Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
English language development materials in Texas: a study of effectiveness and selection
(USC Thesis Other)
English language development materials in Texas: a study of effectiveness and selection
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
1
ENGLISH LANGUAGE DEVELOPMENT MATERIALS IN TEXAS:
A STUDY OF EFFECTIVENESS AND SELECTION
by
Q. Tien Le
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(URBAN EDUCATION POLICY)
December 2017
ii
Dedication
If I have seen further,
it is by standing on the shoulder of giants,
most notably,
my mom and dad.
iii
Acknowledgements
This dissertation would not have been possible without the advice and support of many people.
First and foremost, I would like to thank my mom and dad for a beautiful childhood filled with
books and unfettered dreams. They have never doubted my ability to follow through on my
goals, and, more importantly, they have taught me the importance of humility.
I am grateful to my advisor, Dr. Morgan Polikoff, for honest and timely feedback and for
providing me with the skills needed to complete my dissertation.
I would like to thank Dr. Julie Marsh and Dr. Gary Painter for their insightful feedback on my
dissertation paper.
I am grateful to the district leaders in Texas who openly shared their knowledge with me.
I would like to thank Thuy Huynh and Phong Le for giving me the courage to teach.
I am grateful to Ms. Nori Hepler, my beloved high school chemistry teacher, for being a shining
example of the kind of teacher all students deserve.
I am indebted to Jennifer Yvonne Perry and Yvonne Monteverde for being my moral compass
and for making sure I made it to the finish line.
I am grateful to Eriq Felix and Román Liera for allowing me to join their reading group and for
helping me to become a more critical scholar.
I would like to thank Dr. Alan Green for helping me navigate life in academia and for being
generous with his time.
I am grateful to my students for helping me become a more compassionate human being.
Last, but certainly not least, I am indebted to all those who risk their livelihood so that we may
live in a more just society.
iv
Table of Contents
Dedication ....................................................................................................................................... ii
Acknowledgements ........................................................................................................................ iii
List of Tables .................................................................................................................................. v
List of Figures ............................................................................................................................... vii
Abstract ........................................................................................................................................ viii
Chapter 1: Introduction ................................................................................................................... 1
Chapter 2: English Language Development Curriculum Effects on English Language Proficiency
in Texas ........................................................................................................................................... 4
Chapter 3: Variation in English Language Development Materials—How Social Constructions
Shape the Adoption of ELD Materials.......................................................................................... 43
References ................................................................................................................................... 126
Appendix ..................................................................................................................................... 145
v
List of Tables
Table 1. 1. Number of schools that adopted each curriculum in Texas for grade 3 (N=1,233) ..... 9
Table 1. 2. Construction of the Analytic Sample .......................................................................... 13
Table 1. 3 Covariate-by-Covariate Balancing Details for the Comparison Between No ELD Text
and State-Adopted ELD Text ................................................................................................ 25
Table 1. 4. Covariate-by-Covariate Balancing Details for the Comparison Between Rigby On
Our Way to English and National Geographic Reach .......................................................... 26
Table 1. 5 Average treatment effects of non-adoption relative to state-adopted ELD instructional
materials on TELPAS spring 2015 composite scores ........................................................... 33
Table 1. 6 Average treatment effects of Rigby On Our Way to English relative to National
Geographic Reach on TELPAS spring 2015 composite scores ............................................ 35
Table 1. 7 Falsification Tests: Effects of Non-Adoption versus State-Adoption on STAAR 2015
Math Achievement ................................................................................................................ 37
Table 1. 8 Falsification Tests: Effects of Non-Adoption versus State-Adoption on 2011-2012
3rd-grade TELPAS Composite Scores ................................................................................. 37
Table 2. 1. Types of Programs for EBs ......................................................................................... 51
Table 2. 2. ELD Adoption Patterns in Texas for Kindergarten (N=1,133) .................................. 70
Table 2. 3. Sampling Plan for Interview with District Leaders in Texas (N=28) ......................... 71
Table 2. 4. List of Codes from First Round of Coding ................................................................. 78
Table 2. 5. List of Codes from Second Round of Coding ............................................................ 80
Table 2. 6. Descriptive Characteristics of Non-Adopters, State-Approved Adopters, Off-List
Adopters. ............................................................................................................................... 84
vi
Table 2. 7. Descriptive Statistics of District Characteristics by Publisher ................................... 85
Table 2. 8. Results from Multinomial Logistic Regression of ELD Adoption Patterns on District
Characteristics (Relative Risk Ratios) .................................................................................. 91
Table 2. 9. Social Constructions of EBs and English Language Development Adoption Patterns
............................................................................................................................................... 97
vii
List of Figures
Figure 1. 1. Distribution of Outcome Variable ............................................................................. 16
Figure 2. 1. The Relationship Between Social Constructions and Academic Achievement ........ 46
Figure 2. 2. Social Construction of Target Populations (Schneider & Ingram, 2008) ................. 62
Figure 2. 3. Schneider and Ingram’s Model: Policy Design Predictions (1993) .......................... 65
Figure 2. 4. Distribution of Districts By Social Construction .................................................... 100
viii
Abstract
Federal court cases such as Castañeda v. Pickard (1981) have attempted to standardize
program quality for emergent bilinguals (EBs) by mandating that programs for emergent
bilinguals be effective. However, little is known about how federal mandates for EBs are enacted
at the local level and the extent to which ELD instructional programs are effective. This
dissertation paper focuses on English language development (ELD) curricular materials because
they are one important aspect of educational programming for EBs. To answer questions about
the effectiveness of ELD materials and ELD adoption patterns, I use two studies.
In the first study, I use local linear matching to assess the effectiveness of ELD
instructional materials. I find that schools that do not purchase any ELD curricula have
significantly lower English language proficiency scores relative to schools that purchase state-
adopted ELD materials. These results are robust across various matching models—inverse
probability weights with regression adjustment, kernel matching, and nearest neighbor matching.
There is no significant difference between schools that adopt the two most popular ELD
curricula—Rigby On Our Way to English and National Geographic Reach. This study suggests
that EBs who attend schools that have instructional materials that explicitly foreground English
language proficiency standards outperform schools that do not have such materials.
To understand how federal mandates are implemented locally, I use a mixed-methods
sequential explanatory design to investigate the adoption of ELD instructional materials in
Texas. First, I use a multinomial logistic regression to determine whether or not EBs have
equitable access to ELD materials. Second, I interview district leaders in Texas to understand the
extent to which local influence explains the variation in ELD materials adoption. I find that
ix
certain district characteristics are predictive of ELD adoption patterns. In addition, I find that the
social construction of EBs at the local level can explain some of the variance in adoption patterns
1
Chapter 1: Introduction
Emergent bilinguals represent a rapidly growing and heterogeneous group of students.
About 77% of emergent bilinguals in American public schools speak Spanish, with the next four
most common home languages being Arabic, Chinese, Vietnamese, and Hmong (NCES, 2016).
Five states—California, Texas, Florida, New York, and Illinois—account for 62% of the EBs
enrolled in K-12 public schools in the United States (U.S. Department of Education, 2013). EBs
are more likely to be located in urban areas, comprising 14% of public school enrollment in
cities compared to 4% in rural areas (NCES, 2016). Since 1998, the emergent bilingual (EB)
population in American public schools has grown by 51% (Education Week 2011) and now
accounts for 9.3% of all public school students in the United States (National Center for
Education Statistics (NCES), 2016).
Despite their growing numbers, EBs continue to be underserved. Not only do EBs have
the dual task of learning the dominant language and learning content, EBs are also more likely to
be “triply segregated”—more likely to attend schools with other EBs, minorities, and low-
income students (Orfield, 2001; Orfield & Lee, 2006; Ríos-Aguilar & Gándara, 2012). Gándara
and Orfield (2010) find that linguistic segregation intensifies all the negative impacts of racial
and socioeconomic segregation. In addition, the majority of teachers who teach EBs have had
little or no professional development for teaching EBs (NCES, 2002), and few teachers have
taken a course focused on issues related to EBs (Menken & Antunez, 2001).
To improve the educational outcomes of EBs, federal court cases have required that
programs for EBs be effective. In 1974, the Supreme Court mandated in Lau v. Nichols that
schools must teach language minority students English and provide them with access to content-
area curriculum. Subsequently, Castañeda v. Pickard (1981) further elevated the status of EBs
2
on the educational policy agenda by requiring that educational programs for EBs be based on
sound educational theory, be well-implemented, and produce positive results. One part of a
quality educational program for EBs is English language development (ELD), or language
instruction that is delivered separately from content (Saunders, Goldenberg, & Marcelletti,
2013).
In this dissertation, I investigate the effectiveness of ELD instructional materials, the
distribution of such materials, and how district leaders in Texas go about choosing ELD
materials. Specifically, the two essays in this dissertation address the following research
questions.
1) Non-Adopters vs. State-Adopters: To what extent do English proficiency outcomes differ
amongst schools that do not purchase any ELD curricular materials versus schools that
purchase state-adopted curricular materials?
2) Rigby vs. National Geographic: To what extent do English proficiency outcomes differ
amongst schools that use the two most popular ELD curricular materials (Rigby On Our
Way to English and National Geographic Reach)?
3) To what extent, if any, do ELD adoption patterns vary by district characteristics?
4) To what extent can the social constructions of EBs explain the variance in ELD materials
adoption?
In Chapter 2, I use local linear matching at the school level to estimate the effect of state-
adopted ELD curricula on English language proficiency scores. In addition, I estimate the effect
of the two most popular ELD curricula in Texas--Rigby On Our Way to English and National
Geographic Reach. The findings indicate that schools who purchase state-adopted ELD curricula
3
outperform non-adopters by 0.29 standard deviations. There was no significant difference in
English proficiency scores between Rigby-adopters and adopters of National Geographic Reach.
In Chapter 3, I explore the variation in ELD adoption patterns across Texas. I find that
certain district characteristics are predictive of ELD adoption patterns. Specifically, districts with
higher enrollments, shares of EBs, and per pupil expenditure are more likely to adopt state-
approved ELD materials. The variation in ELD adoption patterns can be partially explained by
how districts socially construct EBs. For example, districts in which EBs were positively
construed and had high numbers had at least one ELD curricula, and, in many cases, had access
to multiple ELD curricula. For districts in which EBs are perceived to be undeserving, resources
were undersubscribed or symbolic.
Together, these findings highlight the importance of systematic ELD practices. Districts
that adopted comprehensive ELD curricula outperformed those districts that did not have a
systematic ELD program. In addition, the findings underline the need for local education
agencies to clearly define the differences between ELD, language arts, and Spanish language
development. Interview data reveal that there is much confusion between the three subject areas,
which may impede the implementation of effective programs for EBs.
4
Chapter 2: English Language Development Curriculum Effects on English Language
Proficiency in Texas
Emergent bilinguals are one of the fastest growing and one of the most underserved
student populations in the United States (Capps et al., 2005; Genesee, Lindholm-Leary,
Saunders, & Christian, 2005; National Education Association, 2008). Since 1998, the emergent
bilingual (EB) population in American public schools has grown by 51% (Education Week,
2011) and now accounts for 9.3% of all public school students in the United States (National
Center for Education Statistics, 2016). Of all emergent bilinguals in the United States, 73% are
Spanish-speaking, and 59% qualify for free or reduced lunch (National Clearinghouse for
English Language Acquisition, 2008). Emergent bilinguals are “triply segregated”—more likely
to attend schools with other EBs, minorities, and low-income students (Orfield, 2001; Orfield &
Lee, 2006; Ríos-Aguilar & Gándara, 2012). On both the math and reading sections of the
National Assessment of Education Progress, a full standard deviation separates emergent
bilinguals and their counterparts (National Center for Education Statistics, 2011).
Following Burke, Morita-Mullaney, and Singh (2016), I use the term emergent bilingual
in lieu of the following: English language learner (ELL), English as a second language (ESL),
and limited English proficient (LEP). Scholars have objected to the terms “limited English
proficient” and “English language learner” (Garcia, 2009) due to its deficit-framing of the
linguistic abilities of emergent bilinguals. The terms also suggest that English proficiency is the
only desired outcome for this population of students (Menken & Solorza, 2014). Referring to this
group of students by an additive name that does not focus on its limitations is an important step
towards the decentering of English monolingual norms. Similar to Burke et al. (2016), I use the
term emergent bilingual with two exceptions. First, the terms ELL, ESL, and LEP may be used
5
when referencing federal and state laws. Second, the terms ELL, ESL, and LEP may be used in
my literature review if the authors of the studies use the terms.
To improve the educational outcomes of emergent bilinguals, federal court cases have
required that programs for emergent bilinguals be effective. Castañeda v. Pickard (1981)
mandated that educational programs for emergent bilinguals be based on sound educational
theory, implemented effectively, and tested for program efficacy. Since research has shown that
textbooks influence the content and methods that teachers use in the classroom (Alajmi, 2009,
2012; Robitaille & Travers, 1992), English language development (ELD) textbooks can play an
integral role in program efficacy. This paper addresses the third mandate in the Castañeda v.
Pickard (1981) ruling by measuring the relationship between ELD curriculum materials and
emergent bilinguals’ English proficiency. Though Texas’ educational policies refer to English
language development as English as a second language, I use the term ELD in this study to refer
to language programs provided to emergent bilinguals to attain English proficiency. ELD is a
more accurate term because English may not necessarily be the second language of emergent
bilinguals.
The association between textbook quality and student achievement has important
implications for local education agencies. Since the costs of textbooks across publishers are quite
similar, districts have the potential to increase student achievement at minimal marginal cost
(Chingos & Whitehurst, 2012). Furthermore, efforts to improve the achievement of emergent
bilinguals by making better curriculum choices may face fewer challenges compared to other
alternatives (e.g., extending the school day, increasing the number of bilingually-certified
teachers), which may be less feasible or politically palatable. Though a few studies have looked
at the relationship between mathematics textbooks and student achievement (Agodini et al.,
6
2010; Bhatt & Koedel, 2012; Bhatt, Koedel, & Lehmann, 2013; Koedel et al., 2016), no studies
have examined the effect of ELD textbooks on the English proficiency of emergent bilinguals at
the state level. Past studies on the English language development of EBs have focused heavily on
program type (bilingual programs versus English-only programs). My study seeks to fill this gap.
This study answers the following research questions:
1.) Non-Adopters vs. State-Adopters: To what extent do English proficiency outcomes differ
amongst schools that do not purchase any ELD curricular materials versus schools that
purchase state-adopted curricular materials?
2.) Rigby vs. National Geographic: To what extent do English proficiency outcomes differ
amongst schools that use the two most popular ELD curricular materials (Rigby On Our
Way to English and National Geographic Reach)?
In the sections that follow, I review the pertinent research literature, describe the data,
outline the study design, and conclude.
Background Information
Research on Textbook Efficacy
The choice of curriculum materials can significantly affect student achievement. Four
recent rigorous, empirical studies examine the relationship between textbooks and student
academic outcomes. First, Agodini et al. (2010) use a randomized control trial across 12 districts
to compare the effects of four elementary mathematics on student achievement in grades 1 and 2.
The authors find that math curricula correspond to large differences in student academic
achievement. For example, by the end of 2
nd
grade, certain curricula were associated with an
increase of 0.17 standard deviations in student test scores. Second, Bhatt & Koedel (2012) use a
statewide, quasi-experimental evaluation that compares three elementary math curricula. The
7
authors also find large differences in the efficacy of curriculum materials. The largest difference
between two curricula was 0.14 standard deviations on a math achievement test. Third, Bhatt,
Koedel, and Lehmann (2013) use textbook data from the state of Florida to study curriculum
quality across various mathematics subtopics. The authors find evidence of variability in
curricular effectiveness across various subtopics within the same textbook. Fourth, Koedel et al.
(2016) use textbook data from California and find that the adoption of Houghton Mifflin’s
California Math is associated with a 0.05-0.08 standard deviation increase in math achievement.
In sum, there is empirical evidence to suggest that textbook efficacy can vary between textbooks
and within the same textbook. These findings suggest that school and district leaders can
improve student achievement at little to no cost by simply making more informed decisions
about curriculum materials.
Though a few studies have looked at the relationship between mathematics textbooks and
student achievement, no studies have examined the effect of ELD textbooks on emergent
bilinguals’ English language proficiency. This study would be the first to use quasi-experimental
methods to assess the effectiveness of ELD instructional materials at the state level.
Textbook Adoption at the State Level
In the United States, there are 19 state-level adoption states and 31 states whose textbook
adoption process occurs at the local level (Association of American Publishers, 2015). Texas is
one of the two largest adoption states. Thus, publishers tailor their books to the needs and
preferences of the Texas State Board of Education and concentrate their selling efforts in Texas
(Bowler, 1978; Stein, Stuen, Carnine, & Long, 2001). In Mike Bowler’s seminal piece on
textbook adoption (1978), he outlines the political nature of the textbook adoption process and
Texas’ national role in this process. Given Texas’ large share of the textbook market, publishers
8
time the production of new textbooks to coincide with the adoption cycle in Texas (Stein, Stuen,
Carnine, & Long, 2001). However, it should be noted that Texas’ national influence on textbook
adoption may have waned since the adoption of Common Core State Standards (CCSS). While
forty-two states and the District of Columbia have adopted the CCSS, Texas has not.
Textbook Adoption at the District Level
Once the Texas State Board of Education decides which instructional materials are
approved by the state, district leaders can decide which curricula they want to adopt for their
particular district. ELD textbook adoption at the district level is particularly complex compared
to, for example, mathematics textbook adoption, because districts are not mandated to purchase
ELD textbooks for their EBs. In addition, the Texas State Board of Education declared in 2011
that districts could use state funds to purchase instructional materials that are not on the State
Board of Education list (Texas Education Agency, 2011). Given the great flexibility that districts
have over ELD textbook adoption, it is important to understand how the extra autonomy has
impacted the English proficiency of EBs.
