Subscribe Today
Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13

Disentangling School- and Student-Level Effects of Desegregation and Resegregation on the Dropout Problem in Urban High Schools: Evidence From the Cleveland Municipal School District, 1977–1998

by Argun Saatcioglu - 2010

Background/Context: Past studies have consistently found modest academic gains for minorities as a result of desegregation. In addition, school effects have tended to be small or even null once student-level nonschool factors are controlled. However, traditional approaches not only treat desegregation as a policy that may be sufficient by itself to improve student performance, but also involve analytical techniques that may mask the beneficial effects of desegregated schools. In reality, student performance is affected by both school and nonschool factors, and the latter is often more influential than the former. Therefore, there is a need to reframe the approach to evaluating desegregation’s academic outcomes. Conceptually, modest gains need not be viewed as a sign of either success or failure. Instead, the outcomes can be judged in light of the inherent limitations characteristic of any reform intended to close the achievement gap, limitations associated with enduring nonschool problems that undermine student performance. Empirically, multilevel analytical procedures can disentangle school- and student-level effects of desegregation to help determine whether the policy can improve the schools, despite limited gains in eventual student performance.

Purpose: This article illustrates the proposed conceptual and empirical approach to desegregation evaluation by focusing on the dropout problem in urban high schools. The principal objective is to estimate the difference that desegregation makes in the effects of high schools on the dropout tendencies of predominantly poor minorities, net of student-level effects, many of which originate from outside the schools.

Research Design: The analysis draws on data from the Cleveland Municipal School District, 1977–1998. Specifically, data on four cohorts were available. The first cohort attended segregated schools until late in high school. The second one gradually desegregated in middle school and attended fully integrated high schools. The third one attended integrated schools from 1st through 12th grade. The fourth cohort attended integrated elementary and middle schools, followed by gradual resegregation in high school. Thus, the analysis estimates the effects of segregated, desegregated, and resegregated high schools while controlling for different degrees of exposure to desegregation prior to high school.

Findings: Minority (Black and Hispanic) dropout rates changed slightly, and only for the second cohort. Student-level nonschool problems, such as poverty, family disruption, and neighborhood disadvantage, had worsening effects over time, which likely countered some of desegregation’s benefits. Yet, desegregation made a considerable difference in the way that high schools aggravated the dropout problem. Much of the difference was explained by key compositional changes such as reductions in minority, poverty, and nontraditional family concentration in the schools for minorities. Resegregation reversed those benefits. The results provide no evidence of White harm. Instead, Whites appear to have benefited from desegregated schools in ways similar to how minorities benefited, although to a lesser extent.

Conclusions: It is fairer to evaluate desegregation in light of its inherent limits. The policy may benefit students in terms of school effects but still fail to reverse eventual performance problems such as dropouts, which are subject to many forces that the schools can do little about. The results suggest that in the absence of equitable “educational” policies, such as desegregation, unequal schooling conditions and outcomes for urban minorities may be further exacerbated.

Reviews of desegregation outcomes have consistently indicated that the academic gains for minority students have been modest at best (Braddock & Eitle, 2004; T. Cook et al., 1984; Crain & Mahard, 1978; Schofield, 2001). Critics of the policy have often drawn on this point to describe desegregation as a futile effort (e.g., Armor, 1995, 2002; Glazer, 1975) and to argue that its social and economic costs have far outweighed its educational benefits (Caldas & Bankston, 2005; Rossell, 1990; Rossell, Armor, & Wallberg, 2002). However, virtually all evaluations of school desegregation have focused on changes in performance at the student level, paying little attention to effects at the school level. Studies have ignored whether desegregation was able to enhance the schools contribution to minority success despite the limited changes in student performance, which is subject to influence by both school and nonschool factors.

Taking a multilevel analytical approach, this article disentangles student- and school-level effects. Because one of the central objectives of desegregation was to close the racial achievement gap, its effectiveness has naturally been considered in terms of the extent to which the performance of minorities approached that of White studentsa norm that is customarily applied to many educational reforms. Yet, as Konstantopoulos and Hedges (2008) noted, judging the effects of reform in terms of individual variation and the achievement gaps between relevant groups may not only be disappointing, but also misleading (p. 1614). This is because although reform may substantially improve school-level effects for disadvantaged students, the eventual performance of the students may continually be undermined by various difficulties outside the schools (Jencks & Meyer, 1990; Raudenbush, 2004), such as neighborhood and family problems (Coleman et al., 1966; Conley & Albright, 2004; Jencks et al., 1972; Kozol, 1988, 1991; Rothstein, 2004), pessimistic views of postsecondary education or occupational prospects (Gates, 2004; Stewart, Stewart, & Simons, 2007), and oppositional norms discouraging high performance (despite the intrinsic value the students may place on success; P. Cook & Ludwig, 1998; Mickelson, 1990; Ogbu, 2003).

Although such nonschool dynamics are often considered beyond the scope of education policy, the potential discrepancy between student- and school-level outcomes of desegregation need not constitute grounds for a total abandonment of the goal of integrated schools because the absence of that policy may exacerbate the distribution of students across the schools by class and race and foster further inequities in school quality (Cashin, 2004; Gold, 2007; Lippman, Burns, & McArthur, 1996; Orfield & Ashkinaze, 1991). As a result, the schools serving disadvantaged minorities may aggravate rather than counteract educational inequality because these schools take on a higher risk of lacking the compositional attributes and the resources and organizational features (educational and administrative practices) necessary for adequate contribution to student performance (Orfield & Eaton, 1996; Kozol, 2004).1

The racial achievement gap, which had steadily narrowed throughout the 1970s and 1980s, has been widening since the early 1990s, when the national retreat from desegregation accelerated and racial isolation in the schools began to intensify (National Center for Education Statistics, 2000, 2005). Although it is difficult to determine the exact size of desegregations effect in narrowing the gap in the 1970s and 1980sbecause much else was changing at the time (Grissmer, Flanagan, & Williamson, 1998; Lee, 2002)a critical question at this point is whether resegregation is making things worse. If segregated schools are helping perpetuate the gap, they pose a challenge greater than currently realized by those who advocate improving minority success through separate but equal schooling.

Therefore, in addressing the issue of desegregation, this article also deals with the consequences of resegregation. More precisely, it examines both student- and school-level effects on performance under conditions of segregated, desegregated, and resegregated education. The focus of the analysis is on the dropout problem in urban high schools that serve predominantly poor Black and Hispanic students. Although much of the debate on desegregation outcomes and the broader concern with the achievement gap have revolved around test scores, low graduation rates among poor minorities have become a particularly pressing problem in recent decades (Belfield & Levin, 2007; Orfield, 2004). Nationwide, 78% of White students graduate from high school, whereas only 54% of Blacks and 57% of Hispanics do (Greene, 2002). In urban districts, the average minority dropout rate is 60%, reaching well over 70% in some of the most disadvantaged and racially isolated school systems (Greene). Dropping out from high school is the ultimate failure for a student in the post-industrial economya failure that usually causes deep and irreversible life-long damage to the student and his family (Orfield & Lee, 2005, p. 37).

Although multilevel evaluations have been mostly absent from desegregation research until recently, the simultaneous examination of student- and school-level effectsfor instance, by means of hierarchical growth models or value-added assessmentis increasingly common in efforts to hold the schools accountable for their influence on student performance as part of evaluating many contemporary reforms (D. B. Downey, von Hippel, & Hughes, 2008; Konstantopoulos & Hedges, 2008). This article retrogressively applies a similar view to school desegregation: Net of student-level effects that originate largely from outside the schools, if integrated urban high schools were to be held accountable for their contribution to the dropout tendencies of minorities, would they fare better or worse than segregated ones?

A judicious evaluation of desegregation should also address the situation of White students. Most reviews have indicated that integrated schools have had no negative impact on White performance, except when a poorly designed or irresponsibly implemented plan placed a few Whites in a majority-non-White school with a high concentration of poverty (Hochschild & Skovronick, 2003).2 Even then, White harm remained limited. However, much of the evidence regarding Whites has come from studies that do not differentiate between student- and school-level effects on performance. Because this study takes a multilevel approach, it asks: Net of student-level effects, were school-level effects on the dropout tendencies of Whites in urban districts better, worse, or the same in desegregated conditions when compared with segregated conditions? Thus, although the basic argument and much of the literature review in the article pertain to minorities, the key argument is tested for Whites as well.

The analysis relies on a unique longitudinal data set from the Cleveland Municipal School District (CMSD), which implemented a comprehensive citywide desegregation policy in its elementary, middle, and high schools between 1979 and 1992. The district resegregated between 1993 and 1998. A rich array of yearly repeated measures was available on four high school cohorts, each consisting of minority and White students (a total of 27,550 students). The cohorts were 5 years apart and consisted of students who started high school (ninth grade) in 1977, 1982, 1987, and 1992. Each cohort was observed for 6 years. Given CMSDs desegregation timeline, the first cohort experienced minimum desegregation in high school, whereas the second and third cohorts attended fully desegregated high schools. The fourth experienced gradual resegregation. Because the changes in the district were not limited to the high schools, the cohorts also varied in terms of exposure to desegregation in elementary and middle schools: The first cohort experienced no desegregation before high school; the second one experienced only a limited degree during middle school; and the third and fourth cohorts were fully desegregated from first through eighth grade. As such, although the analysis concentrates on dropout tendencies in segregated, desegregated, and resegregated high schools, it also accounts for the degree of desegregation exposure prior to high school.


Rigorous evaluations of desegregation focusing on the dropout problem have been rare. An early study by Felice and Richardson (1976) revealed inconclusive findings on the changes in minority dropout rates in integrated schools, particularly when socioeconomic status was taken into account. In Orfield and Ashkinazes (1991) study of schools in the greater Atlanta region, desegregation was found to reduce dropout rates for Blacks, although the analysis did not involve a multivariate design. Using nationwide data and superior modeling strategies, Guryan (2004) determined that desegregation reduced high school dropout rates of Blacks by no more than 23 percentage points. More recently, Reber (2007) complemented Guryans work by analyzing district-level data from Louisiana. Although her results were similar to those of Guryan (3.3% reduction in minority dropout rates), she found that much of the change had to do with increased funding as part of the integration effort rather than with interracial contact.

These findings are consistent with the results of studies on desegregations impact on other academic outcomes, such as grades or test scores. Most studies in the 1970s and 1980s not only revealed limited gains in minority performance but also showed that the effects of school composition and other desegregation-related school factors reduced significantly once the socioeconomic and other student-level nonschool factors were accounted for (see Armor, 2002; T. Cook et al., 1984; and Jencks & Meyer, 1990, for thorough reviews). This has partly to do with the normative focus on student-level performance, along with the heavy reliance on single-level methods of evaluation, which produced regression models that inappropriately mixed student- and school-level predictors of performance. Such models fail to properly disentangle the effects of predictors at different levels. Despite some notable exceptions (e.g., Rumberger & Palardy, 2005), similar conceptual and analytical strategies have been common in more recent studies as well (e.g., Angrist & Lang, 2004; Hanushek, Kain, & Rivkin, 2008; Hanushek & Raymond, 2005; Hoxby, 2000)though some of these studies have found relatively more encouraging results regarding desegregation effects, at least from the perspective of pro-desegregation scholars and policy makers.

School desegregation involves, as do most other reforms, altering the conditions, processes, and resources only within the schools. It does not control the influence of the nonschool context, where the average student spends more than two thirds of his or her time (Hofferth & Sandberg, 2001; Walberg, 1984). Therefore, there is a need to examine school effects on performance as distinct from student-level effects that originate largely from outside the schools. Essentially, the success of any given reform in terms of eventual academic outcomes is inevitably contingent upon two issues: (1) the share of variance in student performance attributable to the schools, and (2) the relative strength of the countervailing forces in the nonschool context. If the share of the school effect is smaller than that of the nonschool context (which it usually is), and if the nonschool forces that undermine student performance are stronger than the school effect, then academic outcomes may not change meaningfully despite potential gains in the extent to which the schools contribute to success. To use a rough analogy, advances in treatment techniques may be significant in the fight against lung cancer, but they may be insufficient to substantially reduce the rate of the disease unless problems outside the medical field change, such as smoking habits, air and water pollution, and limited access to quality health care. Evaluating desegregation from this standpoint is particularly relevant with regard to urban schools, where the prospects for educational achievement of many students are undermined by severe nonschool problems.


According to Swanson (2004), urban high schools that concentrate poor minorities have become dropout factories. Although there may be nothing inferior about racial concentration in and of itself, it is an inherent problem when it entails poverty concentration and perpetual isolation from the mainstream (Orfield & Ashkinaze, 1991). For a variety of interrelated reasons, desegregation is likely to be beneficial in terms of what urban schools can do for the dropout problem. At the very least, the policy can curb the ways in which the schools aggravate the problem of dropouts.

