American Educational ResearchJournal
Fall 2004, Vol. 41, No. 3, pp. 605-631

Accountability in a Postdesegregation Era:
The Continuing Significance of Racial
Segregation in Florida'sSchools
KathrynM. Borman
Universityof SouthFlorida
TamelaMcNultyEitle
University of Miami
Deanna Michael
University of South Florida
David J. Eitle
Florida International University

ReginaldLee, LarryJohnson, DeirdreCobb-Roberts,
ShermanDorn, and BarbaraShircliffe
Universityof SouthFlorida
In the wake of both the end of court-ordered school desegregation and the
growing popularity of accountability as a mechanism to maximize student
achievement, the authors explore the association between racial segregation
and the percentage of students passing high-stakes tests in Florida's schools.
Results suggest that segregation matters in predicting school-level performance on the Florida ComprehensiveAssessment Testafter controlfor other
known andpurportedpredictors of standardized testperformance. Also, these
results suggest that neither recent efforts by the state of Florida to equalize the
funding of education nor current efforts involving high-stakes testing will
close the Black-White achievement gap without consideration of the racial
distribution of students across schools.
accountability, desegregation, education reform, segregation.
KEYwoRDS:

During the past 50 years, Brown v. Board of Education (1954) has shaped

debates about and strategies for achieving equity and excellence in
schooling, not only through court orders to desegregate and decades of court
supervision of school districts, but also because all other policy initiatives
designed to reduce educational inequality since Brown have been implemented in the context of school desegregation. Although many school districtshave now been declared "unitary,"1
a legal status indicating that a district
has removed all vestiges of past discrimination and is released from court

Borman et al.
oversight, the basic ideal represented in Brown-the right to equal educational opportunity-remains a part of the national debate on education and
a significant rationale for state and federal accountability policies today.
Under the force of Brown, a wide variety of race-specific efforts to provide Black students with educational opportunities equal to those of White students were implemented. These efforts included using race as a factor in

KATHRYN
M. BORMAN
is a Professor of Anthropology and Associate Director, David
C. Anchin Center, University of South Florida, 4202 E. Fowler Avenue, EDU160B,
Tampa, FL 33620-5650; e-mail borman@tempest.coedu.usfedu. She specializes in
educational reform and policy as well as evaluation studies.
TAMELA
MCNULTY
EITLEis an Assistant Professor, Department of Sociology,
University of Miami, P.O. Box 248162, Coral Gables, FL 33124-2030; e-mail teitle@
miami.edu. Her areas of specialization include segregation, school politics, and social
organization of education.
DEANNAMICHAEL
is an Assistant Professor, Department of Social and Psychological Foundations, University of South Florida, 140 Seventh Avenue South, COQ236B,
St. Petersburg, FL 33701; e-mail dmichael@stpt.usfedu. Her research area is the history of educational policy, and she is writing a book on President Jimmy Carter's
involvement in education policy.

DAVID
is an Assistant Professor, Department of CriminalJustice, Florida
J. EITLE
International University, 355A PCA Building, University Park, FL 33199; e-mail
eitled@fiu.edu. His research interests include exploring how the institutional and
organizational characteristicsof schools and their surrounding communities influence
student violence, misbehavior, and substance use on campus.
REGINALDLEE
is a Research Associate, David C. Anchin Center, University of South
Florida, 4202 E. Fowler Avenue, EDU162, Tampa, FL 33620-5650; e-mail Rlee@
tempest.coedu.usfedu. His research has focused on minority student representation
in special education and school reform.

LARRY
is an AssistantProfessor, College of Education, Universityof South
JOHNSON
Florida, 140 Seventh Avenue South, COQ 201, St. Petersburg, FL 33701; e-mail
He is writing a book on the development of community colleges.
ljohnson@stpt.usf.edu.
DEIRDRE
COBB-ROBERTS
is an Associate Professor, Department
of Psychologicaland
Social Foundations, University of South Florida, 4202 E. Fowler Avenue, EDU162,
Her areas of specialTampa, FL33620-5650; e-mail Cobb-Rob@tempest.coedu.usf.edu.
ization include the history of American higher education and multiculturaleducation.
SHERMAN
DORNis an Associate Professor, Department of Psychological and Social
Foundations, University of South Florida, 4202 E. Fowler Avenue, EDU162, Tampa,
FL33620-5650; e-mail
His research interests include educational
dorn@mail.usf.edu.
policy.
BARBARA
SHIRCLIFFE
is an Associate Professor, Departmentof Socialand Psychological Foundations, University of South Florida, 4202 E. Fowler Avenue, EDU162,
Tampa, FL 33620-5650; e-mail Shirclif@tempest.coedu.usf.edu. She specializes in
school segregation, desegregation, and community institutions.

606

Accountability in a Postdesegregation Era
student assignment, desegregating teaching staffs, busing students to achieve
racial balance, and creating magnet schools to diversify school populations
(Orfield & Eaton, 1996). One of the original justificationsfor school desegregation was the contention that Black students would have increased access to
educational resources and opportunities if they attended schools with White
students. It was argued that this access would have a substantive impact on
Black students' educational achievement, postsecondary educational opportunities, and future life chances. However, extant research investigating the
relationshipbetween desegregation and academic achievement among Blacks
has not provided unequivocal evidence in support of this fundamental argument. While studies that have focused on the long-term consequences of
school desegregation for Blacks have produced largely consistent results and
demonstrated positive relationships (see Wells & Crain, 1994, for a review of
earlier research), considerably more debate and inconsistency remain in the
findings gleaned from studies exploring the short-termeffects of school desegregation on achievement (see Braddock & Eitle, 2003, for a review of previous
research; see Crain& Mahard,1983, for a meta-analysis).
Consistent with the idea (forwarded in Brown and in other school desegregation cases) that access to predominantlyWhite institutionswould enhance
the social mobility and life chances of Blacks, research on long-term consequences has provided evidence that attending a desegregated school is associated with obtaining employment in desegregated occupational work groups
(Braddock & McPartland,1989) and attending desegregated colleges (Braddock, 1980). However, many studies (see Crain& Mahard,1983) on the shortterm outcomes of school desegregation have failed to reveal any positive
effects. These studies have focused almost exclusively on immediate effects on
standardized test scores, with little or no consideration of implementation or
context (Brown, 1996; Grant, 1996), although considerable evidence indicates
that resegregation within schools is a problem (Eyler, Cook, & Ward, 1983;
Kaeser, 1979; Mills & Bryan, 1976). Studies of school desegregation that have
considered second-generation segregation (allocation of opportunities within
schools), implementation (teacher attitudes, classroom organization), and
allocation of other educational resources (funding, quality of teachers) as
mediating factors have shown that resegregation may undermine the goals
of desegregation (Condron & Roscigno, 2003; Eitle & Eitle, in press; Epstein,
1985; Meier, Stewart, & England, 1989; Mickelson, 2001). Hence, the failure
of earlier studies to account for second-generation school segregation may
help explain the inconsistency of these studies in regardto finding support for
a positive association between school desegregation and short-termacademic
achievement among Black students. Nonetheless, questions remain as to
whether or not school desegregation will produce the academic achievement
gains among minority students assumed in the Brown decision.
Despite the unresolved questions regarding the short-termeffectiveness
of desegregation (in producing academic achievement gains among Black students) and whether racialintegrationwas actuallybeing achieved, Orfield and
Eaton (1996) argued that the Supreme Courtdecisions of the 1990s effectively
607