Prior research has shown that district leaders consult a variety of sources to make
textbook adoption decisions. Using data from over 150 interviews with curriculum decision-
makers in eight states, Zeringue et al. (2010) find that district leaders have limited information
on textbook efficacy and, therefore, tend to make adoption decisions based on other factors such
as teacher acceptance and advice from neighboring districts. In addition, Zeringue et al. (2010)
find that 80% of curriculum leaders seek out information about textbook use, effectiveness, and
acceptance of the materials. Thus, if more research was available regarding textbook efficacy,
curriculum leaders may use this information to make better choices.
ELD Textbooks in Texas
9
The 2011-2012 school year is the most recent ELD textbook adoption cycle in the state of
Texas. During this adoption cycle, the state released a list of six state-approved ELD curricula.
Of the six, one is an online program and five are traditional, textbook curricula. The textbooks
vary along several dimensions including number of lessons on forms of English, opportunities to
learn new vocabulary words, and the ratio of listening/speaking activities to reading/writing
activities. Of the five traditional curricula, Rigby On Our Way to English and National
Geographic Reach are the most popular, comprising 69% of adoption patterns in Texas’s
elementary schools (see Table 1.1). The other three traditional curricula are McGraw-Hill Texas
Treasure Chest, Santillana Spotlight on English, and Pearson Longman Cornerstone. Alloy
Interactive provides a web-based ELD curricula called ESL Reading Smart.
Table 1. 1. Number of schools that adopted each curriculum in Texas for grade 3 (N=1,233)
Number of Schools % of Schools
State-adopted Texts
National Geographic Reach 551 44%
Rigby On Our Way to English 310 25%
McGraw-Hill Texas Treasure Chest 83 7%
Pearson Longman Cornerstone 20 2%
Santillana Spotlight on English 0 0%
Alloy Interactive 0 0%
Non-State-Adopted Book 16 1%
No ELD textbook adoption 253 20%
With 551 schools across 48 districts adopting National Geographic Reach, it is the most
frequently adopted ELD curricula and represents 44% of adoption patterns. Dr. Deborah Short,
one of the book’s authors, is the creator of the Sheltered Instruction Observation Protocol
(SIOP), which is a research-based and validated classroom tool designed to improve instruction
for EBs. SIOP was developed through a 7-year research study (1996-2003) sponsored by the
National Center for Research on Education, Diversity & Excellence and funded by the U.S.
Department of Education. Short’s extensive work on SIOP makes her a leading expert on
10
developing language proficiency amongst EBs. National Geographic Reach was first published
in 2011 and the second edition is set to come in 2017. A student anthology in the National
Geographic Reach series is usually divided into eight units. Each unit includes multiple
opportunities to use language frames, share aloud with a partner, learn conversational and
academic vocabulary, and practice language-learning strategies (e.g., predicting, rereading and
retelling). The anthology includes a mix of fiction and non-fiction texts.
With 310 schools across 37 districts adopting Rigby On Our Way to English, it is the
second most popular ELD curricula in Texas. Rigby is a registered trademark of Houghton
Mifflin Harcourt. On Our Way to English is marketed as a comprehensive ELD program that
focuses on academic language, vocabulary development, content-based instruction, and oral
language. The text is typically divided into eight units with each unit providing opportunities to
engage with fiction and non-fiction texts. Dr. Robert Marzano, one of the authors of the series,
has written extensively about building academic vocabulary and building prior knowledge. On
Our Way to English was first published in 2011 and is available in a Texas-specific version and a
national version. In contrast, a Texas-specific version of National Geographic Reach does not
exist.
Data
The data for the proposed study come from the state of Texas.
1
Texas serves roughly
5,000,000 students who are spread across approximately 1,250 public school districts. The racial
composition of students in Texas’ public schools is as follows: 52% Hispanic, 30% White, 13%
Black, 4% Asian, and 2% multiracial (Texas Education Agency, 2014). Roughly 18% of students
1
This material is based upon work supported by the National Science Foundation under Grant No. 1445654 to the
University of Southern California. Any opinions, findings, and conclusions or recommendations expressed in this
material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.
11
in Texas’ public schools are identified as English language learners (Texas Education Agency,
2014).
Texas is an ideal state for the study of English language development curriculum
materials on emergent bilinguals’ English proficiency for several reasons. First, Texas is home to
830,000 EBs, the second highest number of EBs in the United States.
2
Second, the textbook
adoption process is centralized in Texas, which allows for a more manageable analysis.
Textbook adoption in Texas occurs at the district-level with the exception of open-enrollment
charter schools (Texas Education Agency, 2015). Third, Texas has roughly 1,200 public school
districts, which allows for high statistical power for student achievement analyses. Lastly, since
Texas grants great autonomy to districts over ELD textbook adoption, there is variation in the
kinds of ELD instructional materials that are adopted. Districts in Texas report using 7 different
state-approved ELD curricula and 21 different non-state-approved ELD curricula. In sum, there
is variation in the type of ELD instructional materials adopted by districts.
Textbook information was collected from districts using the Texas Education Agency
(TEA) online data portal. The TEA collects information, which is publicly available through
Freedom of Information Law (FOIL) request, on instructional materials allotment (IMAT). I was
able to obtain ELD textbook purchases for the years 2011-2015. The following textbook
information is available for each district: publisher, textbook title, subject, grades used, unit
price, quantity of units purchased, and adoption year. I find 418 districts on the IMAT list for
years 2011-2015, which leaves 810 districts with missing information on ELD textbook
adoption.
2
With 1.5 million EBs, California has the highest number of EBs in the United States (National Center for
Education Statistics, 2012).
12
Since the IMAT report only captures purchases made using state funds, it does not
capture instructional materials that were purchased using local funds nor instructional materials
that were purchased prior to 2011. Therefore, it cannot be assumed that districts that are not on
the 2011-2015 IMAT list did not purchase ELD instructional materials. The other 810 districts
might have purchased ELD textbooks prior to 201l. Alternatively, these 810 districts may not
have purchased any ELD instructional materials. To obtain information on ELD instructional
materials for the remaining 810 districts, I submitted Public Information Requests to each district
with missing ELD textbook information and received responses from 743 districts, yielding an
overall response rate of 94%.
Analytic Sample
I focus my analysis on elementary schools in Texas that serve emergent bilinguals. Since
the study focuses on elementary English language development books, I drop middle and high
schools from the sample as elementary ELD textbooks are intended for elementary-aged students
and are typically not purchased for older students. In addition, elementary schools that do not
have emergent bilinguals (n = 82) are excluded from the study since they do not have a need for
ELD curricular materials.
Table 1.2 provides details about the construction of the analytic sample starting with the
universe of schools in Texas that are eligible for the study (N=3,085). After five exclusion
criteria were applied
3
, 1,233 elementary schools (40%) remain in the sample. First, schools that
did not adopt the same ELD curriculum in grades K-3 were dropped from the sample. Roughly
three percent of schools were excluded because they indicated using more than one ELD
3
The exclusion criteria used in this study are similar to the ones used in Koedel et al.’s (2016) study on the
effectiveness of math textbooks in California.
13
curricula in grades K-3. Though it is possible to examine mixed-treatment effects, there are too
few schools for an effective analysis.
Table 1. 2. Construction of the Analytic Sample
Schools % of total Districts % of total
Initial Universe 3,085
651
Reason for exclusion
Non-uniform adopted, grades k-3 -95 3.1 -47 7.2
Missing outcome variable (TELPAS 2015) -1482 48.2 -410 63.0
Missing textbook information -21 0.7 -3 0.5
HISD or DISD -230 7.5 -2 0.3
Missing covariates 0 0 0 0
Off-list Adopter -16 0.5 -6 0.9
Final Analytic Sample 1,233 40.0 183 28.1
Second, elementary schools that did not report English language proficiency scores for
spring 2015 were eliminated from the study. There are two scenarios in which the TEA would
mask Texas English Language Proficiency Assessment System (TELPAS) scores: 1.) when few
students in a group are evaluated and it may be possible to identify particular students or 2.)
when all the students in a group have the same result (Texas Education Agency, 2015). Masking
refers to the use of special symbols to conceal data to protect student confidentiality. Due to
masking of student data, 48% of schools did not report TELPAS scores for spring 2015 and
were, thus, dropped from the study.
Third, less than one percent of schools were eliminated from the study due to missing
information regarding ELD adoption. Specifically, 21 schools or 0.7% of the sample were
dropped due to missing ELD textbook information. Though public information requests were
submitted to all districts in Texas with missing ELD textbook information, 3 districts out of the
14
remaining districts in the study did not respond. These 3 districts represent 21 schools. Given
these data restrictions, the results from this study are only generalizable to elementary schools in
Texas that adopted the same ELD curriculum in grades K-3, reported TELPAS scores in spring
2015, and reported ELD textbook adoption.
Fourth, Houston Independent School District (HISD) and Dallas Independent School
District (DISD) were dropped because both districts are much larger than all other districts in the
state which made it difficult to find comparable matches. After dropping HISD and DISD from
the sample, 1,233 schools and 191 districts remain. The final analytic sample retains 41% of the
universe of schools and 29% of the universe of districts.
Lastly, off-list adopters (N=16) were dropped from the study as they comprise a small
group of districts. Any statistical inferences made about this group will be unreliable. After
dropping off-list adopters 1,233 schools and 183 districts remain. The final analytic sample
retains 40% of the universe of schools and 28% of the universe of districts.
ELD Textbook Adoption Patterns
In the final analytic sample, I find that the majority of elementary schools in Texas that
service emergent bilinguals purchase state-adopted ELD materials (see Table 1). While roughly
1% of the final analytic sample purchased materials that were not on the state-adopted list, 79%
of schools purchased materials on the state-adopted list. Approximately 20% of schools in the
final sample did not purchase any ELD materials even though they did have students identified
as EB in their school. Of the ELD materials on the state-adopted list, National Geographic
Reach and Rigby On Our Way to English were the most popular with 69% of schools in the final
analytic sample purchasing from one of these two publishers.
School-Level Variables
15
Though ELD textbook adoption typically occurs at the district level, I match at the school
level for two reasons. First, matching is a “data hungry” process that requires a large sample size
(Heckman, Ichimura, & Todd, 1998). Second, schools within the same district can vary greatly
in their demographics, test scores, teacher experience, etc. Conducting my analysis at the school
level will allow me to adjust for both school- and district-level characteristics (Bhatt & Koedel,
2012; Bhatt & Koedel & Lehmann, 2013).
School-level data is drawn from the Texas Education Agency online data portal.
Information about students and teachers are available at the school level. With regards to
teachers, the following information is available: experience, gender, race, and credential type. In
terms of students, the following information is available at the school level: race, gender,
free/reduced price lunch eligibility, special education eligibility, English language learner status,
district average math and reading score on the State of Texas Assessments of Academic
Readiness (STAAR), and district average English language proficiency scores on the TELPAS
(Texas English Language Proficiency Assessment System). Since the last year of ELD textbook
adoption in Texas was the 2011-2012 school year, I will be using achievement and demographic
data from 2010-2011 to match schools.
Outcome Variables
School-average composite scores on the spring 2015 TELPAS assessment serve as the
outcome variable. The TELPAS assessment is delivered on a yearly basis to monitor EBs’
progress in attaining English proficiency. Since there are four language modalities—listening,
speaking, writing, and reading—the TELPAS exam is divided into four sections. Students are
given a rating of 1-4 on each subtest where 1 represents Beginning, 2 represents Intermediate, 3
is Advanced, and 4 represents Advanced High (TEA, 2015). In order to arrive at the composite
16
score, TEA weights each section as follows: listening (0.10), speaking (0.10), reading (0.50), and
writing (0.30). Using student-level TELPAS scores, I create school-average composite TELPAS
scores which are normally distributed (see Figure 1.1).
Figure 1. 1. Distribution of Outcome Variable
Since the STAAR reading assessment may appear to be very similar to the TELPAS
exam, I clarify the differences between the two assessments and justify using TELPAS scores as
the outcome variable. While the reading portion of the STAAR assesses students’ mastery of
content standards for English language arts, the TELPAS assessment focuses on language
standards—not content standards. The reading portion of the STAAR assessment only measures
reading abilities and does not include a listening and speaking portion. Since the TELPAS exam
covers all four language modalities (listening, speaking, reading, and writing), the TELPAS is
better positioned to assess English proficiency. Increased English proficiency may lead to greater
access to content standards; therefore, it is possible that increases in TELPAS scores lead to
increases in STAAR reading scores. However, since TELPAS scores are a more direct measure
of English proficiency than STAAR reading scores, I use TELPAS scores as my outcome
variable.
17
Research Design
To model the association between various ELD curriculum materials and EBs’ English
proficiency, I use local linear matching. Bhatt and Koedel (2012) used a similar matching
approach to evaluate relative curriculum effectiveness in the state of Indiana. Since districts can
self-select into the treatment (i.e. the ELD textbook), selection bias may affect our coefficient
estimates. Matching on school characteristics reduces group differences across all the observed
variables used in the propensity score equation (Shadish, Cook & Campbell, 2002). Matching
can allow for causal inferences if the conditional independence assumption is met (Heckman,
Ichmura, & Todd, 1997; Rosenbaum & Rubin, 1983).
The conditional independence assumption requires that potential outcomes be
independent of textbook choice conditional on the observables. In other words, the assumption
requires that all relevant variables be included in the model. The conditional independence
assumption will not be satisfied if unobserved characteristics influence both the treatment and
the outcome. By using an estimated propensity score, I can match schools based on their
observable characteristics to minimize the likelihood that unobserved characteristics may
influence the outcome. Matching has two main advantages over a simple regression analysis.
First, matching has fewer restrictions on functional form (Black & Smith, 2004). Second,
matching can more accurately predict the relationship between covariates and the outcome
variable by restricting the influence of noncomparable treatment and control units (Black &
Smith, 2004).
Of schools that adopt ELD materials in the final analytic sample, 92% adopted new ELD
curricula in 2011-2012. Thus, third graders in 2014-2015 will have potentially been exposed to
the ELD curricula for four years. If an ELD curriculum is effective, then districts using this
18
curriculum should have higher TELPAS scores for their EBs compared to districts that did not
adopt this particular curriculum.
Following Bhatt and Koedel (2012), I use the term “total treatment effects” to account for
the possibility of the estimates capturing systematic differences between treatment and control
schools. For example, if one curriculum has a teacher’s guide that is easier to navigate, the
estimates will reflect this difference. Based on the series of falsification tests I conducted for the
study, I find little evidence to suggest that there are systematic differences between treatment and
control schools. However, it should be noted that the estimates from this study will capture “total
treatment effects.”
Propensity Score Estimation
In order for propensity score matching to yield unbiased estimates of the effectiveness of
ELD textbooks, two types of covariates should be included: 1.) pretest measures of the outcome
variable and 2.) direct measures of the selection process (Cook, Shadish, & Wong, 2009; Cook &
Steiner, 2010; Glazerman, Levy & Myers, 2003; Steiner & Cook, 2013). The first type of
covariate is easily obtainable through TEA. I include school average TELPAS composite scores
from spring 2011
4
in the selection model. Since ELD adoption began in the 2011-2012 school
year, I use pre-adoption TELPAS scores as pretest measure of the outcome variable. The second
type of covariate is slightly harder to obtain. Steiner and Cook (2013) suggest investigating the
actual selection process before starting the study because the omission of important variables can
seriously increase bias in the estimates (Heckman, Ichimura, & Todd, 1997). Shadish and Steiner
(2010) recommend interviewing relevant individuals to identify factors that influence selection
into treatment and studying the pertinent research literature. In addition, economic theory and
4
The higher the correlation between pretest and posttest, the greater the bias reduction (Steiner & Cook, 2013). The
correlation between school-average TELPAS scores in 2011 and school-average TELPAS scores in 2015 is 0.43.
19
previous empirical findings should guide the choice of variables to include in the propensity
score estimation
5
(Caliendo & Sabine, 2008).
Heckman et al. (1997) find that selection models that include variables related to
treatment assignment are better able to predict treatment assignment than models that only
include demographic characteristics like education and race. The covariates that are used to
estimate the propensity score should include all of the key factors affecting both the outcome
variable and the treatment assignment (Smith & Todd, 2005). When creating a list of covariates
for the propensity score estimation, there is a tradeoff between bias and variance. Including an
irrelevant variable can lead to increased variance (Augursky & Schmidt, 2001), while excluding
a potentially important confounder can lead to increased bias (Heckman, Ichimura, & Todd,
1997; Stuart, 2010).
I investigate the selection mechanism using two methods. First, I conduct in-depth
interviews with 28 district leaders in Texas which I will describe in the following paragraphs.
Second, I review previous research on textbook effectiveness to identify potential covariates
(Rubin, 2001).
District leader interviews. To understand the factors influencing ELD textbook
adoption, I conducted 28 in-depth interviews with district leaders using structured open-ended
interviewing protocols (see appendix). I focus on district leaders because textbook adoption
typically occurs at the district-level in the state of Texas. The districts selected for interviews
represented variation along two dimensions: percentage of EBs (e.g., low and high) and ELD
textbook adoption patterns (3 possible patterns)—yielding 6 total cells. I used stratified sampling
to identify districts. Interviews lasted approximately 45 minutes and were audio recorded and
transcribed. Interview data was analyzed using NVivo software.
5
I use a binomial logistic model for all research questions to estimate the propensity score.
