions about the likely effects of decisions and so requires some understanding of the relevant social mechanisms and causal processes. Some goals may be achievable in other situations but completely unrealistic here and now. Moreover, choices made in the present may well change what can be done in the longer term. While the full effects of the available options are never certain, decision makers aiming to realize particular values need to pay attention to both the direct and immediate effects of their decisions and to the dynamics that those decisions are likely to set in motion. Perhaps, for example, the use of a random-assignment process to allocate scarce spaces in oversubscribed schools could produce more socioeconomically integrated schools than would be achieved by a policy of assignment by neighborhood, since it would reduce the role of proximity to the school (and hence housing prices) in the allocation of students to schools. At the same time, however, the introduction of a lottery approach might drive more affluent parents, who have chosen their residential location already, out of the public school system altogether and into even more segregated and socially divisive private schools. If both of these effects were likely, one could not avoid hard judgments about how to balance them, and one would want as full information as possible about which dynamic might dominate. One would also need to assess the capacity of decision makers—including oneself—to “sell” the policy so as to minimize those reactions that might undermine its ability to achieve the desired goals. The second kind of plausibility requires a preliminary judgment about whether the actions are feasible in the sense that there is some prospect of them being implemented. Decision makers need to avoid two temptations. One is to overestimate their own power and consider options that they should know have little chance of being put into effect. The other is to underestimate their space for action and disregard

80

Chapter Four

options that, though ambitious, might in fact have a prospect of being adopted. Political constraints affect what is feasible. These include not only the electoral considerations confronting democratically elected politicians competing for votes but also less obvious ways in which decision makers have to keep an eye on how their decisions will be received by others. A school principal may have good reasons to introduce a change in the curriculum that she knows would be unpopular with an influential group of parents or teachers. She will then fi nd herself balancing the cost of that unpopularity against the benefit to be had from the curriculum change. Judgments about the likely effects of decisions and about which decisions it is feasible to implement are always probabilistic. This uncertainty creates room for disagreement about what should be done in terms of policy even between those who agree about how to weight all the relevant values. Often uncertainties are even greater when assessing long-term outcomes. In both the long run and short run, multiple values need consideration—from the multiple educational goods to childhood goods and other independent values. The uncertainty of the effects of choices on these values will differ among them, and evaluation of the full effects needs to consider the relative uncertainty. We warn especially against erring too much on the side of caution in identifying which actions are plausible. One should be willing to consider innovative or radical actions that may differ from those on the current policy agenda. It is not until step 3 that one gets into the business of carefully assessing how likely it is that any given proposal would in fact achieve the desired goals and at what cost in terms of other values. At this stage the aim is simply to begin to focus one’s thinking and to make the task in 3 manageable by identifying a practicable set of proposals to work with. The question here is, what actions would be worth investigating seriously as an attempt to achieve the values identified in step 1? Often some of the key decisions are obvious candidates simply because they are already on the policy agenda. If so, the main task of step 3 is to evaluate already existing proposals. All three chapters in the second part of the book take this form. But we want to leave open the possibility that the best ways forward include policies that are not currently on the table, perhaps some that are not normally considered education policies at all. Even within the existing range of policy options, it may well be that the best version is some distance from anything currently being given much attention.

Combining Values and Evidence

81

3. Assess the Options in Light of the Values and Evidence The third step is to assess, as best one can, how well each of the key decisions identified in step 2 is likely to promote the values identified in step 1 in light of the best available evidence. Perhaps, though initially attractive, more thorough investigation suggests that one option will not actually help at all or will help only with respect to some values with costs to others or is likely to produce only negative consequences. Data- driven decision making is a mantra of many recent education reform discussions. The website of the Institute of Education Sciences, the research arm of the US Department of Education, states, “By identifying what works, what doesn’t, and why, we aim to improve educational outcomes for all students, particularly those at risk of failure.” The Center for Research and Reform in Education at Johns Hopkins University conducts research into how school districts can use data- driven reform to improve student achievement. We have already made clear that decision making should not be—indeed cannot be—data driven. It must be value driven: talk of “what works” makes no sense unless one has a valueinformed goal in mind. But decision making must be data informed. Indeed, if it is not data informed, one’s values cannot guide it effectively because every decision is a shot in the dark. Even the best evidence requires careful interpretation. Consider the gold-standard source of evidence in the social sciences: the well- designed and well-implemented random- controlled trial (RCT). RCTs test interventions by choosing a population and randomly selecting some for the intervention while using the others as a control. With large enough populations, RCTs can reliably pick up an average effect. But effect sizes vary within the affected population and may well be negative for some. Suppose an RCT evaluating a math curriculum in forty sites fi nds that it has a significant positive effect on average but that for eight of the sites there is no effect and for eight more the effect is negative. Knowing that the average effect is positive is not a good enough reason to adopt it in your site. You need to know whether yours is relevantly like those in which it had a positive effect or relevantly like those where the effect was absent or negative. The RCT does not give you that information. It does not even tell you what the relevant similarities are.7 Even with the best available information about the present and the past, decisions about the future rely on judgments as well as knowledge. Even when an RCT shows that the effect of an intervention was

82

Chapter Four

positive everywhere, it still does not follow that it would produce similar results in another context, so knowledge of context and the relevant variables that make the intervention effective is important. Take, for example, the California Class- Size Reduction program. Evidence from the student-teacher achievement ratio (STAR) project in Tennessee, which used a large-scale RCT, concluded that students in smaller classes in K through 3 outperformed students in larger classes and that the effect was particularly pronounced for minority and inner- city children. California initially spent $1 billion per annum and then $1.6 billion per annum on reducing class sizes across the state. This had no discernible benefits and in particular no effect on inner- city and minority children. Among various explanations offered for the ineffectiveness of the program is the suggestion that the increased hiring in affluent districts drew more experienced teachers from poorer districts, which, in turn, had to hire inexperienced and in many cases unqualified teachers. The result was that inner- city and minority students had less stable learning environments and less skilled teachers than before the intervention. 8 To make decision making even more difficult, the most wide-ranging questions—designing systems for school fi nance or governance, choosing certification requirements for teachers, or setting up charter school laws—are often those least directly informed by empirical research.9 No randomized experiments have tested (or in many cases can test) the efficacy of policies in these areas, and even correlational, quasiexperimental studies are rare and often infeasible. Instead, decision makers tend to draw on data that are peripherally related to the decision in question, on anecdotal information, and on their own understandings about which mechanisms it is important to consider. In choosing certification requirements, for example, they may rely on information about whether teachers who have taken more math courses in college are more effective at increasing student learning. While relevant, this information can also mislead. Even if teachers who have taken more math are more effective, mandating additional courses in a certification system may bring unintended consequences. For example, some suppliers may start providing poor- quality, low- cost coursework options that meet the requirement but do not improve teacher performance. Alternatively, rather than math courses improving teacher effectiveness, it may be simply that teachers who are more effective for other reasons choose to take more math classes. In practice, then, all judgments about the evidence, like judgments about which decisions are likely to realize partic-

Combining Values and Evidence

83

ular values and which are likely to be implemented in the fi rst place, are probabilistic. Many different factors will need to be taken into consideration at this stage. Judgments are required, for example, about what agents are actually capable of in the circumstances, and about whether incentives are well calibrated to induce those actors to act as the policy requires. Policy reforms that look great on paper can flop because they overestimate what classroom teachers, or school principals, are actually competent to deliver. It is easy to misunderstand the incentive structures faced by those teachers and principals, or simply to fail to provide the resources (funding, training) that would be needed for the reform to achieve its desired effects. The key judgments at this stage then are the following: judging the quality of the available research and drawing conclusions about the likely effects of different options in one’s own situation from the best available evidence about what produces valued outcomes elsewhere. 4. Establish What Is the Best Policy Overall in the Circumstances The last step is to identify the policy that brings the greatest expected return in terms of the values identified at the fi rst stage. This involves comparing the various actions identified (in step 2) in terms of their probable effects (assessed in step 3) to see which is likely to produce the best overall outcome available in the circumstances. One can think of this, in terms familiar from decision theory, as working out which of the actions available to us in our circumstances yields the greatest expected utility—with “utility” here understood as a weighted index or combination of the values. As we have emphasized in chapters 2 and 3, there are several such values and, given the inevitability of trade- offs, these have to be balanced or weighed against each other to evaluate overall outcomes. One also needs to factor in the various probabilities of achieving those outcomes. Working out which is the best policy, all things considered, is thus a complex enterprise. One proposal might have a very good chance of bringing about modest improvements with respect to values that are only moderately weighty, while another is less likely to achieve aims that, if achieved, would be hugely valuable. A third might be very likely to improve the situation substantially with respect to one value while risking making it worse with respect to others. One cannot predict the overall

84

Chapter Four

return to any decision with much confidence, but one can try to assess which decision has the best expected outcome in light of the available evidence.

Conclusion Some readers may fi nd the method we have outlined in this chapter intimidating. We hope that in the following three chapters, seeing it applied to a range of policy domains will allay their concerns. Indeed, the four-step approach is really little more than a codification of the way in which people already go about the business of making decisions. The approach is a systematic, explicitly analytical presentation of the elements that are almost inevitably involved in the making of a considered decision. It is common sense that the aim is to improve things, to make them better (or at least avoid making them worse), and obvious that decision makers face trade- offs between different ways in which they could be better (or worse) as well as having to assess what is actually likely to happen as a result of any particular decision. Whatever decision is made will in effect reflect judgments of the kind we have set out. Our goal is mainly to bring those judgments to consciousness—to break them down into their different elements, and to show how those elements can be combined—so that decision makers can think and act with greater clarity and precision. We do not pursue our proposed approach all the way to step 4. Decision makers need to make a judgment about which policy looks best overall, but we are not decision makers. It is not our purpose to advance any particular policy agenda or advocate any particular decisions. Our aim is simply to help those responsible for making decisions to understand more clearly the elements involved in doing that well.

chapter five

School Finance

S

chool systems require both funding and a governance structure. In the next three chapters we address decisions about fi nance and governance and focus on the values that are most relevant and on the empirical evidence that helps to clarify the potential trade- offs between them. Central to our discussion is how those decisions affect the level, mix, and distribution of educational goods. In this chapter we consider funding, while the next two examine two major governance issues: holding schools accountable and decentralizing decision-making power to parents and schools. In each chapter we identify the values at stake and a few key decisions that policy makers must make. We then bring evidence to bear on how the decisions are likely to affect the relevant values. Left to their own devices, many families would invest less in their children’s educational development than would be desirable for social purposes and even for the good of their own children. Some families lack access to sufficient resources to invest and, because educational goods are not good collateral for a loan, borrowing will be difficult even if the long-run benefits of education spending far outweigh the costs. So governments fund schools in part to promote distributive goals. Even families with sufficient funds may invest too little because they primarily consider the benefits that education will generate for their own child to the exclusion of the additional benefits it can bring to others. Each student’s educational goods benefit others through a range of mechanisms including economic spillovers in the form of greater productivity, lower crime, better citizenship, and better health. Education also promotes people’s democratic competencies, their capacity to regard others as equals, and their capacity for personal relationships, all of which can

86

Chapter 5

lead to better decision making and social interactions. So governments fund schools to increase the production of these educational goods in light of their social benefits. Tax-funded free public education has become a standard in much of the world, but funding levels vary tremendously across countries and even within countries. In 2012, the average spending per student across OECD countries was approximately $8,954 per student. Figure 5.1 plots the spending per pupil and spending as a percent of gross domestic product. The United States spent $11,732 per pupil (3.5 percent of GDP), while the United Kingdom spent $10,056 per pupil (3.4 percent of GDP), Poland spent $6,764 per pupil (3.7 percent of GDP), and Chile spent $4,522 per pupil (3.2 percent of GDP). Similarly, spending on elementary and secondary schools varies mean-

Figure 5.1. Spending on elementary and secondary education across select Organization of Economic Co- operation and Development (OECD) countries (2012).

School Finance

87

Figure 5.2. Current per-pupil expenditures for public elementary and secondary education in the United States (fi scal year 2013).

ingfully across US states. Figure 5.2 shows that, for school year 2012/13, while Arizona, Idaho, Oklahoma, and Utah all spent less than $8,000 per pupil for public elementary and secondary education, Alaska, Connecticut, Massachusetts, New Jersey, New York, Vermont, and Washington, DC, all spent more than $15,000 per pupil. The average expenditure across the United States was $10,763 per pupil. The level of government that makes decisions about school funding also varies across countries. In the Netherlands, for example, the central government has almost exclusive responsibility for funding schools. In contrast, the United States is more decentralized. State governments are constitutionally responsible for school funding, and most of them delegate at least some funding discretion to local school districts. In this chapter, we take the perspective of policy makers at what we call the central level, deciding on policy for the local actors. Readers should bear in mind that the discussion applies to any two levels (e.g., national to state or province, state to local district, or local district to school).

88

Chapter 5

Identifying the Main Values In Play A primary purpose of schools is to develop educational goods—the knowledge, skills, dispositions, and attitudes needed to flourish and contribute to the flourishing of others. Funding for schools contributes directly to the overall quantity of educational goods. Although in some cases funding decisions may also influence the mix of educational goods, we focus here on the overall level and thus on the distributive questions that are at the heart of school funding decisions. Our main concern is how various funding decisions are likely to affect the distributive values we described in chapter 2: adequacy, equality, and benefiting the less advantaged. In the next two chapters we discuss policy decisions that are more clearly and deliberately linked to the mix of goods. While equality in US policy discussion has typically referred to equality of funding, we are concerned with equality of educational goods. The understanding of adequacy invoked in debates about funding is more aligned with ours, namely, funding sufficient to ensure that everyone has adequate educational goods.1 Financing plays a major role in the extent to which either adequacy or equality, however conceived, can be achieved. To provide adequate educational goods for all children, funding must be adequate. Similarly, achieving equality of educational goods (if that is possible) will require an appropriate distribution of funds across and within schools. 2 We focus primarily on the complications involved in achieving equality or adequacy of educational goods. We also briefly discuss how funding decisions can affect the extent to which the education system benefits the less advantaged. Funding decisions also affect the realization of what we have identified as independent values. More funding for schools leaves less for other goods, such as housing or entertainment. Opportunity costs are unavoidable and always force trade- offs among different values. Financing decisions can also affect the level and distribution of childhood goods. A well-funded school can afford a more aesthetically pleasing physical plant or interesting or exciting playgrounds that enhance children’s enjoyment of the school day independently of any strictly educational benefit. Someone running a school on a shoestring may have to make tradeoffs between educational benefits (such as investing in high- quality mathematics teachers) and childhood goods (such as spending on better sporting facilities).

School Finance

89

The independent value of parents’ interests with respect to their children is also in play. Differential parental investment in children outside the school system can complicate and frustrate efforts to achieve distributive values through school fi nancing. Exactly what parents should be permitted to do in terms of investing in their own children outside of school time is a matter of philosophical debate. 3 Individual decision makers may have their own views about how best to balance respect for parents’ interests against distributive values (and, indeed, against other values), but they are operating in a context where particular views about that balance prevail. Most liberal democracies have assumed that parents have a right to considerable freedom to spend additionally on their own children. They may spend their private funds on private tutoring, books and computers, and summer enrichment programs that help to generate educational goods. Parents are usually permitted to withdraw their children from the public education system altogether. Decision makers operate within this context, and of course their decisions must reflect those de facto constraints on the pursuit of distributive values. Other independent values play a similarly constraining role. Individuals are typically granted extensive freedom of choice with respect to residential location and occupation. Some liberal democracies conscript individuals into the military and therefore do not have to pay them market wages, but we know of no liberal democracy in which it would be feasible to conscript teachers. If the government could do that effectively, schooling might be less fi nancially costly overall, especially for schools that have difficulty in recruiting teachers. However, we take freedom of occupational choice as a constraint on policy. Similarly, if the government could decide where people lived, it could, for example, create socioeconomically mixed neighborhoods, which would reduce the inequality of student need between schools. Again, for the purposes of this discussion, we take adults’ free choice of residence as a constraint.4 Finally, respect for the democratic process is relevant to funding decisions. Suppose one were a policy maker who would like to allocate funds to reach equal, or even adequate, distributions of educational goods. Respect for the democratically expressed views of one’s fellow citizens might lead one to draw back from taking decisions in pursuit of that goal even if one could do so without counterproductive results. In that case, one’s decision would reflect one’s view about the relative importance of democratic respect as compared with the distributive value in question.

Chapter 5

90

In practice, however, democratic processes constrain decision makers irrespective of any judgment about their moral weight. Policy makers operate in contexts where they cannot depart too far from the democratic will without coming up against institutional constraints such as the electoral system.

Identifying the Key Decisions Relevant to Those Values Governments play a significant role in developing educational goods and attending to how they are distributed, and they do so primarily through the school system. Schools, however, are not the only policy lever with which governments can promote the development of educational goods. Investments in housing, health, early childhood development, and other social programs all contribute to educational goods. In some situations, investments such as these may be the most effective ways to pursue the values we have identified. Even if policy makers cared only about the production and distribution of educational goods, they would have to decide how much to spend on schooling and how much on those complementary areas. But of course they do not and should not care only about educational goods. Other policy arenas, such as transportation and health care, compete with education for resources and attention. So the government must decide how much to spend on education generally as well as how much to spend on schools in particular. Here we focus on decisions about school spending assuming that expenditure in other areas is given. Policy makers at the central level determine how much funding to provide and how to distribute it across local jurisdictions (or schools if the relevant local level is the school). They can also dictate the extent to which local governments are allowed to raise additional funds for schools, and they can require local governments to spend the funds they receive in specific ways. We focus here on three key decisions, with greatest emphasis on the fi rst. Differentiation of funding. Should the central policy makers provide more funds to certain local governments than to others to compensate for differences in the cost of schooling or the needs of students? If so, how much more?

Supplementation. Should local governments be allowed to supplement funds from the central government through taxes or other means?

School Finance

91

Autonomy. Should local governments have autonomy to determine how to spend the money they receive from the central government, or should the central policy makers dictate how the money is to be spent?

Differentiation of Funding Higher-level policy makers have to decide how much to spend and how to distribute the funds across lower-level units. One option is not to differentiate at all: to provide funds on a flat-rate per-pupil basis. Districts and schools differ, however, in two ways that are important for decisions about whether to differentiate. First, they cannot purchase the same resources with the same budget. Wages for teachers, secretarial workers, and maintenance staff vary across local labor markets. Real estate varies in price depending on the location of the school. Transporting children costs more in rural areas than in urban and suburban areas. Hence, a monetary unit of funding varies in the quality of educational resources it purchases. We refer to this kind of variation as variation in costs. A decision maker wanting to ensure that all lower-level units have equally valuable resources available to them would differentiate funding to compensate for costs. Second, districts and schools are not homogenous with respect to the kinds of students they educate. Some districts, and some schools, have proportionately more children from poor homes, more children with disabilities, more children from homes with low levels of parental education. Some have more students whose fi rst language is not English. To achieve the same level of educational goods per student (equality of educational goods), units with such characteristics will require more funds. We refer to this kind of variation as variation in student needs. The decision, then, is whether to differentiate funding in order to compensate for cost or for student need or for both, and, if so, how much to differentiate. The governments of most developed countries make some effort to differentiate funding to compensate, as least in part, for cost and need differences. In the United Kingdom, the funding formula includes a pupil premium for students receiving subsidized lunch (“on free school meals”) since these students come from lower-income families with fewer resources. The amount in question varies depending on the age of the child, but overall, in 2015/16, schools received around £1,000 more for each student in this category than for others. 5 From 1987 to 2006, the Netherlands had a weighted funding system for primary schools that pro-

92

Chapter 5

vided 90 percent more funding per child to the schools chosen by disadvantaged students. Here, disadvantage was defi ned by immigrant status and low education level of the parents.6 In the United States, the federal government’s Elementary and Secondary Education Act directs supplemental funds to support students from poor backgrounds, while the Individuals with Disabilities Education Act provides states with block grants and requires them to provide services for students with disabilities. As things stand, states within the United States vary enormously in how they treat student disadvantage in funding formulas. Most do not provide additional funding for students in poverty, but some do. By one estimate, in 2013, districts in Delaware with 30 percent of students in poverty received 81 percent more in state and local revenue than did districts with no students in poverty. Delaware is an outlier, but Minnesota, Utah, Ohio, New Jersey, South Dakota, Tennessee, Massachusetts, Indiana, and North Carolina all spent at least 10 percent more in these higher-poverty districts. However, at the other end of the spectrum, Nevada, North Dakota, Illinois, Maine, Missouri, Vermont, Idaho, Wyoming, and Alabama all spent at least 10 percent less on districts with 30 percent poverty than on no-poverty districts.7 Differences in the relative spending on students in poverty are unlikely to stem from true differences in student need across these states or countries. Instead they result from a political process with imperfect information as well as powerful constituencies not wanting to reallocate funding away from higher-income areas. States also differ in how they compensate for disabilities. Some States give block grants to districts for special education while others break disabilities down into categories, each with different weights. Moreover, the weights vary considerably: schools in Arizona receive nearly five times more for a student with impaired hearing than for one without; in Kentucky, the same student attracts less than 2.5 times the regular funding; in Tennessee, no disability attracts more than 1.9 times the regular funding.8 Supplementation Central policy makers need to decide not only how much to distribute to districts or schools but also whether to allow these local entities to supplement funds with local revenues raised through local taxes, fees, or voluntary contributions from parents. In the United States, some states, such as Illinois, Pennsylvania, and New Jersey, grant districts taxing au-

School Finance

93

thority to raise addition local funds for school operating expenditures, while others, such as California and Washington, do not. Some other states, such as Michigan and Texas, allow some supplementation but significantly restrict it. All states allow private contributions to local school districts, but some school districts place restrictions on the amount that individual schools may raise through private contributions. Similarly, most publically funded voucher programs place limits on the additional tuition that schools may charge parents.9 Supplementation would almost certainly increase inequality of resources because the additional funds are likely to be concentrated in districts with higher taxable wealth or in schools serving higher-income families. In so far as policy makers care only about adequacy of educational goods and provide sufficient funding for that to be achieved, they may fi nd the supplementation straightforward. However, in so far as equality matters, supplementation will be more problematic. Moreover, allowing supplementation may reduce the willingness of higher-income taxpayers to support funding from the central government as it will be cheaper for them to raise money locally for only their own schools.10 As a result, such supplementation may even confl ict with adequacy.11 Local supplementation may also generate benefits. At least in the short run, supplementation would increase overall spending and so may increase educational goods. In addition, local spending fi nanced through local taxation has the potential to improve school management because of the stake that local taxpayers have in making good use of their tax payments, again increasing the overall level of educational goods. Local supplementation may also help to alleviate some of the political obstacles to school funding that arise through the democratic process and from parents’ interest in their children. That is, allowing local supplementation may affect voters’ willingness to support a school funding level that would not otherwise have been politically feasible. Autonomy The third decision concerns how much and what kinds of discretion districts and schools should have over the spending of funds (whether those funds come from central government or from supplementation). Until recently, for example, California provided most of its funds for school districts through programs that restricted their use to specific purposes such as teacher professional development. Recent reforms in that state

94

Chapter 5

have led to greater authority at the district level to choose how to spend the dollars.12 Allowing greater local discretion will clearly lead to greater local variation in how funds are spent. This discretion may lead to better decisions if local decision makers are more knowledgeable about their needs and opportunities than central decision makers but to worse decisions if they are less knowledgeable or if they direct their efforts to inappropriate goals.