Borman et al.
moved the debate from questions regardinghow to best achieve school desegregation to how to dismantle it. In the case of school districtsthat have earned
"unitarystatus"from the courts, race-based policies for student assignment are
no longer a legal option for ensuring equal educational opportunity, even if
racialsegregation and isolation still characterizethe experience of minoritystudents in those districts.
Today, the focus of federal and state policymakers has clearly shifted from
concerns about racial segregation as an obstacle to achieving educational
equality to an emphasis on accountability as a mechanism to enhance academic achievement. The logic of the accountability mechanism is relatively
simple: Setting high standardsgoverning what students should know and be
able to do at each grade level and institutinghigh-stakes tests (with both incentives and disincentives for school districts,schools, and students to maximize
performance on these tests) are the best ways to improve academic achievement or to "leave no child behind." Thus, in the past decade, two majortrends
have emerged with regard to efforts to maximize academic achievement
among minorities: the waning of the importance of racial integration as a
mechanism for improving academic achievement and attainment among
Blacks and other minorities and the implementation of high-stakes testing as
a mechanism designed to improve academic achievement overall. Accountability through schoolwide testing at "strategic"grade levels, coupled with the
dismantling of desegregation, potentially obscures a core aspect of the Brown
decision: the idea that a separate education is inherently unequal.
The nexus between the dismantling of desegregation and the emphasis
on high-stakes testing as a means to maximize student achievement is the
focus of this article. In order to situate our inquiry in this nexus and establish
its particularcontributions, we briefly document the history of segregation/
desegregation in Florida and the series of policy steps that Floridahas undertaken during the past half century to enhance equity in educational outcomes.
This history demonstrates the difficulties faced by states and schools in reaching such goals.
We then examine the relationship between variations in school segregation and school-level performance on the FloridaComprehensive Assessment
Test (FCAT)in an attempt to understand whether such variationshelp explain
students' performance on Florida'shigh-stakes tests. Evidence that school segregation is associated with test performance in Floridawould raise considerable doubt about both the fundamentalfairness of using such accountability
mechanisms to target incentives and disincentives toward school districts,
schools, and students and the effectivenessof high-stakes testing mechanisms
in maximizing student academic achievement if the racial distribution of students across schools is not considered. Because Florida has been a national
leader in anticipating and implementing the provisions of the No Child Left
Behind Act, and has done so amid increasingly diverse enrollments, increasing school segregation, and the granting of unitary status to many Florida
districts, findings from this state may help inform national debates about
implementing accountability-based reform in the postdesegregation era.
608

The Florida Context
Priorto 1954, in Florida,like other southern states, there were dual systems of
education, one for White students and another for Blacks. In this section, we
discuss the desegregation of Florida districts, state-legislated educational
reform programs designed to achieve equity, the dismantling of desegregation
in Florida, and current accountability reforms. However, readers should be
aware that all of this occurred in a context of demographic shifts, including
rapid growth in the Hispanic2and minority populations. These demographic
changes resulted in a substantialincrease in the percentage of Floridastudents
attending predominantly minority schools. Nonetheless, Hispanic students
were not considered in the original desegregation cases and remained outside
the scope of Florida'sdesegregation orders despite the state's changing demographics and the 1973 Keyesv. Denver SchoolDistrictdecision.3 Hence, the history we outline largely focuses on the educational opportunities of Black
students in the state.
Avoiding Desegregation: The Minimum Foundation Program
and the FloridaPupil Assignment Law
Nine years before the 1954 Brown decision, FloridaGovernor MillardCaldwell
appointed a blue ribbon committee to suggest needed changes in the state's
educational policies. The legislation that came from the committee's report,
establishing the MinimumFoundation Program(MFP),was intended to equalize educational funding among districtsby distributingstate funds based on a
county's ability to support its schools (Colburn & Scher, 1980). With this bill,
the legislature set a minimum statewide property tax level to allow each of the
state's 67 counties to raise a certain percentage of the funds needed for each
teaching unit (White, 1981). Unlike past funding legislation, the MFPguaranteed funding to a school whether it served Black students or White students.
Local districts,whether wealthy or impoverished, were now able to hire qualified teachers and offer students a basic level of education.
The explicit reason given by Caldwell and his supporters for increasing
state support for education was to attractmore business to the state in the
post-World War II period. However, this increasing financial support for education coincided with the legal efforts of the National Association for the
Advancement of Colored People (NAACP)to pressure state governments to
desegregate (Colburn & Scher, 1980). By increasing funding for poorer school
districts and pressing for equitable distribution of state funds for Black and
White schools, many White Floridapoliticians hoped to prove that Black children would be provided an education equal to that provided to the state's
White children, even if in separate schools (Colburn, 1996). In fact, both
Florida Attorney General Richard Ervin and School Superintendent Thomas
Bailey believed that school segregation was constitutionally defensible given
the state's achievements in equalizing school funding. Under the MFP, the
racial gap in per-pupil expenditures had been reduced. In the 1945-1946
school year, expenses for all public schools in Floridaper pupil enrolled were
609

Borman et al.
$91.80 for White students and $52.61 for Black students (Biennial Report,
1946). By the 1954-1955 school year, expenses per pupil were $193.04 for
White students and $179.07 for Black students (Biennial
DurReport, 1955).
these
the
increased
from
of
funds
on
Black
children
57%
ing
years,
spent
those spent on White children to 93%. In a 1952 report, Bailey wrote of the
"rapidprogress" made toward equalizing Black and White schools, demonstratingFlorida's"good faith."Bailey's (1952) comments indicating that he did
not believe anyone would challenge the Florida constitution in view of this
substantial progress suggest that school officials planned to attempt to prove
that separate meant equal in Florida.
In 1955, on the same day the U.S. Supreme Court demanded that southern districts assume responsibility for desegregating schools, Florida Governor LeRoy Collins signed the Florida Pupil Assignment Law. This law, which
was strengthened in 1956, empowered county school boards to assign pupils
to schools on the basis of "sociological, psychological and like intangible
socio-scientific factors"(Hillsborough County School District, 1960). The law
essentially provided a loophole for schools wishing to avoid desegregation.
For example, the application for pupil special assignment in Hillsborough
County (Tampa), Florida,was six pages long and requested information from
the receiving principal about the "compatibilityof the student to the education program"and the "psychological effect of [the] prospective student on
your local community" (Hillsborough County School District, 1960). If the
principal indicated that the student would not "fitin,"the school board could
reject the request. Because many AfricanAmerican children were 2 to 3 years
behind their White counterparts, school boards could use a number of tests
to limit Black enrollment in White schools.
It was not until 1964 and the passage of the Civil Rights Act that school
districts in Florida, along with those in other southern states, faced increasing pressure to develop desegregation plans that actually resulted in school
desegregation. During the late 1960s and 1970s, federal courts ruled that
school districts had to desegregate and should use transportation to do so.
The Supreme Courtupheld the constitutionality of busing in the famous 1971
decision Swann v. Charlotte-Mecklenburg(Orfield, 1978).
Achieving Equity:School Desegregation and Florida'sAccountability Act
Despite the anti-busing campaigns led by Governor Claude Kirk, several
counties throughout Florida adopted court-ordered mandatory desegregation plans, and many involved busing. An early part of many of these desegregation plans was the dissolution of formerly all-Black schools and the
transferof students to White schools (Eitle, in press; Shircliffe,2002). In some
districts, this strategy was met with frustration and despondency by Black
parents, who felt that the burden of desegregation was being placed disproportionately on Black students (Shircliffe, 2002). By the mid-1980s, magnet
schools had become a popular method for achieving school desegregation in
Floridadistricts,and a few districts,such as HillsboroughCounty (Tampa),were
610

Accountability in a Postdesegregation Era
actually being heralded as models of successful desegregation (Clotfelter,
2004; Sinclair, 1978; U.S. Commission on Civil Rights, 1977). Other districts
such as Duval (Jacksonville),Miami-Dade,and Orange (Orlando) County preserved a number of majorityBlack schools that seemed untouched by school
desegregation efforts. The extent of school desegregation varied from district
to district, with many schools becoming more racially balanced and others
experiencing little change.
Simultaneous with the actual implementation of court-ordered desegregation in Florida,Governor Reubin Askew pushed to reform the MFPto close
some of the loopholes that allowed wealthy counties to avoid taxing themselves equitably. Askew (1973) argued that funding equality would make genuine educational opportunitypossible. The FloridaEducationFinance Program
was established to replace the MFP;this program restructuredthe funding formula based on full-time equivalency and local contributions, with the state
making up any difference from the minimum level of required funding (Citizens EducationCommittee, 1971). The new funding system also addressed the
costs of different grades, special programs, cost-of-living differences in each
county, and the cost of busing in rural areas (Citizens Education Committee,
1971). Through this complex formula, Governor Askew and the legislature
intended to address the varying economic conditions in the school districts.
This funding formula was accompanied by the Florida Accountability
Act, described by Askew as a means to assess schools and to "morefully predict the needs and prescribe the programs for each individual as an individual" (Askew, 1971). From 1972 through 1978, the legislature variously added
and removed provisions to the Florida Accountability Act, including requirements that preservice teachers take a state examination before they could be
certified and that principals and superintendents justify their budgets in
annual reports outlining the effectiveness of their expenditures (Herrington
& McDonald, 2001; Kimbrough, Alexander, & Wattenbarger, 1984). As part
of evaluating the effectiveness of expenditures, students were required to be
assessed in Grades 3, 5, 8, and 11 and to pass an additional test in order to
graduate from high school (Herrington & McDonald, 2001). Information on
the numbers of students who passed and failed these tests was included in
district and school annual reports to the public. This focus on assessments
and the graduation requirement reflected the national push in the 1970s for
schools to go "back to basics" by centering on reading, writing, and mathematics (White, 1981).
Under the Florida Accountability Act, the only group held accountable
in any concrete way was high school seniors. The students who failed the
test were most often impoverished and minority students who had not had
access to the same curricula, or even books, to which students in predominantly White and affluent schools had access (Tractenberg & Kahn, 1980).
This approach to accountability looked to the students to attain goals set by
the legislature and the State Department of Education, even though resources
to achieve those goals had been denied many students in the past. In 1979,
the Tampa Times reported that despite 8 years of integrated schooling in
611