20
In the initial cycle of interview data review, which required at least three readings of the
full transcript, I used structural codes, which entail little interpretation. Structural coding “acts as
a labeling and indexing device, allowing researchers to quickly access data likely to be relevant
to a particular analysis from a larger data set” (Namey, 2008, p. 141). Though interviews covered
a range of topics, including program model for EBs, I subjected the entire transcript to this initial
coding, since district leaders spoke about instructional materials in response to questions not
directly about instructional materials (e.g., bilingual model). The first cycle of coding revealed
52 structural codes related to English language development textbook adoption. These structural
codes were further categorized and analyzed in the second stage of review.
In the second cycle of coding, the primary goal is to create categorical, thematic, and
conceptual organization from the array of codes that emerged from first cycle coding (Saldana,
2013). I used pattern coding in the later stages of data analysis to group codes into a smaller
number of themes. Five main factors emerge from the interview data. The following list of
factors shape district leaders’ decisions around ELD textbook adoption: number of EBs,
academic achievement of EBs, district leaders’ beliefs about second language acquisition, district
budget, and program model for EBs. I discuss each of these factors in greater detail in the
paragraphs that follow.
One factor that district leaders take into consideration when deciding which ELD
curriculum to adopt is the number of EBs in their district. Six out of 28 district leaders cited EB
enrollment as a factor for ELD textbook adoption. For example, a curriculum facilitator in
district 15 states that “the number of students that we have identified [as ESL]” is a major
determinant in whether or not ELD materials are purchased. This district leader elaborates”
21
If there’s a high number of students then, of course, the need [for ELD instructional
materials] would be greater. But if there is a low number of students then our expert
teachers in the bilingual department will be able to modify their instruction to meet the
needs of their ESL or ELL students.
For this particular district, bilingual teachers are expected to meet the linguistic needs of
EBs in the event that ELD materials are not purchased. To model the treatment assignment, I
include the number of EBs—at both the school and district level--in the propensity score.
Second, district leaders consider EBs’ academic achievement when making curricular
decisions. Eight out of 28 district leaders report EBs’ test scores (e.g., scores on the STAAR test,
scores on language proficiency tests) as a factor that influences ELD textbook adoption. District
Leader 18 shares, “We’re a pretty successful district when you look at state scores. What we
have found is that a lot of the times the basals don’t get you where you need to get. We pull from
a lot of different resources.” Since this district leader seems to be satisfied by her district’s test
scores, she does not see a need to purchase ELD materials. Instead, she believes that the district’s
current curriculum, which seems to be a mixture of various resources, is sufficient to meet the
needs of the students.
A third factor that influences ELD materials purchasing is district leaders’ beliefs about
second language acquisition. Thirteen out of 28 district leaders use their beliefs about second
language acquisition to make decisions about ELD materials. Some leaders believed that second
language learning is not much different from first language learning; thus, EBs do not need
additional materials. For instance, District Leader 10 states, “We don’t have a particular program
that we offer or provide for [EBs]…We just rely on good instruction.” This district did not
purchase any ELD curricular materials because school administration believes that the
22
instruction that all students receive should be effective for EBs as well. Since it is difficult to
directly assess district leaders’ beliefs about second language acquisition for every district in
Texas, I adjust for the following Census characteristics at the district level: median income and
the percentage of non-English speakers in the district. Since beliefs about language learning may
vary based on professional status (Horwitz, 1999), I include median income to proxy for
professional status. As classroom and social peers influence learning (Arnold, 1999), I include
the percentage of non-English speakers in the district as a means of capturing the variation in
beliefs about second language acquisition. Furthermore, I choose percentage of non-English
speakers because a student’s comfort in the classroom is correlated with oral language
proficiency (Krashen, 1981). EBs who go to school with other EBs may feel more comfortable in
the classroom as they have peers that can relate to their academic experiences.
Fourth, cost is another factor that districts weigh in their decision to purchase ELD texts.
Sixteen out of 28 district leaders report that cost is an important consideration in ELD materials
purchasing. An elementary school principal from District 21 states, “We are a small, rural
district. So we also have to think about using our money wisely too…the main thing we have to
make sure of is that it’s something we can afford. Price would be the big decision-maker.” Along
the same lines, District Leader 15 states that “the budget is always a factor.” I include district per
pupil revenue along with district instructional expenditures to capture each district’s financial
ability to purchase ELD materials.
Fifth, a district’s program model for EBs emerges as another factor influencing ELD
materials purchasing. Eight out of 28 district leaders state that the district’s program model for
EBs impacts whether or not ELD materials are adopted. Districts that have bilingual programs
are less likely to purchase ELD texts because they are already purchasing texts in other
23
languages, usually Spanish, for their EBs which reduces the amount of money that is available
for ELD texts. In addition, districts that have bilingual programs also have bilingual teachers and
some districts use bilingual teachers’ knowledge of second language acquisition to supplant ELD
instructional materials. In the propensity score estimation, I include a dichotomous variable
indicating the presence of a bilingual program.
Prior research. In addition to district leader interviews, prior research on textbook
adoption can help to model the selection mechanism (Caliendo & Sabine, 2008; Heckman,
Ichimura & Todd, 1997). For instance, Bhatt and Koedel (2012), in modeling the association
between textbook adoption and achievement, use the following covariates in their propensity
score model: enrollment, attendance rates, demographics (race, language status, free lunch
status), achievement outcomes from prior years, per pupil expenditures, and Census measures
(e.g., median household income, share of adults with a high school diploma). Table 1.3 contains
the final list of covariates used to estimate the propensity score.
Final List of Covariates. When creating a list of covariates for the propensity score
estimation, there is a tradeoff between bias and variance. Including an irrelevant variable can
lead to increased variance (Augursky & Schmidt, 2001), while excluding a potentially important
confounder can lead to increased bias (Heckman, Ichimura & Todd, 1997; Stuart, 2010). To
minimize bias and variance, I start with a comprehensive list of all potential covariates that
emerge from district leader interviews and prior research. The covariates that are used to
estimate the propensity score should include all of the key factors affecting both the outcome
variable and the treatment assignment (Smith & Todd, 2005). In addition, Shadish and Steiner
(2010) posit that the most important factor in propensity score analysis is the quality of the
24
pretest measure. Following their recommendation, I include district average TELPAS scores
from 2011 as a pretest measure for the outcome variable, TELPAS scores from spring 2015.
The final list of covariates is as follows. At the school level, I include the following
characteristics from the 2010-2011 school year: enrollment, average years of teacher experience,
number of bilingual teachers, percentage of students who qualify for special education,
percentage of Hispanic students, percentage of EBs, bilingual program indicator, school average
math and ELA achievement. At the district level, I include the following variables from the
2010-2011 school year: enrollment, total per pupil revenue, percentage of EBs, per pupil
expenditure, urbanicity, district average math and ELA scores, and district average TELPAS
composite scores. From the 2010 Census, I include the following two variables in the propensity
score estimation for each district: median household income and percentage of the population
that speaks a language other than English.
I use a binomial logistic model for all research questions to estimate the propensity score.
Though a linear probability model may be used, logistic models have fewer shortcomings. For
example, linear probability models do not bound the outcome to [0,1], and the error term is not
normally distributed.
Balancing Tests
I test for covariate balance using three different tests. Multiple tests are used because
there is no clearly preferred test for balance, and different tests may yield different results (Smith
& Todd, 2005b). Since there are two research questions, the propensity score must be calculated
separately. While the first propensity score measures the probability of not choosing an ELD
text, the second propensity score measure the likelihood of adopting Rigby On Our Way to
English. For this reason, covariate balancing is assessed twice (Table 1.3 and 1.4).
25
Table 1. 3 Covariate-by-Covariate Balancing Details for the Comparison Between No ELD Text and State-Adopted ELD Text
Mean for
Treated
Mean for
Control
Standardized
difference
t-test p-value
Variance(T)/
Variance(C)
School Characteristics
Enrollment 619.59 632.34 -0.072 -0.85 0.397 0.77
Share EB 0.41 0.41 -0.017 -0.2 0.841 1.49
Share Hispanic 0.68 0.72 -0.172 -1.96 0.05 1.15
Share SPED 0.07 0.07 0.09 0.96 0.335 1.04
Average Teacher Years of Experience 10.69 10.99 -0.112 -1.13 0.257 0.86
Number of Bilingual Teachers 24.24 28.67 -0.255 -2.5 0.013* 0.78
Bilingual Program Indicator 0.9 0.94 -0.107 -1.47 0.142 --
School Outcomes
School Average Math Score 590.62 592.07 -0.05 -0.56 0.578 1.08
School Average ELL Math score 586.74 587.76 -0.028 -0.3 0.764 1.02
School Average ELA Score 601.15 599.41 0.054 0.61 0.539 1.26
School Average ELL ELA Score 587.26 586.19 0.025 0.26 0.793 0.87
District Characteristics
District Enrollment 31360.03 29779.48 0.039 0.69 0.488 0.74
Share ELL 22.44 21.03 0.112 1.26 0.21 1.76
Per Pupil Expenditure 10924.73 11211.18 -0.118 -1.33 0.183 3.77
Total Revenue Per Pupil 1348.79 1399.63 -0.093 -0.92 0.36 1.4
Urbanicity 4.08 3.93 0.051 0.5 0.618 0.58
District Outcomes
District Avg TELPAS Score 3.08 3.08 -0.004 -0.04 0.966 0.93
District Average Math Score 596.02 595.68 0.018 0.2 0.844 1.03
District Avg Math Score for EBs 587 587.13 -0.006 -0.07 0.945 1.61
District Average ELA Score 609.87 608.22 0.077 0.87 0.384 1.07
District Avg ELA Score for EBs 587.98 585.54 0.104 1.12 0.262 1.46
District-Area Characteristics (Census)
Median Household Income 47976.89 46785.11 0.08 0.82 0.413 1.09
Share Non-English Speakers 42.85 43.21 -0.014 -0.15 0.879 1.13
Propensity score 0.34 0.34 0 0.00 0.997 1
26
Note. *p<0.05; **p<0.01; ***p<0.001
Table 1. 4. Covariate-by-Covariate Balancing Details for the Comparison Between Rigby On Our Way to English and National
Geographic Reach
Mean for
Treated
Mean for
Control
Standardize
d difference
t-test p-value
Variance(T)/
Variance(C)
School Characteristics
Enrollment 670.99 695.17 -0.126 -1.53 0.127 1.07
Share EB 39.2 38.01 0.067 0.91 0.365 1.16
Share Hispanic 0.71 0.69 0.103 1.21 0.225 0.86
Share SPED 7.4 7.78 -0.15 -1.94 0.053 1.45
Average Teacher Years of Experience 10.92 11.09 -0.066 -0.77 0.439 0.93
Number of Bilingual Teachers 21.85 18.61 0.188 2.14 0.033* 1.1
Bilingual Program Indicator 0.88 0.91 -0.097 -1.26 0.209 --
School Outcomes
School Average Math Score 585.29 585.61 -0.011 -0.13 0.895 0.84
School Average ELL Math score 583.44 580.96 0.068 0.84 0.404 0.84
School Average ELA Score 594.97 596.91 -0.06 -0.76 0.446 1.08
School Average ELL ELA Score 583.58 583.74 -0.004 -0.05 0.962 0.93
District Characteristics
District Enrollment 37628.28 36155.02 0.032 0.81 0.418 0.97
Share ELL 20.75 19.81 0.098 1.25 0.213 0.45
Per Pupil Expenditure 11195.11 11306.95 -0.067 -0.69 0.49 0.86
Total Revenue Per Pupil 1305.36 1222.28 0.167 1.84 0.067 1.03
Urbanicity 2.46 2.68 -0.083 -0.91 0.361 1.05
District Outcomes
District Avg TELPAS Score 3.15 3.15 -0.02 -0.24 0.81 0.9
District Average Math Score 592.09 593.71 -0.088 -1.08 0.282 0.8
District Avg Math Score for EBs 583.21 584.67 -0.081 -0.95 0.344 0.73
District Average ELA Score 605.4 606.88 -0.074 -0.91 0.362 0.95
District Avg ELA Score for EBs 584.19 585.35 -0.058 -0.71 0.48 0.87
District-Area Characteristics (Census)
27
Median Household Income 47996.23 50695.91 -0.189 -2 0.046* 0.81
Share Non-English Speakers 47.95 46.59 0.06 0.67 0.505 0.97
Propensity score 0.52 0.52 0.001 0.01 0.99 1
Note. *p<0.05; **p<0.01; ***p<0.001
Running head: VARIATION IN ENGLISH LANGUAGE DEVELOPMENT MATERIALS 28
28
Following Rubin (2001), I calculate the standardized difference in the mean propensity
score in the two groups. To minimize bias, the standardized difference should be zero or close to
zero. Though there are no formal criteria for evaluating standardized differences, Rosenbaum
and Rubin (2001) suggest that values greater than 0.20 should be interpreted as large. The
standardized difference for the propensity score is 0.001 for both research questions.
Second, I calculate the variance ratios of the propensity score distribution in the treatment
and control groups (Rubin, 2001). To minimize bias, the variance ratios should ideally lie
between 0.8 and 1.25. Variance ratios smaller than 0.5 or greater than 2.0 are considered extreme
(Rubin, 2001). The variance ratio of the propensity scores are 1.00 for both research questions.
I extend the first two balancing tests to include individual covariates. Though Rubin
(2001) recommends balancing on propensity score alone, I also calculate the standardized
difference and variance ratio for each covariate in the propensity score model. By doing so, I can
gain a more robust understanding of the comparability of the control and treatment groups. For
the first research question, I find that the variance ratio for all covariates is in the range of
acceptability [0.5, 2.0] except for district per pupil expenditure which has a variance ratio of
3.77. State-adopters have a wider range of per pupil expenditures than non-adopters. For the
second research question, the variance ratio for share of EBs is 0.45 which is slightly outside the
range of acceptability. Schools that adopt National Geographic Reach have a larger range of
percentage of EBs compared to school that adopt Rigby On Our Way to English. The absolute
value of the standardized difference is less than 0.20 for all covariates except for the number of
bilingual teachers which has a standardized difference of -0.26. For the first research question,
non-adopters, on average, have 29 bilingual teachers while state-adopters average 24 bilingual
teachers per campus.
29
Third, I present results from hypothesis tests. Though hypothesis tests and p-values
should not be used as measures of balance (August, 2007; Imai, King, and Stuart, 2010; Stuart,
2010), they are still common in balancing tests. Results from hypothesis testing may be
misleading since small differences in means may be statistically significant in a large sample.
Though results from hypothesis tests are not given much consideration, I present them to create a
more robust understanding of covariate balance. For the first research question, treatment and
control schools differ significantly on one variable—number of bilingual teachers. For the
second research question, treatment and control schools differ significantly on two variables—
number of bilingual teachers and median household income.
Overall, these tests suggest that the individual covariates and the propensity scores are
well balanced between treated and control units. The standardized difference for the propensity
score is 0.001 the variance ratio is 1 for both balancing tests.
Matching Algorithm
Matching algorithms assume that the outcome variable is independent of treatment
assignment conditional on a set of observable characteristics (Smith & Todd, 2005). In addition,
it is assumed that for all conditioning variables, there is a positive probability of assignment to
treatment or control conditions. This assumption implies that a match can be found for all treated
cases. Matching estimators perform well in replicating the results of a randomized control trial
when the following criteria are met: 1.) the same data sources are used for both control and
treated units, 2.) control and treated units reside in the same labor market, and 3.) the data
contain a rich set of covariates that impact both treatment assignment and the outcome variable
(Heckman, Ichimura, & Todd, 1997; Heckman et al., 1998; Smith & Todd, 2005). There is
strong evidence suggesting that all three conditions are met in this study. First, covariates and
outcome variables are the same for both treatment and control schools and is also measured in
30
the same way for both groups. Second, all treatment and control schools reside in the same
geographic area—Texas. Third, conditioning variables adjust for a wide array of
characteristics—reading and math achievement, English proficiency, teacher quality,
demographic characteristics, and median household income. Since these foundational
requirements have been met, it is appropriate to proceed with matching.
Four commonly used matching algorithms are: kernel matching, local linear matching,
inverted probability weights with regression adjustment (IPW-RA), and nearest-neighbor
matching. In the following paragraphs, I outline each matching algorithm and justify my choice
of kernel matching.
Nearest neighbor matching, one of the most common and easiest to implement, selects
for each treated case i the control case with the smallest distance from i (Smith & Todd, 2005;
Stuart, 2010). Since nearest neighbor matching discards control units that were not selected as
matches, this method may have reduced power compared to other matching algorithms. An
additional concern is that this nearest neighbor matching may lead to poor matches. Caliper
matching, a variant of nearest neighbor matching, alleviates the concern regarding bad matches
by imposing a caliper; only matches within a certain caliper are selected. Though caliper
matching can improve standard errors relative to nearest-neighbor matching, caliper matching
can lead to greater bias (Dehejia & Wahba, 2002). Matching with replacement can also avoid
bad matches by allowing a control unit to be matched to several different treatment units. When
using nearest neighbor matching, the number of matches needs to be specified. Selecting
multiple control units for each treated case may increase bias since the 2
nd
and 3
rd
match are, by
definition, further away from the 1
st
closest match (Stuart 2010).
With inverse probability weighting, units that are underrepresented are weighted more
heavily. Units that are overrepresented are downweighted. Inverse probability weighting can be
31
combined with an additional covariance adjustment, that is, by regressing the outcome on key
covariates; this method will be known henceforth as IPW-RA. Covariance adjustment can
correct for residual bias due to a misspecified propensity score model (Steiner & Cook, 2013).
The regression is as follows:
(3)
where Y
i
is the school average TELPAS composite score, Z
i
is the treatment indicator, the
treatment effect, X
i
the vector of covariates, and β is the corresponding coefficient vector (Ho et
al., 2007; Rubin, 1979). Combining propensity score methods and covariance adjustments
reduces residual bias due to a misspecified propensity score model, but only if the outcome
model is correctly specified (Robins & Rotnitzy, 1995). If both models are misspecified, it is
possible for bias to increase (Schafer & Kang, 2007). However, an additional covariance
adjustment usually reduces bias and standard errors because it corrects for residual bias due to
inexact matches (Steiner & Cook, 2013).