Assessing the Options in Light of the Values and Evidence Decision making requires information about the likely effects of each option, which in turn involves bringing social science evidence to bear and paying close attention to measurement issues. In the case of some funding decisions, measurement challenges play a particularly large role. Differentiation of Funding The challenge for designers of school finance systems is to identify the level of spending for each local area or school that is necessary to meet the distributive goal in question, such as adequacy, equality, or benefiting the less advantaged. This task is fraught with technical and political difficulty. We are not talking here primarily about the related challenge of determining the appropriate average level of funding for the typical local district or school. As we have noted, that decision will depend heavily on trade- offs between educational goods and other goods as well as on the relative productivity of spending on schools compared with other investments that also contribute to educational goods. We are talking rather about the decision that central policy makers face about whether to adjust funding to account for differences in costs and needs across areas and how much to do so. The methods that social science researchers have developed to help policy makers address the decision about levels partly overlap with those needed to address the decision about differentiation, but only partly.13 We start by considering together the goals of adequacy and equal distribution of educational goods. An obvious challenge to achieving either goal is that the costs of one set of crucial inputs, namely high- quality teachers and other school personnel, vary across different regional labor markets. A second barrier is that the needs of individual students

School Finance

95

vary depending on such factors as the education and income levels of their parents, whether they are native speakers, and whether they have physical or cognitive impairments that affect learning. People who become teachers or school leaders could have chosen—and until midcareer many of them still could choose—to enter a different profession. The cost of keeping them from making that choice varies depending on how well those other professions are paid locally and on educators’ perceptions of their working conditions, which may be influenced by the mix of students in the school. In addition, students vary in how well prepared they are for school and in how much and what kind of support they get for their academic and other relevant learning before attending and outside of the school. Higher wages are needed, presumably, to induce talented educators to prefer the more difficult job of ensuring that highly disadvantaged children reach the threshold for adequacy of educational goods to the less daunting job of ensuring that outcome for advantaged children. Even more would need to be spent to induce enough talented educators to educate disadvantaged children so that they enjoyed a level of educational goods equal to that achieved by more advantaged children. Given the available evidence and the technical tools available to social scientists, some cost differentials are relatively easy to adjust for provided policy makers have the will and political space to do so. For example, it is not difficult to assess reasonable transportation costs, and many states provide additional funds to geographically large districts with sparse enrollment to compensate for their extra costs. Similarly, some states, such as New Jersey, adjust their formulas for distributing state grants to districts based on geographic cost indexes to account for differences in the price of recruiting equally talented teachers across labor markets.14 Importantly, though, central policy makers would not want to adjust for differences in the actual salaries that school districts pay their teachers, because that would create incentives for districts to raise salaries simply to obtain more state funding. Not all cost differences can be measured easily, and compensating for them is not straightforward. The functioning of the school board and the level of support provided by the local community affect how well funds are used. A poorly managed district will not generate the same level of educational goods with a given sum as a well-managed district with the same population and costs. Policy makers may not want to reward poor performance even as they recognize that poorly performing schools may

96

Chapter 5

need new resources to improve. As a result, adjusting for costs for the sake of adequacy or equality is not a simple task even given the plethora of evidence about differences in cost. So far we have only talked about cost adjustments. What is the evidence about how funding would have to be differentiated in order to compensate for different student needs? Some needs—for example those arising from severe cognitive disabilities—cannot be compensated for sufficiently to ensure equal educational goods. But it might, in principle, be possible to compensate for other needs, including less extreme differences in abilities to learn. In so far as students have had different opportunities to develop educational goods outside of school, schools can compensate by giving greater opportunities to more needy students, and this costs money. How much is far from clear. A number of recent studies have used different methods and have reported a range of estimates of the differential spending needed to compensate for student poverty. Based on the cost-function approach that uses regression models to estimate costs, researchers have suggested that in New York, the funding needed to achieve specified outcome goals for children from low-income families is two to three times higher than that for more advantaged students. Based on the professional judgment panel method, estimates for California are somewhat more modest and range from 17 percent more to more than 100 percent more. Similar studies show that over and above any extra funding needed to educate poor students, roughly another 30  percent increase is needed for students who are English-language learners.15 These estimates are imprecise at best but all show that substantial funding is required to compensate for differences in need. As we have said, the US federal government and most states make some attempt to target need with differential funds. But they do not explicitly use research-based estimates, and the amounts are typically small. Moreover, as we mentioned earlier, states differ. While in 2013 Minnesota provided 33 percent more for high-poverty districts than for low-poverty districts, Illinois provided 18 percent less.16 But in neither state do children from the poorer districts come close to reaching the same level of educational attainment as children from richer districts. The situation is similar in the United Kingdom, which spends double the average amount on its disadvantaged pupils and still does not achieve close to parity. Can differentiating spending to compensate for need have much effect on the distribution of educational goods? One source of evidence is

School Finance

97

the “summer learning gap” research in the United States. With its long summer breaks, the United States gives us a rough way of estimating the different effects of being in school on children from different social backgrounds. More disadvantaged students enter kindergarten with lower educational achievement than more advantaged students. But several studies have shown that during the school year the gap does not increase much—and then, during the long summer, it increases sharply. The return to school in September does not reduce the gap, but again, the gap does not increase much if at all until the next long summer break. The inference is that, up to some limit, more schooling (which requires more funding) helps somewhat to equalize educational goods compared with what they would have been without the schooling.17 Research on the effects of US school fi nance reforms also provides evidence that funding can improve educational outcomes, fi nding that a 10 percent increase in per-pupil spending each year for all twelve years of public school leads to about 7 percent higher wages and that these effects are larger for children from low-income families.18 Even if social scientists established exactly how much additional funding would compensate for specific needs, implementing that differentiation would be practically complex. Highly resourced parents can respond to differentiation of funding by giving their children even greater opportunities outside of school or by paying more for their children to attend private schools.19 This response would, then, change the estimates of what would be needed to equalize the level of educational goods. Another challenge for policy makers is that providing more funds to schools with students who have learning-related disabilities can create an incentive for schools and districts to overestimate their need. One study of schools in Texas, for example, documents large changes in the rates of classifying students in need of special education service in response to changes in fiscal incentives to do so. 20 The more general concern is that if central policy makers provide more funds to districts or schools whose students learn at low levels in order to compensate them for greater need, the schools then have an incentive to keep their students’ scores low enough to maintain their eligibility for the additional funds. Now let us consider, separately, the distributive goal of benefiting the less advantaged. Because we understand the less advantaged in terms of their all-things- considered prospects for flourishing, it is not a straightforward matter of simply directing more funds to educating the least advantaged children. In a highly competitive economy that rewards educa-

98

Chapter 5

tion, spending more on the socially and cognitively less advantaged will benefit their longer-term prospects, other things being equal. But the prospects for flourishing for the disadvantaged depend partly on the development of the talent of other students. To illustrate, consider health. The less advantaged will experience more physical and mental health deficits over the life course than the more advantaged. They will need high- quality nurses, primary care physicians and medical care facilities, and they will benefit from improved pharmaceuticals and medical technologies. The cognitively disabled—especially those who are severely cognitively disabled—will be unable to compete in the higher end of the labor market and will depend on high- quality caregiving and technologies that improve their lives. Benefiting the less advantaged supports investment in the highly talented as long as there is reason to believe that enough of the return to that developed talent will redound to the benefit of the less advantaged, which depends in part on the structure of the economic and social institutions under which they live. The difficulty of determining exactly what distribution of funds is needed to achieve any given desired distribution of educational goods mean that policy makers cannot know for sure whether the school finance system is allocating enough funds to one group or the other. They can see, however, the results of the current funding and governance system. While some school systems produce more equal outcomes than others, in all systems, on average students from low-income families leave school with lower educational goods on all standard measures. Concern for equality and benefiting the disadvantaged supports spending considerably more on these low-income students. Whether to invest more on these students through school funding depends on how useful this funding is likely to be in improving their educational development relative to other potential investments such as early childhood care or after school programs as well as housing and health. We have focused here solely on compensating for cost and need and the likely effect of such measures on the distribution of educational goods. But, as we mentioned earlier, spending in schools can affect children’s daily lived experience—the extent to which they enjoy childhood goods. A more aesthetically pleasing environment can affect their mood, for example, which is important for reasons beyond its effects on their educational development. An intermittent smell of sewage and the sight of an automobile factory right outside the school may depress educational performance relative to the smell of freshly mown grass and the

School Finance

99

sight of a sports field, but that is not the only reason why it is worse. We do not know of any high- quality studies that show how fi nancing decisions affect the aesthetic environment or other features of school life that affect childhood goods, but decision makers concerned with those goods as well as with the production and distribution of educational goods might seek out or commission such studies. Supplementation Should local decision makers and parents be permitted to supplement the resources received from the central government with public or private funds? Two considerations support local supplementation, whether at the district or school level. One is that it is likely to lead to higher overall spending on schools. Indirect evidence of that pattern at the district level comes from California, where a well-known school fi nance court case (Serrano v. Priest ) in the late 1970s removed the right of local school districts to raise local tax revenues to supplement the funding provided by the state. Research suggests that the elimination of local supplemental funds accounted for approximately 40 percent of the fall in spending in that state between 1970 and 1990. 21 The main mechanism through which local supplementation at the district level may lead to higher overall spending is that wealthier families, who tend to be geographically concentrated, are more willing to support higher taxes on themselves when taxation is local and when they are assured that the revenues will be spent on their own children than when taxes are channeled through the state to all districts. At the school level, the use of school fees or additional tuition, if permitted, is likely to be more feasible and to raise more revenue in schools serving affluent families than in those serving poor families. 22 Local supplementation may also improve the effectiveness with which resources are used within the local community. Local control of funding levels may make local school leaders more responsive to the interest of local parents and voters. If state leaders controlled the whole budget, school personnel might be more responsive to state leaders than to local leaders and, because state leaders have more difficulty gathering information about the conditions within local schools, local personnel might fi nd it easier to shirk responsibilities. Local school boards in the United States have hiring authority for school superintendents and the associated oversight over them. If they do not control tax rates and the size of

100

Chapter 5

budgets, highly skilled residents may be deterred from seeking election to school boards. Local control of resource levels may also make it easier to respond to particular or unpredicted needs of a district. If, for example, something happens in the community that means students need more support, local leaders can respond more rapidly than state leaders by raising revenue from local taxes. Set against these potential advantages of local supplementation are the likely adverse distributive consequences. Wealthier and higherincome communities are not surprisingly more likely to spend more on schools than low-wealth areas. 23 Moreover, allowing local supplementation may reduce political support for school funding at the state level from those affluent families who, when local supplementation is allowed, may be inclined to support lower state-level and higher local-level taxation. 24 A system with local supplementation will usually result in more spending in schools with children from affluent families. Spending in states that allow local supplementation, such as Pennsylvania, is more unequal across school districts than in states that do not, such as Washington or California. Less clear, however, is how unequal school spending generated by local supplementation will affect the distribution of educational goods. Parents have considerable discretion over how much of their own resources to spend on their children. Increases in income inequality, especially at the top end of the income distribution, have been accompanied by striking increases in high-income families’ spending on enrichment for their children and increases in the gaps between the enrichment spending of high- and low-income families. 25 Between 1972 and 2006, in the United States, enrichment spending (on music lessons, athletics, private tuition, college preparation, and other similar programs) increased from $3,536 to $8,872 per child among families in the top quintile of the income distribution but only from $835 to $1,315 per child among families in the bottom quintile (in 2008 dollars). 26 When the highest-income parents cannot raise money for schools through local taxes, they can spend private resources on supplementary education. When their children encounter low- quality teaching, they can step in with fi nancial or personal resources. Restricting local supplementation may not, in fact, dramatically reduce the greater opportunities that these children have, or even reduce them at all. The restriction, however, may adversely affect the families below the very top of the income distribution more,

School Finance

101

since these families are more dependent on public services. Because middle-income families are less likely (because less able) to offset reductions in public school spending, policy makers may be more successful in using the school fi nance system to affect the distribution of educational goods between low- and middle-income children than between children of upper-income families and others. It is worth recalling that, when the relevant distributive principle is benefiting the less advantaged, what matters is neither the gap in spending nor the gap in educational goods but the prospective flourishing of the less advantaged. If local supplementation causes inequality and increases the educational goods of the more advantaged but we have good reason to suppose these additional resources will be harnessed to benefit the less advantaged, either through transfers or because the increased economic growth made possible by the enhanced human capital enables them to flourish more, then this consideration, at least, supports local supplementation. The decision involves a judgment about whether benefits to the more advantaged will in fact redound to the benefit of the less advantaged in the specific circumstances. Autonomy In addition to deciding the amount of funds for each district or school, central government policy makers must also decide how much discretion to cede to the districts or schools over how to spend the funds. Should the local officials be free to use the funds as they see fit, or should they be directed to use them in specific ways? Countries and even states make very different decisions regarding decision-making authority over resource allocation. In the United States, most funding goes to districts, and then districts choose how to allocate those funds. In turn, districts pass some of that authority to the school level. However, some states place substantial restrictions on how the money is spent. As examples, some states set a minimum class size or have a teacher salary schedule, which together determine how much money a school spends on its teachers. Similarly, some states require districts to spend funds on mentoring programs for teachers or curricular packages that schools and teachers are required to use. Until recently, schools in Mexico were not allowed to choose their teachers. Instead, the central authority assigned teachers to each school, as is still the case for many schools in China. In the United

102

Chapter 5

States, districts sometimes hire teachers and assign them to schools, although often with some input from educators at the school level. In most of the United Kingdom, individual schools hire teachers directly. Suppose that state policy makers, animated by a concern with equalizing educational goods, decide to provide additional funds to districts with high proportions of low-income children. Simply providing the additional funds does not guarantee a more equal distribution of educational goods. A district receiving more funds merely has more money and, absent directions from the central policy makers, it can spend it how and on whom it wants. A district with 100 percent low-income children cannot spend the additional funds on higher-income children, but it can spend them ineffectively. A district with, say, 50 percent low-income children might be under considerable pressure from higher-income parents, who on average are more likely to have the time and resources needed to lobby school and district officials to improve provision for (by spending more money on) their children. One way that higher-level units of government can attempt to influence who benefits from additional funds is by establishing accountability measures of the kind we discuss in chapter 6. A more common method is to require those at the local level to spend the money in a specific way. State-level funding programs in the United States are often extremely complex because they include numerous “categorical” funding streams, such as funds specifically earmarked for transportation costs for rural areas or for spending on children with special educational needs or on reducing class sizes. Districts and schools that do not spend the funding on the required purposes are in violation of the terms of the aid. Directing funds to be used for specific purposes is especially attractive to higher-level decision makers when they have reason not to trust the lower-level decision makers to base their decisions on similar values or to know how to spend funds effectively. To give a straightforward example, if state decision makers believe that, left to their own devices, districts will neglect low-income students because local politicians do not care about equality, legally requiring them to spend additional funds on those children will be an appealing measure. Similarly, if state decision makers believe that new teachers are not receiving sufficient mentoring, they may require local districts to devote some of their state funding to that purpose. They can then monitor those dollars to make sure they are spent accordingly.

School Finance

103

At the same time, specifying in too much detail how funds should be used has clear drawbacks. Local actors are better placed to understand their local context. While some districts may benefit substantially from a required mentoring program for new teachers, others may put the funds to better use. When local decision makers’ values are in line with those of the higher-level decision makers and the latter have some reasonable level of confidence in local abilities and knowledge, granting discretion to the local level can be sensible, and locals can make use of their contextual knowledge. Moreover, central decision makers are often less able to make corrections in response to inevitable mistakes in expenditure choices. The class-size reduction policy in California provides an example. High- quality research from Tennessee showed that, other things being equal, lower-income children benefit from lower class sizes in the elementary years. 27 Using this evidence, the state of California imposed a cap on class sizes. Many districts had to hire new teachers as a result. But the new teachers were not, on average, as qualified as the existing teaching force, and wealthier districts were able to hire more experienced and better- qualified teachers away from districts with higher concentrations of lower-income students. The result was that children with higher levels of educational need were taught by, on average, less qualified teachers than before the cap was introduced. 28 The specific design of the US education governance system might give the impression that respect for democracy supports allowing local communities wide latitude in educational decision making. For a long time, the practice in the United States was that districts raised most of their revenues locally, and locally elected school boards decided how to spend those funds. But even if all the funds were raised locally, respect for democracy does not require local decision making. We can set aside the fact that voting rates in local school board elections are lower than voting rates in state and federal elections. The boundaries of school districts are set by higher-level units. As long as those higher-level units are themselves subject to democratic governance, specifying how funds should be spent is as democratically legitimate as authorizing the local units to raise funds and govern schools. So, just as respect for democracy does not require that local units be permitted to supplement central funding, it also does not demand that local units have latitude over how to spend locally raised or centrally provided funds. It is worth noting at this point that the separation between fi nance

104

Chapter 5

and governance is somewhat arbitrary. It seems logical to think that decision-making authority for a particular decision should rest with the decision maker most likely to make the best decision (which, in turn, is that which best serves the values that really matter). That decision maker must have access to information about the relevant needs and the options available and must have the time or other resources to make good decisions. In some respects local decision makers may have better information and better incentives to serve children well. In other respects central authorities are better placed. Context also matters. If local actors are given clear incentives to pursue specific goals (as with testbased accountability described in the next chapter), it may make sense for them to have decision-making authority. Accountability is less likely to be effective without this local control.29 Autonomy will lead to differences across local areas in spending. Whether this differentiation is beneficial will depend on local abilities, resources, and goals. Empirical evidence about these characteristics can help central decision makers, and voters, determine the likely benefits of autonomy.

Conclusion Decisions about fi nance are integral to all school systems. Each system requires a myriad of decisions about how to raise, distribute, and spend funds. The framework that we provide for identifying the values in play and bringing empirical research to bear on the likely effects of policy choices on those values can serve as a base for making more informed decisions even in light of the political processes that go into fi nal policy outcomes. Here, we have focused on three sets of school fi nance choices in particular: whether and how much to adjust for costs and needs differences, whether to allow local supplementation to state dollars, and to what extent and in what ways to specify how funds are used. The choice of how to distribute funds to schools or school districts so as to adjust for differences in costs and needs depends on the weights given to different goals. Adjustments for costs that can be measured and are outside of the control of school leaders are beneficial on distributive grounds and are unlikely to affect how well schools are managed and the overall effectiveness of how educational goods are produced.

School Finance

105

Adjustments for cost differences that are within the control of school leaders may lead to distortions in how schools are run that either benefit or (more likely) hurt the effective production of education goods. How much to make these cost adjustments—for example, giving more money to poorly run school systems—again depends on how one weighs distributive values against overall productivity. The same concerns hold for a number of possible adjustments for differences in student needs across schools. Do policy makers provide greater support for lower-achieving students, or does such support lead districts or schools to reduce their focus on student achievement? Our analysis also sheds light on the relative merits of allowing local supplementation of funds. Because some parents have the opportunity to provide additional educational goods to their children, either through school contributions or spending directly on their child, even the no-supplementation approach may not fully realize any given distributive goal. However, supplementation is likely to lead to greater spending and greater variation in spending across schools. Currently, little research has provided evidence on whether this increased variation— with its expected greater spending in already affluent districts—leads to greater variation in educational goods. Such research would allow a better assessment of the merits of this policy and would illustrate how our framework can be useful to social scientists designing research as well as those making policy decisions. Even without this information, our analysis suggests that if policy makers place a higher weight on the distributive value of equality over other goals, then restricting supplementation is likely to be beneficial. However, if one values the overall level of educational goods more than how they are distributed, then allowing local supplementation may be preferable as long as it does not substantially reduce school spending in lower-income districts. On the question of how much autonomy to allow local jurisdictions in the use of funds provided by the central government, our analysis again points to trade- offs among valued outcomes. Greater autonomy will almost surely lead to greater variation in spending and in the effectiveness in spending. If local decision makers have greater knowledge than central decision makers of the needs and opportunities in their area and their goals are aligned with the goals of central decision makers, then autonomy will likely lead to better decisions overall. However, in practice, there is often a trade- off in which some local decision makers make bet-

Chapter 5

106

ter decisions and some make worse. In the next two chapters we address potential policy approaches for better reaching goals in light of the likely trade- offs. Overall, school fi nance decisions illustrate the complexity of decisions facing education policy makers. These decisions require an understanding of the political forces and local incentives. They can also affect a variety of valued goals. The production of educational goods and the distribution of those goods across students are the most evident values in play, but a range of independent values from childhood goods to respect for democracy and parents’ interests in their children are also salient. Our framework can help clarify the goals and their trade- offs even if individuals make different judgments about how to weigh the values when making or advocating policy decisions. *

*

*

Appendix Any estimate of resource costs and needs is specific to the current knowledge or “technology” of schooling. Innovations in curriculum or instruction, for example, may reduce the cost of achieving a given education goal and, in some cases, investment in research and development may be a better use of funds for improving outcomes than additional dollars for current instruction. The governance structure and the incentives it creates can also influence the development of technology, spurring new approaches to benefit students. 30 The extant approaches for estimating funding levels needed to obtain a given set of outcomes for students by adjusting for cost and need differences across schools and districts have various strengths and weaknesses. Most common is the professional judgment method. It involves panels of educators designing prototype schools that they believe will provide adequate educational opportunities. Consultants hired to conduct the study then attach costs to these prototypes. The disadvantages of this method are clear—panel members are unlikely to know the most effective use of funds for a given population, and the schools they put together may simply reflect their own biased preferences. Some professional judgment panels are supplemented with research evidence on effective use of resources, such as the average estimated effects of small versus large class size or the effects of mentoring for entering teachers.

School Finance

107

A second approach, sometimes referred to as the evidence-based approach for costing schools, ends up being very similar to the professional judgment approach, because evidence on the effectiveness of different types of investments is sparse. Another popular approach is the cost-function method, which uses regression models of current per-pupil spending at the school district level on district characteristics and student outcomes. The idea here is to develop a statistical model describing the relationship between spending and student outcomes. The problem with the approach is that although the relationship may be due in part to the causal effect of spending on student outcomes, it may also be due to other factors. For example, communities with greater interest in education may both spend more and have higher achievement even though the spending did not cause achievement gains. Alternatively, communities with greater needs may both spend more and have lower achievement, but, again, the relationship between spending and achievement may not be causal. If researchers had perfect measures of differences across districts in all factors that influence spending, then they could account for them in the costfunction approach and produce a good model for predicting how much money would be needed to reach a given level of achievement. However, these measures of cost and need differences are not available and, as a result, the cost-function approach is often not a reliable method for estimating adequate funding. 31 In practice, many states have used professional judgment panels to estimate the cost of providing an adequate education. Some have supplemented these panels with evidence on specific school inputs and with cost-function estimates of the necessary spending levels. Given the imprecision of these techniques, it is not surprising that states using a formula that allocates different funding levels per pupil to school districts, based on their student characteristics, vary greatly in the weights that go into the formula.

chapter six

School Accountability

I

n the previous chapter we showed how values and evidence are interrelated in decision making about school funding. We now turn to a second set of decisions that are central to education decision making: how to ensure that schools are doing their part to contribute to the production and valuable distribution of educational goods as well as to other relevant values. Schooling is a vast public enterprise. In the United States, fi fty million students attend almost one hundred thousand public schools. Because schools are funded by publicly collected revenue, policy makers have a responsibility to assure that the money is well spent. We use the term accountability to refer to the various methods policy makers can use to ensure that schools use the resources they have—such as their teachers, their discretionary funding, and their facilities—as productively as possible toward desired ends. Without some form of accountability, schools might waste resources, perhaps because of mismanagement, lack of knowledge, limited capacity, or insufficient attention to desired goals. One approach to school oversight is regulatory policy. Basic regulations prohibit school officials from using funds fraudulently and require schools to operate in accordance with legal provisions such as those related to student safety or the hiring or fi ring of teachers. Regulations might well also apply to specific inputs. Schools might be required, for example, to meet certain standards with respect to classroom space per student, books in the library, teachers with appropriate credentials, or class size. Although regulations are clearly important policy levers, they primarily set floors on quality and are not well designed for quality improvement above the floors.

School Accountability

109

We focus in this chapter primarily on incentive-based systems designed to hold schools accountable for student outcomes or for internal school processes and practices. These accountability systems complement or substitute for the regulation of school inputs. They typically include two components—information and consequences—although to varying degrees. Most accountability systems provide information on how well a school is doing based on one or more relevant metrics that may or may not be linked to specified standards. That information can then be used directly by the school to make better decisions, by higher levels of authority to intervene in some manner, and by parents who are making decisions about where to live or which schools they want their children to attend. Thus, information is the starting point for accountability. Most accountability systems also involve consequences for schools that perform particularly well or badly. Those can be positive or negative and provide incentives for the schools to improve their performance. Positive consequences might include public recognition and praise, bonuses for the school’s teachers, or greater school autonomy. In some cases the consequence may simply be additional revenue associated with the school’s ability to attract more students. Negative consequences might include the naming and shaming of low-performing schools, loss of revenue associated with the outflow of students, fi ring of principals or teachers, and, in the extreme, school restructuring or school closure. The clearer and more negative the consequences, the higher the stakes for schools. These higher stakes may increase the incentives for educators to respond to do what policy makers intend them to do, but they can also lead to frustration, especially if educators do not believe the needed changes are under their control or are in the best interests of students.

