Achieving High Educational Standards for All: Conference Summary (2002)

Ronald F. Ferguson with Jal Mehta¹

Evidence from the middle of the 20th century provided little if any reason to expect that closing skill gaps in reading and math could substantially reduce inequality in black-white earnings (Cutright, 1972, 1974; Jencks et al., 1972). Today, at the beginning of the 21st century, conditions are different because the value of such skills to employers has grown, and racial barriers to employment have weakened (Murnane et al., 1995). By the late 1980s, disparities in reading and math skill predicted half or more of the hourly earnings gap between black and white young adults (Johnson and Neal, 1998; Ferguson, 1995).² Discrimination on the basis of race or ethnicity still affects who gets some jobs (e.g., Fix and Struyk, 1991). Indeed, reading and math scores are much stronger predictors of what people earn, once employed, than of whether they are employed. Nonetheless, because skills help to determine earnings, skill disparities among racial and ethnic groups help to perpetuate historical inequities in every aspect of life that depends on financial resources.

The good news is that achievement gaps among racial and ethnic groups in the United States are smaller than they were several decades ago. The reading-score gap between black and white 17-year-olds in the National Assessment of Education Progress (NAEP) was less than half as large in 1988 as in 1971, when the Educational Testing Service (ETS) first administered the NAEP Trend Assessment (also see Lloyd et al., this

volume). Similarly, the gap between Hispanics and whites was 40 percent smaller in 1990 than in 1975.³ For both blacks and Hispanics, there were years around 1990 when the gap with whites in math scores was more than 40 percent narrower than in 1973. The bad news is that progress stopped around 1990.⁴ In 1999, when the latest NAEP test was administered, large differences remained between average scores for blacks and Hispanics on one hand and whites and Asians on the other.⁵

Focusing on test-score disparities, this paper concerns what researchers have learned about equalizing educational opportunities and outcomes among racial groups, primarily blacks and whites, in the last half-century. While progress is evident and many milestones have been achieved, especially in civil rights, policy measures focused on rights, resources, and testing requirements for students have not achieved their full promise for raising achievement and narrowing gaps. Failure to foster high-quality instructional practices in all schools and classrooms and for all students is strongly implicated in these disappointing results. Now is a time to supplement other policies with a more determined, high-quality research-based emphasis on improving what happens in classrooms. We agree that the types of incentives being imposed by the current standards movement are important, but principals and teachers need help knowing how best to respond to them. Chapter 6 in this volume describes what progress might entail, by discussing instructional regimes that recent research has shown to be effective. It reports experience introducing those regimes into schools and classrooms and working to make them routine.

This paper provides some historical background with an emphasis on what research has shown about the effectiveness of past policies. We present a historical overview that touches on a number of topics related to rights, resources, and requirements in education reform over the past half-century. Then we focus in more detail on research about ways that desegregation, grouping and tracking practices, and class sizes relate to achievement disparities. We focus on these topics because of their interdependence with instructional quality and their historical and contemporary policy importance. For example, we ask, “Are grouping and tracking practices among the reasons that racial desegregation seems to produce only small achievement gains, and how does the answer relate to instructional quality?” And “Do we know enough about class size effects to justify strong claims about the advantages of class size reductions for raising achievement, compared with investments in instructional quality?” Our aim is to present an informed perspective on what research has established and what remains to be learned about a number of important questions.

One hundred and five years ago, the U.S. Supreme Court upheld the doctrine of “separate but equal” in the case of Plessy v. Ferguson. Although the conflict was over passenger accommodations on the East Louisiana Railroad, the “separate but equal” doctrine that the Court’s decision affirmed was codified in state laws governing schools and virtually all other types of public accommodations in the South, where the majority of black Americans lived. Representing an eight-person majority, Justice Henry Brown wrote: “The object of the [Fourteenth] Amendment was undoubtedly to enforce the absolute equality of the two races before the law, but in the nature of things it could not have been intended to abolish distinctions based upon color, or to enforce social, as distinguished from political equality, or a commingling of the two races upon terms unsatisfactory to either.”

Half a century later, the doctrine of separate but equal still dominated the South, but the question being litigated was whether enforced segregation in public schools deprived black children of equal protection under the U.S. Constitution. On May 17, 1954, Chief Justice Earl Warren of the U.S. Supreme Court issued the court’s decision in the case of Brown v. Board of Education (Martin, 1998). The Court’s opinion granted that it might be possible with segregation to achieve equality of “tangible factors”— things that money can buy—but the Court rejected the idea that separate could be equal or that laws maintaining segregation could provide equal protection under the U.S. Constitution. Informed by the work of social scientists, including the black psychologist Kenneth Clark, the justices wrote the following about the harm that segregation was doing to black children: “To separate them from others of similar age and qualifications solely because of their race generates a feeling of inferiority as to their status in the community that may affect their hearts and minds in a way unlikely ever to be undone” (Martin, 1998). Thus, Brown v. Board of Education was not merely about equality of resources. It was also about children’s “hearts and minds” and “status in the community.” The decision struck down the doctrine of separate but equal. It was a landmark event.⁶

In challenging the separate but equal doctrine of the Jim Crow South, the plaintiffs in Brown v. Board of Education aimed to challenge white supremacist ideology and the moral injustice of forced segregation. In addition, they hoped that giving black children access to the schools and classrooms in which white children studied would help to equalize educational resources and academic outcomes. Unfortunately, implementation of the court order was exceedingly slow and limited (see below). Most of the school integration that actually happened in the South took place after the Civil Rights Act of 1964 and after other court orders took

effect in the late 1960s and early 1970s. Evidence regarding the impacts of desegregation on achievement and other outcomes when it finally happened is mixed (see below).

Parallel to the desegregation cases, there have been dozens of school finance cases in state and federal courts, with plaintiffs alleging that patterns of school funding violated state or federal constitutions (see Rebell, this volume). When successful, these cases have helped to increase spending in low-income districts (Evans et al., 1997, 1999).⁷

After the mid-1970s, forced integration was no longer the standard judicial remedy for segregation, and the desegregation cases came to resemble the school finance cases, especially in the North. They focused increasingly on state aid and compensatory education. James E. Ryan writes, “In sum, school desegregation and school finance litigation have converged around money. That poor and minority schools will remain separate from white and wealthier schools [because they are in different political jurisdictions] appears to be taken as a given, and, if anything, is reinforced by the fact that advocates are fighting not over integration but resources” (1999:272). Courts in the 1990s began releasing districts from desegregation orders issued in the 1970s. The likely result is that court-ordered desegregation will soon be only a memory.

Around the time that the Civil Rights Act of 1964 set the wheels in motion to enforce desegregation orders, the War on Poverty introduced the federal Head Start program (in 1965) in order to give children from disadvantaged homes a “head start” on school success. In addition, Chapter I (now Title I) of the Elementary and Secondary Education Act (ESEA) of 1965 was intended to supplement academic resources for low-income children who needed extra support in the early grades. Head Start and

Title I were not explicitly race targeted, but a major motivation among supporters was to reduce racial inequities. Over the years, recipients of these services have included large numbers of poor minority children. Title I is a funding stream to supplement school-level resources and is not a highly prescriptive intervention. Schools have great discretion in how the funds are used. Before reforms in 1994, federal legislation targeted Title I funds to the early elementary grades, with the intention that funds should assist only the students in those grades who were most in need of supplemental support. However, reforms passed in 1994 encouraged support for students across all grade levels, not only the early elementary years. They also encouraged whole-school reforms in high-poverty schools and an increased emphasis on accountability. Some critics of the reforms, including Farkas and Hall (2000), argue that the reforms of 1994 dilute the focus on high-need students by spreading funds across all grade levels and, furthermore, that whole-school programs encourage

substitution of Title I funds for local spending that would have happened anyway.⁸

Others (e.g., Slavin, 2001) applaud the reforms and argue that, with refinements, they offer the potential for broad-based improvements in high-poverty schools. Indeed, Slavin (2001:236) asserts, “Whatever the average effects of Title I resources may be, Title I is the crucial resource for reforming the education of students in high-poverty schools. Whenever an inner-city or poor rural school is found to be achieving outstanding results with its students by implementing innovative strategies, these innovations are almost invariably funded primarily by Title I.” This includes the Success For All (SFA) Program, a whole-school reform designed and disseminated by Slavin and his associates at Johns Hopkins University. Success For All has been adopted in hundreds of schools across the nation.