Kernel matching is a recently developed nonparametric matching estimator that relies on
weaker assumptions than the previous two methods. Smith and Todd (2005) describe kernel-
based matching as a weighted regression of the counterfactual outcome on an intercept with
weights given by the kernel weights. While nearest neighbor and IPW-RA matching requires the
conditional independence assumption to be met, kernel matching only requires that the outcome
under the control condition for the treated cases to be independent of the treatment assignment,
conditional on observed covariates (Heckman et al., 1998). In addition, kernel-based matching
can reduce variance relative to other matching techniques because it relies on more information.
While common matching techniques like nearest neighbor matching only use a few observations
32
from the comparison group to construct the counterfactual, kernel-based matching uses a
weighted average of comparison units near the treatment unit.
Local linear matching is a type of kernel matching that provides several advantages over
kernel matching. While the estimated intercept in kernel matching provides the estimate of the
counterfactual mean, local linear matching also includes a linear term in addition to the intercept.
A linear term is useful whenever control units are distributed asymmetrically around treatment
units (e.g., where there are gaps in the distribution of the propensity score or near the extreme
values of the propensity score).
I choose local linear matching (LLR) over other matching methods for three main
reasons. First, local linear matching can reduce variance relative to other matching techniques
because it relies on more information. While common matching techniques like nearest neighbor
matching only use a few observations from the comparison group to construct the counterfactual,
local linear matching uses a weighted average of comparison units near the treatment unit. Since
more information is used, there is lower variance which leads to more precise estimates of the
treatment effect. Second, local linear matching includes a linear term in the propensity while
kernel matching does not. This added intercept is an advantage whenever there are gaps in the
propensity score. Lastly, local linear matching relies on weaker assumptions compared to other
matching approaches. Rosenbaum and Rubin’s (1983) framework for propensity score matching
requires strongly ignorable treatment assignment and common support; kernel-based matching
methods such as local linear matching replaces these two assumptions with weaker ones
(Heckman et al., 1998). Instead of assuming strongly ignorable treatment assignment, Heckman
and colleagues only require that the outcome under the control condition for the treated cases to
be independent of the treatment assignment, conditional on observed covariates (Guo & Fraser,
2010). Instead of assuming full independence, Heckman and colleagues impose mean
33
independence. Though local linear matching is my method of choice, I also include results from
kernel matching, inverse probability weighting with regression adjustment (IPW-RA), and
nearest neighbor matching with 1, 2, and 3 neighbors (which will be abbreviated K1, K2, and
K3, respectively).
Common Support
Following Dehejia and Wahba (2002), I trim the sample to determine the support region.
I drop control units that are below the minimum propensity score for treated units. However,
there are no control units below the minimum propensity score for treated units for either
research question. Thus, zero control units were dropped. I also assess for common support by
visually inspecting a probability density distribution of the propensity score in both groups.
Lechner (2000b) argues that common support can be assessed visually, and there is no need to
implement more complicated, formal processes.
Results
Research Question 1: Non-Adopters vs. State-Adopters
Using local linear matching, I compare the effect of not purchasing an ELD text to the
effect of purchasing a state-adopted ELD text. I find that elementary schools that do not purchase
any elementary ELD texts score 0.29 standard deviations lower on school-average TELPAS
composite scores (see Table 1.5). Following Bhatt, Koedel and Lehman (2013), standard errors
are estimated by bootstrapping using 250 repetitions and clustered at the district level.
Table 1. 5 Average treatment effects of non-adoption relative to state-adopted ELD instructional
materials on TELPAS spring 2015 composite scores
LLR Kernel IPWRA K1 K2 K3
Treatment effect -0.29* -0.30* -0.34*** -0.30*** -0.33*** -0.35***
(0.14) (0.13) (0.07) (0.09) (0.08) (0.07)
N (schools) 1,014 1,014 1,014 1,014 1,014 1,014
34
Note. Standard errors are in parentheses. Standard errors are estimated by bootstrapping using
250 repetitions and clustered at the district level for the local linear and kernel matching models.
Robust standard errors are calculated for K1 matching and IPWRA models.
*p<0.05; **p<0.01; ***p<0.001
To address the robustness of the findings, I compare results across all matching models.
The results from the kernel matching estimator are similar to the estimates from the local linear
matching model with non-adopters scoring 0.30 standard deviations lower than state-adopters.
Results from the IPWRA and nearest neighbor matching models are similar in magnitude with
non-adopters scoring 0.30-0.34 standard deviations lower than state-adopters. The standard
errors in these models are roughly half the size of the standard errors in the local linear matching
model. One reason for the large standard errors for the LLR and kernel matching models may be
due to the use of bootstrapping methods. The nearest neighbor matching and IPW-RA models
rely on Abadie-Imbens standard errors (Abadie & Imbens, 2008) instead of bootstrapped
standard errors.
Regardless of the type of matching estimator used, a common pattern emerges across all
six models; schools that do not adopt any ELD curricula score have TELPAS composites scores
that are roughly 0.29 standard deviations lower than schools that purchase state-adopted ELD
curricula. As previously mentioned, these results are only generalizable to schools in Texas
whose TELPAS scores are not masked, who use the same ELD curricula in grades K-3, who did
not have missing textbook data, and had fewer than 150,000 students.
Research Question 2: Rigby On Our Way to English vs. National Geographic Reach
When comparing the Rigby on Our Way to English curricula to National Geographic
(NG) Reach, I find that schools that use Rigby have slightly higher TELPAS composite scores in
all outcome models but these differences are not significant (see Table 1.6). The standard errors
for these estimates are slightly larger than the standard errors for the previous research question
35
due to the difference in sample size. It is possible that there is a significant treatment effect in the
population; however, this study may not be able to identify the treatment effect given the limited
sample size. Relatedly, this study was not able to assess the effectiveness of any other ELD
curricula due to small sample size.
Table 1. 6 Average treatment effects of Rigby On Our Way to English relative to National
Geographic Reach on TELPAS spring 2015 composite scores
LLR Kernel IPWRA K1 K2 K3
Treatment effect 0.05 0.05 0.14 0.13 0.16* 0.17*
(0.17) (0.16) (0.08) (0.08) (0.07) (0.07)
N (schools) 706 706 706 706 706 706
Note. Standard errors are in parentheses. Standard errors are estimated by bootstrapping using
250 repetitions and clustered at the district level for the local linear and kernel matching models.
Robust standard errors are calculated for K1 matching and IPWRA models.
*p<0.05; **p<0.01; ***p<0.001
The nonsignificant differences in TELPAS scores between Rigby-adopters and NG-
adopters is, for the most part, mirrored in the kernel matching, IPWRA, and K-1 matching
models. However, the K-2 and K-3 matching models find that Rigby-adopters have significantly
higher TELPAS composite scores. According to the K-2 and K-3 matching models, Rigby users
score 0.16 standard deviations higher on TELPAS composite scores relative to schools that adopt
National Geographic Reach. The standard errors are smaller in the K2 and K3 models because
there is more information being used. However, these estimates should be interpreted with
caution as bias increases as more control units are matched to each treated unit.
Falsification Tests
Propensity score matching will not provide causal estimates if the conditional
independence assumption (CIA) is violated. The CIA can be violated if there are systematic
differences in school or district quality across the various curriculum adopters that are not
captured by the covariates in the model. If these differences are correlated with curriculum
36
adoptions and student achievement, then the estimates may be biased. For example, if there are
differences in commitment to improving achievement for EBs across the various curriculum
adopters, this could bias the estimates. Though it is impossible to rule out all possible sources of
bias, I provide evidence about the general reliability of the findings using three falsification tests.
First, I estimate ELD curriculum effects on math achievement for 3
rd
grade EBs (see
Table 1.7). If there are unobserved differences across adopters that are not captured by the
covariates in the model, then we might see significant differences between treated and control
units on math scores for 3
rd
-grade EBs in 2015. The confounding factor (e.g., commitment to
improving educational outcomes for emergent bilinguals) will be present in the form of non-zero
curriculum effects. If curriculum effects are not significantly different from zero for unexposed
cohorts, then it is the curricula themselves, not other unobserved differences, that are driving the
estimates. Since English language proficiency is correlated with math achievement (Grant, Cook,
& Phakiti, 2011), it is possible that non-adopters have lower math achievement than state-
adopters. At most, these spillover effects will be small and no larger than the effects from the
main findings. I find that schools that do not adopt any ELD curricula score 0.04 standard
deviations lower on the state math test than scores that do adopt ELD curricula; however, this
difference is not significant. Indeed, these spillover effects are smaller than the main treatment
effect of 0.29 standard deviations. While the estimates from the kernel and local linear matching
models are negative, the other models yield positive treatment effects on STAAR 2015 math
achievement. However, these differences are not concerning as the magnitude of the estimates
are relatively small and non-significant. There appears to be no evidence that the primary
findings are driven by unobserved differences between non-adopters and state-adopters.
37
Table 1. 7 Falsification Tests: Effects of Non-Adoption versus State-Adoption on STAAR 2015
Math Achievement
LLR Kernel IPWRA K1 K2 K3
Treatment effect -0.03 -0.05 0.03 0.12 0.10 0.09
(0.14) (0.14) (0.07) (0.08) (0.08) (0.08)
N (schools) 1,076 1,076 1,076 1,076 1,076 1,076
Note. Standard errors are in parentheses. Standard errors are estimated by bootstrapping using
250 repetitions and clustered at the district level for the local linear and kernel matching models.
Robust standard errors are calculated for K1 matching and IPWRA models.
*p<0.05; **p<0.01; ***p<0.001
In the second falsification test, I use 3
rd
grade TELPAS scores from spring 2012 as the
outcome variable. By the end of the 2011-2012 school year, 3
rd
-graders will have been exposed
to the ELD curricula for no more than one year. Since ELD curricular materials were adopted
during the 2011-2012 school year, it is unlikely that the materials had a large impact on English
proficiency less than a year after implementation. If there are significant curricular impacts, then
the impact should be smaller in magnitude than the impact on TELPAS spring 2015 scores.
Indeed, there are no significant differences between 3
rd
graders in treatment and control schools
during the 2012 TELPAS test (see Table 1.8). In addition, the estimates are all negative which is
consistent with the finding from the main analysis that non-adoption is associated with lower
English language proficiency. Less than one year after implementation, I find that non-adopters
score 0.08 standard deviations lower than state-adopters on the state English language
proficiency test; however, this difference is not significant.
Table 1. 8 Falsification Tests: Effects of Non-Adoption versus State-Adoption on 2011-2012 3rd-
grade TELPAS Composite Scores
LLR Kernel IPWRA K1 K2 K3
Treatment effect -0.08 -0.06 -0.02 -0.04 -0.05 -0.07
(0.14) (0.15) (0.08) (0.08) (0.08) (0.08)
N (schools) 1,028 1,028 1,028 1,028 1,028 1,028
38
Note. Standard errors are in parentheses. Standard errors are estimated by bootstrapping using
250 repetitions and clustered at the district level for the local linear and kernel matching models.
Robust standard errors are calculated for K1 matching and IPWRA models.
*p<0.05; **p<0.01; ***p<0.001
In the third falsification test, I use a comparative interrupted time series (CITS) model to
estimate the impact of ELD adoption on English language proficiency. In the CITS regression, I
adjust for the same school and district characteristics that are used in the propensity score
estimation. The results from the CITS model mirrors the results from the local linear matching
model with non-adopters scoring 0.30 standard deviations lower than state-adopters (p<0.01).
With regards to the second research question, the CITS model reveals no significant differences
in spring 2015 English language proficiency score between schools that adopt Rigby and schools
that adopt National Geographic. Given the robustness of results across various quasi-
experimental methods, there is strong evidence to suggest that state-adopted ELD textbooks have
a positive impact on emergent bilinguals’ English language proficiency.
In sum, all three falsification tests point towards the general reliability of the findings.
None of the tests suggests that there are systematic differences between treatment and control
schools. In addition, I employed a different type of quasi-experimental method—comparative
interrupted time series—and recovered similar treatment effects for both research questions.
Discussion
To make causal claims about the effectiveness of ELD textbooks, the conditional
independence assumption must be met. In other words, the selection process needs to be
accurately modeled so that potential outcomes are independent of textbook choice conditional on
observables. If relevant unobserved characteristics are omitted from the model, then the resulting
estimates may be biased. However, Heckman et al. (1997) argues that mismatching and
misweighting of the data is numerically more important than bias due to selection on
39
observables. Though selection bias is a real threat to internal validity, selection bias may not be
as big of a threat as mismatching and misweighting. Thus, having balanced groups is extremely
important in obtaining unbiased estimates of the treatment effect.
There are several reasons to believe that the estimates are unbiased. First, I use multiple
balancing tests to assess balance, and all tests show that the covariates are well balanced. The
standardized difference of the propensity score is nearly zero and the variance ratio is close to 1.
Second, I include pretest measures of the outcome variable, which some researchers argue is the
most important factor in propensity score analysis (Shadish & Steiner, 2010). Third, through
interviewing district leaders in Texas, I am able to study the selection process. This allows me to
choose a rich set of relevant covariates to estimate the propensity score. Fourth, the treatment
effect remains robust across various matching models. Though local linear matching was my
model of choice due to its reliance on weaker assumptions, I also use inverse probability
weighting and nearest neighbor matching. The estimates generally remain the same across all
models. Non-adopters score 0.29 standard deviations lower on the TELPAS composite score
relative to state-adopters. I also use a CITS regression to model the association between ELD
textbook adoption and English language proficiency and was able to recover treatment effects
that are nearly identical to the estimates produced by the local linear matching model.
These curriculum effect estimates may be lower bound estimates because interviews with
district leaders reveal that ELD texts are used modestly in many districts. For example, a district
leader acknowledges that their adopted ELD curricula, Rigby On Our Way to English, is “not
used very well.” Three district leaders report using their ELD curricula only with newcomers,
which represent a small portion of their EB population. If ELD texts were implemented as they
were intended, it is possible that the treatment effects will be even larger.
40
It is also important to discuss what this study is not suggesting. This study does not
suggest that off-list ELD materials, or ELD materials that are not on the state-adopted list, are
ineffective for ELD instruction. Off-list ELD curricula such as EL Achieve and Carousel of
IDEAS may also be effective but due to small sample size I was not able to evaluate the
effectiveness of off-list texts.
As previously noted, the data limitations in this study restrict the generalizability of the
findings. The results from this study are only generalizable to elementary schools in Texas that
adopted the same ELD curriculum in grades K-3, reported TELPAS scores in spring 2015,
reported ELD textbook adoption, and had fewer than 150,000 students. In addition, the results
are not generalizable to schools that adopted off-list texts as they were excluded from the study.
Conclusion
This study makes two contributions to the research literature on curricular effectiveness.
This study is the first to assess the effectiveness of ELD curricular materials at the state level.
Prior studies on textbook effectiveness have focused solely on math and English language arts
textbooks. Furthermore, this study makes use of a unique dataset comprised of ELD curricular
materials adoptions in the state of Texas. No prior studies have looked at curricular effectiveness
in Texas. Texas is an appropriate state for the proposed study as it services the second highest
number of EBs in the United States.
Second, the study contributes to the larger research literature on English language
development which is sparse. Saunders, Goldenberg, and Marcelletti (2008) conduct the most
comprehensive review of ELD instructional practices and find that little research examines the
effects of ELD instruction on EBs’ academic achievement. In fact, since ELD research is so
scant, the authors had to rely on second language research in other fields to inform their ELD
guidelines. For example, some of the studies included in their synthesis focused on English
41
learners in Quebec, Canada. The context for learning English in French-speaking Quebec is
substantively different than learning English in the United States. My study contributes directly
to the research literature on ELD because it focuses on English language learners in the United
States.
Future work is needed to understand how teachers across districts are implementing the
adopted ELD curriculum. Again, these estimates may be lower bound estimates as district
leaders report varying levels of ELD curricula usage. The current study is not able to capture the
variation across and within districts with regards to ELD curriculum implementation. It is
possible that the impact of ELD curriculum on EBs’ achievement is mediated by the degree of
implementation. If there is a significant relationship between degree of implementation and EBs’
academic outcomes, then it is important to understand the nature of that relationship.
Furthermore, if implementation fidelity is mediated by certain factors (e.g., practitioners’ beliefs
about second language acquisition), then it is of import to identify these mediating factors so that
barriers to implementation can be addressed.
Qualitative data from interviews with district leaders reveal that there is confusion about
the purpose of ELD instruction and how it differs from instruction for monolingual English-
speakers. For instance, some district leaders believe that instruction for emergent bilinguals is
“just good teaching” (JGT), and, therefore, there is no need to purchase materials specifically for
emergent bilinguals. In other words, whatever works for monolingual English-speakers is
sufficient to address the academic and linguistic needs of emergent bilinguals. This finding is not
new as previous research has found that a “just good teaching” approach is not adequate to meet
the academic and linguistic needs of emergent bilinguals (De Jong & Harper, 2005). Ten years
after De Jong and Harper coined the term JGT to describe this phenomena, I find that the JGT
view of teaching EBs is still pervasive. In short, I reiterate De Jong and Harper’s (2005) call to
42
explicitly include emergent bilinguals at all levels of educational policy and practice. If there are
expectations that content standards be explicitly and rigorously addressed, the same expectation
should exist for English language proficiency standards. Otherwise, a double standard exists
based on language which is effectively a proxy for national origin (Lau v. Nichols, 1974).