Identifying the Main Values In Play Ideally, schools would be held accountable for their contribution to the full range of educational goods. These would include, among others, the cognitive skills and knowledge that are valuable in the workplace or enable good citizenship along with the other dispositions and attitudes that enable individuals to flourish and to contribute to the flourishing of others. Moreover, a good accountability system would attend not only to the level of educational goods but also to how they are distributed both

110

Chapter 6

across and within schools. At the same time, the system would balance the level and distribution of educational goods against a concern for various independent values. The most salient one in this case is likely to be childhood goods—that is, the quality of children’s experiences when they are in school. In practice, accountability systems differ in the emphasis they place on specific types of educational goods. This variation may occur because some educational goods are easier to measure than others or because different groups with different priorities—for example, local voters, parents, educators, or policy makers—make the decisions about what is valuable. Further, policy makers may differ in the weights they place on the value of some capacities relative to others. Some policy makers may care particularly about the cognitive knowledge and skills that are valued in the labor market or those that are needed for democratic competence. Others may believe that the most important educational goods for schools to foster are those that are valued by parents. Still others may give greater weight to personal autonomy or respect for others even if parents do not view those as essential activities of schools. In practice, choices among educational goods are central to the design of accountability systems. The design of an accountability system may also affect the distribution of educational goods, both overall and within individual schools. Some accountability systems are intentionally designed to promote distributive goals such as adequacy or the reduction of gaps in educational goods across groups of schools or groups of students. If the goal were for all students to reach a threshold level of achievement, for example, policy makers might focus attention on the proportion of students within each school who meet that level and provide incentives for all schools to raise that proportion to some target level without regard to whether some schools reached higher levels. A system designed to promote equality among groups of students might focus on average test score gaps between groups such as those defi ned by their income, race or ethnicity, or disability status. One designed to improve standards at the bottom of the distribution might incorporate a program of capacity building in schools serving large numbers of less advantaged students. An accountability regime may also have unintended distributive effects. Consider an accountability system aimed at improving student learning across all schools. If schools serving more advantaged students can better respond to the incentives the accountability system creates,

School Accountability

111

accountability could increase inequality of educational goods. Additionally, if accountability pressure makes low-performing schools less attractive places to teach, it may widen the distribution of teacher quality across schools. Alternatively, if, in the absence of accountability, schools serving lower-performing students would have made the least effective use of their resources, the students in those schools could benefit most from an accountability system. Finally, three independent values are especially salient—childhood goods, parents’ interests, and respect for democratic processes. An accountability system would reduce childhood goods, for example, if it shifted the attention of school personnel toward the development of specified educational goods and away from the provision of an environment with opportunities for exploration and play. Of course, policy makers who valued childhood goods could avoid that outcome by choosing an approach that better balances educational goods with childhood goods. Some forms of accountability could also confl ict with parents’ interests. That would be the case if, for example, policy makers hold schools accountable for educational goods that are not highly valued by parents or if policy makers’ attention to specific distributive goals clashes with affluent parents’ interest in their own children. Other forms of accountability, including some we discuss in the next chapter on school autonomy and parental choice, place great weight on parents’ decisions. So even if an accountability system succeeds in achieving a better level and distribution of educational goods, it may do so at a cost to other values. In these cases, trade- offs among values will be required.

Identifying the Key Decisions Relevant to Those Values As we have noted, the regulation of inputs can importantly affect the level and distribution of educational goods. Consider, for example, regulations specifying maximum class size. To the extent that researchers can identify a class size above which student outcomes are likely to decline, one could expect a binding regulation that limits classes to be below that threshold to generate more educational goods, all else being constant, than the alternative of no restriction. Moreover, the limit could promote one or more of the distributive values if, as the evidence suggests, smaller class sizes disproportionately benefit students from disadvantaged backgrounds. Similarly, regulations requiring that teachers be

Chapter 6

112

certified may also affect the level and distribution of educational goods. If certified teachers are more effective than those without certification, for example, one would expect a requirement that all teachers be certified to increase the level of educational goods, especially for the most disadvantaged students who might otherwise face the weakest teachers. Regulations are part of all education systems, but they sometimes do little to affect the quality of schools above the floor. In some cases, moreover, they may be a costly and clumsy way to achieve various educational goals. Hence, policy makers may choose to turn to incentive-based accountability as a substitute for or a complement to regulations. As they do so, they face the following two key decisions: 1. Should they impose some form of administrative accountability? 2. If so, for what should they hold schools accountable: student outcomes or internal school policies and practices?

We defi ne an administrative accountability system as one in which central policy makers determine the criteria for accountability, gather and report the school-level measures relevant to those criteria, and establish consequences to reward school success or punish school failure. This approach contrasts with two other broad types of accountability—political accountability and market-based accountability. These are approaches in which other groups of actors defi ne what it is that schools should be doing. The decision of whether or not to use some form of administrative accountability thus revolves around the strengths and limitations of the other approaches as well as the potential for administrative accountability itself to realize the values at stake. We briefly describe the other approaches here, leaving to the following section an assessment of how well they serve particular values. Political accountability involves elected officials making decisions about schools and their decisions being answerable to the electorate. The clearest case is in the context of relatively small local school districts of the type historically common in the US education system. Within each district, local voters express their preferences for schooling through the ballot box by voting for the school board members who represent their preferences. If voters are displeased with the performance of the schools, they can replace officials with candidates who promise to operate the local schools more in line with their preferences. While this particular structure is specific to the United States, political accountabil-

School Accountability

113

ity operates wherever any democratically elected body makes decisions about education. In contrast, market-based accountability for schools starts with families being able to choose the schools their children attend. Historically, high-income families have been able to do this through their choice of neighborhood or by opting out of the public school system in favor of private schools. The mobility of these high-income families provides incentives for school officials to pay attention to school quality because they know that unsatisfied families can withdraw their children and their support for public funding of schools. If school choice is disconnected from choice of neighborhood, as it would be with a deliberately market- driven approach, all parents, and not only those with high incomes, would be in a better position to hold their children’s schools to account for the things they value. Thus, at least in theory, schools are held accountable for the educational goods preferred by parents. Some form of political accountability is inevitable in a democracy because disgruntled voters can always seek to change the government through the ballot box. Similarly, as long as housing markets and private schools exist, market accountability is in operation, which is tilted to the preferences of more affluent families. Policy makers have to decide whether to supplement regulation—and the political and market-based accountability forces already in place—with an administrative accountability system, and if so, what form that system should take. Assuming some reliance on administrative accountability, what should schools be held accountable for? Policy makers designing an administrative accountability system face the problem that there are no perfect measures of the school “quality” they are aiming to promote.1 One option is to focus on things that are fully under the control of the school, such as its internal processes and practices. Included here might be the school’s procedures for identifying and addressing the needs of children, its efforts to promote respect and to provide healthy, safe, and secure environments for children, and its success in providing a coherent curriculum, all of which would need to be identified through some formal system of school inspections. An alternative option is to assess school quality in terms of student outcomes, often measured in practice by student test scores. The former approach typically consists of inspection systems in which professionally trained external review teams visit individual schools on a periodic basis, use formal rubrics to evaluate a range of practices, and

114

Chapter 6

write public reports highlighting the strengths and weaknesses of the school. In contrast, accountability systems based on student outcomes typically judge schools based on the test scores of their students in a limited number of subjects, such as math and reading, with clear criteria for what constitutes good or bad school performance. On either approach, successful schools may be singled out for public praise or possibly fi nancial bonuses, while schools judged unsuccessful may face sanctions. For more than a dozen years, all schools in the United States have been subject to outcomes-based accountability primarily as a result of the federal No Child Left Behind (NCLB) Act of 2001 that held all schools accountable for their students’ test scores in math and reading. Even before the passage of that law, many US states had introduced their own test-based accountability systems. In contrast, other developed countries—such as New Zealand and the Netherlands, both of which have systems of self-governing schools and strong formal school choice arrangements—rely heavily on process-based accountability in the form of inspection systems. Some countries—such as England, with its league tables that report student test results by school and its Office for Standards in Education—have a mixed system. According to a 2011 report from the Organization for Economic Cooperation and Development (OECD), the United States is unusual in placing so much more emphasis on outcomes-based accountability than on some form of school inspection, especially for elementary (or primary) schools. 2 Of the thirty-two OECD countries reviewed in the report, only the United States, Turkey, and the French part of Belgium use any form of test-based accountability at the elementary level. In contrast, twenty-two countries operate inspection systems for schools at that level. 3 The situation for secondary schools is more varied: thirteen countries use outcomes-based accountability, and twenty-three countries have some form of inspection system. As is evident from the differing choices that countries have made, there is no consensus on how best to hold individual schools accountable.

Assessing the Options in Light of the Values We address outcomes-based accountability and school inspection sequentially, starting with the decision to impose some form of administrative accountability.

School Accountability

115

Administrative Accountability Some have argued that the local political approach common in the United States during much of the twentieth century worked quite well, at least as measured by the overall level of educational goods. Some historians think, for example, that competition among local jurisdictions led to the faster development of high- quality secondary education programs than would have occurred had the system been more centralized.4 At the same time, local political accountability of this form can generate large differences—and inequalities—across communities in terms of what schools do and the educational goods their students develop. One community might place great value on the performance of athletics teams while another emphasizes academic success. As we discussed in chapter 5, economic inequalities across local communities are likely to translate into large disparities in spending. Demographic, economic, and social changes make such a highly localized system increasingly ill suited to the requirements of schooling today. The current high levels of movement of people and fi rms across states and nations and the potential for large inequalities across communities associated with differences in their income or wealth mean that such a system will benefit some children much more than others. Further, the growing national and international competition for skilled workers may render local voters less well positioned to make decisions about children’s education than in the past, when many adults stayed local and could be expected to work in the same types of jobs as their parents. Hastened by court cases that highlighted the constitutional responsibility of state governments to assure general and uniform systems, the trend in the United States has been to shift the setting of educational goals and policies toward the central government. This increases the distance between policy makers, who are making many of the policy and regulatory decisions, and schools, which are delivering education services. Since regulations are rarely subtle enough to enhance the quality of schools above a minimum (albeit perhaps a high minimum) level, higher-level policy makers may seek more incentive-based ways to get assurance that the schools are meeting expected goals or benchmarks. Administrative accountability is one such option. Market-based accountability, which we discuss in more detail in the next chapter, is an alternative option. Yet even if one places heavy weight

116

Chapter 6

on the independent value of parents’ interests, some form of administrative accountability may still be valuable. First, for education markets to work well, consumers—in this case parents—need good information about schools’ quality. Second, policy makers may place greater weight on some educational goods than do families. Some parents, for example, may want their children in schools with children from similar backgrounds as their own, or in nearby schools, even if those schools are of low quality. Parents also may only consider the flourishing of their own children instead of taking into account how their children can contribute to the well-being of others. Third, because they have more options and more cultural capital, middle- class families are better able than lowerincome families to work the education system to their advantage. So policy makers may want to use accountability for distributive purposes—to counteract the power of more affluent parents in the competition for limited resources. Policy makers have good reasons to devise some form of administrative accountability to ensure that schools are making effective use of their resources and are doing so in a manner consistent with specific distributive principles. How much to rely on administrative accountability will vary depending on contextual factors that determine the feasibility and desirability of the other approaches as well as the need for additional accountability. Also relevant is how policy makers weigh the level and distribution of the educational goods that they are able to incorporate into an administrative accountability system against other educational goods and parents’ interests that might be adversely affected by such a system. Their decision, then, will depend on the capacity of the central government to implement an administrative accountability system that enhances their preferred balance between the values in play. Assessing School Accountability For what should policy makers hold schools accountable? Should schools be accountable for student outcomes or internal school policies and practices or both? Assessing Outcomes- Based Accountability Given that some form of administrative accountability is desirable, outcomes-based accountability for schools often has great intuitive appeal. Indeed, holding schools accountable for the outcomes of their students would seem to

School Accountability

117

be fully consistent with the central idea of this book, namely, that education policy makers should keep their eyes on the prize—the capacity for children to flourish and to contribute to the flourishing of others. Unfortunately, valued outcomes are often difficult to measure, and some are more difficult to measure than others. Unless schools are held accountable for all valued outcomes, the incentive for schools to direct resources to the ones for which they are held accountable could lead to lower levels of resources for the others, with the net effect on educational goods being unclear. Further, schools have only limited control over their students’ outcomes, because families and communities also play an important role in the generation of educational goods. Unless policy makers can devise effective methods for measuring the contribution of schools to student outcomes rather than the outcomes themselves, efforts to hold schools accountable for those outcomes could unduly burden schools serving concentrations of disadvantaged students and perhaps induce some schools to react in counterproductive ways. When using outcomes-based accountability, policy makers have to rely on student outcomes they can measure. In practice, that typically means student performance on tests in a few key subjects such as math and reading supplemented perhaps by measures of student attendance or high school graduation rates. Thus, the focus tends to be on cognitive  knowledge and skills to the exclusion of other educational goods. In the United States, this focus on cognitive skills was partly driven by a desire to raise academic standards in order to promote economic development. Some early adopters of accountability at the state level, for example, were southern states, such as South Carolina and North Carolina, whose academic standards had long been low, making it difficult for them to compete for high-wage jobs with other states. Similarly, at the national level, NCLB grew out of the national movement to raise standards so that US students would be more competitive with those in other developed countries who were outperforming them on international tests and so were likely to outcompete US graduates for jobs. The heavy use of this type of accountability in the United States has provided grist for extensive social science research into its effects. Researchers have examined the effects of programs that were fi rst introduced by states and districts during the 1990s and then more recently the effects of the 2001 federal NCLB Act that applies to schools across the country. This research has often used test scores at the student level

118

Chapter 6

that were required as part of the accountability systems. Also available for analysis are test scores from the National Assessment of Educational Progress (NAEP), often called the Nation’s Report Card, which have been reported for representative samples of students in grades four and eight at the state level since the early 1990s and nationally for children age nine and thirteen since the early 1970s. Less readily available, however, is information on other educational goods. We draw on key elements of this research to assess the likely effects of outcomes-based accountability on the values in play. Effects on educational goods. The fi rst and most obvious question concerns the extent to which outcomes-based accountability has raised the level of educational goods, starting with the cognitive knowledge and skills that are directly incentivized. Any discussion of evidence about the NCLB effects of test-based accountability in the United States should recognize that the federal policy, like many of the state accountability systems, has relied largely on test scores covering reading and mathematics. They have not extensively used information about knowledge or skills in art, music, social science, science, or languages let alone about the acquisition of skills and dispositions pertaining directly to whether a person can have successful personal relationships or become a good citizen. If schools are held accountable for only a subset of the educational goods that matter, they have strong incentives to focus on those goods at the expense of others. Not surprisingly, many schools responded to the focus on reading and math by allocating more time to those subjects and less to other subjects and activities, including recess. A nationally representative survey of 349 school districts between 2001 and 2007 shows that schools raised instructional time (measured in minutes per week) in English and math quite significantly and reduced time for social studies, science, art and music, physical education, and recess. 5 National surveys by the Center on Education Policy yield similar patterns. To the extent that time devoted to these other subjects contributes to the development of other capacities— such as the capacity for democratic competence or personal fulfi llment— that are important for flourishing, any gains in cognitive skills in math and reading must be weighed against the loss of these other capacities. Unfortunately, we have little direct evidence on the harm to those other educational goods, in part because of the difficulty of measuring them. In principle, one could develop administrative accountability systems

School Accountability

119

that apply to a much wider range of outcomes than test scores in math and reading. In fact, the 2015 federal legislation that replaced NCLB empowers and encourages states to introduce additional measures.6 One should not, however, underestimate the difficulty of developing reliable measures of educational goods other than cognitive skills. Producing good measures of dispositions and attitudes involved in competent democratic citizenship would take a great deal of research and experimentation. Moreover, some potentially reliable measures of such dispositions would be ill suited for accountability purposes. Consider, for example, the current use in one chain of US charter schools (see chap. 7 for discussion of charter schools) of a ‘Character Report Card.” 7 Each child has a card, and each teacher regularly grades the student on a one-throughfive scale with respect to twenty-four behaviors that display “grit,” “selfcontrol,” “optimism,” “gratitude,” “social intelligence,” and “curiosity.” Even if these reports could be reliable and useful for the purposes of tracking a student’s progress and deliberating about interventions, the incentives for teachers to misassess the students would be hard to resist if the reports were used for purposes of accountability. The resulting distortions would render the reports useful neither for holding schools accountable nor for judging what would be helpful for individual students. Indeed, developing valid measures of student achievement in math and reading alone is itself challenging. How well test scores measure actual achievement in those areas depends on the design of the tests. Reliable and valid tests are expensive to develop, and very similar questions must appear repeatedly on tests, both so that the reliability of those questions can be assessed and so that the scores are comparable over time. Because no test can cover all aspects of a subject area, there are also concerns about the extent to which teachers have incentives to teach narrowly to the test rather than to the broader set of constructs that the test is designed to measure. Even with good measures of student achievement, it is difficult to gauge how much a school contributes to the learning of its students. NCLB, at least in its original incarnation, as well as some of the state accountability systems, held schools accountable for the level of their students’ test scores rather than the improvements in those scores. A school with low test scores but in which the students improve considerably during the year may well be more effective than one with high test scores where students do not improve so much. Over time, many states requested waivers from NCLB and moved toward measuring schools by

120

Chapter 6

their “value added” as revealed by gains in test scores. But value-added measures have their own problems. Unless sample sizes and the differences in value added are large, it is impossible to tell which school is better because there is so much error in the measure. Further, improvement in student performance, just like the level of performance, is affected by forces outside and beyond the control of the school, so it can be difficult to identify the contribution made by the school itself. Any interpretation of the evidence about test-based accountability should keep in mind the limited content coverage, the difficulty of reliable measurement, and the problems of converting measures of student outcomes into measures of school effectiveness. Although the results are somewhat mixed, the research suggests that test-based accountability as used in several states in the 1990s and in all since 2002 has generated at most small positive effects, typically more apparent for math than for reading, on student achievement. Somewhat larger effects appear to have emerged from early experiences with test-based accountability at the state level compared to the subsequent nationwide implementation of NCLB, though assessing the effect of a nationwide program is difficult. The most convincing of the US studies are those that use scores on a low-stakes test as the outcome measure, such as the National Assessment of Education Progress (NAEP), rather than on the high-stakes test that is used as part of the accountability system. Since NCLB was implemented, NAEP has shown steeper upward trends in both math and reading for nine-year- olds but not for thirteen- or seventeen-year- olds in either subject. The most recent scores, however, show significant declines in both subjects in 2015.8 Careful studies based on serious efforts to isolate the causal effects of NCLB typically fi nd clear but at most moderate positive effects for math scores in fourth grade, less clear effects for math in eighth grade, and generally little or no effects in reading.9 On a more positive note, student graduation rates have increased meaningfully since the act was implemented, especially for black and Hispanic students, in contrast to the previous thirty years, when graduation rates had remained flat.10 Evidence from the United Kingdom provides somewhat stronger evidence that test-based accountability can raise student achievement, especially for students in low-performing schools. In particular, student achievement measured at the secondary level in Wales dropped relative to England when Wales stopped publishing school performance ta-

School Accountability

121

bles.11 It is worth noting that the performance tables used for secondary schools in the United Kingdom index a much greater variety of tests of a larger range of educational goods than are assessed in the United States. A possible explanation of the relatively limited impacts on the knowledge and skills targeted by the US accountability system is the potential for schools to game the accountability system. Forms of gaming include the selective assignment of low-achieving students to special education programs, selective disciplinary policies that keep low-performing students out of the testing pool, teaching test taking instead of the true content, or even outright cheating. Tests that are immune to gaming and that require students to learn the full range of desirable knowledge and skills are expensive and difficult to design and maintain.12 The limited effects on measured improvement may also reflect flaws in two assumptions underlying the US accountability regime. One is that schools exhibit low levels of student achievement because school personnel are simply not working hard enough or are not smart enough to raise student achievement. The other is that schools and teachers know how to raise student achievement. Instead, perhaps schools have limited control over student outcomes, school leaders do not know what to do to improve their schools, or teachers simply do not have the capacity or knowledge to teach to higher standards. Incentives can only work when the agents can and know how to do those things for which they are being rewarded. If they lack the relevant knowledge and skills, resources should be devoted to enabling them to gain them. Some states did invest substantial additional resources when implementing their outcomes-based accountability, and these have tended to see more positive results. In Massachusetts, for example, math NAEP test scores rose by 19 scale points for both fourth and eighth graders between 2000 and 2007, much bigger than the average increase typically associated with NCLB. A plausible explanation for the gains in Massachusetts is that its education reform strategy was far more comprehensive than test-based accountability alone, though test-based accountability was arguably a necessary component of the reform. In particular, the Massachusetts program included a substantial increase in funding (more than doubling in ten years), new learning standards, revised student assessments based on clear curriculum frameworks, revised teacher licensing and professional development programs, early childhood programs, and parental choice and the creation of new charter schools.13

122

Chapter 6

Effects on distributive values. The likely distributive effects of test-based accountability depend on its structure and goals. Unlike many of the state accountability systems, both NCLB and the system in Texas on which it was modeled were explicitly designed to raise the achievement of lower-performing students and reduce gaps in proficiency rates between subgroups of students both within and across schools. All schools were required to gradually raise the performance of all their students to proficiency. In addition, NCLB held each individual school accountable not only for the average performance of its students but also for each subgroup of students defi ned by their economic, racial or ethnic, or disability status. This provision was intended to induce each school to raise the achievement of all its students, not just those whose test scores were the easiest to increase. One crucial design feature of NCLB and also of some of the state programs that preceded it is that the criterion for success is stated in terms of the percentage of each school’s students that meet a specified proficiency standard. So the program is apparently designed to achieve a goal of adequacy with respect to at least some educational goods. But again, some schools may not know how to achieve that goal, and some, especially those serving disadvantaged students, may fi nd it difficult or impossible. Furthermore, the incentives established by the regime were not entirely benign. Focusing exclusively on the threshold gives schools incentives to ignore both students who are likely to perform well above the proficiency standard and those who are unlikely to reach it and to focus instead on students close to the threshold, sometimes referred to as “bubble” kids.14 Nonetheless, the NAEP does show a larger increase in achievement for the lowest-performing students (tenth percentile) than for high-performing students. That is, the variation across students narrowed somewhat after NCLB was introduced. Rates of proficiency are not the only possible outcome measure for an accountability system. An alternative approach is to set the target in terms of average gains in student test scores. This approach requires test score data on individual students that can be matched over time, and it is difficult to do well. Nonetheless, several states, including South Carolina and North Carolina, used gains along these lines well before the introduction of NCLB. Such an approach is more consistent with the goal of raising overall achievement but may not be as effective at addressing the performance of low-performing students.15 Historically, black and Hispanic students in the United States have,

School Accountability

123

on average, scored lower on tests of academic achievement and attained lower levels of education than white students. Trends in outcomes by race and ethnicity provide some evidence of the distributive effects of outcomes-based accountability. Between 2004 and 2012, test scores for blacks and Hispanics increased more than those for whites in both math and reading. Some researchers also fi nd a substantial increase in high school graduation rates between 2000 and 2010, especially for black and Hispanic students.16 These rates had been flat during the previous twenty years, especially for blacks. It remains unclear, however, to what extent these improvements with respect to distributive values can be attributed to pressure from accountability.17 It is not surprising that these moves to hold schools accountable have had only a small impact on the distribution of achievement. While accountability pressures have undoubtedly induced individual schools to pay more attention to their low-performing students, test-based accountability does not address many of the barriers that impede the learning of disadvantaged students, such as their poor health and the economic and racial segregation of many residential neighborhoods. As figure 3.7 in chapter 3 illustrates, even countries with very strong education systems and social supports are not able to eliminate achievement gaps (in that case measured by PISA scores) between children from families with low and high socioeconomic status, although some are more successful with their disadvantaged students than others. Effects on independent values. Outcomes-based accountability may also affect independent values, particularly childhood goods. These policies ignore those goods and may unintentionally reduce them if some of the stress associated with school-level pressure is transferred to students or if they reduce the time or resources devoted to the quality of children’s experience in school, for example, through reduction in recess time. Children often respond negatively to high-stakes tests. Various qualitative studies show that many students experience anxiety, anger, pessimism, boredom, loss of motivation, or “less love of learning” as a result of their participation in high-stakes testing.18 Some research has documented stomachaches or headaches indicative of anxiety.19 A recent study found a higher rate of student misbehaviors in schools declared to be failing to meet the NCLB accountability standard than in those just above the failure cutoff. 20 That said, not all studies fi nd a negative effect on student behavior or engagement. One study found that accountability

124

Chapter 6

pressure increased students’ stated enjoyment of math and apparently made them no more anxious about standardized tests.21 Overall. The evidence so far on the effects of outcomes-based accountability shows that the task of raising student achievement is not easy and that giving schools incentives to do so produces quite limited gains. At the same time, as is evident from the Massachusetts experience, a well-structured accountability system that is embedded in a more comprehensive reform package can have quite large positive effects on student performance as well as reducing achievement gaps. 22 In sum, outcomes-based accountability as it has been used in the United States creates incentives for schools to focus on only a small subset of valued educational goods, typically the skills and knowledge measured by test scores in math and reading, and pays no heed to childhood goods. Moreover, even for the outcomes it targets, accountability systems are hindered by not being able to differentiate the effects of schools from those due to other factors. When schools are held accountable for outcomes that are outside their control, they may well respond to accountability pressure in unintended and often undesirable ways. How much weight to place on these concerns depends on the extent to which accountability pressures produce negative effects as well as on one’s judgments about the different values at stake. Assessing Inspection Systems When school accountability takes the form of an inspection or review process, external review teams visit individual schools on a periodic basis and write public reports highlighting the strengths and weaknesses of the school. The school inspectors are professionally trained and use formal rubrics or protocols designed to produce consistent measures of school quality. In combination with school self-reports and fi nancial and other documents provided in advance of their visit, inspectors aim to gain a solid understanding of the policies and practices within a school. They typically examine compliance with rules and regulations, the quality of instruction, student performance, and the quality of the school climate. Financial management is typically reviewed through a separate audit process. In addition, in some countries inspectors look at perception and satisfaction levels of students, parents, and staff. When they fi nd evidence of shortcomings, the school is typically required to