Neither of the two large-scale evaluations of Title I has reached the conclusion that it substantially narrows achievement gaps between disadvantaged and middle-class students, as policy makers intended (Puma et al., 1997; Carter, 1983). Reanalyses by Borman et al. (2001) of the data from the study conducted by Puma et al. (1997) and a quantitative synthesis by Borman and D’Agustino (2001) of state-level studies produce somewhat more optimistic conclusions, but none of the studies finds effects that are impressively large.

To put these findings in proper perspective, it should be noted that all of the estimates depend on contestable decisions about how to estimate what would have happened to Title I student achievement in the absence of Title I support. Even if Title I has failed to narrow the achievement gap between disadvantaged and middle-class students, it might nonetheless have helped to keep the gap from widening, and to a degree that existing studies have no way to reliably estimate. The most definitive and defensible methodology for this purpose would be to randomly assign students to a treatment group that receives Title I support or to a control group that does not. There have been no such random assignment studies of Title I, probably because it would seem unfair to the control group. Again, the possibility exists that outcomes might have been worse in the absence of Title I, but existing studies have no way of establishing it because they have not used random assignment (or, alternatively, carefully executed quasi-experiments that use comparison instead of control group designs).

Findings on the effectiveness of Head Start are somewhat more positive than those for Title I. Specifically, most studies find that Head Start raises school readiness, as measured by achievement test scores (see the discussion of this point in Oden et al., 2000). However, most also find that the initial advantage fades during the elementary years, such that achieve-

ment scores among Head Start graduates eventually resemble those of nonparticipants from similar backgrounds. The most likely reason for the fade-out is that Head Start graduates attend inferior schools that fail to motivate them sufficiently and to build optimally on the skills they bring (Campbell and Frey, 1970; Lee and Loeb, 1995; Currie and Thomas, 1995). There is evidence (though not much) that, with favorable conditions, fadeout is not inevitable. For example, preschool programs, including Head Start, have sometimes shown sustained benefits (including but not limited to test scores) all the way into adulthood (Barnett, 1992; Oden et al., 2000). Increasing the long-term sustainability of the gains generated by Head Start depends almost surely on improving the primary and secondary schools that Head Start graduates attend, including those assisted by Title I.

While the federal government was introducing Head Start and Title I in 1965, local districts were continuing a century-long trend toward reducing class sizes for children of all backgrounds. Classes historically have been larger in schools that blacks have attended (Coleman et al., 1966; Boozer, Krueger and Wolkon, 1992). However, class-size reductions have been larger for blacks than for whites. By 1990, the national pupil/ teacher ratio for all races and ethnicities in elementary school classrooms was only 70 percent of what it was in 1965 (18.9 pupils per teacher in 1990 versus 27.6 in 1965), and there was no clear remaining difference among racial groups.⁹ Most of the reduction that took place after 1965 was complete by 1980 (Table 65, U.S. Department of Education, 2000). Debate continues about the effect on achievement, but there are reasons to believe that it was positive at the elementary school level, especially for blacks (see below).

The period from the mid-1960s to the early 1980s was also a time when schools went “back to basics.” The back-to-basics movement spread rapidly during the 1970s in response to media attention to such things as falling SAT scores. It was driven to a substantial degree by parental concern that their children were not acquiring basic skills. It produced systemic and results-based accountability reforms that were precursors to those of today, raising many of the same issues about “teaching to the test” and diluting curricula. Jennifer O’Day and Marshall Smith write (1993:258):

By 1985, 35 states were mandating statewide minimum competency testing (MCT) and 11 required passing such tests as a prerequisite for graduation. Some also used the scores to determine eligibility for remedial programs and promotions.

O’Day and Smith believe that instruction focused heavily on basic skills in preparation for minimum competency tests was among the important reasons that black students’ scores rose on the NAEP during the 1970s (for 9-year-olds) and during the 1980s (for 17-year-olds). However, analysts who have looked closely at the timing of the gains are skeptical for a number of reasons (Office of Technology Assessment, 1992: Chapter 1). First, scores rose on tests even in states without MCT. Second, scores began to rise before MCT could have had much of an impact. Third, all states were reporting performance of their students on nationally normed achievement tests above the national average, which is a statistical impossibility. Fourth, the degree to which NAEP and SAT scores rose varied from place to place in ways that seem inconsistent with MCT as the explanation. It seems most likely that MCT and rising scores were both products of the movement to strengthen basic skills, but that MCT was not a key causal factor in the rise in NAEP scores.

The critique that ultimately weakened the basic skills movement was that it did not focus enough on higher-order thinking. Students, it was argued, needed much more than basic skills. This meant they needed teachers who had more than basic skills themselves. Attention during the 1980s shifted to improving the quality of new teachers. Only 3 states required initial certification testing of new teachers in 1980, but 42 states did by 1990. States also adopted measures encouraging students, including minorities, to take more academically advanced courses (see the discussion and statistics in Ferguson, 2001b).

From the late 1980s to the present, the nation has searched actively at both state and federal levels for ways of improving whole schools and whole school systems. Ideas about “systemic reform” and “standards-based accountability” have been influential at every level of policy making.¹⁰ A more extensive discussion of this latter period is beyond the scope of this chapter, but see the papers in Ravitch (2001).

Compared with 1954, much has changed and much has not. On one hand, segregation is still high and, as Gary Orfield mentioned at the conference, improving schools for the most disadvantaged children with current levels of segregation and isolation is a gargantuan task. On the other hand, political leaders at all levels of society are claiming public education as their number one concern, and they are talking publicly and opti-

mistically about prospects for improving outcomes for even the most disadvantaged children. That’s progress from 1954, but there remains a long way to go.

In 1955, the Supreme Court issued the implementation order for Brown v. Board of Education, known as Brown II. In it, the Court ordered Southern states to desegregate their schools with “all deliberate speed.” However, it defined neither “desegregation” nor “all deliberate speed.” Instead, the ruling left the interpretation and enforcement of Brown II to federal district courts in the South.¹¹ Under heavy pressure from local Southern politicians, schools remained heavily segregated, with only 1 out of 50 Southern black children attending integrated schools in 1964 (Orfield and Eaton, 1996:7). For roughly a decade, the decision in Brown v. Board of Education did little to raise achievement among Southern children, because it did little to affect the conditions of their schooling.

Southern patterns of segregation in public schools persisted until the Civil Rights Act of 1964 prohibited discrimination in all schools receiving federal dollars. Four years later, in another landmark decision, Green v. County School Board, the U.S. Supreme Court declared that segregated or dual systems of public education had to be dismantled “root and branch.” The mandated desegregation applied to facilities, staff and faculty, extracurricular activities, and transportation. By 1970, Southern schools were less segregated than schools in any other region.

Desegregation in the North and the West faced a different set of challenges, because large-scale white suburbanization after World War II had left too few whites in cities to achieve meaningful integration without crossing city-suburb lines. Interdistrict desegregation plans sprung up in the wake of the civil rights legislation, and segregation decreased in the North. But there was a limit to how far this movement would go. In 1974 the Supreme Court in Milliken v. Bradley overruled a metropolitan-wide desegregation plan under which children from Detroit would have integrated with children from the mostly white suburbs. Absent a court finding that suburban districts had conspired to maintain segregation in Detroit schools, the Court ruled that there was no legal reason that the suburbs of Detroit should be part of the remedy.¹² This made it effectively impossible for Detroit to achieve extensive integration, since there were too few whites left in the city. Instead, in a companion case, Milliken v. Bradley II, the remedy approved by the Court required the state of Michigan to help fund remedial and compensatory education programs. Court rulings in the Milliken cases, combined with continuing outmigration from

cities to suburbs by the largely white middle class, effectively foreclosed the possibility of meaningful desegregation in the North and the West.

In recent years, segregation has begun to increase due to continuing suburbanization of whites and because courts are no longer writing and enforcing desegregation orders.

A number of studies in the 1960s and 1970s evaluated the effects of the desegregation orders. Reviews of this literature have pooled estimates from multiple studies to reach summary conclusions. They suggest the following: (1) white achievement is entirely unaffected by desegregation;¹³ (2) desegregation does not lead to an increase in black mathematics achievement; (3) desegregation does tend to raise black reading scores, but by relatively small amounts, probably between 0.06 and 0.26 standard deviations; and (4) gains are likely to be greatest among the youngest children (Cook, 1984; Schofield, 1995). These studies have been subject to a number of methodological criticisms, the most important of which is that the time frame for the majority of the studies is far too short (Crain and Mahard, 1983). Many studies estimate effects on achievement after only one year, and none estimates the effect of desegregation on the cumulative achievement of black students over a number of years (Jencks and Mayer, 1990). Another problem is that the studies rarely attend to the details of implementation, and thus the factors that create greater gains from integration in some schools than others are almost entirely unknown. Finally, it is unclear whether the effects of court-ordered plans from nearly three decades ago can be fairly generalized to today (Schofield, 1995).