First, desegregation breaks the racial and poverty concentration in the schools. Racially mixed education, which has been pivotal in reducing prejudice (Sonleitner & Wood, 1996), is an important stimulant of positive school climate when managed well, resulting in higher levels of cooperation and stronger norms of performance (McClendon, 1974; Pettigrew & Tropp, 2006). In particular, poor students interaction with affluent and near-affluent peers can provide the motivation to work harder (Betts, Zau, & Rice, 2003; Coleman et al., 1966; Wells & Crain, 1997), which is likely to increase the odds of graduation. Such beneficial peer effects are particularly important for poor minority adolescents (Kasen, Cohen, & Brook, 1998). Although the racial contact argument has fallen into disusein part because it has been employed in a purely symbolic and disembodied fashion to suggest that minorities will learn better if they sit next to Whites (for a review, see Braddock & Eitle, 2004)recent research by cognitive psychologists has shown that integration helps create environments for all students to break out of their traditional patterns of thinking and to adopt more complex and critical thinking styles (Antonio et al., 2004; Terenzini, Cabrera, Colbeck, Bjorklund, & Parente, 2001). This, in turn, broadens students expectations and potentially reduces the likelihood of dropping out (Rumberger, 1987, 2004; Wigfield & Eccles, 2002), though the extent of eventual behavioral changes may depend considerably on a variety of factors outside the schools (Duncan, Biosjoly, Levy, Kremer, & Eccles, 2003; Gurin, Dey, Hurtado, & Gurin, 2002).

The second important school-level benefit of desegregation has to do with possible improvements in the schools social capital. Low levels of parental and community involvement in racially isolated urban high schools is a significant problem (Nettles, 1991; Sheldon, 2003; Sheldon & Epstein, 2002). Because of the high frequency of nontraditional (single-parent) family structure among poor minorities (Besharov & West, 2002); low levels of education, occupational skills, resources, and time among urban parents (D. M. Downey & Coyne, 1990; Garfinkel & McLanahan, 1986; Lareau, 2003; McLoyd, Jayaratne, Ceballo, & Borquez, 1994); and limited collective efficacy in disadvantaged neighborhoods (Morenoff, Sampson, & Raudenbush, 2001), the schools often lack social capital in the form of parental input and broader community engagement that more affluent high schools tend to enjoy (Henig, Hula, Orr, & Pedescleaux, 1999; Pong, 1998). Desegregation potentially mitigates the problem by redistributing parental and community support for the schools such that residents in more resourceful neighborhoods remain committed not only to the quality of their local schools but also to that of other schools in socially disadvantaged areas (Orfield & Eaton, 1996; Orfield et al., 1997). Though rarely addressed and no longer a central concern, securing the social and political commitment of predominantly affluent White communities to the quality of all schools was an important motive behind many desegregation efforts (D. Bell, 2004; Kluger, 1976; Mondale & Patton, 2001).

Third, desegregation also implies considerable improvements in tangible resources. Racially concentrated urban high schools typically lack adequate funds (Baker & Green, 2005; Gold, 2007). They also suffer from overcrowding (Kozol, 2004), poor facilities, low teacher and staff qualifications (Clotfelter, Ladd, Vigdor, 2005; Lankford, Loeb, & Wyckoff, 2002), and inferior supplies and school services (Phillips & Chin, 2004). All these school-level problems have been found to increase students risk of dropping out (Bryk & Thum, 1989; Rumberger & Palardy, 2004; Rumberger & Thomas, 2000). Whether the implementation is limited to a single district or involves a broader regional focus, an inherent part of any desegregation program is to bring the funding for previously isolated minority schools up to the same levels that prevail at majority-White schools (Hochschild, 1984; Hochschild & Skovronick, 2003; Irons, 2002; Orfield, 1978; Patterson, 2001). As such, the policy potentially reverses many of the resource-related school problems that aggravate dropout tendencies among disadvantaged minorities. It wasnt that we wanted our children to go to school with white children, Vivian Scales, one of the original plaintiffs in Brown, had argued, That was not the gist of it all. . . .  We wanted our children to have a better and equal education, which we knew that we were not getting (quoted in Mondale & Patton, 2001, p. 137).

Fourth, desegregated high schools for minorities are likely to be characterized by organizational features (educational and administrative practices) that are superior to those at segregated schools in terms of confronting the dropout problem. Curricular and pedagogical improvements are fundamental aspects of this process (Hochschild, 1984; Irons, 2002; Wells, 2002). In addition, desegregated schools maintain fairer disciplinary procedures and higher emphasis on classroom reasoning (Raudenbush, Fotiu, & Cheong, 1998). They also implement tracking policies in a less discriminatory fashion (Lucas, 1999; Southworth & Mickelson, 2007). Additionally, such schools are less likely to allow push out exams, social retention, and graduation-discouraging advice from teachers and staff (Natriello, 1995). By contrast, racially isolated schools are often contexts of negative student-teacher relations (Croninger & Lee, 2001), and they lack activities to bolster student engagement and expectations (Finn, 1989). In a recent empirical examination of school-level effects, Rumberger and Palardy (2005) found that teacher expectations, the amount of homework, and the number of rigorous classes that students took were among the key mechanisms by which racial and poverty concentration in the schools undermined student performance (see also D. Mayer, Mullens, & Moore, 2000). Finally, high schools serving disadvantaged minorities are more likely to adopt and vigorously implement explicit dropout prevention programs when they are under pressure to improve graduation rates as part of desegregation orders (Erickson & Simon, 1998).

The fifth and final basis for the favorable effects of urban high schools on minority dropout rates under a desegregation plan has to do with the positive role that integrated elementary and middle schools play. Though much of the dropout problem occurs at the high school level, dropping out is a problem that culminates from an early age because of several detrimental experiences in elementary and middle grades (Alexander, Entwisle, & Horsey, 1997). Therefore, if desegregation benefits the students prior to high school, it is likely to facilitate high schools task of curbing dropout tendencies. Significant indirect evidence supports this view, from desegregation studies that address student-level outcomes. Jencks et al. (1972), for instance, determined that although the academic benefits of desegregation are relatively small for minority students in the short run (23 percentage points), they can accumulate to a considerable degree (up to 15 percentage points) if students attend integrated schools through all grades (see also T. Cook et al., 1984; Crain & Mahard, 1978; St. John, 1975). Reviewing various desegregation studies, S. Mayer and Jencks (1989) also predicted that more than 10 years in a predominantly White school can make a meaningful difference in the success of poor minorities. More recently, Mickelson (2001) found that the extent of time minority students spent in segregated elementary grades negatively affected their test scores and track placement in high school. Aside from desegregation research, dropout studies have consistently found that leaving high school before graduation is highly correlated with grade retention, low grades, and low test scores in early school years (Rumberger, 1995, 2004; Rumberger & Larson, 1998; Stearns, Moller, Blau, & Potochnick, 2007). Although studies addressing the issue of accumulated advantage (and disadvantage) focus primarily on changes in student-level performance over time, part of that dynamic possibly transpires in terms of school-level contributions to performance.


Regardless of the extent of reform, urban high schools have to contend with a number of nonschool problems that significantly influence minority students dropout tendencies. Most studies highlight economic, family, and neighborhood difficulties.

Low income and low wealth are strong predictors of dropping out (Astone & McLanahan, 1991; Pong & Ju, 2000; Rumberger & Larsen, 1998). Economic impoverishment not only deprives the students of tangible resources (adequate residential context, educational supplies, extracurricular activities, and so on) but also hinders their hopes and ambition for success (Linver, Fulingi, & Brooks-Gunn, 2004). Family structure and processes are also significant problems. Single-parent minority families, particularly those that are poor and female-headed, tend to compound dropout tendencies (Astone & McLanahan, 1991; Rumberger, 2004). Although single parenthood in the urban context does not guarantee ineffective childrearing,3 it tends to be associated with family processes and styles insufficient to support educational attainment (McLoyd et al., 1994; McNeal, 1997; Rumberger, 1995; Teachman, Paasch, & Carver, 1996). In particular, weak relationships among poor single parents and their children (e.g., lack of psychological engagement and proper monitoring) increase the odds of dropping out (Entwisle, Alexander, & Olson, 2005; McNeal, 1999). Moreover, because low-income single-parent families often find it difficult to maintain a stable residence, students experience a high rate of residential mobility, which implies frequent school changes as well. Both problems aggravate the risk of dropping out (Astone & McLanahan, 1994; Rumberger, 1995, 2003).

Chances of high school graduation are also undermined by impoverished community and spatial conditions. Neighborhoods characterized by racial and poverty concentration often lack the quality of community relations and trust, and the level of cooperation common in affluent neighborhoods (Coleman, 1988; Massey & Denton, 1993; Sampson, 2001; Sampson & Groves, 1989; Wilson, 1987, 1996; Wilson & Taub, 2007). These circumstances are strongly correlated with high dropout rates (Brooks-Gunn, Guo, & Furstenberg, 1993; Crane, 1991; Ensminger, Lamkin, & Jacobson, 1996; Hallinan & Williams, 1990; South, Baumer, & Lutz, 2003). Though some disadvantaged communities manage to create a context that nurtures childrens growth and developmentoften by means of tremendous collective efficacy (Schutz, 2006)many of them are unable to maintain strong norms against deviance and academic failure and suffer from the scarcity of affluent role models, prevalence of negative peer effects, and opportunities for criminal involvement.

Altogether, the effects of adverse economic, family, and neighborhood circumstances can reach levels that mask the contribution of the schools to student performance (Anyon, 1997; Gamoran, 1992; Noguera, 2003; Rothstein, 2004). Moreover, such problems interact in complex ways over time that have deeper impacts on students in terms of attitudes, cultural styles, and social competencies, which in turn impede academic performance (e.g., Lareau, 2003; Lewis, 1966; MacLeod, 1987). As Swidler (1986) noted, students growing up in poverty may develop a type of rationality that, although at odds with affluent-style success, is critical for survival in their own life world. According to Mickelson (1990), despite a positive attitude toward education, disadvantaged minorities may perform at low levels because in their life experience, high performance in school may not be as highly rewarded as other behaviors arebehaviors typically considered unproductive from an affluent perspective.


One of the most contentious episodes in the school desegregation controversy had to do with efforts to include predominantly White affluent suburbs in urban desegregation programs. Following the Supreme Courts Milliken ruling (Milliken v. Bradley, 1974),4 concerns for White harm were reflected mainly in the attitudes of working-class ethnic Whites residing in urban districts (Orfield, 1978). Though some suburbanized and others enrolled their children in private schools in response to integration, many whose children took part in citywide programs felt particularly threatened, given their near-affluent status and their own struggle for upward mobility (Sitkoff, 1993). The exclusion of suburban Whites from urban desegregation plans also increased the burden on working-class Whites because they were the only White group geographically available to help achieve and maintain racial balance in desegregated schools (Clotfelter, 2004).5 The resulting resistance not only created a volatile political situation but also helped perpetuate the fear that desegregation will harm White students.

Evidence has largely disproved the White harm hypothesis, irrespective of class status (Braddock & Eitle, 2004; T. Cook et al., 1984; Schofield, 2001). In fact, some studies have found that desegregation was moderately beneficial to White success (e.g., Mickelson, 2001). Yet, questions about White harm have typically been addressed in reference to performance at the student level, not in terms of the school-level effects on performance. As far as the dropout problem among urban Whites is concerned, there is good reason to expect favorable school-level effects under desegregated conditions because when implemented well, desegregation programs commonly aim at improving schools for both Whites and minorities rather than compromise the quality of majority-White schools to enhance minority schools in a zero-sum fashion (Clark, 1965; Hochschild, 1984; Orfield, 1978). Although the extent of discrepancies between planning and actual implementation may vary by case, proper monitoring and evaluation mechanisms are central in maintaining both high quality and equity across all schools in a given jurisdiction (Foster, 1973; Metz, 1986). When those mechanisms are in place and are effectively used, majority-White urban high schools are likely to experience minimum resource losses and may instead receive additional funds from local, state, and federal agencies for program implementation (an inherent aspect of many desegregation programs in the 1970s and 1980s).

In addition, White students can benefit from positive peer effects, improved curriculum, and other new administrative and educational practices, just as minority students do. After all, although there are potential quality differences between White and minority schools in urban school systems, White high schools are not dramatically better (Balfanz & Legters, 2004; Lippman et al., 1996), given the broad socioeconomic disadvantage affecting all urban residents.

Disentangling student- and school-level effects on dropout rates for Whites in urban districts is also important because, just as in the case of minority students, the effects of improved school quality may be insufficient to overcome the deleterious influence of nonschool problems. Although White students tend to suffer from a relatively lower degree of economic, family, neighborhood, and other contextual problems, the nature and impact of those difficulties are still considerable (Hauser, Simmons, & Pager, 2004). As a matter of fact, the slightly lower degree of nonschool disadvantage puts urban Whites at a higher risk of dropping out than it does minorities. For example, better access for Whites to low-level jobs in the urban labor market (Bane & Elwood, 1986; Ihlanfeldt & Sjoquist, 1990, 1991; ORegan, 1993) makes them more likely than minorities to drop out of high school (Hauser et al.). Thus, the question is whether desegregated schools counteract or aggravate the problem.