School Accountability

125

develop a plan to address the problems and is subject to a subsequent review within a relatively short period of time. Some inspection systems, such as the one in England, are clearly summative and judgmental. Others, such as the continually evolving system in New Zealand, are also intended to be more formative, with the inspection process itself driving and supporting school improvement. An advantage of an inspection system is that inspectors can examine a variety of internal school processes that could be useful for generating a range of educational goods. At the same time, an effective inspectorate system requires well-trained inspectors who can identify quality schools while at the same time allowing for the variety of different approaches that may be used to benefit students. As we discuss below, research on the effects of such systems is far more limited than that on test-based accountability. Effects on educational goods, distributive values, and independent values. Little evidence is available on whether inspectorate systems improve outcomes for students or positively affect the distribution of educational goods. Similarly, little is known about their effects on independent values such as childhood goods. Nonetheless, some description of these programs provides insights into how such systems might affect the values in play. Compared to outcomes-based accountability, the inspection approach starts with a much broader conception of school quality or, in our terminology, of the schools’ contribution to the production of educational and childhood goods. However, it requires an understanding of the processes that lead to the development of those goods and an ability to measure those processes. One concern with this approach is that reasonable people might well disagree about the extent to which specific school processes and practices promote the desired educational and childhood goods. Consider, for example, New Zealand’s experience. An innovative part of New Zealand’s major reform package of the early 1990s—a package that turned operating responsibility over to individual schools— was the establishment of an Education Review Office (ERO) designed to monitor school quality through periodic school visits. The initial intent was for the ERO to evaluate each school in terms of its own mission statement. In practice, however, the vagueness of the mission statements made that approach unworkable. With no clear model, the ERO

126

Chapter 6

then chose to hold schools accountable primarily for their compliance with regulations spelled out in the form of national guidelines for school policies and practices. These reviews concentrated primarily on management procedures and often became mechanistic. Concerns arose that the standards for success were biased toward the types of procedures that worked best in middle- class schools. Starting in 2003, the country introduced a new planning and reporting framework for schools under which the ERO started focusing much more on process questions. How well is the school using information on student achievement, both formally and informally, to develop programs to meet the needs of individuals and groups of students? How well is the school using time for learning purposes? How effective are the systems for identifying and meeting staff professional development needs? How well does the school establish partnerships around learning with its community?23 With the arrival of a new chief review officer in 2010, the New Zealand Education Review Office modified its approach once again, this time by strengthening the link between external evaluation and internal self-review. This change aimed to further its two-part goal of holding schools accountable and of promoting school improvement: evaluation for accountability purposes involves reporting on compliance, goals, and standards, while the improvement agenda involves assisting schools to improve through self-review. 24 Many inspectorate systems—including those in England, the Netherlands, New York City, and the Charlotte-Mecklenburg schools in North Carolina—focus on how well schools make use of data on student achievement to allocate resources within schools and on the coherence of policies for supporting student learning. This use of achievement data raises some of the same issues, such as limited content coverage, that we discussed in relation to outcomes-based accountability. However, the external reviewers are in a position to judge school quality based not only on how the students are doing overall but also on how well the schools use their resources to address the particular circumstances of their students. That focus is designed to promote both higher overall student outcomes and the distributive goal of assuring that all students attend a school that meets an adequacy standard, where adequacy is defi ned in terms of the quality of internal school processes. Not surprisingly, inspection systems are not free from problems. One is the danger that the reviewers will emphasize processes that are not in fact directly related to valued educational outcomes. On the other hand,

School Accountability

127

the review protocol could be so broad that the reports are not useful for distinguishing between those processes that contribute significantly to educational goods and those that do not. Moreover, educators may try to mislead the inspectors about what is really going on in the schools, as has been documented for the inspection program in England.25 Further, because it is based on human judgment, the observational approach may be subject to variation in ratings that reflect differences across reviewers rather than differences in school quality. Unless the reviewers are well trained and the reports are tested for reliability across reviewers, the system could provide misleading information. Depending on how they are operated, the reviews can also impose substantial burdens on the schools being evaluated and can be extremely stressful for school personnel as was true, for example, during the early years of the English program and also in Northern Ireland. 26 The most commonly discussed concern about the inspection approach, at least in the United States, is its high cost. The more frequently schools are visited and the longer the visits, the higher the costs will be, where costs include reviewers’ time and the costs of their training as well as the time spent by school officials preparing for and responding to reviews. Presumably, it was largely because of the high costs associated with its five- day school visits in the early years of the English program that the length of the visits was reduced. Similarly, the high cost of external reviewers from outside New York City was one of the reasons for that city’s switch to internal reviewers. Nonetheless, as far as the values discussed in this book are concerned, the inspection approach has four possible strengths. The fi rst is that it avoids the distortions, such as teaching to the test and the narrowing of the curriculum, associated with test-based accountability, and it can promote the production of educational goods that are not subject to quantitative measurement. Among the standards used by the Dutch inspectors, for example, is one concerned with whether staff behave respectfully toward students and whether the school promotes respect among the students. Standards of this type are consistent with the view that schools should encourage students to treat others as equals, one of the capacities we have noted as being important to flourishing and to contributing to the flourishing of others. Second, the inspection process provides potentially useful information to school officials and to higher-level managers, not just incentives for performance. Because the information relates to internal school

128

Chapter 6

processes and practices, the school itself is likely to be in a position to change many of them. Third, by generating rich information about children’s experience while they are in school, the inspections often explicitly promote the production of childhood goods that might be ignored in a test-based system. Although the specific components may be hard to measure, inspectors have the potential to determine whether a school’s practices are generally consistent with the delivery of a safe and caring schooling experience. Finally, inspection systems can provide information about the overall education system as well as about individual schools. The inspectors can use their observations in the schools to hold policy makers publicly accountable for factors that are outside the control of individual schools, such as the availability of adequate fiscal resources and the appropriate training of teachers. That was the situation, for example, in the initial ERO in New Zealand. Assessments of the system as a whole may be more likely if inspection is run by an independent agency separate from the department or ministry of education. Even if the inspectorate has less autonomy, as in the Netherlands, it can still productively contribute to the policy-making process within the ministry through its access to comprehensive information on what is happening at the school level. Although school inspection systems have the potential to contribute to high levels of a variety of educational goods and to promote one or more distributive values, it is difficult in practice to determine how much impact they have. They are simply one component of a larger education system, so it is hard to single out their effects on valued outcomes. Moreover, the objects of accountability are sufficiently broad and hard to measure that evidence of any type is limited. Nonetheless, both descriptive and analytical studies generate some positive fi ndings. Survey responses by school leaders and teachers and analysis of progress reported in English Ofsted reviews indicate positive impacts on school policies and practices. 27 And careful quantitative studies of individual inspection reports in both England and the Netherlands fi nd small positive effects on student test scores. 28 With respect to distributive principles, a recent study based on data from the Dutch inspection system sheds light on differences in school practices across schools serving different student groups. The authors were exploring the extent to which the patterns across schools were consistent with the intent of the accountability system, namely, to assure an

School Accountability

129

adequate quality of policies and practices across schools. 29 They fi nd no average difference across subgroups with respect to the presence of robust school-wide procedures for assuring the well-being and safety of their pupils and teachers and for promoting respect among pupils. In addition, the evidence indicates at most small shortfalls in the highly disadvantaged schools in terms of teacher practices such as making efficient use of instruction time and engaging students actively in their own learning. Large deficits do appear, however, with respect to the ability of the highly disadvantaged schools to meet their students’ needs, presumably because, for many students, those needs are so great. This fi nding emerges despite the far greater number of teachers and other adults in those schools compared to the schools serving advantaged students. Overall. We have far less evidence with which to assess the effectiveness of inspection systems than we have for outcomes-based accountability, in part because fi nding such evidence is more difficult. The contribution of the social science researcher to date has been mainly descriptive. What types of activities have inspection systems in different countries tried to assess, what problems have arisen in their implementation, and what are their potential strengths with respect to the production of educational goods? Such evidence as is available suggests that this type of approach can be implemented at scale (as it is in several countries) but may require substantial resources in order to be done well. That said, a policy maker whose fi rst choice would be a well-resourced inspection system might nevertheless decide in favor of an outcomes-based system on the grounds that resources were lacking or could be better spent in other ways.

Conclusion There is no single best way to hold schools accountable. In terms of broad approaches, which strategy to favor will depend on context and judgments about the weight of different values. A high weight on the values of local political control or parents’ interests, for example, would support a significant role for local political accountability or marketbased accountability. Our discussion suggests that most governmental units would do well to supplement either of these broad approaches with

130

Chapter 6

some form of administrative accountability, that is, one in which policy makers make explicit choices about what they want schools to do and measure schools’ success at achieving these goals. With respect to administrative accountability, the choice between outcomes-based accountability and inspection systems that focus on internal school policies and practices revolves largely around the extent to which there are feasible measures of valued educational goods or of the processes that produce them. On the one hand, if policy makers choose outcomes-based accountability and are able to measure only a limited set of outcomes, such as student test scores in reading and math, they are letting schools off the hook for other educational goods, such as skills and dispositions that lead to personal fulfi llment and healthy personal relationships. On the other hand, if they use inspections to look only at regulatory compliance, they will also miss many goals for students. If policy makers use a richer inspection approach or combine the two approaches, they may be able to promote a broader range of educational goods, including dispositions and attitudes. The distributive consequences are unclear. Accountability systems often aim specifically to improve outcomes for the lowest-performing students and students from the lowest-performing groups. If well designed, this approach could lead to distributive benefits, particularly with regard to the incentivized outcomes. However, the effects are not clear when the broader range of educational goods is considered. For schools with ample resources serving advantaged students, an accountability regime that focuses narrowly on academic skills may not interfere very much with their ability to contribute to the production of other educational goods. Moreover, the families of children in those schools may be in a good position to generate other educational goods through the out- of-school activities—such as music lessons, sports teams, church activities, and summer camps and summer travel—they provide for their children. The situation may well differ for children from disadvantaged families attending schools with limited resources. In this case, the policy decision to focus accountability pressures on math and reading may benefit such children in terms of those particular skills but perhaps at the cost of other skills, knowledge, dispositions, and attitudes that contribute to flourishing. For this group of children, the failure to provide a rich schooling experience is more likely to translate into low overall educational goods than for their more advantaged counterparts because their

School Accountability

131

families are less well positioned to compensate for any shortfalls in the school. Policy makers concerned about flourishing and its distribution must attend to the broad range of educational goods as they design a system for holding schools accountable. How much emphasis policy makers should place on outcomes-based accountability versus an inspection approach will depend both on the value that policy makers place on different types of educational goods and on their ability and willingness to monitor processes. Neither a test-based system without good tests that are closely linked to a rich curriculum taught by well-trained teachers nor an inspection system without professionally trained inspectors will further the generation of educational goods. A separate consideration relates to the value of childhood goods such as play, wonder, and curiosity. Policy makers may have different views about how important such goods are relative to various educational goods in their particular contexts. In any case, accountability for childhood goods cannot be achieved by attention to student outcomes alone. An alternative is for external reviewers to visit schools on a periodic basic to see whether the environment is caring and whether children feel secure and are able to experience the joys of childhood. The public reports generated from such visits provide signals to schools about their strengths and weaknesses and can also provide useful information to parents. Although individual parents will inevitably have complaints about a school’s treatment of their children, systematic information about the school environment can put such complaints into perspective while at the same time providing assurance that the school is providing an appropriate schooling experience for all groups of children. In sum, school accountability has a central role to play in any education system. While countries, states, or school districts make the crucial decisions about funding, the organization of schools, the qualifications of teachers, or the design of specific programs, the schools themselves are where the rubber hits the road. What happens within schools is central to the generation of educational goods. For that reason, higher-level policy makers have a responsibility to hold schools accountable for making good use of the resources provided to them. We have emphasized the importance of their keeping the full range of educational goods in mind when designing accountability mechanisms.

chapter seven

School Autonomy and Parental Choice

O

ver the past twenty years, the management of schools has been increasingly devolved to the school site. District and state leaders have given school leadership teams more control over matters ranging from budget allocation to teacher hiring. This trend is apparent in countries around the developed world, including Australia, Canada, Chile, New Zealand, the United Kingdom, Sweden, and the United States, and the World Bank has promoted it in economically developing countries. Local autonomy allows school leaders to respond to needs and opportunities, potentially enabling them to make best use of available resources. It also allows schools to differentiate themselves; one size no longer needs to fit all. Giving school leaders power over resources, hiring, and educational approaches is likely to diversify schools: for example, instructional strategies and extracurricular offerings can vary. As schools differentiate themselves, students and their families have more opportunity to satisfy their preferences for one school or educational approach over another. So it is not surprising that policies aimed at increasing local autonomy often increase school choice for families. School diversification and school choice are symbiotic. As schools differentiate, parents will advocate for choice so their children can attend a local school that suits their needs and tastes. Similarly, choice gives schools greater opportunity and greater incentive to differentiate themselves from other schools. When families can choose schools, schools no longer need to cater to all or even most students. Instead, a school leader

School Autonomy and Parental Choice

133

can focus on the preferences of a minority of parents as long as the minority is large enough to support the school’s operation. School choice comes in many forms. Families with the means can choose schools by purchasing housing nearby. Realtors advertise school performance when selling houses because parents (and even adults without children) consider the quality of the schools when they choose where to live. Within school district boundaries, some districts in the United States have long offered limited choice through magnets and other specialized schools. This intradistrict choice has increased substantially in recent years: many large urban districts now ask families to list preferences that they take into account when assigning students to schools (sometimes using lotteries when a school is oversubscribed). Cambridge, Massachusetts, implemented a school choice program as early as 1980. Today, Boston, Chicago, Milwaukee, New Orleans, New York, San Francisco, and many other districts have choice systems. Many districts aim to have a “portfolio” of differentiated schools that appeal to different parents and students. Some areas also have interdistrict choice, including, for example, the entire state of Minnesota. Here students can cross district boundaries as long as school spaces are available. Residential choice, intradistrict choice, and interdistrict choice all maintain district governance and the combination of political and administrative accountability described in chapter 6. Yet other forms of school choice work outside the district system. Home schooling and private schooling provide alternatives to district-led school systems— though, like residential choice, these choices are limited to families with the resources (fi nancial or other) to take up these opportunities. Two additional forms of choice—school vouchers and charter schools (or similar semiautonomous schools of choice outside of the United States)— use public funds to support schools outside traditional school districts. Vouchers provide funds for families to choose private schools, while charter schools are publicly funded schools with a more autonomous governance structure than traditional public schools. Charter schools can be single schools or groups of schools, both of which can be, and in some states are, considered their own district, though subject to different rules from traditional school districts. We focus on charter schools for several reasons. In addition to being the fastest-growing kind of school of choice in the United States, they resemble those in other countries more than do voucher-supported private

134

Chapter 7

schools. Whereas “open enrollment” forms of school choice do not fundamentally change the governance structure for schools and voucher programs rely on private school governance, charter schools are an intermediate case: control over resources moves from the state or district to the school level but without the abandonment of public oversight. Because oversight is retained, charter schools in the United States (like their analogues in several other countries) have developed in tandem with a policy focus on outcomes-based accountability, which typically binds them far more than would be the case for voucher-supported private schools. Despite the fact that they are a relatively recent phenomenon, charter schools elicit questions of choice and local autonomy that have a long history in education policy discussions. So the policy decisions and the ways of thinking about them that we discuss below are internationally applicable to a range of schools of this type and to a broad set of education policy concerns.

Charter Schools Under traditional governance arrangements, the government (whether it is a local government or a higher-level government) establishes, funds, and runs public schools and assigns children to those schools, usually by neighborhood. To get the desired outcomes, governments regulate how schools behave—by determining funding, teacher licensing requirements, class size, curriculum, or other aspects of schooling. But the democratic process favors majorities over minorities and, especially when school boards are elected separately from other parts of local government, it is likely to favor well- organized groups of voters. And because parents with the resources can choose to live in neighborhoods with schools they prefer, schools that serve more affluent neighborhoods are subject to more market pressure than those serving relatively disadvantaged students. Recent US reforms have sought to change this traditional model by expanding parental choice of schools, by allowing schools to enter the public system subject to less regulation, and by increasing outcomesbased accountability. Charter schools are run by nongovernmental organizations that must obtain a charter agreement from an authorized chartering agency. They are funded by taxes and are not allowed to charge additional tuition. While charter agreements usually exempt the

School Autonomy and Parental Choice

135

school from many of the input regulations that apply to traditional public schools, such as the requirement to hire certified teachers or to maintain a minimum class size, agreements typically require charters to meet performance goals or risk closure. Charter authorizers vary between states and include school districts, universities, state boards of education, and independent, government-appointed, charter-authorizing bodies. Charter applications come from various groups of parents, teachers, universities, nonprofit organizations, and, in states that permit it, forprofit companies. Charter schools are schools of choice to which families voluntarily apply. If oversubscribed, the schools must select students by lottery.1 In 1991, Minnesota was the fi rst state to introduce chartering legislation, with its fi rst charter schools opening the following year. Public debate before the introduction of that legislation focused on the value both of providing for needs that were not well met in the regular public schools and of allowing for educational innovation. Among the fi rst three schools was a Montessori school. Whereas the initial advocates for charter schools focused on creating more diverse alternatives within the public system, in recent years different arguments have been advanced. Supporters of such schools share the view that regulation alone has not done well at getting schools to produce the desired outcomes. Although this impetus is similar to that for increased test-based accountability for traditional public schools, the theory of action behind the policies is somewhat different. Charter schools aim to increase flexibility within a system of outcomes-based accountability. As defi ned by the California Charter Schools Association, for example, charters are “public schools of choice that serve all student populations, are tuition-free, and have more flexibility and greater accountability.” 2 From the 2003/4 to the 2013/14 school years, enrollment in charter schools in the United States increased from 0.8 million to 2.5 million students in approximately sixty-five hundred charter schools. This enrollment represents only a small fraction (5.1 percent) of the approximately 49.5 million students in US public schools. 3 It is somewhat more than the homeschooled population (1.8 million in 20124) and is substantially smaller than private school enrollment of 5.4 million students (10 percent of the elementary and secondary student population) in 2013/14. 5 Three kinds of operators run charter schools. Most are run by individual independent operators, which currently operate approximately 64 percent of all charters.6 Education management organizations (EMOs) run

136

Chapter 7

about 15 percent of the schools.7 EMOs are for-profit organizations that provide a range of services to the individual schools they operate. Some give their schools substantial autonomy while others implement common policies for curriculum and instruction. A third kind of operator, and the type that has garnered particularly wide attention, are charter management organizations (CMOs), which are nonprofit organizations that are similar in many ways to EMOs. Among the many CMOs, the Knowledge Is Power Program (KIPP), Uncommon Schools, and Achievement First charter schools are perhaps the best known. KIPP currently runs two hundred schools in twenty states and the District of Columbia, serving almost eighty thousand students;8 Uncommon Schools runs fortynine schools in six cities;9 and Achievement First runs thirty-two schools in three states.10 These CMOs share a “no-excuses” approach to schooling. This approach is characterized by a culture of high expectations for every student, a strict behavioral and disciplinary code, increased instructional time, high-dosage tutoring, frequent use of data to inform instruction, and a college preparatory curriculum for all students.11 While CMOs make up only between 20 and 25 percent of charter schools, they are both the most visible and the most rapidly growing sector. In 2009/10, CMOs served 237,591 students. Two years later, 445,052 students attended CMOrun schools.12 From a policy perspective, the most obvious question often appears to be what are the effects of a specific charter school, or type of school, on the children who attend. Certainly a good deal of public debate is conducted at that level. But a more important policy question is how will charter schools affect the system of which they are a part. Think of members of a chartering authority deciding whether to accept a proposal for a particular charter school. Ideally, the authority members will want to consider the probable effects not only on the children who are likely to attend but also on the surrounding schools. Think of different decision makers—legislators deciding whether to introduce charter- enabling legislation, and if so, what particular features that legislation should have. They should make the decision that will have the most positive effect on the performance of the entire state’s school system. The evidence could be that some kinds of charter legislation will reduce overall performance even though the charter schools themselves improve outcomes for the children who attend. The very same evidence could simultaneously suggest that different kinds of charter legislation would improve the performance of the system as a whole. The effects

School Autonomy and Parental Choice

137

of some particular version of charter legislation might vary by state. If Florida’s public schools are much lower quality than New Hampshire’s, or if the possible suppliers of new schools in Florida are much stronger than those in New Hampshire, then chartering legislation that would weaken New Hampshire’s public system might improve Florida’s. This distinction between the effects on the children who attend a particular kind of school and the effects on the system as a whole is important in the following discussion, as are the differences between kinds of charter schools and kinds of chartering legislation. Identifying the Main Values In Play The main values we are interested in are what we have called educational goods—the knowledge, skills, attitudes, and dispositions that enable individuals to flourish and to contribute to the flourishing of others—and how they are distributed. As we have explained, however, other values also play a role in education decision making. These independent values include childhood goods, parents’ interests, respect for the democratic process, and freedom of residential and occupational choice. All these independent values are relevant when considering charter schools. The intrinsic value of childhood comes into play, for example, in discussions of the “no- excuses” schools, when questions are raised about whether children should have more play time and more variety of work. Freedom of choice and parents’ interests are also obviously and directly relevant as charter schools expand options for parents. We return to these independent values in the fi nal section of this chapter. We begin, however, with educational goods and their distribution. Educational Goods In chapter 1 we translated the general concept of educational goods into various capacities that contribute to the ultimate goal of flourishing. These are the capacities for economic productivity, personal autonomy, democratic competence, treating others as equals, healthy interpersonal relationships, and personal fulfi llment. The introduction and expansion of charter schools has the potential to increase the overall level of educational goods, but the extent to which it does so, and which particular goods are produced, will depend on many policy decisions. There are four pathways or mechanisms by which charter schools might generate more educational goods: (1) school differentiation could

138

Chapter 7

facilitate greater matching of schools and students, (2) self-governance and freer entry of schools combined with public oversight and accountability could improve school leadership and attract high- quality teachers, (3) experimentation could generate innovations within the charter sector that would then be transferred to traditional public schools, and (4) incentives to compete for students could make both charter schools and traditional public schools more attentive to the educational needs of their students. While the logic of the charter school model suggests that such schools may generate more educational goods, each of the four mechanisms has potential drawbacks as well. Classroom activity is opaque to administrators, let alone to parents, and choice may not benefit students if parents are poor judges of what their children need to flourish or have poor understanding of what actually happens in the classroom. To the extent that charter schools are segregated by race or socioeconomic status, they may reduce students’ capacity to treat others as moral equals. Greater autonomy may reduce rather than improve schools’ quality if school leaders lack capacity or if the authorizing agencies do a poor job of evaluating charter applications. Experimentation may not lead to progress if innovations go unnoticed or are difficult to transfer to other schools. Finally, incentives to compete may distract educators from their educational mission or induce them to allocate resources in ways that attract families but do little to improve school quality. So empirical evidence is needed to assess the extent to which charter schools increase educational goods. The mix of educational goods is also at stake largely because of the two accountability mechanisms that apply to charter schools. Public oversight in the form of test-based accountability, for example, gives charter schools incentives to develop the cognitive skills measured by tests, possibly at the expense of other educational goods. Accountability through parental choice provides incentives of another type. Parents, for example, may care most about some educational goods (e.g., those that contribute to the subsequent economic productivity of their children) and less about other goods (e.g., those that support democratic competence). We refer the reader to our more detailed discussion in chapter 6 of how accountability regimes affect the mix of goods. Here we simply note that charter school reforms change the governance of schools and the accountability regimes in which schools operate. As a result they can influence the mix of educational goods.