Borman et al.
Hillsborough County, only 89% of Black 11th graders had passed the communication section of Florida's functional literacy test, and only 38% had
passed the math section. Conversely, 99% of White students (Asians and
American Indians were included as White) had passed the communication
section, and 81% had passed the math section (Brunais, 1979).
Once Florida implemented testing requirements, lawyers acting on
behalf of students who failed the graduation test filed a class-action lawsuit
challenging the denial of a high school diploma. In Debra P. v. Turlington
(1979), the plaintiffs charged that the test contained culturally biased material and that withholding a diploma on the basis of test scores discriminated
against minority students. The federal court concurred and charged the state
with providing a test based on information that students were actually taught.
The judge also decreed that the test could not be administered until students
who had been under the dual education system in Florida had graduated.
The court delayed the implementation of the test until the 1982-1983 school
year (Debra P. v. Turlington, 1979; Tractenberg & Kahn, 1980).
That year the administration of Governor Bob Graham developed the
Raise Achievement in Secondary Education program, which increased the
course requirements for high school graduation, significantly increased state
funding for education from kindergartento the university level, and instituted
a state-mandated curriculum and graduation testing (Herrington & McDonald, 2001). The mandate established in Debra P. v. Turlington that the state
test students only on material they had been taught was now supported by
the state-mandated curriculumand increased funding, and so all districts,rich
and poor, would offer the same basic curricula. This funding, however,
addressed only the basic curriculum.Wealthy districtscontinued to offer more
than the basics, while impoverished ones such as Gadsden (a historically
Black rural district) could provide only the minimum requirements.
Dismantling Desegregation and Instituting Accountability:
Unitary Status and the A+ Plan
The administration of Governor Lawton Chiles (1991-1998) continued the
state's commitment to testing as a means of accountability and incorporated
the business goals of America 2000 into educational reform.At the close of the
20th century, Florida'spopulation had grown to more than 15 million, making
it the fourth most populous state in the union. Many analysts attribute the
growth of national and internationalmarkets as the catalyst for Sunbelt development and related educational policies aimed at creating "good"workers for
Florida'sburgeoning industrialeconomy (Cobb, 1990). The need for an educated workforce now became preeminent. To ensure the production of welltrained workers, Florida'sschools, it was reasoned, must be held accountable
for the quality of their graduates (Florida Commission on Education Reform
and Accountability, 1994).
The educational reform package, Blueprint 2000, answered this need by
decentralizing school management while ensuring that accountability contin612

Accountability in a Postdesegregation Era
ued to be focused on the school and the student. Decentralizationwould give
control of education back to the local districts,to "encourage innovation, risk
taking, and successful student learning,"while the state set standardsand coordinated the comprehensive system and assessment testing (FloridaDepartment
of Education, 1998). Localcontrol consisted of assigning school personnel, parents, and community members responsibility for schools' budgets and programs. These "stakeholders"would sit on school advisory committees and
make decisions about how to address the needs of the students and the community. They would also design programs to achieve the goals established by
the state and be held accountable for student progress, which would be measured by the FCAT,the assessment package for Blueprint 2000 that was based
on statewide tests to be developed by the department of education. Rewards
and sanctions for schools were to be determined by the state. If schools made
adequate progress, they could receive financial rewards, special school designations, and publicity.The state developed the FloridaSchool RecognitionProgram to reward schools that either maintained high levels of achievement or
showed exemplary improvement. If a school did not show adequate progress
after 3 years, it would be subject to actions by the State Board of Education,
the governor, and the cabinet (Florida Department of Education, 1998).
In 1995, the commissioner of education, Frank Brogan, publicized lists
of the lowest performing schools (Hegarty, 1995). This publication of school
names served as the beginning of the implementation of sanctions against
schools that did not make adequate progress. These sanctions did not take
the economic or social conditions of schools into account. Schools located in
high poverty areas were expected to perform as well as those located in privileged, economically thrivingcommunities. By the end of the 1990s, Blueprint
2000 had become the A+ Plan under Governor Jeb Bush. The supposed
premise of the A+ Plan is that every child can learn and no child should be
left behind. This premise mirrorsthat of the more recent reauthorization of
the Elementaryand Secondary Education Act ("No Child LeftBehind") passed
by Congress and signed into law by President George W. Bush in 2001.
A major aspect of Governor Bush's A+ Plan is ensuring the accountability of the state's public education sector. The A+ Plan uses student performance on the criterion-referenced portion of the FCAT in reading and
mathematics to determine the overall performance of schools and rank these
schools using an A through F grading system. In 1999-2000, the second year
of the state's implementation of the plan, school grades were based on the
performance of students in Grade 4, 5, 8, or 10. Schools showing improvement to A status were provided incentives in the form of stipends awarded
to teachers and principals, while beleaguered D and F schools were offered
little real assistance. More recently, this lack of any real assistance for failing
schools has become even more evident and disturbing to educators. In 2002,
the Florida Legislatureappropriated more than $122 million from lottery proceeds to provide high-performing schools $100 per student. While many
schools accepted the funds, some A schools refused them, dismayed that their
colleagues in F schools received no resources to improve student performance.
613

Borman et al.
To address this criticism, the Bush administration elected to send out a call
for volunteers to serve as tutors and coaches for F schools as an approach
to assisting these schools, most of which were located in resource-poor districts facing many complex social and economic challenges.
Despite the aforementioned court orders and substantialdecreases in the
segregation of Black students during the 1980s, in the 1990s Florida'sschools
became increasinglymore segregated, particularlyin the largesturban districts.
Six of the seven largest school districtsin Floridaexperienced increases in levels of segregation, with the Hillsborough County school district,once heralded
as a model of successful desegregation, seeing the largest increases in school
segregation during the past decade (Eitle, 2003). Even under these conditions
of increasing segregation, Florida districts have sought and obtained unitary
status and abandoned mandatory programs designed to achieve district-wide
desegregation. Federal courts have viewed patterns of increasing school segregation resulting from factors such as changing demographics due to the
growth of White suburbs, the expansion of city ghettos, and immigration as
beyond the control of local school boards. Therefore, as counties such as
Hillsborough moved from mandatory to voluntary desegregation plans during the 1990s, school officials were not held accountable for resegregation
unless plaintiffs could prove they caused it. Nevertheless, racial segregation
concentrates poverty in majorityBlack schools while reducing poverty in predominantly White schools (Orfield & Eaton, 1996). This reality probably has
consequences for educational equity.

Summary
The state of Floridahas implemented myriad policies, including school desegregation, in an attempt to increase the academic achievement of Florida'sstudents generally and to reduce earlier inequities in both opportunities and
achievement for Florida's Black students. However, the emphasis on and
implementation of accountability plans have led to a virtual neglect of the
issue of race and achievement at a time when school segregation among
Florida schools is accelerating. Given the incentives and disincentives fundamental to such accountability mechanisms, we argue that an examination of
the possible influence of racialsegregation on school-level FCATperformance
is not only timely but of great importance because of the high-stakes implications associated with performance on the FCAT.Thus, our primaryresearch
question was whether school-level variationin racialsegregation is associated
with school-level performance on the FCAT.
To answer this question, we examine the relationship between two
measures of segregation (racial composition and racial balance) and the percentage of students passing the FCAT while controlling for other salient
school-level predictors of standardized test performance. Specifically, we
consider whether or not per-pupil expenditures, an important tool used by
the state of Florida (beyond integration efforts) to diminish the gap in educational achievement between Black and White students, are an important
614

Accountability in a Postdesegregation Era
predictor of FCAT performance in the context of racial composition and
racial balance variables. We assert that evidence that per-pupil expenditures
are not associated with FCATperformance, coupled with evidence that racial
segregation measures are significant predictors of test performance, should
raise concerns about both the fairness and the effectiveness of using highstakes testing without consideration of the role of segregation in predicting
test performance differences.

