CAN SCHOOL INTEGRATION INCREASE STUDENT ACHIEVEMENT? EVIDENCE FROM HARTFORD PUBLIC SCHOOLS

A Thesis submitted to the Faculty of the

Graduate School of Arts and Sciences of Georgetown University

in partial fulfillment of the requirements for the degree of

Masters of Public Policy in Public Policy

By

Lucretia Anne Witte, B.A.

Washington, DC April 12, 2016

ii

Copyright 2016 by Lucretia Anne Witte

All Rights Reserved

iii

CAN SCHOOL INTEGRATION INCREASE STUDENT ACHIEVEMENT? EVIDENCE FROM HARTFORD PUBLIC SCHOOLS

Lucretia Anne Witte, B.A.

Thesis Advisor: Erica Johnson, Ph.D.

ABSTRACT

Since Brown vs. Board of Ed prohibited de jure school segregation, the national black-

white achievement gap has reflected trends in school integration. Hundreds of programs

and policies since then have had only incremental impacts on the disparities in

educational outcomes based on race and socioeconomic status. In Hartford, CT, the 1999

Sheff vs. O’Neill case required the State of Connecticut to create an integration program

to reduce racial isolation in Hartford’s segregated inner city schools. As a result,

Hartford implemented a proactive integration policy. Between 2008 and 2013 the

number of poor black and Hispanic students attending an integrated school in Hartford

increased twenty percentage points. A difference-in-difference approach was used to test

whether this policy had an effect on reading, writing and math proficiency rates, holding

demographic factors constant. The study finds that integration status has a positive and

highly significant effect on proficiency rates in math and reading and no effect on writing

proficiency.

iv

O, let my land be a land where Liberty

Is crowned with no false patriotic wreath,

But opportunity is real, and life is free,

Equality is in the air we breathe.

Langston Hughes, “Let America Be America Again”

v

TABLE OF CONTENTS

Introduction ..........................................................................................................................1 Background ..........................................................................................................................4 Literature Review: Does Integration Impact What Students Achieve? ............................12

Conceptual Model: How Can Integration Improve Student Outcomes? ..........................19 Empirical Strategy: Connecticut District Level Data and The Difference In Differences Model .................................................................................................................................22

Results ................................................................................................................................27

Potential Limitations ..........................................................................................................35 Directions for Further Research .........................................................................................36 Policy Implications ............................................................................................................37 Conclusions ........................................................................................................................38 References ..........................................................................................................................39

vi

LIST OF FIGURES

Figure 1: White Flight across America, between 1960 and 1970 ........................................2 Figure 2: The Black-White Achievement Gap Between 1971-2012 ...................................3 Figure 3: Map of Integration Efforts in 1966 .......................................................................7

Figure 4: Hartford Public Schools Regional School Choice Program .................................9 Figure 5: Towns in the Sheff Open Choice Region ...........................................................18 Figure 6: Conceptual Model for How School Integration Impacts Student Outcomes .....20 Figure 7: Population Density of Connecticut Counties, with City Names ........................24 Figure 8: Regression Equation ...........................................................................................26

LIST OF TABLES

Table I: Percent Of Black Students In Schools With More Than Half Minority Students By Region, 1968-1980 .........................................................................................................4 Table II: Percent of Minority Hartford Residents Attending Integrated Schools ..............10 Table III: Descriptive Statistics on Connecticut Counties .................................................23

Table IV: Student Demographics in Comparison Cities ....................................................25 Table V: Fully Controlled Model Using Math, Reading, and Writing Proficiency Scores............................................................................................................................................28 Table VI: The Impact of Integration on Proficiency Rates in Suburban Open Choice Districts ..............................................................................................................................31 Table VII: The Relationship Between School Type on Math Proficiency Rates ..............33

1

INTRODUCTION

America is a diverse society in which educational differences have the

potential to become a progressively larger source of inequality and social

conflict. Many people now recognize that eliminating these differences

has become a moral and pragmatic imperative. (Miller, 1995)

In the US we believe that education is the great equalizer, but we have acted on

that belief inconsistently throughout our nation’s history. Today, the black-white

achievement gap persists despite broad school reform efforts ranging from increased

spending to higher accountability to better teacher education. This thesis will use a

difference-in-difference analysis of an integration effort in Hartford, Connecticut to

determine whether purposeful integration increased student achievement, with the hope

that Hartford’s outcomes can support policy decisions in other cities.

On May 7, 1954, the Supreme Court ruled that segregation in schools was illegal

(Brown v. Board of Ed, 1954). This case marked a promise of better opportunity for the

next generation. During the 1960s, schools across the country integrated in compliance

with this ruling. Integration did reduce the black-white achievement gap: between 1970

and 1988, the black-white achievement gap narrowed by 46% (NCES, 2012). Further

studies determined that higher achievement led to better life outcomes: blacks who spent

five or more years in desegregated schools earned 25% more than those that did not

(Johnson, 2011). They were less likely to be incarcerated, and more likely to be healthy

as adults (ibid). Integration had no negative effect on whites’ earnings, incarceration, or

health.

2

Yet over the years, segregation crept back. Detroit’s 1974 Mililken v. Bradley

case decided that segregation was constitutional as long as it was not a product of the

district’s explicit policy (Miliken v. Bradley, 1974). In the early 2000s, federal judges

struck down mandates that had ordered deliberate integration policies (Greenberg,

Grewal, Kalogrides, Reardon, 2011). Meanwhile, between 1960-1970 whites began

moving out of cities, leaving minorities behind in the urban core. Neighborhoods and

schools became resegregated (Fry and Taylor, 2012, Frankenberg, Lee, and Orfield 2003;

Orfield, Kucsera, and Siegel-Hawley 2012). Even though integration was shown to have

positive effects on minority students and no negative effects on white students, many

districts did not pursue desegregation efforts once federal mandates were removed.

Within these re-segregated schools, expectations of students, resources, and traditions

were different (Kozol, 2005). These inequities perpetrated an achievement gap, which

Figure 1: White flight across America, between 1960 and 1970. Source:NYSTROM/Herff Jones,

2015

3

contributes to a gap in social outcomes between between whites and minority students

(Hanushek, Kain and Rivkin, 2009, Perry 2010).

Figure 2: The Black-White Achievement Gap between 1971-2012. Source: NCES, 2014.

Myriad public and non-profit efforts have been targeted at the racial achievement

gap.1 Head Start, Title I, local at-risk funding, teacher evaluations, instructional reforms,

and non-profits like Teach for America all focus on providing more equitable educational

opportunities for minority students.2 But despite this multi-sector, multifaceted push for

1 The Department of Education lists 22 nationwide initiatives and projects; there are more than two million non-profits in the US, according to the National Center for Charitable Statistics at the Urban Institute. If one-twentieth of those non-profits did work related to education, that would be over one hundred thousand programs related to education reform. In addition to these, there are also state and local education initiatives and programs. 2 Mission statements for these organizations each mention equalizing educational opportunity for students from all backgrounds. Head Start: “The National Head Start Association (NHSA) is a nonprofit

4

educational equity, the black-white student achievement gap has actually widened

slightly since 1988, when integration was at its peak (NCES, 2012).

Table I: Percent Of Black Students In Schools With More Than Half Minority Students

By Region, 1968-1980

Total

U.S.

Southern

States

Border

States

Northeast

States

Midwest

States

West

States

1968 76.6 80.9 71.6 66.8 77.3 72.2

1972 63.6 55.3 67.2 69.9 75.3 68.1

1976 62.4 54.9 60.1 72.5 70.3 67.4

1980 62.9 57.1 59.2 79.9 69.5 66.8

Change

1968-80

-13.7 -23.8 -12.4 +13.1 -7.8 -5.4

Can renewed integration efforts increase student achievement? A few cities have

recommitted to integration efforts, with mixed results. This thesis will examine whether

school integration efforts in Hartford, Connecticut reduced the achievement gap for

minority students between 2007-2013.

BACKGROUND

The unique story initiated by Connecticut’s Supreme Court is of great

regional and national importance, not as a grand solution to a very deeply

embedded problem but as an important example of what can be done

through the wise use of choice programs with clear civil rights objectives

and major educational innovations. (Orfield, 2015)

organization committed to the belief that every child, regardless of circumstances at birth, has the ability to succeed in life.” Title I: “Title I… provides financial assistance to local educational agencies (LEAs) and schools with high numbers or high percentages of children from low-income families to help ensure that all children meet challenging state academic standards.” Teach for America: “Our mission is to enlist, develop, and mobilize as many as possible of our nation's most promising future leaders to grow and strengthen the movement for educational equity and excellence.”

5

Sheff vs. O’Neill: The Catalyst for Integration Efforts in Hartford, CT

In 1989 in Hartford, Connecticut, the parents of 4th grader Milo Sheff and

seventeen other children began a civil action against the state for providing a racially

isolated and therefore unequal education to their children. Some plaintiffs were black,

some were Latino or white. Even though segregation in their schools was no result of

state policy—rather as a result of racially homogenous neighborhoods— the plaintiffs

claimed that the state was responsible for proactively integrating these schools.

