Fall 22

District-LevelSchoolChoiceandRacialEthnicTestScoreGaps.pdf

District-Level School Choice and Racial/Ethnic Test Score Gaps

Lorraine Blatt Elizabeth Votruba-Drzal University of Pittsburgh

The rapid expansion of school choice is restructuring public education in the United States. This study examines associations between charter and magnet school enrollment, White-Black and White-Hispanic segregation, and test score gaps at the district level from 2009 to 2015 in third to eighth grade using the Stanford Education Data Archive and the U.S. Department of Education’s Common Core of Data. Robust findings indicate that higher charter school enrollment is associated with larger White-Black test score gaps and this effect is mediated by White-Black segregation. There is also evidence that magnet school enrollment is associated with White-Hispanic test score gaps. Overall, this study suggests that the expansion of school choice may have negative implications for structural education equity.

KEYWORDS: charter schools, magnet schools, school segregation, achieve- ment gap

School choice is restructuring public education in the United States in the form of open enrollment, means tested vouchers, magnet schools, and

charter schools. Magnet and charter schools in particular are expanding rap- idly. Between 2001 and 2017, the number of students in the United States attending a magnet or charter more than tripled (National Center for Education Statistics [NCES], 2018). As of 2017, over 2.5 million public school students attended magnets and over 3 million attended charters, together comprising over 11% of all public school students in the United States

LORRAINE BLATT is a doctoral student in the developmental psychology program at the University of Pittsburgh, 3420 Forbes Avenue, Pittsburgh, PA 15260, Area Cube 634; email: [email protected]. Her research interests center on how structural inequities and education policies affect child development.

ELIZABETH VOTRUBA-DRZAL is a professor of developmental psychology and a senior sci- entist at the Learning Research and Development Center at the University of Pittsburgh. Her scholarship is interdisciplinary and focuses on how socioeconomic status and race/ethnicity interface to shape learning opportunities and development over the life course.

American Educational Research Journal

December 2021, Vol. 58, No. 6, pp. 1178–1224

DOI: 10.3102/0002831221999405

Article reuse guidelines: sagepub.com/journals-permissions

� 2021 AERA. https://journals.sagepub.com/home/aer

http://crossmark.crossref.org/dialog/?doi=10.3102%2F0002831221999405&domain=pdf&date_stamp=2021-04-19

(NCES, 2018). In many districts, magnets and charters encompass a significant proportion of the available public schools. For example, Los Angeles, Miami- Dade, and Houston school districts all have over 100 magnets (Polikoff & Hardaway, 2017) and in Washington D.C., Detroit, and New Orleans, charters make up more than 40% of public school enrollment (National Alliance for Public Charter Schools [NAPCS], 2012).

Students typically apply and are admitted to magnet and charter schools via lottery, or in some cases, on a first come first serve or test score basis (Gleason et al., 2010; Goldring & Swain, 2019). Magnet schools typically offer a thematic curriculum and aim to promote racial/ethnic integration by enroll- ing students from multiple catchment areas within or across districts (Goldring & Swain, 2019). The missions of charter schools range from serving as lab schools to test novel teaching pedagogies to increasing competition in public education markets, which is hypothesized to incentivize public school improvement (Mickelson et al., 2013; NAPCS, 2012). Magnets are predomi- nantly operated under the jurisdiction of the same public school districts as traditional public schools while charters usually have their own governing bodies that can include for-profit educational management organizations (Ertas & Roch, 2014; Mickelson et al., 2013). Thus, these schools likely have distinct impacts on district-level academic outcomes due to differences in enrollment, governing, accountability, and funding structures.

Along with the rise of school choice is an increase in research examining the efficacy of magnet and charter schools, which is largely conducted at the local level and provides important insights to inform local school choice policy. However, when considering federal policy decisions and federal incentives and guidelines for states and districts, it is also important to understand how school choice influences district-level academic outcomes on a national scale. Furthermore, looking between schools within a district does less to advance a broader understanding of what contributes to structural education inequity because evidence suggests that most of what explains variance in achievement disparities occurs between rather than within districts (Reardon et al., 2019a).

Structural education inequity, which in this article refers specifically to the institutional and systemic oppression of Black and Hispanic students concur- rent with the privileging of non-Hispanic White students in public education, remains a critical issue in the United States (Nieto, 2005). Persistent gaps between White students’ test scores and the test scores of Black and Hispanic students on a macro scale signal inequitable access to opportunity, resources, and support more broadly and therefore serve as one indicator of structural education inequity (Reardon et al., 2019b).

There has been a general decline in racial/ethnic test score gaps since 1940; however, from 1990 to 2000 both White-Black and White-Hispanic test score gaps widened or stabilized (Lee, 2002). In the 2014–2015 school year, Black students in third through eighth grade scored an average of 1 SD below White students on standardized math and English language arts

School Choice and Racial/Ethnic Test Score Gaps

1179

(ELA) tests. Hispanic students scored between 0.7 and 0.8 SD on average below White students on math and ELA tests, respectively (author calculations from the Stanford Education Data Archive (SEDA; Reardon et al., 2018)).

Test score gaps vary significantly by school district, and patterns of racial/ ethnic segregation are among the strongest correlates of these gaps (Reardon et al., 2019a). Prior research suggests that the influx of school choice is asso- ciated with increases in segregation at the district level (Frankenberg et al., 2010; Harris, 2018; Monarrez et al., 2019). Therefore, this study uses SEDA ver- sion 2.1, a publicly available national data set, to investigate whether school choice enrollment is associated with test score gaps at the district level in third through eighth grade, and whether racial/ethnic segregation mediates this association.

Background

School Choice and Student Test Scores

Overall, there is great variability in associations between school choice and students’ test performance in elementary and middle school, likely due to heterogeneity in school choice landscapes across different contexts. Methodological approaches to studies examining the impact of school choice on students’ academic outcomes also vary widely, particularly in their treat- ment of comparison groups and their attention to omitted variable bias (e.g., Rapa et al., 2018). Here, we summarize the results of selected large-scale studies that address threats to internal validity through research design or ana- lytic approach.

A quasi-experimental examination of five urban school districts in four states found no overall effect of magnet schools in grades K–12 (Wang et al., 2017). Differences in academic outcomes between magnet and tradi- tional public school students varied widely depending on a range of school-level factors, including student engagement, professional develop- ment, teacher support, and partnerships (Wang et al., 2017). Another study used value-added models with student fixed effects to examine effects of mag- net schools in a large southwestern school district (Harris, 2019) and found that magnet students in first through eighth grade had the same or worse test scores compared with students in traditional public schools. Some of the effects were related to peer composition on dimensions of race, poverty, and academic track (Harris, 2019).

Large-scale charter school studies also yield mixed findings at the local level due to contextual differences. For example, two causal studies leverag- ing charter middle school lotteries as instrumental variables, one examining 33 schools in 13 states (Clark et al., 2015) and another examining 28 schools across 15 states (Gleason et al., 2010), found no overall achievement effect for charter attendance; some charters’ students exhibited better test scores and

Blatt, Votruba-Drzal

1180

some worse, compared with other students who applied. These studies found variation in test scores related to urbanicity and socioeconomic status.

When discussing district-level education equity, it is also essential to con- sider students who remain in traditional public schools in districts where choice is expanding. The indirect impacts of school choice on traditional pub- lic school students also vary depending on context. For example, evidence in a large southwestern district using an instrumental variables approach finds that charter expansion undermines the test performance of elementary stu- dents in traditional public schools (Imberman, 2011). However, a study in New York City employing a difference-in-difference approach found elemen- tary students in traditional public schools experienced slight increases in test scores resulting from the presence of charters (Cordes, 2018).

Consequently, there is evidence that school choice can be disadvanta- geous, advantageous, or unrelated to the test performance of students both in public schools of choice and in traditional public schools. The disparate impacts of school choice merit additional research to delineate how these effects map onto differences in racial/ethnic test score gaps across districts throughout the United States. It is also important to better understand the mech- anisms that drive differences in associations between school choice enrollment and test score gaps. One potential mechanism is district segregation.

School Choice and Segregation

By some measures, U.S. public schools are at least as segregated by race/ ethnicity as they were in 1954, when Brown v. Board of Education deemed segregation unconstitutional (Donnor & Dixson, 2013). As of 2016, over 18% of schools were 90% to 100% non-White, up from 6% of schools in 1988 (Frankenberg et al., 2019). While much of this change is attributable to factors such as the increasing racial and ethnic heterogeneity of the U.S. pop- ulation and exclusionary zoning, it is also predicted by the proliferation of school choice policies that can serve as catalysts for racial/ethnic segregation (Donnor & Dixson, 2013; Fiel, 2013; Rothwell, 2012).

National studies of districts with charter schools find that charter students are more likely to be in segregated schools and that charter growth relates to increases in segregation (Monarrez et al., 2019; Vasquez Heilig et al., 2019). Additionally, even for charters that disproportionately enroll Black or Hispanic students, transfers of Black and Hispanic students from traditional public schools to charter schools can contribute to district-level segregation (Bifulco & Ladd, 2007; Garcia, 2008; Kotok et al., 2017). Studies find that char- ter expansion is especially predictive of White-Black segregation (Fiel, 2013; Frankenberg et al., 2010).

There is less research linking magnet schools to increasing segregation at the district level, likely because many magnet programs were established to address federal desegregation mandates (Goldring & Swain, 2019). Magnet

School Choice and Racial/Ethnic Test Score Gaps

1181

schools became an especially useful tool for school districts that had to rely on a metropolitan remedy due to Milliken v. Bradley; these were metropolitan districts with desegregation orders that did not have the option of interdistrict mandates despite surrounding suburban areas actively contributing to de facto segregation (Gordon, 1994). However, it is possible for a district to become more segregated, even as its magnet schools become more integrated (Harris, 2018). For example, in some urban, predominantly non-White school districts, magnet schools attract a majority of the White students, which fosters within-school integration in a few schools while amplifying segregation at the district level (Kimelberg & Billingham, 2013). Furthermore, while federally mandated desegregation orders and metropolitan remedies established an ini- tial relationship between magnets and integration, once districts began to achieve unitary status, the role of magnets shifted away from integration (Goldring & Swain, 2019; Harris, 2019). In fact, Supreme Court rulings starting in the mid-1990s, and continuing with more recent cases such as Parents Involved in Community Schools v. Seattle, increased restrictions on the consid- eration of race/ethnicity in public school enrollment overall, particularly for dis- tricts with unitary status (Goldring & Smrekar, 2002; Goldring & Swain, 2019; Harris, 2019). Thus, many magnet schools are increasing their focus on higher achievement while decreasing integration efforts (Harris, 2019), and in some cities magnets are increasingly racially isolated (e.g., Grooms & Williams, 2015). Therefore, when examining associations between school choice and structural education inequity, attention to magnet schools is crucial.

While much of school segregation is rooted in neighborhood segregation (Rothwell, 2012), school choice also independently contributes to segregation. In fact, many districts in the United States are becoming more racially/ethnically diverse while the schools in those districts are concurrently becoming more segregated (Coughlan, 2018; Mader et al., 2018; Monarrez, 2018; Siegel- Hawley, 2014). There is also evidence that as White students’ neighborhoods become more racially/ethnically heterogenous, White students are more likely to apply to magnets and charters, leading to increased school segregation (Goldring & Swain, 2019; Renzulli & Evans, 2005). One reason for this may be that for some parents, the idea of sending their children to the traditional public schools deterred them from living in certain neighborhoods, and magnet or charter schools potentially remove that deterrent. This is supported by evi- dence that an influx of school choice is associated with rising gentrification (Pearman & Swain, 2017). Persistent links between school segregation and structural education inequity suggest that segregation resulting from the prolif- eration of school choice likely has implications for racial/ethnic test score gaps.

Segregation and Test Score Gaps

School segregation by race/ethnicity is one of the largest correlates of district-level test score gaps (Reardon et al., 2019a, 2019b). Most evidence

Blatt, Votruba-Drzal

1182

finds that school segregation is associated with lower resourced schools and inferior school quality (e.g., facilities, range of course offerings, access to counselors, student to teacher ratio, and teacher qualifications, absenteeism, and retention) for minoritized students, which affects educational outcomes into adulthood (Johnson, 2019). For example, for Black students in particular, a 15% increase in segregation for at least half of a student’s schooling years is associated with a 7–percentage point decrease in likelihood of college atten- dance, a 7% reduction in earnings, and a 3.5–percentage point increase in likelihood of incarceration (Johnson, 2019).

In general, school segregation concentrates social and economic disad- vantage in the lowest resourced schools that are also majority Black and or Hispanic (Rothstein, 2015). Thus, when school choice segregates districts by race/ethnicity this is accompanied by socioeconomic segregation and cou- pled with an inequitable distribution of resources. For example, Bifulco et al. (2009) found that students with college educated parents were more likely to choose a magnet or charter if their traditionally assigned public school had a large proportion of students without college educated parents. Ni (2012) found that in predominantly urban and low-income areas, student transfers from traditional public schools to charters resulted in both the charter and tra- ditional public school enrollment becoming stratified by race, socioeconomic, and special education status (Ni, 2012). In this case, charter school expansion segregated minoritized, low-income, and special needs students in the most underresourced traditional public schools (Ni, 2012). Hence, if the expansion of school choice segregates school districts in a way that further concentrates disadvantage in the most underresourced schools where the enrollment is majority Black and or Hispanic students, then White-Black and White- Hispanic test score gaps may widen (Bifulco & Ladd, 2007).

There is also evidence that racial/ethnic school segregation has dispropor- tionately negative impacts on the academic performance of minoritized stu- dents independently of socioeconomic status (e.g., Hanushek et al., 2009; Mickelson, 2001). Research examining math test score gaps over thirty years finds that despite increases in economic mobility for Hispanic and Black fami- lies, school segregation is also growing which leads to racial/ethnic isolation and results in significant increases in test score gaps (Berends & Peñaloza, 2010). Hence, segregation may also widen test score gaps due to processes related to racial/ethnic inequity beyond gaps in schools’ socioeconomic resour- ces, such as unequal social capital (Bankston & Caldas, 1996), teachers’ racial biases (T. M. Scott et al., 2019), racist disciplinary practices (Skiba, 2015), and language barriers for Spanish-speaking students (Reardon & Galindo, 2009).

In addition to the negative associations between segregation and educa- tion equity, there are positive benefits of integration. Studies find that inte- grated schools improve academic outcomes for Black students in particular, with positive or null effects for White students (Johnson, 2019; Linn & Welner, 2007). Beyond academic outcomes, integration also reduces White

School Choice and Racial/Ethnic Test Score Gaps

1183

students’ racial prejudice, increases the racial diversity of their social net- works, and shapes students’ political identities in adulthood (Billings et al., 2020; Johnson, 2019). Thus, not only is the exacerbation of segregation actively detrimental to education equity, but efforts intentionally focused on integration have the capacity to narrow disparities in academic outcomes across race and ethnicity while improving equity.

However, there is also some evidence that school choice has the capacity to improve educational outcomes for minoritized students, despite increasing segregation, in instances where schools of choice have more resources than traditional public schools (Whitehurst et al., 2016). In other words, Black and Hispanic students in racially/ethnically homogenous schools of choice may fare better than they would in more heterogeneous traditional public schools if the schools of choice have greater resources and are of superior quality to the traditional public schools students would otherwise be enrolled in. If school choice is primarily working to improve the quality of education for Black and Hispanic students, then district-level White-Black and White- Hispanic test score gaps may narrow as a result of school choice expansion (Dobbie & Fryer, 2011; Whitehurst et al., 2016). Given these competing pos- sibilities, it is important to examine links between school choice, segregation, and test score gaps to better understand how this widespread public educa- tion reform is associated with structural education inequity.

The Current Study

The current study examines the relationship between district-level school choice and test score gaps in third through eighth grades for 4,613 school dis- tricts in the United States from 2009 to 2015 by addressing two main research questions:

Research Question 1: What is the association between magnet and charter school enrollment and White-Black and White-Hispanic math and ELA test score gaps at the district level?

Research Question 2: Are there indirect effects of magnet and charter school enroll- ment on White-Black and White-Hispanic math and ELA test score gaps at the district level that operate through White-Black and White-Hispanic segregation?

Method

Stanford Education Data Archive Version 2.1

Most of the data analyzed in this study come from SEDA (Reardon et al., 2018), a publicly available data set that includes national district-level data for grades three through eight from the 2008–2009 to 2014–2015 school years. SEDA includes data for 12,065 school districts in total which enroll roughly

Blatt, Votruba-Drzal

1184

35 million third through eighth graders. The cases included in SEDA represent 89.5% of all possible district-level cases across subject, grade, and year. The 10.5% of missing cases in the full SEDA are due to suppressed data for districts with less than 95% participation for standardized tests, suppressed individual estimates with a standard error greater than 2 SDs on the state-standardized scale, test score data not reported to EdFacts, and data identified as incorrect due to data entry errors.

Participants

There are 4,613 school districts in SEDA with enough racial/ethnic hetero- geneity to measure White-Black and or White-Hispanic math and or ELA test score gaps. These districts represent 38% of the districts included in SEDA, but they serve almost 29 million third through eighth graders. This is 83% of all public school students in third through eighth grade enrolled in districts included in SEDA and an even larger proportion of Black and Hispanic stu- dents. Districts are considered to have adequate racial/ethnic heterogeneity if they have at least 20 students of each racial/ethnic subgroup in at least one grade in third through eighth grade in at least one school year from 2008– 2009 to 2014–2015 (e.g., for White-Black test score gaps, districts must have at least 20 Black students and 20 White students in at least one grade in at least one year). SEDA chose a 20 student minimum based on the disclosure risk agreement with the National Center for Education Statistics. Of the 4,613 dis- tricts that meet the criteria described above, 95% of observations have complete data for every variable analyzed in the study. To account for the 5% of missing data, the study used multiple imputation methods with chained equations to create 20 complete data sets in Stata 15.0 (Royston, 2004, 2005).

A total of 2,171 of the districts in the analyses have enough heterogeneity to examine both White-Black and White-Hispanic math and or ELA test score gaps. An additional 743 districts only have enough heterogeneity to examine White-Black math and or ELA test score gaps (resulting in a total of 2,914 dis- tricts analyzed for the White-Black test score gaps (math and reading test score gaps have different sample sizes due to differences in available data per test per district)), and an additional 1,699 districts only have enough heterogeneity to examine White-Hispanic math and or ELA test score gaps (resulting in a total of 3,870 districts analyzed for the White-Hispanic test score gaps). Of the total 4,613 districts, about 31% have school choice in the form of magnets and or charters (Table 1), compared with only 15% of the 12,065 districts in SEDA overall. Of the 31%, 7% have both magnets and charters, 4% have magnets and no charters, and 20% have charters and no magnets.