Beyond studies of court-ordered desegregation, a parallel literature seeks to understand whether natural variation in the level of school integration can explain differences in student achievement, controlling for family background factors. While these studies use nationally representative data for schools that are not operating under desegregation orders, they cannot overcome the possibility that selection bias has affected their findings. In this context, the selection bias issue is that black families who send their children to integrated schools may differ in unmeasured ways from black families that do not. Therefore, some of the estimated effects that studies attribute to integration might instead be the result of these unmeasured differences in families and children. For example, we know that, on average, more-advantaged blacks in nationally representative

data are the ones more likely to attend integrated schools. Some characteristics, such as parental ambition, are not accounted for in the research. Therefore, the positive effect of integration is likely to be overestimated if more ambitious black parents are the ones more likely to integrate (which seems likely).

With this caveat in mind, Jencks and Mayer (1990) in their review of the literature on the racial composition of schools suggest that the best evidence comes from the Equality of Educational Opportunity study (Coleman, 1966) and the National Longitudinal Survey (NLS, 1972). They show (by one achievement-score measure) that blacks who attended predominantly white Northern schools during the late 1960s and early 1970s scored 0.30 standard deviations higher than blacks who attended all-black Northern schools. Phillips (2001) analyzed the same issue using the Prospects data—a nationally representative sample collected in the early 1990s to study the effect of Title I. Phillips found effects of integrated schools on black reading achievement of about 13 percent of the black-white reading gap, but she found no effect on math achievement. She speculates that the repeated findings in the literature that reading, but not math achievement, is affected by integration, suggest that black students in integrated schools benefit not so much from better instruction or more advanced curricula as from interaction with teachers and peers who speak the mainstream dialect.

There has been some research on ways that integration affects other important outcomes besides test scores. The latter include rates of high school completion, college attendance and completion, and lower rates of delinquency and teen childbearing (Mayer, 1991). Authors who emphasize this longer list of impacts are careful to point out that the advantages of integration probably come less from racial mixing per se than from middle-class educational environments (Orfield and Eaton 1996:57):

Consistent with this view, there is a considerable literature that documents the reduced opportunities available in schools that have extremely high concentrations of poverty. Schools in high-poverty areas are less likely to offer college preparatory classes, and they have much higher rates of teachers’ teaching out of subject areas, greater teacher turnover, and lower test scores. Parents are less likely to be involved in school

affairs, less able to ensure high standards, and less likely to pressure administrators to fire or transfer bad teachers (Kahlenberg, 2001). While there has been considerable debate in the academic literature about how much these school-level factors affect achievement independent of student background, it is undeniably true that the interaction of poverty, segregation, and inadequate school resources heavily disadvantages the poor and minority students who attend high-poverty schools.

Accordingly, the most striking effects of integration have been measured in case studies of interdistrict programs in which a limited number of city children are bused to suburban schools. For example, Wells and Crain (1997) found that of the students who remained in the suburban schools they studied in the St. Louis metropolitan area, approximately 50 percent graduated from high school in 1994, compared with 24 percent of all students in central-city public schools. Of the suburban stayers, 68 percent of those who graduated from high school went on to college, about two-thirds entered four-year colleges and the rest two-year colleges. Of the much smaller proportion who graduated from city schools, about 50 percent went to college, with one-third of those attending four-year institutions. Rosenbaum’s (1995; Rosenbaum et al., 1993) study of the Gautreaux program concerned a semirandomized program comparing families that moved to the Chicago suburbs with those who moved to other urban locations. Rosenbaum found greatly increased educational outcomes among suburban movers. Specifically, suburban students (who experienced both residential and school integration) were four times less likely to drop out of school, more than twice as likely to attend college, and almost seven times more likely to attend four-year colleges compared with city students (Rosenbaum et al., 1993:1533).

Unfortunately, there are methodological problems with both the St. Louis and the Chicago studies, so their findings are far from definitive. The Wells and Crain study compares the students who stay in suburban schools with mean values for city public schoolchildren, and there is substantial reason to think that this might be a source of selection bias. The Rosenbaum study may also suffer from selection bias, because the response rate for those whom they followed up was fairly low, raising the risk of self-selection.

With a more robust study design, the federally sponsored residential mobility program Moving To Opportunity (MTO) began operating in 1994 in Baltimore, Boston, Chicago, Los Angeles, and New York. Qualified households were low-income families with children living in public housing or Section 8 project-based housing in selected high-poverty census tracts. Families who volunteered were assigned randomly to one of three groups. The experimental group was offered rental subsidies that could be used only for private-market housing in census tracts with 1990

poverty rates under 10 percent. A “Section 8-only” comparison group received rental subsidies, but with no requirement to live in a low-poverty census tract. Members of the control group received no subsidy at all. Because of the random assignment design and careful implementation, this project is not subject to the same concerns about sample selection bias that plague other studies.

So far, only the Baltimore study of Moving To Opportunity has reported education findings. Ludwig et al. (2001) report that elementary schoolchildren in the experimental group scored higher than the control group by about a quarter of a standard deviation in both reading and math. Elementary schoolchildren in the Section 8-only group score higher than the controls in reading but not math. Achievement data for adolescents are too limited to draw specific conclusions about effects on test scores. There are, however, some general conclusions: “The teens in the experimental and Section 8-only groups experience a higher incidence of grade retentions than controls, and may experience more disciplinary actions and school dropout as well. These findings are generally robust to problems of missing data and to decisions about the specific estimation approach that is used” (Ludwig, 2001:4). There is no way to know at this time what the long-term effects will be.

All in all, evidence on the impacts of racial and income mixing varies from study to study. A tentative conclusion is that a combination of racial and socioeconomic mixing often improve outcomes—particularly nontest score outcomes—for nonwhite students who might otherwise attend low-income segregated schools. However, the research generally suffers from methodological problems that limit the ability to draw firm, confident conclusions.

As the decades-long struggle to achieve integration continues, supports and incentives for schools to maximize achievement outcomes for minority children after integration occurs need to be strengthened so that all children will be well served. The work that Brown v. Board of Education and other desegregation cases began remains undone as long as segregation remains as extensive as it is and as long as children’s hearts and minds are not nurtured in ways that propel them toward their potential in whatever schools they attend. To maximize the value of mixing, colocation of students under the same school building roof is only the first step in a more elaborate process of social and academic adjustment.

Why are the effects of integration so small? If integrated (especially middle-class) schools tend to be superior to segregated schools in their capacity to produce positive student outcomes and those capacities are

applied effectively to benefit all students who attend them, then we should expect to find more positive gains from integration than most studies show. Surely, there is more potential for a poor child to thrive in a school that is mixed by race, ethnicity, and socioeconomic status than in one in which teachers and administrators are overwhelmed by a high concentration of children who are poor and socially isolated from mainstream society. However, as reviewed above, research findings tell mixed stories about the degrees to which such potential has been harvested over the past several decades. One possibility is that when poor and minority children entering a school are systematically different in preparation and identity from the children that the school is accustomed to serving, integration without accommodation may yield far less than the full potential of the opportunity. The following passage is illustrative (Ladson-Billings, 1994:60-61):

The quotation is from Gloria Ladson-Billing’s book, Dreamkeepers: Successful Teachers of African American Children. It reports an experience from the early 1960s. Even today, however, it remains true in racially integrated schools that black students are overrepresented in lower level-classes. It also seems to be common (or at least it remains a common perception) that lower level-classes are less well taught than those at higher levels.

Consequently, tracking and ability grouping are leading suspects for why integration has not produced greater benefits for minority children (Oakes, 1985; Braddock and Slavin, 1993). They provide means by which students attending the same schools may nevertheless have different instructional experiences. Because of differences in their family backgrounds and academic preparation (Phillips et al., 1998) and perhaps also because

of biases (see below), children are frequently grouped for instruction in combinations that are more homogeneous by race and socioeconomic background than the school is as a whole. However, as we posit below, grouping and tracking are not necessary in order for some children to be served less effectively than others, especially if schools do not accommodate well their instructional needs.