In 1995, Judge Harry Hammer ruled in favor of the state, writing in his decision

that Connecticut was not responsible for correcting educational inequities due to de facto

segregation, only for ensuring that no de jure segregation took place.3 In 1996, John

Brittain, a civil rights lawyer, appealed Hammer’s decision. The initial ruling was

overturned by the Connecticut Supreme Court on the basis that Connecticut had an

obligation to provide students with equal educational opportunities.4 According to their

decision, that meant the state had to provide students an education that was not impaired

by racial and ethnic isolation (Sheff vs. O’Neill, 1999). In compliance with the court’s

ruling, the legislature called for additional spending on magnet schools, voluntary

interdistrict transfer programs, and interdistrict coop programs. It prohibited redrawing

of district boundaries (ibid).

3 Precedent for Hammer’s decision could be found in Miliken v. Bradley, among other cases.

4 While there was some language in the Connecticut State Constitution about equitable school funding, the Supreme Court’s decision deviated from other US Supreme Court precedents that have held that only de jure segregation is unconstitutional. Relevant language from the eigtht article of the Constitution of the State of Connecticut reads, “The interest of [the school fund] shall be inviolably appropriated to the support and encouragement of the public schools throughout the state, and for the equal benefit of all the people thereof….and no law shall ever be made, authorizing such fund to be diverted to any other use than the encouragement and support of public schools, among the several school societies, as justice and equity

shall require.”

6

In 2003, a legal settlement between the plaintiffs and the state formalized the

definitions, integration targets and programs the state would support to comply with the

Sheff vs. O’Neill ruling. The state funded two main programs: the Interdistrict Magnet

Program and the Project Choice Transfer Program (Dougherty, Estevez, Wanzer and

Tatem, 2006). The goal was that at least 30% of poor black and Hispanic students living

in downtown Hartford would attend integrated5 schools by June 2007.

This settlement effectively created a natural experiment in Hartford. Because the

ruling concerned Hartford and its suburban ring but not the other segregated school

systems in Connecticut, any change in achievement trends in Hartford that did not occur

in other cities could be attributed to the integration policy. If integration efforts narrowed

the achievement gap, integration could begin to be investigated as a strategy for school

reform.

Hartford: An Ethnographic and Political History of Inequality in Connecticut’s Capitol

Due to a strong manufacturing and shipping industry, Hartford was the wealthiest

city in the country by the late 19th century (Clark, 1876). Today it is the second-

wealthiest city in the nation (Lee, 2015). Attracted by its wealth and booming

manufacturing sector, immigrants flooded to Hartford during the first half of the

twentieth century, including a tide of Puerto Rican immigrants right after World War II.

But in the late 1950s, as suburbs around Hartford sprang up to make room for its new

5 A note about language: For the purpose of this study, “integrated” and “desegregated” mean different things. While all schools in the US were ordered to be desegregated by Brown vs. Board of Ed in 1954, being desegregated is necessary but not sufficient for being integrated. If a desegregated school’s neighborhood has only black and Hispanic residents, then it is, in the language of the Sheff case, racially isolated. This distinction can also be thought of in terms of de jure and de facto segregation. While segregation in Hartford was de facto, the court ruled that the Department of Education had a responsibility to create integrated educational environments. According to the terms of the settlement agreement, schools were considered integrated if they had at least 25% but not more than 75% black and Hispanic children in attendance.

7

residents, the city’s core began to decline. Businesses moved to less expensive locations

in the suburbs and wealthy whites followed them at increasing rates. In the 1990s,

Hartford’s population shrank by 11 percent (1999 Census).

Figure 3: Map of Integration Efforts in 1966, The Hartford Times.

Source: Hartford Public Library.

Despite its historic wealth, a series of recent economic policies, such as high taxes

on everything from income to sales to unemployment insurance, have made Connecticut

one of the most hostile climates to small business in the country (Powell, 2013). By

2012, Connecticut was last in the country for economic growth; that year it was the only

state whose economy actually shrunk (BEA, 2012). This took a toll on employment and

economic mobility prospects for residents. That year, 17.3% of the workforce was

unemployed and more than a quarter lived below the poverty line (De Avila, 2012).

Between 1996 and 2006, the number of small businesses in the state declined by 2.2%,

when the average change in all other states was a 10% increase (ibid). As a result of a

8

barren employment landscape, crime is pervasive in some parts of the urban core. In

2013, Hartford was the 24th most dangerous city in the US and the 4th most dangerous

city under 200,000 people (Warner, Fuchs and Lubin, 2013; FBI, 2013).

Limited economic opportunity and high crime rates contribute to the state of

housing and school segregation between inner-city Hartford and its suburban neighbors.

In 1987-88, the school year before Sheff vs. O’Neill was filed, there were 25,000 students

in the Hartford Public School District, 90.5% of whom were ethnic minorities, while the

surrounding suburbs were 2-30% minority (Brenz, 2013). Virtually all students who

attended segregated schools were also poor, and achieved at much lower rates than their

white affluent peers (Orfield, 2015). This led to Elizabeth Sheff’s 1989 decision to sue

the state on grounds that isolated education is not equal education, a conclusion that

much subsequent econometric research has justified (NCES report 2015-018, among

others). Importantly, due to its longstanding history of anti-slavery sentiment, Hartford

as a city was open to Sheff’s view on the state’s responsibility: teachers and principals

testified against the state in the Sheff proceedings, giving evidence that was crucial to

both the Supreme Court’s decision and to public sentiment about the case (Green, 1993).

The 2003 and 2008 Sheff Settlements: Goals, Programs and Results

In the initial 2003 Sheff settlement, the goal was that 30% of public school minority

students living in Hartford would have an integrated educational experience by June

2007. The settlement defined “integrated” schools as schools where minorities

constituted more than 25% and less than 75% of students. There were three main

initiatives through which the state sought to achieve this goal:

9

1) Interdistrict Magnet Schools. These schools, mostly located in the urban core,

were designed to attract suburban students to attend school in the city. They

offered specialized curricula, often with new facilities like a performing arts

center or a planetarium.

2) The Open Choice Program. This program allows inner city minorities to transfer

to a suburban district school.6 The hosting district specifies how many Open

Choice students they will accept each year.

Figure 4: Hartford Public Schools offers students the opportunity to choose from a

variety of school options through the Regional School Choice Office Lottery. HPS

partners with other magnet and charter providers as well as suburban open choice

districts.

6 Open Choice districts include Avon, Berlin, Bolton, Bristol, Canton, Cromwell, East Granby, East Windsor, Ellington, Enfield, Farmington, Glastonbury, Granby, Hartford, Newington, Plainville, Portland, Region 10, Rocky Hill, Simsbury, Somers, South Windsor, Southington, Suffield, Vernon, West Hartford, Wethersfield, Windsor, and Windsor Locks. Seats are available in elementary, middle and high schools.

10

3) The Reverse Choice Program. This much smaller program allows students from

the suburbs to transfer in to an urban district school.

Table II:

Percent of Minority Hartford Residents Attending Integrated Schools

Year Goal Actual

2002-2003 30% 10% 2004-2005 30% 13%

2006-2007 30% 17% 2008-2009 41% -- 2010-2011 41% -- 2012-2013 41% 37% 2015-2016 47.5% 45.5%

Data for this table was compiled from settlement documents and the Connecticut

Mirror’s reports on integration levels. Empty boxes represent years when integration

was not measured.

Students from either the Hartford urban core or the hosting districts were eligible to

participate in both programs. Spots were allotted through the Greater Hartford Regional

School Choice Lottery.

In 2003 when Sheff and the State of Connecticut reached this initial settlement,

only 11% of Hartford minority students were receiving an integrated education (Orfield,

2015). By 2007 that number had grown to 16.9%, but it had still fallen far short of its

goal of 30% (Orfield, 2009). Some magnet schools had too many minorities (>75%)

while others had too few (<25%) (Rioual, 2013). Further, many minorities attending

magnets were actually coming into Hartford from the suburbs, where their home districts

would have been more integrated than the options available to city residents (ibid). As a

result, the state and the plaintiffs negotiated a new settlement in 2008. This time, they

aimed for 80% of lottery applicants to be accommodated in an integrated education

environment by June 2013 (Connecticut General Assembly, 2008), with a fallback goal

11

of 41% of poor black and Hispanic students from the urban core being enrolled in

reduced-isolation settings. The state promised more resources and greater political

efforts to meet the increased goals. There was dramatic improvement—by 2013, 43% of

Hartford minority students attended a racially integrated school (Orfield, 2015).

Demand for the programs operated under the integration efforts is high. Many

more families enter the lottery than can be accommodated in either of the programs

(Dougherty, 2014). Even though Hartford surpassed its 2013 fallback goal of 41% of

minority students in reduced isolation schools in 2013, it did not succeed in placing 80%

of lottery applicants. Sheff and the state are currently in ongoing negotiations about

future requirements, with another ruling expected in the next 1-2 years (Megan, 2015).