Castañeda v. Pickard (1981) mandated that educational programs for emergent bilinguals
be tested for program efficacy, which is what I attempt to do in this study. Using local linear
matching, I find that non-adopters have lower English proficiency relative to state-adopters.
Schools that do not have curricula that foreground English language proficiency should consider
obtaining instructional materials that explicitly addresses forms of English, provides ample
opportunities to develop all four language modalities, and distinguishes between conversational
and academic language. Though there is evidence to suggest that ELD instructional materials can
improve English language proficiency, much more work needs to be done to adequately meet the
academic and linguistic needs of emergent bilinguals (e.g., tracking of emergent bilinguals,
resegregation efforts that concentrate emergent bilinguals in low-income schools). Implementing
high quality ELD curricula is only one part of an educational agenda that seeks to redress
educational inequities faced by emergent bilinguals.
43
Chapter 3: Variation in English Language Development Materials—How Social
Constructions Shape the Adoption of ELD Materials
Emergent bilinguals are one of the fastest growing and one of the most underserved
student populations in the United States (Capps et al., 2005; Genesee, Lindholm-Leary,
Saunders, & Christian, 2005; National Education Association, 2008). Since 1998, the emergent
bilingual (EB) population in American public schools has grown by 51% (Education Week
2011) and now accounts for 9.3% of all public school students in the United States (National
Center for Education Statistics (NCES), 2016). Emergent bilinguals represent a heterogeneous
population composed of various language backgrounds and immigration statuses. About 77% of
emergent bilinguals in American public schools speak Spanish, with the next four most popular
languages being Arabic, Chinese, Vietnamese, and Hmong (NCES, 2016).
Despite their growing numbers, EBs continue to be underserved in American public
schools. They are more likely to have less effective teachers compared to non-EBs (Ballantyne,
Sanderman, & Levy, 2008; Darling-Hammond, 2010). Furthermore, EBs are more likely to be
“triply segregated”—more likely to attend schools with other EBs, minorities, and low-income
students (Orfield, 2001; Orfield & Lee, 2006; Ríos-Aguilar & Gándara, 2012). Gándara and
Gándara and Orfield (2010) find that linguistic segregation intensifies all the negative impacts of
racial and socioeconomic segregation. On both the math and reading sections of the National
Assessment of Education Progress, a full standard deviation separates emergent bilinguals and
their counterparts (National Center for Education Statistics, 2011).
To improve the educational outcomes of EBs, federal court cases have required that
programs for EBs be effective. In 1974, the Supreme Court mandated in Lau v. Nichols that
schools must teach language minority students English and provide them with access to content-
44
area curriculum. Subsequently, Castañeda v. Pickard (1981) further elevated the status of EBs
on the educational policy agenda by requiring that educational programs for EBs be based on
sound educational theory, be well-implemented, and produce positive results. These seminal
decisions established the right of EBs to high-quality language-learning programs (Slama, 2014).
One part of a quality educational program for EBs is English language development (ELD), or
language instruction that is delivered separately from content (Saunders, Goldenberg, &
Marcelletti, 2013). Saunders et al. (2013) find that a separate, daily block of time should be
devoted to ELD instruction and that the likelihood of sustaining an effective ELD program
increases when districts make it a priority.
Though the federal government has taken steps towards the standardization of policies
and programs for EBs, research shows that the implementation of federal policies can vary
drastically. Specifically, the loose coupling between means and ends allows for local adaptation
of federal policies (Weick, 1977). Meyer and Rowan (1978) posit that maintaining nominal
control over instructional practices allows schools to adapt instruction to unique local contexts.
However, this greater flexibility may lead to local policy implementers carrying out a policy in a
way that differs from what policymakers originally envisioned (Porter, Fusarelli, & Fusarelli,
2015; Werts & Brewer, 2015). Local influence can help to explain the variation in the
implementation of federal mandates. Specifically, this study looks at how the local context can
shape the implementation of state ELD policies.
I look at curriculum materials
6
because they are widely-used in American public schools
and can impact student exposure to academic and language standards. Recent empirical studies
have examined the effects of instructional materials on student achievement and have found
6
Following Remillard (2005), I use the words “curriculum materials” and “curriculum” to refer to the set of
instructional materials used by a school or district. Others may define a “curriculum” as an overarching framework
that specifies what should be taught. In this study, I refer to the former definition of “curriculum.”
45
substantial differences across different types of curricula (Agodini et al., 2010; Bhatt & Koedel,
2012; Chingos & Whitehurst, 2012; Koedel et al., 2017). These differences are noteworthy
because textbooks covering the same material tend to be similar in price. Using quasi-
experimental methods, Le and Polikoff (2017) find that state-approved ELD curricula are
associated with a 0.29 standard deviation increase in English language proficiency. Thus, local
education agencies may be able to improve EB academic achievement through the adoption of
comprehensive English language development (ELD) curriculum materials.
Given that prior research has shown that ELD materials can improve English proficiency
(Le & Polikoff, 2017), an equity issue arises if the adoption of ELD materials varies by district
characteristics. For example, if districts with larger enrollment are more likely to adopt ELD
curricula, then there may be a systemic inequity whereby EBs have unequal access to ELD
resources. Using statewide textbook data from Texas, I examine the relationship between district
characteristics and ELD adoption patterns.
Decision-making about ELD materials does not happen in a vacuum, and it is important
to account for environmental factors in the decision-making process (Coburn, 2004). One aspect
of the environment that influences decision-making is the social construction, or cultural
characterization, of target populations. Schneider and Ingram (1993) posit that the social
construction of groups plays a powerful role in policy design. Understanding the social
construction of target populations may help to explain why certain EB-serving districts purchase
ELD materials while others do not.
46
Figure 2. 1. The Relationship Between Social Constructions and Academic Achievement
Figure 2.1 outlines the connection between social constructions and academic
achievement. The first arrow links social construction to ELD materials adoption as prior
research suggests that the social construction of target population influences the policies that are
prescribed for that particular group. The second arrow connects ELD curricular materials to
English language proficiency which is supported by recent research (Le & Polikoff, 2017). The
third arrow links English language proficiency to academic achievement which is well-
documented (Aina et al., 2013; Wilson & Komba, 2012). I do not test all these connections in my
study; rather, I focus on the first arrow which represents the link between social constructions
and ELD materials adoption.
Using statewide district-level data and interview data from district leaders in Texas, this
sequential mixed-methods study seeks to understand ELD textbook adoption patterns. The
research questions for the study are as follows:
1. To what extent, if any, do ELD adoption patterns vary by district characteristics?
2. To what extent can the social constructions of EBs explain the variance in ELD materials
adoption?
Social
Construction
of EBs
ELD
Curriculum
Materials
Adoption
English
Language
Proficiency
Academic
Achievement
47
This study proceeds as follows. First, I review the relevant background information.
Second, I outline the conceptual framework undergirding the study. Third, I discuss the data and
the study design. Fourth, I describe the quantitative and qualitative findings. Lastly, I conclude.
Background Information
Terminology
Following Burke, Morita-Mullaney, and Singh (2016), I use the term emergent bilingual
in lieu of the following: English language learner (ELL), English as a second language (ESL),
and limited English proficient (LEP). Scholars have objected to the terms “limited English
proficient” and “English language learner” (Garcia, 2009) due to their deficit framing of the
linguistic abilities of emergent bilinguals. The terms also suggest that English proficiency is the
only desired outcome for this population of students (Menken & Solorza, 2014). Referring to this
group of students by an additive name that does not focus on its limitations is an important step
towards the decentering of English monolingual norms. Similar to Burke et al. (2016), I use the
term emergent bilingual with two exceptions. First, the terms ELL, ESL, and LEP may be used
when referencing federal and state laws. Second, the terms ELL, ESL, and LEP may be used in
my literature review if the authors of the studies use the terms. The terms are used
interchangeably, meaning that, even if different terms are used, I refer to the same group of
students.
Emergent Bilinguals
EBs are a heterogeneous group with varying levels of language proficiency,
socioeconomic standing, and immigration status (National Council of Teachers of English,
2008). While 57% of adolescent EBs were born in the United States, 43% were born in a
different country (National Council of Teachers of English, 2008). Of this group, 27% are
second generation, and 30% are third generation. EBs also vary in how much of their native
48
language is used in the home, with some exclusively using their native language while others
speaking mostly English at home. The obstacles that EBs face outside of the classroom are
numerous: higher likelihood of growing up poor, living in segregated neighborhoods, and being
on the outskirts of the dominant culture and language (Ríos-Aguilar & Gándara, 2012).
The obstacles for EBs continue in the classroom as well. EBs are at increased risk for
academic failure because they face the dual task of developing academic English and content
knowledge (Coleman & Goldenberg, 2012; Mora-Flores, 2011; Scarcella, 2003). Not only are
EBs less likely to be assigned fully-credentialed teachers, they are also more likely to be
assigned to beginning teachers (Gándara et al., 2003). The majority of teachers who teach EBs
have had little or no professional development for teaching EBs (NCES, 2002), and few teachers
have taken a course focused on issues related to EBs (Menken & Antunez, 2001). In addition,
many teachers feel inadequately prepared to teach EBs (Gándara, Maxwell-Jolly, & Driscoll,
2005). As a result, educational outcomes differ significantly between EBs and non-EBs. As an
example, in the 2013-2014 school year, the high school graduation rate for EBs was 63%,
compared to 82% for all students (U.S. Department of Education, 2016).
To adequately address the needs of EBs, schools need to accommodate the social,
cultural, and linguistic diversity of the EB population (Gil & Bardack, 2010). Some practitioners
assume that EB students will benefit equally from the same instructional approach delivered to
non-EBs. However, providing high quality educational programs for EBs requires respecting and
integrating EBs’ native language and home culture (Adams & Jones, 2006). Also, schools use
language styles that are common with mainstream families but are less accessible for children
from different cultures (McLaughlin, 1992). Thus, effective programs for EBs need to explicitly
target and teach these language functions (e.g., idiomatic expressions) to EBs.
49
Knowledge of second language acquisition is also needed to work effectively with EBs
(Lucas, Villegas, & Freedson-Gonzalez, 2008). Some EBs may be able to use English fluently in
informal settings but still experience significant literacy challenges in school. Because informal
conversations often focus on speakers’ personal experiences, informal language is relatively
accessible to EBs (Lucas et al., 2008). In contract, academic language is more removed from
personal and shared experiences, making meaning more dependent on the language itself instead
of context. Academic language can be especially challenging for EBs because of its reliance on
specialized vocabulary and language forms (Schleppegrell, 2004). Practitioners who understand
the difference between conversational language and academic language are in a better position to
provide the linguistic scaffolding EBs need to access content (Lucas et al., 2008). O’Day (2009)
find that instructional approaches designed for EBs, such as differentiation, are not effective
unless teachers have a background in second language acquisition.
The Every Student Succeeds Act (2015) requires that all states identify EBs, measure
their English language proficiency, and include these students in state testing programs. Most
states identify EBs using a home language survey which is given to students upon enrollment
into the school system (Samson & Collins, 2012). For children whose home language is not
English, their English language proficiency is assessed using a state approved standardized test.
For example, the state of Texas uses the Texas English Language Proficiency Assessment
System (TELPAS) to determine a child’s English proficiency in the four language modalities:
reading, writing, speaking, and listening. Each state determines the cutoff score for being
considered proficient in English. All students who score below this threshold are entitled to
appropriate instructional programs and funding until they are reclassified as fluent English
proficient (RFEP). Title III of the ESSA allocates federal money to districts to help schools meet
the linguistic needs of EBs (ESSA, 2015).
50
What Types of Programs are Afforded to EBs? There are three broad categories of
programs for EBs: 1.) bilingual, 2.) English-only, and 3.) structured English immersion. Table
2.1 provides an overview of each program.
First, I define the meaning of bilingual program and describe different types of bilingual
programs. Bilingual models, generally, instruct EBs by using both the students’ native language
and English. Bilingual programs can be further subdivided into early-exit transitional bilingual,
late-exit transitional bilingual, one-way dual language, and two-way dual language programs.
Dual language programs differ from transitional programs in that students are expected to
develop and maintain literacy in the native language and English (Brisk, 1998; Flores, Sheets &
Clark, 2011; Francis, Lesaux, & August, 2006). In contrast, transitional bilingual programs begin
instruction in the child’s native language initially but will ultimately transition students into an
English-only setting. In addition, transitional bilingual programs serve language minority
students while two-way dual language programs are typically composed of language minority
students and native English-speakers (Flores, Sheets & Clark, 2011). One-way dual language
programs are comprised entirely of EBs and the goal is to develop literacy in both English and
students’ native language. The main difference between a two-way and a one-way dual language
program is the composition of students, with two-way programs enrolling both EBs and native-
English speakers. Slavin et al. (2011) conducted the first longitudinal, random assignment study
to compare the effectiveness of transitional bilingual education and English-immersion
approaches with EBs. The authors found that Spanish-speaking children learn to read English
equally well in both types of programs. Despite numerous studies showing that bilingual
programs are more effective than English immersion programs (August & Shanahan, 2006;
Collier & Thomas, 2004; Greene, 1998; Rolstad, Mahoney, & Glass, 2005), only a small fraction
of EBs are enrolled in bilingual education (Gándara & Hopkins, 2010).
51
Table 2. 1. Types of Programs for EBs
Type Definition
Early-exit transitional
bilingual
- Instruction begins in the child’s native language but will
ultimately transition students into an English-only setting.
- Students typically transition into all English instruction by grade
3
- Comprised only of EBs
Late-exit transitional
bilingual
- Instruction begins in the child’s native language but will
ultimately transition students into an English-only setting.
- Students typically transition into all English instruction by grade
6
- Comprised only of EBs
One-way dual language - Instruction begins in the minority language and English is
gradually added with no more than 50% of the instructional day
being spent in the majority language (English).
- Comprised only of EBs
Two-way dual language - Instruction begins in the minority language and English is
gradually added with no more than 50% of the instructional day
being spent in the majority language (English)
- Comprised of both EBs and native English speakers
English-only - Instruction is delivered only in English
Structured English
immersion
- Unmodified English is gradually phased in
- Instruction relies on special techniques to make English
accessible to EBs
The second type of EB program, English-only, does not use students’ native language,
and EBs may be placed in classrooms with students who were never EBs at the onset. In
practice, English-only programs may contain elements of both structured immersion and
English-only programs. Pure English-only programs are uncommon (Francis, Lesaux, & August,
2006).
The third type of EB program, structured English immersion, refers to a “well-planned,
gradual phase-in of unmodified English instruction relying initially on special techniques to
52
make content delivered in English accessible to English language learners” (Francis, Lesaux, &
August, 2006, pg. 366). Even though English immersion programs, on average, produce lower
achievement gains for EBs compared to bilingual programs (August & Shanahan, 2006;
Cummins, 1979, 2000; Goldenberg, 2008; Greene, 1998; Rolstad, Mahoney & Glass, 2005;
Willig, 1985), some states--Arizona, and Massachusetts--have enacted restrictive policies that
mandate English immersion as the default program for EBs (Gándara & Hopkins, 2010).
Though the definitions of these three programs may seem clearly delineated, there is
great variability within each program type. For example, the use of students’ native language is
highly variable even within a bilingual model due to district administrators’ interpretation of
language policies and teachers’ beliefs and language skills (Gándara et al., 2000). The amount of
instructional time spent in each language is rarely accounted for in program evaluation studies
which may led to biased results (August & Shanahan, 2006). For example, some prior studies
have found that English immersion programs produce greater achievement gains for EBs
compared to bilingual programs (Baker, 1998; Porter, 1990; Rossell & Baker, 1996); however,
authors of these studies rarely conduct classroom observations to assess fidelity to the model
(Francis, Lesaux, & August, 2006). It is possible that teachers are implementing bilingual
programs differently in classrooms and that implementation differs dramatically from the
original design. The success of native language instruction depends heavily on how it is
implemented (Gomez, Freeman, & Freeman, 2010).
Research on English Language Development
Regardless of the program model (i.e., bilingual, English-only, structured English
immersion), research shows that EBs benefit from a separate, daily block of ELD (Goldenberg,
Saunders, & Marcelletti, 2013). The purpose of ELD is to provide students with the English
proficiency needed to succeed academically (Snow & Katz, 2010). Though the focus of ELD is
53
to develop academic language development, proficiency also requires the mastery of the social
and pragmatic uses of language. Goldenberg, Saunders, and Marcelletti (2013) define ELD
instruction as instruction that focuses specifically on helping emergent bilinguals develop
English language skills and that is delivered separately from academic content.
ELD may be configured in a variety of ways. Some schools may choose to deliver ELD
instruction during a designated ELD block while other schools may utilize a pull-out model in
which emergent bilinguals leave the classroom for a specified period of time. ELD instruction
may take place in mainstream classrooms, structured immersion classes, during sheltered content
instruction, or in bilingual classrooms. Regardless of how local education agencies choose to
configure ELD, research shows that the following six elements should be present for effective
ELD instruction.
First, there is robust evidence suggesting that a separate, daily bock of time should be
devoted to ELD instruction (O’Brien, 2007; Saunders et al., 2013). Saunders, Foorman, and
Carlson (2006) find that EBs who were provided a separate ELD instructional block significantly
outperformed EBs whose ELD block was integrated within the language arts block. Their study
included more than 1,200 students located in 25 schools across California and Texas. Similarly,
O’Brien (2007) examined oral language outcomes amongst Spanish-speaking EBs in three
different ELD configurations: 1.) classrooms with a separate ELD block using an explicit ELD
program, 2.) classrooms with a separate ELD block using ELD material from various sources,
and 3.) classrooms without a separate ELD block where teachers were integrating ELD
instruction with language arts instruction. The author finds that students in the first condition had
significantly higher oral language outcomes than students in the other two conditions.