Ability grouping refers to elementary school practices that separate children for instruction either within or between classrooms, based on teachers’ judgments. Ability grouping after elementary school often occurs in the context of what historically has been called tracking and what more recently has been called “leveling” because tracking has acquired a pejorative connotation associated with more rigid structures of the past (Loveless, 1998). Courses at higher levels cover more advanced material and may require more work. Currently, the standard arrangement is that no student is officially forbidden from entering a course at any level. Nonetheless, race, gender, and socioeconomic imbalances frequently develop. Explanations include differences in proficiency, in the advice received from parents, counselors, and teachers, and students’ own preferences to be with their friends.¹⁵

For at least the past century, there have been recurrent debates among educators about whether ability grouping and tracking are helpful or harmful, especially for low achievers and minority students.¹⁶ In its standard form, the debate confuses at least three questions, on which we comment in what follows. The first concerns whether groups at all levels receive the same quality of instruction and, if not, what the implications are for whether the grouping should continue. To the extent that research addresses the question, the most common conclusion is that children in lower ability groups (and tracks) receive a lower quality of instruction than those in higher ability groups (and tracks) and therefore that grouping and tracking hurt students at the lower levels (Oakes, 1985; Finley, 1984; Schwartz, 1981; Metz, 1978; Gamoran and Berends, 1987).

Authors who conclude that ability grouping hurts students in lower-level groups or tracks and that it should therefore be abandoned assume that the same students would receive superior instruction in more heterogeneous groups or classrooms. However, there is ample evidence that even in heterogeneous, mixed-ability classrooms, low-achieving students often receive inferior treatment. Thomas Good (1987) reviewed the literature comprising studies of classroom observations aimed at detecting teacher biases. Citing multiple studies for each item, he identified the following ways that teachers were found to treat “highs” and “lows” differently when they were in the same classroom. They did so by waiting less time for “lows” to answer; giving low achievers answers or calling on someone else rather than trying to improve their responses (by giving

clues or repeating or rephrasing questions); rewarding inappropriate behavior or incorrect answers by low achievers; criticizing low achievers more often for failure; praising low achievers less often than highs for success; failing to give feedback to the public responses of low achievers; paying less attention to low achievers or interacting with them less frequently; calling on low achievers less often to respond to questions; seating low achievers farther away from the teacher; demanding less from low achievers (e.g., teaching them less, accepting low-quality or even incorrect answers, providing unsolicited help); interacting with low achievers more privately than publicly and monitoring and structuring their activities more closely; grading tests or assignments in a different manner, in which the high achievers but not the low achievers are given the benefit of the doubt in borderline cases; having less friendly interaction with low achievers, including less smiling and fewer other nonverbal indicators of support; providing briefer and less informative feedback to the questions of low achievers; providing less eye contact and other nonverbal communication of attention and responsiveness; evidencing less use of effective but time-consuming instructional methods with low achievers when time is limited; evidencing less acceptance and use of low achievers’ ideas.

We think researchers are correct when they conclude that instruction for lower-level tracks is routinely inferior to what higher-level tracks receive at middle school and high school levels. An analogous statement may apply for ability groups at the elementary school level. However, it does not follow directly that moving students into heterogeneously grouped arrangements will typically improve learning outcomes. As Good’s list reminds us, even in heterogeneously grouped schools and classrooms, minority students in integrated schools may receive inferior instruction if they are overrepresented among low achievers (Phillips et al., 1998) or among students from whom not much is expected (Ferguson, 1998a).¹⁷

It is important to note here that lower-level groups and classrooms are sometimes not even exposed to important concepts that they need to know in order to preserve future learning options (such as more advanced math classes). Sometimes this is given as a reason to abolish ability grouping and tracking (or leveling). There are surely circumstances in which the only way to give students what they need academically is to move them to heterogeneously grouped classrooms. However, there are surely also situations in which enriching the curriculum and quality of instruction while remaining within the ability-grouped regime is the most academically responsible option. As Loveless (1998) suggests, the answer to what is best will depend on the specific circumstances of the schools and classrooms under consideration.

A second question to consider is whether ability grouping matters if

the curriculum and the quality of instruction are similar across different grouping arrangements. The best studies to address this question are experimental and quasi-experimental studies in which students are randomly assigned (or carefully matched) to be grouped by ability, or not. Kulik (1992) presents findings from a meta-analysis of many such studies. The studies find overwhelmingly that ability grouping makes no difference to learning if there is no tailoring of curriculum or instruction to fit the proficiency of the group. In other words, if what is taught is the same, classmates’ proficiencies seem not to matter.¹⁸

Kulik’s analysis also indicates that when students are placed in ways that match their proficiencies and instruction and curriculum are tailored to the needs of the students in the group, students at all levels can benefit at least modestly from ability-grouped instruction. Less time is wasted on material that is too elementary or too advanced for any given group member, and instruction that serves one student well is more likely to serve others well too. Furthermore, contrary to the belief that ability grouping harms the self-esteem of low achievers, Kulik’s (1992) review found that ability grouping tends to lower self-esteem slightly among the high group and raise it slightly among the low group. Presumably, this is because compared with more heterogeneous arrangements, interpersonal comparisons within the high group are less favorable for the typical member and the opposite is true within the low group.

A final issue concerns whether students are placed in ways that match their proficiencies and potentials. Even if instruction is of high quality and tailored, some students may be misplaced. If teachers or guidance counselors use race, gender, or socioeconomic status as indicators of current or potential proficiency, there may be race, gender or socioeconomic bias in the placements. However, Ferguson’s conclusion after reviewing the literature (Ferguson, 1998a, 1998b) is that most placements are predicted by measures of past performance (and sometimes parental education), leaving little independent explanatory power for race.¹⁹

Ultimately, our conclusion is that how children are grouped for instruction seems less important than how well they are taught. For example, in the classroom that Gloria Ladson-Billings described in the quotation above, it is not at all clear that merging the entire “basic” class with students from the higher-level section would have been the best option for Gloria’s classmates. It extremely clear, however, that the basic class was being poorly taught. As summarized in Kulik’s (1992) review, experimental and quasi-experimental studies indicate that tailoring instruction to student skill levels and being responsive to progress produces the best outcomes, assuming that all of the instruction is of high quality.²⁰ However, as Good’s (1987) review shows, there are reasons to believe that students in regular classrooms who have weaker skills than their class-

mates (or who are racially different—Ferguson, 1998a) are often taught less effectively, even in heterogeneously grouped arrangements. The issues involved are complex, implicating racial and social class biases as well as problems with teachers’ instructional skills.²¹ Any simple, unidimensional, yes/no perspective with regard to ability grouping and tracking may be correct at some times, in some places, but it surely is wrong as a generalization.

The discussion of the past few pages is motivated by the question of why integration typically produces only meager achievement gains. It is impossible to make a reliable generalization based on the existing literature about whether disproportionate placements in lower-level ability groups and curriculum tracks are important to the answer. However, if nonwhite, non-Asian children in integrated schools are overrepresented among students who enter with less preparation than their peers, and if students with less preparation are taught less effectively in both ability-grouped and heterogeneously grouped classrooms, then the quality of instruction is part of the problem, no matter what the grouping arrangements may be. Racial biases in placement may be common as well, but the few studies that seek to measure this bias, controlling for scores and past performance, tend not to find it.²²

One of the most highly touted education policies of the Clinton presidency was a major class size reduction initiative that distributed $1.3 billion to help school districts recruit, hire, and train new teachers for the 2000-2001 school year. According to a progress report on the web site of the U.S. Department of Education (2000:1):

On December 22, 2000, Congress appropriated an additional $1.6 billion to cover the class size initiative during the 2001-2002 school year. However, President Bush was not persuaded that class size reductions should be a priority. Hence, the new Bush administration proposed to end the class size initiative and shift the funds toward efforts to improve teacher quality (Bush, 2001:12):

Below, we discuss the state of research on class size and consider implications with regard to investments in class size versus efforts to improve instructional quality.

The Bush administration’s deemphasis on class size was surely informed by the work of economist Erik Hanushek, based in part on the widely cited literature reviews that he has conducted over a period of two decades (e.g., Hanushek, 1986, 1997). Hanushek tabulates the findings from what economists call education production function studies. Education production functions use data on child, family, classroom, school, and community characteristics as predictors of academic outcomes such as test scores and graduation rates. In most such studies, the data have been generated for purposes unrelated to the study and seldom include all of the variables that the researcher’s theory suggests should be included. Indeed, a host of problems with both data and methodology make many of the studies heavily flawed.

Nonetheless, for class sizes, expenditures, teacher experience, and teacher education (i.e., master’s degrees), Hanushek’s literature reviews tabulate estimates of the degree to which these education “inputs” affect academic outcomes. Because it counts the number of estimates in each category, the method is known as “vote counting.” His well-known and widely quoted conclusion based on his reviews is that there is no consistent relationship between these school resources and school achievement. Instead, based on other work by himself and others (Rivkin et al., 2000; Sanders and Horn, 1995), he emphasizes that far and away the most important schooling input is teacher quality, but that teacher quality is difficult to measure with the types of data available for education production function analysis. The only possible exceptions, he has written, are direct measures of teachers’ own academic skills (Hanushek, 1986; Ferguson, 2000), and even these are imprecise.