Table 1 highlights a range of differences in districts based on school choice, which extend beyond the presence or absence of magnets and char- ters. For example, 7% of districts without magnets and charters are urban com- pared with 51% of districts with both magnets and charters. Additionally,

School Choice and Racial/Ethnic Test Score Gaps

1185

T a b le

G ro

u p

M e a n

s b

y S

c h

o o

l C

h o

ic e

C o

m p

o s it

io n

F ro

m 2 0 0 9

to 2 0 1 5

fo r

P u

b li

c S

c h

o o

l D

is tr

ic ts

W it

h R

a c ia

l/ E

th n

ic H

e te

ro g

e n

e it

D is

tr ic

ts W

it h

N o

M ag

n e t o r

C h ar

te r Sc

h o o ls

(n =

3 ,2

2 6 )

D is

tr ic

ts W

it h

B o th

M ag

n e t an

C h ar

te r Sc

h o o ls

(n =

3 0 4 )

D is

tr ic

ts W

it h

Ju st

M ag

n e t

Sc h o o ls

(n =

1 8 1 )

D is

tr ic

ts W

it h

Ju st

C h ar

te r

Sc h o o ls

(n =

9 0 2 )

M SD

M e an

d is

tr ic

t m

at h

te st

sc o re

(N A E P

sc al

e )

2 5 3 .7

6 1 4 .4

5 2 4 7 .1

4 1 1 .5

8 2 4 8 .9

0 1 3 .4

1 2 4 9 .8

3 1 3 .1

M e an

W h it e -B

la ck

m at

h te

st sc

o re

g ap

1 8 .8

3 7 .1

4 2 3 .0

8 8 .2

0 2 1 .2

6 7 .9

2 1 9 .6

6 7 .3

M e an

W h it e -H

is p an

ic m

at h

te st

sc o re

g ap

1 2 .9

5 7 .1

5 1 6 .8

0 8 .1

6 1 5 .2

8 7 .8

0 1 3 .8

3 6 .9

M e an

d is

tr ic

t E LA

te st

sc o re

(N A E P

sc al

e )

2 3 5 .9

8 1 4 .8

2 2 2 8 .9

6 1 2 .3

0 2 3 2 .4

6 1 3 .4

2 2 3 1 .5

1 1 3 .0

M e an

W h it e -B

la ck

E LA

te st

sc o re

g ap

1 9 .8

0 7 .9

2 2 4 .8

5 9 .9

0 2 3 .2

5 9 .2

2 2 0 .8

3 8 .4

M e an

W h it e -H

is p an

ic E LA

te st

sc o re

g ap

1 6 .2

1 8 .3

8 2 1 .5

9 1 0 .4

9 1 9 .9

0 9 .9

9 1 8 .1

9 8 .2

M e an

d is

tr ic

t e n ro

ll m

e n t/

1 0 0 0

3 .2

4 3 .6

8 2 8 .6

3 6 0 .0

2 1 0 .1

2 1 5 .0

1 8 .6

8 1 2 .2

M e an

so ci

o e co

n o m

ic st

at u s

co m

p o si

te 0 .0

9 0 .9

9 2

0 .4

4 1 .0

4 2

0 .2

2 1 .0

5 2

0 .1

1 0 .9

P e r p u p il

e x p e n d it u re

s/ 1 0 0 0

1 2 .4

8 4 .2

3 1 2 .1

0 4 .0

8 1 3 .0

3 4 .0

6 1 1 .3

8 3 .6

G in

i co

e ff ic

ie n t

0 .3

7 0 .0

6 0 .4

1 0 .0

5 0 .3

9 0 .0

5 0 .3

8 0 .0

P e rc

e n t N

at iv

e A m

e ri ca

n p e r g ra

d e

1 4

1 2

1 1

1 4

P e rc

e n t A si

an p e r g ra

d e

4 7

4 6

5 6

4 6

P e rc

e n t H

is p an

ic p e r g ra

d e

2 0

2 1

2 7

2 2

2 1

2 2

2 9

2 5

P e rc

e n t B la

ck p e r g ra

d e

1 3

1 7

2 7

2 5

2 4

1 4

1 8

P e rc

e n t W

h it e

p e r g ra

d e

6 2

2 2

4 1

2 4

4 8

2 4

5 2

2 6

P e rc

e n t o f st

u d e n ts

w h o

ar e

E LL

6 8

1 0

9 8

9 1 0

1 0

W h it e -B

la ck

se g re

g at

io n

0 .0

4 0 .0

6 0 .1

7 0 .1

4 0 .0

9 0 .0

9 0 .1

0 0 .1

W h it e -H

is p an

ic se

g re

g at

io n

0 .0

3 0 .0

5 0 .1

1 0 .0

9 0 .1

0 0 .0

8 0 .0

W h it e -B

ac k

d if fe

re n ce

in fa

m il y

in co

m e

0 .6

4 0 .5

2 0 .6

9 0 .4

1 0 .6

7 0 .3

8 0 .6

3 0 .4

(c o n

ti n

u ed

)

1186

T a b le

1 (c

o n

ti n

u e d

)

D is

tr ic

ts W

it h

N o

M ag

n e t o r

C h ar

te r Sc

h o o ls

(n =

3 ,2

2 6 )

D is

tr ic

ts W

it h

B o th

M ag

n e t an

C h ar

te r Sc

h o o ls

(n =

3 0 4 )

D is

tr ic

ts W

it h

Ju st

M ag

n e t

Sc h o o ls

(n =

1 8 1 )

D is

tr ic

ts W

it h

Ju st

C h ar

te r

Sc h o o ls

(n =

9 0 2 )

M SD

W h it e -H

is p an

ic d if fe

re n ce

in fa

m il y

in co

m e

0 .6

1 0 .4

8 0 .6

8 0 .3

7 0 .7

1 0 .3

8 0 .6

2 0 .4

W h it e -B

la ck

d if fe

re n ce

in p ar

e n t e d u ca

ti o n

0 .1

9 0 .4

4 0 .2

8 0 .3

6 0 .1

8 0 .3

8 0 .1

6 0 .4

W h it e -H

is p an

ic d if fe

re n ce

in p ar

e n t e d u ca

ti o n

0 .6

2 0 .4

9 0 .7

4 0 .3

9 0 .6

6 0 .3

7 0 .7

1 0 .4

W h it e -B

la ck

d if fe

re n ce

in fr

e e

lu n ch

ra te

s 0 .0

2 0 .0

4 0 .1

1 0 .1

0 0 .0

7 0 .0

5 0 .0

W h it e -H

is p an

ic d if fe

re n ce

in fr

e e

lu n ch

ra te

s 0 .0

2 0 .0

4 0 .1

2 0 .1

0 0 .0

7 0 .0

8 0 .0

6 0 .0

P e rc

e n t u rb

an 7

2 5

5 1

4 8

3 6

4 6

2 4

4 1

P e rc

e n t su

b u rb

3 9

4 7

3 7

4 6

4 4

4 7

3 7

4 6

P e rc

e n t to

w n

2 7

4 1

7 2 3

1 3

3 1

2 0

3 7

P e rc

e n t ru

ra l

2 7

4 1

5 1 9

8 2 2

1 9

3 5

N o te

. E LA

= E n g li sh

la n g u ag

e ar

ts ; E LL

= E n g li sh

la n g u ag

e le

ar n e r;

N A E P

= N

at io

n al

A ss

e ss

m e n t o f E d u ca

ti o n al

P ro

g re

ss .

1187

districts with both magnet and charter schools have over eight times the public school student enrollment as districts without magnets and charters. Districts with choice are also substantially less affluent, less White, and more racially and ethnically segregated on average; they also demonstrate lower test scores and larger test score gaps compared with districts without choice. Tables 2 and 3 provide additional descriptive information on the variables included in this study.

Measures

Test Score Gaps

SEDA obtained test scores from the U.S. Department of Education’s (2017) EdFacts database and standardized them according to the National Assessment of Educational Progress scale in order to be comparable across districts and states. SEDA includes test score gaps for districts from 2008–2009 to 2014– 2015. Within each district and each school year, the mean test score gaps are included for both math and ELA tests for each grade, from third through eighth. The analyses examine math and ELA test score gaps separately since they are descriptively different. For example, in these data, ELA test score gaps are gen- erally higher than math test score gaps (see Tables 2 and 3).

The White-Black test score gaps represent the difference between the mean White students’ test scores and mean Black students’ test scores for a given test subject, grade, school year, and school district. The same calcula- tion is used for White-Hispanic test score gaps. The Black and White students’ test scores in SEDA represent non-Hispanic students.

Magnet and Charter Enrollment

SEDA measures the cumulative percentage of third through eighth graders in each district enrolled in charter schools. In practice, charters are not linked to geographic school districts, so SEDA identified the geographic school district where charters are physically located (only brick and mortar charters are included). In addition to analyzing SEDA data, this study indepen- dently sourced and merged magnet school data from the U.S. Department of Education’s (2017) Common Core of Data (CCD). The CCD includes school- level data on magnet school enrollment which we aggregated to the district level to compute the cumulative percentage of third through eighth graders enrolled in magnet schools in each district. The CCD does not include com- plete administrative data on magnet schools in Massachusetts, New Jersey, New York, Ohio, and Vermont. So, we adjusted the classification of magnet schools in these five states in the CCD files based on a list of all the magnet schools in these states obtained by the first author from Magnet Schools of America (2018) based on a directory compiled by researchers at the University of North Carolina Charlotte.

Blatt, Votruba-Drzal

1188

Table 2

Descriptive Statistics for Analyses of White-Black Test Score Gaps

Math ELA

Observations (Level 1 N) 79,270 85,143 Districts: (Level 2 N) 2,890 2,908 States: (Level 3 N) 51 51

Variable M SD Minimum Maximum

Mean district math test performance (NAEP scale) 252.07 21.41 190.56 345.07 Mean White-Black math test score gap 19.95 9.36 224.09 84.66 Mean district ELA test performance (NAEP scale) 233.41 22.52 153.49 317.00 Mean White-Black ELA test score gap 21.39 10.50 225.84 147.54 Total district membership/1,000 9.84 23.93 0.36 758.30 Socioeconomic status composite 20.22 1.00 24.11 2.98 Per pupil expenditures/1,000 12.05 3.71 2.84 42.21 Gini coefficient 0.39 0.05 0.16 0.58 Percent Native American per grade 1 2 0 48 Percent Asian per grade 4 6 0 77 Percent Hispanic per grade 17 20 0 99 Percent Black per grade 24 20 0 98 Percent White per grade 54 23 0 100 White-Black Segregation 0.08 0.11 0.00 0.90 Percent of students who are ELL 6 8 0 69 White-Black difference in family income 0.67 0.44 21.97 2.96 White-Black difference in parent education 0.18 0.35 21.21 1.98 Percent enrolled in charter schools 3 8 0 99 Percent enrolled in magnet schools 3 12 0 100 Urbanicity (%)

Urban 23 Suburb (reference) 40 Town 16 Rural 21

Grade (%) Three 16 Four 17 Five 17 Six 17 Seven 17 Eight (reference) 17

Year (%) 2008–2009 15 2009–2010 15 2010–2011 15 2011–2012 16 2012–2013 15 2013–2014 12 2014–2015 (reference) 12

Note. ELA = English language arts; ELL = English language learner; NAEP = National Assessment of Educational Progress.

School Choice and Racial/Ethnic Test Score Gaps

1189

Table 3

Descriptive Statistics for Analyses of White-Hispanic Test Score Gaps

Math ELA

Observations (Level 1 N) 94,433 101,810 Districts: (Level 2 N) 3,840 3,857 States: (Level 3 N) 50 50

Variable M SD Minimum Maximum

Mean district math test performance (NAEP scale) 252.85 21.52 181.26 343.10 Mean White-Hispanic math test score gap 14.00 9.16 240.94 80.63 Mean district ELA test performance (NAEP scale) 234.02 22.96 157.70 312.22 Mean White-Hispanic ELA test score gap 17.94 10.88 243.13 153.68 Total district membership/1000 8.76 22.08 0.22 758.30 Socioeconomic status composite 0.08 0.92 23.60 2.76 Per pupil expenditures/1,000 12.02 4.12 3.51 47.06 Gini coefficient 0.37 0.05 0.14 0.58 Percent Native American per grade 1 4 0 92 Percent Asian per grade 4 7 0 77 Percent Hispanic per grade 28 22 0 100 Percent Black per grade 12 15 0 98 Percent White per grade 55 24 0 100 White-Hispanic Segregation 0.06 0.09 0.00 0.89 Percent of Hispanic population from Central America 6 9 0 100 Percent of Hispanic population from Cuba 1 4 0 65 Percent of Hispanic population from Mexico 68 30 0 100 Percent of Hispanic population from Puerto Rico 11 18 0 100 Percent of Hispanic population from South America 5 9 0 83 Percent of students who are ELL 9 10 0 73 White-Hispanic difference in family income 0.64 0.43 22.18 2.48 White-Hispanic difference in parent education 0.74 0.40 20.94 2.37 Percent enrolled in charter schools 3 8 0 99 Percent enrolled in magnet schools 2 10 0 100 Urbanicity (%)

Urban 20 Suburb (reference) 40 Town 22 Rural 18

Grade (%) Three 17 Four 17 Five 17 Six 17 Seven 16 Eight (reference) 16

Year (%) 2008–2009 13 2009–2010 14 2010–2011 15 2011–2012 17 2012–2013 16 2013–2014 11 2014–2015 (reference) 13

Note. ELA = English language arts; ELL = English language learner; NAEP = National Assessment of Educational Progress.

Blatt, Votruba-Drzal

1190

Segregation

SEDA includes a measure of segregation based on Theil’s (1972) informa- tion theory index. The White-Black and White-Hispanic information index variables are calculated by computing the mean deviation of a student’s school’s racial and ethnic composition from the school’s district-wide racial and ethnic composition. The variable is calculated individually for each grade from third through eighth in each school year from 2009 to 2015. If the varia- ble’s value is close to one, this indicates high racial/ethnic segregation. This is a useful operationalization because the point of reference is the district-wide integration that is possible given the racial/ethnic composition of the school district as opposed to measures of segregation that simply look at students’ level of exposure to students of other races and ethnicities.

As a robustness check, however, we also ran the analyses using exposure indices which measure the average percent of White students in the average Black or Hispanic student’s school and vice versa. This is an important check because if a predominantly White school district is adjacent to a predominantly Black school district, it is possible that both of these districts would score rel- atively low on the Theil index given their low capacity for within-district inte- gration. Additionally, catchment areas of magnet and charter schools do not always correspond to the boundaries of school districts, so they may be work- ing to segregate or integrate school districts beyond those in which they are physically located. The findings of indirect effects of the exposure indices are not included in the results because they largely replicate the size, direc- tion, and significance of the analyses using the Theil index.

Covariates

The analyses control for several factors that are associated with district-level school choice and racial/ethnic test score gaps. Districts’ mean test scores are included to control for baseline district test performance. Total district enrollment is included because school choice is more prevalent in larger districts (NAPCS, 2012; Polikoff & Hardaway, 2017). District urbanicity is reflected in a series of dummy variables for urban, suburban (reference group), rural, and town. District-level socioeconomic status is measured with a composite variable that includes a district’s median income, proportion of adults with a bachelor’s degree or higher, the poverty rate in households with at least one child aged 5 to 17 years, unemployment rate, proportion of households participating in the Supplemental Nutrition Assistance Program, and proportion of households headed by a single mother. District per pupil expenditures are also included (per pupil expenditures for 2014–2015 are not included in SEDA version 2.1, so the values for the 2014– 2015 school year reflect the same values as the 2013–2014 school year). District income inequality is controlled for using a Gini coefficient, with values closer to zero indicating equal income distribution in a district and values closer to one reflecting inequality in the income distribution (Von Hippel & Powers, 2015).

School Choice and Racial/Ethnic Test Score Gaps

1191

The racial composition of the school district is accounted for with the per- cent of students in each grade in each district who are Black and Hispanic. The total percent of English language learners in a district is included as well. For the analyses examining the White-Hispanic test score gaps, the percent of the Hispanic population in a district that is Central American Cuban, Mexican, Puerto Rican, and South American (these reflect the categories in SEDA) are included as covariates to account for the heterogeneity of the Hispanic pop- ulation in the United States. The analyses also control for standard deviation differences in districts’ White and Black parents’ as well as White and Hispanic parents’ income and educational attainment to account for the inter- section of race/ethnicity and socioeconomic status. Importantly, racial/ethnic differences in free lunch rates (included in Table 1) are not included in the final analyses due to high intercorrelations with the racial/ethnic segregation variables; this is discussed further in the limitations. Finally, the analyses con- trol for all grade levels using dummy variables for grades three through eight (reference group) and all school years using dummy variables for each school year from 2008–2009 to 2014–2015 (reference group).

Analytic Approach

The first research question examines the association of charter and mag- net school enrollment and the White-Black and White-Hispanic test score gaps at the district level. This association is tested using multilevel random intercept and longitudinal fixed effects modeling. We ran the models sepa- rately for each test score gap (White-Black math, White-Black ELA, White- Hispanic math, and White-Hispanic ELA). Our three-level random intercept models include a random intercept at the district and state levels. We con- ducted all of our analyses using Stata 15.0.

Level 1 : Yijk5p0jk1 ppjkapijk1eijk ð1Þ

In Equation 1, Yijk represents the racial/ethnic test score gap in grade/year (i), nested in district (j), and in state (k). Here, ppjk (p is the within district param- eter) represents the coefficients for predictors apijk that vary within district by grade-level and or school year (percent of students enrolled in charters, per- cent of students enrolled in magnets, urbanicity, mean district test perfor- mance, district size, per pupil expenditures, racial/ethnic composition, percent of English language learners, grade-level, and school year), and eijk

is the within-district random effect.

Level 2 : p0jk5b00k1b0qkXqjk1r0jk ð2Þ

In Equation 2, b0qk (q is the between district parameter) represents the coef- ficients for predictors Xqjk that only vary in the data between districts (socio- economic status, socioeconomic inequality, percent of a district’s Hispanic

Blatt, Votruba-Drzal

1192

population representing different nationalities, and racial/ethnic differences in family income and parent education), and r0jk is the district-level random effect.

Level 3 : b00k5 g0001u00k ð3Þ

In Equation 3, u00k represents the state-level random effect to account for dis- tricts being nested within states. While there are no state-level predictors in the models, this level is necessary because school choice policies vary by state and this likely influences the associations in these analyses. For example, the 12 states that were awarded federal ‘‘Race to the Top’’ grants expanded sup- port for charter schools (Boser, 2012). Conversely, some states have caps on the number of charters permitted statewide and as of 2016, seven states did not have laws allowing for charter schools at all (Center for Education Reform, 2018).

A general concern of analyzing data at the school district level is omitted variable bias. There are remaining factors at the individual student, school, and district level that influence district-level racial/ethnic test score gaps and school choice that are not measured in these data. Obvious omitted var- iables include other forms of school choice such as vouchers and open enroll- ment. We address omitted variable bias with longitudinal fixed effects models that replicate the random intercept models, but also control for each variable’s district mean for a given grade from 2008–2009 to 2014–2015. The benefit of this approach is that it removes variance explained by unmeasured time- invariant variables (Miller et al., 2016). So, each district serves as its own coun- terfactual over time by examining whether deviation from a district’s mean magnet and charter enrollment predicts deviation from a district’s mean White-Black or White-Hispanic test score gap. In these models, the covariates are also measured as deviations from their district means with the exception of invariant covariates including urbanicity, socioeconomic status, the Gini coef- ficient, White-Black/White-Hispanic differences in family income, White- Black/White-Hispanic differences in parent education, and composition of the Hispanic population (percent Central American, Cuban, Mexican, Puerto Rican, and South American). While many of these covariates may be time variant in reality, SEDA only includes one measure of these variables from 2009 to 2015. The longitudinal fixed effects models control for these time invariant variables because they may still predict the degree to which a district deviates from its mean test score gap over time.