Second, ELD should explicitly teach elements of English such as grammar, vocabulary,
functions, and conventions. Exposure to a second language can lead to the development of oral
54
fluency, but it may not be enough to master the morphological, syntactic, and pragmatic features
of a language (Spada & Lightbrown, 2008). Explicit attention to elements of English is likely to
facilitate second language acquisition. In their meta-analysis of 79 studies, Norris and Ortega
(2000) find that methods that explicitly attend to features of language were more than twice as
effective as implicit approaches that did not draw students’ attention to specific language
features.
Third, there is preliminary evidence that ELD instruction should address both academic
and conversational
7
language development. Academic language is the language used in
educational settings for the purposes of imparting new information, acquiring new knowledge
and skills, and describing abstract ideas (Chamot & O’Malley, 1994). In other words, academic
language refers to the specialized vocabulary, grammar, and functional skills associated with
academic instruction. Academic language is often contrasted with every day conversational
language which relies on more common and less technical words (Stevens, Butler, and
Castellon-Wellington, 2000). A recent review by the Institutes of Education Sciences failed to
find strong empirical evidence for the teaching of academic English (Gersten et al., 2007).
However, this does not mean that focusing on academic language is a futile endeavor; rather,
there are not enough rigorous, empirical studies about academic language to make conclusions at
the moment.
Fourth, ELD instruction should emphasize listening and speaking but can include reading
and writing. Instruction time outside of the ELD block typically focuses on literacy and content;
thus, the ELD block is an opportunity for teachers to make listening and speaking a priority
(Saunders & Goldenberg, 2013). Research shows that more effective ELD approaches focus on
7
Scholars have used various terms to refer to conversational language. In this study, I use the terms conversational,
social, and informal language interchangeably.
55
oral language more than half of the time (O’Brien, 2007; Saunders, Foorman, & Carlson, 2006).
The development of oral language proficiency is important because of its positive impact on
English reading achievement (Carlisle, Beeman, & Spharim, 1999).
Fifth, ELD instruction should include carefully planned interactive activities. Simply
pairing EBs with more proficient EBs is not sufficient for improving language proficiency.
Saunders and O’Brien (2006) discover that carefully structured tasks need to be created to
facilitate language development. Similarly, Keck et al. (2006) conduct a meta-analysis of
interactive activities and language learning outcomes and find that learning outcomes were
stronger when EBs had to learn language forms that were necessary to the successful completion
of the task. After reviewing quasi-experimental studies, Lyster (2004) arrive at a similar
conclusion—to support language development, interactive tasks should require EBs to use the
targeted language form. In their review of effective reading programs for Spanish-dominant EBs,
Cheung and Slavin (2012) find that effective reading programs make extensive use of
cooperative learning which gives EBs multiple opportunities to interact with peers in meaningful
contexts. Cooperative learning is a structured interactive activity in which small teams of
students work together to improve their understanding of a subject. Though a reading program is
not the same as an ELD program, I include the Cheung and Slavin (2012) study in my review of
ELD as the authors include an ELD program in their review.
Lastly, the probability of sustaining an effective ELD program increases when local
education agencies make it a priority. Similar to other areas of academic achievement, the
direction set by district leadership is likely to influence what is taught in the classroom. Parrish et
al. (2006) find that high-achieving schools with high concentrations of EBs shared various
characteristics that facilitated high academic achievement. One of those common characteristics
is a schoolwide focus on ELD and standards-based instruction.
56
In sum, research shows that ELD is an important component of effective programs for
EBs (Saunders, Goldenberg, & Marcelletti, 2013). ELD curricula can provide districts a
systematic way of covering language standards across the different grade levels.
Textbook Adoption
Commercially published curriculum materials are a ubiquitous resource in schools (Ball
& Cohen, 1999). Publishers, working with a team of educators and researchers, create
instructional materials that are aligned with state standards. Textbook adoption varies from state
to state, with some states creating a list of approved materials. In the United States, there are 19
state-adoption states and 31 open-territory states (Association of American Publishers, 2015). In
open-territory states, the choice of curricular materials is unrestricted by the states and decisions
about funding and timing of adoptions are made locally. In contrast, state-adoption states
exercise greater control over textbook adoption by providing state funding for the purchase of
state-approved materials and by providing guidelines for textbook adoption.
Texas is one of the two largest adoption states. Thus, publishers tailor their books to the
needs and preferences of the Texas State Board of Education and concentrate their selling efforts
in Texas (Bowler, 1978; Stein, Stuen, Carnine, & Long, 2001). In Mike Bowler’s seminal piece
on textbook adoption (1978), he outlines the political nature of the textbook adoption process
and Texas’ national role in this process. Given Texas’ large share of the textbook market,
publishers time the production of new textbooks to coincide with the adoption cycle in Texas
(Stein, Stuen, Carnine, & Long, 2001). However, it should be noted that Texas’ national
influence on textbook adoption may have waned since the adoption of Common Core State
Standards (CCSS). Since the vast majority of the textbook market adopted CCSS, publishers are
now tailoring their materials to align with CCSS. While 42 states and the District of Columbia
have adopted the CCSS, Texas has not (CCSS Initiative, 2015).
57
Texas is a state-adoption state, meaning that the Texas State Board of Education decides
which instructional materials are approved by the state. Each district is given an annual allotment
from the state instructional materials fund for each student enrolled in the district on a date
during the preceding school year specified by the commissioner (Texas Senate Bill 6, 2011).
Each school district and open-enrollment charter in Texas must annually certify that for each
core subject (e.g., English language arts, math, science, and social studies) students are provided
with instructional materials that cover all elements of the Texas Essential Knowledge and Skills
(TEKS). Recently, lawmakers have allowed for greater flexibility in textbook adoption by
passing Texas Senate Bill 6 in 2011. The new legislation allows districts to use their
instructional materials funds to purchase materials that are not on the state adoption list.
Previously, districts could only use their instructional materials funds to purchase state-approved
materials.
Limited research is available to describe how textbook adoption decisions are made at the
district level. In addition, there is little evidence that textbook-adoption decisions made by state
and/or district personnel are informed by evidence on curricular efficacy (Chingos & Whitehurst,
2012). Zeringue et al. (2010) is the only study to date that looks at textbook adoption at the
district level. After conducting in-depth interviews with over 150 curriculum decision-makers in
eight states, Zeringue et al. (2010) find that district leaders have limited information on textbook
efficacy and, therefore, tend to make adoption decisions based on other factors such as teacher
acceptance and advice from neighboring districts. Furthermore, Zeringue et al. (2010) find that
80% of curriculum leaders seek out information about textbook use, effectiveness, and
acceptance of the materials. Thus, the authors conclude, if more research was available regarding
textbook efficacy, curriculum leaders may use this information to make better choices.
58
Though studies on mathematics textbook adoption can shed light on ELD materials
adoption, they might not be entirely applicable to ELD adoption. The vast majority of research
on textbook efficacy and textbook adoption looks at textbooks in core content areas such as
math, English language arts, science, and social studies. Unlike materials in the core subjects,
ELD materials foreground language standards and not content standards. Therefore, the way that
district leaders make decisions about ELD materials may be different than how they decide
which math textbook to purchase. Morita-Mullaney (2016) interviewed district leaders in Indiana
and found that content standards are privileged over language standards, complicating the
implementation of language standards. The privileging of content standards may lead district
leaders to choose ELD materials based on how they address content standards, rather than
language standards. This study extends previous research on textbook adoption and usage by
examining how district leaders make decisions about ELD instructional materials.
English Language Development Curricular Materials in Texas
Texas does not require EB-serving districts to purchase ELD textbooks, which allows for
great variation in ELD adoption patterns. While some districts purchase ELD instructional
materials, other districts use the regular curriculum and ask teachers to modify the instruction as
needed. The variation in ELD curricula was further increased by the passage of Senate Bill 6 in
2011 which, as noted previously, allowed districts to use state funds to purchase non-state-
approved materials and online materials. Though the federal government sought to standardize
policies and programs for EBs, it is evident that states and local education agencies have great
flexibility in how they implement programs for EBs.
The 2011-2012 school year was the last ELD textbook adoption cycle in the state of
Texas. During this adoption cycle, the state released a list of six state-approved ELD curricula
for grades K-5. Of the six, one is an online program and five are traditional, textbook curricula.
59
The textbooks vary along several dimensions including number of lessons on forms of English,
opportunities to learn new vocabulary words, and the ratio of listening/speaking activities to
reading/writing activities. Of the five traditional curricula, Rigby On Our Way to English and
National Geographic Reach are the most popular.
With 118 districts adopting Rigby On Our Way to English, it is the most popular ELD
curricula in Texas. Rigby is a registered trademark of Houghton Mifflin Harcourt. On Our Way
to English is marketed as a comprehensive ELD program that focuses on academic language,
vocabulary development, content-based instruction, and oral language. The text is typically
divided into eight units with each unit providing opportunities to engage with fiction and non-
fiction texts. Dr. Robert Marzano, one of the authors of the series, has written extensively about
building academic vocabulary and building prior knowledge. On Our Way To English was first
published in 2011 and is available in a Texas-specific version and a national version.
With 98 districts adopting National Geographic Reach, it is the second most popular
ELD curricula in Texas. Dr. Deborah Short, one of the book’s authors, is the creator of the
Sheltered Instruction Observation Protocol (SIOP), which is a research-based and validated
classroom tool designed to improve instruction for EBs. SIOP was developed through a seven-
year research study (1996-2003) sponsored by the National Center for Research on Education,
Diversity & Excellence and funded by the U.S. Department of Education. Short’s extensive work
on SIOP makes her a leading expert on developing language proficiency amongst EBs. National
Geographic Reach was first published in 2011 and the second edition is set to come in 2017. A
student anthology in the National Geographic Reach series is usually divided into eight units.
Each unit includes multiple opportunities to use language frames, share aloud with a partner,
learn conversational and academic vocabulary, and practice language-learning strategies (e.g.,
predicting, rereading and retelling). The anthology includes a mix of fiction and non-fiction
60
texts. It should be noted that a Texas-specific version of National Geographic Reach does not
exist.
McGraw-Hill Texas Treasure Chest, adopted by 96 districts in Texas, is marketed as a
comprehensive ELD program for EBs in grades K-6 that provides instruction in the four
language domains of listening, speaking, reading, and writing. While the Texas Treasure Chest
series specifically addresses Texas English Language Proficiency Standards, the Treasure Chest
series is a national version that is designed to meet English language proficiency standards across
multiple states.
Adopted by 35 districts in Texas, Pearson Longman Cornerstone provides
comprehensive ELD material for EBs in grades K-5. This series utilizes the instructional
principles of Understanding by Design, which allows students to demonstrate their mastery of
skills through both formal and informal assessments. In addition, the series provides focused
instruction that is designed to accelerate EBs’ language acquisition, vocabulary, reading
comprehension, and oral and written communication skills. Pearson Longman Cornerstone does
not have a Texas-specific version.
After analyzing state instructional materials reports and responses to public information
requests, I find that five districts in Texas use Santillana Spotlight on English. The series is
marketed as a comprehensive, standards-based program designed to help EBs in grades K-6
develop English proficiency and access academic content. Each student anthology contains eight
thematic units designed to develop language, literacy, and content-area vocabulary. There is both
a Texas-specific version of this series as well as a national version.
Alloy Interactive is the publisher of ESL ReadingSmart which is a web-based learning
program designed to accelerate English language development. The online program integrates
language arts standards with English language proficiency standards. Each lesson in ESL
61
ReadingSmart contains activities and reading selections at a variety of English proficiency
levels. The individualized program also has a specific module for newcomers, or students who
have recently immigrated to the United States. According to state reports on instructional
materials adoption, three districts purchased Alloy Interactive during the 2011-2015 school years.
Conceptual Framework: Social Construction of Target Populations
If there is variation in ELD adoption patterns across districts in Texas, then it is likely
that local education agencies exert influence over ELD decision-making. I use Schneider and
Ingram’s (1993) model of the social construction of target populations
8
to understand district
leaders’ decision-making around ELD instructional materials. Schneider and Ingram posit that
the characterization, or social construction, of groups can determine the kinds of benefits or
burdens directed towards them. Social constructions are “stereotypes about particular groups of
people that have been created by politics, culture, socialization, history, the media, literature,
religion and the like” (Schneider & Ingram, 1993, p. 335). The social construction of target
populations framework was developed to help explain why some policies seem to perpetuate
injustice and produce an unequal citizenship (Ingram, Schneider, & deLeon, 2007). In this study,
I use the framework to understand how the social constructions of EBs influence the types of
instructional resources that are directed towards them.
Two-Dimensional Typology
The social construction typology consists of two dimensions (see Figure 2.2). The
horizontal axis shows whether the social construction of the target population is positive or
negative. While positive constructions may portray target populations as deserving, good, or
intelligent, negative constructions portray target groups as undeserving, lazy, or selfish. There
8
The term “target population” is used to identify those groups chosen to receive benefits and
burdens.
62
are numerous evaluative dimensions that can be used to portray target populations. The social
constructions of target populations may change over time depending on environmental factors.
For example, political events may cause positively constructed groups to become construed
negatively. The vertical axis shows whether the power resources of the target population are high
or low. Power is a multidimensional concept that can refer to such things as the ability of the
group to mobilize, wealth, access to decision-making, or the size of the group.
Figure 2. 2. Social Construction of Target Populations (Schneider & Ingram, 2008)
Though labels have been applied to the four corners of the typology, the dimensions
should be thought of as a continuum rather than as discrete categories (Schneider & Ingram,
2008). Advantaged groups, such as the elderly and veterans, are both powerful and positively
63
constructed (see Figure 2.2). Contenders are perceived to be powerful but are negatively
constructed, such as unions and the rich. Though they carry positive constructions, dependents,
such as mothers and children, have little power. Not only are deviants, such as criminals,
negatively constructed, they tend to carry little political power. Policies send messages about the
worthiness of target populations and these messages can become internalized by both the target
population and society as large.
A critical implication is that social constructions can shape policy design which can then
influence identity, attitudes, and political participation of groups. Policy design can have
fundamental social and political consequences on both material welfare and social reputation of
groups (Ingram, Schneider, & deLeon, 2007).
Social Constructions and Policy Design
Scholars have provided evidence that social constructions can influence policy design
choices (Nicholson-Crotty & Meier, 2005). Policy designs, in turn, can shape institutions and
society through the instrumental (resource) effects of the policy and the symbolic (interpretive)
effects (Ingram, Schneider, & deLeon, 2007). Therefore, policy designs can impact public
opinion, the distribution of resources, and the legitimacy of various knowledge systems. During
the policy-designing process, differences in social constructions are taken into account, leading
to variation in the types of policies that are prescribed for different target groups. Policy designs
allocate material resources, structure opportunities for participation, and send messages about the
target population which can affect the participation patterns of the target group and others.
Messages convey whose interests are important, who belongs, and whose voice will be heard.
Policy designs have the potential to reproduce prevailing institutional norms and power
64
relationships; however, they also have the potential to disrupt dominant norms and introduce
change.
One notable example of how perceptions of groups have been translated into policy is the
Comprehensive Crime Control Act of 1984. The debate about the causes of criminality can be
summarized by two dominant schools of thought—the structuralist and culturalist perspectives.
While structuralists argue that historical inequities explain why some groups cannot break the
cycle of poverty and crime (Clark, 1970; Morris, 1994), culturalists insist that the culture of
certain groups is the cause of crime. By the early 1980’s, the culturalist perspective became
popular (Beckett, 1997), particularly through the very public positions of prominent culturalists
like Patrick Moynihan and Ronald Reagan. Culturalists tended to portray criminals as
undeserving of government resources. The changing assumptions about the cause of crime was
coupled with the increasing marginalization of criminals. Criminality became associated with
minority groups (Tonry, 1993). The changing perceptions of criminality led to wide support for
the Comprehensive Crime Control Act. In 1984, the legislation was approved by an
overwhelming majority in both the House of Representatives (316-91) and the Senate (78-11).
The bill allocated considerable burdens which include, but are not limited to: increasing federal
penalties for marijuana possession, eliminating discretionary parole release, and expanding the
government’s ability to seize the assets of certain offenders (Nicholson-Crotty & Meier, 2005).
As illustrated above, Schneider and Ingram’s (1993) theory of social construction of
target populations can be used to predict policy designs (see Figure 2.3). The degree to which
groups are negatively or positively constructed can determine the allocation of benefits and
burdens. While beneficial policy will be oversubscribed to groups with significant political
power, burdensome policies will go mainly to groups with little power or those who are
negatively constructed. Beneficial policies confer rights, subsidies, and resources that are desired
65
by the group. The public is not likely to object to the creation of beneficial policies as they are
being directed towards deserving groups. In contrast, burdensome policies limit the freedoms of
a target group. Policy burdens can come in the form of sanctions, penalties, fines, prison terms,
etc.
Construction
Positive Negative
Strong
Advantaged
Prediction:
Beneficial
policies
oversubscribed
Contender
Prediction:
Complex,
confusing
policies; hidden
benefits; empty
burdens
Weak
Dependent
Prediction:
Symbolic and/or
inadequate
policies are
subscribed
Deviant
Prediction:
Beneficial
policies
undersubscribed;
burdens
subscribed
Figure 2. 3. Schneider and Ingram’s Model: Policy Design Predictions (1993)
The theory of social construction predicts that advantaged groups, due to their positive
construction and access to power, are likely to receive beneficial policies such as tax credits,
material resources, training, and free information. If burdens are allocated to advantaged groups,
they will often be voluntary rather than mandated. Policy designs for advantaged groups will
include opportunities for the target group to voice concerns.