There may be future settlements which require the city to meet this growing demand.

Several studies have been done using Hartford data from 2003 to 2008, which

find a positive effect of integration on student achievement scores in math and reading

(e.g., Bifulco, 2009; Cobb, 2011). Further, positive effects of magnet schools on student

achievement have been found in Hartford (Orfield, 2015). But no study has yet been

done to evaluate the impact of integration on academic outcomes since the Phase II

settlement efforts in 2008. Because the percent of students attending an integrated school

increased from just below 17% to 43% in the five years following those renewed

integration efforts, any impact on student achievement in Hartford during those years can

add to the body of evidence for policymakers considering whether integration can work

as a citywide education reform strategy.

12

LITERATURE REVIEW:

DOES INTEGRATION IMPACT WHAT STUDENTS ACHIEVE?

In the 1966 Coleman Report, formally titled Equality of Educational Opportunity,

researchers from Johns Hopkins concluded that poor black students learn better in well-

integrated classrooms (Coleman et al, 1966). Over the subsequent 50 years, researchers

explored and tested this conclusion to inform policy decisions, particularly in light of the

growing achievement gap amongst our country’s high and low achievers. Research has

shown that school segregation impacts student achievement when it concentrates race,

when it concentrates poverty, and when it reduces students’ self-worth.

Racial Diversity Promotes Achievement: Theory and Evidence

According to sociological theory, attending a racially diverse school develops

self-awareness and critical thinking as students learn about people who are different from

them and prepares them to succeed in a racially integrated world. This exposure may

stimulate their creativity, ability to see diverse perspectives, and curiosity about the

world, leading to higher academic achievement through the cultivation of attitudes and

mindsets (Schofield, 1995, and Wells & Crain, 1994).

There are mixed results on whether higher levels of integration lead to higher

student achievement. Cook and Evans (2000) found that little of the black-white gap in

NAEP scores can be attributed to segregation level. Card and Rothstein (2007) found

that neighborhood segregation had consistently negative effects on student achievement,

but that school segregation had no independent effect. The most recent result, Hanushek,

Kain and Rivkin (2009), found that the more segregated a school was, the worse its black

13

students performed academically, while segregation had insignificant effects on white

achievement. The magnitude of their finding was significant: according to their analysis,

if all Texas schools had the mean share of black enrollment, the black-white achievement

gap for 7th graders would be reduced by 10% (Hanushek, Kain & Rivkin, 352). One

potential explanation why these studies found such different results is that it was hard to

find exogenous variation in racial composition in the data used for these studies.

Socioeconomic Status is the Best Predictor of Student Success: Theory and Evidence

Theory about poverty and its influence on student achievement suggests that low

socioeconomic status (SES) influences student preparation for school, parental

involvement in education, and student aspirations beyond school (American

Psychological Association, 2014). Students in low-SES households are less likely to

have developed academic and social skills before starting school (Lacour, 2011). Once

they start school, they are less likely to receive help from their parents with homework or

to have parents engage in school activities. (US Department of Health and Human

Services, 2000). As they reach high school, they are less likely to stay enrolled and

graduate and are more likely to feel as though material is either not relevant or not

interesting to them (Yazzie-Mintz, 2007).

The empirical evidence on the implications of poverty on student achievement is

far less mixed than the effects of integration. Studies have shown that students’ own

socioeconomic status and that of their peers is positively related to achievement. The

Coleman Report found that the best predictor of student academic achievement was

family socioeconomic status, and the second-best predictor was the socioeconomic status

14

of the other students in the school (Coleman, 1966). Rusk (2002) studied public schools

in Madison-Dane County, WI, and found that a 1% increase in middle class classmates

was associated with a 0.64 percentage point gain in reading and a 0.72 percentage point

gain in math. Rumberger and Palardy (2005) found that a school’s SES was as impactful

as a family’s SES on student achievement growth in math, science, reading and history.

The PISA 2006 report found neighborhood effects of socioeconomic status: students

from any socioeconomic background performed better in a high SES school than a low

SES school (PISA, 2006). Perry (2010) found that increases in a school’s mean SES

were associated with consistent increases in student achievement, and that this

relationship held regardless of students’ individual SES.

Self-Worth: The Psychological Theory Behind an Omitted Variable

The final avenue through which segregation impacts achievement is students’ self

worth. Sociological theory postulates that students who attend segregated schools

develop an attitude that they are not “worth it” when they see how run-down their school

is or how little their teachers care (Epps, 1975). Recent ethnographies of race relations in

the US have cited broken windows and trash left out on the street in poor minority

communities as physical manifestations of the world’s value for them (Coates, 2015,

among others).

While this theory does not have the empirical evidence that the other two do, it is

included here because psychological studies of black children that connected low self-

esteem to segregated schooling were the main social science evidence of harm from

segregation cited in Brown vs. Board of Education (Hanushek, 2009). In spite of

15

numerous qualitative studies of the self-worth effect, quantitative social science research

has not explored this theory or its relationship to race and SES. Many studies instead try

to isolate the effects of schools’ racial and socioeconomic composition, when in fact the

two constructs are very much interwoven in the lived realities of schools (Mickelson,

2010). Without explicit studies of the impact of segregation on students’ self image, the

psychological effects should be considered an omitted variable. When students attend

schools in both racially and socioeconomically isolated neighborhoods, the self-worth

effect most likely amplifies the negative effects of race and SES on student achievement.

Results of Integration Efforts on Student Achievement

Brown vs. Board of Ed’s prohibition of de jure segregation was correlated with an

increase in student achievement. In 1988 when integration rates peaked, the black-white

achievement gap had closed 54% (The Nation’s Report Card, NCES 2012). After 1988,

the trends in de facto segregation began to undo the progress made by Brown vs. Board of

Ed and the achievement gap began to widen again.

Based on the substantial empirical evidence that integration is positively

correlated with achievement growth, a number of cities have used proactive integration as

an education reform strategy. Some have found positive effects on student achievement

or psychological metrics, but others have demonstrated no or negative impact.

A few examples of integration with positive results include Louisville and

Jefferson County, Kentucky; Seattle, Washington; LaCrosse, Wisconsin; and Cambridge,

Massachusetts (Semuels, 2015; Kahlenberg, 2012; and Cowen Institute, 2011). In 2008,

Orfield, et. al., reported that students who attended integrated schools in Jefferson County

16

and Seattle “have reported that interracial schooling experiences have been valuable and

have made them better prepared to live and work in diverse communities.” (Orfield, 104).

Kahlenberg (2000) shows that through peer effects, economic integration in LaCrosse

supports academic growth.

Those cities where integration efforts were ineffective include Detroit, which only

made it two years before being released from its court order by the Milliken v. Bradley

case, and Metco in Boston, where public opposition shut the program down. Detroit did

not collect enough data for an empirical evaluation of the program’s effect. Boston lasted

long enough that economists could evaluate impact: Angrist and Lang (2004) found that

there was little positive impact on student achievement, but that blacks in receiving

districts could be negatively impacted by the arrival of lower-achieving black students.

Research finds that integration methods and public opinion are crucial to the

positive relationship with student achievement. Today, the most successful form of

integration has been the magnet school, which pulls parents, students and faculty from

diverse communities into a new school (Kahlenberg, 2000). Following that, open choice

programs with friendly receiving districts have been correlated with positive effects on

student achievement. Receiving districts which did not voluntarily extend spaces to

urban students, as in Boston’s Metco and the Detroit program, fare worst.

The comparison between Louisville, KY and Detroit, MI shows how two

originally similar districts have diverged since their decisions about integration. In 1972

when the courts ordered these two districts to integrate, their demographics looked

similar: they had equally segregated schools and proportionally-sized racial ghettos

(Orfield, 2015). In 1974 their paths diverged: Miliken v. Bradley freed Detroit from the

17

integration mandate, while Louisville kept theirs. In 2000, the average black student in

Detroit attended school with less than 2% white students, whereas black students in

Louisville were likely to attend schools that were half white (ibid). A comparison of

student achievement on NAEP fourth grade reading tests in 2011 showed that twice as

many students scored at goal or higher in Louisville as in Detroit (NAEP District

Summary Tables, 2011). Further, there are stark economic differences: in 2010 Detroit

had 28 jobs per hundred residents while Louisville had 61 (Orfield, 2015). Between

2005-2015, Detroit lost 25% of its population while Louisville’s increased by 7% (ibid).

There are many factors besides their school integration policies that have influenced these

cities’ evolution, but this case study invites researchers to consider and investigate

whether school integration policies can impact housing, race relations, economic

opportunity and growth of enterprise in a city that chooses to integrate.

18

The Impact of Segregation in Hartford

Figure 5: Towns in the Sheff Open Choice Region.

Source: CT Regional School Choice Office, 2015.

In Hartford, research on the impact of integration on student achievement was

done prior to 2012 when integration levels were less than 30% (e.g., Bifulco, 2009, and

Cobb, 2011). These analyses determined that the effects of integration programs were

positive and statistically significant on reading and math scores of middle and high

school students (Cobb, 2009). In the years since those studies came out, integration rates

have risen from 30% of minority students in Hartford attending integrated schools to

45.5% (Thomas, 2013). The data from the past three years thus provides more variation

in integration rates. This thesis will use the Hartford data from the years between 2007-

2014 to show the impact of the integration policy on student test scores.