The Cleveland Municipal School District served nearly 120,000 students in 1973, of whom 45% were White. The NAACP sued the district that year for intentional segregation and discrimination against minority students. The court ruled in favor of the plaintiffs in 1976, and CMSD began implementing a citywide desegregation plan in fall 1979 (Reed v. Rhodes, 1976).6 Besides busing for racial mixing, the plan involved funding increases in previously majority-non-White schools to equalize them with previously majority-White schools. Curricular innovations, along with new testing, reading, counseling, guidance, safety, community relations, and extracurricular programs, were also implemented across all schools (Reed v. Rhodes, 1978). Finally, a limited magnet school program was instituted to facilitate voluntary desegregation. To keep CMSDs costs at a minimum, the cost of busing and funding equalization was picked up largely by Ohio, and much of the cost of the remaining initiatives was covered by federal resources, such as the Emergency School Aid Act.

All aspects of implementation were closely evaluated by the Office on School Monitoring and Community Relations (OSMCR), a local agency established by the court in 1978 to produce periodic status reports.7 OSCMR began its operations in early 1979. Besides racial mixing and other practices, funding equalization across the schools was an issue that the agency was particularly vigilant in monitoring, given the history of financial inequity, mismanagement, and inefficiency in the district.8

The schools remained integrated until 1993, when total student population was at 70,000, with only 20% White enrollment. Given the significant decline in the number of White students (which had been going on since the early 1970s) and Clevelands severe social and economic downturn in the late 1980s and early 1990s (driven primarily by intensified deindustrialization and social disorganization in the area; see Chow & Coulton, 1998; Miller & Wheeler, 1997), the court permitted gradual resegregation in 1993. Although racial mixing was abandoned in phases, a consent decree in 1994 (Reed v. Rhodes, 1994) mandated OSCMR to continue monitoring the district for equity across the schools until the district was declared unitary in 1998 with about 67,000 students, of whom only 18 percent were White. The equity mandate addressed a wide range of issues, including funds, facilities, supplies, teacher and staff credentials, curricula, and several other policies.

As part of its ruling in 1976, the federal court required CMSD to archive yearly personal, socioeconomic, and academic information on all its students. These archives formed the basis of the data used in this article. In particular, data on four high school cohorts were obtained, each cohort consisting of minority (Black and Hispanic)9 and White students. As shown in Figure 1, the cohorts were picked 5 years apart, in specific reference to CMSDs timeline of integration. The students in each cohort were observed from ninth grade onward. Therefore, time-varying data in the study pertain only to high school years. However, being 5 years apart, the cohorts varied in terms of the timing and exposure to desegregation both during and before high school.

Figure 1. Variation in the Degree and Timing of Desegregation Exposure across the Cohorts Included in the Sampling Frame of the Study

click to enlarge

Students in the first cohort had minimal exposure to desegregation. Those in the second cohort attended segregated schools up until late middle school, followed by desegregation. The third cohort attended desegregated schools the entire time. Finally, those in the fourth cohort attended desegregated elementary and middle schools, followed by gradual resegregation in high school. For each cohort, the observation window was limited to 6 years. The first 4 years (up to 12th grade) constitute the ideal time it would take to graduate. The next two account for grade retention in high school. Very few students remained enrolled beyond 6 years.10

The sizes of the cohorts varied between 6,339 and 8,550 (Tables A1 and A2 in the appendix provide more detailed information on cohort sizes, racial breakdowns, and yearly changes). In total, the data set included yearly records for 27,550 high school students, 73% of whom were minority (about 67% Black). The mean panel size was 2.96 years (3.09 for minorities and 2.65 for Whites). For the purposes of this study, magnet and vocational school students were excluded from the analysis, leaving those enrolled in regular, bilingual, special, and gifted education programs.11 Regular education students constituted over 80% of the total sample.

Because high school students initial enrollment dates in the district vary, there are two ways to define each cohort. The cohort sizes reported here are based on an unconstrained definition, where all members of the cohort are included regardless of when they had enrolled in CMSD prior to high school (i.e., any time between first and ninth grade). The constrained definition involves a smaller segment of the master data because each cohort includes only those students who had been enrolled in CMSD since first grade (i.e., the native CMSDers). Although the constrained definition controls out the potential effect of enrollment in other districts prior to high school, the number of students with limited pre-high-school CMSD histories (e.g., less than 2 years) was notably small. The majority of the latecomers to CMSD had enrolled in the district in early, middle, or late elementary school, which implies sufficient exposure to the school and nonschool effects in the district. Therefore, the unconstrained cohort definition was chosen for the analysis. However, the entire analysis was also replicated using data based on the constrained definition, which revealed essentially the same insights.12


Multilevel growth models are capable of disentangling student- and school-level effects on academic performance using longitudinal data (Raudenbush & Bryk, 2002; Snijders & Bosker, 1999). Ideally, a four-level logistic growth model is suitable for the analysis. Level 1 would estimate the dropout likelihood as a function of time over 6 years (within-student equation); Level 2 would estimate the effects of student-level characteristics (between-student equation); Level 3 would estimate school-level effects (between-schools equation); and Level 4 would estimate cohort-level effectsfor instance, by means of cohort membership dummiesbecause cohort membership is a nesting variable as well (between-cohort equation). However, testing this model, or even less complex versions of it, proved computationally infeasible.13

Therefore, a simpler approach was pursued, involving two stages. In the first stage, shown next, a three-level hierarchical logistic growth model was fitted. Level 1 estimated the log-odds (logit) of dropping out as a function of time, controlling for time-varying student-level characteristics that were treated as fixed effects (see Singer & Willet, 2003, for this approach at Level 1). Level 2 estimated the effect of the key time-invariant student characteristic, namely race.14 School-level effects (the third level in the ideal four-level model) were omitted, and cohort-level effects were entered at Level 3 in the form of cohort membership dummies. In the absence of an explicit school-level equation, cohort membership dummies would, for the most part, pick up the influence of different schooling treatments in the districtsegregation, desegregation, and resegregation in high school, as well as the pre-high-school exposure to desegregation (as shown in Figure 1).15 Basically, viewing the district as a collection of schools, the cohort membership dummies at Level 3 were an indirect means to estimate the change in school(ing) effects over time. It should be noted that within the time frame of this study, no programmatic, systemwide reform policy other than school desegregation was implemented at CMSD (see Saatcioglu & Carl, 2008).16 The downside of the three-level model was that the effects of specific school-level variables, such as compositional characteristics, remained unobserved (these were estimated in Stage 2 of the analysis). The model was specified as:


where i = student ID, t = year, and c = cohort membership. Dropping out (DR) was a binary outcome (1 = dropout, 0 = nondropout) based on a detailed withdrawal classification in CMSD archives. Dropouts were students who withdrew because of enrollment in adult education, full-time employment, illness or being over-age, enrollment in the armed forces, expulsion, or simply being lost to the system. Unlike in most dropout studies, nondropout withdrawals, such as transfers to other districts or to private schools, were effectively identified and measurement biases reduced.17 Time at Level 1 was coded from 0 and 5, referring to the 6-year observation window for each cohort. Its coefficient (β1ic) and the intercept (β0ic) at Level 1 were specified as random terms, contingent upon the effects of minority status (min; 1 = Black or Hispanic) at Level 2 and of cohort membership (coh2, coh3, coh4) at Level 3.

The vector of time-varying student-level controls at Level 1 (X(k-1)ic) comprised eight predictors. Because these controls were treated as fixed effects, the coefficient for each one represented the average influence of the variable, over time, on dropout likelihood. Nonschool controls included poverty status (1 = on free or reduced-price lunch), nontraditional family structure (1 = single- or no-parent family), levels of neighborhood personal crime and minority concentration (both percentage measures),18 and residential mobility (1 = different residence from previous year).19 Another important measure was school mobility (1 = different school from previous year), which accounted for the effect of changes in school context.20 Although students typically changed schools because of desegregation (and resegregation), some changes were due to residential mobility (except for in ninth grade, when the majority of students had to have left their middle school buildings to start high school in new buildings). Busing (1 = bused to school) was included to control for the effects of transportation, often for desegregation purposes (but at times for other reasons as well). Finally, two separate academic performance measures were included. Grade retention, which strongly affects dropout tendencies (Natriello, 1986; Rumberger, 2004), was entered as a lagged measure (1 = held back the previous year).21 Truancy was specified to account for attainment in the current year (percent of unexcused absence days). The yearly means and standard deviations of the time-varying predictors are presented in Tables A1 and A2 in the appendix.

In the second stage of the analysis, a two-level hierarchical logistic regression model was tested. This was a static model fitted separately for each cohort, and it was fitted twicefirst on data for the 1st year (ninth grade for all students), and then on data for the 4th year (12th grade for most students). This approach is based on earlier work by Goldschmidt and Wang (1999), who analyzed student- and school-level effects on dropout tendencies in early and late high school. The model examines the differences between initial dynamics in high school and the dynamics toward the end, conditional on the probability that the student had not dropped out of high school for at least 3 years. Thus, the two-level model is complementary to the three-level growth model in Stage 1. Whereas the latter model examines the influence of different schooling treatments (segregation, desegregation, resegregation) in the district by race, the former tests more closely the effects of particular school-level variables, such as compositional characteristics, at the beginning and near the end of the 6-year time window by cohort and race. The model was specified as:

click to enlarge

where i = student ID and j = school building ID. At Level 1, the logit of dropping out was regressed on the effect of minority status (min), gender (gen; 1=male), and the same vector of time-varying student-level controls (X(k-2)ij) used in the model in Stage 1. Only the intercept (λ0ij) and the coefficient for min1ij) were defined as random terms, contingent upon the effects of a vector of seven school-level predictors (Smj) at Level 2. These predictors included minority concentration (percent of non-Whites), poverty concentration (percent on free or reduced-price lunch), nontraditional family concentration (percent from single- or no-parent families), average neighborhood personal crime (the mean of neighborhood personal crime percentages for all students in the school), percentage of school mobility (proportion of students who were at a different school the previous year), percentage of residential mobility (proportion of students living at a residential address different from previous year), and school size (total number of students). Such characteristics often reflect school quality in public education (Gamoran, 1996; Pong, 1998) and are particularly subject to change as a district desegregates or resegregates. The yearly means and standard deviations for the school-level predictors are presented in Tables A3 and A4 in the appendix.

Data on school resources and organizational featuresfunds, facilities, supplies, teacher and staff credentials, curricula, and key educational and administrative practiceswere unavailable. However, the potential variation in these areas was kept to a minimum across the schools because of the federal courts mandate for equity. As noted earlier, OSMCR vigilantly monitored the schools compliance with the court order from 1979 to 1998, with particular emphasis on funding equalization. The downside of this, from an inquiry standpoint, is that it precludes the observation of school effects due to changes in resources and organizational features during much of the period analyzed in this study (even if the data for relevant variables were available). Nevertheless, because desegregation did not start until very late for the first cohort, that cohort is one in which school-level disparities, if any, were likely to have been at least partly due to unequal resources and organizational features. The rest of the cohorts were largely uniform in that regard, even the fourth one, which experienced resegregation.

Another important set of predictors omitted from Level 2because of unavailable dataare measures of teacher and staff skill and experience in the schools, the distribution of which was not monitored by the court. It is, however, assumed here that skill and experience correlate positively with formal credentials, the distribution of which was monitored by OSMCR.


Figure 1 shows the cumulative percentage of dropouts in each cohort by race. Each observation is the proportion of total dropouts from a particular race since starting high school in a given cohort. For instance, by the end of the 19771978 school year, 3.6% of minorities who had started out in Cohort 1 dropped out of CMSD. This equals 200 of the initial 5,574 Blacks and Hispanics in the cohort. The number accumulated to 1,228, or 22% of the initial 5,574 by the 4th year. In the 6th year, it reached 1,861, or 33%. A similar pattern is evident in Cohort 1 for Whites as well, whose total number was 2,932 in the initial year. Overall, Figure 1 shows that, except in Cohort 2, about one fourth of the students (22% of minorities and 26% of Whites) dropped out of CMSD by their 4th year in high school. As expected, the White dropout rate was higher than the minority dropout rate in almost all the years, due most likely to better access to low-level jobs for Whites in the urban labor marketalthough the racial difference appears to have closed steadily over time.22

Figure 2. Yearly Cumulative Percentage of Dropouts by Cohort and Race

click to enlarge

       Note. The values in the figure were generated using information from Tables A1 and A2 in the appendix. The value in any given year is the proportion of all students of a particular race who started in the relevant cohort (e.g., 5,574 minorities in Cohort 1; see Table A1) but have dropped out by that year. For example, 200, or 3.6%, of the 5,574 minorities in Cohort 1 dropped out in the 19771978 school year.  The following year, another 151 dropped out, adding up to 351 minorities, or 6.3% of the initial 5,574. In Year 3, an additional 312 dropped out, reaching 11.9% of the initial 5,574 minorities in the cohort.