Method
Data
The data for this study were obtained from the Florida Department of Education. The state of Florida is composed of 67 public school districts, one in
each county, serving more than 2.3 million students. Significantchanges have
occurred in the state's student population during the past decade. There was
a 26%increase in the student population from 1991-2000, and Black and Hispanic student enrollments increased by 31% and 89%, respectively, during
this period. While these changes have not been uniform across the state, they
do affect a large proportion of Florida's students. For example, in 1991 only
17%of public school students in Florida attended school in a districtthat was
at least 50%minority;by the year 2000, however, 41%of Floridapublic school
students were enrolled in such districts.The Florida SchoolsIndicators Report
and the SchoolAdvisoryCouncil Reportfor the academic year 1999-2000 were
the sources for the data analyzed here.4 The Florida Department of Education collects comprehensive information from each school in each district
at scheduled times during the school year. In the present analyses, data for
elementary, middle, and high schools were assessed separately.
Dependent Variables
On the basis of scaled FCATscores, students are classified into one of five
achievement levels for math and reading performance.5 Our two dependent
variables captured the percentages of students in a school (among those who
were tested during 4th, 5th, 8th, and 10th grades) achieving Level 3, 4, or 5
on the math and reading portions of the test. Classification at one of these
levels represents a passing score and demonstrates at least partial success
with the most challenging content articulatedin the Sunshine State Standards
(Florida State Board of Education, 1996).
Independent Variables
Segregation Measures
We used two different measures to capture racial segregation at the school
level. Racial composition, a categorical measure of the relative percentage of
a school's students who were Black ("percentage Black"), comprised four
615

Borman et al.
categories: (a) more than 90% Black, (b) 50% to 90% Black, (c) 15%to 50%
Black, and (d) less than 15%Black (the omitted category). While percentage
of the student body that is Black is often measured as an interval-level variable in studies investigating the impact of school desegregation or segregation (Bankston & Caldas, 1996; Crain & Mahard, 1983; Hoxby, 2002), there
is recent evidence that there are important tipping points in which differences in the racial composition of the student body have meaningful influences on the academic environment of the school (Brown, 1999). However,
additional analyses (available from the authors upon request) in which percentage Black was substituted for the categorical variable produced results
similar to those reported here.
In addition to the racial composition measure, we constructed a threecategory measure of racial balance in which each school was classified as
Black segregated, integrated, or White (or non-Black) segregated. A school
was defined as a Black segregated school if the Black enrollment exceeded
by more than 15%the average percentage Black enrollment in the school district.This measure, based on the definition of an integrated school as one that
is within 15%of the Black enrollment in the entire school district,was used
in many of the court cases in Florida and elsewhere as representing a legal
definition of when a school was desegregated (Mickelson, 2001; U.S. Commission on Civil Rights, 1971; Valencia, 2000). A school was defined as White
segregated6 if its enrollment was 15% (or more) below the district average
percentage Black, whereas a school was defined as integrated if the percentage of students who were Black fell within 15%(plus or minus) of the district
percentage Black. This is an important additional component (beyond racial
composition) for understanding how segregation may influence school-level
standardized test performance, because districts that have relatively large
Black enrollments will have many schools with largely Black student bodies,
and these schools will not be considered Black segregated if they are racially
balanced. Finally,we analyzed the racial composition measures and the racial
balance measures in separate models because of the confounding influence
of these two variables on school-level test performance.7
OtherMeasures
We evaluated the importance of the segregation measures in terms of predicting the dependent variables in the context of assumed indicators of
school-level standardized test performance. We included six such measures:
percentage of students who are Hispanic ("percentage Hispanic");per-pupil
expenditures; instructionalquality; percentage of students who enroll for the
first time, reenroll (return to the school), or withdraw during the school year,
divided by a given school's total enrollment ("percentage mobility");percentage of students eligible for the free lunch program("percentagepoverty");and
average class size. Per-pupil expenditures are considered to capture resource
differences and were included because funding equalization has been a key
focus of educational reform in Florida during the past half century, although
616

Accountability in a Postdesegregation Era
claims regarding any positive effects of expenditures on achievement are
highly contested in the literature(Nyhan & Alkadry, 1999). As a result of the
multicollinearitythat exists among the three available indicators of per-pupil
expenditures (i.e., per-pupil dollar expenditures for regular,at-risk,and exceptional educational programs), we conducted a principal-components analysis
to extract a single factor representing such expenditures. Higher scores on this
factor represented higher levels of resources invested per pupil.
Percentage Hispanic is an important variable in the state of Florida
(a) given that in many school districts, particularlyin southern Florida, more
than 20% of the student body is Hispanic and (b) given previous research
indicating that percentage Hispanic is inversely associated with standardized
test performance (Natriello & Pallas, 2001; Valencia, 2000; Valenzuela, 1999).
Instructionalquality was an index constructed by summing the z-score transformations of two items: percentage of teachers with a master's degree and
percentage of school staff who are specialists. Higher scores on this index
represented greater instructional quality. As mentioned, percentage mobility
was measured as the number of students enrolling for the first time, reenrolling (returning to the school), or withdrawing during the school year
divided by total enrollment. Increases in student mobility should be associated with poorer school-level performance on standardized tests, since such
movement reflects a disruption in the learning process for students (Entwisle,
Alexander, & Olson, 1997).
Percentage poverty, measured as the percentage of students eligible for
the free lunch program, is well established as an important predictor of student achievement. Unfortunately, this measure was available for consideration only in the elementary and middle school analyses. Finally, average class
size was included because of the inconsistency in past research regardingthe
relationship between class size and academic achievement (Finn & Achilles,
1990; Hanushek, 1997; Nyhan & Alkadry, 1999) and because the state of
Florida recently passed a state referendum dictating that class sizes in Florida
public schools be reduced to ensure smaller classes. In the elementary school
analyses, class size was measured as average class size in a given school. Class
sizes in the middle and high school analyses were measured as the average
number of students in language arts and math classes as reported by each
school to the state of Florida. Such measures are more accurate indicators of
class size than typical student/teacher ratios.
Analytic Strategy
We evaluated the role of school segregation in predicting school-level variations in the percentage of students passing the FCATin two stages. First,
we examined mean differences in the percentage of students passing the
FCAT across three distinct groups of schools-Black segregated schools,
integrated schools, and White segregated schools---to establish whether segregated schools have significantly lower percentages of students passing the
FCATthan integrated and predominantly White schools. Second, we used
617

Borman et al.
multivariate models to evaluate the influence of segregation in the context
of other important predictors of aggregate-level measures of student test performance. Schools were clustered within larger units (i.e., school districts),
and thus standard ordinary least squares regression techniques were inappropriate because the assumption that the error terms were uncorrelated
across observations was probably violated. To correct for the effect of clustering, we used a procedure available in Stata 7.0 that adjusts for clustering
of units within larger contexts. In the Stata estimator, a Huber-White correction is used for standard errors. Finally, variance inflation factors calculated
for each of the models analyzed suggested no significant concerns about
multicollinearity;none of these factors approached 4, and the mean inflation
factor for each model analyzed was less than 2.

Results
Tables 1, 2, and 3 present selected variable means for Black segregated, integrated, and White segregated elementary, middle, and high schools. As
expected, significantly lower percentages of students in Black segregated
Table 1

MeanScores on Selected Variables:Segregated Black,
Segregated White,and IntegratedElementarySchools
Black
segregated
schools
(B; n = 309)

Integrated
schools
(I; n = 920)

White
segregated
schools
(W; n = 318)

passing FCATreading
%O/

31.85

54.63

55.83

O/passing FCATmath

27.25

46.38

51.25

Per-pupilexpenditures

.45

-.09

-.18

Qualityof instruction

.03

-.12

.30

Average classroom size

22.75

23.24

24.65

mobility
%/o

44.13

33.51

29.28

% poverty

81.29

50.15

43.40

Type of Variable

ap < .05 (two-tailedtest).

618

Scheff6 test
of significant
differencesa
B vs. I
B vs. W
B vs. I
B vs. W
I vs. W
B vs. I
B vs. W
I vs. W
B vs. I
B vs. W
I vs. W
B vs. W
I vs. W
B vs. I
B vs. W
I vs. W
B vs. I
B vs. W
I vs. W

Table 2

MeanScores on Selected Variables:Segregated Black,
Segregated White,and IntegratedMiddleSchools
Black
segregated
schools
(B; n = 80)

Integrated
schools
(I; n = 367)

White
segregated
schools
(W; n = 66)

%passing FCATreading

23.68

42.11

43.29

%passing FCATmath

32.84

54.55

54.91

Per-pupilexpenditures

.33

-.02

-.26

Quality of instruction

.00

-.07

.35

Average classroom size

25.93

26.13

29.58

%mobility

39.37

29.65

26.67

%poverty

67.34

42.03

40.55

Type of Variable

Scheff6test
of significant
differencesa
B vs. I
B vs. W
B vs. I
B vs. W
B vs. I
B vs. W
I vs. W
B vs. W
I vs. W
B vs. W
I vs. W
B vs. I
B vs. W
B vs. I
B vs. W

ap< .05 (two-tailedtest).