19

CONCEPTUAL MODEL

HOW CAN INTEGRATION IMPROVE STUDENT OUTCOMES?

I hypothesize that higher levels of school integration improve academic

achievement for black and Hispanic students without affecting white students. I expect

that the more even the ratio of black and Hispanic students to white and Asian students,

the greater the improvement in student outcomes for blacks and Hispanics.

To derive the conceptual model, I returned to the research, which suggests that

integrating schools changes student outcomes predominantly through school quality, peer

effects, and psychological effects. Some researchers have found that integration mainly

acts through improved school quality (Rumberger & Palardy, 2005). Others have found

that the effect comes from peer effects; that is, minorities benefit from being in a school

with whites or Asians (Hanushek, Kain & Rivkin, 2000). The following figure presents a

conceptual model for how racial integration could catalyze improved educational

outcomes. Consistent with the existing literature, this model suggests that a more even

ratio of blacks and Hispanics to whites and Asians will improve student outcomes for

blacks and Hispanics without significant impact on whites and Asians.

20

Figure 6: Conceptual Model for How School Integration Impacts Student Outcomes.

School quality, the first channel through which integration impacts achievement, can be

determined by racial composition through increased funds or better advocacy. In the

context of Hartford, white and Asian students attending integrated schools are almost all

from a middle or high SES households (Orfield, 2015). High-SES households pay more

in taxes, which means more money is available for school budgets. Further, more than

40% of low-income students don’t get a fair share of state and local education funding

(US Department of Education, 2011). While low-income schools receive supplements

from federal programs like Title I, this money often comes with restrictions that could

impact how effectively it is deployed (Formula Fairness Campaign, 2015).

School quality could also be impacted through parent involvement. Parental

advocacy can include demanding better information about what students are learning,

pushing to recruit and retain higher quality teachers, volunteering at school activities, or

even working on homework with one’s child. These behaviors have been shown to have

21

strong positive effects on student achievement (Hill & Taylor, 2004). According to the

U.S. Department of Health and Human Services, only 36% of low-income parents were

involved with three or more school activities on a regular basis, compared with 59% of

parents above the poverty line (HHS, 2015). When students from low social-capital

families attend an integrated school, they benefit from the increased parent engagement.

Rumberger and Palardy (2005) found that the most important channels through which

school quality impacts achievement for black and Hispanic students are increased teacher

expectations, more homework assigned, more rigorous courses taught, and students

feeling safer at school.

Peer effects are the second channel through which integration impacts student

academic outcomes. Children from affluent families are more likely to be academically

advanced coming into school, while children from low SES families begin kindergarten

with less linguistic capability than their middle-class and affluent peers (Purcell-Gates,

McIntyre & Freppon, 1995). Students from affluent families are more likely to

demonstrate constructive learning behaviors, whereas low SES is associated with higher

levels of emotional and behavioral difficulties like anxiety, depression and ADD/ADHD

(Weissman et al., 1984; Goodman, 1999; Spencer et al., 2002), and higher levels of

aggression (Molnar et al., 2008). Research suggests that low SES students benefit from

the academic preparation of their peers and assimilate to standards of behavior

demonstrated by high SES students (Coleman, 1966; Rusk, 2002; Perry, 2010).

Peer effects can also impact student achievement via social skills. Attending an

integrated school has been shown to make students more comfortable around others of

different backgrounds and less racially prejudiced (Wells, Holme, Revilla, & Atanda,

22

2004). In the 2006 Parents Involved case, 553 social scientists signed an amicus brief

stating that integrated education exposes students to cultural knowledge and social

perspectives and promotes critical thinking (Amici Brief for the American Council on

Education, 2007). Black and Hispanic students also graduate and attend college at a

higher rate when attending integrated schools (Bachman, 1971). These skills and further

education in turn sets them up for better job prospects beyond graduation (Garda, 2011).

The third channel through which integration impacts student outcomes for low-

income black and Hispanic students is via psychological effects. Kenneth and Mamie

Clark found that black children were likely to view black as “bad and ugly” and white as

“good and pretty” (Clark and Clark, 1939). This trend is combined with the impact of

low SES. According to the American Psychological Association, lower SES is linked to

negative psychological outcomes like depression, learning disabilities, and a tendency to

give up on tasks, while high SES corresponds with positive outcomes like optimism, self-

esteem and perceived control (2015).

EMPIRICAL STRATEGY:

CONNECTICUT DISTRICT LEVEL DATA

AND THE DIFFERENCE IN DIFFERENCES MODEL

A difference in difference approach can be used to test the hypothesis about

whether the integration program changed the trajectory of student outcomes in Hartford.

The study will compare academic achievement trends in Hartford integrated schools (the

“treatment group”) to achievement trends in other similar cities in Connecticut (“the

control group”), before and after the integration efforts. Based on the difference between

23

treatment and control before 2008 and the difference after, an integration effect on

achievement can be estimated.

In late 2008, the state of Connecticut reached a new settlement where the state

agreed to increase the proportion of inner-city minority students attending integrated, or

“reduced isolation,” schools. The pre-integration years are 2006-2008. Between 2008

and 2011 the proportion of inner city minority students in integrated schools increased by

20 percentage points. The post-integration years will be considered 2011-2013.

All school districts in Connecticut for which the state reports achievement data

are included in the sample. Below are some descriptive statistics at the county level.

Table III:

Descriptive Statistics on Connecticut Counties7

County Name 2010 Pop Avg Adult Ed % FRL % Black/Hisp AvgInc

Fairfield 916,829 11.81 22.6 25.6 $82,283

Hartford 894,014 11.56 25.9 30.2 $64,967

Litchfield 189,927 11.87 17.8 7.8 $71,338

Middlesex 165,676 11.93 13.3 10.5 $76,994

New Haven 862,477 11.80 31.4 30.6 $61,996

New London 274,055 11.93 24.6 17.5 $66,583

Tolland 152,691 11.93 13.5 8.4 $80,529

Windham 118,428 11.93 29.4 12.4 $59,333

Fairfield, Hartford and New Haven counties are home to the state’s largest cities but also

to some of the most affluent and well-educated suburbs. For example, Fairfield County

has nearly 23% of students that qualify for free and reduced lunch, yet the mean family

income there is over $82,000. The poor black and Hispanic families in Bridgeport, a city

7 Full variable names: Column I: 2010 county population, Column II: Average Adult Educational Attainment, Column III: Percent of students receiving free or reduced lunch, Column IV: Percent of students that are black or Hispanic, and Column V: 2013 Average Income.

24

in Fairfield County, earn much less than $82,000 a year, but these disparities are muted

by the pockets of extreme wealth that exist in other parts of the county. In Fairfield,

Hartford and New Haven counties, the percent of students receiving free and reduced

lunch tracks the percent of black and Hispanic students quite closely, whereas in counties

without big cities, such as Litchfield or Windham, there are poor white and Asian

students, too. The combination of urban poverty and racial segregation in these cities has

sparked a charter movement, independent of the Sheff integration settlement.

Figure 7: Population Density of Connecticut Counties, with City Names

Bridgeport and New Haven are of particular interest as comparison districts.

They have similar attributes to Hartford: urban cores surrounded by suburbs, similar

crime rates and similar economic opportunities (Rizzo, 2013). Hartford has a greater poor

black and Hispanic population that New Haven, but a smaller one than Bridgeport.

Importantly, all three were trending similarly during the period of interest. All three

cities are in close proximity and thus subject to similar shocks from the recession. All

25

three experienced an increase in the presence of charter schools and Teach for America

(Connecticut State Department of Education, 2015). All three implemented Common

Core State Standards in 2010 (CT SDE, 2010). One notable difference was that Hartford

built a cadre of magnet schools during the early 2000s to serve as the primary integration

strategy for the Sheff settlement, while magnet growth in other cities stagnated.

Table IV:

Student Demographics in Comparison Cities8

City % BH %FRL %SpecEd %ELL %Mag %Char

Bridgeport 92.4 87.2 9.8 7.7 1.9 44.8

Hartford 64.0 53.9 14.3 4.6 18.4 36.0

New Haven 54.7 44.7 7.1 3.8 10.1 34.0

Stamford 62.5 53.9 9.1 10.9 5.7 20.0

Waterbury 73.9 77.3 16.3 11.4 11.2 0.0

All other CT 12.4 29.4 14.6 2.6 0.6 0.0

The state of Connecticut sponsors multiple data sets which can be combined to

give a picture of how integration impacted student achievement.