The first major challenge to Hanushek’s summary of the literature came from statistician Larry Hedges and his colleagues. Hedges argued that Hanushek’s technique for summarizing the literature was likely to produce misleading results because it can easily fail to detect real effects.

Hedges, Laine, and Greenwald use formal methods of meta-analysis, which have greater statistical power (see Hedges et al., 1994a, 1994b; Greenwald et al., 1996; Hedges and Greenwald, 1996). Analyzing most of the same studies as Hanushek, they find that several kinds of resources, including class size, have beneficial effects on student outcomes. In the debate that has ensued, each side has applied assumptions that, if correct, favor its own position. These assumptions are not testable, so there is an intellectual standoff.²³ Our own view is that the literature on educational production functions is sufficiently flawed, especially in the way that it treats class size, that neither method is very reliable.²⁴

The latest and most effective challenge comes from economist Alan Krueger, who challenges the way that Hanushek selected estimates from the education production function literature to include in his summaries. For example, Hanushek’s 1997 review included 277 estimates of class size effects from 59 different studies. Different numbers of estimates were taken from different studies. Two studies contributed 24 estimates each, and 17 studies contributed only one apiece.

Since each estimate counted equally in the analysis, some studies counted much more than others did in a manner that affected the findings. According to Krueger and Diane Whitmore (2001), “The number of estimates Hanushek extracted from a study is systematically related to the study’s findings, with fewer estimates taken from studies that tend to find positive effects of smaller classes or greater expenditures per student” (p. 5). Krueger and Whitmore do not question Hanushek’s motives, but they do question his judgment. Krueger (2000) offers three alternative ways to combine estimates from the 59 studies. He writes, “In sum, all three of these alternatives to Hanushek’s weighting scheme produce results that point in the opposite direction of his findings: all three find that smaller class sizes are positively related to performance, and that the pattern of [positive] results in the 59 studies is unlikely to have arisen by chance” (p. 12).

Hanushek’s (2000) response to Krueger’s critique focuses on the differential quality of the studies from which he drew the estimates. Hanushek’s preference is to give greater weight to the studies that he regards as higher quality, and he argues that Krueger’s suggested schemes rely on “placing heavy weight on lower-quality and biased econometric estimates.” For example, he thinks that Krueger’s schemes give too much weight to studies that confound the effects of class size and other resources with unmeasured features of state-level policy regimes.

Hanushek’s summary of the education production function literature should no longer be treated as the definitive word on what that literature shows. Like Hanushek, both Hedges and Krueger are highly respected researchers. There are genuine, legitimate differences of professional judg-

ment that lead these researchers to contradict one another in the conclusions they draw from the education production function literature. Our judgments are more similar to those of Krueger and Hedges than to Hanushek, but there are others who agree with Hanushek. In any case, when Hanushek’s summary was the only one available it was regarded by many as definitive, but there is no research consensus today on what the education production function literature shows regarding the effectiveness of class size reductions.

Consensus is growing among researchers, including all those cited directly above, that the only way to reliably estimate the effect of class size is by doing experimental studies that randomly assign otherwise similar students (and teachers as well) to classes of different sizes. Experimental studies avoid most of the methodological pitfalls of education production function studies and are less controversial to interpret. In a perfectly executed class size experiment, class size differences would be the only systematic source of differential achievement among students who get assigned to different class sizes.

Tennessee’s Project Star, funded by the Tennessee state legislature in 1985, is the largest experimental study of class size ever conducted (see Word et al., 1990; Nye et al., 1993, 1994a, 1994b; Mosteller, 1995; Boyd-Zaharias, 1994). It randomly assigned 11,600 students and teachers to small classes (13 to 17 students) or large classes (22 to 25 students).²⁵ Students began with their assigned class sizes in kindergarten and continued through 3rd grade. A number of researchers have examined the Project Star data and all found that students in smaller classes outperformed those in the larger classes. Moreover, the effects were roughly twice as large for blacks as for whites. Black students in small classes in kindergarten through 3rd grade scored 7 to 10 percentile points higher than blacks in regular classes, while whites in small classes scored 3 to 4 percentile points higher than whites in regular classes (Finn and Achilles, 1990; Krueger and Whitmore, 2001). Krueger and Whitmore show that students from small classes kept their advantage in achievement all the way through high school (although in standardized units, it diminished) and they were also more likely to take ACT or SAT college admissions exams. For each outcome variable, the advantage of small classes was greater for blacks than for whites.

Students in small classes did not move into regular sized classes until after they completed the 3rd grade. The advantage of being in a small class measured in standardized units did not grow much after the first

year that a student was in a small class. This has led some, including Hanushek, to argue that the entire advantage accrues during the first year, and that the additional years probably are not necessary. A competing view is that students in small classes had to learn more each year than students in large classes to maintain the advantage that they did (as measured in standard deviation or percentile units). Since there was no arrangement to move some students into regular classes after only one or two years in small classes, there is no way to know who is right about the extra years’ payoff.

There have also been questions raised about whether other imperfections in the way the experiment was executed might have biased the results. However, analyses that test the sensitivity of the results to implementation imperfections have found no evidence that they distorted the findings (Nye and Hedges, 2000; Krueger, 1999). Furthermore, Nye and Hedges point out that the magnitude of the estimated effects in Project Star are quite close to those that Glass and Smith (1979) estimated in their meta-analysis of small-scale class size experiments from before 1980.

Krueger and Whitmore present four additional sets of findings that are far from definitive but nonetheless worth mentioning. First, using the education production function literature, Krueger and Whitmore (2001) examine class size effects from studies that provide separate estimates by race, socioeconomic status, or achievement level. In general, just as in Project Star, effects appear to be larger for low achievers, minorities, and students from low socioeconomic status households. Second, Krueger and Whitmore show using the Tennessee Star data that blacks in majority white schools and whites in majority black schools experience gains similar to their classmates. This suggests that the average behaviors or instructional needs of classmates or the skills of teachers in different schools may be the key reasons that small-class effects are higher in schools in which more blacks attend.²⁶ Third, in a very basic cost-benefit calculation, Krueger (1999) shows that the discounted present value of expected future earnings gains due to small classes appears to be sufficient to cover the costs of reductions in class size.²⁷ The calculation is admittedly rather crude, but it responds to those who say categorically that class size reductions in the early grades are a bad investment.

Finally, Krueger and Whitmore (2001) use the elementary school pupil/teacher ratio to predict the black-white test score gap among 4th graders for the 1970s through the 1990s. Like Ferguson (1998b), they found that the relationship between falling class sizes and the narrowing of the black-white test score gap among 4th graders on the NAEP is very close to what one would have predicted based on class size effects estimated in Project Star (also see Grissmer et al., 1998; Grissmer et al., 2000). The

authors are appropriately cautious, however, acknowledging that there were many other forces aside from class size that could also have contributed to changes in the test score gap.

Recently, California and Wisconsin have joined Tennessee as sites for research on the importance of class size. The California class size initiative adopted in 1996 allocates additional funds per student to classes of 20 or fewer students. Evaluators report: “A small positive gain in achievement is associated with being in a reduced class size, and this gain is realized by all groups of students, regardless of their demographic characteristics” (Bohrnstedt, 2001:2). They acknowledge, however, that it is difficult to know for sure that these gains are due to the changes in class size, as opposed to other reforms in the same classrooms.

Schools serving low-income, minority and English language learning students have had difficulty competing for additional teachers and have suffered what Bohrnstedt (2001) says is “a far greater decline in the qualifications of teachers than other schools.” They have also faced greater problems finding space for additional classrooms. Hence, they have been slower to implement the initiative and have received disproportionately less revenue for class size reduction (which accrues after class sizes are reduced). It is possible that disadvantaged students fortunate enough to be placed in smaller classes have benefited from this initiative, while those whose schools have been unable to implement it have suffered. To the degree that research shows that low-income and minority students benefit most from reduced class sizes, the California initiative seems poorly targeted. Other states should pay close attention to the California experience as they design their own initiatives, in order to avoid disadvantaging the students who most need the benefits of smaller class sizes.