School Autonomy and Parental Choice

139

Distributive Values Charter schools can also influence how educational goods are distributed. They can affect the variation in the level of educational goods between individual students and between groups of students. All of our distributive principles—adequacy, equality, and benefiting the less advantaged—are in play in discussions about charter schools. The lighter regulation of charter schools could lead to greater variation in school quality across schools and thus more inequality in the distribution of educational goods across students. At least in theory, regulations set a floor on quality by limiting class sizes, by requiring teacher certification and training, and through a range of other mechanisms. The autonomy given to charter schools could potentially generate some very good schools that would not have existed if subject to standard regulations. But it may also lead to some schools that make poor choices in allocating resources, resulting in consistently large and ineffective classrooms, or that hire uncertified teachers without the skills and knowledge necessary to support students and help them learn. This differentiation of schools would be beneficial to students who end up in successful schools but detrimental to those who end up in experiments that failed. At stake here are both whether the weaker schools meet an adequacy standard and whether the differentiation generates inequality of educational goods across students. The downsides of reduced-input regulation may be mitigated in part by public oversight in the form of sanctioning or closing ineffective charter schools, but this safety mechanism relies heavily on both the actions of chartering authorities and the measures of effectiveness (e.g., performance on state standardized tests) used to capture the failures of poor schools. Parental choice could also mitigate the potential disadvantages of deregulation if parents withdraw their students from poor schools. In theory, this greater role for chartering authorities and parents could reduce the range of quality by leading very poor schools to close, effectively setting a higher floor on quality than past input regulations on traditional public schools have achieved. These processes can lead to changes in the overall distribution of school quality, but they can also affect which students attend the more effective schools. Before the introduction of charter schools, parents with the resources to move across school districts or access private schools had far greater choice of schools than did parents without

140

Chapter 7

these resources. Presumably the increased choice associated with charter schools will benefit most those families who had least choice before reform. As a result, we would expect any positive (or negative) effects of charter schools to be concentrated among lower-income families. If these effects are positive, then the policy would reduce inequality of educational goods between students from different income groups. Charter schools could also affect within-group inequalities. These within-group effects stem largely from differences in families’ use of the choices available to them. Among families with limited financial resources, those with greater ability to understand the choices that they face and more interest in and time to invest in their child’s education may be more likely to make use of charter schools. In a system pervaded by charter schools (such as New Orleans), such families are likely to make better choices. These differences could lead to gains for the better- off children within the group of low-income families. This withingroup increase in variation, driven by benefits to the relatively advantaged, may not be of particular concern if between-group inequality decreases. However, the movement of relatively more advantaged students into the charter school sector could adversely affect the students remaining in the noncharter public schools by removing the most active parents and potentially reducing positive classroom-level peer effects. Similarly, the less informed families that do make choices may end up in schools that are worse than their traditional public schools. These changes in the within-group distribution of educational goods would be problematic for our distributive values. Identifying the Key Decisions Relevant to Those Values Charter school policies involve many decisions: which institutions have chartering authority, what kinds of organizations are allowed to run schools (individual operators, nonprofit CMOs and/or for-profit EMOs), how much freedom charter schools have when admitting students. We cannot address the full range of choices in this short chapter but instead focus on three that are likely to be particularly important for the values we have identified. Scale. Should policy makers deliberately keep the charter school sector small relative to the number of traditional public schools?

School Autonomy and Parental Choice

141

Regulation. Should most of the regulations that apply to public schools be waived for charter schools?

Public oversight. Should the education system aggressively hold charter schools accountable for test performance and other measures of student outcomes or school processes?

The Scale of Charters Although currently only about 5 percent of all US students attend charter schools, this figure can be misleading because charter school students are unevenly distributed across districts. Many districts have none, while in 2014/15, such students accounted for more than 10 percent of public school students in more than 160 districts nationwide, more than 40 percent in Kansas City, Missouri; Washington, DC; Flint, Michigan; and Detroit, Michigan, and more than 90 percent in New Orleans.13 These differences reflect different decisions by policy makers about whether to allow charters at all and, if so, on what scale. Should charters be limited to a small subset of schools operating largely at the fringe of the public school system? Or should an entire public school system, or a large part of it, be converted into a system of charter schools? The number of charter schools in a given area will reflect both the supply of organizations interested in running these schools and direct policy decisions about how many charter schools to allow. Even policy makers who are enthusiastic about the potential effects of charter schools may choose to limit the scale. Their choice will depend on the mechanism by which they believe charter schools would benefit students. Consider the possibility that current input regulations that apply to traditional public schools are significantly detrimental to students. This could be because the regulations keep potential excellent teachers or school leaders out of schools and restrict local educators from responding to opportunities for improvement and students’ needs. In this situation, the deregulation inherent in the charter school approach has the potential to generate more educational goods, which would argue in favor of a large charter school sector. Similarly, suppose that parental choice is a strong force for school improvement, as it might be if parents have and use good information about the quality of schools when deciding where to send their children. In this case, charter school policies, especially those that allow eas-

142

Chapter 7

ier entry of new schools, would generate benefits by increasing parental choice. Again, this would lead policy makers toward a large charter school sector. Alternatively, consider the suggestion that the main benefit of charter schools stems not from greater efficiency or more parental choice but from experimentation that generates some successes and some failures. The systemic effect of experimentation would come from the spread of new and effective educational approaches from the charter system to other schools. In addition to generating benefits in the form of good new ideas, however, experimentation of this type may well lead to great variation in the quality of charter schools. As a result, policy makers will want to balance the benefits of innovation against the costs of greater variation across schools in the development of educational goods. In this case, policy makers may choose a small charter sector in order to limit the negative distributive effect while getting some of the benefits of experimentation. The Regulation of Charters If they believe the current regulations on traditional public schools usefully set a floor on quality but are detrimental to many students, perhaps because they limit innovation, policy makers may choose to reduce those regulations through the use of charter schools. Even if current regulations have negative effects, on average they still reduce the probability of unacceptably bad schools. This floor undoubtedly benefits some students, particularly the least advantaged, who are more likely to end up in the worst schools. In this case, policy makers’ decisions about the scope of regulations may depend on the scale of the charter sector. If the charter sector is large, then they will want to be careful about which regulations they remove. If they remove too many, a large number of students may end up in schools below the floor that regulations would have established. However, if the charter school sector is small and the benefits of charters arise largely through experimentation, then policy makers may want to impose fewer regulations on charter schools to give them maximum opportunity to innovate. The Public Oversight of Charters Finally, we address the policy choice of how much oversight of charter schools is warranted. Again, the choice depends on the mechanisms in play. If parents are a strong force for school improvement, then there is less need for state oversight.

School Autonomy and Parental Choice

143

Parents have a much broader view of their children than the state does and can consider children’s happiness, social skills, and performance across disciplines. Often the state is limited to consideration of easily measured qualities such as math and language performance. While parents may value a broader array of student outcomes, however, they may be less able to assess the contribution of the school to those outcomes or how well their child would be likely to do in other schools. As a result, parents may not make good choices, and thus parental choice may be a weak force for school improvement. Parents are also likely to be motivated much more by the effects of a choice on their child than by spillover costs or benefits it might have for other students or other members of society. To the extent that this is the case, policy makers should choose greater oversight of schools, although not necessarily in the form of test-based measures. Assessing the Options in Light of the Values and Evidence What decisions policy makers should make on these questions concerning scale, regulation, and oversight will depend on the mechanisms through which charter schools affect student outcomes and how the choices are interrelated. Suppose the evidence indicates that governance changes in the form of greater autonomy and accountability and freer entry of new schools lead to better local decision making than in the more highly regulated, traditional system, with the outcome being higher average school quality with no meaningful increase in the variation in school quality. Then the policy maker may rightly choose a largescale system with little regulation but well- developed public oversight in the form of outcome or process-based accountability, such as the system that the city of New Orleans has tried to develop.14 Alternatively, if the evidence shows that charter schools improve student outcomes on average but also increase the variation in quality across schools, policy makers may still choose a large system but with stronger regulations designed to minimize the variation that comes with a deregulated, decentralized system of schools. Finally, suppose that local autonomy and decision making do little to raise student outcomes on average but instead generate some schools that are very ineffective and others that are excellent and innovative and are based on models that can be scaled and transferred to traditional schools. In this case the policy maker may

144

Chapter 7

choose a small charter sector with few regulations, which allows for experimentation while minimizing the number of students attending lowquality schools. This section considers the evidence on whether and how charter schools are affecting students and uses it to inform the three specific but interconnected decisions about scale, regulation, and oversight. Average Performance of Charters The fi rst relevant consideration is whether charter schools increase the educational goods of the students they educate. We have a fair amount of evidence about the effects of charter schools on student test performance, but far less about its effects on other educational goods. Studies looking only at the students who switch from one sector to the other, and compare their learning when they are in charter schools to their learning in traditional schools, conclude that on average students in charter schools perform approximately the same as similar students in traditional public schools.15 Other research designs that compare students in charters to similar students in traditional public schools, including studies that focus only on charter schools run by CMOs, come to the same conclusion, fi nding little detectable difference in test performance, on average.16 At the same time some states and cities have systematically more effective charter schools and some have systematically less effective schools.17 These differences could, in theory, be driven by differences in the quality of traditional public schools across states and cities. However, some places, such as Boston, that have relatively highperforming traditional public schools have charter schools that are even more successful with the types of students they enroll than the traditional public schools.18 In contrast to the evidence concerning average test performance, far less is known about educational goods that tests fail to measure. One possible source of information—at least about the goods that parents care about—is parent satisfaction surveys. In general, parents of children in charter schools report greater satisfaction than those in traditional schools, and such parents report being more satisfied when their children are in charter schools than when their children were in traditional schools.19 An alternative approach that compares return rates of students to the different kinds of school generates similar conclusions about parental satisfaction but also suggests that, in areas where charter schools are more highly segregated than traditional public schools, par-

School Autonomy and Parental Choice

145

ents of children in mainly white charter schools are more satisfied than those in mainly black ones. 20 More generally, it is possible that parents in some cases care more about the composition of the student body than about the development of educational goods, or wrongly take the former as a proxy for the latter. Moreover, greater satisfaction may stem primarily from the choice process itself rather than from charter schools’ providing students with greater opportunities to develop educational goods that parents value. Charter schools may also affect educational goods that are neither well measured by test scores nor valued by parents. Consider the knowledge, skills, attitudes and dispositions needed for democratic competence. Here the evidence is mixed. One proxy for these goods might be educational attainment: higher attainment seems to help students develop a range of educational goods related to health, voter participation and even happiness. 21 The small research base on the average effects of charter schools on attainment provides some evidence of positive effects. One study of students in Florida and Chicago fi nds that charter school students are 8–10 percent more likely to attend college than observably similar students who attended charter schools in middle school but not in high school. 22 Another showed that one charter school significantly improved students’ likelihood of attending college and decreased their probability of teen pregnancy and incarceration. 23 However, research on Texas charter schools found that, while “no- excuses” charter schools increased four-year college enrollment, other types of charter schools decreased it. 24 Overall, the evidence is far from defi nitive. Another possible proxy for the tendency to promote democratic competence is the level of racial or economic diversity within charter schools. There is some evidence, for example, that diversity in a group leads to more innovative and complex thinking. 25 Moreover such diversity may support the development of the capacity to regard others as equals, an important educational good, by making students more tolerant. 26 Studies of how charters affect segregation do not produce a clear picture. Charter schools are typically more racially segregated than traditional public schools, but they also serve students who would have attended more segregated traditional public schools had charter schools not been available. 27 In some large cities, charters appear to decrease racial segregation as black students moved from schools with very high proportions of their own race to charters with lower proportions. For example, one study in Chicago showed that students who moved from

146

Chapter 7

traditional public schools to charter schools had a decrease of 6 percentage points (90 to 84 percent) in the proportion of black students in their school and in Milwaukee students had a decrease of 8 percentage points (73 to 65 percent). 28 In many other cities or states, however, black students moved to schools with greater shares of their own race. This increase was 9 percentage points in Denver (42 to 51 percent), 3 in Philadelphia (84 to 87 percent), 8 in San Diego (25 to 34 percent), 5 in Ohio (74 to 79  percent), and 15 in Texas (52 to 67 percent), 29 with even larger increases in North Carolina and Texas. 30 Recent work on North Carolina demonstrates that the proportion of charter school students who attend racially segregated schools—that is those that mainly serve white students (e.g., less than 20  percent minority) or mainly serve minority students (e.g., greater than 80 percent minority)—is large and has been growing significantly. 31 The extent to which these differences matter for the development of democratic competences, especially in large urban areas where segregation is already substantial, remains unclear. Effects on Other Schools The evidence about whether charter schools increase the full range of educational goods for their own students, then, is sparse, though growing. But from a systemic perspective, it is also important to determine how the existence of such schools affects students beyond their walls. Those effects could be positive if competition induces other local schools to become more effective. They could be negative if charter schools create disruption for local schools, 32 attract away local schools’ most involved parents, or reduce the funds available to local schools. This last concern arises because charter schools typically serve lower proportions of the students who are most expensive to educate, leaving such students in greater concentrations in the traditional public schools. 33 Alternatively, charter schools could have positive effects on nonlocal traditional public schools through innovation if they establish new and effective educational approaches that are implemented on a wider scale. We address each of these separately. A number of studies have tried to measure how charter schools have affected the performance of traditional local schools. Some fi nd no effect of charters, such as studies in North Carolina and Michigan. 34 Some find negative effects. Research on a southwestern school district, for example, found evidence that increased charter school presence in an area is associated with a decline in test scores in the area’s public, noncharter elementary schools, especially in the year after charter penetration

School Autonomy and Parental Choice

147

increases (but less so in later years). 35 By contrast, two other studies, in Texas and Florida, have shown some positive effects on the math performance of students who remain in traditional public schools. 36 Overall, the results are mixed. In general, charter schools do not appear to have had much of an effect on the performance of nearby noncharter school performance. This might be for a number of reasons. Most of the studies have looked at areas where there are relatively few charter schools; as the numbers grow, their effect on traditional schools (either positive or negative) may increase. Perhaps charters are attracting parents by providing benefits, such as a more congenial or advantaged peer group, that are not transferable to traditional public schools. Alternatively, competition may simply not be a strong force for change. Traditional schools may not have the ability to respond—because of either capacity or regulatory constraints—or they may do so in ways designed to attract more students, such as improving their marketing, rather than by improving their program offerings. 37 Charter schools may affect students in noncharter schools through innovation as well as through competition or disruption. They may develop approaches that lead to improvements in schools anywhere, not necessarily locally. It is difficult to assess the innovation effects of charters. Not only must a charter be unusually effective but also its innovation must spread. There is some evidence of the fi rst step in this process. While on average charter schools appear to be no better than traditional public schools at supporting students’ learning, some have been remarkably successful at improving students’ test performance. 38 For example, one study of charter schools in Boston found that middle school students who were accepted by lottery scored substantially higher in math, English-language arts, and writing than those who were not. If replicated for all children, the improvements were sufficient to put a meaningful dent in the black-white achievement gap. 39 High-performing charter schools—such as those in Boston, as well as KIPP (Knowledge is Power Program) and the Harlem Children’s Zone schools—often use the “no- excuses” approach with its high expectations of every student and a strict behavioral and disciplinary code.40 Charter schools have innovated in other ways, such as broader use of technology and different allocations of teachers to classes (e.g., Rocketship Education), but evidence about the effectiveness of these innovations is limited. To affect noncharter students through innovation, the charters must not only be unusually effective but their innovation must be scalable and

148

Chapter 7

effective for a wider group of students. Evidence about the scalability of KIPP-type programs gives some grounds for optimism. A study in Houston introduced five reforms into low-performing schools that are common in successful charter schools: increased instructional time, a more rigorous approach to building human capital, high- dosage tutoring, frequent use of data to inform instruction, and a culture of high expectations. While these are not new ideas, the charter sector was able to implement them in combination and with intensity in a way rarely seen elsewhere. The study found gains in both reading and math as a result of these changes.41 At the same time, however, it illustrated the difficulties any school system would face in implementing such policies on a large scale, given the structure of the school week and the need for highquality teachers willing to work such long hours. KIPP, for example, depends heavily on young teachers who, at this early stage in their careers, work for lower salaries and are less likely to have family commitments than the average teacher. KIPP schools tend to have high turnover rates for teachers (27 percent in 2015),42 though on average, the turnover rate for charter school teachers is only slightly above that for traditional public school teachers (18.5 percent vs. 15.6 percent in 2012/13).43 This study highlights the potential for charter school reforms to benefit traditional public schools but also the difficulties of doing so. To date we know of no evidence that innovations like these have spread on a large scale. A possible drawback of “no- excuses” programs concerns their effects on other educational goods and on some of the childhood goods we described in chapter 2. While these schools show large improvements in student achievement, there are questions about whether they provide a fully rounded education and whether students are losing out in other ways because of the extended time in school and the regimented nature of the program. To be clear, for many decision-making purposes, the right comparison is not with an unrealistic ideal in which students enjoy childhood goods in abundance but with the reality that they would otherwise be dealing with. Perhaps those goods would be undermined in other ways, such as by disruptive classroom behavior and lax discipline in the hallways. One study found that KIPP schools had positive effects not only on achievement but also on parents’ and students’ attitudes toward school, though it also found that attending a KIPP school increased student-reported behavior problems such as lying to or arguing with parents. It found no effect on parent-reported behavior problems or on other student behaviors, such as self-reported self- control,

School Autonomy and Parental Choice

149

academic self- concept, school engagement, effort/persistence in school, educational aspirations, and illegal activities.44 This evidence suggests that increased achievement at KIPP need not come at the cost of other valued outcomes. However, the families that have chosen KIPP schools are likely to differ from those in other schools, and the type of intensive education that increases their satisfaction may not do the same for a broader set of parents and students. Overall, charter schools with extended time and a culture of high expectations show the potential for charter schools to innovate, and for those innovations to spread, at least within the charter school sector. As these schools have grown and attracted attention, others have emulated their approach—perhaps because chartering agencies see this type of school as their best option for students to improve on state tests. To the extent that these schools start to dominate the charter market, charters may become less innovative, and the potential of charter schools to benefit other schools through innovation may decrease. Whether the process of innovation will expand to other educational approaches is less clear. The Effect of Charters on the Distribution of Educational Goods Suppose that charter schools do not improve the overall level of educational goods. They might nonetheless contribute to distributive values and be supported for that reason. Conversely, even if they do increase the production of educational goods, it would count against them if they worked against distributive goals. We focus here on the value of equalizing educational goods and that of benefiting the less advantaged. School autonomy is likely to lead to greater variation in school effectiveness that, in turn, leads to greater variation in the educational opportunities available to students. On average charter schools are no more effective than traditional public schools, and some charter schools are substantially more effective than their local schools, which means that some charter schools are substantially less effective. One national study found that 31 percent of charter schools posted math gains that were significantly below what their students would have seen if they had enrolled in local traditional public schools.45 A number of studies have identified this increased variation in effectiveness in the charter sector.46 Freer entry of new schools may bring in some that are more effective, some less so, depending on the supply of potentially good school leaders interested in running schools and on the skills and goals of the chartering authorities. As we might expect, charters are more likely than traditional pub-

150

Chapter 7

lic schools to focus their instructional approach on specific missions.47 Some schools focus on academic achievement, some on cultural traditions, some on social or artistic endeavors. This greater variation is likely to affect the distribution of education goods. The greater variation in effectiveness could either increase or decrease differences across groups. One of the initial purposes of charter schools was to expand the schooling options for low-income families, but a major concern has been that they may provide less benefit to lowincome families than to high-income families because the latter are in a better position to exercise choice. In that case, we might expect charter schools to increase the already large differences in opportunities between high-income and low-income families. A few careful studies suggest that at least in some areas, urban charter schools serving low-income students are having the best results, at least as measured by student test scores. Research on Boston, for example, fi nds far fewer effective charters in suburban areas than in urban areas.48 Similarly, a 2013 study by the Center for Research on Education Outcomes (CREDO), covering most charter schools in the country, fi nds more positive effects for black students in poverty than for other students.49 Part of the reason for these differences may be that the demand for charters in urban areas comes from families that are interested in the outcomes measured by state tests, while those in suburban areas are more interested in other outcomes. In any event, the Boston and CREDO research provides little support for the conclusion that charters are exacerbating the achievement differences between students in highincome and low-income areas. Not all studies show these positive results for disadvantaged communities. In North Carolina, for example, the evidence shows that charter schools have increased their share of white students relative to minority students and have increasingly engaged in “cream skimming”: students who select into charter schools tend to be more able (as measured by their previous year’s test scores) or better behaved (as measured by their previous year’s absence rates) than their public school peers. Moreover, the schools serving mainly white students are more successful in satisfying the expectations of the parents than are the charter schools serving mainly minority students. 50 While some individual charter schools may be effectively meeting the needs of minority students, the sector as a whole in North Carolina is increasingly educating white, middle- class

School Autonomy and Parental Choice

151

students, which may exacerbate inequalities between groups defi ned by race if charter schools are more effective. Charters may also increase inequality within low-income communities because of the limited supply of high- quality charter schools and the fact that some low-income families are better able to navigate the system than others. Providing better opportunities for some low-income families may benefit the less advantaged on average even if these opportunities are not available to all low-income families (e.g., only to those who apply for and win lotteries to the best schools) and even if some children actually end up in lower- quality schools because of the wider variety of quality among charters. However, the most disadvantaged among the disadvantaged students—those for whom their parents’ nonapplication to the schools reflects lack of information, interest, or means, and those who are excluded from charters for disciplinary reasons—are likely to be worse off, especially if the charter schools reduce the public funding going to traditional public schools. Considering the Choices How does the evidence help us think about our three policy choices? Should policy makers explicitly limit the scale of the charter school sector to be small relative to the number of traditional public schools? Should most of the regulations that public schools have been subject to be waived for charter schools? And should the education system aggressively hold charter schools accountable for test performance and other measures of student outcomes or school processes? A large charter school sector would be warranted if such schools were systematically more effective and were not significantly more variable than traditional public schools. The extant research shows that neither of these is true, at least on average. More precisely, charter schools have been on average very similar to traditional public schools in their effectiveness at improving student achievement, and they have tended to be more variable, with some substantially better and others substantially worse than the average. They have expanded the options for parents— and thus parents’ ability to make choices for their children—but they have also tended to increase racial segregation across schools. These results call for caution on the large expansion of charter schools and for consideration of the context in which they might be introduced—such as the quality of the traditional schools and the capacity of chartering agencies to provide oversight. Because some states have implemented poli-

152

Chapter 7

cies that have resulted in charter schools that are more effective on average than traditional public schools, a case could be made for large-scale expansion but only if certain conditions are met. Some research suggests that strong oversight—including careful chartering and the willingness to close low-performing schools—may lead to a successful charter school sector, but there is limited evidence about what conditions are necessary to produce that result. The decision about the appropriate level of regulation is closely tied to that about scale. Given the evidence that charters have not been consistently better at developing students’ educational goods and are more variable in quality, relatively strong regulation might be worthwhile if the choice were to have a very large charter sector. However, the greater variation in the charter sector due to deregulation has resulted in some excellent schools. If the goal of the sector is experimentation and learning, then the combination of a relatively small sector and weak regulation would be appropriate. As the sector gets bigger, the greater variation that it produces in student outcomes affects an increasing share of the population, and considerations about distributive values become increasingly important. Overall, as discussed above, there is a trade- off between, on the one hand, reducing variation and setting a floor on charter school quality through regulation and, on the other, increasing the variability to allow for both better matching of students to schools and positive innovation. This trade- off lies at the center of many decisions about charter school design, including how to regulate these schools and how to hold them accountable for their performance. Public oversight can reduce the variation in charter school quality. Chartering authorities can become more careful about the schools they charter and more willing to close down poorly functioning schools. More careful chartering is likely to reduce the potential for innovation, one of the main ways in which charters may generate higher levels of educational goods in the long run. More oversight of operating charters, including closing down low-performing schools, is a clear alternative mechanism, and some research provides evidence that closing schools does improve the quality of the sector. 51 The importance of this mechanism, like regulation, increases as the size of the sector grows and more students are subject to poor schools. The problem with it is similar to the disadvantages of administrative accountability discussed in chapter 6: imperfect measures of quality can lead to inaccurate assessments of schools.

School Autonomy and Parental Choice

153

This discussion of questions of scale, regulation, and oversight is primarily illustrative. Policy makers would doubtless want a fuller consideration of these questions, including an assessment of local capacities. What we have said does, however, exemplify our framework: we have identified the relevant values in play and brought to bear the available evidence about how the choices would affect those values. We leave the fi nal step—weighing these considerations in order to make a decision—to decision makers.

Conclusion In this chapter we highlight the importance of school governance for the educational opportunities available to students and the level of educational goods they obtain. Schools are complicated undertakings. Policy makers, voters, and educators do not agree on the best educational approaches. Indeed, different approaches may be best for different students, and the best approach may change as new technologies become available. In addition, policy makers must always be considering which students they are trying to serve, balancing the value of increasing the overall level of educational goods against that of prioritizing the less advantaged and taking into consideration constraints imposed by the political and market power of more advantaged families. The complexity and changing nature of schooling requires a continuous series of decisions about how best to provide schools. Much of the debate about whether charter schools are a good policy choice depends on whether this form of governance is more likely to realize valued outcomes than the alternatives. Charter schools also illustrate how policies can affect educational variation within normatively important groups as well as across them. How detrimental this increased variation is depends, at least in part, on the context. Charter schools to date have been more variable in their effectiveness at improving student outcomes than traditional public schools in their community. In some urban areas charter schools have proven more effective than the local public schools. These differences introduce greater variation in outcomes for students. If access to higherquality schools has a lot of positive spillovers, then this greater variation may be beneficial overall even if the variation itself makes the distribution of educational goods more unequal.

154

Chapter 7

Finally, charter schools, like the accountability systems discussed in the previous chapter, point to trade- offs among different educational goods. Within the charter school sector, parents and school leaders have greater power over their schools, at least theoretically, while voters, again at least theoretically, have greater influence over traditional public schools through school boards, school site councils, and other institutions. As a result, charters introduce a possible trade- off between the values held by parents and those of voters. Parents are more likely to be interested in educational goods that have direct payoffs for their children rather than spillovers to the general public. School leaders within charters, in turn, have incentives to respond both to parents and to the typically narrow set of test-based measures for which the school is held accountable. Voters and their elected representatives, in contrast, may well value a broader set of educational goods largely because they are in a position to take account of the whole system rather than individual schools. The extent to which they do so may vary across districts and states depending on the size of local districts, the role of the state in education policy, and the culture of the state.