A simple reading of the univariate patterns in Figure 2 can help make the case both for and against school desegregation. A proponent of the policy may interpret the apparent reduction in the cumulative dropout rate for minorities in Cohort 2 (reaching only 15% by the 4th year) as a sign of desegregations beneficial effect on minorities, because Blacks and Hispanics in Cohort 2 were the first minority group to experience a meaningful degree of desegregation in CMSD. An opponent, on the other hand, may view the reduction from Cohort 1 to Cohort 2 as too small to justify continued desegregation and may also raise the question of why the reduction (if it was indeed due to desegregation) largely disappeared in subsequent cohortsespecially in the third one, which remained integrated from Grades 1 through 12. Ultimately, from a policy makers standpoint, the outcomes in Figure 2 are unconvincing to promote school desegregation. The traditional scholarly approach, on the other hand, is likely to evaluate the role of student- and school-level factors in the outcomes by means of linear models that fail to properly disentangle the effects of those factors. As a result, the school-level benefits of desegregation, if any, may be masked.

The results of the hierarchical logistic growth model are shown in Table 1. The model was fitted in four steps. First, a null model, without any predictors, was fitted (Model 1) to produce baseline estimates of variance components at different levels. As seen in the table, the variances at Level 3 (0.352) and Level 2 (0.438) were statistically significant. In hierarchical logistic regression, residual variance at Level 1 (within-student level) is set by default at either 3.286 (π2/3; assuming standard logit distribution for the residuals) or 1.000 (assuming standard probit distribution; Hedeker, 2008; Snijders & Bosker, 1999). This results in a between-cohort intraclass correlation coefficient (ICC) of either 0.087 or 0.197, and a between-student ICC of either 0.107 or 0.245. Staying with the more conservative optionthat is, assuming a standard logit distribution for Level 1 residualsthe cohort level explains 8.7%, and the between-student differences explain 10.7%, of the variance in dropout rates. In other words, although changes in terms of segregated, desegregated, and resegregated schooling did make an important difference (controlling for pre-high-school exposure to desegregation), that difference accounted for a little less than 1/10 of the variation in dropout likelihood. The rest had to do with between- and within-student dynamics that the schools could do little about.

Table 1. Results of Hierarchical Logistic Growth Model Predicting the Likelihood of Dropping Out by Cohort and Race

click to enlarge

In Model 2, cohort membership dummies were entered at Level 3, treating Cohort 1 as the comparison groupthe most segregated set of students in the sample. Cohorts 2, 3, and 4 had significantly better effects on dropout likelihood than did Cohort 1. Cohort 3 had the best impact (-0.372, p < 0.050), followed by Cohort 4 (-0.188, p < 0.050), which was followed by Cohort 2 (-0.127, p < 0.050). Most important, the dummies significantly reduced the variance at the cohort level (0.026, p > 0.100), explaining nearly 94% of it. This means that the dummies adequately captured the cohort-level dynamics in the district, which had mostly to do with different schooling treatments.

Cohort membership dummies were removed in Model 3. Instead, time was entered at Level 1, and minority status was entered at Level 2. The principal objective was to obtain baseline estimates of the coefficients and variance components for these effects. The results indicated that each additional year in a CMSD high school increased the average students odds of dropping out by 10% (odds ratio = 1.103, p < 0.010). The between-student (Level 2) variance for this effect was also significant (0.113, p < 0.010). Being Black or Hispanic, on the other hand, decreased the average odds of dropping out by about 17% (odds ratio = 0.830, p < 0.010), with a significant between-cohort (Level 3) variance (0.112, p < 0.010). Thus, not only were the influences of time and race in the expected direction, but there was also considerable variance around each to be explained.

To help explain that variation to the best extent possible and to adjust the coefficient estimates using the relevant controls, all predictors were entered in Model 4. Not surprisingly, the cohort membership dummies explained 89% of the variance at Level 3. In particular, they rendered nonsignificant the variance of the race effect (0.028, p > 0.010). As for the time-varying student-level controls, because they were specified at Level 1 rather than at Level 2, their effects on the variance components at Level 2 were limited (about 10%). Still, because these controls adjusted the size of the time effect at Level 1 (reducing it from 0.098 to 0.069), they had a likely impact on the variance component of the time effect at Level 2 as well, reducing it from 0.097 (p < 0.050) to 0.072 (p < 0.100). Simply, the between-student variance of the time effect became barely significant once time-varying controls were entered.

As expected, key nonschool problems such as poverty and nontraditional family structure significantly increased the average odds of dropping out over time, each by about 20%. Neighborhood crime level and minority concentration were also important obstacles.

The former increased dropout likelihood by nearly 9%, and the latter increased it by about 6%. Residential and school mobility had negligible, nonsignificant effects. As far as academic performance was concerned, both grade retention (lagged by 1 year) and truancy significantly aggravated the dropout likelihood of the average studentretention did so by 15% and truancy by 12%. The only student-level control that reduced the odds of dropping out was busing, with a slight but significant effect of 4%, meaning that students who were transported to school were 4% less likely to drop out than those who were not transported.

Figure 3 portrays the influences of time, race, and cohort membership, based on their main effects and their two-way and three-way interactions (all but one of which were statistically significant). The dropout likelihood was well below 95% for both minorities and Whites in the ninth grade (Year 1), regardless of cohort. However, the subsequent growth trajectories varied considerably by both cohort and race. For minorities, the early benefits of desegregation are apparent in contrasting Cohorts 1 and 2: The yearly trend in the latter was increasingly better (less harmful) than that in the former. The improvement is attributable to mainly two factors. The first is the compositional, resource, and organizational changes in the high schools, changes that minorities in the first cohort did not experience at all until 1979. The second factor is the gradual exposure of minorities in the second cohort to the desegregation policy starting in middle school.

The situation was even better for minorities in Cohort 3. This is most likely due to long-term exposure to desegregation prior to high school, because the compositional, resource, and organizational attributes of the high schools were equitable for minorities in both Cohort 2 and Cohort 3. In other words, as expected, integration in elementary and middle grades appears to have contributed to lower dropout risks for Blacks and Hispanics in desegregated high schools.

Finally, the consequences of resegregation for minorities are evident in the yearly trend for Cohort 4. Blacks and Hispanics in this cohort appear to have lost ground in contrast to not only those in Cohort 3 but also those in Cohort 2. Much of this loss is attributable mainly to changes in the compositional characteristics of the high schools, because (1) the students in the fourth cohort, like those in the third, attended integrated schools prior to high school, and (2) OSMCR continued to monitor the schools for equity in funding and several other areas throughout the resegregation period.

Figure 3. Yearly Trends in Dropout Likelihood by Cohort and Race Based on the Results of Multilevel Logistic Growth Model

click to enlarge

The trends for Whites suggest that these students benefited from desegregation to a lesser extent than minorities did. This is evident in the relatively smaller reductions in yearly dropout likelihood from Cohort 1 to Cohort 2, and from Cohort 2 to Cohort 3. However, Figure 3 also indicates that Whites lost less ground than minorities did because of resegregation. Specifically, the yearly trend for Whites in Cohort 4 was slightly better than that in Cohort 3, but worse than that in Cohort 2. In simple terms, different schooling treatments (segregation, desegregation, resegregation) appear to have had a smaller effect on Whites than on minorities. Ultimately, the insights from Figures 2 and 3 strongly suggest that desegregation was not a source of White harm in terms of dropout rates at CMSD, a finding consistent with those from prior research.

The results of the two-level hierarchical logistic model, shown in Tables 2 and 3, offer further insights on the likely role of particular school-level variables in the trends observed in Figure 3. As described earlier, this model was fitted separately on the data for the 1st and the 4th year in each cohort. The fitting strategy was the same as in the first stage. Initially, a null model was specified to obtain baseline estimates of variance components for the levels (Model 1). Then, school-level variables were entered to determine the extent to which they explained the variance at the school level (Model 2). These variables were removed in the third step, where minority status, the only random effect, was entered to obtain baseline estimates of its coefficient and associated variance component (Model 3). In the final step, the full model was specified (Model 4).

Table 2. Results of Hierarchical Logistic Model Predicting the Effects of Student and

School Factors on Likelihood of Dropping OutCohorts 1 and 2

click to enlarge

Table 3. Results of Hierarchical Logistic Model Predicting the Effects of Student and

School Factors on Likelihood of Dropping OutCohorts 3 and 4

click to enlarge

The structure of variance components in the tables points to three important patterns.  First, assuming standard logit distribution for Level 1 residual variance, the ICCs suggest that, regardless of cohort, school-level dynamics account for about 7%8% of the variation in early high school, and about 10%11% of the variation in late high school (the estimates are greater under the assumption of standard probit distribution). These findings suggest that CMSD high schools made a somewhat greater difference in dropout tendencies in Year 4 than in Year 1. They are also consistent with the modest extent to which different schooling treatments (captured by the cohort membership dummies) in the three-level growth model were found to account for the variation in dropout likelihood (8.7%).

Second, the school-level predictors explain a considerable proportion of the variation at Level 2. Entering these predictors in the second step of the analysis rendered nonsignificant the between-school variance regardless of year or cohort (borderline significant in one case, Year 4 in Cohort 1). This means that the set of school-level predictorswhich lacked explicit measures for school resources and organizational featureswas both substantively and statistically instrumental.

Third, when race is the only predictor in the model, both its coefficient and its variance component are significant. However, these two estimates lose significance in the full model, where other student- and school-level predictors are included along with cross-level interactions. In other words, the full model not only nullifies the main effect of race (which is consistent with findings from past research on dropouts; see Rumberger, 2004) but also sufficiently explains the between-school (Level 2) variance of the race effect.

To highlight the ways in which the results of the two-level model complement those of the three-level growth model, Figure 4 presents the odds ratios for the school-level effects based on the coefficients shown in Tables 2 and 3. For one thing, any school-level effect estimated by the two-level model is less harmful in early than in late high school, which is consistent not only with past research (Goldschmidt & Wang, 1999) but also with the invariably positive slopes of the lines in Figure 3 that imply increasing dropout likelihood from early to late high school regardless of race and cohort. More important, there are cross-cohort differences in Figure 4 that resonate squarely with those observed in Figure 3. In particular, the effects of several predictors exhibit a U-shaped pattern across the cohorts, starting relatively high in Cohort 1, becoming gradually smaller in desegregated cohorts (Cohorts 2 and 3), and gaining in size again in the resegregated cohort (Cohort 4). In this regard, the two-level model helps explain part of the reason why the slopes of the lines vary the way they do across the cohorts in Figure 3.

Figure 4. Odds Ratios for School-Level Effects on the Likelihood of Dropping Out in Early and

Late High School by Cohort and Race

click to enlarge

click to enlarge

For example, Figure 4 indicates that, in Cohort 1, minority concentration in the high schools increases the average Black or Hispanic students odds of dropping out by 20% in the 1st year and by 39% in the 4th year. Although the odds ratio for the 1st year goes up to 28% in Cohort 2, the one for the 4th year drops to 33%. Both ratios are smaller and statistically nonsignificant in the third cohort. However, by the fourth cohort, the ratio for the 1st year climbs to 40%, whereas the one for the 4th year reaches 49%. The same pattern is evident for poverty concentration, nontraditional family concentration, average neighborhood personal crime, and the percentage of students who experienced residential mobility. Therefore, the more exposed minorities were to desegregation prior to high school, the greater were the beneficial effects of compositional changes for them during high school. Resegregation curbed those advantages to some degree, even though the minorities in Cohort 4 attended integrated schools prior to high school.  These patterns play a likely role in the way the slopes of the lines for minorities shift across cohorts in Figure 3. The exception is the odds ratio for the percentage of students who experienced school mobility. It remains largely stable and nonsignificant across the cohorts.  

Figure 4 also shows that although the results for the discussed effects are considerably similar between the racial groups, they are less pronounced for Whites. This is consistent with Figure 3, where there is a smaller degree of cross-cohort changes in the slopes of the lines for Whites. As such, the results of the two-level model not only reconfirm the lack of desegregations harm on Whites in CMSD but also demonstrate that White students in desegregated urban high schools can benefit from several school-level changes in pretty much the same way that minority students often do.

Despite the broad correspondence between Figures 3 and 4, it is important to refine the interpretation of the U-shaped pattern of the discussed school-level effects in two important ways. First, the size of the effects for minorities in Cohort 1 may partly reflect disparities in school resources and organizational features, because equity in these areas did not start until late in high school for Cohort 1. In the absence of relevant resource and organizational measures, the school-level effects specified in the model may be slightly overestimated for minorities because, as the court case demonstrated, majority-White schools were verifiably superior in terms of resource and organizational inequities prior to desegregation. The problem of omission bias is less of an issue for the estimates in the remaining cohorts, given the courts mandate for equity across all schools between 1979 and 1998. Ironically, had OSMCR not continued to monitor the schools for equity during the resegregation period (19931998), the estimates of the school-level effects for Cohort 4 may have been subject to considerable omission bias.  