Contenders are likely to receive beneficial policy, due to their substantial access to
power, but these benefits may be hard to identify. Benefits to contenders are difficult to pinpoint
66
because policymakers do not want to openly do good things for target groups who are perceived
to be undeserving (Ingram, Schneider, & deLeon, 2007). Contenders may also receive burdens
but these burdens are difficult to enforce due to the political power of the group. Alternatively,
policymakers might allow avenues for contenders to avoid these burdens.
Dependents, positively constructed groups with little power, tend to receive beneficial
policies that are inadequate or limited by funding. Since dependents have less political power
than advantaged groups, dependents are perceived to be less deserving of actual investments than
advantaged groups. When benefits are provided to dependents, they tend to be heavy on rhetoric
and low on material resources. Dependents may also receive symbolic policies that allow policy
makers to show great concern but relieve them of the need to allocate resources (Schneider &
Ingram, 1993).
Deviants receive a disproportionate share of burdens due to their negative construction
and lack of access to power. Some scholars have viewed deviants as a permanent underclass in
the United States who are blamed for society’s ills (Nicholsen-Crotty & Meier, 2005).
Policymakers gain political capital from punishing groups that are perceived to be undeserving
by the public.
As noted earlier, this four-fold classification is not meant to produce discrete categories.
As a result, EBs may reside in the middle of two categories, complicating the prediction of
policy designs. For instance, it is difficult to predict ELD policy designs for districts that believe
that EBs are simultaneously deserving and undeserving. For these groups with mixed views on
EBs, I predict that ELD policy designs will be complex and confusing. The existence of
divergent characterizations of EBs may push policymakers to adopt a variety of policies to
satisfy the multitude of perspectives.
67
Social constructions can vary over time and across contexts. Target groups can reside in
multiple categories at once and can move between categories across time. It is possible for target
populations to lack any kind of positive or negative construction. It is also possible for a target
group to be perceived differently by various audiences. For example, undocumented immigrants
may be viewed as the backbone of America’s workforce while others perceive them to be
lawbreakers who take jobs away from U.S. citizens (Newton, 2005). Though there are many
unanswered questions about social construction of target populations, the current research
literature validates the proposition that the social construction of target groups affects the way
they are treated in policy design.
In sum, a common pattern across many policy arenas is that advantaged groups receive
benefits, deviants are allocated burdens, contenders are the recipients of hidden benefits and
empty burdens, and dependents receive inadequate help.
Operationalizing Social Constructions
Schneider and Ingram (1993) argue that social constructions of target populations are
measurable phenomena. Data from texts, guidelines, speeches, media coverage, and interviews
can be used to understand how target populations are perceived. In this study, I attempt to
categorize the social construction of EBs using data from interviews with district leaders, district
websites, relevant district documents, and the TEA online data portal.
In my study, I operationalize power as the share of EBs in the district. EBs are considered
a “strong” group if they comprise at least 17% of the district’s population. I choose 17% as the
cutoff
9
because, during the 2010-2011 school, EBs comprised 17% of students in Texas’ public
schools. As mentioned previously, I focus on the 2010-2011 school year because textbook
9
This cutoff may seem arbitrary and the limitations of choosing a cutoff are discussed in the
limitations section.
68
adoption decisions are made in the year before the textbooks are implemented which is 2011-
2012 for ELD curricula. The greater the share of EBs, the more likely they are to be visible and
command the attention of policymakers and district leaders. In addition, as the number of EBs
grows, the less they can be ignored given the accountability measures that exist for state
standardized testing in Texas. In the state of Texas, EBs are one of several subgroups for which
adequate yearly progress (AYP) measures are calculated. Local education agencies that fail to
meet AYP for two consecutive years are subject to corrective action (TEA, 2017). For EBs to be
included in AYP participation measures, they must meet a certain size or percentage
requirement.
10
Thus, as the size of EBs in a district surpasses a certain threshold, they are more
likely to demand the attention of district leaders, who presumably want to avoid corrective action
by the state.
Schneider and Ingram (1993) acknowledge that it is possible for beneficial policies to be
prescribed to powerless, negatively constructed groups through court actions and mandates.
Since Castañeda v. Pickard (1981) mandated that programs for EBs be theoretically sound and
empirically tested, it is possible that districts that negatively construct EBs have prescribed
beneficial policies for EBs. This study will look to see if court mandates are enough to overcome
negative constructions of EBs.
Data
Data for the study come from four main sources: Texas Education Agency (TEA), United
States Census, interviews with district leaders in Texas, and district documents. Texas serves
10
For EBs’ AYP participation measures to be evaluated for AYP, a district or campus must have
either: A.) 50 or more students in the group enrolled on the test date for the subject, and the
student group must comprise at least 10 percent of all students enrolled on the test date, or B.)
200 or more students in the group enrolled on the test date, even if that group represents less than
10 percent of all students enrolled on the test date (TEA, 2008).
69
roughly 5,000,000 students who are spread across approximately 1,250 public school districts.
The racial composition of students in Texas’ public schools is as follows: 52% Hispanic, 30%
White, 13% Black, 4% Asian, and 2% multiracial (Texas Education Agency, 2014). Roughly
17% of students in Texas’ public schools are identified as EBs (Texas Education Agency, 2011).
Texas is home to roughly 830,000 EBs, the second highest number of EBs in the United States.
First, TEA provides information on district enrollment, demographics, expenditures,
achievement outcomes, and teacher characteristics. In addition, TEA collects information on
instructional materials allotment (IMAT), which is publicly available through Freedom of
Information Law (FOIL) request. I was able to obtain ELD textbook purchases for the years
2011-2015. The following textbook information is available for each district: publisher, textbook
title, subject, grades used, unit price, quantity of units purchased, and adoption year. I find 418
districts on the IMAT list for years 2011-2015, which leaves 810 districts with missing
information on ELD textbook adoption.
Since the IMAT report only captures purchases made using state funds, it does not
capture instructional materials that were purchased using local funds nor instructional materials
that were purchased prior to 2011. Therefore, it cannot be assumed that districts that are not on
the 2011-2015 IMAT list did not purchase ELD instructional materials. The other 810 districts
might have purchased ELD textbooks prior to 201l. Alternatively, these 810 districts may not
have purchased any ELD instructional materials. To obtain information on ELD instructional
materials for the remaining 810 districts, I submitted Public Information Requests to each district
with missing ELD textbook information and received responses from 743 districts, yielding an
overall response rate of 94%. In other words, I was able to ascertain ELD adoption patterns for
94% of the 1,133 EB-serving districts in Texas. Districts that did not respond to the FOIL request
70
(N=67) are smaller, have higher shares of EBs, and have higher shares of white students. Table
2.2 summarizes the distribution of ELD adoption patterns in Texas.
Table 2. 2. ELD Adoption Patterns in Texas for Kindergarten (N=1,133)
Districts
Rigby On Our Way to English 118
National Geographic Reach 98
McGraw-Hill Texas Treasure Chest 96
Pearson Longman Cornerstone 35
Santillana Spotlight on English 5
Alloy (online) 3
Non-State-Adopted Book 63
No ELD textbook adoption 644
Missing 67
Both 4
Second, district-level Census data was obtained to provide more robust measures of
district characteristics such as income. The TEA provides information on subsidized meal
eligibility which is a poor proxy for economic disadvantage (Michelmore & Dynarski, 2017).
Thus, I supplement TEA data with Census data on median household income. I also include a
measure for the percentage of bachelor degree holders as educational level may influence
educational processes such as textbook adoption.
Third, interviews with 28 district leaders in Texas provide rich information about ELD
policies and practices. The districts selected for interviews represent variation along two
dimensions-- percentage of EBs (e.g., low and high) and ELD textbook adoption pattern (state-
adopted text, off list text, no ELD text)—yielding 6 total cells. I use stratified sampling to
identify districts. Table 2.3 summarizes the number of districts in each cell and how many
district leaders agreed to or refused to participate. Each district was placed into one of the six
cells and then alphabetized. Districts were contacted based on their place on the alphabetized list.
If district leaders did not respond to emails or phone calls, then I moved onto the next district on
71
the alphabetized list. I continued interviewing until I reached theoretical saturation in each cell
(Miles & Huberman, 1994). Lincoln and Guba (1985) posit that theoretical saturation is reached
when interviews with different sources yield similar information. However, I was only able to
interview one district in each of the off-list adoption categories as many of the districts in those
categories did not have a website, making it difficult to find contact information. Interviews
lasted approximately 45 minutes and were audio recorded and transcribed. Interview data were
analyzed using NVivo software.
Table 2. 3. Sampling Plan for Interview with District Leaders in Texas (N=28)
Percentage of EBs
Less than 17% Greater or equal to 17%
No ELD Text Number of completed interviews: 7
Refused to participate/no response: 119
Total districts in cell: 574
Number of completed interviews: 5
Refused to participate/no response: 49
Total districts in cell: 69
State-Adopted Text Number of completed interviews: 10
Refused to participate/no response: 109
Total districts in cell: 291
Number of completed interviews: 4
Refused to participate/no response: 54
Total districts in cell: 69
Off-list Text Number of completed interviews: 1
Refused to participate/no response: 19
Total districts in cell: 55
Number of completed interviews: 1
Refused to participate/no response: 3
Total districts in cell: 8
I used a structured open-ended interviewing protocol to gather information about ELD
practices and instructional materials (see appendix). A structured interview protocol is
appropriate for this study because prior research (Zeringue et al., 2010) has already identified
factors relevant to textbook adoption. I employed an open-ended interviewing method to allow
participants the opportunity to discuss topics that are not included in the protocol but are relevant
to ELD adoption. The open-ended design also allowed me the chance to probe on newly
emerging themes.
72
I focused on district leaders because they play a crucial role in deciding which ELD
textbook to adopt for their particular district. In each participating district, I identified the person
primarily responsible for textbook adoptions. The content of these interviews included questions
about district policies regarding EBs, processes for textbook adoption, professional development,
implementation of English language proficiency standards, programs for EBs, and beliefs about
second language acquisition. These questions were tied to the conceptual framework as they
provided ample information about how districts socially construct EBs. For example, questions
about how district leaders choose certain bilingual/ESL programs reveal the extent to which they
believe a students’ native language is an asset. Districts that utilize an additive bilingual program
such as two-way dual immersion instead of a subtractive program may be more likely to have
positive constructions of EBs. The differences between additive and subtractive bilingual
programs are further discussed in the methods section. Questions about district support for
teachers of EBs can provide information on where district leaders are locating the problem of EB
underachievement. If district leaders are allocating resources to train teachers of EBs, this action
suggests that district leaders believe that one contributing factor to EB underachievement is
institutions that inadequately prepare teachers.
Lastly, district policy documents such as program model description were collected from
schools and/or districts to understand the local policy context. The majority of districts also had a
website containing information about the districts’ bilingual or ESL department. The use of
multiple sources—interview data, district documents, district website—can enhance the
trustworthiness of analysis, which I discuss in the methods section (Miles, Huberman, &
Saldaña, 2013).
Methods
73
I used a mixed-methods sequential explanatory design to study the adoption of ELD
instructional materials. In the first phase, quantitative data were collected and analyzed. The
second, qualitative, phase was used to explain the statistical results that were obtained in the
previous stage (Creswell et al., 2003). A mixed-methods approach is particularly useful when
unexpected results emerge from a quantitative study (Morse, 1991). While averages and trend
lines can hide unique cases, qualitative methods can shed light on those individuals whose
experiences do not reflect the typical case. A fundamental principle of mixed methods research is
that the qualitative and quantitative methods should be strategically mixed to produce an overall
design with complementary strengths and non-overlapping weaknesses (Brewer & Hunter, 1989;
Johnson & Turner, 2003; Webb, Campbell, Schwartz, Sechrest, & Grove, 1981).
Quantitative Data Analysis
To answer the first research question, I use a multinomial logistic regression to test the
association of district characteristics with ELD textbook adoption. There are three outcome
variables for this model: state-approved ELD materials, non-state-approved ELD materials, and
no ELD materials. Following prior research on textbook selection (Bhatt & Koedel, 2012; Le &
Polikoff, 2017), I include covariates related to enrollment, demographics, district expenditures,
education level, and urbanicity. The multinomial logistic regression model is as follows:
icst
(1)
Where
is the percentage of EBs in district d,
is the percentage of white
students in the district,
is log of district enrollment in the district,
is log
of per pupil expenditure in the district,
the percentage of expenditures that is spent
74
on instructional materials in the district,
is average district scores on the reading state
test,
is the median income for the district,
is the percentage of people
with a Bachelor’s degree or higher, and
is a categorical variable taken from the
NCES database that represents the geographic location of the district. Following Nolle et al.
(2007), I collapse the urbanicity variable into four main groups: city, town, rural, and suburban.
Districts in rural areas serve as the reference group. With regards to teacher characteristics,
represents the average years of teacher experience for the district and
is the percentage of white teachers at district d. As the Texas Education
Agency called for new ELD materials to be implemented during the 2011-2012 school year, I
use district variables from the 2010-2011 school year in my analyses. Typically, school districts
make decisions about textbooks in the year prior to implementation; thus, the district
characteristics most relevant to decision-making are the district characteristics from the school
year in which adoption decisions are made. Districts that did not adopt any ELD textbook will
serve as the reference category as they are the largest category.
Districts that adopted both state-approved and off-list ELD materials (N=4) were dropped
from the analysis as they comprise a small group. Any inferences that are made about such a
small group may be unreliable.
Qualitative Data Collection
To answer the second research question about the link between social construction of
EBs and ELD policy design, I use a fairly structured qualitative design since prior research has
already identified constructs that are relevant to textbook adoption and implementation.
Structured designs provide clarity and focus, and researchers can be more selective with the
collection of data compared to looser designs (Miles & Huberman, 1994).
75
Social constructions are derived from a variety of sources (Ingram, Schneider, & deLeon,
2007). In my study, I use interview data and district documents to garner information about how
districts are perceiving EBs. Specifically, I look at district goals, district’s choice of a bilingual
program, district policy on ESL/bilingual certification, and transcripts from interviews with
district leaders.
First, district goals from documents and interviews provide information on how EBs are
framed. While some districts state biliteracy as an explicit goal for EBs, other districts emphasize
English proficiency. These two different goals send different messages about the value of EBs’
linguistic abilities. Districts that strive to develop multilingualism explicitly convey the
importance of students’ native language. Since district goals alone are not sufficient to
understand how districts are constructing EBs, I also consider the type program that districts
choose for EBs.
Second, a district’s choice of bilingual program—additive vs. subtractive—also reveals
how districts are perceiving EBs. According to the Texas Education Code 29.053c, each district
must provide bilingual or special language programs if there are 20 or more EB students in one
grade level. However, the state does not mandate what type of bilingual program must be
provided. There are four main types of bilingual programs that districts can choose from: early-
exit transitional bilingual, late-exit transitional bilingual, one-way dual language, and two-way
dual language. The goal of a transitional bilingual program is to use students’ native language to
transition EBs into English as quickly as possible. Transitional bilingual programs are considered
to be subtractive bilingual programs because the second language is added at the expense of the
first (Cummins, 1994). As students progress in school, they spend more of their school day in
English until their native language is completely phased out. While early-exit transitional
programs aim to transition by 3
rd
grade, late-exit transitional programs allow EBs to spend more
76
time in their native language, with most students transitioning fully into English by 6
th
grade. In
theory, dual language programs strive to create bilingual and biliterate students. One-way dual
language programs exclusively enroll linguistic minorities whose first language is not English. In
contrast, two-way dual language programs strive to enroll equal shares of native English
speakers and linguistic minorities. By doing so, one group serves as a language model for the
other group. Dual language programs are viewed as additive bilingual programs because the
native language continues to be developed while the second language is added; the native
language is not phased out (Cummins, 1994). A district’s choice of bilingual program sends a
message about the value of students’ native languages.
Third, a district’s policy on ESL certification can provide information on who districts
believe are responsible for addressing the opportunity gap between EBs and non-EBs. Districts
that invest in systematic policies that increase the number of ESL-certified teachers locate the
problem of EB underachievement in institutions instead of in individuals. A district that views
systematic training of teachers of EBs as a necessary district goal may have positive
constructions of EBs as they are investing resources into EBs’ education. These districts may
realize that the linguistic needs of EBs are different than the needs of non-EBs; thus, districts
recognize that they must improve the capacity of teachers to work with linguistically and
culturally diverse populations. Furthermore, a district that mandates that teachers must be ESL-
certified is also taking a strong stance about how EBs will be treated within the district.
Providing mandatory training for teachers of EBs indicates that the district is locating the
problem of EB underachievement in school systems and not in EBs.
Lastly, the way that district leaders discuss EBs broadly can provide useful information
about the social construction of EBs. As mentioned previously, district leaders were selected
based on two dimensions—ELD adoption pattern and percentage of EBs—which led to six cells.
77
Each EB-serving district in Texas was placed into one of these six cells. Districts in Texas were
alphabetized in each cell, and districts were contacted based on the order in the list. If the district
had a bilingual/ESL department, then the head of that department was contacted via email or
telephone. If no such department existed, then the director of curriculum and instruction was
invited to participate. Follow-up emails and calls were sent to non-responders. If non-responders
did not reply within three weeks, then the next district on the list was contacted. Approximately
7.3% of district leaders who were contacted agreed to participate in the interview (see Table 2.3).
The low response rate raises concerns about potential differences between responders and non-
responders. These concerns are discussed in the limitations section.
To triangulate the data, I use information from four sources to determine a districts’
social construction of EBs: district documents, interviews with district leaders, district’s choice
of bilingual program, and district’s ESL certification policy. By looking at all four areas
simultaneously, I can arrive at a fuller understanding of how districts are constructing the target
population. The border between various social constructions is blurry. On specific issues,
districts may fit better into a group other than the one in which they were ultimately placed.