Table 3

MeanScores on Selected Variables:Segregated Black,
Segregated White,and IntegratedHighSchools
Black
segregated
schools
(B; n = 42)

Integrated
schools
(I; n = 289)

White
segregated
schools
(W; n = 37)

%passing FCATreading

18.48

28.38

31.97

%passing FCATmath

31.95

53.06

54.81

Per-pupilexpenditures

.21

-.02

-.12

Qualityof instruction
Average classroom size
%mobility

.14
27.83
43.66

-.05
25.83
30.22

.29
29.87
25.04

Type of Variable

Scheffe test
of significant
differencesa
B vs. I
B vs. W
B vs. I
B vs. W
I vs. W
B vs. I
B vs. W
I vs. W
I vs. W
B vs. I
B vs. W

Note. The Floridadepartmentof Educationdoes not reportpovertystatisticsfor high schools.
ap< .05 (two-tailedtest).

619

Borman et al.
schools passed the FCATreading and math sections than students in either
of the other two groups of schools (integrated and White segregated schools).
For instance, it can be seen in Table 1 that, on average, only 31.85%of fifthgrade students at Black segregated schools passed the reading portion of the
FCAT,as compared with 54.63%of fifth-grade students at integrated schools
and 55.83% of fifth graders at White segregated schools. This pattern was
reproduced across middle schools (Table 2) and high schools (Table 3) and
was repeated when mean differences in the percentages of students passing
the math portion of the FCATwere examined.
However, in only two of the six possible comparisons between the mean
FCAT test performance of students in integrated versus White segregated
schools were the differences statisticallysignificant (p < .05), and even in these
comparisons, the gaps between the percentages of students passing the FCAT
at integrated and White segregated schools were substantially smaller (i.e.,
approximately 2%and 5%)than the gaps between Black segregated and integrated schools' FCATperformances (i.e., approximately 21%and 19%,respectively). This pattern of differences reveals that mean FCATtest performance
differed only between Black segregated schools and integrated schools or
between Black segregated and White segregated schools. Schools that were
integrated were roughly comparable in their FCATperformance to schools
with relatively few Black students (i.e., White segregated schools).
When we consider other comparisons, of little surprise is the finding
that Black segregated schools had significantly more impoverished students
than either integrated or White segregated schools. Indeed, at the elementary school level, the distinction was somewhat surprising in its magnitude.
Results showed that more than 80% of students in Black segregated schools
were impoverished, as compared with 50%of students in integrated schools
and 43% of students in White segregated schools. Clearly, the concentration
of' poverty in segregated schools is an important dimension in regard to
understanding differences in standardized test performance (such as performance on the FCAT).We explored this connection furtherin the multivariate
analyses.
Another notable finding is that Black segregated schools evidenced significantly higher per-pupil expenditures than either integrated or White segregated schools, reflecting the aforementioned efforts by Florida to reduce
educational inequities, in addition to Title I funding and funding related to
special needs students. However, average scores capturing quality of instruction suggested that instructionalquality at White segregated middle and high
schools was greater than that at either integrated or Black segregated middle
and high schools, a factor that may have accounted for some of the disparities in FCATperformance. Not surprising, Black segregated schools exhibited
significantly higher percentages of student mobility than either integrated or
White segregated schools, and this may also have contributed to the lower
average FCATperformance among students in these schools. Finally, White
segregated middle and high schools appeared to have largersizes classes than
either Black segregated or integrated middle and high schools. This finding
620

Accountability in a Postdesegregation Era
may again reflect the policies enacted to reduce educational inequities and
the availabilityof federal Title I funds for exceptional students.
Table 4 presents the results of the regression analyses predicting the
percentage of fifth-grade students passing the math portion of the FCATand
the percentage of fourth-grade students passing the reading portion (in
1999-2000, only fourth graders took the reading portion of the FCAT,and
only fifth graders were tested in math). In both the racial balance and racial
Table 4

OrdinaryLeast Squares Regression of Percentage
of ElementarySchool Students Passing the Math
and Reading Portions of the FCAT(n = 1,547)
Type of variable
Control
Per-pupilexpenditures
Instructionalquality
%mobility
% Hispanic
Average class size
% poverty
Segregationa
Racialbalance
Black segregated school
Integratedschool

4th grade:
balance

4th grade:
composition

5th grade:
balance

5th grade:
composition

-1.62*
(0.78)
2.00**
(0.70)
-0.02
(0.02)
-0.12***
(0.03)
-0.10
(0.06)
-0.44***
(0.02)

-1.32
(0.67)
2.30**
(0.64)
-0.03
(0.02)
-.17***
(0.03)
-0.05
(0.04)
-0.38***
(0.03)

0.13
(0.65)
.80
(0.78)
-0.00
(0.02)
-0.01
(0.03)
-0.02
(0.05)
-0.49***
(0.03)

0.19
(0.63)
1.02
(0.84)
-0.00
(0.02)
-0.02
(0.03)
0.00
(0.04)
-0.46***
(0.03)

-8.27***
(1.32)
0.21
(0.87)

Racialcomposition
More than 90%Black
Black
50/o%-90%
Black
15%/o50%
Constant
R2

81.11***
(1.49)
.716

-5.34**
(1.88)
-1.37
(1.32)
-18.19***
(1.80)
-12.65***
(1.40)
-3.88***
(0.86)
80.06***
(1.18)
.735

73.20***
(1.58)
.612

-9.42***
(1.68)
-6.51***
(1.41)
-2.88**
(0.93)
71.85***
(1.28)
.618

Note. Robust standard errors are in parentheses.
aDefault categories for the segregation measures were White segregated schools (for the racial
balance models) and less than 15%Black (for the racial composition models).
*p < .05; **p< .01; ***p< .001 (two-tailed tests).

621

Borman et al.
composition models predicting the percentage of students passing the reading portion of the FCATat each school, three of the nonsegregation variables
were found to be significant predictors of the dependent variable: instructional quality, percentage Hispanic, and percentage poverty. In schools with
a higher percentage of poor students, a higher percentage of Hispanic students, and lower instructional quality, lower percentages of students passed
the reading portion of the FCAT.Furthermore,spending exhibited an inverse
association with the dependent variable in the racial balance model, whereas
it failed to reach statistical significance in the racial composition model.
In regard to the segregation measures, the results of the model including
the measures of racial balance are presented in the first column of Table 4.
This model reveals that significantly lower percentages of students in Black
segregated schools than in White segregated schools passed the fourth-grade
reading FCAT. However, there were no significant differences between
integrated schools and White segregated schools in terms of percentage of
students passing the reading portion of the FCAT.
With regard to the racial composition of the schools, the findings were
significant and in the expected direction. In comparison with schools with
relatively few Black students enrolled (i.e., fewer than 15% of students),
schools with a greater proportion of Black students enrolled evidenced a significantly lower percentage of fourth graders passing the FCATreading test.
For example, in a school with less than 15%Black enrollment (and for which
other variables exhibited values close to their overall means), approximately
54% of the student population would be expected to pass the reading section of the FCAT.However, in a similar school with more than 90% Black
enrollment, only 36%of the students would be expected to pass the reading
section of this test. Schools with larger percentages of Black students performed poorer, independent of other predictors of percentage of students
passing the reading FCAT.
In the model predicting fifth-grade students' performance on the math
portion of the FCAT,only one control variable-percentage poverty-was
found to be significant. All else equal, as percentage poverty increased, the
percentage of students passing the math portion of the FCATdecreased. The
findings for the model including measures of racial balance were parallel to
those reported for the model predicting fourth-grade students' reading FCAT
performance. Approximately 5% fewer students passed the math portion of
the FCATin Black segregated schools than in White segregated schools. However, after control for other factors, integrated schools and White segregated
schools did not vary significantly in their average performance.
Consistent with the predictions regarding percentages of students passing the reading FCAT,the racial composition measures were again predictive of the percentages of students passing the math portion of the FCAT.
Significantly lower proportions of students in schools with high percentages
of Black students than in schools with lower percentages of Black students
passed the math portion of the FCAT.
622