Conclusion

E

ducational decision making involves value judgments. As policy makers and educational practitioners aim for improvements, they need standards to determine what counts as an improvement. Similarly, researchers aiming to provide decision makers with useful information need standards to guide their research questions and choice of outcome measures. Yet both groups typically lack a rich and sophisticated language for talking about values and articulating trade- offs. Impoverished vocabulary can lead to imprecise thought and misguided choices. Researchers may examine the effects of policies on only those outcomes that are most easily measured. Decision makers may base decisions on only the most readily available research. Both may fail to consider the full range of valued outcomes. Our main goal is to enable decision makers to make better decisions in light of their values. We do that by enriching the language available both to them and to the researchers whose work informs their deliberations, by offering a framework for thinking about the goals of education, and by proposing a procedure for making trade- offs among them. We make no attempt to address the collective aspects of decision-making processes, which inevitably involve strategic interactions among the interested parties. Our goal is rather to put individual decision makers in a better position to participate in the broader policy-making process. Discussion of how individuals do or should interact to produce good policy would require an entirely different book.

156

Conclusion

Enriching the Language We have coined educational goods as a compact term for the broad goals of education. Educational goods are the knowledge, skills, attitudes, and dispositions that children develop both for their benefit and for the benefit of others. These goods are varied. In our view they are the capacities that enable individuals to be economically productive, act autonomously, participate effectively in democratic processes, gain personal fulfi llment, and treat others as equals. By identifying educational goods as valued outcomes for students and by articulating specific types of educational good, decision makers can systematically consider the tradeoffs both among different educational goods and between educational goods and other aims. Cognitive skills, as measured, for example, by math and Englishlanguage test scores, are educational goods. Cognitive skills open up labor-market opportunities and help individuals make better choices and support their own children’s development. But cognitive skills are clearly not all that matters. The ability to work with others and to enjoy fulfi lling personal relationships are also educational goods. They are valuable both because they contribute to the individual’s well-being and because they help individuals contribute to the well-being of others. A narrow focus on cognitive skills, and within cognitive skills only on math and language test performance, can sometimes lead to underinvestment in other educational goods, sacrificing long-term benefits for students and for the public more generally. Childhood experiences, including those related to schooling, do more than develop children’s capacity for subsequent flourishing. The wonder that children experience when engaging in new activities and new discoveries can bring them joy in the moment and as memories even if they play no developmental role. So can play and an aesthetically pleasing physical environment. If policy makers and educators ignore these experiences when making decisions about what kinds of schools to create, they are likely to be forfeiting important values. We have introduced a second term, childhood goods, to capture these valuable experiences and distinguish them from educational goods. Some childhood goods are specific to childhood and can only be gained at specific ages or stages of development. Others are quite similar to ben-

Conclusion

157

efits later in life. Policy makers might be willing to sacrifice some shortterm benefits for long-run gains but typically would not give up the former completely. The separate terms childhood goods and educational goods clarify these different kinds of values. We have drawn attention, especially in chapters 6 and 7, to some of the inevitable trade- offs between the immediately beneficial experiences of children and their development of capacities for longer-term flourishing.

Clarifying Distributive Values and Trade- Offs Decision makers typically care not only about the average level of educational goods that students acquire, or of childhood goods that children enjoy, but also about how these goods are distributed. Yet here again, the language commonly used to describe these distributive goals is often imprecise and unhelpful for decision-making purposes. Terms such as equity and social justice help to draw attention to distributive values but do little to illuminate what those values are or help decision makers think about the trade- offs between them. In this book we have focused on three distributive values that are sometimes in alignment and sometimes in opposition, but each specifies a distinct distributive aim. The value of equality leads one to seek to reduce inequalities between individuals, while adequacy is concerned rather with all reaching at least some threshold level—without concern for the potentially unequal distribution of those above the bar. Benefiting the less advantaged similarly ignores inequalities at the top of the distribution while focusing on improving the prospects for the worse off. One decision maker may place substantial weight on the fi rst, while another cares more about the second or the third. Some educational decisions may promote all three distributive goals while others require trade- offs among them; for example, a decision may require sacrificing some equality for the sake of benefiting the less advantaged. Specifying the values rather than accepting the vagueness of common terminology brings out the range of different distributive goals and makes clear the need for trade- offs, which enables decision makers to consider their choices systematically.

158

Conclusion

Accounting for Other Values Producing educational goods and childhood goods, and distributing them in valuable ways, tend to be the main aims of schooling systems and of educational decision making more broadly. But educational decisions are not made in a vacuum, and other values are affected by, and should in turn affect, those decisions. The importance of other values for education decisions may raise the concern that no policy maker or practitioner could ever assess all the relevant values carefully. We have suggested, however, that in practice only a small set of other values—which we have called independent values—tends to be relevant to most educational decision making. These include the value of parents’ interests in their children, respect for the democratic process, and freedom of choice of residence and occupation, as well as other goods such as food, housing, and entertainment. The limited size of this list means that decision makers are in a position to think systematically about the effects of their choices on these other values. Recognizing that manageably few values are regularly at stake in education decisions can bring clarity to education decision making. Respect for the democratic process is clearly relevant in school funding decisions, as is the value of other goods such as housing and entertainment. Parents’ interests are clearly salient in discussions of charter schools. Although we have not specifically addressed the policy issues involved in recruiting teachers and integrating schools, racially and economically, those decisions would clearly be affected by the value that decision makers place on adults’ freedom to choose their occupation and residential locations. Most policy makers would be unwilling, for example, to mandate residential integration in order to integrate schools or to force individuals to teach in particular schools in order to provide equally knowledgeable and effective teachers across schools. Instead, they would need to pursue their educational goals in ways that respected adults’ freedom of occupation and residence.

Combining Values and Evidence Clarifying the values at stake in educational decision making is a key aim of this book, but merely identifying values is not sufficient for effec-

Conclusion

159

tive choices. Decision makers must not only know what they care about but also how their decisions will affect valued outcomes. While empirical social science research will not be able to establish the precise effect of each choice on each value, it can provide evidence about likely effects. Our framework calls for assessment of these likely effects and consideration of the trade- offs that they illuminate. Of course, empirical evidence is only useful if it does, in fact, shed light on the likely effects of decisions on the values at stake. If social science research provides evidence of the effects of a choice on only one valuable outcome, say the development of math skills, it will be far less useful than research that gives information on a richer range of possible effects. For some decisions, math test scores and their distribution might, in fact, hardly be relevant at all. Social scientists can draw the same lessons as policy makers and practitioners from the framework we present. Clarification of the values that are significant in education decision making is also important for social science research. For decisions likely to affect different educational goods, decision makers need information on those different goods. For educational choices likely to affect educational goods and childhood goods, they need information about both kinds of good. For educational choices likely to have effects on parents’ interests or the consumption of other goods, decision makers need information on these factors too. Consciously or unconsciously, educational researchers are usually motivated by value concerns, however vague or diffuse. The framework presented in this book offers a way of sharpening those concerns and a vocabulary for expressing them. Existing research fi ndings can be interpreted and framed in the terms presented here. Moreover, once clear on the range of values at stake, researchers will be better able to focus their efforts on achieving those fi ndings most useful for decision makers Our framework emphasizes the importance of being clear about values and about the need to assess how effective specific interventions are likely to be in helping to realize them. Decision makers must be attuned to the challenges of both measurement and determining causation, and they must be able to make effective use of social science research. In some cases, the research is very specific to particular contexts, and one needs to be careful in generalizing it to other contexts. Thus, the decision maker’s evaluation of the options in light of the values at stake will require understanding of what to draw from existing evidence. What  is distinctive here is that the social science research is used not

160

Conclusion

simply to understand or predict effects, whether intended or unintended, but rather to determine the extent to which specific choices are likely to promote or respect the values at stake. If it is to be useful, empirical evidence must provide information about the implications of policies for those values. As we observed in chapter 4, these observations support a revision of the current enthusiasm for “data- driven” decision making. Evidence is essential for good decision making, but it cannot drive it. Values drive decisions, and while good decision making requires high- quality interpretation of high- quality evidence, even discerning what sort of research to seek and consult depends on knowing what one’s values are.

Using the Language and the Framework In chapters 4, 5, and 6, we used the examples of school fi nance, school accountability, and school autonomy and parental choice to illustrate how this framework can be applied and to shed light on the tensions involved in key policy decisions. In chapter 4, on school fi nance, we showed how the framework helps identify the trade- off between distributive goals and respect for democracy as well as the empirical difficulty of clearly identifying cost and need differences across schools and school districts. In chapter 5, on school accountability, we explained how it helps to identify potential trade- offs among different educational goods, and between specific educational goods and childhood goods, and we emphasized the empirical difficulty in identifying which knowledge, skills, and attitudes developed in school lead to future flourishing. In chapter 6, on school autonomy and parental choice, we highlighted the value of parents’ interest in their children and the importance of better assessing the effects of parental choice of school not only on students who themselves attend charter schools but also on other students who could be affected by the policy. The conclusions in these chapters may seem obvious once identified, but they are frequently left implicit, at best, in the policy-making process. Debates about accountability often focus exclusively on whether accountability policies have increased student test scores with little focus on other educational goods or childhood goods. Debates on school autonomy tend to concentrate almost entirely on whether students attending charter schools perform better rather than on either parents’ inter-

Conclusion

161

ests or more systemic effects. By systematically considering the breadth of educational goods, distributive issues, and other values commonly in play in education polices, we have identified issues that have received insufficient consideration both among decision makers and among researchers who aim to inform their decisions. Much of the research on charter schools and accountability policies, for example, has focused on test performance instead of a broader set of educational goods, and only a few studies have provided evidence of childhood goods or parents’ interests. We hope that our approach will inspire social scientists to attend to that broader range of outcomes and will stimulate funders to encourage research that provides the evidence needed for educational decision makers to make empirically informed and morally responsible decisions.

Notes Introduction 1. On the former, see Finnigan and Daly (2014) and Coburn and Stein (2010). On the latter, see Cartwright and Hardie (2012). 2. See Wolff (2011) for discussion, with examples, of how to think philosophically about evidence in the light of values in public-policy areas other than education. See also Duncan and Murnane (2014) for a model values-informed discussion by social scientists of educational policy questions and Witte (2000, chap. 7) for a nuanced discussion of how different value considerations should influence the interpretation of evidence from an educational voucher program. 3. Levinson 2015. 4. Gamoran 2009. 5. Hanushek and Wößmann 2006 and Huang 2009. 6. Clotfelter, Ladd, and Vigdor 2015. 7. Domina et al. 2014, 2015.

Chapter 1 1. The best-known and most influential contemporary theory of values for education is Gutmann (1987), which prompted a revival of philosophical interest in the topic. Like us, and following Gutmann, recent work has tended to focus on two key questions about values in education: what the aims of educational practices should be and how education should be distributed. We discuss the second, distributive question in chapter 2. On the fi rst, theorists tend to divide into those who, like us, think of the aims as primarily concerned with flourishing (e.g., Brighouse 2008; Grant 2012; White 1995, 2011; Engel 2015; Curren 2000) and those who conceive them rather in terms of democratic values (e.g., Gutmann 1987, Anderson 2007).

164

Notes to pages 20–24

2. Another way of talking about dispositions and attitudes, widely used by philosophers, is as virtues. Virtues, in the Aristotelian sense that philosophers usually talk about them, consist of dispositions to act and corresponding appropriate understandings and valuings. We avoid the language of virtue mainly because in political discourse in much of the English-speaking world, and especially in the United States, it has connotations likely to alienate some policy makers and many social scientists. We are grateful to Randall Curren for pressing us to explain why we avoid that language. 3. The two major kinds of theory of flourishing are objective and subjective. Objective theories identify some precepts and/or activities and pursuits following or participating in which constitutes flourishing. Many religious theories specify a precept such as “abide by God’s commands,” but there are also “objective list” theories that include such items as “maintaining physical and mental health,” “participating in complex and socially valuable work,” and “engaging in intimate personal relationships.” Subjective theories fall into two categories. Some, “conscious state,” theories identify certain kinds of feeling or subjective experience—hedonistic theories, for example, hold that flourishing consists in feeling happy. Other subjective theories specify some relationship between an internal state of the agent and the state of the world—economists, for example, commonly treat the extent to which an agent’s preferences are satisfied as a measure of flourishing. For different objective theories and illuminating discussion, see Griffi n (1985) and Raz (1986); for influential subjective theories, see (Sumner 1996) and Mill ([1867] 2002). Our overall approach has considerable affi nities with the capabilities approach to justice (e.g., Nussbaum 2000) as well as to Raz’s (1986) theory of freedom. See also Sypnowich (2000). 4. The rest of this section contains a greatly condensed and revised version of the fi rst four chapters of Brighouse (2006). 5. See Bowles and Gintis (1976) for an argument that schools in the United States do, in fact, prepare children for economic participation but not in the way that our normative values suggest that they should. Their view is instructive because the values we describe in this chapter and the next would provide strong grounds for criticizing schooling in the United States if it indeed functions as they suggest. 6. We understand that this is not entirely uncontroversial. In particular, some adherents to some religious conceptions of flourishing may believe that knowledge of other religions is, in fact, risky, especially in adolescence, because it may undermine continued affi liation with the religion in which the child has been raised. Notably few, however, believe that acquaintance with their own religious commitments and traditions would be damaging to others. 7. Again, this is not entirely uncontroversial. Moreover, while it seems to be widely accepted among citizens of liberal democracies, there is considerable room for disagreement about the extent to which the relevant knowledge and skills benefit the person who has them rather than benefiting others.

Notes to pages 25–38

165

8. See Gutmann (1987), Callan (1997), Macedo (2003), and Allen et al. (2016) for a sample of different arguments for the conclusion that promoting democratic competence is an important aim of education. 9. On the practical implications of an educational concern with personal fulfi llment, see White (2011). 10. Some education researchers have developed formal methods to attach weights to different valued outcomes within the context of cost- effectiveness analysis. Although that approach typically evaluates projects by the ratio of a project’s costs to the changes in a single outcome of interest, an extended form, called cost-utility analysis, incorporates multiple outcomes by attaching utility weights to each outcome. Within the field of education policy, the weights are based on preferences that are typically elicited from education experts or expert panels (Levin and McEwan 2002). Similar methods could, in principle, be used in the context of our framework.

Chapter 2 1. Most theorists who have addressed distributive justice in education have defended one or another of these rules as a principle of justice, sometime against the others. Like Schouten (2012) and Brighouse and Swift (2014b), we are pluralists: we believe that each rule has some force. In circumstances where they confl ict, judgments must be made about which should take priority. For an accessible general discussion of these three distributive principles, see Swift (2013). 2. For variants of this view see Gutmann (1999), Curren (1994), Tooley (1995), Anderson (2007), and Satz (2007). 3. For defenses of adequacy (or “sufficiency”) as a general principle of justice, see Frankfurt (1987) and Raz (1978). For criticism, see Casal (2007). 4. For defenses of various principles of equality in education see Brighouse (2000), Brighouse and Swift (2014b), and Ben- Shahar (2016). For a discussion of the indeterminacy of the idea of equality of educational opportunity, see Jencks (1988). 5. The classic discussion of positional goods is Hirsch (1976); see also Brighouse and Swift (2006). For their significance specifically in the educational context, see Swift (2003) and Koski and Reich (2007). For a more general treatment of their significance for economic policy, see Frank (1999). 6. For a defense see Schouten (2012). The seminal idea here is Rawls’s difference principle (Rawls 1971). The classic discussion of “priority to the worse off” as an alternative to equality is Parfit (2000). 7. See Gheaus (2015), Brennan (2014), and MacLeod (2010) for variants of this idea. For a discussion of its relevance for schools, see MacLeod (2016). 8. See Engel (2009).

166

Notes to pages 39–61

9. For contrasting views, see Brighouse and Swift (2014a) and Stroud (2015). 10. Friedman 1962. 11. See Tooley (1993, 2000) and Brighouse (1998).

Chapter 3 1. The AFQT is a general measure of trainability and is the primary criterion of enlistment eligibility for the armed forces. It covers a range of topics, including general science, arithmetic reasoning, word knowledge, paragraph comprehension, numerical operations, coding speed, auto and hop information mathematics knowledge, mechanical comprehension, and electronics information. See Neal and Johnson (1996). The study referred to in the text—Murnane, Willett, and Buckley (2012)—is based on the National Longitudinal Survey of Youth (NLSY), which uses a modified version of the AFQT. 2. See Chetty, Friedman, and Rockoff (2011). 3. See Oreopoulus and Salvanes (2011) and Havemen and Wolfe (1984). 4. Moretti 2004. 5. See Dee (2004), Milligan, Moretti, and Oreopoulos (2004), and Verba, Schlozman, and Brady (1995). 6. See Heckman and Rubinstein (2001). Another problem with the GED is that it is notoriously easy. 7. See Heckman, Stixrud, and Urzua (2006) and Cunha and Heckman (2007). 8. Rose v. Council for Better Education, 790 S.W. 2nd 186, 205–1 (Kentucky 1989). 9. See Marshall and Swift (1997). 10. Politicians, activists and even some theorists sometimes use language that seems to contradict this view, as if a group could thrive independently of its members. While understanding the rhetorical purpose of such language, we stand by this individualist claim and think it less controversial than it might seem. 11. The Nation’s Report Cart: Trends in Academic Progress 2012, http://fi les .eric.ed.gov/fulltext/ ED543744.pdf. The NAEP reports two sets of results, one at the national level over a long time period for students at ages nine, thirteen, and seventeen and one for a shorter time period at the national and state level for fourth and eighth graders. We use the long-term trend data for thirteen-yearolds to illustrate the trends since the early 1970s. 12. For math, the 36 point increase in the scores of black students far exceeds the 19 point increase for white students. The achievement gap between the races fell from 46 points in 1973 to 29 points in 2008, with the smallest gap in 1988. 13. Jencks and Phillips 1998. 14. See Neal and Johnson (1996) and Murnane et al. (2012). Calculated as 44.4 – 15.6 = 28.8. 28.8/44.4 = 0.65.

Notes to pages 61–89

167

15. Bertrand and Mullainathan 2004. 16. These scores were standardized as of 2000 to have a mean of 500 and a standard deviation of 100. 17. For further discussion, see Ladd (2012). For more information on Finland, see Sahlberg (2011), and on South Korea, see Ahn (2011). 18. See Schweinhart et al. (2005).

Chapter 4 1. For more general discussion of and theoretical background to our approach, see Swift and White (2008) and Swift (2008). 2. Rothstein 2004. 3. Rothstein 2008. 4. Duncan and Murnane 2014, chap. 7. 5. See Duncan and Murnane (2011) for a compendious presentation of evidence concerning the various ways in which different aspects of social disadvantage affect educational and labor-market success (or lack of it). See also Duncan and Murnane (2014), which combines a careful development of the analyses suggested by the fi ndings of Duncan and Murnane (2011) with evidence of interventions that seem to have had positive effects on the educational goods of disadvantaged groups to produce proposals for policy consideration. 6. On feasibility in general, see Gilabert and Lawford- Smith (2012) and Lawford- Smith (2013). 7. See Cartwright and Hardie (2012). 8. Jepsen and Rifkin 2009. 9. Haskins and Margolis (2014) instructively discuss the way that the fi rst Obama administration designed interventions so that they could be studied rigorously, the promise being that only those programs that evidence showed to be successful would be expanded. Notably, these programs were all microlevel interventions and did not include policies, such as the expansion of charter schools or the adoption of the Common Core State Standards, that were promoted through Race to the Top.

Chapter 5 1. See Ladd and Goertz (2015) chap. 3, 15, and 21. 2. To be clear—distributing funding appropriately is a necessary but not a sufficient condition for achieving any given distributive goal. 3. Brighouse and Swift 2014a. 4. See Rothstein (2013) and Brighouse and Swift (2013) for discussions of how

168

Notes to pages 91–107

the government can legitimately influence where people live without violating their right to free choice of residence. 5. https://www.gov.uk/government/publications/pupil-premium-2015 -to -2016 -allocations/pupil-premium-2015 -to -2016 - conditions- of-grant 6. Ladd and Fiske 2012. 7. Baker et al. 2016. 8. Verstegen 2011. 9. Indeed, the Milwaukee Parental Choice Program prohibits private schools serving voucher students from charging additional tuition altogether—though voluntary private contributions to the schools are not limited (as described by Private School Choice Programs Frequently Asked Questions—2016–17 School Year, http://www.schoolchoicewi.org/fi les/9514/2444/4197/ WPCP_FAQ_2015 -16.pdf). 10. Loeb 2001. 11. See Satz (2007) for why those concerned with adequacy may not be troubled by supplementation. 12. Loeb 2012. 13. For a description of multiple methods use to estimate the expenditures needed to reach educational goals, see Loeb, Bryk, and Hanushek (2008). 14. Duncombe, Nguyen-Hoang, and Yinger 2015. 15. Rumberger and Gandara 2015. 16. Baker et al. 2016. 17. Borman and Boulay 2004. 18. Jackson, Johnson, and Persico 2016. 19. One difference between the United States and the United Kingdom is that in the United Kingdom, a much higher proportion of affluent students attend private schools than in the United States, and the more egalitarian funding regime among public schools in the United Kingdom might play a role in accounting for that. 20. Cullen 2003. 21. Silva and Sonstelie 1995. 22. Brunner and Sonstelie 2003. 23. Poterba 1997; Hoxby 1998. 24. Loeb 2001. 25. Lareau 2003; Kornrich and Furstenberg 2013; Magnuson and Duncan 2006. 26. Kaushal, Magnuson, and Waldfogel 2011. 27. Krueger 1999. 28. Jepsen and Rifkin 2009. 29. Loeb and Strunk 2007. 30. This paragraph comes largely from Loeb, Bryk, and Hanushek (2008). 31. See Costrell, Hanushek, and Loeb (2008).

Notes to pages 113–121

169

Chapter 6 1. Ladd and Loeb 2013. 2. OECD 2011, www.oecd.org/edu/eag2011, Education at a Glance 2011: OECD Indicators, Chart D5.1, Performance and Regulatory Accountability in Public Schools (2009), http://www.oecd.org/edu/skills-beyond-school/48631428.pdf. 3. Finland, a country whose students score very well on international tests, uses neither type of accountability, opting instead for what some might call professional accountability. This approach relies on internal practices at the school level in the form of professional norms and trust. As professionals, teachers are accountable to the norms and standards of their profession and are expected to work together to maintain those standards. 4. Goldin 2001. 5. McMurrer (2007); also reported in Rothstein, Jacobsen, and Wilder (2008, table 3, p. 49). 6. In December 2015, the federal government reauthorized the Elementary and Secondary Education Act and relabeled it the Every Student Succeeds Act (ESSA). This new law devolved most of the responsibility for school accountability to the states. 7. The charter chain is the Knowledge Is Power Program (KIPP). 8. The text refers to results from NAEP long-term trends, which reports test score results by age. Another set of tests, known as “Main NAEP,” reports results by grade and similarly fi nds that any breaks in trend occur only for the younger students, those in grade 4. 9. Dee and Jacob 2011; Wong, Cook and Steiner 2009; Lee and Reeves 2012. 10. Murnane 2013a, 2013b. 11. Burgess, Wilson, and Worth 2013. 12. Cullen and Reback 2006; Deere and Strayer 2001; Figlio and Getzler 2007; Jacob 2005. Studies have found cheating in response to the high-stakes testing under NCLB in New York City, Washington, DC, Houston, Atlanta, and Pennsylvania, among others places. See Trip Gabriel, “Under Pressure, Teachers Tamper with Tests,” New York Times, June 10, 2010 (http://www.nytimes.com/ 2010/06/11/education/11cheat.html?pagewanted=all); Kim Severson, “Systematic Cheating Is Found in Atlanta’s School System,” New York Times, July 5, 2011 (http://www.nytimes.com/2011/07/06/education/06atlanta.html); Sharon Otterman, “In Cheating Cases, Teachers Who Took Risks or Flouted Rules,” New York Times, October 17, 2011 (http://www.nytimes.com/2011/10/18/nyregion/ how - cheating - cases - at - new -york- schools - played - out .html ?pagewanted = all); Jack Gillum and Marisol Bello, “When Standardized Test Scores Soared in D.C., Were the Gains Real?,” USA Today, March 30, 2011 (http://www.usatoday .com/news/education/2011- 03 -28 -1Aschooltesting28 _CV_N.htm).