In Wisconsin, the Student Achievement Guarantee in Education (SAGE) program is a five-year effort initiated by the Wisconsin Department of Public Instruction for the benefit of schools serving low-income students. It is being implemented in 80 schools across the state (Zahorik, 1999). The evaluation of the initiative involves 31 schools in 21 districts and researchers at the University of Wisconsin-Madision are conducting it. The study design is quasi-experimental, not random assignment. Specifically, the evaluators are comparing academic progress in schools in which class sizes are being reduced to 15 students per teacher in the early grades to similar schools that have regular class sizes. The black-white gap in baseline composite scores (for language arts, reading, and math) was roughly 0.75 standard deviation in both the SAGE and the comparison schools. (Other racial and ethnic groups represent such small percent-

ages of the students that their scores are not separately reported.) Average scores for blacks in the SAGE schools were roughly equal to those for blacks in the comparison schools at the time that the study began, and the same was true for whites.

The evidence from the SAGE project presented in Molnar et al. (2000) shows that most of the advantage of small classes was concentrated among black 1st graders. The effect size for the composite language arts, reading, and math score by the end of 1st grade was 0.58 for black students and 0.12 for white students, measured in baseline standard deviation units. By the end of 3rd grade, this had grown to 0.68 for blacks and it had not changed at all for whites.²⁸ The black-white gap in scores at the end of 3rd grade is smaller than at the baseline among students who attended small classes, but larger than at the baseline among students in regular-sized classes.

Our conclusion based on examining the data tables in Molnar et al. (2000) is the following: the main story in SAGE appears to be that black 1st graders in regular classes learn less than black or white children in small classes or white children in regular classes. This is the main source of differences by the end of 3rd grade. During the 2nd and 3rd grades, all groups learn similar amounts, but black children who were 1st graders in regular classes do not make up the deficit from having learned less in 1st grade. This in fact is similar to the results from Tennessee Project Star, in which most of the differential gain comparing small and larger classes was concentrated in the first year that a student spent in small classes, and the differential was roughly twice as large for blacks as for whites.

Similar to what we suggested above in the discussion of Tennessee’s Project Star, there is a possibility in Wisconcin’s SAGE that black children who spend 1st grade in small classes need to remain in small classes for 2nd and 3rd grades in order to retain the gains from the 1st-grade year. It is also possible, however, that they would have retained those gains even if they had moved into regular-sized classes. The only way to answer this question reliably is to randomly move some children from small to regular classes after the first year in small classes—which may be politically difficult to do.

In the end, the policy debate is not really among those who say that class size never matters for anyone and those who say it always matters for everyone. Most people, even Eric Hanushek, are in the middle. Hanushek (2000:42) writes, “Proposed class size reduction policies generally leave no room for localities to decide when and where reductions would be beneficial or detrimental. The existing evidence does not say that class size reductions are never worthwhile and that they should never be taken. It does say that uniform, across-the-board policies—such as those in the current policy debate—are unlikely to be effective.” This is a

valid criticism of the California class size reduction policy that, as discussed above, applies equally to all elementary schools in the state.

Should policy makers focus more on instructional quality and less on reducing class size? In our view, until more extensive and definitive research on class size gets done, class sizes larger than the low 20s are probably ill-advised, especially in elementary schools in which students are disruptive or need lots of individualized attention. Because we lack definitive evidence from other sources, our judgment is based primarily on what teachers themselves report. As mentioned earlier, in surveys over the years by the Educational Research Service, there is a consistent shift in teachers’ responses that class size is a minor problem to responses that it is a major problem at class sizes of 23 to 24 students (Robinson and Wittebods, 1986). (Recall that classrooms of 22 to 25 were the large ones in the Tennessee experiment.) However, based on both the Tennessee Star and the Wisconcin SAGE findings, class sizes in the neighborhood of 15 seem warranted for black kindergarten and 1st grade children in low-income schools (and probably for other disadvantaged children as well, though the research does not as clearly address them). Whether 2nd and 3rd grade children need classes as small as 15, and what the effects of class size might be in the later grades, have not been clearly established.

At the same time that class size should not be neglected, it would be wrongheaded not to make instructional quality the top priority. Researchers in project SAGE are studying the skills and practices that make some teachers more effective in small classrooms than others (Molnar et al., 2000).²⁹ There needs to be more research of this type to inform teacher training and professional development. There also needs to be greater effort to attract skilled people to the profession. Teacher quality matters. In most research that tries to study it, variation in teacher quality accounts for more of the variation in student achievement than any other schooling input, rivaling parental background in its importance. The evidence is clear. Studies that track the differential progress of students who have different teachers find very large teacher-to-teacher differences in effectiveness (Rivkin et al., 2000; Sanders and Horn, 1995). Teachers who score higher on standardized tests tend to produce students who score higher on achievement tests (Greenwald et al., 1996; Ferguson and Ladd, 1996), but schools attended by blacks and Hispanics often attract teachers with lower scores (Ferguson, 2000; Orfield et al., 1984). Concerns about teacher quality that began growing in the 1980s (as symbolized, for example, by the explosion during that decade in initial certification testing)³⁰ are entirely appropriate, indeed, necessary, if the society is serious about raising

standard levels of achievement among all groups of students. There are active debates under way about how to improve instruction. Some favor an emphasis on attracting more talent to the teaching profession (Ballou and Podgursky, 1999), while others work to refine teacher training and professional development (Darling-Hammond and Ball, 1998). We believe that both emphases are important to pursue and potentially quite consequential.

It seems clear that the nation’s future depends fundamentally on the degree to which schools and communities can raise skill levels among children from all racial, ethnic, and socioeconomic backgrounds. Achievement disparities among today’s students foreshadow socioeconomic disparities among tomorrow’s families. Large socioeconomic disparities among families are morally objectionable and politically dangerous for the future of a society.

As we stated in the introduction, the United States has achieved substantial progress in narrowing gaps among racial groups since Brown v. Board of Education in 1954. At the same time, there have been missed opportunities and the gaps remain large. Integration came too slowly and then produced fewer benefits than it should have. Head Start failed to produce as many lasting benefits as it would have if the schools to which graduates matriculated had more often been strong enough to sustain the gains. Title I has been a disappointment, but still has lots of potential. Class size effects are only now beginning to be understood, although they might have been many years ago if there had been more support for experimental research in education. Small classes on the order of 15 students per classroom seem to produce benefits for black children in kindergarten and 1st grade. We ought to be acting on this knowledge while we await better evidence for later grades. In addition, we need much more information about which types of teacher training and professional development regimes produce the best outcomes for the children that trainees end up teaching. The fact that such information has not been developed using high-quality research standards and then widely shared represents an extremely important set of missed opportunities. Examples of underachieved policy potential constitute a very long list, including examples in the current standards movement, which this paper has not addressed (see Ravitch, 2001).

However, the nation’s habit of missing opportunities to improve educational outcomes need not continue to such a degree. Recent assertions by political leaders regarding the need for more research-based practices and policy decisions are encouraging, especially if there is a focus on

improving classroom practices complemented by results-based accountability. If practitioners, political leaders, and researchers can make real the promise of current-day public discourse about the need to provide a high-quality education to all children, then the next 50 years may be more successful at finishing the long-term journey along which Brown v. Board of Education was a key milestone. We are guardedly optimistic and ready to work with others in getting it done. Other chapters in this volume describe some promising approaches to instructional improvement and thereby offer reasons to be hopeful.

Argys, L.M., D.I. Rees, and D. Brewer 1996 Detracking America’s Schools: Equity at Zero Cost? Working paper series #9501, Center for Research on Economic and Social Policy, University of Colorado, Denver.

Ballou, D., and M. Podgursky 1999 Teacher training and licensure: A layman’s guide. [Online]. In Better Teachers, Better Schools, M. Kanstoroom and C.E. Finn, Jr., eds. Washington, DC: Thomas B. Fordham Foundation. Available: http://www.edexcellence.net/better/tchrs/teachers.html. [Accessed November 16, 2001].

Barnett, W.S. 1992 Benefits of compensatory preschool education. The Journal of Human Resources 27(2):279-312.

Bohrnstedt, G. 2001 Class Size Reduction in California: Early Findings Signal Promise and Concerns. [Online]. Available: http://www.air.org/projects/csr.htm. [Accessed November 16, 2001].

Boozer, M.A., A. Krueger, and S. Wolkon 1992 Race and School Quality Since Brown v. Board of Education. Brookings Papers on Economic Activity, Microeconomics. Washington, DC: Brookings Institution Press.

Boozer, M., and C. Rouse 1995 Intraschool Variation in Class Size: Patterns and Implications. Working Paper No. 5144. National Bureau of Economic Research, Cambridge, MA.

Borman, G.D., and J.V. D’Agostino 2001 Title I and student achievement: A quantitative synthesis. In Title I: Compensatory Education at the Crossroads, G.D. Borman, S.C. Stringfield, and R.E. Slavin, eds. Mahwah, NJ: Lawrence Erlbaum Associates.