The dependent variable is student proficiency, measured as percent of students in

the district scoring proficient or above on the Connecticut Mastery Test and the

Connecticut Academic Performance Test in grades 3, 5, 8 and 10 during the years

observed. Students take these tests in math, reading and writing once a year. The test

represents a standard metric of student achievement that is consistent over time. Ideally,

the regression could have been done using 2014 and 2015 data, but the most recent data

available was for 2011-12 and 2012-13. As integration did increase after 2012-2013, any

8 Full variable names: Column I: Percent Black and Hispanic, Column II: Percent of students receiving Free and Reduced Lunch, Column III: Percent of students classified as Special Education, Column IV: Percent of students classified as English Language Learners, Column V: Percent of LEAs that are Magnets, Column VI: Percent of that are Charters.

26

effect shown on the 2012 and 2013 years would be expected to be stronger in 2014 and

2015 as integration levels continued to rise.

Control variables are obtained from the Common Core of Data. These include

racial composition of the district, percent of students qualifying for free and reduced-

price lunch, the percent of Special Education students, the percent of English Language

Learner students, the average adult educational attainment, and the average school staff

per pupil ratio.9 Each of these controls will make the estimate for the impact of

integration on achievement more accurate by controlling for the well-documented effects

of being minority, poor, ELL or special education on student achievement.

Indicator variables were created for the treatment group (integrated schools in Hartford),

for post-2010 (after the 2008 resettlement of the Sheff v. O’Neill integration went into

effect), and for charter/magnet status. The coefficient of interest will be on the

interaction between an indicator variable for treatment and for post-2010.

Regression Equation

Figure 8: Regression equation of interest.

For each variable, I use data from district i and year t.

9 While these data were available at the school level, achievement data was only available at the district level. To perform the analysis, school-level observations were collapsed to generate district averages. This means that the controls are less accurate, because not all schools within a district have the same staff per pupil or the same rates of poverty. The Sheff settlement requires integration at the school level because having an even number of white and non-white students promotes the learning objectives discussed earlier; having controls at the district level makes it harder to control for the things that make individual schools different.

ProficiencyRateit = β0 + β1 (IntegrationStatus) + β2 (Post2010) +

β3(IntegrationStatus*Post2009) + β4 (PercentBlack/Hispit) + β5(PercentFRLit) + β6

(AdultEducAttainmentit) + β7 (ELLit) + β8 (SpecialEdit) + εit

27

I hypothesize that integrating a district will increase student proficiency rates.

While academic, social and labor market outcomes are all important metrics for the

success of an integration effort, only academic data is available at this time. A positive

result in academic gains bodes well for social and labor market indicators when this

population of integrated students graduates.

RESULTS

According to a difference-in-difference analysis, increased integration rates in

Hartford have a positive and statistically significant effect on reading and math

proficiency rates. Implementing integration policy in Hartford as a result of the Sheff

settlement moved integration levels in Hartford from 17% to 37% (see Table I). This 20

percentage point increase in the percent of inner-city black and Hispanic students in

integrated schools is associated with a 5.7 percentage point increase in math proficiency

and a 10.9 percentage point increase in reading proficiency (see the “Integration x Post

2010” coefficients in Table V). Writing proficiency increases were also positive but not

statistically significant.10

10 This may have been because not all grades in the sample reported test scores for writing. Therefore, although each district has a writing proficiency rate, this rate was generated from fewer grades reported.

28

Table V:

Fully Controlled Model Using Math, Reading, and Writing Proficiency Scores

1 2 3

VARIABLES Math Proficiency Reading Proficiency Writing Proficiency Integration Status -7.529*** -10.55*** -5.434**

(2.371) (2.811) (2.550)

Post-2010 4.557*** 6.113*** 3.165***

(0.399) (0.419) (0.334)

Integration*Post-2010 5.720** 10.89*** 2.279

(2.694) (3.190) (3.008)

% Black/Hisp -7.575*** -9.363*** -3.202

(2.746) (2.786) (2.201)

% FRL -26.11*** -31.37*** -23.41***

(3.137) (3.666) (2.748)

% SpEd -46.25*** -46.35*** -42.92***

(6.476) (5.258) (4.814)

% ELL -52.93*** -39.83*** -33.52***

(9.909) (8.250) (7.765)

Avg Adult Ed -0.417 -0.688 -1.211**

(0.656) (0.599) (0.496)

School Staff Per Pupil -12.52* -7.905 -2.923

(7.238) (6.162) (3.757)

Constant 105.9*** 104.3*** 113.6***

(7.606) (6.971) (5.764)

Observations 2,211 2,207 2,213

R-squared 0.656 0.667 0.587

Robust Standard Errors in Parentheses *** p<0.01 ** p<0.05 * p<0.1

In all three models, the controls on student characteristics are negative and

typically highly statistically significant. A 1 percentage point increase in special

education or ELL population tends to have the largest negative effect on predicted

29

proficiency rates, followed by increasing the percent of poor students. A 1 percentage

point increase in black and Hispanic enrollment also has a negative and statistically

significant impact on math and reading proficiency rates, but at a much smaller

magnitude. Average adult education and school staff per pupil did not have statistically

significant effects on proficiency rates.

Attending school in an integrated LEA in Hartford is associated with a negative

and statistically significant effect on proficiency rates. Because race, socioeconomic

status and other factors are already controlled for, this effect can be interpreted as the

effect of being located in Hartford on student proficiency rates. Schools located in

Hartford have lower proficiency rates, on average, than the rest of Connecticut.

Similarly, there is a positive and statistically significant impact of time on student

test scores. Students in the after-policy period of 2011-2013 scored better, on average,

than students tested between 2006-2008. This could be a result of the rising trends in

charter schools, which were not controlled for in this model. It could also be a product of

improved teacher quality due to the increasing presence of Teach for America teachers. It

could even be a product of greater awareness and advocacy for public education in

Connecticut due to changes like the Sheff settlement or implementation of the Common

Core.

Overall, a high level of variation is explained by the model, with an R2 of

approximately 0.65 for both reading and math proficiency rates.

Table IV captures students who attend integrated schools in Hartford city. It does

not include students in the Open Choice component of the integration program. Many

students who get bussed from the inner city out to a suburban school are among only a

30

few transferring Hartford students in their whole school. Before SY 2015-16, the State

added 325 Open Choice seats for Hartford students, but that is distributed over 28

receiving districts in the Hartford suburban ring. That means about 10 new students from

Hartford joined each district last fall, which is at most a handful of new students per

school. These schools do not meet the integration definition set out in Sheff, and from a

statistical perspective, there is likely very little difference on average from the addition of

these few inner-city students.

Studies of the Metco integration program in Boston found that integration had no

effect on incoming students but actually hurt students in receiving districts. Table VI

shows the impact of integration on suburban choice districts. For math, reading and

writing, impacts are negative and of very small magnitude, and none are statistically

significant. Hosting Hartford urban students through the open choice program had no

effect on proficiency rates in the receiving districts.

31

Table VI:

The Impact of Integration on Proficiency Rates in Suburban Open Choice Districts

1 2 3

VARIABLES Math Proficiency Reading Proficiency Writing Proficiency

Suburban Open Choice 0.767** 0.929* 0.972***

(0.375) (0.530) (0.371)

Post 2010 4.521*** 6.277*** 3.136***

(0.414) (0.433) (0.354)

Suburban Open Choice*Post 2010 -0.232 -0.655 -0.670

(0.553) (0.709) (0.519)

% Black/Hisp -8.321*** -10.23*** -3.858*

(2.744) (2.777) (2.214)

% FRL -25.20*** -30.50*** -22.53***

(3.136) (3.664) (2.739)

% SpEd -47.38*** -47.44*** -43.97***

(6.514) (5.138) (5.013)

% ELL -52.08*** -39.03*** -32.57***

(9.917) (8.213) (7.831)

Avg Adult Ed -0.0335 -0.452 -0.730

(0.570) (0.528) (0.461)

School Staff Per Pupil -12.70* -8.049 -3.178

(7.231) (6.168) (3.773)

Constant 101.3*** 101.4*** 107.8***

(6.649) (6.249) (5.403)

Observations 2,211 2,207 2,213

R-squared 0.655 0.665 0.586

Robust Standard Errors in Parentheses *** p<0.01 ** p<0.05 * p<0.1

Some critics of Hartford’s integration policy point to the fact that it rests upon

magnet schools to make integration appealing (Courant Editorial Staff, 2016). Magnet

schools are costly: they cost approximately $2,500 more per student than traditional

32

public schools (Megan, 2015). Yet they are a key tenet of Hartford’s Open Choice

integration strategy. Even though Table II shows that other counties in Connecticut are

wealthier, on average, than Hartford, median household income in Hartford is still 121%

of the median household income nationwide (American Community Survey, 2015). It

may be that on some level, the city’s wealth makes it possible to “afford” integration.

Cities that could not afford to build 19 new magnet schools in the course of a decade

might not be able to execute a voluntary integration strategy. This represents a threat to

the study’s external validity of the model.