170

Notes to pages 121–135

13. Ladd 2010a, table 1, p. 203. 14. See discussion and references in Lauen and Gaddis (2012). 15. Ladd and Lauen 2010. 16. Murnane 2013a, 2013b. 17. A paper that tries to isolate the effects of NCLB on achievement gaps fi nds no overall effect but does fi nd gap reductions in states with more subgroupspecific accountability pressure, more between-school segregation, and larger gaps before the implementation of the policy (Reardon et al. 2013). This result is in keeping with a second study that incorporated earlier trends and concluded that NCLB had mixed—and, at best, small—effects on the achievement gaps defi ned by NCLB’s racial and socioeconomic categories (Lee and Reaves 2012). Two studies of accountability systems in the United States before NCLB come to different conclusions about their effect on racial gaps (Carnoy and Loeb 2002; Hanushek and Raymond 2005). 18. Wheelock, Haney, and Bebell 2000; Jones 2007. 19. Hoffman, Assaf, and Paris 2001. 20. Holbein and Ladd 2014. 21. Reback, Rockoff, and Schwartz 2011. 22. Papay, Murnane, and Willett 2010. 23. Ladd 2010b. 24. ERO Review Office, Framework for School Reviews, 2011. For the theoretical framework underlying this approach see Mutch 2013. 25. de Wolf and Jannsens 2007. 26. Matthews and Sammons 2004; Gray and Gardner 2012. 27. Ofsted 2013; Ryan, Gandha, and Ahn 2013. See also Ladd and Fiske (2016) for positive effects on student achievement of borough-wide inspections in London. 28. Hussain 2012; Luginbuhl, Webbink, and de Wolf 2009. 29. Ladd and Fiske 2012.

Chapter 7 1. Preference can be given, for example, on the basis of geographic location and sibling enrollment. Some countries, such as England, allow similar schools to select at least some of their students directly from the pool of applicants. 2. http://www.calcharters.org/understanding/. 3. National Center for Education Statistics, Charter School Enrollment, updated April 2016, http://nces.ed.gov/programs/coe/indicator_cgb.asp. 4. Snyder and Dillow (2015, table 206.10). Number and percentage of homeschooled students ages five through seventeen with a grade equivalent of kinder-

Notes to pages 135–146

171

garten through twelfth grade by selected child, parent, and household characteristics, 2003, 2007, and 2012. 5. US Department of Education, National Center for Education Statistics, Private School Universe Survey (PSS), 2003–4 through 2013–14; National Elementary and Secondary Enrollment Projection Model, 1972 through 2025. See Snyder and Dillow (2015, table 105.30). 6. Miron and Gulosino 2013. 7. Ibid. 8. http://www.kipp.org/schools. 9. http://www.uncommonschools.org/. 10. http://www.achievementfi rst.org/. 11. See Brighouse and Schouten (2011) for an account of high commitment schools. 12. Miron and Gulosino 2013. 13. National Alliance for Public Charter Schools, A Growing Movement: America’s Largest Charter School Communities, 10th ed., November 2015, http:// www.publiccharters.org/wp - content/uploads/2015/11/enrollmentshare_web.pdf. 14. Childress and Weber 2010. 15. CREDO 2009, 2013; Hanushek et al. 2007; Bifulco and Ladd 2006; Booker et al. 2007. 16. Furgeson et al. 2011. 17. CREDO 2013. 18. CREDO 2013; Abdulkadiroğlu et al. 2011. 19. Schneider, Teske, and Marschal 2002; Buckley and Schneider 2006. 20. Ladd, Clotfelter, and Holbein, forthcoming. 21. On health, see Cutler and Lleras-Muney (2008); on voting, see Sondheimer and Green (2010); on happiness, see Powdthavee, Lekfuangfu, and Wooden (2013). 22. Booker et al. 2011. 23. Dobbie and Fryer 2015. 24. Dobbie and Fryer 2016. 25. Antonio et al. 2004. 26. Janmaat 2012. 27. Frankenberg, Siegel-Hawley and Wang 2010; Ritter et al. 2010. 28. Zimmer et al. 2011. 29. Ibid. 30. Bifulco and Ladd 2006; Booker, Zimmer, and Buddin 2005. 31. Ladd, Clotfelter, and Holbein, forthcoming. 32. Loeb and Kasman 2013. 33. Bifulco and Reback 2014. 34. Bettinger 2005.

172

Notes to pages 147–152

35. Imberman 2011. 36. Booker et al. 2008; Sass 2006. 37. Lubienski 2003; Loeb and Kasman 2013. 38. Abdulkadiroğlu et al. 2011; Dobbie and Fryer 2009. 39. Most studies that exploit the fact that charter schools use admissions lotteries in order to defi ne a control group emphasize that their evidence relates only to students who applied for admission to charter schools in the jurisdiction they have studied. This obviously does not include students who did not apply to charter schools where they were available. 40. Fryer 2011; Angrist, Pathak, and Walters 2011. 41. Fryer 2011. 42. KIPP annual report, http://www.kipp.org/results/national-results. 43. Goldring, Taie, and Riddles 2014. 44. Tuttle et al 2013. 45. CREDO 2013. 46. Hanushek et al 2007. 47. Gross and Pochop 2008. 48. Angrist, Pathak, and Walters 2011. 49. CREDO 2013. 50. Ladd, Clotfelter, and Holbein, forthcoming. 51. CREDO 2013.

References Abdulkadiroğlu, Atila, Joshua D. Angrist, Susan M. Dynarski, Thomas J. Kane, and Parag A. Pathak. 2011. “Accountability and Flexibility in Public Schools: Evidence from Boston’s Charters and Pilots.” Quarterly Journal of Economics 126 (2): 699– 748. Ahn, B. M. 2011. “Education in South Korea: National Treasure or National Headache?” Paper presented at the annual Association for Education Finance and Policy conference, Seattle, WA, March 24– 26. Allen, Danielle, Tommie Shelby, Marcelo Suárez- Orozco, Michael Rebell, and Quiara Alegría Hudes. 2016. Education and Equality. Chicago: University of Chicago Press. Anderson, E. 2007. “Fair Opportunity in Education: A Democratic Equality Perspective.” Ethics 117 (4): 595– 622. Angrist, Joshua D., Parag A. Pathak, and Christopher R. Walters. 2011. Explaining Charter School Effectiveness. NBER Working Paper Series 17332. Cambridge, MA: National Bureau of Economic Research. Antonio, A. L., M. J. Chang, K. Hakuta, D. A. Kenny, S. L. Levin, and J. F. Milem. 2004. “Effects of Racial Diversity on Complex Thinking in College Students.” Psychological Science 15 (8): 507–10. Baker, Bruce, Danielle Farrie, Theresa Luhm, and David G. Sciarra. 2016. “Is School Funding Fair? A National Report Card Fifth Edition.” Education Law Center and Rutgers Graduate School of Education. http://www.schoolfundingfairness.org/ National _Report _Card _2016.pdf. Ben- Shahar, T. 2016. “Equality in Education: Why We Must Go All the Way.” Ethical Theory and Moral Practice 19 (1): 83–100. Bertrand, Marianne, and Sendhil Mullainathan. 2004. “Are Emily and Greg More Employable Than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination.” American Economic Review 94 (4): 991–1013. Bettinger, Eric. 2005. “The Effect of Charter Schools on Charter Students and Public Schools.” Economics of Education Review 24 (2): 133–47.

174

References

Bifulco, R., and H. Ladd. 2006. “The Impact of Charter Schools on Student Achievement.” Education Finance and Policy 1 (1): 50– 90. Bifulco, R., and Randall Reback. 2014. “Fiscal Impacts of Charter Schools: Lessons from New York.” Education Finance and Policy 9 (1): 86–107. Booker, Kevin, Scott M. Gilpatric, Timothy Gronberg, and Dennis Jansen. 2007. “The Impact of Charter School Attendance on Student Performance.” Journal of Public Economics 91 (5/6): 849– 76 ———. 2008. “The Effect of Charter Schools on Traditional Public School Students in Texas: Are Children Who Stay Behind Left Behind?” Journal of Urban Economics 64 (1): 123–45. Booker, Kevin, Tim R. Sass, Brian Gill, and Ron Zimmer. 2008. “Going beyond Test Scores: Evaluating Charter School Impact on Educational Attainment in Chicago and Florida.” RAND Education working paper series WR- 610-BMG. http://www.rand.org/content/dam/rand/pubs/working _papers/2008/ RAND_WR610.pdf. ———. 2011. “The Effects of Charter High Schools on Educational Attainment.” Journal of Labor Economics 29 (2): 377–415. Booker, Kevin, Ron Zimmer, and R. Buddin. 2005. “The Effects of Charter Schools on School Peer Composition.” RAND Education working paper series WR-306-EDU. http://www.rand.org /content/dam /rand/pubs/working _papers/ 2005/ RAND_WR306.pdf. Borman, Geoffry, and Matthew Boulay, eds. 2004. Summer Learning Research, Policies, and Programs. Mahwah, NJ: L. Erlbaum. Bowles, S., and H. Gintis. 1976. Schooling in Capitalist America: Educational Reform and the Contradictions of Economic Life. New York: Basic Books. Brennan. S. 2014. “The Goods of Childhood, Children’s Rights, and the Role of Parents as Advocates and Interpreters.” In Family- Making: Contemporary Ethical Challenges, edited by F. Baylis and C. McLeod, 29–48. Oxford: Oxford University Press. Brighouse, H. 1998. “Why Should States Fund Schools?” British Journal of Educational Studies 46 (2): 138– 52. ———. 2000. School Choice and Social Justice. Oxford: Oxford University Press. ———. 2006. On Education (Thinking in Action). New York: Routledge. ———. 2008. “Education for a Flourishing Life.” Yearbook of the National Society for the Study of Education 107 (1): 58– 71. Brighouse, H., and Gina Schouten. 2011. “Understanding the Context for Existing Research and Reform Proposals.” In Whither Opportunity, edited by Greg J. Duncan and Richard J. Murnane, 507– 22. New York: Russell Sage. Brighouse, H., and A. Swift. 2006. “Equality, Priority and Positional Goods.” Ethics 116 (3): 471– 97. ———. 2013. “Family Values and Schools Policy: Shaping Values and Conferring

References

175

Advantage.” In Education, Justice, and Democracy, edited by D. Allen and R. Reich, 199– 220. Chicago: University of Chicago Press. ———. 2014a. Family Values: The Ethics of Parent- Child Relationships. Princeton, NJ: Princeton University Press. ———. 2014b. “The Place of Educational Equality in Educational Justice.” In Education, Justice and the Human Good, edited by K. Meyer, 14– 33. New York: Routledge. Brunner, Eric, and Jon Sonstelie. 2003. “School Finance Reform and Voluntary Fiscal Federalism.” Journal of Public Economics 87 (9/10): 2157– 85. Buckley, Jack, and Mark Schneider. 2006. “Are Charter School Parents More Satisfied with Schools? Evidence from Washington, DC.” Peabody Journal of Education 81 (1): 57– 78. Burgess, S., D. Wilson, and J. Worth. 2013. “A Natural Experiment in School Accountability: The Impact of School Performance Information on Pupil Progress.” Journal of Public Economics 106: 57– 67. Callan, E. 1997. Creating Citizens: Political Education and Liberal Democracy. Oxford: Oxford University Press. Carnoy, M., and S. Loeb. 2002. “Does External Accountability Affect Student Outcomes? A Cross- State Analysis.”  Educational Evaluation and Policy Analysis 24 (4): 305– 31. Cartwright, Nancy, and Jeremy Hardie. 2012. Evidence- Based Policy: A Practical Guide to Doing It Better. Oxford: Oxford University Press. Casal, P. 2007. “Why Sufficiency Is Not Enough.” Ethics 117 (2): 296– 326. Chetty, Raj, John N. Friedman, and Jonah E. Rockoff. 2011. The Long-Term Impacts of Teachers: Teacher Value-Added and Student Outcomes in Adulthood. NBER Working Paper Series 17699, app. table 4. Cambridge, MA: National Bureau of Economic Research. Childress, Stacey, and James Weber. 2010. Public Education in New Orleans: Pursuing Systemic Change through Entrepreneurship. Harvard Business Review Case Studies. Boston: Harvard Business School. Clotfelter, Charles T., Helen F. Ladd, and Jacob L. Vigdor. 2015. “The Aftermath of Accelerating Algebra: Evidence from District Policy Initiatives.” Journal of Human Resources 50 (1): 159– 88. Coburn, Cynthia, and Mary Kay Stein. 2010. Research and Practice in Education: Building Alliances, Bridging the Divide. Lanham, MD: Rowman and Littlefield. Costrell, Robert, Eric A. Hanushek, and Susanna Loeb. 2008. “What Do Cost Functions Tell Us about the Cost of an Adequate Education?” Peabody Journal of Education 83 (2): 198– 223. CREDO (Center for Research on Education Outcomes). 2009. “Multiple Choice: Charter School Performance in 16 States. “Center for Research on Education Outcomes, Stanford University.

176

References

———. 2013. National Charter School Study. Stanford, CA: Stanford University, Center for Research on Education Outcomes. Cullen, Julie Berry. 2003. “The Impact of Fiscal Incentives on Student Disability Rates.” Journal of Public Economics 87 (7/8):1557– 89. Cullen, Julie Berry, and R. Reback. 2006. “Tinkering Towards Accolades: School Gaming under a Performance Accountability System.” In Improving School Accountability: Check- Ups or Choice, Advances in Applied Microeconomics. Advances in Applied Microeconomics, vol. 14, edited by T. Gronberg and D. Jansen, 1– 34. Amsterdam: Elsevier. Cunha, Flavio, and James J. Heckman. 2007. The Technology of Skill Formation. NBER Working Paper Series 12840. Cambridge, MA: National Bureau of Economic Research. Curren, R. 1994. “Justice and the Threshold of Educational Equality.” Philosophy of Education 50: 239–48. ———. 2000. Aristotle on the Necessity of Public Education. Lanham, MD: Rowman and Littlefield. Cutler, David M., and Adriana Lleras-Muney. 2008. “Education and Health: Evaluating Theories and Evidence.” In Making Americans Healthier: Social and Economic Policy as Health Policy, edited by J. House, R. Schoeni, G. Kaplan, and H. Pollack, 29– 61. New York: Russell Sage Foundation. Dee, Thomas S. 2004. “Are There Civic Returns to Education?” Journal of Public Economics 88 (9/10): 1697–1720. Dee, Thomas, and B. Jacob. 2011. “The Impact of No Child Left Behind on Student Achievement.” Journal of Policy Analysis and Management 30 (3): 418–46. Deere, D., and W. Strayer. 2001. “Putting Schools to the Test: School Accountability, Incentives and Behavior.” Working paper. College Station: Texas A&M University. de Wolf, Inga F., and Frans J. G. Janssens. 2007. “Effects and Side Effects of Inspections and Accountability in Education: An Overview of Empirical Studies.” Oxford Review of Education 33 (3): 379– 96. Dobbie, Will, and Roland Fryer. 2009. Are High Quality Schools Enough to Close the Achievement Gap? Evidence from a Social Experiment in Harlem. NBER Working Paper Series 15473. Cambridge, MA: National Bureau of Economic Research. ———. 2015. “The Medium-Term Impacts of High-Achieving Charter Schools.” Journal of Political Economy 123 (5): 985–1037. ———. 2016. Charter Schools and Labor Market Outcomes. NBER Working Paper Series 22502. Cambridge, MA: National Bureau of Economic Research. Domina, Thurston, Andrew McEachin, Andrew Penner, and Emily Penner. 2015. “Aiming High and Falling Short: California’s 8th Grade Algebra-forAll Effort.”Education and Policy Analysis 37 (3): 275– 95.

References

177

Domina, Thurston, Andrew M. Penner, Emily K. Penner, and Anne Marie Conley. 2014. “Algebra for All: California’s Eighth- Grade Algebra Initiative as Constrained Curricula.” Teachers College Record 116 (8): 1– 32. Duncan, Greg J., and Richard J. Murnane, eds. 2011. Whither Opportunity: Rising Inequality, Schools, and Children’s Life Chances. New York: Russell Sage Foundation. ———. 2014. Restoring Opportunity: The Crisis of Inequality and the Challenge of American Education. Cambridge, MA: Harvard Education Press. Duncombe, William, Phuong Nguyen-Hoang, and John Yinger. 2015. “Measurement of Cost Differentials.” In Handbook of Research in Education Finance and Policy, 2nd ed., edited by Helen F. Ladd and Margaret E. Goertze, 266– 78. New York: Routledge. Engel, S. 2009. Real Kids: Creating Meaning in Everyday Life. Cambridge, MA: Harvard University Press. ———. 2015. The End of the Rainbow: How Educating for Happiness—Not Money—Would Transform Our Schools. New York: New Press. Figlio, D., and L. Getzler. 2007. “Accountability, Ability and Disability: Gaming the System?” In Improving School Accountability: Check- Ups or Choice Advances, Applied Microeconomics. Advances in Applied Microeconomics, vol. 14, edited by T. Gronberg and D. Jansen, 35–49. Amsterdam: Elsevier. Finnigan, Kara, and Alan Daly. 2014. Using Research Evidence in Education: From the Schoolhouse Door to Capitol Hill. New York: Springer. Frank, R. 1999. Luxury Fever: Why Money Fails to Satisfy in an Era of Excess. New York: Simon and Schuster. Frankenberg, E., G. Siegel-Hawley, and J. Wang. 2010. Choice without Equity: Charter School Segregation and the Need for Civil Rights Standards. Los Angeles: Civil Rights Project/Proyecto Derechos Civiles, University of California, Los Angeles. Frankfurt, H. 1987. “Equality as a Moral Ideal.” Ethics 98 (1): 21–43. Friedman, Milton. 1962. Capitalism and Freedom. Chicago: University of Chicago Press. Fryer, Roland G., Jr. 2011. Injecting Successful Charter School Strategies into Traditional Public Schools: Early Results from an Experiment in Houston. NBER Working Paper Series 17494. Cambridge, MA: National Bureau of Economic Research. Furgeson, Johshua, Brian Gill, Joshua Haimson, Alexandra Killewald, Moira McCullough, Ira Nichols-Barrer, Bing-ru Teh, Natalya Verbitsky- Savitz, Melissa Bowen, Allison Demeritt, Paul Hill, and Robin Lake. 2011. CharterSchool Management Organizations: Diverse Strategies and Diverse Student Impacts. Washington, DC: Mathematica Policy Research. Gamoran, A. 2009. Tracking and Inequality: New Directions for Research and

178

References

Practice. WCER Working Paper No. 2009– 6. Madison: Wisconsin Center for Education Research, School of Education, University of Wisconsin– Madison. Gheaus. A. 2015. “The Intrinsic Goods of Childhood and the Just Society.” In The Nature of Children’s Well- Being in Theory and Practice, edited by A. Bagattini and C. MacLeod, 35– 52. New York: Springer. Gilabert, P., and Holly Lawford- Smith. 2012. “Political Feasibility: A Conceptual Exploration.” Political Studies 60 (4): 809– 25. Goldin. C. 2001. “The Human Capital Century and American Leadership: Virtues of the Past.” Journal of Economic History 61 (2): 263– 92. Goldring, Rebecca, Soheyla Taie, and Minsun Riddles. 2014. Teacher Attrition and Mobility: Results from the 2012–13 Teacher Follow- Up Survey First Look. Washington, DC: US Department of Education, National Center for Education Statistics. Grant, Carl A. 2012. “Cultivating Flourishing Lives: A Robust Social Justice Vision of Education.” American Educational Research Journal 49 (5): 910– 34. Gray, C., and J. Gardner. 1999. “The Impact of Schools Inspections.” Oxford Review of Education 25 (4): 455– 69. Griffi n, J. 1985. Well- Being. Oxford: Oxford University Press. Gross, B., and K. Martens Pochop. 2008. “How Charter Schools Organize for Instruction.” In Hopes, Fears, and Reality: A Balanced Look at American Charter Schools in 2008, edited by Robin Lake, 9– 22. Seattle: University of Washington Bothell, Center on Reinventing Public Education. Gutmann, A. 1987. Democratic Education. Princeton, NJ: Princeton University Press. ———. 1999. Democratic Education. Rev. ed., with a new preface and epilogue. Princeton, NJ: Princeton University Press. Hanushek, Eric A., John F. Kain, Steven G. Rivkin, and Gregory F. Branch. 2007. “Charter School Quality and Parental Decision Making with School Choice.” Journal of Public Economics 91 (5/6): 823–48. Hanushek, Eric A., and M. Raymond. 2005. “Does School Accountability Lead to Improved School Performance?” Journal of Policy Analysis and Management 24 (2): 297– 329. Hanushek, Eric A., and L. Wößmann. 2006. “Does Educational Tracking Affect Performance and Inequality? Differences-in-Differences Evidence across Countries.” Economic Journal 116 (510): C63– C76. Haskins, R., and Greg Margolis. 2014. “Obama’s Promise: Using Evidence to Fight the Nation’s Social Problems.” In Using Research Evidence in Education, edited by K. S. Finnegan and A. J. Daly, 133– 60. New York: Springer. Havemen, Robert H., and Barbara L. Wolfe. 1984. “Schooling and Economic Well-Being: The Role of Nonmarket Effects.” Journal of Human Resources 19 (3): 377–407. Heckman, James J., and Yona Rubinstein. 2001. “The Importance of Noncogni-

References

179

tive Skills: Lessons from the GED Testing Program.” American Economic Review 91 (2): 145–49. Heckman, James J., Jora Stixrud, and Sergio Urzua. 2006. “The Effects of Cognitive and Noncognitive Abilities on Labor Market Outcomes and Social Behavior.” Journal of Labor Economics 24 (3): 411– 82. Hirsch, F. 1976. Social Limits to Growth. Cambridge, MA: Harvard University Press. Hoffman, J. V., L. C Assaf, and S. G. Paris. 2001. “High Stakes Testing in Reading: Today in Texas, Tomorrow?” Reading Teacher 54 (5): 482– 92. Holbein J., and H. Ladd. 2014. Accountability Pressure and Non-Achievement Student Behaviors. Working Paper 122. Washington, DC: National Center for the Analysis of Longitudinal Data in Education Research. Hoxby, Caroline M. 1998. “How Much Does School Spending Depend on Family Income? The Historical Origins of the Current School Finance Dilemma.” American Economic Review 88 (2): 309–14. Huang, M. H. 2009. “Classroom Homogeneity and the Distribution of Student Math Performance: A Country-Level Fixed-Effects Analysis.” Social Science Research 38 (4): 781– 91. Hussain, I. 2012. “Subjective Performance Evaluation in the Public Sector: Evidence from School Inspections.” Journal of Human Resources 50 (1): 189– 221. Imberman, Scott. 2011. “The Effect of Charter Schools on Achievement and Behavior of Public School Students.” Journal of Public Economics 95 (7/8): 850– 63. Jackson, Kirabo, Rucker C. Johnson, and Claudia Persico. 2016. “The Effects of School Spending on Educational and Economic Outcomes: Evidence from School Finance Reforms.” Quarterly Journal of Economics 131 (1): 157– 218. Jacob, B. A. 2005. “Accountability, Incentives and Behavior: The Impact of High- Stakes Testing in the Chicago Public Schools.” Journal of Public Economics 89 (5/6): 761– 96. Janmaat, Jan Germen. 2012. “The Effect of Classroom Diversity on Tolerance and Participation in England, Sweden and Germany.” Journal of Ethnic and Migration Studies 38 (1): 21– 39. Jencks, C. 1988. “Whom Must We Treat Equally for Educational Opportunity to be Equal?” Ethics 98 (3): 518– 33. Jencks, C., and Meredith Phillips. 1998. The Black-White Test Score Gap. Washington, DC: Brookings Institution Press. Jepsen, Christopher, and Steven  Rifkin. 2009. “Class Size Reduction and Student Achievement: The Potential Tradeoff between Teacher Quality and Class Size.” Journal of Human Resources 44 (1): 223– 50. Jones, B. D. 2007. “The Unintended Outcomes of High- Stakes Testing.” Journal of Applied School Psychology 23 (2): 65– 86.

180

References

Kaushal, Neeraj, Katherine Magnuson, and Jane Waldfogel. 2011. “How Is Family Income Related to Investments in Children’s Learning? “ In Whither Opportunity? Rising Inequality, Schools, and Children’s Life Chances, edited by Greg J. Duncan and Richard J. Murnane, 187– 206. New York: Russell Sage Foundation. Kornrich, Sabino, and Frank Furstenberg. 2013. “Investing in Children: Changes in Parental Spending on Children, 1972 to 2007.” Demography 50 (1): 1– 23. Koski, W. S., and R. Reich. 2007. “When ‘Adequate’ Isn’t: The Retreat from Equality in Educational Law and Policy and Why it Matters.” Emory Law Journal 56: 545. Krueger, Alan B. 1999. “Experimental Estimates of Education Production Functions.” Quarterly Journal of Economics 115 (2): 497– 532. Ladd, H. 2010a. “Comments on Thomas S. Dee and Brian Jacob, ‘The Impact of No Child Left Behind on Students, Teachers, and Schools.’” In Brookings Papers on Economic Activity (Fall): 149–47. ———. 2010b “Education Inspectorate Systems in New Zealand and the Netherlands.” Education Finance and Policy 5 (3): 378– 92. ———. 2012. “Education and Poverty: Confronting the Evidence.” Journal of Policy Analysis and Management 31 (2): 203– 27. Ladd, H. F., Charles Clotfelter, and John Holbein. Forthcoming. “The Growing Segmentation of the Charter Sector in North Carolina.” Education, Finance, and Policy. Ladd H. F. and Edward B. Fiske. 2012. “Weighted Student Funding in the Netherlands: A Model for the U.S?” Journal of Policy Analysis and Management 30 (3): 470– 98 ———. 2016. “Educational Success in Two Inner London Boroughs: Lessons for the U.S.” Brookings Institution Brown Center Report. Washington, DC: Brookings Institution, Brown Center on Education Policy. Ladd, H. F., and M. Goertz. 2015. Handbook of Research in Education Finance and Policy. 2nd ed. New York: Routledge. Ladd, H. F., and D. Lauen. 2010. “Status vs. Value Added Measures of School Accountability Policies.” Journal of Policy Analysis and Management 29 (3): 426– 50. Ladd, H. F., and S. Loeb. 2013. “The Challenges of Measuring School Quality: Implications for Educational Equity.” In Education, Justice, and Democracy, edited by Danielle Allen and Robert Reich, 19–42. Chicago: University of Chicago Press. Lareau, A. 2003. Unequal Childhoods: Class, Race, and Family Life. Berkeley: University of California Press. Lauen, D., and S. M. Gaddis. 2012. “Shining a Light or Fumbling in the Dark? The Effects of NCLB’s Subgroup Specific Accountability on Student Achievement.” Educational Evaluation and Policy Analysis 34 (2): 185– 208.