Accountability in a Postdesegregation Era
Results from the models examining predictors of middle school students'
performance on the FCATare shown in Table 5. Analyses revealed that two
control variables,percentage mobility and percentage poverty, were negatively
associated with school-level math and reading test performance across all four
models. In schools with greater student mobility and a higher percentage of
students in poverty, fewer students passed the math and reading portions of
the FCAT.In addition, per-pupil expenditures were negatively associated with
Table 5

OrdinaryLeast Squares Regression of Percentage of MiddleSchool
Students Passing the Mathand Reading Portions of the FCAT(n = 513)
Type of variable
Control
Per-pupilexpenditures
Instructionalquality
%mobility
%Hispanic
Average class size
%poverty
Segregationa
Racialbalance
Black segregated school
Integratedschool

8th grade:
balance

8th grade:
composition

8th grade:
balance

-1.80*
(0.88)
1.73
(1.12)
-0.24***
(0.04)
-0.01
(0.04)
0.13
(0.07)
-0.47***
(0.04)

-1.52
(0.87)
1.79
(1.02)
-0.27***
(0.04)
-0.06*
(0.03)
0.17
(0.08)
-0.41***
(0.05)

-2.15*
(0.90)
1.60
(1.02)
-0.25***
(0.06)
0.00
(0.05)
-0.37*
(0.18)
-0.48***
(0.05)

50/6-90%Black
150/6-50%Black

Constant
R2

64.36***
(2.34)
.728

-1.75*
(0.81)
1.63
(0.90)
-0.29***
(0.06)
-0.05
(0.05)
0.30
(0.19)
-0.41***
(0.07)

-5.15
(2.68)
1.25
(1.58)

-2.30
(2.29)
1.45
(1.27)

Racialcomposition
Morethan 90%Black

8th grade:
composition

-11.44**
(3.19)
-8.22**
(2.54)
-0.89
(1.17)
64.04***
(2.91)
.741

90.24***
(5.15)
.671

-13.72**
(4.19)
-12.40**
(3.80)
-1.54
(1.48)
89.32***
(5.30)
.687

Note. Robust standard errors are in parentheses.
aDefault categories for the segregation measures were White segregated schools (for the racial
balance models) and less than 15%Black (for the racial composition models).
*p < .05; **p< .01; ***p< .001 (two-tailed tests).

623

Borman et al.
eighth graders'math performance on the FCATin both the racial balance and
composition models, and such spending was negatively associated with eighth
graders'reading performance in the racialbalance model. While the spending
coefficient did not reach statistical significance in the model predicting percentage of students passing the reading portion of the FCATwhen the racial
composition measures were included, this coefficient did approach statistical
significance (p < .10).
In the racial composition models, the same pattern of findings emerged
as in the analyses of elementary schools focusing on racial composition: As
percentage of Black students increased, there was an expected decrease in the
percentage of students passing the FCAT(both the math and reading portions).
One caveat that should be noted was that the difference between the effect of
a school being between 15% and 50% Black and the default category-less
than 15%Black-failed to reach statisticalsignificance, suggesting that there
was no difference between average performance on the FCATin these two
categories of for schools. This implies that there is a potential threshold racial
composition effect that leads to significantly lower percentages of students
passing the FCAT,ratherthan a simple linear relationshipbetween racialcomposition and school-level test performance. However, the racial balance models revealed that being enrolled in a Black segregated or integrated school (vs.
a White segregated school) was not a significant predictorof either dependent
variable at the middle school level. That is, whether or not a school was racially
balanced relativeto the district'sracialcomposition did not appear to have predictive utility.
A similar portraitof the predictors of high school students' performance
on the FCATis offered by the findings shown in Table 6. Three of the control
variables were found to be statisticallysignificant (p < .05) in the high school
models: instructionalquality, percentage Hispanic, and percentage mobility.
In schools with greater levels of instructionalquality, less mobility, and fewer
Hispanic students, higher percentages of students passed the math and reading portions of the FCATthan in their counterpartschools. Surprisingly,average class size was found to be a significant predictor of school-level FCAT
math test performance (in the racial composition model), but in a direction
opposite from that predicted. That is, increases in class size were associated
with increases in the percentage of students passing the math portion of the
FCAT.Finally, it must be noted that percentage poverty was not available for
inclusion in the high school models. Thus, to the extent that racial composition, racial balance, and percentage poverty are interrelated,the magnitudes
of the race variable coefficients were probably influenced by the exclusion of
the poverty measure at the high school level.
The pattern of findings regarding the association between school racial
composition and the dependent variable at the high school level was similar
to that found in the middle school analyses. Racial composition was negatively associated with the percentage of students passing both the math and
reading portions of the FCAT.However, even in schools with between 15%
and 50% Black students, significantly lower percentages of students passed
624

Table 6

OrdinaryLeast Squares Regression of Percentage of HighSchool
Students Passing the Mathand Reading Portions of the FCAT(n = 368)
Type of variable
Control
Per-pupilexpenditures
Instructionalquality
%mobility
% Hispanic
Average class size
Segregationa
Racialbalance
Black segregated school
Integratedschool

10th grade:
balance

0.33
(1.38)
5.50***
(0.99)
-.26***
(0.07)
-.31***

(0.03)

(0.06)

(0.04)

0.23
(0.12)

-0.03
(0.21)

0.40*
(0.17)

1.67
(1.15)
6.62***
(0.84)
-0.23***
(0.06)
-0.25***

(0.04)
0.05
(0.13)

-21.49***
(4.77)
-4.26
(2.95)

-13.04***
(3.41)
-5.41*
(2.37)

500/-90% Black

15/o50% Black

R2

-2.61
(2.27)
4.71***
(1.06)
-0.28***
(0.07)
-0.27***

0.34
(1.48)
6.12***
(0.91)
-0.24***
(0.05)
-0.25***

Racialcomposition
Morethan 90%Black

Constant

10th grade: 10th grade: 10th grade:
balance
composition
composition

42.67***
(5.49)
.403

-24.39***
(1.79)
-13.66***
(2.09)
-2.36*
(0.93)
34.63***
(4.19)
.484

69.24***
(7.50)
.425

-43.05***
(2.53)
-27.26***
(2.39)
-5.76***
(1.36)
58.49***
(1.36)
.579

Note. Robust standard errors are in parentheses.The Florida department of Education does not
report poverty statistics for high schools.
aDefault categories for the segregation measures were White segregated schools (for the racial
balance models) and less than 15%Black (for the racial composition models).
*p < .05; ***p< .001 (two-tailed tests).

the FCATthan in schools with relatively few Black students (i.e., fewer than
15%).With regard to the racial balance variables, these analyses revealed that
a much lower percentage of students passed the FCATin Black segregated
schools than in White segregated schools. The difference between integrated
schools and White segregated schools failed to reach statisticalsignificance in
terms of predicting the percentage of 10th graders passing the math portion
of the FCAT.In integrated schools, however, a lower percentage of students
passed the reading portion of the test than in White segregated schools.
In summary, two important patterns are revealed in these analyses. First,
the racial composition of the student body is an important predictor of the
625

Borman et al.
percentage of students passing the FCATmath and reading tests, regardless
of the grade at which testing takes place (4th-5th grade, 8th grade, or 10th
grade). Increases in the percentage of Black students enrolled in a school are
associated with fewer students passing the FCAT, independent of other
important predictors such as instructionalquality, average class size, and perpupil expenditures. Second, the racial balance (or imbalance) of schools is
associated with significantly lower percentages of students passing the FCAT
reading and math tests at the elementary and high school levels. Black segregated schools perform significantly less well than White segregated schools
at both the elementary and high school levels. However, in three of the four
models the difference between the effect of being enrolled in an integrated
school and that of being enrolled in a White segregated school was not statistically significant, suggesting that the relationship between the racial balance measures and the dependent variable is not simply an artifact of the
racial composition of schools. Relative racial balance appears to matter only
if schools are Black segregated-students in schools with relatively racially
balanced student populations and students in schools with relatively low percentages of Black students appear to perform in a relatively similar manner
on the FCAT,after control for other salient predictors of school-level standardized test performance.