The second research question examines district-level White-Black and White-Hispanic segregation as pathways through which charter and magnet school enrollment relate to White-Black and White-Hispanic math and ELA test score gaps. Congruently to the analytic approach described for research question one, we ran the models separately for each of the four test score gaps. We calculated indirect effects by multiplying the adjusted effects of

School Choice and Racial/Ethnic Test Score Gaps

1193

school choice enrollment on segregation with the adjusted effects of segrega- tion on test score gaps; the significance of indirect effects was probed using Sobel standard errors (Preacher & Leonardelli, 2010). As a robustness check, we also estimated indirect effects in a structural equation modeling frame- work analyzed in Mplus 8.1. There was consistency across frameworks, so only the models run in Stata 15.0 are included in the results.

Due to the high concentrations of zeroes for the percentage of magnet and charter enrollment in districts, we also ran the analyses with natural log-transformed school choice variables. These models yielded consistent results with slightly larger effect sizes. For ease of interpretation, we present the results with untransformed variables in the article and present the results with natural log-transformed variables as supplementary tables (see Supplemental Tables S1 and S2 in the online version of the journal).

Results

General trends indicate that even when accounting for factors such as district-level income inequality and racial/ethnic socioeconomic disparities, there are robust but small associations between higher charter enrollment and larger White-Black test score gaps and there is some evidence of associ- ations between higher magnet enrollment and larger White-Hispanic gaps. The results also consistently support that segregation is a mechanism through which charter and, to a lesser degree, magnet enrollment predict larger racial/ ethnic test score gaps.

White-Black Test Score Gaps

The three-level random intercept model results in Table 4 and Figure 1 show that for every 10 percentage points of students in a district enrolled in charters, a district’s White-Black math test score gap is 0.3 points (0.03 SD units) larger and the White-Black ELA gap is 0.36 points (0.03 SD) larger. These results are replicated with stronger effect sizes in the longitudinal fixed effects models; math and ELA test score gaps are 0.06 and 0.05 SD larger for every additional 10 percentage points of students in a district enrolled in char- ter schools (Table 5, Figure 2).

These associations are mediated by White-Black segregation. The ran- dom intercept models show that an increase of 10 percentage points of charter enrollment is associated with a 0.02 point (0.13 SD) increase in White-Black segregation, which is associated with a White-Black math gap that is 0.03 points (0.003 SD) larger (Table 6, Figure 3). This is even more pronounced for ELA where the indirect effect of White-Black segregation predicts a gap that is 0.08 points (0.01 SD) larger (Table 6, Figure 3). Similar to the direct asso- ciations, the indirect effects are stronger in the longitudinal fixed effects mod- els. An increase of 10 percentage points in charter enrollment is associated with a 0.02 point (0.6 SD) increase in White-Black segregation, which is

Blatt, Votruba-Drzal

1194

T a b le

R e s u

lt s

fo r

T h

re e -L

e v e l

R a n

d o

m In

te rc

e p

t M

o d

e ls

E x a m

in in

g th

e A

s s o

c ia

ti o

n B

e tw

e e n

D is

tr ic

t- L

e v e l

C h

a rt

e r

a n

d M

a g

n e t

S c h

o o

E n

ro ll

m e n

t a n

d W

h it

e -B

la c k

T e s t

S c o

re G

a p

M at

h E n g li sh

La n g u ag

e A rt s

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

G ra

d e

3 2

1 .6

4 * * *

0 .2

3 2

0 .0

7 1 .9

7 * * *

0 .2

9 0 .0

G ra

d e

4 2

1 .2

4 * * *

0 .1

9 2

0 .0

5 2 .2

5 * * *

0 .2

4 0 .0

G ra

d e

5 2

1 .3

2 * * *

0 .1

5 2

0 .0

5 1 .4

1 * * *

0 .1

9 0 .0

G ra

d e

6 2

0 .6

6 * * *

0 .1

2 2

0 .0

3 0 .9

0 * * *

0 .1

4 0 .0

G ra

d e

7 2

0 .3

5 * * *

0 .0

9 2

0 .0

1 0 .3

7 * * *

0 .1

0 0 .0

2 0 0 8 – 2 0 0 9

2 0 .1

4 0 .0

9 2

0 .0

1 2

0 .5

1 * * *

0 .1

0 2

0 .0

2 0 0 9 – 2 0 1 0

2 0 .5

1 * * *

0 .0

9 2

0 .0

2 2

0 .7

2 * * *

0 .1

0 2

0 .0

2 0 1 0 – 2 0 1 1

2 0 .5

7 * * *

0 .0

9 2

0 .0

2 2

0 .5

0 * * *

0 .1

0 2

0 .0

2 0 1 1 – 2 0 1 2

2 0 .6

8 * * *

0 .0

9 2

0 .0

3 2

0 .8

2 * * *

0 .0

9 2

0 .0

2 0 1 2 – 2 0 1 3

2 0 .3

8 * * *

0 .0

9 2

0 .0

1 2

0 .0

8 0 .0

9 2

0 .0

0 3

2 0 1 3 – 2 0 1 4

2 0 .3

9 * * *

0 .0

9 2

0 .0

1 2

0 .0

7 0 .1

0 2

0 .0

0 2

U rb

an 0 .6

0 * * *

0 .1

8 0 .0

3 0 .6

9 * * *

0 .2

0 0 .0

T o w

n 2

0 .4

4 *

0 .1

7 2

0 .0

2 2

0 .5

2 * *

0 .1

9 2

0 .0

R u ra

l 2

0 .3

7 *

0 .1

5 2

0 .0

2 2

0 .6

4 * * *

0 .1

7 2

0 .0

M e an

d is

tr ic

t te

st p e rf

o rm

an ce

(N A E P

sc al

e )

0 .0

6 * * *

0 .0

0 4

0 .1

5 0 .0

1 0 .0

T o ta

l d is

tr ic

t m

e m

b e rs

h ip

/1 0 0 0

0 .0

2 * * *

0 .0

0 4

0 .0

6 0 .0

2 * * *

0 .0

0 5

0 .0

So ci

o e co

n o m

ic st

at u s

co m

p o si

te 3 .6

7 * * *

0 .1

7 0 .4

0 3 .7

9 * * *

0 .2

0 0 .3

P e r p u p il

e x p e n d it u re

s/ 1 0 0 0

0 .0

5 * *

0 .0

2 0 .0

1 0 .0

2 0 .0

0 5

G in

i co

e ff ic

ie n t

4 4 .8

9 * * *

3 .0

7 0 .2

5 5 5 .0

0 * * *

3 .5

9 0 .2

P e rc

e n t H

is p an

ic p e r g ra

d e

0 .1

9 * *

0 .0

6 0 .0

4 0 .0

8 0 .0

7 0 .0

(c o n

ti n

u ed

)

1195

T a b le

4 (c

o n

ti n

u e d

)

M at

h E n g li sh

La n g u ag

e A rt s

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

P e rc

e n t B la

ck p e r g ra

d e

0 .4

7 * * *

0 .0

6 0 .1

0 0 .5

2 * * *

0 .0

7 0 .1

P e rc

e n t o f st

u d e n ts

w h o

ar e

E LL

0 .2

2 * *

0 .0

8 0 .0

2 0 .1

9 *

0 .0

9 0 .0

W h it e -B

la ck

d if fe

re n ce

in fa

m il y

in co

m e

1 .1

2 * * *

0 .2

4 0 .0

5 1 .3

9 * * *

0 .2

9 0 .0

W h it e -B

la ck

d if fe

re n ce

in p ar

e n t e d u ca

ti o n

6 .2

8 * * *

0 .3

0 0 .2

4 7 .7

8 * * *

0 .3

8 0 .2

P e rc

e n t e n ro

ll e d

in ch

ar te

r sc

h o o ls

0 .3

0 * * *

0 .0

8 0 .0

2 0 .3

6 * * *

0 .0

9 0 .0

P e rc

e n t e n ro

ll e d

in m

ag n e t sc

h o o ls

0 .1

0 0 .0

5 0 .0

1 0 .0

0 2

0 .0

6 0 .0

0 0 2

In te

rc e p t

2 1 5 .7

2 * * *

1 .6

7 2

6 .7

5 * * *

1 .9

D is

tr ic

t v ar

ia n ce

1 8 .8

3 0 .6

0 2 6 .7

2 0 .8

St at

e v ar

ia n ce

7 .1

6 1 .9

0 8 .2

8 2 .2

R e si

d u al

v ar

ia n ce

3 7 .1

3 0 .1

9 4 7 .9

2 0 .2

x 2

2 7 3 2 9 .3

3 * * *

2 9 9 7 9 .7

8 * * *

N o te

. P e rc

e n ta

g e

v ar

ia b le

s ar

e sc

al e d

in 1 0

p e rc

e n t u n it s.

E LL

= E n g li sh

la n g u ag

e le

ar n e r;

N A E P

= N

at io

n al

A ss

e ss

m e n t o f E d u ca

ti o n al

P ro

g re

ss .

* p

\ .0

5 . * * p

\ .0

1 . * * * p

\ .0

0 1 .

1196

associated with White-Black math and ELA gaps that are 0.11 points (0.02 SD) and 0.15 points (0.03 SD) larger (Table 7, Figure 4).

District-level magnet school enrollment does not significantly predict var- iance in White-Black test score gaps (Tables 4 and 5, Figures 1 and 2). However, there are small significant indirect effects of heightened magnet enrollment on larger math and ELA gaps that operate through White-Black segregation in both the random intercept and longitudinal fixed effects mod- els (Tables 6 and 7, Figures 3 and 4).

White-Hispanic Test Score Gaps

While charter enrollment predicts variance in White-Black test score gaps, it is magnet enrollment that drives the school choice effects in the White- Hispanic gaps. In the three-level random intercept models, for every addi- tional 10 percentage points of students in a district enrolled in magnet schools, a district’s White-Hispanic math gap is 0.15 points (0.02 SD) larger and the White-Hispanic ELA gap is 0.16 points (0.02 SD) larger (Table 8, Figure 1). However, these results were not replicated in the longitudinal fixed effects models (Table 9, Figure 2).

There are small positive indirect effects of magnet school enrollment on White-Hispanic test score gaps operating through White-Hispanic segregation in both the random intercept and longitudinal fixed effects models (Tables 10 and 11, Figures 3 and 4). In the random intercept models, a 10–percentage point increase in magnet enrollment is associated with a 0.001 unit (0.02 SD) increase in White-Hispanic segregation, which significantly predicts

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.10

White-Black Math White-Black ELA White-Hispanic Math White-Hispanic ELA

s n

oitai ve

d dra

d nats

pa g er

ocs ts e T

Charter

School

Enrollment

Magnet

School

Enrollment

†

*** ***

** **

ns ns

Figure 1. Standard deviation change of test score gap per 10 percentage points of

students in a district enrolled in charter and magnet schools.

Note. ELA = English language arts. yp \ .1. *p \ .05. **p \ .01. ***p \ .001.

School Choice and Racial/Ethnic Test Score Gaps

1197

T a b le

R e s u

lt s

fo r

L o

n g

it u

d in

a l

F ix

e d

E ff

e c ts

M o

d e ls

E x a m

in in

g th

e A

s s o

c ia

ti o

n B

e tw

e e n

D is

tr ic

t- L

e v e l

C h

a rt

e r

a n

d M

a g

n e t

S c h

o o

E n

ro ll

m e n

t a n

d W

h it

e -B

la c k

T e s t

S c o

re G

a p

M at

h E n g li sh

La n g u ag

e A rt s

V ar

ia b le

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

G ra

d e

3 2

0 .0

1 0 .0

7 2

0 .0

0 1

2 0 .0

1 0 .0

7 2

0 .0

0 0 4

G ra

d e

4 2

0 .0

1 0 .0

7 2

0 .0

0 1

2 0 .0

1 0 .0

7 2

0 .0

0 0 3

G ra

d e

5 2

0 .0

1 0 .0

7 2

0 .0

0 0 4

2 0 .0

1 0 .0

7 2

0 .0

0 0 4

G ra

d e

6 2

0 .0

1 0 .0

7 2

0 .0

0 1

2 0 .0

0 4

0 .0

7 2

0 .0

0 0 2

G ra

d e

7 2

0 .0

1 0 .0

7 2

0 .0

0 1

2 0 .0

0 1

0 .0

7 0 .0

0 0 0

2 0 0 8 – 2 0 0 9

0 .0

5 0 .0

9 0 .0

0 4

2 0 .3

3 * * *

0 .1

0 2

0 .0

2 0 0 9 – 2 0 1 0

2 0 .3

0 * * *

0 .0

8 2

0 .0

2 2

0 .5

2 * * *

0 .0

9 2

0 .0

2 0 1 0 – 2 0 1 1

2 0 .4

7 * * *

0 .0

8 2

0 .0

3 2

0 .4

2 * * *

0 .0

9 2

0 .0

2 0 1 1 – 2 0 1 2

2 0 .5

6 * * *

0 .0

8 2

0 .0

4 2

0 .7

1 * * *

0 .0

8 2

0 .0

2 0 1 2 – 2 0 1 3

2 0 .2

8 * * *

0 .0

7 2

0 .0

2 2

0 .0

2 0 .0

8 2

0 .0

0 1

2 0 1 3 – 2 0 1 4

2 0 .3

1 * * *

0 .0

8 2

0 .0

2 2

0 .0

2 0 .0

9 2

0 .0

0 1

U rb

an 2

0 .0

1 0 .0

6 2

0 .0

0 1

2 0 .0

6 2

0 .0

0 1

T o w

n 2

0 .0

4 0 .0

6 2

0 .0

0 3

2 0 .0

3 0 .0

7 2

0 .0

0 2

R u ra

l 0 .0

1 0 .0

5 0 .0

0 0 4

2 0 .0

4 0 .0

6 2

0 .0

0 3

M e a n

d is

tr ic

t te

st p

e rf

o rm

a n

c e

(N A

E P

sc a le

) 0 .0

4 * * *

0 .0

0 5

0 .0

3 0 .0

2 * * *

0 .0

1 0 .0

T o

ta l

d is

tr ic

t m

e m

b e rs

h ip

/1 ,0

0 0

0 .0

7 * * *

0 .0

2 0 .0

1 0 .0

5 *

0 .0

2 0 .0

So ci

o e co

n o m

ic st

at u s

co m

p o si

te 2

0 .0

0 2

0 .0

3 2

0 .0

0 0 4

2 0 .0

1 0 .0

4 2

0 .0

0 1

P e r

p u

p il

e x

p e n

d it

u re

s/ 1

,0 0

0 0 .0

3 0 .0

2 0 .0

1 2

0 .0

2 0 .0

2 2

0 .0

0 3

G in

i co

e ff ic

ie n t

2 0 .0

1 0 .6

4 2

0 .0

0 0 1

2 0 .0

5 0 .7

3 0 .0

0 0 4

P e rc

e n

t H

is p

a n

ic p

e r

g ra

d e

0 .4

8 * * *

0 .1

1 0 .0

2 0 .3

2 * *

0 .1

2 0 .0

P e rc

e n

t B

la c k

p e r

g ra

d e

2 0 .2

0 *

0 .0

9 2

0 .0

1 2

0 .0

9 0 .1

0 2

0 .0

0 3

(c o n

ti n

u ed

)

1198

T a b le

5 (c

o n

ti n

u e d

)

M at

h E n g li sh

La n g u ag

e A rt s

V ar

ia b le

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

P e rc

e n

t o

f st

u d

e n

ts th

a t

a re

E L L

0 .0

7 0 .0

8 0 .0

0 3

0 .0

8 0 .0

0 4

W h it e -B

la ck

d if fe

re n ce

in fa

m il y

in co

m e

2 0 .0

1 0 .0

5 2

0 .0

0 1

2 0 .0

1 0 .0

6 2

0 .0

0 1

W h it e -B

la ck

d if fe

re n ce

in p ar

e n t e d u ca

ti o n

0 .0

2 0 .0

7 0 .0

0 1

0 .0

2 0 .0

7 0 .0

0 1

P e rc

e n

t e n

ro ll

e d

in c h

a rt

e r

sc h

o o

ls 0 .3

1 * * *

0 .0

8 0 .0

1 0 .2

8 * * *

0 .0

9 0 .0

P e rc

e n

t e n

ro ll

e d

in m

a g n

e t

sc h

o o

ls 2

0 .0

2 0 .0

5 2

0 .0

0 1

2 0 .1

2 0 .0

6 2

0 .0

In te

rc e p t

0 .3

0 0 .2

5 0 .3

5 0 .2

R e si

d u al

v ar

ia n ce

2 8 .5

6 0 .1

5 3 8 .7

7 0 .1

N o te

.P e rc

e n ta

g e

v ar

ia b le

s ar

e sc

al e d

in 1 0

p e rc

e n t u n it s.

V ar

ia b le

s in

b o ld

fa ce

co n tr o lf

o r th

e fi x e d

e ff e ct

s o f d is

tr ic

t m

e an

s fo

r a

g iv

e n

g ra

d e

fr o m

2 0 0 8 – 2 0 0 9

to 2 0 1 4 – 2 0 1 5 . E LL

= E n g li sh

la n g u ag

e le

ar n e r;

N A E P

= N

at io

n al

A ss

e ss

m e n t o f E d u ca

ti o n al

P ro

g re

ss .

* p

\ .0

5 . * * p

\ .0

1 . * * * p

\ .0

0 1 .

1199

a White-Hispanic math gap that is 0.002 points (0.0003 SD) larger, indicating a small significant indirect effect (Table 10, Figure 3). The effect was more pro- nounced for ELA where the indirect effect of magnet enrollment operating through White-Hispanic segregation predicts a gap that is 0.01 points (0.001 SD) larger (Table 10, Figure 3).

There are no significant direct effects of district-level charter enrollment on White-Hispanic test score gaps (Tables 8 and 9, Figures 1 and 2). However, the results suggest there are larger indirect effects of charter enroll- ment than magnet enrollment on White-Hispanic test score gaps operating through White-Hispanic segregation (Tables 10 and 11, Figures 3 and 4). In other words, higher charter enrollment is related to greater White-Hispanic segregation than magnet enrollment which leads to larger indirect effects on White-Hispanic test score gaps.

Discussion

School choice is expanding rapidly in the United States while robust racial/ethnic segregation and test score gaps reflect staunch structural educa- tion inequity. Yet the variability of findings in school choice research con- ducted at the local level makes it difficult to determine best policy practices at the federal level. Consequently, analyses such as those presented in this study are necessary to understand whether school choice may be operating to exacerbate or attenuate structural education inequity on a macro scale.

-0.03

-0.02

-0.01

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.10

White-Black Math White-Black ELA White-Hispanic Math White-Hispanic ELA

s n

oitai ve

d dra

d nats

pa g er

ocs tse T

Charter

School

Enrollment

Magnet

School

Enrollment

***

ns ns

†

Figure 2. Mean difference from district mean standard deviation change in test

score gap per 10 percentage points of students in a district enrolled in charter

and magnet schools.

Note. ELA = English language arts. yp \ .1. *p \ .05. **p \ .01. ***p \ .001.