To test whether magnet schools accounted for the positive relationship between

integration and achievement, Table VII examines the impact of non-district status (either

charter or magnet) on student achievement. First a controlled model was run with an

indicator for non-district (meaning the LEA is run by either a charter or a magnet

network). Next the model was run with an indicator for charter-only LEA, and last

magnet-only LEA. Non-district status has a positive and statistically significant effect on

math and reading proficiency rates (regressions 1 and 4). In regressions 2-3 and 5-6, this

effect decomposes into an even stronger and highly statistically significant effect of

charter status on proficiency rates, and a small negative but not statistically significant

effect of magnet status on proficiency rates. This suggests that that magnet schools are

not the reason for integration’s positive effects in Hartford as critics have claimed.

It is unclear why charters have such a high and statistically significant effect on

proficiency rates. The most obvious explanation is that their instructional and

organizational practices may actually be better than the district schools’ practices.

Another possibility is non-random selection of students into charter schools. In the past,

33

Table VII:

The Relationship Between School Type on Math Proficiency Rates 1 2 3 4 5 6

VARIABLES Math Math Math Read Read Read

% Black/Hisp -7.539*** -8.645*** -5.002* -7.849*** -8.114*** -5.926**

(1.511) (2.890) (2.664) (2.757) (2.780) (2.747)

% Poor -25.15*** -25.46*** -23.36*** -29.32*** -29.25*** -28.06***

(1.568) (2.929) (2.929) (3.397) (3.372) (3.327)

% SpecEd -34.82*** -34.53*** -50.63*** -37.93*** -39.99*** -48.49***

(4.282) (6.750) (6.329) (5.273) (4.821) (5.741)

% ELL -53.33*** -47.44*** -65.76*** -47.74*** -46.17*** -57.28***

(5.882) (10.97) (9.949) (9.387) (9.489) (8.890)

Avg Adult Ed 2.536*** 2.455*** 2.377*** 2.988*** 2.920** 2.900***

(0.383) (0.496) (0.506) (0.515) (0.515) (0.519)

Staff/Student -13.97*** -5.977 -6.553 -7.573 -1.987 -4.426

(2.924) (7.644) (7.542) (6.850) (6.374) (7.001)

Non-District 6.557*** 5.181**

(1.195) (2.067)

Charter 10.34*** 6.392**

(3.188) (2.738)

Magnet -3.938 -0.323

(2.510) (2.275)

Constant 71.78*** 71.54*** 74.15*** 62.26*** 62.46*** 63.88***

(4.592) (5.934) (5.991) (6.171) (6.217) (6.207)

Observations 2,211 2,211 2,211 2,207 2,207 2,207

R-squared 0.634 0.640 0.630 0.628 0.628 0.625

Robust Standard Errors in Parentheses *** p<0.01 ** p<0.05 * p<0.1

some charters have been accused of “cream-skimming” to get the best students (Altonji,

Huang and Taber, 2010, and Cullen and Rivkin, 2003). Another possibility is that

charters classify a high level of students as special education so as to exempt them from

reporting the lowest student scores. These hypotheses could be tested using longitudinal

34

achievement data at the student level to examine whether, on average, students that

matriculated to charters scored higher than their district peers before entering charters, or

whether attending a charter school increases a student’s likelihood of being classified as

special education. Other controls, such as teacher education and salary levels, per pupil

spending, and parent engagement could be included to isolate whether charters

themselves are correlated with higher achievement.

Table VII also demonstrates a potential source of negative bias muting the

coefficient of integration on achievement. During the period in which magnet schools

and integration were used as a reform strategy in Hartford, charter schools were also used

statewide to provide options for poor and minority families.11 Charters improve

academic scores, as is shown by regressions 2 and 5 above, but are highly segregated

(National Association for Public Charter Schools, 2014). In interpreting Table VII and

considering the impact of increased charter presence in Hartford, it is important to

consider that charter schools increase racial isolation (in other words, decrease

integration) but also increase student achievement. This likely muted the effect of

integration on achievement found in Hartford in the regressions from Table V. It also

likely improved the schooling options for minority children in other cities, making the

comparison benchmark higher during years after the proliferation of the charter

movement.

11 According to the National Alliance for Public Charter Schools’ 2014 State-by-State analysis, charters became legal in 1996 and as of 2013–14, there were 18 public charter schools and 6,981 students enrolled in them. 88% of these were located in urban areas, while only 49% of all schools are located in urban areas.

35

POTENTIAL LIMITATIONS

The greatest limitation of this study is that the data is at the district level, which

mutes the effect of integration on individual schools and individual students. Future

studies should use student-level longitudinal data, if possible, to evaluate whether

students performed better in integrated settings than racially-isolated ones.

Another potential threat to the external validity of this study is that magnet

schools are expensive. Even Connecticut is currently engaged in a school funding

lawsuit that may threaten the future of its magnet programs (Courant, 2015). Can other

cities that have lower tax revenues than Hartford afford to create good magnet schools?

If not, are there alternatives that would allow them to follow a voluntary integration

strategy that works?

Attending an integrated school is highly correlated with attending a magnet

school. This opens the internal validity of the study up to a number of threats due to

omitted variable bias. It may be that students who opt into Hartford’s Open Choice

programs have parents who are more committed to their educational outcomes. It may be

that they are smarter, so their parents are more committed to their education. It may be

that they are second and third children of parents who had negative experiences in the

Hartford Public Schools and have committed to ensuring better for their other kids. If any

of these stories were true, it would bias the estimated effect of integration downwards,

towards zero. None of these omitted variables could be controlled for with the data

available.

36

DIRECTIONS FOR FURTHER RESEARCH

Further research should be done to improve the precision and expand the validity of

these results. To confirm that integration has a positive and statistically significant

impact on achievement data, this difference-in-difference model could be replicated using

the following adjustments:

1) Use 2014 and 2015 data to understand the effect of a longer lag and higher levels

of integration. Some of the ways in which integration impacts achievement may

take time to bear out their full effect.

2) Use school- or individual-level achievement results. District-level achievement

data limited the nuance of this analysis. Looking at the effect of attending an

integrated school on school averages or individual scores could be far more

instructive.

3) Change the dependent variable to another metric of student success. Test scores

are one of many metrics of achievement. The effect of integration could be

explored differently by using high school graduation rates, college outcomes,

civic engagement outcomes, or earnings outcomes as the dependent variable

instead.

Based on the positive effect found in this study, another potential question is, what is

the best method for integration? One way to study this relationship would be to look at

the impact of LEA type on achievement. If, as this study found, charters have a consistent

positive association with achievement, such a finding could opt that integration be done

using charters instead of magnets. This question has significant financial sustainability

ramifications as well.

37

Finally, to determine the external validity of school integration as a reform policy, it

would be helpful to study whether city characteristics, such as wealth, size, racial

composition and urban concentration influenced integration’s relationship with student

achievement. This could be done by comparing achievement gains in all cities with

integration policies, and controlling for the characteristics mentioned above. Such a

study could illuminate whether, perhaps, it might be easier to integrate a highly

concentrated city like New York, or harder to integrate a racially dichotomous one.

POLICY IMPLICATIONS

There are three main policy recommendations for the state of Connecticut:

1) Integrate schools using school choice. Based on the positive impacts on

achievement in Hartford, Connecticut should support integration through school

choice. Policies could include easing the regulatory environment for new charter

networks, providing financial or logistical supports to open choice programs in

other cities, and funding new magnet schools.

2) Track school level data in more readily available formats. One key to

understanding the return on investment in Hartford schools is precise, school- or

student-level data on achievement. Connecticut’s data reporting is fragmented

and imprecise. Graduation rates are not available before 2010 and student

achievement scores are missing after 2013. By collecting better data and

improving access for researchers, Connecticut can get a clear picture of how

various policies impact student achievement.

38

3) Commit school funding to support equal opportunities for all children. In a school

funding case currently on the Connecticut Supreme court docket, magnets and

other special programs risk losing their funding. Connecticut should commit to

spending equal amounts on schooling opportunities regardless of a child’s zip

code.

CONCLUSIONS

Increasing integration levels in Hartford, CT by twenty percentage points between

2008 and 2013 had a positive and statistically significant impact on student proficiency

rates in math and reading. Small changes in integration level (suburban districts) had no

effect on proficiency rates. LEA type had an important impact on student achievement,

and the increase in charters in CT likely muted the impact of integration in Hartford

alone. These findings have important policy implications for Connecticut: sustain its

integration efforts through funding and conducive policies, and increase data

transparency for further evaluations. Further research should be done to understand the

longer-term educational and civic outcomes of integration and how its relationship with

achievement varies as city characteristics change.

39

REFERENCES Abrams, L. S., & Gibbs, J. T. 2002. Disrupting the logic of home-school relations: Parent involvement strategies and practices of inclusion and exclusion. Urban Education, 37(3), 384-407.

Ainsworth-Darnell, James W., and Douglas B. Downey. 1998. Assessing the oppositional culture explanation for racial/ethnic differences in school performance. American Sociological Review 63 (August): 536–53.

Altonji, Joseph G., Ching-I Huang and Christopher R. Taber. 2010. Estimating the Cream Skimming Effect of School Choice. NBER Working Paper No. 16579.

Angrist, Joshua D., and Kevin Lang. 2004. Does school integration generate peer effects? Evidence from Boston’s Metco Program. American Economic Review 94 (December): 1613–34.