Second, the drop in the effects of compositional disadvantages in Cohorts 2 and 3 is not a sign that predictors such as minority or poverty concentration became poorer proxies for school quality under the desegregation plan. This may occur when the policy achieves a perfect balance of race and class across the schools by means of forced assignment in a district with equal but segregated proportions of racial and economic groups. Yet, CMSD never had such racial or economic composition within the time frame of this study. More important, its desegregation plan balanced the schools only in terms of certain ranges around the districts overall racial composition, and other compositional factors were rarely, if ever, taken into account in student assignments, allowing considerable variation in school composition (see Tables A3 and A4 in the appendix for the yearly means and standard deviations for the school-level predictors). Finally, had the poor proxy issue been a problem in the analysis, several school-level predictors would have statistically failed not just in Cohort 3 but also in Cohort 2, because high schools in both cohorts operated under the desegregation plan. The difference between the two groups is more likely a result of the third cohorts longer exposure to desegregation prior to high school. Ultimately, there is good reason to interpret the changes in the school-level effects on dropout likelihood in the second and third cohorts as a sign of the educational advantages of integrated schools, advantages that began to erode because of resegregation in the fourth cohort.

The only odds ratio that remains to be interpreted in Figure 4 is that of school size, which follows a similar pattern across the cohorts. Evidently, the larger the high school, the higher the odds of dropping out. The most noticeable pattern in the effects of school size is the extent to which it surges in Cohorts 3 and 4 irrespective of race. A possible reason is that CMSD closed a number of its high schools in the mid-1980s because of fiscal problems. Because the size of the student population remained the same at the time and decreased only at a slow pace in subsequent years, the school closings are likely to have caused considerable overcrowding in the high schools. It appears that the impeding effects of school size was impervious to both desegregation and resegregation in CMSD. Still, minorities were slightly more disadvantaged in that regard, because the effects on Whites were less pronounced.

Figure 5 presents the odds ratios for student-level predictors from the full models shown in Tables 2 and 3. The main effect of race is, as noted earlier, small and nonsignificant, and it is fairly stable across the cohorts. The same is true for gender, although its effect is statistically significant. Consistent with past findings (Goldschmidt & Wang, 1999), the average male in CMSD was slightly less likely than the average female to drop out regardless of cohort. The effects of academic performance were also consistent with past findings. As in the results of the three-level growth model, both grade retention and truancy significantly increased the odds of dropping out, particularly in late high school. Their influence remained large and fairly stable across the cohorts.

School mobility had a small effect, which, contrary to findings from previous studies (Astone & McLanahan, 1994; Rumberger, 1995, 2003), remained statistically nonsignificant irrespective of year and cohort. This can be interpreted as partial evidence that school reassignments for purposes of desegregation (and resegregation) did not contribute to the mobility problem in CMSD enough to aggravate the dropout problem, at least when several other factors are held constant. A similar interpretation applies to the effect of transportation. As seen in Figure 5, regardless of cohort, being bused to school has a small but beneficialand at times significanteffect on dropout likelihood, particularly in early high school. In other words, busing students for desegregation appears not to have had a deleterious influence on the dropout problem. However, the interpretations for the school mobility and busing effects should be taken with caution, because the variance in neither variable solely reflects the districts efforts to maintain or undo racial balance in the schools (students often changed schools for reasons other than desegregation or resegregation).

Figure 5. Odds Ratios for Student-Level Effects on the Likelihood of Dropping Out in Early and Late High School by Cohort

click to enlarge

click to enlarge

Among the key nonschool problems that the schools had to contend with, poverty and nontraditional family structure are the most pronounced impediments to graduation, particularly in late high school. For Cohort 1, eligibility for free or reduced-price lunch increased the average odds of dropping out by 18% in the 1st year and by 26% in the 4th year. These estimates rose to 25% and 44%, respectively, in Cohort 4. Likewise, living in a single- or no-parent household increased the average odds of dropping out by 20% in the 1st year and by 32% in the 4th year in Cohort 1; the odds increased to 32% and 48%, respectively, in Cohort 4.  Similar but less pronounced results are observed for neighborhood personal crime and residential mobility. The ill effects of both problems steadily grew in size and statistical significance, reaching the 10%20% range in the late 1980s (third cohort). Altogether, being poor, growing up in a nontraditional family structure, being exposed to socially unstable neighborhood conditions, and frequent residential changes while in school became increasingly serious problems over time not only within any given cohort but also across the consecutive cohorts. Although the effects of these problems are well established in the dropout literature, the results here uniquely demonstrate the implications of the worsening conditions in the social ecology of an urban school district in the long run. A brief review of the means for several socioeconomic variables in Tables A1 and A2 also reveals the growing extent of the problems in Cleveland throughout the 1980s and 1990s, with particularly dire effects on minorities (see Chow & Coulton, 1998, as well).

The worsening nonschool problems are an important part of the difficulty that CMSD high schools likely had in curbing their students dropout tendencies regardless of reform. Had nonschool problems not intensified that way they did over time in CMSD, the drop in the cumulative percentage of dropouts for minorities in Cohort 2shown in Figure 2may have persisted in subsequent cohorts, at least in the third cohort. Instead, the percentages returned to levels that prevailed in Cohort 1, despite the school-level benefits of desegregation.

An unexpected finding in Figure 5 is the effect of neighborhood minority concentration. It appears that the concentration of Blacks and Hispanics in the average students census tract was somewhat beneficial in terms of the dropout problem regardless of cohort. There are two possible reasons for this. First, neighborhood minority concentration may not be adequately capturing unfavorable effects of community disadvantage in Cleveland, the effects that neighborhood personal crime appears to capture more effectively. Given that CMSD has remained a predominantly minority district in the time frame of this study and that it has traditionally been highly segregated residentially, the beneficial effects of neighborhood minority concentration may simply be a reflection of the lower unadjusted dropout rate for minorities as opposed to Whites (shown earlier in Figure 2). In other words, because Blacks and Hispanics tend to drop out less often than Whites do, census tracts with more Black and Hispanic students residing in them may appear to be beneficial in terms of the dropout problem. The second possible reason for this finding is that the two-level hierarchical model does not break down the effect of neighborhood minority concentration by race. That is, race-specific estimates for Whites and minorities separately may be different. Resolving the potential problem with the neighborhood minority concentration variable ultimately requires further analysis that is beyond the scope of this study.


This study has disentangled the school- and student-level causes of dropouts in Cleveland under conditions of segregation, desegregation, and resegregation, controlling for the degree of pre-high-school exposure to desegregation. The principal objective was to estimate the difference that desegregation made in the effects of schools on the dropout tendencies of predominantly poor minorities, net of student-level effects, many of which originate from outside the schools. The results demonstrate that desegregated high schools likely played a more effective role in counterbalancing student-level nonschool problems than did segregated ones. The benefits were most noticeable when Blacks and Hispanics attended integrated elementary and middle schools. However, although the desegregation policy had favorable consequences for the schools, they were not sufficient to bring about a total resolution to the dropout issue. Consistent with findings from past research, there was no evidence of White harm. Instead, Whites appear to have benefited from desegregated schools in similar ways, although to a lesser degree.

The findings point to an important problem in the debate over the effects and virtues of school desegregation. Because desegregation effects on student performance have traditionally been found to be modest at best, much of the debate has polarized between those who consider modest effects a success, and those who consider such effects a failure. Proponents have also promoted the policy on moral grounds, whereas opponents have advocated alternative strategies to integrate the schools, such as school choice. For these advocates, improvement in minority student performance was never the sole, perhaps not even the most important, purpose of desegregation (Wells, 2002; Wells, Holme, Revilla, & Atanda, 2009). Yet, student performance has been a critical yardstick by which to evaluate the policy, and the debate on it need not be stuck the way it has been. And because of the methodological shortcomings of prior researchthat is, the failure to properly disentangle school- and student-level contributions of desegregation to student performancethe benefits of desegregation at the school level may have been masked.

There is a need for a conceptual and an operational reframing of desegregation effects on student performance. Conceptually, desegregation, like many other reforms, is unlikely to be sufficient by itself to help close the achievement gap when students are continually affected by severe nonschool impediments to performance gains, such as racial isolation, poverty, family disruption, and social disorganization. Still, both the proponents and opponents of the policy tend to evaluate it as if reforming the schools via desegregation or other reforms is the primary means of closing the achievement gap. As W. E. Bell (1979) observed, Although many of the problems and the solutions originate outside the schools, the schools take the brunt of fixing and implementing them (p. 72). Is it not fairer and more politically balanced to consider the academic success or failure of desegregation in light of the fact that integrated schools may not be enough to achieve equal academic performance? In other words, desegregation policy should be evaluated with regard to the inherent limitations on what school reform can achieve in the first place. For instance, as the results of this study show, school-level dynamics in CMSD explain about 10% of the variance in dropout rates (about 20%30% under less conservative empirical assumptions). Therefore, desegregation or any other effort to alter the composition, structure, and organization of CMSD high schools could make only a modest difference in dropout likelihood under the best of circumstances. The modest nature of desegregations impact on dropout rates is a sign of neither success nor failure. Rather, the issue is whether desegregation succeeded in making the difference it can, given the limits of any school reform.

Addressing such a question requires an empirical approach different from those traditionally applied in desegregation evaluation, which may mask school-level effects. Contemporary multilevel analysis techniques help determine the extent to which the changes in the schools explain the difference the schools can make in student performance. The results from CMSD show that when the policy is implemented and monitored effectively, various school characteristics can account for most of the difference that the schools make in minority (and White) dropout rates. Changes in compositional features, such as minority concentration, poverty concentration, nontraditional family concentration, and unstable neighborhood concentration in the schools were critical in this regard. The results also offer partial evidence that funding equity and the equalization of key educational and administrative practices played a role in improving the schoolsat least from Cohort 1 to Cohort 2. Ultimately, although these changes were insufficient to reverse the dropout problem, they turned the average high school into an institution that cushions more effectively the negative effects of intensifying nonschool problems on graduation chances. This is not a small win, given the difficulties the schools face in the urban context. From that standpoint, school desegregation was successful in CMSD.

What are the policy implications of these findings? Is there any reason to feel good about a policy that may improve the schools, yet fail to reverse important educational problems, such as dropouts? Returning to the analogy used earlier in the article is instructive. Improvements in the medical field are important in fighting lung cancer, but they may still fail to eradicate the disease unless problems outside the medical field change, such as smoking habits, air and water pollution, and limited access to quality health care. In this regard, small reductions in the rate of the disease are not a sufficient reason to give up on medical improvements. What is needed, instead, is to continue the medial improvements while also attacking the causes outside the medical context. Desegregation is like that, at least with regard to urban schools. The fact that it delivers at the school level but fails at the student level should not necessarily be grounds to abandon it. Otherwise, unfavorable outcomes may be further exacerbated. As the results from CMSD indicate, during the resegregation period, the students (especially minorities) increasingly suffered from both school and nonschool problems. However, conventional wisdom justified dismantling desegregation rather than broadening its scope and also addressing the problems outside the schools. There was a shared assumption in the district that desegregation failed and that resegregation may be more effective in counteracting nonschool problems and improving student performance. One-race schools are OK, the citys popular Black mayor asserted as the district began reversing its desegregation plan (Theiss, 1993, p. 1A). He described the effort as the best opportunity in 20 years to create a quality school system.

However, the problems in the schools worsened. For the last decade, CMSD has ranked at or near the bottom of a list of 611 districts in Ohio, with a 72% dropout rate in 2001 (Greene, 2002). In 2002, the graduation rate for CMSDs Black students was the lowest in the nation (Greene, 2002). Among ninth graders in 2005, 80% scored below proficient in math, and about half scored below proficient in reading.23

Desegregation is not the only means to improve minority education (Morris, 2008; Watkins, 1996); neither is it an inherent necessity for progress in closing the achievement gap (Walker, 1996). However, by weakening the traditional link between social status and schooling, desegregation in Cleveland interrupted a fundamental pattern underlying the structure of educational inequality in America. In so doing, it enabled many minority children to experience considerable improvement in school quality. As national policy and community preferences have shifted away from desegregation, there has been an increasing focus on ways that seek to equalize minority schools while allowing them to remain separate. Improved funding, along with curricular, pedagogical, and administrative initiatives in urban schools, is a central concern. The results of this study illustrate the potential benefits of such efforts because good schools are important for all children. However, the results also show that compositional factors play a critical role in school quality as well, even when the schools are equalized in other ways. Besides, altering compositional patterns was in part a means of ensuring equity in other aspects of school quality. Thus far, resegregation has not been able to achieve this goal to the same extent that desegregation had (Kozol, 2004). As the nations schools rapidly resegregate (Logan, Oakley, & Stowell, 2008; Orfield, 2001), the goal of equal education is becoming ever more difficult to achieve. As disadvantaged minorities continue to be undermined by severe nonschool problems, their prospects need not be undermined further by separate and unequal schools.