Blatt, Votruba-Drzal

1200

T a b le

R e s u

lt s

fo r

T h

re e -L

e v e l

R a n

d o

m In

te rc

e p

t M

o d

e ls

E x a m

in in

g th

e In

d ir

e c t

E ff

e c ts

o f

W h

it e -B

la c k

S e g

re g

a ti

o n

th e

A s s o

c ia

ti o

n s

B e tw

e e n

D is

tr ic

t- L

e v e l

C h

a rt

e r

a n

d M

a g

n e t

S c h

o o

l E

n ro

ll m

e n

t a n

d W

h it

e -B

la c k

T e s t

S c o

re G

a p

M at

h E n g li sh

La n g u ag

e A rt s

P re

d ic

ti n g

W h it e -B

la ck

Se g re

g at

io n

P re

d ic

ti n g

W h it e -B

la ck

T e st

Sc o re

G ap

P re

d ic

ti n g

W h it e -B

la ck

Se g re

g at

io n

P re

d ic

ti n g

W h it e -B

la ck

T e st

Sc o re

G ap

V ar

ia b le

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

G ra

d e

3 0 .0

6 * * *

0 .0

0 1

0 .2

1 2

1 .7

5 * * *

0 .2

3 2

0 .0

6 0 .0

7 * * *

0 .0

0 2

0 .2

3 1 .6

1 * * *

0 .3

0 0 .0

6 G

ra d e

4 0 .0

6 * * *

0 .0

0 1

0 .1

9 2

1 .3

4 * * *

0 .1

9 2

0 .0

5 0 .0

6 * * *

0 .0

0 1

0 .2

1 1 .9

2 * * *

0 .2

4 0 .0

8 G

ra d e

5 0 .0

5 * * *

0 .0

0 1

0 .1

7 2

1 .4

0 * * *

0 .1

5 2

0 .0

5 0 .0

5 * * *

0 .0

0 1

0 .1

8 1 .1

3 * * *

0 .1

9 0 .0

5 G

ra d e

6 0 .0

2 * * *

0 .0

0 1

0 .0

6 2

0 .6

9 * * *

0 .1

2 2

0 .0

2 0 .0

2 * * *

0 .0

0 1

0 .0

7 0 .8

0 * * *

0 .1

4 0 .0

3 G

ra d e

7 0 .0

0 4 * * *

0 .0

0 1

0 .0

1 2

0 .3

5 * * *

0 .0

9 2

0 .0

1 0 .0

0 5 * * *

0 .0

0 1

0 .0

2 0 .3

5 * * *

0 .1

0 0 .0

1 2 0 0 8 – 2 0 0 9

0 .0

1 * * *

0 .0

0 1

0 .0

2 2

0 .1

5 0 .0

9 2

0 .0

1 0 .0

1 * * *

0 .0

0 1

0 .0

2 2

0 .5

5 * * *

0 .1

0 2

0 .0

2 2 0 0 9 – 2 0 1 0

0 .0

0 5 * * *

0 .0

0 1

0 .0

1 2

0 .5

2 * * *

0 .0

9 2

0 .0

2 0 .0

1 * * *

0 .0

0 1

0 .0

2 2

0 .7

5 * * *

0 .1

0 2

0 .0

3 2 0 1 0 – 2 0 1 1

0 .0

0 2 * * *

0 .0

0 1

0 .0

1 2

0 .5

7 * * *

0 .0

9 2

0 .0

2 0 .0

0 3 * * *

0 .0

0 1

0 .0

1 2

0 .5

2 * * *

0 .1

0 2

0 .0

2 2 0 1 1 – 2 0 1 2

0 .0

0 1

0 .0

0 1

0 .0

0 2

2 0 .6

9 * * *

0 .0

9 2

0 .0

3 0 .0

0 1 * *

0 .0

0 1

0 .0

0 4

2 0 .8

3 * * *

0 .0

9 2

0 .0

3 2 0 1 2 – 2 0 1 3

– 0 .0

0 1

0 .0

0 1

2 0 .0

0 2

2 0 .3

8 * * *

0 .0

9 2

0 .0

1 0 .0

0 0 0

0 .0

0 1

0 .0

0 0 1

2 0 .0

8 0 .0

9 2

0 .0

0 3

2 0 1 3 – 2 0 1 4

0 .0

0 0 4

0 .0

0 1

2 0 .0

0 1

2 0 .3

8 * * *

0 .0

9 2

0 .0

1 2

0 .0

0 0 3

0 .0

0 1

2 0 .0

0 1

2 0 .0

6 0 .1

0 2

0 .0

0 2

U rb

an 0 .0

0 3 *

0 .0

0 1

0 .0

1 0 .5

7 * * *

0 .1

7 0 .0

3 0 .0

0 2

0 .0

0 1

0 .0

1 0 .6

4 * * *

0 .2

0 0 .0

3 T o w

n – 0 .0

0 1

0 .0

0 1

2 0 .0

0 2

2 0 .4

2 *

0 .1

7 2

0 .0

2 – 0 .0

0 1

0 .0

0 1

2 0 .0

0 3

2 0 .4

8 *

0 .1

9 2

0 .0

2 R u ra

l 0 .0

0 1

0 .0

0 1

0 .0

0 3

2 0 .3

7 *

0 .1

5 2

0 .0

2 0 .0

0 1

0 .0

0 1

0 .0

0 2

2 0 .6

4 * * *

0 .1

7 2

0 .0

2 M

e an

d is

tr ic

t te

st p e rf

o rm

an ce

(N A E P

sc al

e )

0 .0

0 0 4 * * *

0 .0

0 0 0

0 .0

8 0 .0

6 * * *

0 .0

0 4

0 .1

4 0 .0

0 0 5 * * *

0 .0

0 0 0

0 .0

9 0 .0

0 7

0 .0

1 0 .0

2 T o ta

l d is

tr ic

t m

e m

b e rs

h ip

/1 0 0 0

0 .0

0 1 * * *

0 .0

0 0 1

0 .3

0 0 .0

2 * * *

0 .0

0 4

0 .0

5 0 .0

0 1 * * *

0 .0

0 0 1

0 .3

0 0 .0

2 * *

0 .0

0 5

0 .0

3 So

ci o e co

n o m

ic st

at u s

co m

p o si

te – 0 .0

0 1

0 .0

0 3

2 0 .0

1 3 .6

7 * * *

0 .1

7 0 .4

0 2

0 .0

1 * *

0 .0

0 3

2 0 .0

8 3 .8

0 * * *

0 .2

0 0 .3

6 P e r p u p il

e x p e n d it u re

s/ 1 0 0 0

0 .0

0 0 2

0 .0

0 0 1

2 0 .0

1 0 .0

5 * *

0 .0

2 0 .0

2 2

0 .0

0 0 1

0 .0

0 0 1

2 0 .0

0 3

0 .0

1 0 .0

2 0 .0

0 5

G in

i co

e ff ic

ie n t

0 .2

3 * * *

0 .0

5 0 .1

1 4 4 .4

5 * * *

3 .0

5 0 .2

4 0 .1

9 * *

0 .0

6 0 .0

8 5 3 .8

0 * * *

3 .5

5 0 .2

6 P e rc

e n t H

is p an

ic p e r g ra

d e

0 .0

1 * * *

0 .0

0 1

0 .1

3 0 .1

8 * *

0 .0

6 0 .0

4 0 .0

1 * * *

0 .0

0 1

0 .1

3 0 .0

4 0 .0

7 0 .0

1 P e rc

e n t B la

ck p e r g ra

d e

0 .0

0 3 * * *

0 .0

0 1

0 .0

5 0 .4

6 * * *

0 .0

6 0 .1

0 0 .0

0 2 * * *

0 .0

0 1

0 .0

4 0 .4

9 * * *

0 .0

6 0 .0

9 P e rc

e n t o f st

u d e n ts

th at

ar e

E LL

– 0 .0

0 1

0 .0

0 1

2 0 .0

1 0 .2

2 * *

0 .0

8 0 .0

2 – 0 .0

0 1 *

0 .0

0 1

2 0 .0

1 0 .1

9 *

0 .0

9 0 .0

1 W

h it e -B

la ck

d if fe

re n ce

in fa

m il y

in co

m e

0 .0

0 2

0 .0

0 2

0 .0

1 1 .1

2 * * *

0 .2

4 0 .0

5 0 .0

0 2

0 .0

0 2

0 .0

1 1 .3

8 * * *

0 .2

9 0 .0

6 W

h it e -B

la ck

d if fe

re n ce

in p ar

e n t e d u ca

ti o n

– 0 .0

0 2

0 .0

0 4

2 0 .0

1 6 .2

9 * * *

0 .3

0 0 .2

4 – 0 .0

0 3

0 .0

1 2

0 .0

1 7 .8

2 * * *

0 .3

8 0 .2

6 P e rc

e n t e n ro

ll e d

in ch

ar te

r sc

h o o ls

0 .0

2 * * *

0 .0

0 1

0 .1

0 0 .2

6 * * *

0 .0

8 0 .0

2 0 .0

2 * * *

0 .0

0 1

0 .1

0 0 .2

5 * *

0 .0

9 0 .0

(c o n

ti n

u ed

)

1201

T a b le

6 (c

o n

ti n

u e d

)

M at

h E n g li sh

La n g u ag

e A rt s

P re

d ic

ti n g

W h it e -B

la ck

Se g re

g at

io n

P re

d ic

ti n g

W h it e -B

la ck

T e st

Sc o re

G ap

P re

d ic

ti n g

W h it e -B

la ck

Se g re

g at

io n

P re

d ic

ti n g

W h it e -B

la ck

T e st

Sc o re

G ap

V ar

ia b le

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

P e rc

e n t e n ro

ll e d

in m

ag n e t sc

h o o ls

0 .0

0 1 * * *

0 .0

0 0 4

0 .0

1 0 .0

9 0 .0

5 0 .0

1 0 .0

0 1 * *

0 .0

0 0 4

0 .0

1 2

0 .0

1 0 .0

6 2

0 .0

0 1

W h it e -B

la ck

se g re

g at

io n

1 .7

8 * * *

0 .5

2 0 .0

2 5 .3

7 * * *

0 .5

8 0 .0

6 In

te rc

e p t

2 0 .1

8 * * *

0 .0

2 2

1 5 .4

1 * * *

1 .6

7 2

0 .1

7 * * *

0 .0

2 2

5 .7

9 * *

1 .9

5 D

is tr ic

t v ar

ia n ce

0 .0

1 0 .0

0 0 2

1 8 .6

2 0 .6

0 0 .0

1 0 .0

0 0 2

2 6 .1

1 0 .8

3 St

at e

v ar

ia n ce

0 .0

0 1

0 .0

0 0 3

7 .1

3 1 .8

9 0 .0

0 1

0 .0

0 0 3

8 .2

0 2 .2

5 R e si

d u al

v ar

ia n ce

0 .0

0 1

0 .0

0 0 0

3 7 .1

3 0 .1

9 0 .0

0 1

0 .0

0 0 0

4 7 .8

8 0 .2

4 In

d ir e ct

e ff e ct

s C h ar

te r sc

h o o l e n ro

ll m

e n t an

d W

h it e -B

la ck

se g re

g at

io n

0 .0

3 * * *

0 .0

1 0 .0

0 2

0 .0

8 * * *

0 .0

1 0 .0

1 M

ag n e t sc

h o o l e n ro

ll m

e n t an

d W

h it e -B

la ck

se g re

g at

io n

0 .0

0 2 *

0 .0

0 1

0 .0

0 0 3

0 .0

1 * *

0 .0

0 2

0 .0

0 1

N o te

. P e rc

e n ta

g e

v ar

ia b le

s ar

e sc

al e d

in 1 0

p e rc

e n t u n it s.

E LL

= E n g li sh

la n g u ag

e le

ar n e r;

N A E P

= N

at io

n al

A ss

e ss

m e n t o f E d u ca

ti o n al

P ro

g re

ss .

* p

\ .0

5 . * * p

\ .0

1 . * * * p

\ .0

0 1 .

1202

Given the proliferation of school choice, it is important to highlight that this study finds no support for the notion that charter or magnet enrollment narrows district-level race/ethnicity test score gaps. On the contrary, the results suggest that school choice enrollment is associated with slightly larger test score gaps at the district level even when controlling for other district characteristics, including socioeconomic status disparities by race/ethnicity. The associations between higher charter enrollment and larger White-Black test score gaps are robust in all model specifications, and associations between higher magnet enrollment and larger White-Hispanic test score gaps are significant only in the random intercept models.

It is useful to consider both the random intercept and longitudinal fixed effects models due to the differences in between-district comparisons versus within-district comparisons that are driving the effects. While the longitudinal fixed effects models provide more compelling causal evidence, these models are only capable of highlighting associations in districts with variability from 2008–2009 to 2014–2015. In other words, if 100% of third through eighth graders in a district are enrolled in magnet schools every year from 2008– 2009 to 2014–2015, that district’s difference from mean value is zero in every wave, the same as it would be in a district with 0% magnet enrollment from 2008–2009 to 2014–2015. As a result, the longitudinal fixed effects estimates are conservative due to constrained variance. This may be one reason why the magnet effects were not replicated for the White-Hispanic test score

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.10

White-Black Math White-Black ELA White-Hispanic Math White-Hispanic ELA

s n

oitai ve

d dra

d na ts

pa g er

ocs tse T

Charter

School

Enrollment

Magnet

School

Enrollment

Segregation

Indirect

Effects

for Charter

Segregation

Indirect

Effects

for Magnet

*** ***

***

** **

Figure 3. Standard deviation change in indirect effect of racial/ethnic segregation

on test score gap per 10 percentage points of students in a district enrolled in char-

ter and magnet schools.

Note. Significance corresponds to the significance of indirect effects. ELA = English language

arts. yp \ .1. *p \ .05. **p \ .01. ***p \ .001.

School Choice and Racial/Ethnic Test Score Gaps

1203

T a b le

R e s u

lt s

fo r

L o

n g

it u

d in

a l F

ix e d

E ff

e c ts

M o

d e ls

E x a m

in in

g th

e In

d ir

e c t

E ff

e c ts

o f

W h

it e -B

la c k

S e g

re g

a ti

o n

th e

A s s o

c ia

ti o

n s

B e tw

e e n

D is

tr ic

t- L

e v e l

C h

a rt

e r

a n

d M

a g

n e t

S c h

o o

l E

n ro

ll m

e n

t a n

d W

h it

e -B

la c k

T e s t

S c o

re G

a p

M at

h E n g li sh

La n g u ag

e A rt s

P re

d ic

ti n g

W h it e -

B la

ck Se

g re

g at

io n

P re

d ic

ti n g

W h it e -B

la ck

T e st

Sc o re

G ap

P re

d ic

ti n g

W h it e -

B la

ck Se

g re

g at

io n

P re

d ic

ti n g

W h it e -B

la ck

T e st

Sc o re

G ap

V ar

ia b le

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

G ra

d e

3 0 .0

0 0 1

0 .0

0 0 3

0 .0

0 2

2 0 .0

1 0 .0

7 2

0 .0

0 1

0 .0

0 0 0

0 .0

0 0 3

0 .0

0 0 5

2 0 .0

1 0 .0

7 2

0 .0

0 0 5

G ra

d e

4 0 .0

0 0 0

0 .0

0 0 3

0 .0

0 0 2

2 0 .0

1 0 .0

7 2

0 .0

0 1

0 .0

0 0 0

0 .0

0 0 3

0 .0

0 0 1

2 0 .0

1 0 .0

7 2

0 .0

0 0 3

G ra

d e

5 – 0 .0

0 0 1

0 .0

0 0 3

2 0 .0

0 1

2 0 .0

1 0 .0

7 0 .0

0 0 4

2 0 .0

0 0 3

0 .0

0 0 3

2 0 .0

0 4

2 0 .0

0 4

0 .0

7 2

0 .0

0 0 2

G ra

d e

6 – 0 .0

0 0 1

0 .0

0 0 3

2 0 .0

0 1

2 0 .0

1 0 .0

7 2

0 .0

0 1

2 0 .0

0 0 2

0 .0

0 0 3

2 0 .0

0 3

2 0 .0

0 2

0 .0

7 2

0 .0

0 0 1

G ra

d e

7 – 0 .0

0 0 1

0 .0

0 0 3

2 0 .0

0 1

2 0 .0

1 0 .0

7 2

0 .0

0 1

2 0 .0

0 0 1

0 .0

0 0 3

2 0 .0

0 2

0 .0

0 0 4

0 .0

7 0 .0

0 0 0

2 0 0 8 – 2 0 0 9

0 .0

0 4 * * *

0 .0

0 0 4

0 .0

5 0 .0

3 0 .0

9 0 .0

0 2

0 .0

0 5 * * *

0 .0

0 0 4

0 .0

6 2

0 .3

7 * * *

0 .1

0 2

0 .0

2 2 0 0 9 – 2 0 1 0

0 .0

0 3 * * *

0 .0

0 0 4

0 .0

3 2

0 .3

2 * * *

0 .0

8 2

0 .0

2 0 .0

0 3 * * *

0 .0

0 0 4

0 .0

4 2

0 .5

5 * * *

0 .0

9 2

0 .0

3 2 0 1 0 – 2 0 1 1

0 .0

0 1 * * *

0 .0

0 0 4

0 .0

2 2

0 .4

8 * * *

0 .0

8 2

0 .0

3 0 .0

0 2 * * *

0 .0

0 0 4

0 .0

2 2

0 .4

4 * * *

0 .0

9 2

0 .0

2 2 0 1 1 – 2 0 1 2

2 0 .0

0 0 1

0 .0

0 0 4

2 0 .0

0 1

2 0 .5

6 * * *

0 .0

8 2

0 .0

4 0 .0

0 0 4

0 .0

0 0 4

0 .0

1 2

0 .7

1 * * *

0 .0

8 2

0 .0

4 2 0 1 2 – 2 0 1 3

– 0 .0

0 1

0 .0

0 0 4

2 0 .0

1 2

0 .2

7 * * *

0 .0

7 2

0 .0

2 2

0 .0

0 0 4

0 .0

0 0 3

2 0 .0

1 2

0 .0

2 0 .0

8 2

0 .0

0 1

2 0 1 3 – 2 0 1 4

– 0 .0

0 1

0 .0

0 0 4

2 0 .0

1 2

0 .3

1 * * *

0 .0

8 2

0 .0

2 2

0 .0

0 1

0 .0

0 0 4

2 0 .0

1 2

0 .0

2 0 .0

9 2

0 .0

0 1

U rb

an 2

0 .0

0 0 1

0 .0

0 0 4

2 0 .0

0 2

2 0 .0

1 0 .0

6 2

0 .0

0 1

2 0 .0

0 0 4

0 .0

0 1

2 0 .0

1 2

0 .0

2 0 .0

6 2

0 .0

0 1

T o w

n 2

0 .0

0 0 3

0 .0

0 0 5

2 0 .0

0 4

2 0 .0

4 0 .0

6 2

0 .0

0 3

2 0 .0

0 0 3

0 .0

0 1

2 0 .0

0 4

2 0 .0

3 0 .0

7 2

0 .0

0 2

R u ra

l 0 .0

0 0 0

0 .0

0 0 4

0 .0

0 0 0

0 .0

0 4

0 .0

5 0 .0

0 0 3

0 .0

0 0 1

0 .0

0 0 5

0 .0

0 1

2 0 .0

4 0 .0

6 2

0 .0

0 3

M e an

d is

tr ic

t te

st p

e rf

o rm

an c e

(N A

E P

sc al

e )