Archibald, Douglas Albrigh. "Magnet Schools, Voluntary Desegregation, and Public Choice Theory: Limits and Possibilities in a Big City School System". University of Wisconsin-Madison, 1988.

Armor, David J. 1972. The evidence on busing. Public Interest 28 (Summer): 90–126. ———. 1995. Forced justice: School desegregation and the law. New York: Oxford University Press.

Austen-Smith, David, and Roland G. Fryer Jr. 2003. The economics of “acting white.” Working Paper no. 9904, National Bureau of Economic Research, Cambridge, MA.

Bacolod, Marigee P., and Justin L. Tobias. 2006. Schools, school quality, and achievement growth: Evidence from the Philippines. Economics of Education Review 25 (December): 619–32.

Bifulco, Robert et al. “Can Interdistrict Choice Boost Student Achievement? The Case of Connecticut’s Interdistrict Magnet School Program.” 31 Educational Evaluation and Policy Analysis 323 (2009).

Bishop, John H., Ferran Mane, Michael Bishop, and Joan Moriarty. 2001. The role of end-of-course exams and minimal competency exams in standards-based reforms. In Brookings papers on education policy, 2001, ed. Diane Ravitch, 267–345. Washington, DC: Brookings Institution Press.

Boardman, Anthony E., and Richard J. Murnane. 1979. Using panel data to improve estimates of the determinants of educational achievement. Sociology of Education 52 (April): 113–21.

Booth, A., & Dunn, J. (Eds.). (1996). Family-school links. Mahwah, NJ: Lawrence Erlbaum.

40

Boozer, Michael A., Alan B. Krueger, and Shari Wolkon. 1992. Race and school quality since Brown v. Board of Education. In Brooking papers: Microeconomics, ed. Martin Neil Baily and Clifford Winston, 269–338. Washington, DC: Brookings Institution Press.

Brock, William A., and Steven N. Durlauf. 2001. Interactions-based models. In Handbook of econometrics, ed. James J. Heckman and Edward E. Leamer, 3297–3380. Amsterdam: North Holland.

Card, David, and Jesse Rothstein. 2007. Racial segregation and the blackwhite test score gap. Journal of Public Economics 91 (December): 2158–84.

Clark, Kenneth, and Mamie Clark. 1939. The development of consciousness of self and the emergence of racial identity in Negro children. Journal of Social Psychology 10 (November): 591–99.

Clotfelter, Charles T. 1976. School desegregation, “tipping,” and private school enrollment. Journal of Human Resources 22 (Winter): 29–50. ———. 2004. After Brown: The rise and retreat of school desegregation. Princeton, NJ: Princeton University Press.

Clotfelter, Charles T., Helen F. Ladd, and Jacob L. Vigdor. 2003. Segregation and resegregation in North Carolina’s public school classrooms. North Carolina Law Review 81 (May): 1463–511.

Cobb, Casey et al., Legally viable desegregation strategies: The case of Connecticut 131, In Integrating Schools in a Changing Society: New Policies and Legal Options For a Multiracial Generation, Erica Frankenberg and Elizabeth DeBray, eds. (2011)

Coleman, James S., Ernest Q. Campbell, Carol J. Hobson, James McPartland, Alexander M. Mood, Frederic D. Weinfeld, and Robert L. York. 1966. Equality of educational opportunity. Washington, DC: Government Printing Office. 380

Coleman, James S., Sara D. Kelley, and John A. Moore. 1975. Trends in school integration, 1968–73. Urban Institute Paper no. 722–03-01, Urban Institute, Washington, DC.

Cook, Michael D., and William N. Evans. 2000. Families or schools? Explaining the convergence in white and black academic performance. Journal of Labor Economics 18 (October): 729–54.

Cook, Philip J., and Jens Ludwig. 1997. Weighing the burden of “acting white”: Are there race differences in attitudes toward education? Journal of Policy Analysis and Management 16 (Spring): 256–78.

Cook, Thomas D. 1984. What have black children gained academically from school integration? Examination of the meta-analytic evidence. In School desegregation and black achievement, ed. T. Cook, R. Crain, N. Miller, W. Stephan, J. Walberg, and P. Wortman, 6–42. Washington, DC: National Institute of Education.

41

Comer, J. P., Ben-Avie, M., Haynes, N. M., & Joyner, E. T. (1999). “Child by child: The Comer process for change in education.” New York: Teacher’s College Press.

Courant Editorial Staff. 2016. Too Many Hartford Magnet Schools? No, But We Need A Plan. Hartford Courant, 27 Jan. 2016. Web. 9 Apr. 2016. <http://www.courant.com/opinion/editorials/hc-ed-wrong-about-too-many-magnet-schools-0127-20160127-story.html>.

Crain, Robert. 1970. School integration and occupational achievement of Negroes. American Journal of Sociology 75 (January): 593–606.

Crain, Robert L., and Rita E. Mahard. 1978. Desegregation and black achievement: A review of the research. Law and Contemporary Problems 42:17–53.

Cullen, Julie Berry and Steven G. Rivkin. 2003. The Role of Special Education in School Choice. University of Chicago Press. URL: http://www.nber.org/chapters/c10086 Cutler, David M., Edward L. Glaeser, and Jacob L. Vigdor. 1999. The rise and decline of the American ghetto. Journal of Political Economy 107 (June): 455–506.

de Carvalho, M. E. (2001). Rethinking family school relations: A critique of parental involvement in schooling. New York: Teacher’s College Press.

Delaney, Stephen B. “Milo Sheff et al. v. William A. O'Neill et al,” 28 Equity & Excellence in Education 14 (1995)

Dougherty, Jack, et al. “Sheff v. O'Neill": Weak Desegregation Remedies and Strong Disincentives in Connecticut, 1996-2008, 103-127 In From the Courtroom to the Classroom: The Shifting Landscape of School Desegregation, Claire Smrekar and Ellen B. Goldring, eds. (March 2009).

Susan E. Eaton and Gary A. Orfield, Continuing the Struggle for Desegregated Schools, In Readings on Equal Education, Volume 13: Forty Years After the Brown Decision - Implications of School Desegregation for U.S. Education, Kofi Lomotey and Charles Teddlie, eds. 5 (1996).

Susan E. Eaton, The Children in Room E4: American Education on Trial, (2006)

Ehrenberg, Ronald G., and Dominic J. Brewer. 1995. Did teachers’ verbal ability and race matter in the 1960s? Coleman revisited. Economics of Education Review 14 (March): 1–21.

Ferguson, Ronald F. 1998a. Comment [on Cook and Ludwig]. In The black-white test score gap, ed. Christopher Jencks and Meredith Phillips, 394–97. Washington, DC: Brookings Institution Press.

42

———. 1998b. Teachers’ perceptions and expectations and the black-white test score gap. In The black-white test score gap, ed. Christopher Jencks and Meredith Phillips, 273–317. Washington, DC: Brookings Institution Press.

———. 2001. A diagnostic analysis of black-white GPA disparities in Shaker Height, Ohio. In Brookings papers on education policy, 2001, ed. Diane Ravitch, 347–414. Washington, DC: Brookings Institution Press.

Ferguson, Ronald F., and Helen F. Ladd. 1996. How and why money matters: An analysis of Alabama schools. In Holding schools accountable: Performance-based reform in education, ed. Helen F. Ladd, 265–98. Washington, DC: Brookings Institution Press.

Fordham, Signithia, and John Ogbu. 1986. Black students’ school success: Coping with the burden of “acting white.” Urban Review 18, no. 3: 176–206.

Fryer, Roland G., Jr., and Steven D. Levitt. 2003. The causes and consequences of distinctively black names. Working Paper no. 9938, National Bureau of Economic Research, Cambridge, MA.

Grant, Ellsworth Strong, and Grant, Marion Hepburn, "The City of Hartford, 1784-1984," Connecticut Historical Society, 1986.

Griliches, Zvi, and Jerry A. Hausman. 1986. Errors in variables in panel data. Journal of Econometrics 31 (February): 93–118.

Grogger, Jeffrey T. 1996. Does school quality explain the recent black/ white wage trend? Journal of Labor Economics 14 (April): 231–53.

Guryan, Jonathan. 2004. Desegregation and black dropout rates. American Economic Review 94 (September): 919–43.

Hanushek, Eric A. 1972. Education and race: An analysis of the educational production process. Cambridge, MA: Health-Lexington.

Hanushek, Eric A., John F. Kain, Jacob M. Markman, and Steven G. Rivkin. 2003. Does peer ability affect student achievement? Journal of Applied Econometrics 18 (September/October): 527–44.

Hanushek, Eric A., John F. Kain, Daniel M. O’Brien, and Steve G. Rivkin. 2005. The market for teacher quality. Working Paper no. 11154, National Bureau of Economic Research, Cambridge, MA.

Hanushek, Eric A., John F. Kain, and Steve G. Rivkin. 2002a. Inferring program effects for specialized populations: Does special education raise achievement for students with disabilities? Review of Economics and Statistics 84 (November): 584–99.