Borman, G.D., K.K. Wong, L.V. Hedges, and J.V. D’Agostino 2001 Coordinating categorical and regular programs: Effects on Title I students’ educational opportunities and outcomes. In Title I: Compensatory Education at the Crossroads, G.D. Borman, S.C. Stringfield, and R.E. Slavin, eds. Mahwah, NJ: Lawrence Erlbaum Associates.

Boyd-Zaharias, J., et al. 1994 Quality Schools Build on a Quality Start. Tennessee State University, Center of Excellence for Research in Basic Skills.

Braddock, J., and R.E. Slavin 1993 Why ability grouping must end: Achieving excellence and equity in American education. Journal of Intergroup Relations 20(2):51-64.

Bush, G.W. 2001 No Child Left Behind. [Online]. Available: http://www.ed.gov/inits/nclb/index.html. [Accessed November 16, 2001].

Campbell, D.T., and P.W. Frey 1970 The implications of learning theory for the fade-out of gains from compensatory education. Pp. 185-223 in Compensatory Education: A National Debate, Vol. 3, J. Hellmuth, ed. New York: Brunner/Mazel.

Carter, L.F. 1983 A Study of Compensatory and Elementary Education: The Sustaining Effects Study. Washington, DC: U.S. Department of Education, Office of Program Evaluation.

Coleman, J., et al. 1966 Equality of Educational Opportunity, Washington, DC: U.S. Office of Education.

Cook, T. 1984 School Desegregation and Black Achievement. Washington, DC: U.S Department of Education.

Crain, R.L., and R.E. Mahard 1978 Desegregation and black achievement: A review of the research. Law and Contemporary Problems 42(3):17-56.

1983 The effect of research methodology on desegregation-achievement studies: a meta-analysis. American Journal of Sociology 88(5):839-854.

Currie, J., and D. Thomas 1995 Does Subsequent School Quality Affect the Long-Term Gains from Head Start? Manuscript. University of California, Los Angeles.

Cutright, P. 1972 Achievement, Mobility, and the Draft: Their Impact on the Earnings of Men. DHEW (SSA) 73-11854. Washington, DC: U.S. Department of Health, Education, and Welfare.

1974 Academic achievement, schooling, and the earnings of white and black men. Journal of Social and Behavioral Sciences 20(3):1-18.

Darling-Hammond, L., and D. Loewenberg Ball 1998 Teaching for High Standards: What Policymakers Need to Know and Be Able to Do. National Commission on Teaching & America’s Future and the Consortium for Policy Research in Education. [Online] Available: http://www.cpre.org/Publications/jre04.pdf [Accessed November 16, 2001].

Dreeben, R., and A. Gamoran 1986 Race, instructions, and learning. American Sociological Review 51:660-669.

Evans, W.N., S. Murray, and R. Schwab 1997 Schoolhouses, courthouses, and statehouses after Serrano. Journal of Policy Analysis and Management 16(1):10-31.

1999 The impact of court-mandated school finance reform. Pp. 72-98 in National Research Council, Equity and Adequacy in Educational Finance: Issues and Perspectives, H.F. Ladd, R. Chalk, and J.S. Hanson, eds. Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.

Farkas, G., and L.S. Hall 2000 Can Title I attain its goal? In Brookings Papers on Education Policy, 2000, D. Ravitch, ed. Washington, DC: Brookings Institution Press.

Ferguson, R.F. 1991 Paying for public education: New evidence on how and why money matters. Harvard Journal on Legislation 28(2):465-498.

1995 Shifting challenges: 50 years of economic change toward black-white earnings equality. Daedalus 124.

1998a Teachers’ perceptions and expectations and the black-white test score gap. In The Black-White Test Score Gap, C. Jencks and M. Phillips, eds. Washington, DC: Brookings Institution Press.

1998b Can schools narrow the black-white test score gap? In The Black-White Test Score Gap, C. Jencks and M. Phillips, eds. Washington, DC: Brookings Institution Press.

2000 Certification test scores, teacher quality and student achievement. In Analytic Issues in the Assessment of Student Achievement, D.W. Grissmer and J.M. Ross, eds. Washington, DC: National Center for Education Statistics.

2001a A diagnostic analysis of black-white GPA disparities in Shaker Heights, Ohio. In Brookings Papers on Education Policy, 2001, D. Ravitch, ed. Washington, DC: Brookings Institution Press.

2001b Test-score trends along racial lines 1971 to 1996: Popular culture and community academic standards. Pp. 348-390 in National Research Council, America Becoming: Racial Trends and Their Consequences, N. Smelser, W.J. Wilson, and F. Mitchell, eds. Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.

Ferguson, R.F., and H.F. Ladd 1996 How and why money matters: An analysis of Alabama schools. In H.F. Ladd, ed., Holding Schools Accountable Performance Based Reform in Education, Washington, DC: Brookings Institution Press.

Finley, M. K. 1984 Teachers and tracking in a comprehensive high school. Sociology of Education 57:233-243.

Finn, J.D., and C.M. Achilles 1990 Answers and questions about class size: A statewide experiment. American Educational Research Journal 27(3):557-577.

Fix, M., and R. Struyk 1991 Clear and Convincing Evidence, Measurement of Discrimination in America. Washington, DC: The Urban Institute.

Fordham, S. 1996 Blacked Out: Dilemmas of Race, Success and Identity at Capital High. Chicago: University of Chicago Press.

Gamoran, A., and M. Berends 1987 The effects of stratification in secondary schools: Synthesis of survey and ethnographic research. Review of Educational Research 57(4):415-435.

Gamoran, A., and R. G. Mare 1989 Secondary school tracking and educational inequality: Compensation, reinforcement, or neutrality? American Journal of Sociology 94:1146-1183.

Garet, M.S., and B. Delaney 1988 Students, courses and stratification. Sociology of Education 61:61-77.

Glass, G.V., and M.E. Smith 1979 Meta-analysis of research on class size and achievement. Educational Evaluation and Policy Analysis 1:2-16.

Good, T.L. 1987 Two decades of research on teacher expectations: Findings and future directions. Journal of Teacher Education 38(4):32-47.

Greenwald, R., L.V. Hedges, and R.D. Laine 1996 The effect of school resources on student achievement. Review of Educational Research 66:361-96.

Grissmer, D. 1999 Class size effects: Assessing the evidence, its policy implications, and future research agenda. Educational Evaluation and Policy Analysis 21(2):231-248.

Grissmer, D., A. Flanagan, and S. Williamson 1998 Why did the black-white score gap narrow in the 1970s and 1980s? In The Black-White Test Score Gap, C. Jencks and M. Phillips, eds. Washington, DC: Brookings Institution Press.

Grissmer, D., A. Flanagan, J. Kawata, and S. Williamson 2000 Improving Student Achievement: What State NAEP Test Scores Tell Us. Santa Monica, CA: RAND Corporation.

Haller, E.J. 1985 Pupil race and elementary school ability grouping: Are teachers biased against black children? American Educational Research Journal 22(4):465-483.

Hanushek, E.A. 1986 The economics of schooling: Production efficiency in public schools. Journal of Economic Literature 24:1141-1177.

1997 Assessing the effects of school resources on student performance: An update. Educational Evaluation and Policy Analysis 19(2):141-164.

2000 Evidence, Politics, and the Class Size Debate (Response to Alan Krueger). Working Paper No. 121. October. The Economic Policy Institute, Washington, DC.

Hedges, L.V., and R. Greenwald 1996 Have times changed? The relation between school resources and student performance. Pp. 74-92 in Does Money Matter? The Effect of School Resources on Student Achievement and Adult Success, Gary Burtless, ed. Washington, DC: Brookings Institution Press.

Hedges, L.V., R.D. Laine, and R. Greenwald 1994a Does money matter? A meta-analysis of studies of the effects of differential inputs on student outcomes. Educational Researcher 23(4):5-14.

1994b Money does matter somewhere: A reply to Hanushek. Educational Researcher 23(4):9-10.

Jencks, C., et al. 1972 Inequality: A Reassessment of the Effect of Family and Schooling in America. New York: Basic Books.

Jencks, C., and S.E. Mayer 1990 The social consequences of growing up in a poor neighborhood. Pp. 111-186 in National Research Council, Inner-City Poverty in the United States, L. Lynn and M. McGeary, eds. Committee on National Urban Policy, Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.

Johnson, W.R., and D. Neal 1998 Basic skills and the black-white earnings gap. In The Black-White Test Score Gap, C. Jencks and M. Phillips, eds. Washington, DC: Brookings Institution Press.

Kahlenberg, R.D. 2001 All Together Now: Creating Middle-Class Schools Through Public School Choice. Washington, DC: Brookings Institution Press.