0 .0

0 0 1 * *

0 .0

0 0 0

0 .0

1 0 .0

4 * * *

0 .0

0 5

0 .0

3 0 .0

0 0 2 * * *

0 .0

0 0 0

0 .0

3 0 .0

2 * * *

0 .0

1 0 .0

1 T

o ta

l d

is tr

ic t

m e m

b e rs

h ip

/1 ,0

0 0

0 .0

0 0 2 *

0 .0

0 0 1

0 .0

1 0 .0

7 * * *

0 .0

2 0 .0

1 0 .0

0 0 3 * * *

0 .0

0 0 1

0 .0

1 0 .0

5 *

0 .0

2 0 .0

1 So

ci o e co

n o m

ic st

at u s

co m

p o si

te 2

0 .0

0 0 3

0 .0

0 0 3

2 0 .0

1 2

0 .0

0 1

0 .0

3 0 .0

0 0 2

2 0 .0

0 0 4

0 .0

0 0 3

2 0 .0

2 2

0 .0

1 0 .0

4 2

0 .0

0 1

P e r

p u

p il

ex p

e n

d it

u re

s/ 1

,0 0

0 2

0 .0

0 0 2 *

0 .0

0 0 1

2 0 .0

1 0 .0

3 0 .0

2 0 .0

1 2

0 .0

0 0 2 * *

0 .0

0 0 1

2 0 .0

1 2

0 .0

1 0 .0

2 2

0 .0

0 3

G in

i co

e ff ic

ie n t

– 0 .0

0 5

0 .0

1 2

0 .0

1 0 .0

2 0 .6

4 0 .0

0 0 1

2 0 .0

1 0 .0

1 2

0 .0

1 2

0 .0

2 0 .7

3 2

0 .0

0 0 1

P e rc

e n

t H

is p

an ic

p e r

g ra

d e

0 .0

0 3 * * *

0 .0

0 1

0 .0

2 0 .4

7 * * *

0 .1

1 0 .0

2 0 .0

0 4 * * *

0 .0

0 0 5

0 .0

3 0 .2

9 *

0 .1

2 0 .0

1 P

e rc

e n

t B

la c k

p e r

g ra

d e

0 .0

0 5 * * *

0 .0

0 0 4

0 .0

4 2

0 .2

3 *

0 .0

9 2

0 .0

1 0 .0

0 5 * * *

0 .0

0 0 4

0 .0

4 2

0 .1

4 0 .1

0 2

0 .0

0 5

P e rc

e n

t o

f st

u d

e n

ts th

at ar

e E

L L

– 0 .0

0 1 * * *

0 .0

0 0 4

2 0 .0

1 0 .0

8 0 .0

0 4

2 0 .0

0 1 * * *

0 .0

0 0 3

2 0 .0

1 0 .0

9 0 .0

8 0 .0

0 4

W h it e -B

la ck

d if fe

re n ce

in fa

m il y

in co

m e

2 0 .0

0 0 1

0 .0

0 0 5

2 0 .0

0 1

2 0 .0

1 0 .0

5 2

0 .0

0 1

2 0 .0

0 0 1

0 .0

0 1

2 0 .0

0 2

2 0 .0

1 0 .0

6 2

0 .0

0 0 5

(c o n

ti n

u ed

)

1204

T a b le

7 (c

o n

ti n

u e d

)

M at

h E n g li sh

La n g u ag

e A rt s

P re

d ic

ti n g

W h it e -

B la

ck Se

g re

g at

io n

P re

d ic

ti n g

W h it e -B

la ck

T e st

Sc o re

G ap

P re

d ic

ti n g

W h it e -

B la

ck Se

g re

g at

io n

P re

d ic

ti n g

W h it e -B

la ck

T e st

Sc o re

G ap

V ar

ia b le

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

SE St

an d ar

d iz

e d

C o e ff ic

ie n t

W h it e -B

la ck

d if fe

re n ce

in p ar

e n t e d u ca

ti o n

2 0 .0

0 0 4

0 .0

0 1

2 0 .0

0 5

0 .0

2 0 .0

7 0 .0

0 1

2 0 .0

0 0 1

0 .0

0 1

2 0 .0

0 2

0 .0

2 0 .0

7 0 .0

0 1

P e rc

e n t e n ro

ll e d

in ch

ar te

r sc

h o o ls

0 .0

2 * * *

0 .0

0 0 4

0 .1

5 0 .1

9 *

0 .0

8 0 .0

1 0 .0

2 * * *

0 .0

0 0 4

0 .1

5 0 .1

1 0 .0

9 0 .0

0 4

P e rc

e n t e n ro

ll e d

in m

ag n e t sc

h o o ls

0 .0

0 1 * * *

0 .0

0 0 3

0 .0

1 2

0 .0

3 0 .0

5 2

0 .0

0 2

0 .0

0 1 * * *

0 .0

0 0 2

0 .0

1 2

0 .1

3 *

0 .0

6 2

0 .0

1 W

h it e -B

la ck

se g re

g at

io n

6 .9

2 * * *

0 .7

2 0 .0

3 9 .2

7 * * *

0 .8

1 0 .0

4 In

te rc

e p t

0 .0

0 1

0 .0

0 2

0 .2

9 0 .2

5 0 .0

0 2

0 .0

0 3

0 .3

5 0 .2

8 R e si

d u al

v ar

ia n ce

0 .0

0 1

0 .0

0 0 0

2 8 .5

2 0 .1

5 0 .0

0 1

0 .0

0 0 0

3 8 .6

9 0 .1

9 In

d ir e ct

e ff e ct

s C h ar

te r sc

h o o l e n ro

ll m

e n t an

d W

h it e -B

la ck

se g re

g at

io n

0 .1

1 * * *

0 .0

1 0 .0

1 0 .1

5 * * *

0 .0

1 0 .0

1 M

ag n e t sc

h o o l e n ro

ll m

e n t an

d W

h it e -B

la ck

se g re

g at

io n

0 .0

1 * *

0 .0

0 2

0 .0

0 0 4

0 .0

1 * * *

0 .0

0 2

0 .0

0 1

N o te

. P e rc

e n ta

g e

v ar

ia b le

s ar

e sc

al e d

in 1 0

p e rc

e n t u n it s.

V ar

ia b le

s in

b o ld

fa ce

co n tr o l fo

r th

e fi x e d

e ff e ct

s o f d is

tr ic

t m

e an

s fo

r a

g iv

e n

g ra

d e

fr o m

2 0 0 8 – 2 0 0 9

to 2 0 1 4 – 2 0 1 5 . E LL

= E n g li sh

la n g u ag

e le

ar n e r;

N A E P

= N

at io

n al

A ss

e ss

m e n t o f E d u ca

ti o n al

P ro

g re

ss .

* p

\ .0

5 . * * p

\ .0

1 . * * * p

\ .0

0 1 .

1205

gaps. This is further supported by the fact that magnet enrollment only grew about 20% from 2009 to 2015 while charter enrollment grew about 65% (NCES, 2018). Consequently, the fact that the longitudinal fixed effects models show even stronger effects for associations between higher charter enroll- ment and larger White-Black test score gaps at the district level makes a strong case for the robustness of this finding.

The indirect effects analyses suggest that racial/ethnic segregation is one mechanism through which magnet and even more so charter enrollment may be exacerbating education inequity for both Black and Hispanic students. For example, in the longitudinal fixed effects models, White-Black segregation accounts for over half of the association between charter enrollment and the White-Black ELA test score gap. This finding is consistent with literature that demonstrates associations between increases in charter enrollment and racial/ethnic segregation with widening test score gaps (Bifulco & Ladd, 2007; Fiel, 2013; Frankenberg et al., 2010). The indirect effects of higher mag- net enrollment and greater segregation are also consistent with research that demonstrates magnet schools can be associated with greater district-wide seg- regation (Harris, 2018).

However, segregation is a much weaker mediator of associations between magnet enrollment and test score gaps than of charter enrollment and gaps. While there are theoretical similarities in how charters and magnets

-0.03

-0.02

-0.01

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.10

White-Black Math White-Black ELA White-Hispanic Math White-Hispanic ELA

s n

oitai ve

d dra

d nats

pa g er

ocs tse T

Charter

School

Enrollment

Magnet

School

Enrollment

Segregation

Indirect

Effects

for Charter

Segregation

Indirect

Effects

for Magnet

***

*** ***

* ***

Figure 4. Mean difference from district mean standard deviation change in indirect

effect of racial/ethnic segregation on mean difference from district mean standard

deviation change in test score gap per 10 percentage points of students in a district

enrolled in charter and magnet schools.

Note. Significance corresponds to the significance of indirect effects. ELA = English language

arts. yp \ .1. *p \ .05. **p \ .01. ***p \ .001.

Blatt, Votruba-Drzal

1206

T a b le

R e s u

lt s

fo r

T h

re e -L

e v e l R

a n

d o

m In

te rc

e p

t M

o d

e ls

E x a m

in in

g th

e A

s s o

c ia

ti o

n B

e tw

e e n

D is

tr ic

t- L

e v e l

C h

a rt

e r

a n

d M

a g

n e t

S c h

o o

E n

ro ll

m e n

t a n

d W

h it

e -H

is p

a n

ic T

e s t

S c o

re G

a p

M at

h E n g li sh

La n g u ag

e A rt s

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

G ra

d e

3 2

1 .8

1 * * *

0 .2

1 2

0 .0

8 2 .0

2 * * *

0 .2

8 0 .0

G ra

d e

4 2

1 .8

4 * * *

0 .1

7 2

0 .0

8 1 .6

9 * * *

0 .2

3 0 .0

G ra

d e

5 2

1 .6

0 * * *

0 .1

4 2

0 .0

7 1 .3

2 * * *

0 .1

8 0 .0

G ra

d e

6 2

0 .9

3 * * *

0 .1

1 2

0 .0

4 0 .4

9 * * *

0 .1

3 0 .0

G ra

d e

7 2

0 .2

9 * * *

0 .0

9 2

0 .0

1 0 .3

2 * * *

0 .1

0 0 .0

2 0 0 8 – 2 0 0 9

0 .7

0 * * *

0 .0

8 0 .0

3 2 .5

5 * * *

0 .1

0 0 .0

2 0 0 9 – 2 0 1 0

0 .0

3 0 .0

8 0 .0

0 1

1 .9

8 * * *

0 .0

9 0 .0

2 0 1 0 – 2 0 1 1

2 0 .4

5 * * *

0 .0

8 2

0 .0

2 1 .2

8 * * *

0 .0

9 0 .0

2 0 1 1 – 2 0 1 2

2 0 .9

2 * * *

0 .0

7 2

0 .0

4 0 .1

1 0 .0

9 0 .0

0 4

2 0 1 2 – 2 0 1 3

2 0 .5

5 * * *

0 .0

7 2

0 .0

2 1 .0

5 * * *

0 .0

9 0 .0

2 0 1 3 – 2 0 1 4

2 0 .4

4 * * *

0 .0

8 2

0 .0

2 0 .9

6 * * *

0 .0

9 0 .0

U rb

an 0 .7

4 * * *

0 .1

6 0 .0

3 0 .9

3 * * *

0 .1

9 0 .0

T o w

n 2

0 .2

0 0 .1

5 2

0 .0

1 2

0 .2

3 0 .1

7 2

0 .0

R u ra

l 2

0 .2

6 0 .1

4 2

0 .0

1 2

0 .2

1 0 .1

6 2

0 .0

M e an

d is

tr ic

t te

st p e rf

o rm

an ce

(N A E P

sc al

e )

0 .0

2 * * *

0 .0

0 4

0 .0

5 2

0 .0

1 *

0 .0

0 5

2 0 .0

T o ta

l d is

tr ic

t m

e m

b e rs

h ip

/1 ,0

0 0

0 .0

3 * * *

0 .0

0 4

0 .0

6 0 .0

3 * * *

0 .0

1 0 .0

So ci

o e co

n o m

ic st

at u s

co m

p o si

te 4 .3

3 * * *

0 .1

7 0 .4

4 4 .6

3 * * *

0 .2

1 0 .4

P e r p u p il

e x p e n d it u re

s/ 1 ,0

0 0

0 .0

0 5

0 .0

1 0 .0

0 2

0 .0

2 0 .0

G in

i co

e ff ic

ie n t

4 7 .5

2 * * *

2 .6

0 0 .2

6 5 8 .3

4 * * *

3 .3

7 0 .3

P e rc

e n t H

is p an

ic p e r g ra

d e

0 .0

6 0 .0

5 0 .0

1 0 .2

3 * * *

0 .0

6 0 .0

(c o n

ti n

u ed

)

1207

T a b le

8 (c

o n

ti n

u e d

)

M at

h E n g li sh

La n g u ag

e A rt s

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

P e rc

e n t B la

ck p e r g ra

d e

0 .4

1 * * *

0 .0

7 0 .0

7 0 .4

1* * *

0 .0

8 0 .0

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

C e n tr al

A m

e ri ca

0 .0

9 0 .1

2 0 .0

1 0 .1

0 0 .1

5 0 .0

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

C u b a

2 0 .1

0 0 .2

2 2

0 .0

0 4

2 0 .0

6 0 .2

6 2

0 .0

0 3

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

M e x ic

o 2

0 .0

8 0 .0

9 2

0 .0

3 0 .0

1 0 .1

1 0 .0

0 5

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

P u e rt o

R ic

o 0 .1

7 0 .1

0 0 .0

4 0 .1

6 0 .1

2 0 .0

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

So u th

A m

e ri ca

0 .0

4 0 .1

4 0 .0

0 3

0 .1

8 0 .1

7 0 .0

P e rc

e n t o f st

u d e n ts

th at

ar e

E LL

0 .2

2 * * *

0 .0

5 0 .0

2 0 .2

4* * *

0 .0

6 0 .0

W h it e -H

is p an

ic d if fe

re n ce

in fa

m il y

in co

m e

1 .4

4 * * *

0 .1

9 0 .0

7 1 .5

0* * *

0 .2

6 0 .0

W h it e -H

is p an

ic d if fe

re n ce

in p ar

e n t e d u ca

ti o n

5 .1

8 * * *

0 .2

6 0 .2

3 7 .6

8* * *

0 .3

3 0 .3

P e rc

e n t e n ro

ll e d

in ch

ar te

r sc

h o o ls

0 .0

2 0 .0

6 0 .0

0 2

0 .0

9 0 .0

7 0 .0

P e rc

e n t e n ro

ll e d

in m

ag n e t sc

h o o ls

0 .1

5 * *

0 .0

5 0 .0

2 0 .1

6* *

0 .0

6 0 .0

In te

rc e p t

2 1 3 .5

4 * * *

1 .6

7 2

1 2 .9

9 * * *

2 .0

D is

tr ic

t v ar

ia n ce

2 0 .1

4 0 .5

7 2 9 .7

7 0 .8

St at

e v ar

ia n ce

6 .7

8 1 .6

4 4 .6

1 1 .1

R e si

d u al

v ar

ia n ce

3 4 .8

5 0 .1

7 5 0 .0

6 0 .2

x 2

3 6 8 3 7 .6

0 * * *

3 5 4 3 5 .4

4 * * *

N o te

. P e rc

e n ta

g e

v ar

ia b le

s ar

e sc

al e d

in 1 0

p e rc

e n t u n it s.

E LL

= E n g li sh

la n g u ag

e le

ar n e r;

N A E P

= N

at io

n al

A ss

e ss

m e n t o f E d u ca

ti o n al

P ro

g re

ss .

* p

\ .0

5 . * * p

\ .0

1 . * * * p

\ .0

0 1 .

1208

T a b le

R e s u

lt s

fo r

L o

n g

it u

d in

a l

F ix

e d

E ff

e c ts

M o

d e ls

E x a m

in in

g th

e A

s s o

c ia

ti o

n B

e tw

e e n

D is

tr ic

t- L

e v e l

C h

a rt

e r

a n

d M

a g

n e t

S c h

o o

E n

ro ll

m e n

t a n

d W

h it

e -H

is p

a n

ic T

e s t

S c o

re G

a p

M at

h E n g li sh

La n g u ag

e A rt s

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

G ra

d e

3 0 .0

6 0 .0

0 2

0 .0

3 0 .0

7 0 .0

0 2

G ra

d e

4 0 .0

3 0 .0

6 0 .0

0 2

0 .0

3 0 .0

7 0 .0

0 2

G ra

d e

5 0 .0

3 0 .0

6 0 .0

0 2

0 .0

3 0 .0

7 0 .0

0 2

G ra

d e

6 0 .0

3 0 .0

6 0 .0

0 2

0 .0

2 0 .0

7 0 .0

0 1

G ra

d e

7 2

0 .0

0 1

0 .0

7 – 0 .0

0 0 1

0 .0

1 0 .0

7 0 .0

0 1

2 0 0 8 – 2 0 0 9

0 .4

8 * * *

0 .0

7 0 .0

3 2 .1

4 * * *

0 .0

9 0 .1

2 0 0 9 – 2 0 1 0

2 0 .1

3 0 .0

7 2

0 .0

1 1 .6

1 * * *

0 .0

9 0 .0

2 0 1 0 – 2 0 1 1

2 0 .5

7 * * *

0 .0

7 2

0 .0

4 0 .9

9 * * *

0 .0

8 0 .0

2 0 1 1 – 2 0 1 2

2 0 .8

9 * * *

0 .0

6 2

0 .0

6 2

0 .0

2 0 .0

8 2

0 .0

0 1

2 0 1 2 – 2 0 1 3

2 0 .5

2 * * *

0 .0

6 2

0 .0

4 0 .9

1 * * *

0 .0

7 0 .0

2 0 1 3 – 2 0 1 4

2 0 .4

0 * * *

0 .0

7 2

0 .0

2 0 .8

8 * * *

0 .0

8 0 .0

U rb

an 2

0 .0

2 0 .0

5 2

0 .0

0 1

0 .0

0 0 .0

6 0 .0

0 0 1

T o w

n 2

0 .0

3 0 .0

5 2

0 .0

0 3

2 0 .0

6 2

0 .0

0 1

R u ra

l 0 .0

1 0 .0

5 0 .0

0 0 5

0 .0

6 0 .0

0 3

M e a n

d is

tr ic

t te

st p

e rf

o rm

a n

c e

(N A

E P

sc a le

) 0 .0

1 *

0 .0

0 4

0 .0

1 0 .0

0 4

T o

ta l

d is

tr ic

t m

e m

b e rs

h ip

/1 ,0

0 0

0 .0

7 * * *

0 .0

2 0 .0

1 0 .1

1 * * *

0 .0

2 0 .0

So ci

o e co

n o m

ic st

at u s

co m

p o si

te 0 .0

0 0

0 .0

3 0 .0

0 0 0

0 .0

2 0 .0

4 0 .0

0 3

P e r

p u

p il

e x

p e n

d it

u re

s/ 1

,0 0

0 2

0 .0

3 *

0 .0

1 2

0 .0

1 2

0 .0

2 0 .0

2 2

0 .0

0 4

G in

i co

e ff ic

ie n t

2 0 .0

2 0 .2

9 – 0 .0

0 0 2

2 0 .2

2 0 .3

5 2

0 .0

0 2

P e rc

e n

t H

is p

a n

ic p

e r

g ra

d e

2 0 .2

0 * *

0 .0

7 2

0 .0

1 2

0 .1

9 *

0 .0

8 2

0 .0

(c o n

ti n

u ed

)

1209

T a b le

9 (c

o n

ti n

u e d

)

M at

h E n g li sh

La n g u ag

e A rt s

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

P e rc

e n

t B

la c k

p e r

g ra

d e

0 .2

7 *

0 .1

3 0 .0

1 0 .3

4 *

0 .1

5 0 .0

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

C e n tr al

A m

e ri ca

2 0 .0

1 0 .0

5 2

0 .0

0 1

0 .0

1 0 .0

6 0 .0

0 2

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

C u b a

2 0 .0

0 1

0 .0

3 – 0 .0

0 0 1

0 .0

1 0 .0

3 0 .0

0 0 5

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

M e x ic

o 0 .0

0 3

0 .0

5 0 .0

0 2

0 .0

0 0 2

0 .0

6 0 .0

0 0 1

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

P u e rt o

R ic

o 2

0 .0

0 4

0 .0

2 2

0 .0

0 1

0 .0

1 0 .0

2 0 .0

0 3

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

So u th

A m

e ri ca

2 0 .0

0 2

0 .0

2 – 0 .0

0 0 3

2 0 .0

0 1

0 .0

2 – 0 .0

0 0 1

P e rc

e n

t o

f st

u d

e n

ts th

a t

a re

E L L

0 .1

2 *

0 .0

5 0 .0

1 0 .1

2 *

0 .0

5 0 .0

W h it e -H

is p an

ic d if fe

re n ce

in fa

m il y

in co

m e

0 .0

0 1

0 .0

5 0 .0

0 0 1

0 .0

0 1

0 .0

1 0 .0

0 0 1

W h it e -H

is p an

ic d if fe

re n ce

in p ar

e n t e d u ca

ti o n

2 0 .0

5 2

0 .0

0 1

2 0 .0

6 0 .0

1 2

0 .0

0 4

P e rc

e n

t e n

ro ll

e d

in c h

a rt

e r

sc h

o o

ls 2

0 .0

0 2

0 .0

6 – 0 .0

0 0 1

0 .0

7 0 .0

0 3

P e rc

e n

t e n

ro ll

e d

in m

a g n

e t

sc h

o o

ls 0 .0

2 0 .0

5 0 .0

0 1

2 0 .0

5 0 .0

6 2

0 .0

0 2

In te

rc e p t

0 .3

7 0 .2

6 2

1 .0

0 * * *

0 .3

R e si

d u al

v ar

ia n ce

2 6 .5

4 0 .1

3 3 9 .5

3 0 .1

N o te

. P e rc

e n ta

g e

v ar

ia b le

s ar

e sc

al e d

in 1 0

p e rc

e n t u n it s.