Conclusion
In the 50 years since the landmark decision in the Brown case, the legacy of
this decision continues to structure the debate about how to best close the
achievement gap between Black and White students. Unfortunately, there is
growing evidence that the logic and spirit of Brown-that separate is inherently unequal-are no longer of central import in the educational policies
currently being implemented to raise academic achievement standards generally and to close the race gap specifically. We have documented the history of one state's efforts to close the Black-White achievement gap in the
decades since Brown. Consistent with national trends, Florida has moved
away from racial integration as a solution to the Black-White achievement
gap problem and has now turned its attention to an accountability mechanism complete with disincentives (and the withholding of incentives) for
schools whose student bodies do not meet performance benchmarks.
The results of our analyses suggest great caution in abandoning school
integrationas a mechanism to close the racialgap. Controllingfor other known
and argued predictors of school-level standardized test performance, including per-pupil expenditures and percentage poverty, our analyses revealed that
race still matters:Both the racialcomposition of a school and whether a school
was Black segregated (relative to the school district'sracial composition) predicted the percentage of students passing the FCAT.Indeed, the fact that students in integrated schools did not perform significantly worse on the FCAT
test than students in White segregated schools can be interpretedas primafacie
evidence that integratingBlack and White students makes a difference in terms
626

Accountability in a Postdesegregation Era
of school-level performance on high-stakes tests. While the models including
measures of racial balance did not explain as much of the variation in the
dependent variable as the racialcomposition variables, the overall picture that
these analyses paint is one wherein segregated schools can be viewed as
institutions of concentrated disadvantage (Wilson, 1987). Such schools must
grapple with several adverse factors that simultaneously limit the academic
achievement of their student bodies. Furthermore,our analyses suggest that
policies that attemptto resolve the achievement gap by funding equity or classroom size changes may not be successful if they do not accept the premise of
Brown-that integration is fundamental to ensuring educational equality.
While our study provides an important examination of how the distribution of race influences school-level high-stakes test performance, it involves
some obvious limitations. For example, our analysis was cross sectional in
nature. Arguably, a superior test would consist of a longitudinal analysis
attempting to determine whether changes in the predictor variables predict
changes in the percentages of students passing the FCATover time. It is our
hope that, as the FCATcontinues to be employed as a form of high-stakes
testing over the next few years, additional analyses can be conducted to better assess the dynamic aspects of the relationships examined in the present
study. On a related note, we were able to evaluate only a relatively static state
of the outcomes of extant educational policies in Florida,namely de facto segregation (court-ordered segregation policies had concluded) and educational
equity policies and their relationship to school-level FCATperformance. A
superior test would be to evaluate how specific changes in Florida's educational equity policies influence changes in the dependent variable over time.
However, the FCAThas not been used for a sufficiently long time period to
allow such analyses. Furthermore,we examined only school-level factors and
their relationship with the dependent variable; we did not employ measures
of local community or school district measures to evaluate variations in the
dependent variable.8Furtheranalyses including such measures may provide
greater clarification of how segregation influences school-level high-stakes
testing performance. Finally, the present results do not inform us about the
influence of racial segregation on individual test performance. Such analyses
are crucial to develop a comprehensive understanding of the consequences
of school segregation for educational achievement.
In conclusion, we acknowledge that we have not tested the utility of
accountability mechanisms for raising educational achievement or for closing
the racialgap in educational achievement. Rather,our analyses represent only
an initial foray into examining whether distribution of students by race is an
important component in predicting school-level FCATperformance. We suggest that our results lead to an important conclusion: Using accountability
mechanisms to evaluate schools and dole out incentives and disincentives
without taking into account the racial distributionof students is clearly unfair
and probably will not maximize the efficiency of such initiatives in accomplishing their objectives. School districts that are not guided by the logic of
Brown and fail to take into consideration the importance of racial integration
627

Borman et al.
will likely face difficult problems ahead as disincentives accumulate for failing schools with large Black enrollments. Given the national popularity of
high-stakes testing, the FCATrepresents a harbingerof what is likely to occur
throughout the United States. We implore policymakers to remember the
legacy of Brown when considering the use of accountability mechanisms.
Notes
We gratefullyacknowledge the support of the Spencer Foundationthrough a major
grant to the firstauthor establishingthe Consortiumfor EducationResearchin Floridain
the David C. Anchin Center,College of Education,Universityof South Florida.We would
also like to thank Lee Anne Bell, the editor of this special issue, and MaenetteBenham,
the editor of AERJ,for their encouragement,as well as the anonymous reviewers whose
helpful comments improved the article.
'In Green v. County School Board of New Kent County (1968), the Supreme Court
ruled that desegregation must be achieved with respect to several factors,including student enrollments, facilities, staff, extracurricularactivities, and transportation.After this
decision, these "Greenfactors"were commonly used as goals for desegregationplans and
more recently have become a court standardfor assessing whether school districtsare to
be legally declared "unitary"(Orfield& Eaton, 1996).
aWeuse the term Hispanic ratherthan Latinoto be consistent with the term used by
the state of Floridain describingstudent demographics.
3Keyesv. Denver SchoolDistrictNo. 1 (1973) was the firstschool desegregation case
to recognize Latinos'rightto desegregation.
4TheFloridaSchools IndicatorsReportcan be found at http://info.doe.state.fl.us/fsir/.
The School Advisory Council Report can be found at http://www.firn.edu/doe/eias/
eiaspubs/pdf/sacrfin.pdf.
5Theerrorassociated with classificationof students into achievement levels on the
FCATis discussed in a technical report available at http://www.firn.edu/doe/sas/pdf/
fc00tech.pdf.
6Weoriginallyincluded a Hispanicsegregatedschool measure as well but found no
differencebetween this measureand the Whitesegregatedminoritymeasure;thus, we collapsed the two measuresinto one categoryreferredto here as White(non-Black)segregated.
7Inseparateanalyses, we also included the district-levelmeasure of percentageBlack
as a controlvariable.Because this variablefailed to reach statisticalsignificanceand failed
to influence the predictiveutilityof the dichotomous measureof segregatedschool status,
we omitted it from the analyses reportedhere.
8Althoughour measure of Black segregated school status is a school-level measure,
its value was determinedrelativeto the racialcomposition of the school district.

References
Askew, R. (1971, April 6). Governor'saddress. Tallahassee: Florida State Archives.
Askew, R. (1973, April 3). Address to the legislature.Tallahassee: FloridaStateArchives.
Bailey, T. (1952, February 6). Present status of Negro education in Florida's public
schools. Tallahassee: Florida State Archives.
Bankston, C., & Caldas, C. (1996). MajorityAfricanAmerican schools and social injustice: The influence of de facto segregation on academic achievement. Social
Forces, 75, 535-552.
Biennial report: Report of the superintendent of public instruction of the state of
Florida. (1946). Tallahassee, FL:T. J. Appleyard.
Biennial report: Report of the superintendent of public instruction of the state of
Florida. (1955). Tallahassee, FL:T. J. Appleyard.
Braddock, J. H. (1980). The perpetuation of segregation across levels of education:
A behavioral assessment of the contact hypothesis. Sociology of Education, 53,
178-186.
628