The author wishes to thank the special issue editors Roslyn A. Mickelson and Kathryn Borman, as well as the anonymous reviewers, for valuable comments on earlier drafts. Special thanks also go to John L. Rury, Vicki Peyton, Eric P. Bettinger, Bruce D. Baker, Eric H. Neilsen, Claudia J. Coulton, and Jagdip Singh.


1. This is not to say that racially identifiable schools serving minorities are of inherently low quality (see, for example, Shujaa, 1996; Walker, 1996). Exceptions, however, do not invalidate the dominant trend in which segregation, particularly in large urban districts, ensures low school quality for African Americans and Hispanics (Hochschild & Skovronick, 2003; Kozol, 2004; Lippman et al., 1996).

2. Although attending such schools has traditionally been harmful to Black and Hispanic performance as well, the potential damage to Whites was always the predominant central concern in the desegregation era.

3. See Jarrett (1992, 1995, 1997) and Furstenberg and Cherlin (1993) concerning the exceptional skills and effort on the part of some poor, minority single parents in supporting their childrens success.

4. In Milliken v. Bradley, the Court barred the inclusion of suburban districts in urban desegregation plans unless the suburbs themselves were found guilty of fostering urban segregation.

5. It is important to note the that Millikens exclusion of the suburbs from urban desegregation plans also increased the burden on poor Black and Hispanic students, because these students would have to be transported over long distances to White neighborhoods within their own districts, rather than relatively shorter distances to neighborhoods in adjacent suburban districts (Clotfelter, 2004).

6. The State of Ohio was also found liable for failing to monitor CMSDs discriminatory practices.

7. These reports were prepared on a monthly and yearly basis. OSMCR also produced various comprehensive reports every few years to present more detailed information on the districts efforts to comply with the desegregation order and on the academic, social, and other outcomes of the program.

8. Funding discrimination against minority students and wasteful use of financial resources in the district were frequently addressed in the court hearings and in the eventual ruling (see Reed v. Rhodes, 1976). As a matter of fact, Federal Judge Frank J. Battisti preferred to appoint an independent financial expertDaniel R. McCarthy, a local tax lawyerrather than an educational specialist as head of OSMCR. McCarthy stayed on until 1989, when Daniel J. McMullen, a lawyer and former clerk to Judge Battisti, took over and stayed on until 1997. For evidence on the agencys efforts to maintain funding equity in the district, see Office on School Monitoring and Community Relations (1991), and Reed v. Rhodes, 1996).

9. As in most urban districts, discrimination against Blacks was the basis of the court case. Following the initial filing of the lawsuit, Hispanics also wanted to formally join the case on the plaintiff side but were instead granted friend of the court status. At the time, the proportion of Hispanic students was about 2% of total enrollment. By 1995, it would increase to 7%. Asians and other ethnicities were excluded from the study, given their consistently small group sizes (about 2 1/2% of the student population in any given year) and their uniquely stable schools and socioeconomic microcosms. Most students in this excluded group benefited from relatively superior educational and community circumstances by Cleveland standards. The Asian community, for instance, resides in a largely isolated area within the district, which, in many ways, resembles the inner-ring suburbs of the city. Asian students are often the top-performing ethnic group in CMSD.

10. About 2% in any given cohort.

11. Magnet school students were excluded because most magnet programs creamed the small proportion of affluent or near-affluent students in the district (see Saatcioglu, 2007), whose chances of dropping out were considerably lower than those of the average CMSD student. This pattern, which would bias the analysis of school effects, is common in many urban school systems (Smrekar & Goldring, 1999; Yu & Taylor, 1997). Magnet schools were also superior to regular schools because they received additional funds through the federal Magnet School Assistance Program. Magnet enrollment in the district remained around 5% of all students until 1992. From 1993 onward, it gradually rose to 11% as officials relied increasingly on voluntary desegregation as the main integration strategy in the resegregation era. Vocational school students were also excluded to limit potential biases estimation. The typical vocational student in CMSD enrolled in his or her program for the specific purpose of graduating with low-level manual or service labor skills. Given this purposeful engagement with the school, the students were considerably less likely than others to drop out of high school. The proportion of vocational students in CMSD never exceeded 4% of total enrollment in the time frame of this study.

12. This approach was deemed more efficient than (1) trying to determine an optimal enrollment year by which to cut latecomers from the study, or (2) testing the differences of the results from alternative cohort definitions for statistical significance, which would have further complicated the analysis.

13. The difficulties had to do with the dichotomous nature of the dependent variable, the number of levels, the large size of the cohorts, and, most important, the cross-classified nesting of students within schools. Multilevel growth models in much of the published literature involve data sets in which students remain in a given school over time. That was not the case here, because students in any given cohort frequently changed schools because of desegregation and resegregation, as well as residential mobility. Cross-classifying students by school and yearinvolving more than 30 high schools over a period of 21 yearsrequired a complex equation of school-level effects at Level 3 (Raudenbush & Bryk, 2002; Wright, Sanders, & Rivers, 2006). Given the additional cohort membership dummies at Level 4, the student-level effects at Level 2, and the time effect at Level 1, the overall model involved a significantly high number of three-way and four-way interaction terms and random variance terms. The model was operationalized both in STATA 10.1, using xtmelogit, and in SAS 9.1, using GLIMMIX, on a 3.00GHz, Pentium 4 PC with 4GB RAM and 9999 MB virtual memory. In either case, the procedure failed despite several different convergence alternatives and strategies to subsample the data.

As an alternative, the levels in the model were rearranged for computational parsimony on the same hardware platform. First, time-varying student-level controls were entered as fixed effects at Level 1, but this did not solve the problem. Next, as an additional change, cohort membership was treated as a time-invariant student characteristic at Level 2, rather than a nested district-level effect at Level 4. Specifically, Level 2 involved the main effects of race and cohort membership dummies, along with race*cohort membership interactions. Although this model ran relatively more efficiently, it too ultimately failed to converge, most likely because of the complexity introduced by cross-classified school-level effects. During the estimation, each convergence iteration in STATA 10.1 lasted about 4 hours, and each of the several attempts to fit the model failed after about 20 iterations (meaning 7080 hours). The specification for both the ideal four-level model and its three-level versions described previously are available from the author upon request. The modeling approach presented in this article was the most sophisticated alternative that worked given the hardware limitations. Although not as complex, it suffices given the basic argument of the article.

14. In a separate analysis, not shown here, gender was also included at Level 2 and was found to have a nonsignificant effect with a negligible size. It was excluded from the analysis presented here for purposes of simplicity in visualizing race- and cohort-specific dropout trends based on the results of the model.

15. The cohort membership dummies would also capture the influence of any other cohort effects left over from educational dynamics that transpire through the schools. These leftover effects may have to do with factors such as the generationally determined attitudinal or normative differences across the cohorts, or the changing demographics of the entire district over time.

16. In 1993, the district implemented a limited expansion in its magnet programs, under the name Vision 21. But, as noted earlier, students in magnet schools were excluded from the analysis. In 1996, the State of Ohio started a small voucher initiative in CMSD, the Cleveland Scholarship Program. Although CMSD records did not indicate which students enrolled in this program, the total number was likely to be quite low, because the program was in its pilot stages. As such, the voucher initiative was unlikely to have had any significant effect on schooling and other dynamics in the district at the time.

17. The number of nondropout withdrawals in each year for each cohort by race can be calculated using the information in the third and fourth columns (Cohort Size and Dropout) provided in Tables A1 and A2 in the appendix. It should also be noted that nearly 75% of nondropout withdrawals involved transfers to other school systems or to private schools in the greater metropolitan Cleveland region (especially the inner-ring suburbs). In the majority of these cases, the withdrawing student responded to a specific question regarding the name for the particular district or school that he or she was transferring to. Therefore, the majority of nondropout withdrawals were likely to have remained in school and to not have dropped out, at least for 1 more year after leaving CMSD. This is an important point because the dropout code for nondropout withdrawals was set to 0 rather than 1 for their exit year from the data set. Nevertheless, coding these cases the other way around, as dropouts, had only negligible effects on the eventual results of the analysis. Ideally, a multinomial hierarchical logistic growth model can be used to estimate the likelihood of dropouts and nondropout withdrawals in the same procedure. If some students who were nondropout withdrawals dropped out of their new districts or private schools within a few years time (perhaps because of deleterious effects that had originated from CMSD schools), such dynamics remain unaccounted for in this analysis.

18. Crime and minority concentration are conceptualized in combination to reflect the degree of neighborhood or community disadvantage. To generate these variables, the residential address of each student in each year was geo-coded, which helped determine the students residential census tract. Yearly personal crime percentages for Cleveland tracts, compiled from official police reports, were available from the Center on Urban Poverty and Social Change at Case Western Reserve University. Personal crime is a frequently used aggregate social disorganization measure consisting of homicide, assault, aggravated assault, rape, and robbery (Jargowsky, 1997). Neighborhood minority concentration was obtained by calculating the percentage of minority students in each census tract.

19. The data on this lagged measure for the first year in each cohort were obtained with special permission from CMSD because access to records outside the time window of the sampling frame was required.

20. ibid.

21. ibid.

22. See Wang and Minor (2002) on the shrinking size of the urban labor market in the Cleveland region.

23. Data are available from the Ohio Department of Education (http://www.ode.state.oh.us).


Alexander, K. L., Entwisle, D. R., & Horsey, C. (1997). From first grade forward: Early foundations of high school dropout. Sociology of Education, 70, 87107.

Angrist, J. D., & Lang, K. (2004). Does school integration generate peer effects? Evidence from Boston's METCO Program. American Economic Review, 94, 16131634.

Antonio, A. L., Chang, M. J., Hakuta, K., Kenny, D. A., Levin, S., & Milem, J. F. (2004). Effects of racial diversity on complex thinking in college students. Psychological Science, 15, 507510.

Anyon, J. (1997). Ghetto schooling: A political economy of urban educational reform. New York: Teachers College Press.

Armor, D. J. (1995). Forced justice: School desegregation and the law. New York: Oxford University Press.

Armor, D. J. (2002). Desegregation and academic achievement. In C. H. Rossell, D J. Armor, & H. J. Walberg (Ed.), School desegregation in the 21st century (pp. 147187). Westport, CT: Praeger.

Astone, N. M., & McLanahan, S. S. (1991). Family structure, parental practices and high school completion. American Sociological Review, 56, 309320.

Astone, N. M., & McLanahan, S. S. (1994). Family structure, residential mobility, and school dropout: A research note. Demography, 31, 575584.

Baker, B. D., & Green, P. C. (2005). Tricks of the trade: State legislative actions in school finance policy that perpetuate racial disparities in the post-Brown era. American Journal of Education, 111, 372413.

Balfanz, R., & Legters N. E. (2004). Locating the dropout crisis: Which high schools produce the nation's dropouts? In G. Orfield (Ed.), Dropouts in America: Confronting the graduation rate crisis (pp. 5784). Cambridge, MA: Harvard Education Press.

Bane, M. J., & Elwood, D. T. (1986). Slipping into and out of poverty: The dynamics of spells. Journal of Human Resources, 21, 123.

Belfield, C. R., & Levin, H. (2007). The education attainment gap: Who's affected, how much, and why it matters In C. R. Bellfield & H. M. Levin (Ed.), The price we pay: Economic and social consequences of inadequate education (pp. 120). Washington, DC: Brookings Institution Press.

Bell, D. (2004). Silent covenants: Brown v. Board of Education and the unfulfilled hopes for racial reform. New York: Oxford University Press.

Bell, W. E. (1979). Obstacles in urban school renewal. Theory Into Practice, 18, 6572.

Besharov, D. J., & West, A. (2002). African-American marriage patterns. In A. Thernstrom & S. Thernstrom (Ed.), Beyond the color line: New perspectives on race and ethnicity in America (pp. 95114). Stanford, CA: Hoover Institution Press.

Betts, J., Zau, A. C., & Rice, L. A. (2003). Determinants of student achievement: New evidence from San Diego. San Francisco: Public Policy Institute of California.

Braddock, J. H., & Eitle, T. M. (2004). The effects of school desegregation. In J. A. Banks & C. A. M. Banks (Ed.), Handbook of research on multicultural education (pp. 828843). San Francisco: Jossey-Bass.

Brooks-Gunn, J., Guo, G., & Furstenberg, F. F. (1993). Who drops out of and who continues beyond high school? A 20-year follow-up of black urban youth. Journal of Research on Adolescence, 3, 271294.

Bryk, A. S., & Thum, Y. M. (1989). The effects of high school organization on dropping out: An exploratory investigation. American Educational Research Journal, 26, 353383.