V ar

ia b le

s in

b o ld

fa ce

co n tr o l fo

r th

e fi x e d

e ff e ct

s o f d is

tr ic

t m

e an

s fo

r a

g iv

e n

g ra

d e

fr o m

2 0 0 8 – 2 0 0 9

to 2 0 1 4 – 2 0 1 5 . E LL

= E n g li sh

la n g u ag

e le

ar n e r;

N A E P

= N

at io

n al

A ss

e ss

m e n t o f E d u ca

ti o n al

P ro

g re

ss .

* p

\ .0

5 . * * p

\ .0

1 . * * * p

\ .0

0 1 .

1210

T a b le

1 0

R e s u

lt s

fo r

T h

re e -L

e v e lR

a n

d o

m In

te rc

e p

t M

o d

e ls

E x a m

in in

g th

e In

d ir

e c t E

ff e c ts

o f W

h it

e -H

is p

a n

ic S

e g

re g

a ti

o n

th e

A s s o

c ia

ti o

n s

B e tw

e e n

D is

tr ic

t- L

e v e l

C h

a rt

e r

a n

d M

a g

n e t

S c h

o o

l E

n ro

ll m

e n

t a n

d W

h it

e -H

is p

a n

ic T

e s t

S c o

re G

a p

M at

h E n g li sh

La n g u ag

e A rt s

P re

d ic

ti n g

W h it e -

H is

p an

ic Se

g re

g at

io n

P re

d ic

ti n g

W h it e -

H is

p an

ic T e st

Sc o re

G ap

P re

d ic

ti n g

W h it e -

H is

p an

ic Se

g re

g at

io n

P re

d ic

ti n g

W h it e -

H is

p an

ic T e st

Sc o re

G ap

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

G ra

d e

3 0 .0

6 * * *

0 .0

0 1

0 .2

6 2

1 .9

1 * * *

0 .2

1 2

0 .0

8 0 .0

6 * * *

0 .0

0 1

0 .2

7 1 .3

7 * * *

0 .2

8 0 .0

G ra

d e

4 0 .0

5 * * *

0 .0

0 1

0 .2

4 2

1 .9

3 * * *

0 .1

8 2

0 .0

8 0 .0

6 * * *

0 .0

0 1

0 .2

5 1 .0

8 * * *

0 .2

3 0 .0

G ra

d e

5 0 .0

5 * * *

0 .0

0 1

0 .2

0 2

1 .6

8 * * *

0 .1

4 2

0 .0

7 0 .0

5 * * *

0 .0

0 1

0 .2

1 0 .8

0 * * *

0 .1

8 0 .0

G ra

d e

6 0 .0

2 * * *

0 .0

0 1

0 .0

7 2

0 .9

6 * * *

0 .1

1 2

0 .0

4 0 .0

2 * * *

0 .0

0 1

0 .0

8 0 .3

1 *

0 .1

3 0 .0

G ra

d e

7 0 .0

0 3 * * *

0 .0

0 0 5

0 .0

1 2

0 .3

0 * * *

0 .0

9 2

0 .0

1 0 .0

0 4 * * *

0 .0

0 0 5

0 .0

2 0 .2

7 * *

0 .1

0 0 .0

2 0 0 8 – 2 0 0 9

0 .0

1 * * *

0 .0

0 0 5

0 .0

4 0 .6

8 * * *

0 .0

8 0 .0

3 0 .0

1 * * *

0 .0

0 0 5

0 .0

4 2 .4

5 * * *

0 .1

0 0 .0

2 0 0 9 – 2 0 1 0

0 .0

1 * * *

0 .0

0 0 5

0 .0

3 0 .0

2 0 .0

8 0 .0

0 1

0 .0

1 * * *

0 .0

0 0 5

0 .0

3 1 .8

9 * * *

0 .0

9 0 .0

2 0 1 0 – 2 0 1 1

0 .0

0 4 * * *

0 .0

0 0 4

0 .0

2 2

0 .4

6 * * *

0 .0

8 2

0 .0

2 0 .0

0 4 * * *

0 .0

0 0 4

0 .0

2 1 .2

3 * * *

0 .0

9 0 .0

2 0 1 1 – 2 0 1 2

0 .0

0 1 * * *

0 .0

0 0 4

0 .0

1 2

0 .9

2 * * *

0 .0

7 2

0 .0

4 0 .0

0 1 * *

0 .0

0 0 4

0 .0

1 0 .0

9 0 .0

0 3

2 0 1 2 – 2 0 1 3

0 .0

0 0 5

0 .0

0 0 4

0 .0

0 2

2 0 .5

5 * * *

0 .0

7 2

0 .0

2 0 .0

0 0 4

0 .0

0 0 4

0 .0

0 2

1 .0

5 * * *

0 .0

9 0 .0

2 0 1 3 – 2 0 1 4

2 0 .0

0 0 5

0 .0

0 0 5

2 0 .0

0 2

2 0 .4

4 * * *

0 .0

8 2

0 .0

2 2

0 .0

0 0 5

0 .0

0 0 4

2 0 .0

0 2

0 .9

7 * * *

0 .0

9 0 .0

U rb

an 0 .0

1 * * *

0 .0

0 1

0 .0

3 0 .7

2 * * *

0 .1

6 0 .0

3 0 .0

0 5 * * *

0 .0

0 1

0 .0

2 0 .8

2 * * *

0 .1

9 0 .0

T o w

n 2

0 .0

1 * * *

0 .0

0 1

2 0 .0

2 2

0 .1

8 0 .1

5 2

0 .0

1 2

0 .0

1 * * *

0 .0

0 1

2 0 .0

3 2

0 .1

0 0 .1

7 2

0 .0

0 4

R u ra

l 2

0 .0

0 3 * * *

0 .0

0 1

2 0 .0

1 2

0 .2

5 0 .1

4 2

0 .0

1 2

0 .0

0 3 * * *

0 .0

0 1

2 0 .0

1 2

0 .1

4 0 .1

6 2

0 .0

0 5

M e an

d is

tr ic

t te

st p e rf

o rm

an ce

(N A E P

sc al

e )

0 .0

0 0 3 * * *

0 .0

0 0 0

0 .0

8 0 .0

2 * * *

0 .0

0 4

0 .0

5 0 .0

0 0 4 * * *

0 .0

0 0 0

0 .1

0 2

0 .0

1 * *

0 .0

0 5

2 0 .0

T o ta

l d is

tr ic

t m

e m

b e rs

h ip

/1 ,0

0 0

0 .0

0 1 * * *

0 .0

0 0 0

0 .3

0 0 .0

2 * * *

0 .0

0 4

0 .0

6 0 .0

0 1 * * *

0 .0

0 0 0

0 .3

0 0 .0

2 * *

0 .0

0 5

0 .0

So ci

o e co

n o m

ic st

at u s

co m

p o si

te 0 .0

1 * * *

0 .0

0 2

0 .0

8 4 .3

2 * * *

0 .1

7 0 .4

3 0 .0

0 3

0 .0

0 2

0 .0

3 4 .6

0 * * *

0 .2

1 0 .3

P e r p u p il

e x p e n d it u re

s/ 1 ,0

0 0

0 .0

0 0 2 *

0 .0

0 0 1

0 .0

1 0 .0

0 5

0 .0

1 0 .0

0 2

0 .0

0 0 1

0 .0

0 0 1

0 .0

1 0 .0

2 0 .0

G in

i co

e ff ic

ie n t

0 .3

4 * * *

0 .0

4 0 .2

0 4 7 .0

4 * * *

2 .5

9 0 .2

6 0 .3

2 * * *

0 .0

3 0 .1

8 5 5 .3

2 * * *

3 .2

7 0 .2

P e rc

e n t H

is p an

ic p e r g ra

d e

0 .0

0 3 * * *

0 .0

0 0 4

0 .0

7 0 .0

5 0 .0

1 0 .0

0 3 * * *

0 .0

0 0 4

0 .0

7 0 .2

0 * * *

0 .0

6 0 .0

P e rc

e n t B la

ck p e r g ra

d e

0 .0

0 5 * * *

0 .0

0 1

0 .0

8 0 .4

0 * * *

0 .0

7 0 .0

0 4 * * *

0 .0

0 1

0 .0

7 0 .3

2 * * *

0 .0

8 0 .0

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

C e n tr al

A m

e ri ca

2 0 .0

0 0 3

0 .0

0 1

2 0 .0

0 3

0 .0

9 0 .1

2 0 .0

1 0 .0

0 0 2

0 .0

0 2

0 .0

0 2

0 .1

0 0 .1

5 0 .0

(c o n

ti n

u ed

)

1211

T a b le

1 0

(c o

n ti

n u

e d

)

M at

h E n g li sh

La n g u ag

e A rt s

P re

d ic

ti n g

W h it e -

H is

p an

ic Se

g re

g at

io n

P re

d ic

ti n g

W h it e -

H is

p an

ic T e st

Sc o re

G ap

P re

d ic

ti n g

W h it e -

H is

p an

ic Se

g re

g at

io n

P re

d ic

ti n g

W h it e -

H is

p an

ic T e st

Sc o re

G ap

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

C u b a

2 0 .0

0 1

0 .0

0 3

2 0 .0

0 3

2 0 .1

0 0 .2

2 2

0 .0

0 4

2 0 .0

0 1

0 .0

0 3

2 0 .0

0 4

2 0 .0

5 0 .2

6 2

0 .0

0 2

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

M e x ic

o 2

0 .0

0 2 *

0 .0

0 1

2 0 .0

8 2

0 .0

8 0 .0

9 2

0 .0

3 2

0 .0

0 2

0 .0

0 1

2 0 .0

7 0 .0

3 0 .1

0 0 .0

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

P u e rt o

R ic

o 2

0 .0

0 2

0 .0

0 1

2 0 .0

3 0 .1

7 0 .1

0 0 .0

4 2

0 .0

0 2

0 .0

0 1

2 0 .0

3 0 .1

8 0 .1

1 0 .0

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

So u th

A m

e ri ca

2 0 .0

0 1

0 .0

0 2

2 0 .0

1 0 .0

4 0 .1

3 0 .0

0 4

2 0 .0

0 2

0 .0

0 2

2 0 .0

2 0 .1

9 0 .1

6 0 .0

P e rc

e n t o f st

u d e n ts

th at

ar e

E LL

0 .0

0 0 3

0 .0

0 0 3

0 .0

0 3

0 .2

1 * * *

0 .0

5 0 .0

2 2

0 .0

0 0 1

0 .0

0 0 3

2 0 .0

0 1

0 .2

4 * * *

0 .0

6 0 .0

W h it e -H

is p an

ic d if fe

re n ce

in fa

m il y

in co

m e

0 .0

0 4 *

0 .0

0 2

0 .0

2 1 .4

3 * * *

0 .1

8 0 .0

7 0 .0

0 5 * *

0 .0

0 2

0 .0

2 1 .4

6 * * *

0 .2

6 0 .0

W h it e -H

is p an

ic d if fe

re n ce

in p ar

e n t e d u ca

ti o n

0 .0

3 * * *

0 .0

0 3

0 .1

2 5 .1

4 * * *

0 .2

6 0 .2

3 0 .0

3 * * *

0 .0

0 4

0 .1

3 7 .4

1 * * *

0 .3

2 0 .2

P e rc

e n t e n ro

ll e d

in ch

ar te

r sc

h o o ls

0 .0

1 * * *

0 .0

0 0 4

0 .0

8 0 .0

1 0 .0

6 0 .0

0 0 5

0 .0

1 * * *

0 .0

0 0 4

0 .0

8 2

0 .0

1 0 .0

7 2

0 .0

0 1

P e rc

e n t e n ro

ll e d

in m

ag n e t sc

h o o ls

0 .0

0 1 * * *

0 .0

0 0 3

0 .0

2 0 .1

4 * *

0 .0

5 0 .0

2 0 .0

0 1 * * *

0 .0

0 0 3

0 .0

1 0 .1

5 *

0 .0

6 0 .0

W h it e -H

is p an

ic se

g re

g at

io n

1 .7

0 * *

0 .5

5 0 .0

2 1 1 .1

1 * * *

0 .6

3 0 .0

In te

rc e p t

2 0 .2

1 * * *

0 .0

2 1 .4

4 * * *

0 .5

2 2

0 .2

2 * * *

0 .0

2 1 .4

7 *

0 .6

D is

tr ic

t v ar

ia n ce

0 .0

0 4

0 .0

0 0 1

1 9 .9

1 0 .5

7 0 .0

0 3

0 .0

0 0 1

2 8 .1

7 0 .8

St at

e v ar

ia n ce

0 .0

0 0 3

0 .0

0 0 1

6 .7

6 1 .6

3 0 .0

0 0 3

0 .0

0 0 1

4 .4

3 1 .1

R e si

d u al

v ar

ia n ce

0 .0

0 1

0 .0

0 0 0

3 4 .8

6 0 .1

7 0 .0

0 1

0 .0

0 0 0

5 0 .0

0 0 .2

In d ir e ct

e ff e ct

C h ar

te r sc

h o o l e n ro

ll m

e n t an

d W

h it e -H

is p an

ic se

g re

g at

io n

0 .0

1 * *

0 .0

0 5

0 .0

0 1

0 .0

9 * * *

0 .0

1 0 .0

M ag

n e t sc

h o o l e n ro

ll m

e n t an

d W

h it e -H

is p an

ic se

g re

g at

io n

0 .0

0 2 *

0 .0

0 1

0 .0

0 0 3

0 .0

1 * *

0 .0

0 4

0 .0

0 1

N o te

. P e rc

e n ta

g e

v ar

ia b le

s ar

e sc

al e d

in 1 0

p e rc

e n t u n it s.

E LL

= E n g li sh

la n g u ag

e le

ar n e r;

N A E P

= N

at io

n al

A ss

e ss

m e n t o f E d u ca

ti o n al

P ro

g re

ss .

* p

\ .0

5 . * * p

\ .0

1 . * * * p

\ .0

0 1 .

1212

T a b le

1 1

R e s u

lt s

fo r

L o

n g

it u

d in

a l F

ix e d

E ff

e c ts

M o

d e ls

E x a m

in in

g th

e In

d ir

e c t

E ff

e c ts

o f

W h

it e -H

is p

a n

ic S

e g

re g

a ti

o n

th e

A s s o

c ia

ti o

n s

B e tw

e e n

D is

tr ic

t- L

e v e l

C h

a rt

e r

a n

d M

a g

n e t

S c h

o o

l E

n ro

ll m

e n

t a n

d W

h it

e -H

is p

a n

ic T

e s t

S c o

re G

a p

M at

h E n g li sh

La n g u ag

e A rt s

P re

d ic

ti n g

W h it e -

H is

p an

ic Se

g re

g at

io n

P re

d ic

ti n g

W h it e -

H is

p an

ic T e st

Sc o re

G ap

P re

d ic

ti n g

W h it e -

H is

p an

ic Se

g re

g at

io n

P re

d ic

ti n g

W h it e -H

is p an

T e st

Sc o re

G ap

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

G ra

d e

3 2

0 .0

0 0 3

0 .0

0 0 3

2 0 .0

0 1

0 .0

3 0 .0

6 0 .0

0 2

2 0 .0

0 1 *

0 .0

0 0 2

2 0 .0

0 2

0 .0

3 0 .0

7 0 .0

0 2

G ra

d e

4 2

0 .0

0 0 3

0 .0

0 0 3

2 0 .0

0 2

0 .0

3 0 .0

6 0 .0

0 2

2 0 .0

0 0 4

0 .0

0 0 2

2 0 .0

0 2

0 .0

4 0 .0

7 0 .0

0 2

G ra

d e

5 2

0 .0

0 0 3

0 .0

0 0 3

2 0 .0

0 1

0 .0

3 0 .0

6 0 .0

0 2

2 0 .0

0 1 *

0 .0

0 0 2

2 0 .0

0 2

0 .0

3 0 .0

7 0 .0

0 2

G ra

d e

6 0 .0

0 0 1

0 .0

0 0 3

0 .0

0 0 3

0 .0

3 0 .0

6 0 .0

0 2

2 0 .0

0 0 1

0 .0

0 0 2

2 0 .0

0 0 5

0 .0

2 0 .0

7 0 .0

0 1

G ra

d e

7 0 .0

0 0 0

0 .0

0 0 3

0 .0

0 0 0

2 0 .0

0 1

0 .0

7 2

0 .0

0 0 1

2 0 .0

0 0 1

0 .0

0 0 2

2 0 .0

0 0 2

0 .0

1 0 .0

7 0 .0

0 1

2 0 0 8 – 2 0 0 9

0 .0

1 * * *

0 .0

0 0 3

0 .0

3 0 .4

2 * * *

0 .0

8 0 .0

3 0 .0

1 * * *

0 .0

0 0 3

0 .0

3 2 .0

5 * * *

0 .0

9 0 .1

2 0 0 9 – 2 0 1 0

0 .0

1 * * *

0 .0

0 0 3

0 .0

2 2

0 .1

7 *

0 .0

7 2

0 .0

1 0 .0

1 * * *

0 .0

0 0 3

0 .0

2 1 .5

5 * * *

0 .0

9 0 .0

2 0 1 0 – 2 0 1 1

0 .0

0 2 * * *

0 .0

0 0 3

0 .0

1 2

0 .5

8 * * *

0 .0

7 2

0 .0

4 0 .0

0 2 * * *

0 .0

0 0 3

0 .0

1 0 .9

6 * * *

0 .0

8 0 .0

2 0 1 1 – 2 0 1 2

0 .0

0 0 4

0 .0

0 0 3

0 .0

0 2

2 0 .8

9 * * *

0 .0

6 2

0 .0

6 0 .0

0 0 4

0 .0

0 0 3

0 .0

0 2

2 0 .0

3 0 .0

8 2

0 .0

0 2

2 0 1 2 – 2 0 1 3

2 0 .0

0 0 2

0 .0

0 0 3

2 0 .0

0 1

2 0 .5

2 * * *

0 .0

6 2

0 .0

4 2

0 .0

0 0 1

0 .0

0 0 3

2 0 .0

0 0 5

0 .9

1 * * *

0 .0

7 0 .0

2 0 1 3 – 2 0 1 4

2 0 .0

0 1

0 .0

0 0 3

2 0 .0

0 2

2 0 .3

9 * * *

0 .0

7 2

0 .0

2 2

0 .0

0 1 *

0 .0

0 0 3

2 0 .0

0 2

0 .8

9 * * *

0 .0

8 0 .0

U rb

an 2

0 .0

0 0 3

0 .0

0 0 2

2 0 .0

0 1

2 0 .0

1 0 .0

5 2

0 .0

0 1

2 0 .0

0 0 3

0 .0

0 0 2

2 0 .0

0 1

0 .0

1 0 .0

6 0 .0

0 0 5

T o w

n 2

0 .0

0 0 4

0 .0

0 0 2

2 0 .0

0 2

2 0 .0

3 0 .0

5 2

0 .0

0 2

2 0 .0

0 1 *

0 .0

0 0 2

2 0 .0

0 2

2 0 .0

1 0 .0

6 2

0 .0

0 1

R u ra

l 2

0 .0

0 0 3

0 .0

0 0 2

2 0 .0

0 1

0 .0

1 0 .0

5 0 .0

0 1

2 0 .0

0 0 3

0 .0

0 0 2

2 0 .0

0 1

0 .0

6 0 .0

0 4

M e an

d is

tr ic

t te

st p

e rf

o rm

an c e

(N A

E P

sc al

e )