References

181

Lawford- Smith, Holly. 2013. “Understanding Political Feasibility.” Journal of Political Philosophy 21 (3): 243– 59. Lee, J., and T. Reeves. 2012. “Revisiting the Impact of NCLB High Stakes School Accountability, Capacity and Resources.” Educational Evaluation and Policy Analysis 34 (2): 209– 31. Levinson, M. 2012. No Citizen Left Behind. Cambridge, MA: Harvard University Press. ———. 2015. “The Ethics of Pandering in Boston Public Schools’ School Assignment Plan.” Theory and Research in Education 13 (1): 38– 55. Loeb, Susanna. 2001. “Estimating the Effects of School Finance Reform: A Framework for a Federalist System.” Journal of Public Economics 80 (2): 225–47. ———. 2012. Getting Down to Facts: Five Years Later. Policy Analysis for California Education Report, May. http://edpolicyinca.org/sites/default/fi les/ PACE _Facts %20Brochure_Final.pdf. Loeb, Susanna, Anthony Bryk, and Eric Hanushek. 2008. “Getting Down to Facts: School Finance and Governance in California.” Education Finance and Policy 3 (1): 1–19. Loeb, Susanna, and Matthew Kasman. 2013. “Principals’ Perceptions of Competition for Students in Milwaukee Schools.” Education Finance and Policy 8 (1): 43– 73. Loeb, Susanna, and Katharine O. Strunk. 2007. “Accountability and Local Control: Incentive Response with and without Authority over Resource Generation and Allocation.” Education Finance and Policy 2 (1): 10– 39. Lubienski, C. 2003. “Innovation in Education Markets: Theory and Evidence on the Impact of Competition and Choice in Charter Schools.” American Educational Research Journal 40 (2): 395–443. Luginbuhl, R., D. Webbink, and I. de Wolf. 2009. “Do Inspections Improve Primary School Performance?” Educational Evaluation and Policy Analysis 31 (3): 221– 37. Macedo, S. 2003. Diversity and Distrust: Civic Education in a Multicultural Democracy. Cambridge, MA: Harvard University Press. MacLeod, C. 2010. “Primary Goods, Capabilities and Children.” In Measuring Justice: Primary Goods and Capabilities, edited by H. Brighouse and I. Robeyns, 174– 93. Cambridge: Cambridge University Press. ———. 2016. “Just Schools and Good Childhoods: Non-preparatory Dimensions of Educational Justice.” Journal of Applied Philosophy. doi:10.1111/ japp.12227. Magnuson, K. A., and G. D. Duncan. 2006. “The Role of Family Socioeconomic Resources in the Black-White Test Score Gap among Young Children.” Developmental Review 26: 365– 99. Marshall, Gordon, and Adam Swift. 1997. “Meritocratic Equality of Oppor-

182

References

tunity: Economic Efficiency, Social Justice, or Both?” Policy Studies 18 (1): 35–48. Matthews, P., and P. Sammons. 2004. Improvement through Inspections: An Evaluation of the Impact of Ofsted’s Work. London: Institute of Education, University of London. McMurrer, J. 2007. Choices, Changes, and Challenges: Curriculum and Instruction in the NCLB Era. From the Capital to the Classroom: Year 5 of the No Child Left Behind Act. Washington, DC: Center for Education Policy, George Washington University. Mill, J. S. (1867) 2002. Utilitarianism. Oxford Philosophical Texts. Oxford: Oxford University Press. Milligan, Kevin, Enrico Moretti, and Philip Oreopoulos. 2004. “Does Education Improve Citizenship? Evidence from the United States and the United Kingdom.” Journal of Public Economics 88 (9/10): 1667– 95. Miron, Gary, and Charisse Gulosino. 2013. Profi les of For- Profi t and Nonprofi t Education Management Organizations: Fourteenth Edition—2011– 2012. Boulder, CO: National Education Policy Center, School of Education, University of Colorado Boulder. Moretti, Enrico. 2004. “Workers’ Education, Spillovers, and Productivity: Evidence from Plant-Level Production Functions.” American Economic Review 94 (3): 656– 90. Murnane, Richard J. 2013a. U.S High School Graduation Rates: Patterns and Explanation. NBER Working Paper Series 18701. Cambridge, MA: National Bureau of Economic Research. ———. 2013b. “U.S. High School Graduation Rates: Patterns and Explanations.” Journal of Economic Literature 51 (2): 370–422. Murnane, Richard J., John B. Willett, and Katie Buckley. 2012. “Recent Trends in the Role of Cognitive Skills in Accounting for the Black-White Wage Differential.” Unpublished manuscript. Mutch, Carol. 2013. “Developing a Conceptual Framework for School Review.” In A Developmental and Negotiated Approach to School Self-Evaluation, edited by M. Lai and S. Kushne, 91–107, Advances in Program Evaluation vol. 14. Bingley, UK: Emerald. Neal, Derek A., and William R. Johnson. 1996. “The Role of Premarket Factors in Black-White Wage Differences.” Journal of Political Economy 104 (5): 869– 95. Nussbaum, M. 2000. Women and Human Development: The Capability Approach. Oxford: Oxford University Press. Ofsted (Office for Standards in Education). 2013. School Inspection Handbook: Handbook for Inspecting Schools in England under Section 5 of the Education Act 2005 (as amended by the Education Act 2011). Manchester, UK: Office of Standards in Education, Children’s Services and Skills.

References

183

Oreopoulos, Philip, and Kjell G. Salvanes. 2011. “Priceless: The Nonpecuniary Benefits of Schooling.” Journal of Economic Perspectives 25 (1): 159– 84. Papay, J. P., R. Murnane, and J. Willett. 2010. Income- Based Inequality in Educational Outcomes: Learning from State Longitudinal Data Systems. NBER Working Paper Series 20802. Cambridge, MA: National Bureau of Economic Research. Parfit, D. 2000. “Equality or Priority?” In The Ideal of Equality, edited by M. Clayton and A. Williams, 81–126. Houndmills, Basingstoke, Hampshire: Macmillan. Poterba, James. 1997. “Demographic Structure and the Political Economy of Public Education.” Journal of Policy Analysis and Management 16 (1): 48– 66. Powdthavee, Nattavudh, Warn N. Lekfuangfu, and Mark Wooden. 2013. The Marginal Income Effect of Education on Happiness: Estimating the Direct and Indirect Effects of Compulsory Schooling on Well- Being in Australia. Melbourne Institute Working Paper, no. 13/16. Parkville, VIC: University of Melbourne, Melbourne Institute of Applied Economic and Social Research. Rawls, J. 1971. A Theory of Justice. Cambridge, MA: Harvard University Press. Raz, J. 1978. “Principles of Equality.” Mind (87) 347: 321–42. ———. 1986. The Morality of Freedom. Oxford: Oxford University Press. Reardon, S. F., E. Greenberg, D. Kalogrides, K. A. Shores, and R. A. Valentino. 2013. “Left Behind? The Effect of No Child Left Behind on Academic Achievement Gaps.” Unpublished manuscript. Reback, R., J. Rockoff, and H. Schwartz. 2011. Under Pressure: Job Security, Resource Allocation, and Productivity in Schools under NCLB. NBER Working Paper Series 16745. Cambridge, MA: National Bureau of Economic Research. Ritter, Gary, Nathan Jensen, Brian Kisida, and Josh B. McGee. 2010. “A Closer Look at Charter Schools and Segregation.” EdNext 10 (3): 69– 73. Rothstein, Richard. 2004. Class and Schools: Using Social, Economic and Educational Reform to Close the Black-White Achievement Gap. New York: Teachers College Press. ———. 2008. “Whose Problem Is Poverty?” Educational Leadership 65 (7): 8–13. ———. 2013. “Racial Segregation and Black Student Achievement.” In Education, Justice and Democracy, edited by D. Allen and R. Reich, 173– 99. Chicago: Chicago University Press. Rothstein, R., R. Jacobsen, and T. Wilder. 2008. Grading Education: Getting Accountability Right. Washington, DC: Economic Policy Institute. Rumberger, Russell W., and Patricio Gandara. 2015. “Resource Needs for Educating Linguistic Minority Students.” In Handbook of Research in Education Finance and Policy, 2nd ed., edited by Helen F. Ladd and Margaret E. Goertze, 585– 606. London: Routledge. Ryan, K. E., T. Gandha, and J. Ahn. 2013. School Self-Evaluation and Inspec-

184

References

tion for Improving U.S. Schools? Boulder, CO: National Education Policy Center. Sahlberg, Pasi. 2011. Finnish Lessons: What Can the World Learn from Educational Change in Finland? New York: Teachers College Press. Sass, Tim R. 2006. “Charter Schools and Student Achievement in Florida.” Education Finance and Policy 1 (1): 91–122. Satz, D. 2007. “Equality, Adequacy, and Education for Citizenship.” Ethics 117 (4): 623–48. Schneider, Mark, Paul Teske, and Melissa Marschal. 2002. Choosing Schools: Consumer Choice and the Quality of American Schools. Princeton, NJ: Princeton University Press. Schouten, G. 2012. “Fair Educational Opportunity and the Distribution of Natural Ability: Toward a Prioritarian Principle of Educational Justice.” Journal of Philosophy of Education 46 (3): 472– 91. Schweinhart, L. J., J. Montie, Z. Xiang, W. S. Barnett, C. R. Belfield, and M.  Nores. 2005. Lifetime Effects: The High/Scope Perry Preschool Study through Age 40. Ypsilanti, MI: High Scope Press. Silva, Fabio, and Jon Sonstelie. 1995. “Did Serrano Cause a Decline in School Spending?” National Tax Journal 48 (2): 199– 215. Snyder, Thomas D., and Sally A. Dillow. 2015. Digest of Education Statistics 2013. Washington, DC: National Center for Education Statistics, Institute of Education Sciences, US Department of Education. Sondheimer, Rachel Milstein, and Donald P. Green. 2010. “Using Experiments to Estimate the Effects of Education on Voter Turnout.” American Journal of Political Science 54 (1): 174– 89. Stroud, Sarah. 2015. “Egalitarian Family Values?” Law, Ethics and Philosophy 3: 180– 92. Sumner, L. W. 1996. Welfare, Happiness, and Ethics. Oxford: Oxford University Press. Swift, A. 2003. How Not to Be a Hypocrite: School Choice for the Morally Perplexed Parent. New York: Routledge. ———. 2008. “The Value of Philosophy in Nonideal Circumstances.” Social Theory and Practice 34 (3): 363– 87. ———. 2013. Political Philosophy: A Beginners’ Guide for Students and Politicians. 3rd ed. Cambridge: Polity Press. Swift, A., and Stuart White. 2008. “Political Theory, Social Science and Real Politics.” In Political Theory: Methods and Approaches, edited by D. Leopold and M. Stears, 49– 70. Oxford: Oxford University Press. Sypnowich, C. 2000. “Egalitarianism Renewed.” In Canadian Political Philosophy: Contemporary Reflections, edited by R. Beiner and W. Norman, 118– 31. Oxford: Oxford University Press.

References

185

Tooley, J. 1993. “Equality of Educational Opportunity without the State?” Studies in Philosophy and Education 12 (2–4): 153– 63. ———. 1995. Disestablishing the School: De- bunking Justifi cations for State Intervention in Education. New York: Routledge. ———. 2000. Reclaiming Education. London: Continuum International. Tuttle, Christina Clark, Brian Gill, Philip Gleason, Virginia Knechtel, Ira Nichols-Barrer, and Alexandra Resch. 2013. KIPP Middle Schools: Impacts on Achievement and Other Outcomes. Washington, DC: Mathematica Policy Research. Verba, Sidney, Kay Lehman Schlozman, and Henry E. Brady. 1995. Voice and Equality: Civic Voluntarism in American Politics. Cambridge, MA: Harvard University Press. Verstegen, Deborah A. 2011. “Public Education Finance Systems in the United States and Funding Policies for Populations with Special Educational Needs.” Education Policy Analysis Archives 19 (21): 1– 30. Wheelock, A., W. Haney, and D. Bebell. 2000. “What Can Student Drawings Tell Us about High-stakes Testing in Massachusetts?” Teachers College Record. http://www.tcrecord.org/ library/Abstract.asp?ContentId=10634. White, J. P. 1995. Education and Well- Being in a Secular Universe. London: Institution of Education, University of London. ———. 2011. Exploring Well- Being in Schools: A Guide to Making Children’s Lives More Fulfi lling. New York: Routledge. Witte, John F. 2000. A Market Approach to Education. Princeton, NJ: Princeton University Press. Wolff, Jonathan. 2011. Ethics and Public Policy: A Philosophical Examination. New York: Routledge. Wong, M., T. D. Cook, and P. M. Steiner. 2009. “No Child Left Behind: An Interim Evaluation of Its Effects on Learning Using Two Interrupted Time Series Each with Its Own Non- equivalent Comparison Series.” Institute for Policy Research, Northwestern University, WP- 09-11. http://www.ipr.northwestern.edu/publications/papers/2009/ipr-wp - 09 -11.html. Zimmer, Ron, Brian Gill, Kevin Booker, Stephane Lavertu, and John Witte. 2011. “Do Charter Schools ‘Cream Skim’ Students and Increase RacialEthnic Segregation?” In School Choice and School Improvement, edited by M. Berends, M. G. Springer, D. Ballou, and H. Walberg, 215– 33. Cambridge, MA: Harvard University Press.

Index ability grouping, defi ned, 11 accountability, 109, 110, 111, 113, 114; administrative, 112, 133; defi ned, 108; market-based, 112; political, 112, 133; two approaches, 108– 9. See also market-based accountability; outcomes-based accountability; political accountability achievement, 4, 12, 44– 69, 81, 119, 122– 24,126, 147, 150– 51; and analysis of gaps, 45,57– 67; defi ned, 44; as an educational good, 44– 69; gap, 4, 45, 124 Achievement First, 136 adequacy, 2, 32, 50– 54, 62, 66, 88, 93– 97, 110, 122, 126, 138, 157; and achievement, 50– 51, 53– 54, 62, 66; defi ned, 52; and funding, 88, 93– 97; as principle, 32, 52; and school accountability, 110, 126 administrative accountability system, 112– 31, 152; defi ned, 112 Armed Forces Qualifying Test (AFQT), 45 autonomy, 8, 22– 24, 61, 90, 101– 5; defi ned, 23; and schools, 74. See also school autonomy and parental choice benefiting the less advantaged, 2, 9–10, 34– 36, 94, 157; and achievement, 56, 62, 66; Boston Public Schools’ home-based choice plan, 9–10; and charter schools, 138, 149– 51, 153 Boston Public Schools, 9; as educational good, 44– 69; and funding, 88, 94, 97–101 Brown v. Board of Education, 1954, 61 bubble kids, 122

California Charter Schools Association, 135 California Class- size Reduction program, 82, 103 capacities for flourishing, 2, 22– 27, 76, 85, 110, 118, 127, 130, 137– 38, 145, 156 categorical funding streams, 102 Center for Research and Reform in Education, 81 Center for Research on Education Outcomes (CREDO), 150 Center on Education Policy, 118 central government, 75, 87, 90, 101, 105, 115–16; and accountability, 116; and school funding, 87, 90, 101, 105 central-level policy makers, 87, 90, 92– 97, 101– 5, 112 Charlotte-Mecklenburg schools, 126 charter management organizations (CMOs), 136, 140, 144, defi ned, 136 charter report card, 119 charter schools, 121, 133– 38, 140–48, 151– 53, 160– 61; in comparison to traditional schools, 149, 151– 54; defi ned, 133, 135; and school choice, 133– 34, 160– 61 childhood, 2, 37– 38 childhood goods, 2, 13, 37– 38, 88, 98, 110– 11, 124– 29, 131, 137, 148, 156; and accountability, 110–11, 131; and charter schools, 137, 148; defi ned, 2; and finance, 88, 98; listed, 37– 38 class size, 82, 101– 3, 106, 108, 111, 134– 35, 139, cognitive skills, 2, 20, 23, 44–47, 52, 67– 68, 109–10, 117–19, 138, 156

188 common good, 30 cost-function method, 96, 107 costs, variation in, 91 cream skimming, 150 data, 81, 160 decision Makers, 1– 3, 9–14, 31, 69, 74– 84, 99, 102, 104, 106, 136, 159 decisions, 1– 3, 5, 20, 76, 88, 90– 92, 155; and funding, 88, 90; and impact on education, 1– 2 democracy, 30 democratic competence, 22, 24– 25, 76, 85, 110, 118–19, 137– 38, 145, 156, defi ned 24 differentiation of funding, 90– 92, 94– 99, 104 dignity, 2, 26 distributive principles, 7, 32– 36, 44, 52– 57, 128 distributive values, 2, 31– 36, 77, 88– 89, 110, 122– 23, 125– 29, 131, 139–40, 149, 152; and accountability systems, 110, 122– 23, 125– 29, 131; and charter schools, 139–40, 149, 152; and funding 88 Duncan, Greg, 74 economic, social, and cultural status (ESCS), 63– 65 education, 1– 2, 4, 7– 8, 14, 19, 30– 39, 42– 43, 68– 69, 73– 78, 80– 88, 103– 9, 112–18, 123, 128, 134, 137, 141, 148–49, 155– 59; and charter schools, 148–49; children’s, 4, 7, 32, 38– 39, 68, 75, 77, 115; defi ned, 1; public, 36, 86– 87, 89; spending on, 37, 41, 85, 90, 107; 42, 116 educational goods: 2, 9, 8–14, 19– 22, 25, 44, 51, 76, 88, 109–10, 117–19, 121, 124, 130, 137, 145,156, 160; and accountability, 108– 31; and charter schools, 137– 38, 141, 144, 148, 154; defi ned, 2, 19, 44, 76, 88, 109, 137, 156; and funding, 86, 88–107 education management organizations (EMOs), 135– 36; defi ned, 136 Education Review Office (ERO), 125– 26, 128 Elementary and Secondary Education Act, 92 elementary education, 86– 87, 114

Index empiricism, 1, 4– 7, 12, 25, 74, 82, 104, 138, 159– 61; evidence, 4– 5, 74, 104, 138, 159– 60; questions, 7, 12, 56 equality, 9–10, 12–13, 32– 34, 54– 55, 61– 63, 66– 67, 88, 93– 98, 102, 110, 138, 149– 51, 157; and achievement, 54– 55, 61– 63, 66– 67; and charter schools, 138, 149– 51; defi ned, 10; and funding, 88, 93– 98; and school accountability systems, 110 equity, 2, 30, 43, 157 evidence-based approach for costing schools, 107 flourishing, 1– 2, 20– 22, 44–48, 51, 62, 67– 69, 98, 117; and achievement, 44–45, 47–48, 51, 62, 67– 69; and school accountability, 117; six relevant capacities, 22, 47 framework, 3, 75– 84, 105, 103, 153, 155, 159 freedom, of resistance and occupation, 37, 40–43 137, 158 Friedman, Milton, 41–42 funding, 10, 30– 31, 42, 69, 83, 85–107, chapter 5, 108, 113, 121, 131, 134, 151, 158 GDP, 86 General Educational Development program (GED), 46, 68 geographic cost indexes, 95 goods, use of word, 2. See also childhood goods; educational goods group-based stereotypes, 58 Harlem Children’s Zone schools, 147 Heckman, James, 46 incentive-based systems, 109, 112, 115–16, independent values, 19, 31, 36–42, 52, 67, 77, 89– 90, 93, 99, 106, 110–11, 116, 123– 29, 137, 158, 160– 61; and accountability, 110–11, 123– 29; defi ned, 31, 36; and funding, 88 Individuals with Disabilities Education Act, 92 Institute of Education Sciences, 81 internal processes and practices, 113–14, 124– 31 IQ, 59

Index Jencks, Christopher, 61 judgments, 4– 5, 23– 24, 29, 41, 69, 73, 76, 79– 84, 106, 124, 129, 155; about evidence, 82; probabilistic, 82– 83; and values, 4– 5, 73, 106, 124, 129, 155 Knowledge Is Power Program (KIPP), 136, 147–48 language, 2, 6, 7, 74, 91, 96, 118, 143, 147,155– 57, 160 lottery, 135, 147 market-based accountability, 112–15, 129 markets, 41–43, 112 method, 1– 2, 22, 69, 73– 78, 81, 96, 102, 106– 7; for integrating values and evidence, 69, 73– 76, 81 Minnesota Family Investment Program, 75 mixed-ability classrooms, 11 Montessori schools, 135 Murnane, Richard J., 74 National Assessment of Educational Progress (NAEP), 58, 60, 118, 120– 22 Nation’s Report Card. See National Assessment of Educational Progress (NAEP) New Hope program, 74– 75 No Child Left Behind Act (NCLB) 2001, 114, 117–19, 121– 23 no- excuses schools, 136– 37, 145–48 noncognitive skills, 46–47 normative standards, 4, 49– 51, 61, 65 Office for Standards in Education, 114 Ofsted reviews, 128 open enrollment, 134 Organization of Economic Co- operation and Development (OECD), 63– 65, 86, 114 outcomes-based accountability, 104, 113– 14, 116–18, 123– 29, 131, 134– 35, 143, 154; and administrators, 113–14, 131, 134– 35, 143; and inspection, 125– 29 parents, 1, 3– 5, 7, 9–11, 19, 23– 25, 30, 37– 42, 52, 67, 77, 80, 89, 92– 93, 99, 105– 6, 110–13, 129, 132– 54, 158– 61; choices,

189 132– 54; as decision makers, 1, 3– 5, 7, 11, 30, 80, 99; interests, 37, 42, 38– 39, 52, 67, 77, 89, 93, 99, 106, 111, 116, 129, 137, 158, 160– 61. See also school autonomy and parental choice Perry Preschool Project, 68 plausibility, 79 policy, 1– 6, 74, 73– 84; evaluation of, 3– 5, 73– 84 political accountability, 112–13, 115, 129, 133 positional goods, 33, 53, 62– 63, 66, 69; defi ned 33 professional judgment method, 96, 106 proficiency grouping, 10–14 Program of International Student Assessment (PISA), 64, 123 random- controlled trial (RCT), 81– 82 regulatory policy, 108, 111–13, 115, 130, 134, 141–44 residential location, 40 Rocketship Education, 147 Rose school fi nance case, 1989, 52 Rothstein, Richard, 74– 75 SAT, 74 school accountability, 73, 102, 108– 31, 138, 160 school autonomy and parental choice, 73, 132– 54, 160 school choice, 132– 33, 139–40 school diversification and differentiation, 132– 33, 137, 139 school fi nancing, 73, 85–107, 160 secondary education, 86– 87, 114, 121 Serrano v. Priest, 99 social justice, 2, 30 social science, 2– 3, 13, 44, 81, 94, 117–18 social scientists and researchers, 3, 5, 67, 94, 129, 159 socioeconomic status (SES), 63– 64 standard deviation, 49 streaming, 11 student-teacher achievement ratio (STAR), 82 student needs, variation in, 91 summer learning gap, 97 supplementation, 90, 92– 93, 99–101, 104– 5

Index

190 test-based accountability, 104, 114,118, 120– 23, 125, 127,128, 131, 135, 138, 154. See also outcomes-based accountability Tooley, James, 42 tracking, 11 trade- offs, 3–14, 31, 35, 44, 51– 52, 69, 77, 83– 85, 88, 94, 105– 6, 111, 154– 60; and achievement, 44, 51– 52, 69; and charter schools, 152, 154; among distributive values, 9–11, 35, 157, 160; among educational goods, 7– 9, 11–14, 35, 154– 57, 160; and fi nance and governance, 85, 88, 94, 105, 160; and school accountability, 111

Uncommon Schools, 136 utility, 83 value, 1– 6, 19, 30–43, 73–131, 155, 158– 69; in combination with evidence, 73– 84, 158– 60; defi ned, 4; and fi nance and governance, 85–107; and school accountability, 108– 31; three kinds of, 19 vouchers, 133– 34 weighted funding system, 91 World Bank, 132 World War II veterans, 46