Caldas, S. J., & Bankston, C. L. (2005). Forced to fail: The paradox of school desegregation. Westport, CT: Praeger.

Cashin, S. (2004). The failures of integration: How race and class are undermining the American dream. New York: Perseus Books.

Chow, J., & Coulton, C. (1998). Was there a social transformation of urban neighborhoods in the 1980s? A decade of worsening social conditions in Cleveland, OH. Urban Studies, 35, 13591375.

Clark, K. B. (1965). Dark ghetto: Dilemmas of social power. New York: Harper & Row.

Clotfelter, C. T. (2004). After Brown: The rise and retreat of school desegregation. Princeton, NJ: Princeton University Press.

Clotfelter, C., Ladd, H., & Vigdor, J. (2005). Who teaches whom? Race and the distribution of novice teachers. Economics of Education Review, 24, 377392.

Coleman, J. S. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94, S95S120.

Coleman, J. S., Campbell, E., Hobson, C., McPartland, J., Mood, A., Weinfeld, F. D., et al. (1966). Equality of educational opportunity. Washington, DC: Department of Health, Education, and Welfare.

Conley, D., & Albright, K. (Eds.). (2004). After the bell: Family background, public policy, and educational success. New York: Routledge.

Cook, P. J., & Ludwig, J. (1998). The burden of "acting White": Do Black adolescents disparage academic achievement? In C. Jenks & M. Phillips (Eds.), The Black-White test score gap (pp. 375400). Washington, DC: Brookings Institution Press.

Cook, T. D., Armor, D., Crain, R., Miller, N., Stephan, W., Walberg, H., et al. (1984). School desegregation and Black achievement. Washington, DC: U.S. Department of Education, National Institute of Education.

Crain, R. L., & Mahard, R. (1978). Desegregation and Black achievement. A review of the research. Law and Contemporary Problems, 42, 1758.

Crane, J. (1991). The epidemic theory of ghettos and neighborhood effects on dropping out and teenage childbearing. American Journal of Sociology, 96, 12261259.

Croninger, R., & Lee, V. (2001). Social capital and dropping out of high school: Benefits to at-risk students of teachers' support and guidance. Teachers College Record, 103, 548581.

Downey, D. B., von Hippel, P. T., & Hughes, M. (2008). Are "failing" schools really failing? Using seasonal comparison to evaluate school effectiveness. Sociology of Education, 81, 242270.

Downey, D. M., & Coyne, J. C. (1990). Children of depressed parents: An integrative review. Psychological Bulletin, 108, 5076.

Duncan, G. J., Biosjoly, J., Levy, D. M., Kremer, M., & Eccles, J. (2003). Empathy or antipathy? Consequences of racially diverse peers on attitudes and behaviors (Working Paper). Evanston, IL: Joint Center for Policy Research, Northwestern University.

Ensminger, M. E., Lamkin, R. P., & Jacobson, N. (1996). School leaving: A longitudinal perspective including neighborhood effects. Child Development, 67, 24002416.

Entwisle, D. R., Alexander, K. L., & Olson, L. S. (2005). First grade and educational attainment by age 22: A new story. American Journal of Sociology, 110, 14581502.

Erickson, R. J., & Simon, R. J. (1998). The use of social science data in Supreme Court decisions. Champaign: University of Illinois Press.

Felice, L. G., & Richardson, R. L. (1976). The effects of busing and school desegregation on majority and minority student dropout rates: An evaluation of school socio-economic composition and teachers' expectations. Journal of Educational Research, 70, 242246.

Finn, J. D. (1989). Withdrawing from school. Review of Educational Research, 59, 117142.

Foster, G. (1973). Desegregating urban schools: A review of techniques. Harvard Educational Review, 43, 536.

Furstenberg, F. F., & Cherlin, A. J. (1993). How families manage risk and opportunity in dangerous neighborhoods. In W. J. Wilson (Ed.), Sociology and the public agenda (pp. 231258). Newbury Park, CA: Sage.

Gamoran, A. (1992). Social factors in education. In M. C. Alkin (Ed.), Encyclopedia of educational research (pp. 12221229). New York: Macmillan.

Gamoran, A. (1996). Student achievement in public magnet, public comprehensive, and private city high schools. Educational Evaluation and Policy Analysis, 18, 118.

Garfinkel, I., & McLanahan, S. S. (1986). Single mothers and their children: A new American dilemma. Washington, DC: Urban Institute Press.

Gates, H. L. (2004). America behind the color line: Dialogues with African Americans. New York: Warner Books.

Glazer, N. (1975). Affirmative discrimination: Ethnic inequality and public policy. New York: Basic Books.

Gold, B. A. (2007). Still separate and unequal: Segregation and the future of urban school reform. New York: Teachers College Press.

Goldschmidt, P., & Wang, J. (1999). When can schools affect dropout behavior? A longitudinal multilevel analysis. American Educational Research Journal, 36, 715738.

Greene, J. P. (2002). High school graduation rates in the United States. New York: Manhattan Institute for Policy Research.

Grissmer, D., Flanagan, A., & Williamson, S. (1998). Why did the Black-White gap narrow in the 1970s and 1980s? In C. Jencks & M. Phillips (Eds.), The Black-White test score gap (pp. 181226). Washington, DC: Brookings Institution Press.

Gurin, P., Dey, E. L., Hurtado, S., & Gurin, G. (2002). Diversity and higher education: Theory and impact on educational outcomes. Harvard Educational Review, 330, 330366.

Guryan, J. (2004). Desegregation and black dropout rates. American Economic Review, 94, 919943.

Hallinan, M. T., & Williams, R. A. (1990). Students' characteristics and the peer-influence process Sociology of Education, 63, 122132.

Hanushek, E. A., Kain, J. F., & Rivkin, S. G. (2008). New evidence about Brown v. Board of Education: The complex effects of school racial composition on achievement (Working Paper 8441). Cambridge, MA: National Bureau of Economic Research.

Hanushek, E. A., & Raymond, M. E. (2005). Does student accountability lead to improved student performance? Journal of Policy Analysis and Management, 24, 297327.

Hauser, R. M., Simmons, S. J., & Pager, D. I. (2004). High school dropout, race/ethnicity, and social background from the 1970s to the 1990s. In G. Orfield (Ed.), Dropouts in America: Confronting the graduation crisis (pp. 85106). Cambridge, MA: Harvard Education Press.

Hedeker, D. (2008). Multilevel models for ordinal and nominal variables. In J. de Leeuw & E. Meijer (Eds.), Handbook of multilevel analysis (pp. 237274). New York: Springer.

Henig, J. R., Hula, R. C., Orr, M., & Pedescleaux, D. S. (1999). The color of school reform: Race, politics, and the challenge of urban education. Princeton, NJ: Princeton University Press.

Hochschild, J. H. (1984). The new American dilemma: Liberal democracy and school desegregation. New Haven, CT: Yale University Press.

Hochschild, J. H., & Skovronick N. (2003). The American dream and the public schools. New York: Oxford University Press.

Hofferth, S. L., & Sandberg, J. F. (2001). How American children spend their time. Journal of Marriage and the Family, 63, 295308.

Hoxby, C. M. (2000). Peer effects in the classroom: Learning from gender and race variation (Working Paper No. 7867). Cambridge, MA: National Bureau of Economic Research.

Ihlanfeldt, K. R., & Sjoquist, D. L. (1990). Job accessibility and racial differences in youth employment rates. American Economic Review, 80, 267276.

Ihlanfeldt, K. R., & Sjoquist, D. L. (1991). The role of space in determining the occupations of Black and White workers. Regional Science and Urban Economics, 21, 295315.

Irons, P. H. (2002). Jim Crow's children: The broken promise of the Brown decision New York: Viking.

Jargowsky, P. A. (1997). Poverty and place: Ghettos, barrios, and the American city. New York: Russell Sage Foundation.

Jarrett, R. L. (1992). A family case study: An examination of the underclass debate. In J. F. Gilgun, G. Handel, & K. Daley (Ed.), Qualitative methods in family research (pp. 172197). Newbury Park, CA: Sage.

Jarrett, R. L. (1995). Growing up poor: The family experiences of socially mobile youth in low-income African-American neighborhoods. Journal of Adolescent Research, 10, 111135.

Jarrett, R. L. (1997). Resilience among low-income African-American youth: An ethnographic perspective. Ethos, 25, 112.

Jencks, C., & Mayer, S. (1990). The social consequences of growing up in a poor neighborhood. In L. E. Lynn & M. G. H. McGeary (Ed.), Inner-city poverty in the United States (pp. 111186). Washington, DC: National Academy Press.

Jencks, C., Smith, M., Acland, H., Bane, M. J., Cohen, D., Gintis, H., et al. (1972). Inequality: A reassessment of the effect of family and schooling in America. New York: Harper Press.

Kasen, S., Cohen, P., & Brook, J. S. (1998). Adolescent school experiences and dropout, adolescent pregnancy, and young adult deviant behavior. Journal of Adolescent Research, 13, 4972.

Kluger, R. (1976). Simple justice: The history of the Brown v. Board of Education and Black America's struggle for equality. New York: Vintage Books.

Konstantopoulos, S. H., & Hedges, L. V. (2008). How large an effect can we expect from school reforms? Teachers College Record, 110, 16111638.

Kozol, J. (1988). Rachel and her children: Homeless families in America. New York: Crown.

Kozol, J. (1991). Savage inequalities: Children in America's schools. New York: Crown.

Kozol, J. (2004). The shame of the nation: The restoration of apartheid schooling in America. New York: Crown.

Lankford, H., Loeb, S., & Wyckoff, J. (2002). Teacher sorting and the plight of urban schools: A descriptive analysis. Educational Evaluation and Policy Analysis, 110, 3762.

Lareau, A. (2003). Unequal childhoods: Class, race, and family life. Berkeley: University of California Press.

Lee, J. (2002). Racial and ethnic achievement gap trends: Reversing the progress toward equity? Educational Researcher, 31(1), 312.

Lewis, O. (1966). La vida: A Puerto Rican family in the culture of poverty. New York: Random House.

Linver, M. R., Fulingi, A. S., & Brooks-Gunn, J. (2004). How do parents matter? Income, interactions, and intervention during childhood. In D. Conley & K. Albright (Ed.), After the bell: Family background, public policy, and educational success (pp. 2550). New York: Routledge.

Lippman, L., Burns, S., & McArthur, E. (1996). Urban schools: The challenge of locations and poverty (No. NCES 96-184). Washington, DC: U.S. Department of Education, Office of Educational Research and Improvement.

Logan, J. R., Oakley, D., & Stowell, J. (2008). School segregation in metropolitan regions, 19702000: The impacts of policy choices on public education. American Journal of Sociology, 113, 16111644.

Lucas, S. R. (1999). Tracking inequality: Stratification and mobility in American high schools. New York: Teachers College Press.

MacLeod, J. (1987). Ain't no makin' it: Aspirations and attainment in a low-income neighborhood. Boulder, CO: Westview Press.

Massey, D. S., & Denton, N. A. (1993). American apartheid: Segregation and the making of the underclass. Cambridge, MA: Harvard University Press.

Mayer, D., Mullens, J., & Moore, M. (2000). Monitoring school quality: An indicators report. Washington, DC: U.S. Department of Education, NCES, Office of Educational Research and Improvement.

Mayer, S. E., & Jencks, C. (1989). Growing up in poor neighborhoods: How much does it matter? Science, 243(4897), 14411445.

McClendon, M. J. (1974). Interracial contact and the reduction of prejudice. Sociological Focus, 7, 4765.

McLoyd, V. C., Jayaratne, T. E., Ceballo, R., & Borquez, J. (1994). Unemployment and work Interruption among African American single mothers: Effects on parenting and adolescent socioemotional functioning. Child Development, 65, 562589.

McNeal, R. B. (1997). High school dropouts: A closer examination of school effects. Social Science Quarterly, 78, 209222.

McNeal, R. B. (1999). Parental involvement as social capital: Differential effects on science, achievement, truancy, and dropping out. Social Forces, 78, 117144.

Metz, M. H. (1986). Different by design: The context and character of three magnet schools. New York: Routledge & Kegan Paul.

Mickelson, R. A. (1990). The attitude-achievement paradox among black adolescents. Sociology of Education, 63, 4461.

Mickelson, R. A. (2001). Subverting Swann: Tracking and second generation segregation in the Charlotte-Mecklenburg Schools. American Educational Research Journal, 38, 215252.

Miller, C. P., & Wheeler, R. A. (1997). Cleveland: A concise history, 1796-996 (2nd ed.). Bloomington: Indiana University Press.

Milliken v. Bradley, 418 U.S. 717 (1974).

Mondale, S., & Patton, S. B. (Ed.). (2001). School: The story of American public education. Boston: Beacon Press.

Morenoff, J., Sampson, R. J., & Raudenbush, S. (2001). Neighborhood inequality, collective efficacy, and the spatial dynamics of urban violence. Criminology, 39, 517560.