0 .0

0 0 1 * * *

0 .0

0 0 0

0 .0

0 4

0 .0

1 *

0 .0

0 4

0 .0

1 0 .0

0 0 1 * * *

0 .0

0 0 0

0 .0

1 0 .0

0 0 .0

1 0 .0

0 3

T o

ta l

d is

tr ic

t m

e m

b e rs

h ip

/1 ,0

0 0

0 .0

0 0 2 * *

0 .0

0 0 1

0 .0

0 2

0 .0

7 * * *

0 .0

2 0 .0

1 0 .0

0 0 2 * *

0 .0

0 0 1

0 .0

0 2

0 .1

1 * * *

0 .0

2 0 .0

So ci

o e co

n o m

ic st

at u s

co m

p o si

te 2

0 .0

0 0 1

0 .0

0 0 1

2 0 .0

0 1

0 .0

0 0 4

0 .0

3 0 .0

0 0 1

2 0 .0

0 0 1

0 .0

0 0 1

2 0 .0

0 1

0 .0

2 0 .0

4 0 .0

0 3

P e r

p u

p il

ex p

e n

d it

u re

s/ 1

,0 0

0 0 .0

0 0 0

0 .0

0 0 1

0 .0

0 0 0

2 0 .0

3 *

0 .0

1 2

0 .0

1 0 .0

0 0 0

0 .0

0 0 1

2 0 .0

0 0 2

2 0 .0

2 2

0 .0

0 4

G in

i co

e ff ic

ie n t

2 0 .0

0 0 1

0 .0

0 2

0 .0

0 0 0

2 0 .0

8 0 .5

5 2

0 .0

0 1

0 .0

0 2

0 .0

0 2

0 .0

0 1

0 .0

9 0 .6

5 0 .0

0 1

P e rc

e n

t H

is p

an ic

p e r

g ra

d e

2 0 .0

0 1 * *

0 .0

0 0 3

2 0 .0

0 3

2 0 .1

9 * *

0 .0

7 2

0 .0

1 2

0 .0

0 0 5

0 .0

0 0 3

2 0 .0

0 2

2 0 .1

9 *

0 .0

8 2

0 .0

P e rc

e n

t B

la c k

p e r

g ra

d e

0 .0

1 * * *

0 .0

0 1

0 .0

1 0 .2

1 0 .1

3 0 .0

1 0 .0

1 * * *

0 .0

0 1

0 .0

1 0 .2

5 0 .1

5 0 .0

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

C e n tr al

A m

e ri ca

0 .0

0 0 3 *

0 .0

0 0 1

2 0 .0

0 3

0 .0

0 1

0 .0

3 0 .0

0 0 2

2 0 .0

0 0 3 * *

0 .0

0 0 1

2 0 .0

0 3

0 .0

1 0 .0

3 0 .0

0 2

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

C u b a

2 0 .0

0 0 2

0 .0

0 0 2

2 0 .0

0 1

0 .0

1 0 .0

5 0 .0

0 0 4

2 0 .0

0 0 2

0 .0

0 0 2

2 0 .0

0 1

0 .0

0 2

0 .0

6 0 .0

0 0 1

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

M e x ic

o 2

0 .0

0 0 3 * * *

0 .0

0 0 1

2 0 .0

1 2

0 .0

0 2

0 .0

2 2

0 .0

0 1

2 0 .0

0 0 3 * * *

0 .0

0 0 1

2 0 .0

1 0 .0

2 0 .0

(c o n

ti n

u ed

)

1213

T a b le

1 1

(c o

n ti

n u

e d

)

M at

h E n g li sh

La n g u ag

e A rt s

P re

d ic

ti n g

W h it e -

H is

p an

ic Se

g re

g at

io n

P re

d ic

ti n g

W h it e -

H is

p an

ic T e st

Sc o re

G ap

P re

d ic

ti n g

W h it e -

H is

p an

ic Se

g re

g at

io n

P re

d ic

ti n g

W h it e -H

is p an

T e st

Sc o re

G ap

V ar

ia b le

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

C o e ff ic

ie n t

St an

d ar

d iz

e d

C o e ff ic

ie n t

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

P u e rt o

R ic

o 2

0 .0

0 0 2 *

0 .0

0 0 1

2 0 .0

0 5

0 .0

0 0 1

0 .0

2 0 .0

0 0 0

2 0 .0

0 0 3 * *

0 .0

0 0 1

2 0 .0

1 0 .0

0 2

0 .0

2 0 .0

0 1

P e rc

e n t o f H

is p an

ic p o p u la

ti o n

fr o m

So u th

A m

e ri ca

2 0 .0

0 0 3 * *

0 .0

0 0 1

2 0 .0

0 3

0 .0

0 1

0 .0

3 0 .0

0 0 1

2 0 .0

0 0 4 * * *

0 .0

0 0 1

2 0 .0

0 4

2 0 .0

3 2

0 .0

0 2

P e rc

e n

t o

f st

u d

e n

ts th

at ar

e E

L L

2 0 .0

0 1 * *

0 .0

0 0 2

2 0 .0

0 2

0 .1

3 * *

0 .0

5 0 .0

1 2

0 .0

0 1 * * *

0 .0

0 0 2

2 0 .0

0 3

0 .1

3 *

0 .0

5 0 .0

W h it e -H

is p an

ic d if fe

re n ce

in fa

m il y

in co

m e

2 0 .0

0 0 2

0 .0

0 0 2

2 0 .0

0 1

0 .0

0 1

0 .0

5 0 .0

0 0 1

2 0 .0

0 0 1

0 .0

0 0 2

0 .0

0 0 4

0 .0

0 1

0 .0

6 0 .0

0 0 1

W h it e -H

is p an

ic d if fe

re n ce

in p ar

e n t e d u ca

ti o n

0 .0

0 1 * * *

0 .0

0 0 2

0 .0

0 4

2 0 .0

5 2

0 .0

0 2

0 .0

0 1 * *

0 .0

0 0 2

0 .0

0 3

2 0 .0

7 0 .0

7 2

0 .0

0 4

P e rc

e n

t e n

ro ll

e d

in ch

ar te

r sc

h o

o ls

0 .0

1 * * *

0 .0

0 0 2

0 .0

2 2

0 .0

6 0 .0

6 2

0 .0

0 3

0 .0

1 * * *

0 .0

0 0 2

0 .0

3 2

0 .0

3 0 .0

7 2

0 .0

0 1

P e rc

e n

t e n

ro ll

e d

in m

ag n

e t

sc h

o o

ls 0 .0

0 1 *

0 .0

0 0 2

0 .0

0 2

0 .0

1 0 .0

5 0 .0

0 1

0 .0

0 0 5 *

0 .0

0 0 2

0 .0

0 2

2 0 .0

6 0 .0

6 2

0 .0

0 3

W h

it e -H

is p

an ic

se g re

g at

io n

7 .9

4 * * *

0 .7

9 0 .1

3 1 2 .0

6 * * *

0 .9

3 0 .1

In te

rc e p t

2 0 .0

0 0 2

0 .0

0 1

0 .3

8 0 .5

5 2

0 .0

0 0 4

0 .0

0 1

2 0 .9

8 * *

0 .6

R e si

d u al

v ar

ia n ce

0 .0

0 0 4

0 .0

0 0 0

2 6 .2

7 0 .1

3 0 .0

0 0 4

0 .0

0 0 0

3 9 .4

7 0 .1

In d ir e ct

e ff e ct

C h ar

te r sc

h o o l e n ro

ll m

e n t an

d W

h it e -H

is p an

ic se

g re

g at

io n

0 .0

5 * * *

0 .0

1 0 .0

0 3

0 .1

0 * * *

0 .0

1 0 .0

0 5

M ag

n e t sc

h o o l e n ro

ll m

e n t an

d W

h it e -H

is p an

ic se

g re

g at

io n

0 .0

0 4 *

0 .0

0 2

0 .0

0 0 2

0 .0

1 *

0 .0

0 3

0 .0

0 0 3

N o te

. P e rc

e n ta

g e

v ar

ia b le

s ar

e sc

al e d

in 1 0

p e rc

e n t u n it s.

V ar

ia b le

s in

b o ld

fa ce

co n tr o l fo

r th

e fi x e d

e ff e ct

s o f d is

tr ic

t m

e an

s fo

r a

g iv

e n

g ra

d e

fr o m

2 0 0 8 – 2 0 0 9

to 2 0 1 4 – 2 0 1 5 . E LL

= E n g li sh

la n g u ag

e le

ar n e r;

N A E P

= N

at io

n al

A ss

e ss

m e n t o f E d u ca

ti o n al

P ro

g re

ss .

* p

\ .0

5 . * * p

\ .0

1 . * * * p

\ .0

0 1 .

1214

can potentially affect district-level segregation and test score gaps, the school types still have fundamental differences that may explain why the charter find- ings are stronger. For example, at the school level, charters are predominantly racially and ethnically segregated where magnets tend to be more integrated (Bifulco & Ladd, 2007; Harris, 2018). So, charters may be further segregating districts in addition to contributing to school-level segregation, while magnets may be further segregating districts but increasing opportunities for students in the district to receive integrated education, which may buffer some of the negative effects of district-level segregation. Additionally, as previously men- tioned, charters and magnets have different enrollment, governing, account- ability, and funding structures. Public school districts often choose how to allocate resources to magnets, but charters are often governed privately and can therefore drain resources from traditional public schools for charter schools that do not serve the most disadvantaged students in a district (Dee & Fu, 2004; Mickelson et al., 2018; Riel et al., 2018).

In addition to differences between magnet and charter findings, the pat- terns of associations between school choice and test score gaps differ for White-Black and White-Hispanic gaps. This is consistent with general differ- ences in test score gap trends across different minoritized groups. For exam- ple, the White-Hispanic test score gap is generally smaller than the White- Black gap (Tables 2 and 3). Additionally, an examination of test score gaps from kindergarten to fifth grade reveals that the White-Hispanic test score gap narrows while the White-Black test score gap widens during these grades, and socioeconomic status may explain more of the White-Hispanic gap than the White-Black gap (Reardon & Galindo, 2009). The finding that charter enrollment is associated with White-Black test score gaps and not White- Hispanic test score gaps (except through segregation) is also consistent with evidence that Black students are disproportionately enrolled in charter schools and are more likely to be in segregated charters than Hispanic stu- dents (Frankenberg & Lee, 2003). In other words, the association may be more salient for Black students because they are more likely to be exposed to and affected by the proliferation of charters.

The association between magnet enrollment and White-Hispanic test score gaps is much weaker and less robust than the charter association with White-Black test score gaps. One potential explanation relates to Harris’s (2019) finding that Hispanic students were less likely to be enrolled in mag- nets compared with White students. Thus, it is possible that in districts with large proportions of magnet schools and Hispanic students, Hispanic students are more likely to be in segregated schools than they would be otherwise. However, given the less robust evidence for this finding in this study, more research is necessary to determine the validity of this association.

Overall, the effect sizes in this study are small and, unsurprisingly, the results suggest that factors such as district-level socioeconomic status, district income inequality, and racial/ethnic differences in parent education are much

School Choice and Racial/Ethnic Test Score Gaps

1215

larger predictors of test score gaps than school choice enrollment. However, considering the millions of students attending schools of choice each year and the districts, states, and independent organizations opening schools of choice at an expeditious rate with, at most, the goal of reducing racial/ethnic dispar- ities and, at least, without the goal of exacerbating inequities, even null or small effects supporting the latter should raise concern. Furthermore, small effect sizes are not unusual in macro-level school choice research due to great contextual variability. For example, one national study of charters found a 0.005 SD growth in math for charter students (Cremata et al., 2013 in Rapa et al., 2018). A White-Black math test score gap that is 0.06 SD larger per 10 percentage points of students in a district enrolled in charters (Figure 2) should be equally as compelling as a 0.005 SD growth in math when consid- ering charter expansion. Additionally, the small effects in this study are con- sistent with prior macro-level research linking increases in charter enrollment to segregation that finds small effects in part due to the immense variability across districts (Monarrez et al., 2019).

It is also important to reiterate that the lack of evidence that school choice enrollment is associated with less segregation and smaller test score gaps at the district level should be a concern. If the goal is to eliminate racial/ethnic disparities, school choice expansion may not achieve this goal without attend- ing to other important processes, such as racial/ethnic segregation. In fact, given that much of school segregation is attributable to neighborhood segre- gation, school choice theoretically provides a unique opportunity to integrate schools by drawing on many catchment areas so that students’ schools do not depend on the value of their parents’ homes (Rothwell, 2012; Riel et al., 2018). Furthermore, previously mentioned evidence of the positive effects of inte- gration certainly provides a compelling argument in favor of choice policies that result in integrated schooling opportunities for all students within and between districts (Johnson, 2019; Linn & Welner, 2007). Magnet schools, espe- cially those that rely on interdistrict enrollment, are already capitalizing on this (Goldring & Swain, 2019). Yet, as previously mentioned, there is an increasing trend of school choice segregating districts even as catchment areas in the dis- trict become more racially/ethnically heterogenous (Coughlan, 2018; Mader et al., 2018; Monarrez, 2018; Siegel-Hawley, 2014). Thus, school choice, and charter school choice in particular, could better capitalize on the capacity for district integration, or at least improve efforts to prevent school choice from actively inhibiting district integration.

There are many barriers to school integration such as a reduction in fed- eral desegregation mandates, the legality of de facto segregation, and the increasing legal difficulties to establishing enrollment policies based on racial/ethnic composition (Goldring & Smrekar, 2002; Goldring & Swain, 2019; Johnson, 2019; National Coalition on School Diversity [NCSD], 2020a, 2020b; Reardon et al., 2012). However, there are current efforts in progress to address these difficulties. For example, the ‘‘Strength in Diversity Act’’ is

Blatt, Votruba-Drzal

1216

a federal policy initiative to provide local funding for voluntary integration (NCSD, 2020b). States and districts can also adjust their implementation of existing federal policies to reduce the likelihood of school choice exacerbat- ing educational inequities through segregation. For example, districts can work with states via their ‘‘Every Student Succeeds Act’’ (ESSA) improvement plans to receive and allocate funds to improve integration efforts through school choice (NCSD, 2020a). Title IV of ESSA offers assistance for magnets and charters that explicitly prioritizes both socioeconomic and racial diversity (NCSD, 2020a). Furthermore, there are opportunities to improve federal pol- icy as it relates to school choice and education equity. For example, charter expansion is incentivized at the federal level (e.g., ‘‘Race to the Top’’) without any accountability structures for racial/ethnic equity (Boser, 2012; Frankenberg et al., 2019; J. Scott et al., 2020). Given the robust association between charter enrollment and district-level White-Black test score gaps, this is certainly an opportunity for intervention.

Thus, despite small effect sizes and the fact that school choice policies do not solely have the capacity to eliminate test score gaps, especially without broader comprehensive social policies, it is still important to intervene where the proliferation of school choice impedes progress toward education equity.

Limitations and Future Research

This secondary analysis has several limitations. The largest limitation is the inability to draw causal conclusions about whether school choice expansion is driving an increase in racial/ethnic test score gaps. The longitudinal fixed effects analyses provide compelling causal inference; however, they include time var- iant covariates operationalized with time invariant variables preventing a true longitudinal fixed effects analysis. Variable measurement also presents limita- tions to the types of questions the analyses can answer. For example, SEDAver- sion 2.1 (Reardon et al., 2018) only includes aggregated charter enrollment for third through eighth grade, so this study cannot examine enrollment by grade. This is important to consider given differences in the academic and psycholog- ical development of third graders versus eighth graders. Additionally, about 45% of all magnet students and 17% of all charter students in the United States are in secondary/high school (NCES, 2018). Students in grades nine and above are not included in this study since those grades are not captured in SEDA, and it is possible the associations would look different in older grades. Thus, the findings do not generalize to associations between school choice and test score gaps outside grades three through eight.

This study is also limited by the oversimplification of achievement dispar- ities as White-Black and White-Hispanic test score gaps. There are significant intersections between race and ethnicity, and collapsing students into single racial or ethnic categories masks the more complex relationship linking race and ethnicity to academic achievement. Furthermore, there are inherent

School Choice and Racial/Ethnic Test Score Gaps

1217

confounds with test scores as a marker of academic achievement, ranging from stereotype threat to racial/ethnic inequities in achievement-based track- ing (Berlak, 2001). However, as previously mentioned, when there are differ- ences between White students’ average test scores and Black and Hispanic students’ average test scores at the district level, this is an indicator of struc- tural education inequity and inequitable access to opportunity, resources, and support at a macro level (Reardon et al., 2019b).

Additionally, as previously mentioned, the analyses do not control for dif- ferences in free lunch rates in the average White student’s school versus the average Black or Hispanic student’s school in a district because this measure is correlated with the racial/ethnic segregation measures at 0.8 or higher. Therefore, despite rigorous controls for district-level socioeconomic factors and socioeconomic disparities by race/ethnicity, this study does not disentan- gle racial/ethnic segregation from socioeconomic segregation by race/ethnic- ity. There is strong evidence linking racial/ethnic differences in school poverty to the association between racial/ethnic segregation and test score gaps (Reardon et al., 2019b). Thus, it may be important for future research to examine these associations with socioeconomic test score gaps to provide a more comprehensive understanding of and possible solutions for associa- tions between school choice and structural education inequity.

Furthermore, in order to measure test score gaps, districts had to meet the criteria of having at least 20 students in at least one grade in at least one school year. So, extremely segregated districts (e.g., districts that are almost entirely composed of students of a single race/ethnicity) are not included in this study, and these districts are arguably the largest contributors to segregation and least amenable to policy solutions for integration (Holme & Finnigan, 2013). Additionally, when considering racial/ethnic segregation, it is impor- tant to highlight that district segregation is just one level at which segregation can inhibit students’ academic outcomes. For example, even within integrated schools there is capacity for within school segregation through mechanisms, such as inequitable gifted and talented programs, that may undermine efforts of inter and intradistrict integration (Roda, 2015).