43

———. 2002b. New evidence about Brown v. Board of Education: The complex effects of school racial composition on achievement. Working Paper no. 8741, National Bureau of Economic Research, Cambridge, MA.

———. 2004a. Disruption versus Tiebout improvement: The costs and benefits of switching schools. Journal of Public Economics 88, nos. 9–10: 1721–46.

———. 2004b. Why public schools lose teachers. Journal of Human Resources 39, no. 2:326–54.

Hanushek, Eric A., and Margaret E. Raymond. 2005. Does school accountability lead to improved student performance? Journal of Policy Analysis and Management 24 (Spring): 297–327.

Hanushek, Eric A., and Steve G. Rivkin. 2009. Harming the best: How schools affect the black-white achievement gap. Journal of Policy Analysis and Management 28 (Summer): 366–93.

Holmes, Gloria Graves and Susan Clarke. “To Choose or Not to Choose: Equity in Connecticut in the Wake of Sheff v. O'Neill.” 38 Equity and Excellence in Education 3 (February 2005).

Hoxby, Caroline Minter. 2000. Peer effects in the classroom: Learning from gender and race variation. Working Paper no. 7867, National Bureau of Economic Research, Cambridge, MA.

———. 2003: The Economics of School Choice. National Bureau of Economic Research. University of Chicago Press, February 22-24, 2001.

Iceland, John, and Daniel H. Weinberg. 2002. Racial and ethnic residential segregation in the United States, 1980–2000. Census 2000 Special Reports. Washington, DC: U.S. Bureau of the Census.

Kahlenberg, Richard D. “From All Walks of Life: New Hope for School Integration.” American Educator, Winter 2012-2013. Web: http://www.aft.org/sites/default/files/periodicals/Kahlenberg.pdf. Accessed October 30, 2015.

Kahlenberg, Richard D., ed. The Future of School Integration: Socioeconomic Diversity

as an Education Reform Strategy. The Century Foundation Press, February 27, 2012.

Kahlenberg, Richard D. (02/01/2001). All Together Now: The Case for Economic

Integration of the Public Schools. Brookings Institution Press. . (ISBN: 0-8157-4810-8, 978-0-8157-4810-6).

Kain, John F., and Joseph J. Persky. 1969. Alternatives to the gilded ghetto. Public Interest 14 (Winter): 74–87.

44

Kling, Jeffrey R., Jeffrey B. Liebman, and Lawrence F. Katz. 2007. Ex- 382 Hanushek et al. perimental analysis of neighborhood effects. Econometrica 75 (January): 83–119.

Lacour, Misty and Laura D. Tissington. 2011. The effects of poverty on academic achievement. Educational Research and Reviews Vol. 6 (7), pp. 522-527. Linn, Robert L., and Kevin G. Welner, eds. 2007. Race-conscious policies for assigning students to schools: Social science research and the Supreme Court cases. Washington, DC: National Academy of Education.

Ludwig, Jens, Helen F. Ladd, and Greg J. Duncan. 2001. Urban poverty and educational outcomes. In Brookings-Wharton papers on urban affairs, 2001, ed. William G. Gale and Janet Rothenberg Pack, 147–88. Washington, DC: Brookings Institution Press.

Manski, Charles F. 1993. Identification of endogenous social effects: The reflection problem. Review of Economic Studies 60 (July): 531–42.

McDermott, Kathryn A. et al. “Have Connecticut's Desegregation Policies Produced Desegregation?” 35 Equity & Excellence in Education 18 (2002).

McWhorter, John H. 2000. Losing the race: Self-sabotage in black America. New York: Free Press.

Moffitt, Robert A. 2001. Policy interventions, low-level equilibria, and social interactions. In Social dynamics, ed. Steven Durlauf and H. Peyton Young, 45–82. Cambridge, MA: MIT Press.

NAEP 1996 mathematics: Crossstate data compendium for the grade 4 and grade 8 assessment. Washington, DC: National Center for Education Statistics.

———. 2000. NAEP 1996 trends in academic progress. Rev. ed. Washington, DC: National Center for Education Statistics.

Ogbu, John U. 2003. Black American students in an affluent suburb: A study of academic disengagement. Mahwah, NJ: Erlbaum.

Orfield, Gary, and Jongyeon Ee. “Connecticut School Integration: Moving forward as the Northeast Retreats.” The Civil Rights Project, April 2015.

Oreopoulos, Philip. 2003. The long-run consequences of living in a poor neighborhood. Quarterly Journal of Economics 118 (November): 1533–75.

Parker, Dennis. “Racial and Ethnic Isolation, Post ‘Sheff v. O'Neill’.” 32 Equity & Excellence in Education 5 (1999)

Price, Janet R. and Jane R. Stern. 1986. Magnet Schools as a Strategy for Integration and School Reform. Yale Law & Policy Review, Vol. 5 Issue 2, Article 4, 1986.

45

Purcell-Gates, Victoria, McIntyre, Ellen, and Freppon, Penny A. “Learning Written Storybook Language in School: A Comparison of Low-SES Children in Skills-Based and Whole Language Classrooms.”American Educational Research Journal. Vol. 32, No. 3 (Autumn, 1995), pp. 659-685. Racial isolation in the public schools. Washington, DC: Government Printing Office. U.S. Department of Education. 1997.

Reber, Sarah J. 2005. Court-ordered desegregation: Successes and failures integrating American schools since Brown versus Board of Education. Journal of Human Resources 40 (Summer): 559–90.

Rivkin, Steven G. 1994. Residential segregation and school integration. Sociology of Education 67 (October): 279–92.

———. 2000. School desegregation, academic attainment, and earnings. Journal of Human Resources 35 (Spring): 333–46.

———. 2005. Cumulative nature of learning and specification bias in education research. Photocopy, Department of Economics, Amherst College.

Rivkin, Steven G., Eric A. Hanushek, and John F. Kain. 2005. Teachers, schools, and academic achievement. Econometrica 73 (March): 417–58.

Rivkin, Steven G., and Finis Welch. 2006. Neighborhood segregation and school integration. In Handbook of the economics of education, ed. Eric A. Hanushek and Finis Welch, 1019–49. Amsterdam: North Holland.

Rizzo, Kevin. "Crime in America: Top 10 Most Dangerous Cities Under 200,000." Law Street. N.p., 25 Nov. 2013. Web. 9 Apr. 2016. <http://lawstreetmedia.com/blogs/crime/10-dangerous-small/>. Rosenbaum, James E. 1995. Changing the geography of opportunity by expanding residential choice: Lessons from the Gautreaux Program. Housing Policy Debate 6, no. 1:231–69. Rothstein, Jesse. 2008. Teacher quality in educational production: Tracking, decay, and student achievement. Working Paper no. 14442, National Bureau of Economic Research, Cambridge, MA.

Rumberger, Russell W. and Gregory J. Palardy. “Indicators of High School Performance Test Scores, Dropout Rates, and Transfer Rates as Alternative Indicators of High School Performance.” American Educational Research Journal Vol. 42, No. 3 (2005).

Schofield, Janet Ward. 1995. Review of research on school desegregation’s impact on elementary and secondary school students. In Handbook of School Racial Composition and Achievement 383 research on multicultural education, ed. James A. Banks and Cherry A. McGee Banks, 597–616. New York: Macmillan.

46

Sekou, Bilal (aka Darryl L. McMiller). “ ‘I Support School Integration, But…’ Sheff v.

O'Neill More than Ten Years Later and No End in Sight.” Equity & Excellence in Education, 97 (2009).

The Sheff Movement: Qualified Integrated Education for All Children, 2014. Web. http://sheffmovement.org/ Accessed: October 30, 2015.

Steele, Claude M., and Joshua Aronson. 1998. Stereotype threat and the test performance of academically successful African Americans. In The black-white test score gap, ed. Christopher Jencks and Meredith Phillips, 401–27. Washington, DC: Brookings Institution Press.

Thernstrom, Stephan, and Abigail Thernstrom. 2003. No excuses: Closing the racial gap in learning. New York: Simon & Schuster.

Tobias, Justin L. 2004. Assessing assessments of school performance: The case of California. American Statistician 58, no. 1:55–63.

Todd, Petra E., and Kenneth I. Wolpin. 2003. On the specification and estimation of the production function for cognitive achievement. Economic Journal 113 (February): F3–F33. U.S. Commission on Civil Rights. 1967.

Welch, Finis, and Audrey Light. 1987. New evidence on school desegregation. Washington, DC: U.S. Commission on Civil Rights.

Wooldridge, Jeffrey M. 2002. Econometric analysis of cross section and panel data. Cambridge, MA: MIT Press.

Yazze-Mintz, Ethan. "Voices of Students on Engagement: A Report on the 2006 High School Survey of Student Engagement, Center for Evaluation and Education Policy, Indiana University, 2007." Center for Evaluation and Education Policy, Indiana University. ERIC, 1 Jan. 2007. Web. 09 Apr. 2016. <http://eric.ed.gov/?id=ED495758>.