Morris, J. E. (2008). Research, ideology, and the Brown decision: Counter-narratives on the historical and contemporary representation of Black schooling. Teachers College Record, 110, 713732.

National Center for Education Statistics. (2000). NAEP 1999: Trends in academic progress: Three decades of student performance. Washington, DC: U.S. Department of Education.

National Center for Education Statistics. (2005). NAEP 2004: Trends in academic progress: Three decades of student performance. Washington, DC: U.S. Department of Education.

Natriello, G. (Ed.). (1986). School dropouts: Patterns and policies. New York: Teachers College Press.

Natriello, G. (1995). Dropouts: Definitions, causes, consequences and remedies. In P. W. Cookson & B. Schneider (Eds.), Transforming schools (pp. 107128). New York: Garland.

Nettles, S. M. (1991). Community involvement and disadvantaged students. Review of Educational Research, 61, 379406.

Noguera, P. (2003). City schools and the American dream: Reclaiming the promise of public education. New York: Teachers College Press.

Office on School Monitoring and Community Relations. (1991). Periodic status report. Cleveland, OH: Cleveland Municipal School District.

Ogbu, J. U. (2003). Black American students in an affluent suburb: A study of academic engagement. Mahwah, NJ: Erlbaum.

O'Regan, K. M. (1993). The effect of social networks and concentrated poverty on Black and Hispanic youth unemployment. Annals of Regional Science, 27, 327342.

Orfield, G. (1978). Must we bus? Segregated schools and national policy. Washington, DC: Brookings Institution.

Orfield, G. (2001). Schools more separate: Consequences of a decade of resegregation. Cambridge, MA: The Harvard Civil Rights Project.

Orfield, G. (Ed.). (2004). Dropouts in America: Confronting the graduation crisis. Cambridge, MA: Harvard Education Press.

Orfield, G., Arenson, J., Jackson, T., Bohrer, C., Gavin, D., & Kalejs, E. (1997). City-suburban desegregation: Parent and student perspectives in metropolitan Boston. Cambridge, MA: The Civil Rights Project.

Orfield, G., & Ashkinaze, C. (1991). The closing door: Conservative policy and black opportunity. Chicago: University of Chicago Press.

Orfield, G., & Eaton, S. E. (1996). Dismantling desegregation: The quiet reversal of Brown v. Board of Education. New York: The New Press.

Orfield, G., & Lee, C. (2005). Why segregation matters: Poverty and educational inequality. Cambridge, MA: The Civil Rights Project, Harvard University.

Patterson, J. T. (2001). Brown v. Board of Education: A civil rights milestone and its troubled legacy. New York: Oxford University Press.

Pettigrew, T. F., & Tropp, L. (2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90, 751783.

Phillips, M., & Chin, T. (2004). School inequality: What do we know? In K. Neckerman (Ed.), Social inequality (pp. 417569). New York: Russell Sage Foundation.

Pong, S. (1998). The school compositional effect of single parenthood on 10th-grade achievement. Sociology of Education, 71, 2342.

Pong, S. L., & Ju, D. B. (2000). The effects of change in family structure and income on dropping out of middle and high school. Journal of Family Issues, 21, 147169.

Rabe-Hesketh, S., & Skrondal, A. (2008). Multilevel and longitudinal modeling using Stata. Second edition. College Station, TX: Stata Press.

Raudenbush, S. W. (2004). Schooling, statistics, and poverty: Can we measure school improvement? Princeton, NJ: Policy Evaluation and Research Center, Educational Testing Service.

Raudenbush, S. W., & Bryk, A. S., (2002). Hierarchical linear models: Applications and data analysis methods. Second Edition. Newbury Park, CA: Sage.

Raudenbush, S. W., Fotiu, R. P., & Cheong. Y. F. (1998). Inequality of access to educational resources: A national report card for eighth-grade math. Educational Evaluation and Policy Analysis, 20, 253267.

Reber, S. J. (2007). Desegregation and educational attainment for Blacks (Working Paper No. 13193). Cambridge, MA: National Bureau of Economic Research.

Reed v. Rhodes, 422 F. Supp. 708 (ND Ohio 1976).

Reed v. Rhodes, 455 F. Supp. 569 (N. D. Ohio 1978).

Reed v. Rhodes, 869 F. Supp. 1265 (N.D. Ohio 1994).

Reed v. Rhodes, 934 F. Supp. 1533 (N.D. Ohio 1996).

Rossell, C. H. (1990). The carrot or the stick for school desegregation policy. Philadelphia: Temple University Press.

Rossell, C. H., Armor, D. J., & Walberg, H. J. (Eds.). (2002). School desegregation in the 21st century. Westport, CT: Praeger.

Rothstein, R. (2004). Class and schools: Using social, economic, and educational reform to close the Black-White achievement gap. New York: Teachers College Press.

Rumberger, R. W. (1987). High school dropouts: A review of issues and evidence. Review of Educational Research, 57, 101121.

Rumberger, R. W. (1995). Dropping out of middle school: A multilevel analysis of students and schools. American Educational Research Journal, 32, 583625.

Rumberger, R. W. (2003). The causes and consequences of student mobility. Journal of Negro Education, 72, 621.

Rumberger, R. W. (2004). Why students drop out of school. In G. Orfield (Ed.), Dropouts in America: Confronting the graduation rate crisis (pp. 131155). Cambridge, MA: Harvard Education Press.

Rumberger, R. W., & Larson, K. A. (1998). Student mobility and the increased risk of high school dropout. American Journal of Education, 107, 135.

Rumberger, R. W., & Palardy, G. J. (2004, April). Test scores, dropout rates, and transfer rates as alternative indicators of school performance. Paper presented at the annual meeting of the American Educational Research Association, New Orleans, LA.

Rumberger, R. W., & Palardy, G. J. (2005). Does segregation still matter? The impact of student composition on academic achievement in high school. Teachers College Record, 107, 19992045.

Rumberger, R. W., & Thomas, S. L. (2000). The distribution of dropout and turnover rates among urban and suburban high schools. Sociology of Education, 73, 3967.

Saatcioglu, A. (2007). Latent conflict in urban public education: Silent domination and the institutionalization of discriminatory organizational forms. Unpublished doctoral dissertation, Case Western Reserve University, Cleveland, OH.

Saatcioglu, A., & Carl, J. (2008). The discursive turn in school desegregation: Evidence from the Cleveland Municipal School District, 19731998 (Working Paper). Lawrence: University of Kansas.

St. John, N. (1975). School desegregation: Outcomes for children. New York: Wiley.

Sampson, R. J. (2001). How do communities undergird or undermine human development? Relevant contexts and social mechanisms. In A. Booth & A. C. Crouter (Eds.), Does it take a village? Community effects on children, adolescents, and families (pp. 330). Mahwah, NJ: Erlbaum.

Sampson, R. J., & Groves, W. B. (1989). Community structure and crime: Testing social disorganization theory. American Journal of Sociology, 94, 774802.

Schofield, J. W. (2001). Review of research on school desegregation's impact on elementary and secondary students. In J. A. Banks & C. A. M. Banks (Ed.), Handbook of research on multicultural education (pp. 597616). San Francisco: Jossey-Bass.

Schutz, A. (2006). Home is a prison in the global city: The tragic failure of school-based community engagement strategies. Review of Educational Research, 76, 691743.

Sheldon, S. B. (2003). Linking school-family-community partnerships in urban elementary schools to student achievement on state tests. Urban Review, 35, 149165.

Sheldon, S. B., & Epstein, J. L. (2002). Improving student behavior and school discipline with family and community involvement. Education and Urban Society, 35, 426.

Shujaa, M. J. (Ed.). (1996). Beyond desegregation: The politics of quality in African American schooling. Thousand Oaks, CA: Corwin Press.

Singer, J. D., & Willet, J. B. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. New York: Oxford University Press.

Sitkoff, H. (1993). The struggle for Black equality, 19541992 (Rev. ed.). New York: Hill and Wang.

Smrekar, C., & Goldring, E. (1999). School choice in urban America: Magnet schools and the pursuit of equity. New York: Teachers College Press.

Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA: Sage.

Sonleitner, N., & Wood, P. B. (1996). The effect of childhood interracial contact on antiblack prejudice. International Journal of Intercultural Relations, 20, 117.

South, S. J., Baumer, E. P., & Lutz, A. (2003). Interpreting community effects on youth educational attainment. Youth and Society, 35, 336.

Southworth, S., & Mickelson, R. A. (2007). The interactive effects of race, gender, and school composition on college track placement. Social Forces, 86, 497523.

Stearns, E., Moller, S., Blau, J., & Potochnick, S. (2007). Staying back and dropping out: The relationship between grade retention and school dropout. Sociology of Education, 80, 210240.

Stewart, E. B., Stewart, E. A., & Simons, R. L. (2007). The effect of neighborhood context on the college aspirations of African American adolescents. American Educational Research Journal, 44, 896919.

Swanson, C. B. (2004). Sketching a portrait of public high school graduation: Who graduates? Who doesn't? In G. Orfield (Ed.), Dropouts in America: Confronting the graduation rate crisis (pp. 1340). Cambridge, MA: Harvard Education Press.

Swidler, A. (1986). Culture in action: Symbols and strategies. American Sociological Review, 51, 273286.

Teachman, J. D., Paasch, K., & Carver K. (1996). School capital and dropping out of school. Journal of Marriage and the Family, 58, 773783.

Terenzini, P. T., Cabrera, A. F., Colbeck, C. L., Bjorklund, S. A., & Parente, J. M. (2001). Racial and ethnic diversity in the classroom: Does it promote student learning? Journal of Higher Education, 72, 509531.

Theiss, E. (1993, January 23). White says one-race schools OK. The Plain Dealer [Cleveland], p. 1A.

Walberg, H. J. (1984). Families as partners in educational productivity. Phi Delta Kappan, 65, 397400.

Walker, V. S. (1996). Their highest potential: An African American school community in the segregated South. Chapel Hill: University of North Carolina Press.

Wang, F., & Minor, W. W. (2002). Where the jobs are: Employment access and crime patterns in Cleveland. Annals of the Association of American Geographers, 92, 435450.

Watkins, W. H. (1996). Reclaiming historical visions of quality schooling: The legacy of early 20th-century Black intellectuals. In M. J. Shujaa (Ed.), Beyond desegregation: The politics of quality in African American schooling (pp. 528). Thousand Oaks, CA: Corwin Press.

Wells, A. S. (2002). The "consequences" of school desegregation: The mismatch between the research and the rationale. Hastings Constitutional Law Quarterly, 28, 771797.

Wells, A. S., & Crain, R. (1997). Crossing the color line: African-American students in White suburban high schools. New Haven, CT: Yale University Press.

Wells, A. S., Holme, J. J., Revilla, A. T., & Atanda, A. K. (2009). Both sides now: The story of school desegregation's graduates. Berkeley: University of California Press.

Wigfield, A., & Eccles, J. S. (Ed.). (2002). Development of achievement motivation. San Diego, CA: Academic Press.

Wilson, W. J. (1987). The truly disadvantaged: The inner city, the underclass, and public policy. Chicago: University of Chicago Press.

Wilson, W. J. (1996). When work disappears: The world of the new urban poor. New York: Vintage Books.

Wilson, W. J., & Taub, R. P. (2007). There goes the neighborhood: Racial, ethnic, and class tensions in four Chicago neighborhoods and their meaning for America. New York: Vintage Books.

Wright, S. P., Sanders, W. L., & Rivers, J. S. (2006). Measurement of academic growth of individual students toward variable and meaningful academic standards. In R. L. Lissitz (Ed.), Longitudinal and value added models of student performance (pp. 385399). Maple Grove, MN: JAM Press.

Yu, C. M., & Taylor, W. L. (1997). Difficult choices: Do magnet schools serve children in need? Washington, DC: Citizens' Commission on Civil Rights.


click to enlarge

click to enlarge

click to enlarge

click to enlarge

Cite This Article as: Teachers College Record Volume 112 Number 5, 2010, p. 1391-1442
https://www.tcrecord.org ID Number: 15676, Date Accessed: 10/21/2021 8:52:48 PM

Purchase Reprint Rights for this article or review
Article Tools
Related Articles

Related Discussion
Post a Comment | Read All

About the Author
  • Argun Saatcioglu
    University of Kansas
    E-mail Author
    ARGUN SAATCIOGLU is an assistant professor of educational leadership and policy studies and courtesy assistant professor of sociology at the University of Kansas. His research focuses on the sociology and politics of education, and the sociology of organizations. Currently, he is completing projects on the relationship between nonschool problems and school effectiveness in urban districts, the politics of discourse around school desegregation (with Jim C. Carl), regional and historical patterns in the development of “suburban advantage” in educational stratification in the United States (with John L. Rury), and the effects of school board social capital on academic and financial district outcomes (with Suzanne Moore and Gökçe Sargut).
Member Center
In Print
This Month's Issue