Finally, this study exclusively considers brick and mortar magnet and charter schools and does not generalize to all school choice. Magnets and charters only comprise a piece, albeit a large piece, of the overall school choice landscape. Future analyses should also investigate cyber charters, open enrollment, vouchers, and private school choice to examine a more complete story about the relationship between school choice and education inequity. Moreover, as previously mentioned, this study finds null direct asso- ciations between charter enrollment and White-Hispanic test score gaps (except through segregation), and the associations with magnet enrollment and White-Hispanic gaps are small and do not hold across model specifica- tions. Thus, more work is needed specifically on how school choice is affect- ing Hispanic/Latinx populations.

Blatt, Votruba-Drzal

1218

Overall, given the limitations in this study, future research should explore more comprehensive operationalizations of education inequity, segregation, and school choice, as well as other potential mediators related to more prox- imal measures of children’s schooling experiences such as instructional qual- ity and disciplinary practices. It is also important to examine how these associations influence longer-term outcomes such as high school and college completion. Despite limitations, by analyzing data from SEDA, a publicly available national district-level data set, this study provides novel information on macro-level associations between school choice enrollment, racial/ethnic segregation, and education inequity across thousands of school districts in the United States. While individual districts and states have unique school choice policies with distinct impacts on education inequity, this study highlights that there are robust patterns nationally at the district-level. This study finds no evi- dence to suggest school choice enrollment reduces district-level test score gaps in third through eighth grade. In particular, this study finds greater char- ter enrollment is associated with larger White-Black test score gaps, and this relationship is mediated by White-Black segregation. These macro-level find- ings are especially important to consider when making federal policy deci- sions and determining federal incentives and guidelines for state- and district-level policies regarding school choice expansion.

Note

The data in this article come from SEDAVersion 2.1, which was the most current at the time of data analysis. The most current SEDA data can be accessed at https://edopportunity .org/get-the-data/. The authors would like to thank Drs. Sean F. Reardon, Andrew D. Ho., Benjamin R. Shear, Erin M. Fahle, Demetra Kalogrides, and Richard DiSalvo for creating the Stanford Education Data Archive (SEDA) Version 2.1 and making it publicly accessible. We would also like to thank Magnet Schools of America (MSA) for providing information on the names of magnet schools in states with missing magnet school data in the Common Core of Data (CCD). Additionally, we would like to thank Leyana Johnson for her assistance with coding the MSA and CCD data. Finally, this study was strengthened by thoughtful feedback from Dr. Heather Bachman, Dr. Francis Alvin Pearman, and anonymous reviewers for AERJ.

References

Bankston, C. L., III, & Caldas, S. J. (1996). Majority African American schools and social injustice: The influence of de facto segregation on academic achievement. Social Forces, 75(2), 535–555. https://doi.org/10.1093/sf/75.2.535

Berends, M., & Peñaloza, R. (2010). Increasing racial isolation and test score gaps in mathematics: A 30-year perspective. Teachers College Record, 112(4), 978–1007.

Berlak, H. (2001). Race and the achievement gap In H. S. Shapiro & D. E. Purpel (Eds.), Critical social issues in American education: Democracy and meaning in a glob- alizing world (pp. 227–235). Routledge.

Bifulco, R., & Ladd, H. F. (2007). School choice, racial segregation and test-score gaps: Evidence from North Carolina’s charter school program. Journal of Policy Analysis and Management, 26(1), 31–56. https://doi.org/10.1002/pam.20226

School Choice and Racial/Ethnic Test Score Gaps

1219

Bifulco, R., Ladd, H. F., & Ross, S. L. (2009). The effects of public school choice on those left behind: Evidence from Durham, North Carolina. Peabody Journal of Education, 84(2), 130–149. https://doi.org/10.1080/01619560902810104

Billings, S. B., Chyn, E., & Haggag, K. (2020). The long-run effects of school racial diver- sity on political identity (No. w27302). National Bureau of Economic Research. https://doi.org/10.3386/w27302

Boser, U. (2012). Race to the top: What have we learned from the states so far? A state- by-state evaluation of race to the top performance. Center for American Progress. https://www.americanprogress.org/issues/education-k-12/reports/2012/03/26/ 11220/race-to-the-top-what-have-we-learned-from-the-states-so-far/

Center for Education Reform. (2018). Laws and legislations. https://edreform.com/ issues/choice-charter-schools/laws-legislation/

Clark, M. A., Gleason, P. M., Tuttle, C. C., & Silverberg, M. K. (2015). Do charter schools improve student achievement? Educational Evaluation and Policy Analysis, 37(4), 419–436. https://doi.org/10.3102/0162373714558292

Cordes, S. A. (2018). In pursuit of the common good: The spillover effects of charter schools on public school students in New York City. Education Finance and Policy, 13(4), 484–512. https://doi.org/10.1162/edfp_a_00240

Coughlan, R. W. (2018). Divergent trends in neighborhood and school segregation in the age of school choice. Peabody Journal of Education, 93(4), 349–366. https:// doi.org/10.1080/0161956X.2018.1488385

Dee, T. S., & Fu, H. (2004). Do charter schools skim students or drain resources? Economics of Education Review, 23(3), 259–271. https://doi.org/10.1016/ j.econedurev.2003.10.001

Dobbie, W., & Fryer, R. G., Jr. (2011). Are high-quality schools enough to increase achievement among the poor? Evidence from the Harlem children’s zone. American Economic Journal, 3(3), 158–187. https://doi.org/10.1257/app.3.3.158

Donnor, J. K., & Dixson, A. (Eds.). (2013). The resegregation of schools: Education and race in the twenty-first century. Routledge. https://doi.org/10.4324/9780203522226

Ertas, N., & Roch, C. H. (2014). Charter schools, equity, and student enrollments: The role of for profit educational management organizations. Education and Urban Society, 46(5), 548–579. https://doi.org/10.1177/0013124512458118

Fiel, J. E. (2013). Decomposing school resegregation: Social closure, racial imbalance, and racial isolation. American Sociological Review, 78(5), 828–848. https:// doi.org/10.1177/0003122413496252

Frankenberg, E., Jongyeon, E., Ayscue, J. B., & Orfield, G. (2019). Harming our com- mon future: America’s segregated schools 65 years after Brown. Civil Rights Project/Proyecto Derechos Civiles. https://www.civilrightsproject.ucla.edu/ research/k-12-education/integration-and-diversity/harming-our-common- future-americas-segregated-schools-65-years-after-brown https://doi.org/10.11 77/0003122413496252

Frankenberg, E., & Lee, C. (2003). Charter schools and race: A lost opportunity for inte- grated education. Education Policy Analysis Archives, 11(32). https://doi.org/ 10.14507/epaa.v11n32.2003

Frankenberg, E., Siegel-Hawley, G., & Wang, J. (2010). Choice without equity: Charter school segregation and the need for civil rights standards. Civil Rights Project/ Proyecto Derechos Civiles. https://civilrightsproject.ucla.edu/research/k-12-edu cation/integration-and-diversity/choice-without-equity-2009-report https://doi .org/10.14507/epaa.v19n1.2011

Garcia, D. R. (2008). The impact of school choice on racial segregation in charter schools. Educational Policy, 22(6), 805–829. https://doi.org/10.1177/08959048 07310043

Blatt, Votruba-Drzal

1220

Gleason, P., Clark, M., Tuttle, C. C., & Dwoyer, E. (2010). The evaluation of charter school impacts: Final report. National Center for Education Evaluation and Regional Assistance. https://ies.ed.gov/ncee/pubs/20104029/

Goldring, E., & Smrekar, C. (2002). Magnet schools: Reform and race in urban educa- tion. Clearing House, 76(1), 13–15. https://doi.org/10.1080/00098650209604939

Goldring, E. B., & Swain, W. A. (2019). Perspectives on magnet schools. In M. Berends, A. Primus, & M. G. Springer (Eds.), Handbook of research on school choice (pp. 241–258). Routledge. https://doi.org/10.4324/9781351210447-18

Gordon, W. M. (1994). The implementation of desegregation plans since Brown. Journal of Negro Education, 63(3), 310–322. https://doi.org/10.2307/2967183

Grooms, A. A., & Williams, S. M. (2015). The reversed role of magnets in St. Louis: Implications for Black student outcomes. Urban Education, 50(4), 454–473. https://doi.org/10.1177/0042085913516131

Hanushek, E., Kain, J., & Rivkin, S. (2009). New evidence about Brown v. Board of Education: The complex effects of school racial composition on achievement. Journal of Labor Economics, 27(3), 349–383. https://doi.org/10.1086/600386

Harris, J. C. (2018). Integrating urban schools in a modern context: Roadblocks and challenges with the use of magnet schools. Urban Education. Advance online publication. https://doi.org/10.1177/0042085918802616

Harris, J. C. (2019). Changing context: Do magnet schools improve student achieve- ment in a modern setting? Journal of School Choice, 13(3), 305–334. https:// doi.org/10.1080/15582159.2019.1594605

Holme, J. J., & Finnigan, K. S. (2013). School diversity, school district fragmentation and metropolitan policy. Teachers College Record, 115(11), 1–29. http:// gs4a.org/wp-content/uploads/2015/05/Finnigan-School-Diversity-School- District-Fragmentation-and-Metropolitan-Policy.pdf

Imberman, S. A. (2011). The effect of charter schools on achievement and behavior of public school students. Journal of Public Economics, 95(7–8), 850–863. https:// doi.org/10.1016/j.jpubeco.2011.02.003

Johnson, R. C. (2019). Children of the dream: Why school integration works. Hachette. Kimelberg, S. M., & Billingham, C. M. (2013). Attitudes toward diversity and the school

choice process: Middle-class parents in a segregated urban public school district. Urban Education, 48(2), 198–231. https://doi.org/10.1177/0042085912449629

Kotok, S., Frankenberg, E., Schafft, K. A., Mann, B. A., & Fuller, E. J. (2017). School choice, racial segregation, and poverty concentration: Evidence from Pennsylvania charter school transfers. Educational Policy, 31(4), 415–447. https://doi.org/10.1177/0895904815604112

Lee, J. (2002). Racial and ethnic achievement gap trends: Reversing the progress toward equity? Educational Researcher, 31(1), 3–12. https://doi.org/10.3102/ 0013189X031001003

Linn, R. L., & Welner, K. G. (2007). Race-conscious policies for assigning students to schools: Social science research and the Supreme Court cases. National Academy of Education. https://nepc.colorado.edu/sites/default/files/Brief-NAE.pdf

Mader, N., Hemphill, C., & Abbas, Q. (2018). The paradox of choice: How school choice divides New York City elementary schools. New School Center for New York City Affairs. http://www.centernyc.org/the-paradox-of-choice

Magnet Schools of America. (2018). A snapshot of magnet schools in America. https:// magnet.edu/getinvolved/research-studies/snapshot-of-magnet-schools-report

Mickelson, R. A. (2001). Subverting Swann: First- and second-generation segregation in the Charlotte-Mecklenburg schools. American Educational Research Journal, 38(2), 215–252. https://doi.org/10.3102/00028312038002215

School Choice and Racial/Ethnic Test Score Gaps

1221

Mickelson, R. A., Bottia, M., & Southworth, S. (2013). School choice and segregation by race, ethnicity, class, and achievement. In K. G. Welner, P. H. Hinchey, W. I. Mathis, & G. Miron (Eds.), Exploring the school choice universe: Evidence and recommendations (pp. 167–192). National Education Policy Center.

Mickelson, R. A., Giersch, J., Nelson, A. H., & Bottia, M. C. (2018). Do charter schools undermine efforts to create racially and socioeconomically diverse public schools? In I. C. Rotberg & J. L. Glazer (Eds.), Choosing charters: Better schools or more segregation? (pp. 116–132). Teachers College Press.

Miller, P., Henry, D., & Votruba-Drzal, E. (2016). Strengthening causal inference in developmental research. Child Development Perspectives, 10(4), 275–280. https://doi.org/10.1111/cdep.12202

Monarrez, T. (2018). Segregated neighborhoods, segregated schools? Urban Institute. https://www.urban.org/features/segregated-neighborhoods-segregated-schools

Monarrez, T., Kisida, B., & Chingos, M. (2019). Charter school effects on school segre- gation. Urban Institute. https://www.urban.org/research/publication/charter- school-effects-school-segregation

National Alliance for Public Charter Schools. (2012). A growing movement: America’s largest charter school communities (7th annual ed.). Author.

National Center for Education Statistics. (2018). Table 216.20. Number and enrollment of public elementary and secondary schools, by school level, type, and charter and magnet status: Selected years, 1990-91 through 2016-17. https://nces.ed.gov/ programs/digest/d18/tables/dt18_216.20.asp

The National Coalition on School Diversity. (2020a). Policy brief 10: Including racial and socioeconomic diversity in ESSA district plans. https://school-diversity.org/ pb10/

The National Coalition on School Diversity. (2020b). Policy brief 11: Model state school integration policies. https://school-diversity.org/pb11/

Ni, Y. (2012). The sorting effect of charter schools on student composition in traditional public schools. Educational Policy, 26(2), 215–242. https://doi.org/10.1177/ 0895904810386598

Nieto, S. (2005). Public education in the twentieth century and beyond: High hopes, broken promises, and an uncertain future. Harvard Educational Review, 75(1), 43–64. https://doi.org/10.17763/haer.75.1.31r4357517211566

Pearman, F. A., & Swain, W. A. (2017). School choice, gentrification, and the variable significance of racial stratification in urban neighborhoods. Sociology of Education, 90(3), 213–235. https://doi.org/10.1177/0038040717710494

Polikoff, M., & Hardaway, T. (2017). Don’t forget magnet schools when thinking about school choice. Brookings Institution. https://www.brookings.edu/researc h/dont- forget-magnet-schools-when-thinking-about-school-choice/

Preacher, K. J., & Leonardelli, G. J. (2010). Calculation for the Sobel test: An interactive calculation tool for mediation tests. http://quantpsy.org/sobel/sobel.htm

Rapa, L. J., Katsiyannis, A., & Ennis, R. P. (2018). Charter school enrollment effects: A review of results from recent large-scale studies. Journal of Child and Family Studies, 27(10), 3132–3140. https://doi.org/10.1007/s10826-018-1143-z

Reardon, S. F., & Galindo, C. (2009). The Hispanic-White achievement gap in math and reading in the elementary grades. American Educational Research Journal, 46(3), 853–891. https://doi.org/10.3102/0002831209333184

Reardon, S. F., Grewal, E., Kalogrides, D., & Greenberg, E. (2012). Brown fades: The end of court ordered school desegregation and the resegregation of American public schools. Journal of Policy Analysis and Management, 31(4), 876–904. https://doi.org/10.1002/pam.21649

Blatt, Votruba-Drzal

1222

Reardon, S. F., Ho, A. D., Shear, B. R., Fahle, E. M., Kalogrides, D., & DiSalvo, R. (2018). Stanford Education Data Archive (Version 2.1) [Data set]. http://purl.stanford.edu/ db586ns4974

Reardon, S. F., Kalogrides, D., & Shores, K. (2019a). The geography of racial/ethnic test score gaps. American Journal of Sociology, 124(4), 1164–1221. https://doi.org/ 10.1086/700678

Reardon, S. F., Weathers, E. S., Fahle, E. M., Jang, H., & Kalogrides, D. (2019b). Is sep- arate still unequal? New evidence on school segregation and racial academic achievement gaps (CEPA Working Paper No. 19-06). Stanford Center for Education Policy Analysis. https://cepa.stanford.edu/sites/default/files/wp19- 06-v092019.pdf

Renzulli, L. A., & Evans, L. (2005). School choice, charter schools, and White flight. Social Problems, 52(3), 398–418. https://doi.org/10.1525/sp.2005.52.3.398

Riel, V., Parcel, T. L., Mickelson, R. A., & Smith, S. S. (2018). Do magnet and charter schools exacerbate or ameliorate inequality? Sociology Compass, 12(9), e12617. https://doi.org/10.1111/soc4.12617

Roda, A. (2015). Inequality in gifted and talented programs: Parental choices about sta- tus, school opportunity, and second-generation segregation. Palgrave Macmillan.

Rothstein, R. (2015). The racial achievement gap, segregated schools, and segregated neighborhoods: A constitutional insult. Race and Social Problems, 7(1), 21–30. https://doi.org/10.1007/s12552-014-9134-1

Rothwell, J. (2012). Housing costs, zoning, and access to high-scoring schools. Brookings Institution. https://www.brookings.edu/research/housing-costs-zon ing-and-access-to-high-scoring-schools/

Royston, P. (2004). Multiple imputation of missing values. Stata Journal, 4(3), 227–241. https://doi.org/10.1177/1536867X0400400301

Royston, P. (2005). Multiple imputation of missing values: Update of ice. Stata Journal, 5(4), 527–536. https://doi.org/10.1177/1536867X0500500404

Scott, J., Siegel-Hawley, G., DeBray, E., Frankenberg, E., & McDermott, K. (2020). An agenda for restoring civil rights in K-12 federal education policy. National Education Policy Center. https://nepc.colorado.edu/publication/restoring-civil- rights

Scott, T. M., Gage, N., Hirn, R., & Han, H. (2019). Teacher and student race as a predic- tor for negative feedback during instruction. School Psychology, 34(1), 22–31. https://doi.org/10.1177/1536867X0500500404

Siegel-Hawley, G. (2014). Race, choice and Richmond public schools: New possibili- ties and ongoing challenges for diversity in urban districts. Urban Review, 46(4), 507–534. https://doi.org/10.1007/s11256-014-0277-6

Skiba, R. J. (2015). Interventions to address racial/ethnic disparities in school disci- pline: Can systems reform be race-neutral? In R. Bangs & L. E. Davis (Eds.), Race and social problems: Restructuring inequality (pp. 107–124). Springer Science 1 Business Media. https://doi.org/10.1007/s11256-014-0277-6

Theil, H. (1972). Statistical decomposition analysis (Vol. 14). North-Holland. U.S. Department of Education. (2017). Common Core of Data: America’s Public

Schools [Data set]. https://nces.ed.gov/ccd/ccddata.asp Vasquez Heilig, J., Brewer, T. J., & Williams, Y. (2019). Choice without inclusion?

Comparing the intensity of racial segregation in charters and public schools at the local, state and national levels. Education Sciences, 9(3), 205. https:// doi.org/10.3390/educsci9030205

Von Hippel, P. T., & Powers, D. A. (2015). RPME: Stata module to compute Robust Pareto midpoint estimator. Statistical Software Components S457962. https:// ideas.repec.org/c/boc/bocode/s457962.html

School Choice and Racial/Ethnic Test Score Gaps

1223

Wang, J., Schweig, J. D., & Herman, J. L. (2017). Is there a magnet-school effect? A mul- tisite study of MSAP-funded magnet schools. Journal of Education for Students Placed at Risk, 22(2), 77–99. https://doi.org/10.1080/10824669.2017.1292853

Whitehurst, G. J., Reeves, R., & Rodrigue, E. (2016). Segregation, race, and charter schools: What do we know? Brookings Institution. https://www.brookings.edu/ research/segregation-race-and-charter-schools-what-do-we-know/.

Manuscript received September 26, 2019 Final revision received January 5, 2021

Accepted January 29, 2021

Blatt, Votruba-Drzal

1224