Krueger, A.B. 1999 Experimental estimates of educational production functions. Quarterly Journal of Economics 114(2):497-532.

2000 Understanding the Magnitude and Effect of Class Size on Student Achievement. Working Paper No. 121. October. The Economic Policy Institute, Washington, DC.

Krueger, A.B., and D. Whitmore 2001 Would Smaller Classes Help Close the Black-White Achievement Gap? Paper presented at the Brookings Institution Conference, Closing the Gap: Promising Approaches to Reducing the Achievement Gap. January.

Kulik, J.A. 1992 An Analysis of the Research on Ability Grouping: Historical and Contemporary Perspectives. A publication of the National Center on the Gifted and Talented, University of Connecticut.

Ladson-Billings, G. 1994 DreamKeepers: Successful Teachers of African-American Children. San Francisco: Jossey-Bass Publishers.

Lazear, E.P. 1999 Education Production. Working Paper No. 7349. National Bureau of Economic Research, Cambridge, MA.

Lee, V.E., and A.S. Bryk 1988 Curriculum tracking as mediating the social distribution of high school achievement. Sociology of Education 61: 78-94.

Lee, V.E., and S. Loeb 1995 Where do Head Start attendees end up? One reason why Head Start effects fade out. Educational Evaluation and Policy Analysis 17(1):62-82.

Loveless, T. 1998 The Tracking and Ability Grouping Debate. [Online]. Available: http://www.edexcellence.net/library/track.html. [Accessed November 16, 2001]. Thomas B. Fordham Foundation.

Ludwig, J., H.F. Ladd, and G.J. Duncan 2001 Urban poverty and educational outcomes. In Brookings-Wharton Papers on Urban Affairs 2001, W.G. Gale and J.R. Pack, eds. Washington, DC: Brookings Institution Press.

Martin, W.E., ed. 1998 Brown v. Board of Education: A Brief History With Documents. Boston: Bedford/St. Martin’s.

Mayer, S.E. 1991 How much does a high school’s racial and socioeconomic mix affect graduation and teenage fertility rates? In The Urban Underclass, C. Jencks and P.E. Peterson, eds. Washington, DC: Brookings Institution Press.

Metz, M.H. 1978 Classrooms and Corridors: The Crisis of Authority in Desegregated Secondary Schools. Berkeley: University of California Press.

Molnar, A., P. Smith, and J. Zahorik 2000 1999-2000 Evaluation Results of the Student Achievement Guarantee in Education (SAGE) Program. [Online]. Available: http://www.uwm.edu/Dept/CERAI/sage.html. [Accessed November 19, 2001].

Mosteller, F. 1995 The Tennessee study of class size in the early school grades. The Future of Children 5(2):113-127.

Murnane, R.B., J.B. Willett, and F. Levy 1995 The growing importance of cognitive skills in wage determination. Review of Economics and Statistics 77(2):251-266.

Nye, B.A., et al. 1993 Tennessee’s bold experiment: Using research to inform policy and practice. Tennessee Education 22(3):10-17.

1994a An Initial Evaluation of the Tennessee Department of Education “At-Risk” Student/Teacher Ratio Reduction Project in Sixteen Counties, 1990 Through 1993. Tennessee State University, Center of Excellence for Research in Basic Skills.

1994b The Lasting Benefits Study Seventh Grade: Executive Summary. Tennessee State University, Center of Excellence for Research in Basic Skills.

Nye, B., and L.V. Hedges 2000 The effects of small classes on academic achievement: The results of the Tennessee class size experiment. American Educational Research Journal 37:123-151.

Oakes, J. 1985 Keeping Track: How Schools Structure Inequality. New Haven: Yale University Press.

O’Day, J.A., and M.S. Smith 1993 Systemic reform and educational opportunity. In Designing Coherent Education Policy: Improving the System., S. Fuhrman, ed. San Francisco: Jossey-Bass Publishers.

Oden, S., L.J. Schweinhart, and D.P. Weikart 2000 Into Adulthood: A Study of the Effects of Head Start. Ypsilanti, MI: High/Scope Press.

Office of Technology Assessment 1992 Testing in American Schools. Washington, DC: U.S. Government Printing Office.

Orfield, G., and S.E. Eaton 1996 Dismantling Desegregation: The Quiet Reversal of Brown v. Board of Education. New York: The New Press.

Orfield, G., and J.T. Yun 1999 Resegregation in American Schools. The Civil Rights Project, Harvard University.

Orfield, G., H. Mitzel, T. Austin, R. Bentley, D. Bice, M. Dwyer, L. Gidlow, J. Herschensohn, B. Hibino, T. Kelly, A. Kuhns, M. Lee, C. Rabinowitz, J. Spoerl, A. Vosnos, and J. Wolf 1984 The Chicago Study of Access and Choice in Higher Education: A Report to the Illinois Senate Committee on Higher Education. University of Chicago, Committee on Public Policy Studies Research Project. Available on ERIC.

Pallas, A.M., D.R. Entwisle, K.L. Alexander, and M.F. Stluka 1994 Ability-group effects: Instructional, social, or institutional? Sociology of Education 67:27-46.

Phillips, M. 2001 Do African-American and Latino Children Learn More in Preedominantly White Schools? Unpublished paper, School of Public Policy and Social Research, UCLA.

Phillips, M., J. Brooks-Gunn, G.J. Duncan, P. Klebanov, and J. Crane 1998 Family background, parenting practices, and the black-white test score gap. In The Black-White Test Score Gap, C. Jencks and M. Phillips, eds. Washington, DC: Brookings Institution Press.

Puma, M.J., et al. 1997 Prospects: Final Report on Student Outcomes. Washington, DC: U.S. Department of Education, Planning and Evaluation Service, Office of the Undersecretary.

Ravitch, D., ed. 2001 Brookings Papers on Education Policy, 2001. Washington, DC: Brookings Institution Press.

Rivkin, S.G., E.A. Hanushek, and J.F. Kain 2000 Teachers, Schools, and Academic Achievement. Working Paper No. 6691 (revised). National Bureau of Economic Research, Cambridge, MA.

Robinson, G.E., and J.H. Wittebods 1986 Class Size Research: A Related Cluster Analysis for Decision Making. Research Brief. Arlington, VA: Educational Research Service.

Rosenbaum, J.E.

1995 Changing the geography of opportunity by expanding residential choice: Lessons from the Gautreaux Program. Housing Policy Debate 6:231-269.

Rosenbaum, J.E., N. Fishman, A. Brett, and P. Meaden 1993 Can the Kerner Commission’s housing strategy improve employment, education, and social integration for low-income blacks? North Carolina Law Review 71:1519-1556.

Ryan, J.E. 1999 Schools, race and money. Yale Law Journal 109:249-316.

Sanders, W.L., and S.P. Horn 1995 The Tennessee Value-Added Assessment System (TVAA): Mixed model methodology in educational assessment. In Teacher Evaluation: Guide to Effective Practice, A.J. Shinkfield and D.L. Stufflebeam, eds. Boston: Kluwer Academic Publishers.

Schofield, J.W. 1995 Review of research on school desegregation’s impact on elementary and secondary school students. Pp. 597-616 in Handbook of Research on Multicultural Education, J.A. Banks and C.A. McGee Banks, eds. New York: Macmillan.

Schwartz, F. 1981 Supporting or subverting learning: Peer group patterns in four tracked schools. Anthropology and Education Quarterly 12:99-121.

Slavin, R.E. 2001 How Title I Can Become the Engine of Reform in America’s Schools. In Title I: Compensatory Education at the Crossroads, G.D. Borman, S.C. Stringfield and R.E. Slavin, eds. Mahwah, NJ: Lawrence Erlbaum Associates.

Sorensen, A.B., and M. Hallinan. 1984 Effects of race on assignment to ability groups. In The Social Context of Instruction, P. L. Peterson, L.C. Wilkinson, and M. Hallinan, eds. New York: Academic Press.

U.S. Department of Education 1996 Digest of Education Statistics. Washington, DC: U.S. Department of Education.

2000 Digest of Education Statistics. Washington, DC: U.S. Department of Education.

Wells, A.S., and R.L. Crain 1997 Stepping Over the Color Line: African American Students in White Suburban Schools. New Haven: Yale University Press.

Word, E., et al. 1990 Student/Teacher Achievement Ratio (STAR) Tennessee’s K-3 Class Size Study, Final Summary Report, 1985-1990. Nashville, TN: Tennessee State Department of Education.

Zahorik, J.A. 1999 Reducing class size leads to individualized instruction. Educational Leadership 57(1):50-53.