Accountability in a Postdesegregation Era
Braddock, J. H., & Eitle, T. M. (2003). The effects of school desegregation. In J. A.
Banks & C. M. Banks (Eds.), Handbook of research on multicultural education
(2nd ed., pp. 828-846). San Francisco:Jossey-Bass.
Braddock, J. H., & McPartland,J. M. (1989). Social psychological processes that perpetuate racial segregation: The relationship between school and employment
desegregation. Journal ofBlack Studies, 19, 267-289.
Brown v. Board of Education of Topeka, Kansas, 347 U.S. 483 (1954).
Brown, K. (1996). Revisiting the Supreme Court'sopinion in Brown v. Board ofEducation from a multiculturalistperspective. In E. C. Lagemann & L. P. Miller(Eds.),
Brown v. Board ofEducation: Thechallengefor today's schools (pp. 44-53). New
York: Teachers College Press.
Brown, S. (1999, August). High school racial composition: Balancing excellence and
equity. Paper presented at the meeting of the American Sociological Association,
Chicago.
Brunais, A. (1979, September 17). Black children have not yet gone to the head of
the class. Tampa Times, pp. B1, B2.
Citizens Education Committee. (1971). Partial transcript of meeting of Citizens Education Committee. Tallahassee, FL:Author.
Clotfelter, C. T. (2004). After Brown: The rise and retreat of school desegregation.
Princeton, NJ: Princeton University Press.
Cobb, J. A. (1990). The Sunbelt south: Industrializationin regional, national, and international perspective. In R. A. Mohl (Ed.), Searching for the Sunbelt. Historical
perspectives on a region (pp. 25-46). Knoxville: University of Tennessee Press.
Colburn, D. R. (1996). Florida politics in the twentieth century. In M. Gannon (Ed.),
The new history ofFlorida (pp. 344-372). Tallahassee: University Press of Florida.
Colburn, D. R., & Scher, R. K. (1980). Florida's gubernatorial politics in the twentieth
century. Tallahassee: University Press of Florida.
Condron, D. J., & Roscigno, V. J. (2003). Disparities within: Unequal spending and
achievement in an urban school district. Sociology ofEducation, 76, 18-36.
Crain, R., & Mahard, R. E. (1983). The effects of research methodology on desegregation achievement studies: A meta-analysis. American Journal ofSociology, 88,
839-854.
Debra P. v. Turlington, 474 F. Supp. 244 (M.C. FL. 1979).
Eitle, T. M. (2003, August). Segregation, diversity, and accountability in Florida
schools. Paper presented at the meeting of the American Sociological Association,
Atlanta, GA.
Eitle, T. M. (in press). Diversity, desegregation, and accountability in Florida districts.
In K. Borman & S. Dorn (Eds.), The strugglefor equity: Educational reform in
Florida. Gainesville: University Press of Florida.
Eitle, T. M., & Eitle, D. J. (in press). Inequality, segregation, and the overrepresentation
of AfricanAmericans in school suspensions. Sociological Perspectives.
Entwisle, D. R., Alexander, K. L., & Olson, L. S. (1997). Children, schools, and
inequality. Boulder, CO: Westview Press.
Epstein, J. L. (1985). After the bus arrives: Resegregation in desegregated schools.
Journal of Social Issues, 41, 23-43.
Eyler,J., Cook, V., & Ward, L. (1983). Resegregation: Segregation within desegregated
schools. In C. H. Rossell & W. D. Hawley (Eds.), The consequences of school
desegregation (pp. 126-162). Philadelphia: Temple University Press.
Finn, J. D., & Achilles, C. M. (1990). Answers and questions about class size: A
statewide experiment. American Educational ResearchJournal, 27, 557-577.
Florida Commission on Education Reform and Accountability. (1994). Blueprint 2000.
Tallahassee, FL:Author.

629

Borman et al.
Florida Department of Education. (1996). Sunshine state standards. Tallahassee, FL:
State Board of Education.
Florida Department of Education. (1998). Florida's system ofschool improvement and
accountability. Tallahassee, FL:Bureau of CareerDevelopment and Educational
Improvement.
Grant, C. A. (1996). Reflections on the promise of Brown and multiculturaleducation.
In E. C. Lagemann & L. P. Miller (Eds.), Brown v. Board ofEducation: The challenge for today's schools (pp. 107-121). New York: Teachers College Press.
Green v. County School Board of New Kent County, 391 U.S. 430 (1968).
Hanushek, E. A. (1997). Assessing the effects of school resources on student performance: An update. Educational Evaluation and Policy Analysis, 19, 141-164.
Hegarty, S. (1995, December 27). He's making waves in Florida education. St. Petersburg Times, p. Al.
Herrington, C., & McDonald, V. M. (2001). Accountability as a school reform strategy:
A 30-year perspective on Florida. In C. D. Herrington & K. Kasten (Eds.), Florida
2001: Educational policy alternatives (pp. 7-34). Jacksonville: Florida Institute
of Education.
Hillsborough County School District. (1960, September 21). Application for pupil
special assignment. Unpublished document.
Hoxby, C. (2002). The power of peers. Education Next, 2, 57-64.
Kaeser, S. C. (1979). Suspensions in school discipline. Education and Urban Society,
11, 465-486.
Keyes v. Denver School District No. 1, 413 U.S. 189 (1973).
Kimbrough, R., Alexander, K., & Wattenbarger,J. (1984). Government and education.
In M.J. Dauer (Ed.), Florida'spolitics and government (2nd ed., pp. 422-448).
Gainesville: University Press of Florida.
Meier, K., Stewart,J., & England, R. E. (1989). Race, class, and education: Thepolitics
ofsecond-generation discrimination. Madison: University of Wisconsin Press.
Mickelson, R. A. (2001). Subverting Swann: First-and second-generation segregation
in the Charlotte-Mecklenburgschools. American Educational ResearchJournal,
38, 215-252.
Mills, R., & Bryan, M. M. (1976). Testing... grouping: Thenew segregation in southern
schools?Atlanta,GA: Southern Regional Council.
Natriello, G., & Pallas, A. M. (2001). The development and impact of high-stakes testing. In G. Orfield & M. L. Kornhaber (Eds.), Raising standards or raising barriers?Inequality and high-stakestesting in public education (pp. 19-38). New York:
Century Foundation Press.
Nyhan, R. C., & Alkadry, M. G. (1999). The impact of school resources on student
achievement. Journal of Education Finance, 25, 211-227.
Orfield, G. (1978). Must we bus?Segregated schools and nationalpolicy. Washington,
DC: Brookings Institute.
Orfield, G., & Eaton, S. (1996). Dismantling desegregation: The quiet reversal of
Brown v. Board ofEducation. New York: New Press.
Shircliffe, B. J. (2002). Desegregation and the historically Black high school. Urban
Review, 34, 135-158.
Sinclair,C. (1978,June 17). Desegregation'squiet success. WashingtonPost, pp. Al, A10.
Swann v. Charlotte-MecklenburgSchools, 402 U.S. 1, 15 (1971).
Tractenberg, P. L., & Kahn, L. (1980). State minimum competency testing programs:
Legal implications of minimum competency testing: Debra P. and beyond.
Washington, DC: National Institute of Education.
U.S. Commission on Civil Rights. (1971). Mexican American education study, Report
1: Ethnic isolation of Mexican Americans in thepublic schools of the Southwest.
Washington, DC: U.S. Government Printing Office.
630

ErEra
a
iin
n
aa Postdesegregation
Accountability
Postdesegregation
Accountability
U.S.
Commission
on
school
Civil
on
Statement
U.S.
Commission
on
Civil
(1977).
on
Statement
(1977).
desegredesegremetropolitan
metropolitan
Rights.
Rights.
school
AA report
ted
vil
StStates
ates
Commission
Commission onon CiCivil
gatgation:
ion:
Washington,
reportoof
fthe the UniUnited
RiRights.
ghts.
Washington, DC:DC:
Author.
Author.
R.R. R.R. ((2000).
:
students
nd
ththe
e
2000).
students inin TexasTexas:
ValValencia,
encia,
nority
Inequalities
schooling
Inequalities aand
schooling ofof miminority
Historical
nd
Historicalaand
conditions. Hispanicjournal
ences,
contemporary
contemporary conditions.
HispanicJournal ofBehavioral
ofBehavioral SciSciences,
445-459.
445-459.
222,
,
aand
nd
ththe
e
.S.
-Mexican
Subtractive
1999).
Valenzuela,
Subtractive schooling:
Valenzuela, A.A. ((1999).
polpolitics
itics
oof
f
youtyouth
h
schooling: UU.S.-Mexican
StState
ate
f
NeNew
w
YoYork
rk
PrPress.
es .
caricaring.
ng.
AlAlbany:
bany:
University
University oof
A.A. SS.,
oof
f
ththe
e
fects
R.R. LL.
.
aand
nd
((1994).
1994).
WelWells,
ls,
ain,
., && CrCrain,
heory
long-term
Perpetuation
long-term efeffects
Perpetuation ttheory
sschool
chool
ew
531-556.
531-556.
4,
desegregation.
desegregation. ReviReview
ofEducational
of Educational Research,
Research, 664,
A.A. 0.0. ((1981).
ThThe
e
orida:
evolution
StState
ate
nd
education
1981).
evolution
WhiWhite,
te,
education inin FlFlorida:
leadership
publpublic
ic
leadership aand
oof
f
a a syst
InIn R.R. KK.
TThe
-he
education
nd
ththe
e
.
Goodenow
.
WhiWhite
te
education aand
Goodenow && AA.
OO.
.
((Eds.),
Eds.),
system.
em.
rrise
ise
e
nnew
ew
SoutSouth
h
HaHall.
l.
on:
237-249).
237-249). BostBoston:
oof
f ththe
((pp.
p.
W.W. JJ.
nd
tthe
he
TThe
he
nner
1987).
WiWilson,
lson,
. ((1987).
underclass,
underclass, aand
ttruly
ruly
ccity,
ity,
disadvantaged:
disadvantaged: 7beThe iinner
PrPress.
es .
publicpolicy.
cago:
publicpolicy. ChiChicago:
University
cago
University ofof ChiChicago
December9, 2003
receivedDecember
2003
Manuscript
Manuscriptreceived
Revision
receivedMay
2004
Revisionreceived
10, 2004
May10,
2004
June 1, 2004
Accepted
AcceptedJune

631