Qualitative data analysis proceeded through multiple and multi-level reviews. In the
initial cycle of review, which required at least three readings of the full transcript, I used
structural codes, which entail little interpretation. Structural coding “acts as a labeling and
indexing device, allowing researchers to quickly access data likely to be relevant to a particular
analysis from a larger data set” (Namey et al., 2008, p. 141). Though interviews covered a range
of topics, including EB program model, I subjected the entire transcript to this initial coding,
since district leaders spoke about instructional materials in response to questions not directly
about instructional materials (e.g., bilingual model). The first cycle of coding revealed 52
78
structural codes and subcodes which were further categorized and analyzed in the second stage
of review.
Table 2. 4. List of Codes from First Round of Coding
Codes from first round of coding
ESL Certification
Adoption Process
o Committee Selection
o Publisher Presentation
o Rubric Timing
District Characteristics
District Leader Characteristics
o District leader beliefs
o District Leader Experience
English language Proficiency Standards
Factors for textbook adoption
o Achievement of EBs
o Addresses multiple language backgrounds
o Advocacy by district leader
o Chance encounters
o Committee review
o Content and language
o Cost
o Cultural relevance
o Differentiation
o District budget
o District leader beliefs about second language acquisition
o District or school enrollment
o Ease of implementation
o EB enrollment
o Home school connection
o Leveled readers
o Professional development
o Program model
o Reviews or research
o Standards
o Student engagement
o Teacher acceptance
o Technology
o Urbanicity
o Visuals and colors
Future ESL adoption
Goal of ELD Instruction
Information that District Leader Would like to have
79
Materials Used for EBs
Program Model for EBs
Quality of Materials for EBs
Questionable practices
o Deficit framing
o Isolating EBs
o No ELD block
o Non-EB materials used for EB purposes
o Rosetta Stone
Construction of EBs
o Positive
o Negative
o Hybrid
80
Table 2. 5. List of Codes from Second Round of Coding
Codes from second round of coding
Positive Construction of EBs
o Native language as an asset
o Native culture as an asset
o Standardization of programs for EBs
o leadership buy-in
o Systematic training of teachers of EBs
Reimbursement for ESL certification
Stipend for ESL certification
Negative construction of EBs
o Leadership lacks training to work with EBs
o Limited understanding of ELD
o Limited understanding of various bilingual programs
o Nonperformative speech acts
o Anything goes
o Conflating EB status with IEP
o Conflating ELD materials with materials for struggling readers
o Conflating ELD materials with non-English materials
o ELD as “just good teaching”
In the second cycle of coding, the primary goal is to create categorical, thematic, and
conceptual organization from the array of codes that emerged from first cycle coding (Saldaña,
2013). I used pattern coding in the later stages of data analysis to group codes into a smaller
number of themes that were tied to the conceptual framework. For example, codes that emerged
from the second cycle of coding were related to how EBs were socially constructed (e.g.,
positive, negative, hybrid). Coded data were analyzed using matrix displays (Miles, Huberman &
Saldaña, 2013) to identify patterns, trends, and disconfirming cases. Matrix displays included
information on district demographics, ELD adoption, type of bilingual program, EB-related
programs, and overall construction of EBs. Tables 2.4 and 2.5 list all of the codes for each round
of coding.
81
Miles and Huberman (1994) recommend memoing while coding to record emerging ideas
and patterns. Glaser (1978) defines a memo as the “theorizing write-up of ideas about codes and
their relationships as they strike the analyst while coding” (p. 83-4). Memos help researchers
move from the descriptive to the conceptual level and also helps to inform subsequent
interviews. I memoed about each interview and about my preliminary analyses. At times, my
reflections led me to probe more on certain issues during an interview because previously
collected interview data revealed that there may be other environmental factors influencing ELD
textbook adoption that I had not considered. For example, initial interviews revealed that district
support for ESL certification played a role in how districts structured ELD. Therefore, in
subsequent interview I included a question specifically about ESL certification.
Trustworthiness. Trustworthiness, or the extent to which the results of a study are
believable, is established through the rigor of the study (Merriam & Tisdell, 2009). Lincoln and
Guba (1985) recommend that qualitative researchers use the following criteria to establish
trustworthiness: transferability, dependability, and credibility.
Transferability. Transferability is the extent to which findings can be applied to other
contexts. To enhance the possibility that results can be transferred to another setting, researchers
can pay special attention to the selection of the study sample (Merriam & Tisdell, 2009). As
noted previously, I use stratified sampling to ensure that districts represent variation along two
dimensions—ELD textbook adoption and percentage of EBs. In the final sample, the percentage
of EBs range from 3% to 33% and ELD adoption patterns range from no adoption to the
adoption of multiple ELD curricula. Sampling along these two dimensions led to great variation
in other district demographics. For example, district enrollment ranged from 500 to 60,000 and
the number of EBs in each district ranged from 20 to 10,000. The wide range in district
demographics allows the researcher to arrive at results that have a greater range of application
82
(Patton, 2015). That is, including a variety of sites enables more readers to apply the study’s
findings to their own context.
Consistency. Lincoln and Guba (1985) conceptualize reliability in qualitative research as
consistency. Consistency entails that, given the data collected, an outside researcher will agree
that the results make sense. That is, the results are consistent with the data collected. There are
several ways to arrive at consistency. Data triangulation is one such technique and involves the
use of different data sources to capture multiple dimensions of the same phenomenon (Johnson
& Christensen, 2012). By conducting interviews and analyzing state and district documents, I
can increase the likelihood that my findings represent a more comprehensive and accurate view
of ELD materials adoption.
Credibility. There are several strategies that researchers can employ to ensure that their
findings are credible. One way to develop credibility is to use research methods that are well
established in the field (Shenton, 2004). I revised an interview protocol used in a prior study
about textbook adoption (Campbell & Polikoff, 2017). I edited the questions so that they were
relevant to ELD adoption instead of math textbook adoption. In addition, I added questions about
bilingual program models and English language proficiency standards since they may influence
how districts structure ELD programs and practices. Furthermore, I consulted prior research on
textbook adoption (Zeringue et al., 2010) to ensure that my interview protocol captured all
relevant aspects of the phenomenon under study.
Results
Research Question 1: To what extent, if any, do ELD adoption patterns vary by district
characteristics?
83
Descriptive statistics by ELD adoption pattern. Across the 1,066 districts for which I
have obtained ELD adoption patterns, 644 (60%) are non-adopters, 355 (34%) are state-adopters,
63 (6%) are off-list adopters, and 4 (0.4%) adopted both state-approved and off-list materials
11
.
As mentioned previously, the adoption of ELD materials is not required in the state of Texas so
districts have a choice of purchasing or not purchasing an ELD curricula. In the absence of a
state mandate, I find that the majority (60%) of EB-serving districts do not adopt any ELD
materials. However, of those EB-serving districts that do purchase ELD materials, 85% of them
choose to purchase something from the state-approved list, indicating that the Texas State Board
of Education’s approval still plays an important role in textbook adoption. Even though Texas
Senate Bill 6 (2011) allows districts to use state funds to purchase off-list materials, the majority
of districts do not take advantage of this added flexibility. This is a striking finding as it seems
that, when districts are given choices, they take advantage of certain choices but not all.
11
Again, the four districts that adopted both types of materials are dropped from the analysis as
they are a small group and any inferences that can be made are unreliable.
84
Table 2. 6. Descriptive Characteristics of Non-Adopters, State-Approved Adopters, Off-List
Adopters.
Non-Adopters
State-Approved
adopters
Off-List Adopters
Mean SD Mean SD Mean SD
District Characteristics
Enrollment 2,060 6,798 9,345*** 19,442 1,487 2,805
Percent White 0.56 0.28 0.44*** 0.28 0.58 0.24
Percent Black 0.10 0.16 0.11 0.15 0.07 0.08
Percent Hispanic 0.32 0.26 0.43*** 0.29 0.33 0.23
Percent Native 0.01 0.01 0.00* 0.01 0.01 0.00
Percent API 0.01 0.02 0.02*** 0.04 0.01 0.06
Percent Economically
Disadvantaged 0.59 0.20 0.60 0.20 0.56 0.17
Percent SPED 0.10 0.05 0.09* 0.03 0.1 0.02
Percent EB 0.07 0.10 0.11*** 0.11 0.07 0.08
Per pupil expenditure 13,126 6,392 12,743 4,727 15,141* 8,075
Percent Instructional
Expenditures 0.61 0.05 0.63*** 0.04 0.6 0.04
City 0.12 0.32 0.18* 0.38 0.05 0.21
Town 0.17 0.38 0.21 0.41 0.14 0.35
Suburban 0.05 0.21 0.16*** 0.37 0.05 0.21
Rural 0.67 0.47 0.45*** 0.5 0.76 0.43
District Outcomes
TELPAS 2.6 0.7 2.6 0.5 2.7 0.6
Reading Scores 603 37 602 31 603 32
Reading Scores for EBs 561 39 568 36 560 35
Math Scores 582 37 584 32 579 31
Math Scores for EBs 557 37 566** 35 552 36
District-Area
Characteristics (Census)
Median Income 46,364 13,511 48,285 17,259 47,614 14,527
Percent with Bachelors
Degree or Higher 0.17 0.09 0.19*** 0.11 0.18 0.08
Percent Foreign Born 0.06 0.07 0.11*** 0.09 0.06 0.04
Percent Speakers of Non-
English Languages 0.18 0.18 0.29*** 0.23 0.18 0.15
Note. Non-adopters serve as the comparison group.
*p <0.05. **p<0.01. ***p<0.001.
Running head: VARIATION IN ENGLISH LANGUAGE DEVELOPMENT MATERIALS 85
85
Table 2. 7. Descriptive Statistics of District Characteristics by Publisher
Rigby On Our
Way to English
National
Geographic Reach
McGraw-Hill
Treasure Chest
Pearson
Longman
Cornerstone
Santillana
Spotlight on
English
Harcourt
Moving into
English
Rosetta Stone
Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD
District Characteristics
Enrollment 7,780 13,111 19,552*** 31,223 3,940* 6,899 3,122* 5,869 2,053 1,605 732* 619 2,219 2,845
Percent White 0.46 0.28 0.35** 0.28 0.48 0.28 0.49 0.26 0.44 0.32 0.57 0.27 0.57 0.26
Percent Black 0.12 0.15 0.12 0.14 0.1 0.14 0.13 0.18 0.07 0.05 0.08 0.11 0.07 0.05
Percent Hispanic 0.4 0.28 0.51** 0.29 0.4 0.29 0.37 0.24 0.47 0.33 0.34 0.24 0.35 0.27
Percent Native 0.01 0.01 0 0 0 0 0 0 0.01 0 0 0 0.01 0
Percent API 0.02 0.04 0.02 0.03 0.02 0.04 0.01 0.02 0.02 0.02 0 0 0.01 0.01
Percent Economically
Disadvantaged 0.6 0.17 0.62 0.22 0.57 0.22 0.61 0.19 0.63 0.22 0.58 0.19 0.57 0.18
Percent SPED 0.09 0.04 0.09 0.02 0.09 0.03 0.1 0.03 0.09 0.02 0.1 0.02 0.1 0.02
Percent EB 0.1 0.09 0.16*** 0.13 0.08 0.09 0.09 0.12 0.1 0.09 0.07 0.09 0.07 0.06
Per pupil expenditure 12,345 3,786 12,115 4,244 13,670 6,086 13,138 4,616 10,453 4,312 14,408 5,728 18,004*** 12386
Percent Instructional
Expenditures 0.62 0.04 0.64** 0.04 0.62 0.04 0.62 0.05 0.67* 0.09 0.60* 0.04 0.59* 0.05
City 0.18 0.38 0.27 0.44 0.08* 0.28 0.11 0.32 0.60* 0.55 0.10 0.30 0.08 0.28
Town 0.25 0.43 0.12 0.33 0.24 0.43 0.26 0.44 0.00 0.00 0.14 0.36 0.00 0.00
Suburban 0.14 0.34 0.29** 0.45 0.13 0.33 0.06 0.24 0.00 0.00 0.00 0.00 0.00 0.00
Rural 0.44 0.50 0.33 0.47 0.55 0.50 0.57 0.50 0.40 0.55 0.76** 0.44 0.92*** 0.28
Urbanicity 6.8 3.6 5.2 3.9 7.7 3.1 7.7 3.1 4.6 5.5 9 3.1 9.5 2.4
District Outcomes
TELPAS 2.6 0.4 2.6 0.4 2.5 0.6 2.7 0.7 2.7 0.4 2.7 0.4 2.9 0.4
TELPAS (std)
Reading Scores 602 31 604 26 602 35 596 34 604 22 610 36 610 19
VARIATION IN ENGLISH LANGUAGE DEVELOPMENT MATERIALS 86
86
Reading Scores for EBs 568 35 574 28 559 43 564 44 573 21 557 40 566 29
Math Scores 584 30 587 28 582 39 576 31 586 47 588 34 579 21
Math Scores for EBs 568 34 575 29 553* 38 558 40 569 40 564 38 550 36
District-Area
Characteristics
(Census)
Median Income 46,931 14,924 49,129 17,423 49,895 20,495 47,501 16,280 41,448 5,440 46,147 11,689 44,673 17,179
Percent with Bachelors
Degree or Higher 0.18 0.09 0.22** 0.12 0.19 0.13 0.18 0.08 0.14 0.03 0.17 0.07 0.17 0.09
Percent Foreign Born 0.09 0.07 0.15*** 0.1 0.09 0.07 0.07 0.07 0.05 0.01 0.06* 0.04 0.06 0.05
Percent Speakers of
Non-English Languages 0.26 0.22 0.36** 0.26 0.26 0.21 0.20 0.17 0.18 0.17 0.18 0.15 0.2 0.22
Note. T-tests are done with Rigby On Our Way to English as the comparison
Running head: ENGLISH LANGUAGE DEVELOPMENT CURRICULUM EFFECTS
87
87
Though the majority of EB-serving districts in Texas did not purchase any ELD
instructional materials during the 2011-2012 adoption cycle, the majority of EBs are located in
districts that purchase state-approved ELD materials. Of the roughly 830,000 EBs in the state of
Texas, 76% are in districts that purchase state-approved materials, 22% are in districts that do
not purchase any ELD materials, and about 1% attend districts that purchase off-list ELD
materials. While the average district size for state-adopters is 9,300, non-adopting districts
average 2,100 students. Of the 63 districts in the study that adopted non-state-approved ELD
materials, 76% of these districts are located in rural areas, as defined by the National Center for
Education Statistics (2016).
There are several demographic differences between these three groups (see Table 2.6).
Significance testing was conducted between non-adopters and each of the two other groups.
Results from hypothesis testing are reported in Table 2.6. State-adopters average 11% EBs while
non-adopters and off-list adopters both average 7%
12
. While adopters of state-approved materials
are predominantly minority (56% non-white), the other districts are majority white. Interestingly,
districts that adopt state-approved ELD materials have the highest share of economically
disadvantaged students (60%) but also have the highest median income ($48,300) of the three
groups. According to data from the 2010 Census, state-adopters have a higher share of speakers
of languages other than English (29%) and a higher percentage of foreign-born individuals
(11%) compared to the other two groups. Furthermore, state-adopters have the highest share of
individuals with a bachelors degree (19%) and are more likely to be located in more urban areas.
12
These numbers may seem perplexing given that 17% of students in Texas are EBs. However, these numbers can
be explained by the high degree of linguistic segregation. There are 241 districts with high percentages of EBs
(greater than 17%) and these districts serve 71% of the EBs in the state of Texas. In other words, 20% of schools in
Texas serve 71% of the EB population.
88
There are also notable differences between these three groups in terms of student
spending and EB achievement. Off-list adopters spend $15,100 per pupil while non-adopters
spend significantly much less per student, $13,100. One possible explanation for the difference
in per pupil funding is that off-list adopters are more likely to be small, rural districts, and Texas
provides additional funding for sparsely populated districts (TEA, 2013). State-adopters spend a
significantly greater share of their budget on instructional expenditures compared to non-
adopters. EBs in districts that adopt state-approved ELD materials score significantly higher on
the state math assessment compared to EBs in non-adopting districts.
Overall, state-adopters are located in more urban areas, have larger enrollments, higher
share of EBs, and higher shares of racial minorities compared to the other two groups.
Descriptive statistics by publisher. Next, I take a closer look at district characteristics
by disaggregating the data by publisher (see Table 2.7). I include seven publishers in this
analysis: five state-approved publishers and two off-list publishers. I include the top five state-
approved publishers because they comprise over 80% of the ELD purchases in Texas. I add the
top two off-list publishers—Moving into English and Rosetta Stone—to the analysis to gain a
deeper understanding of what kinds of districts adopt off-list materials. Significance testing was
conducted between Rigby On Our Way to English and each of the other publishers. Results from
hypothesis testing are reported in Table 2.7, with Rigby On Our Way to English serving as the
comparison.
There are several notable differences in district characteristics across these seven
publishers. There is great variation in district size across the seven publishers. Districts that adopt
National Geographic Reach average about 19,600 students while districts that adopt Harcourt
Moving into English are significantly smaller, with about 700 students. Test scores for EBs on
89
the state standardized test also vary greatly by publisher. While districts that adopt National
Geographic Reach have the highest math scores for their EBs, districts that adopt Rosetta Stone
have the lowest math scores for EBs. Interestingly, TELPAS scores are highest among districts
that adopt Rosetta Stone. However, the numbers in this cell should be interpreted with caution as
there are only 13 districts in this group and only 6 of those districts had sufficient number of EBs
to report TELPAS scores. With regards to reading scores for EBs, districts that adopt National
Geographic Reach have the highest scores and districts that adopt Harcourt Moving into English
have the lowest scores. While districts that adopt either Harcourt Moving into English or Rosetta
Stone are majority white, districts that adopt one of the five state-approved texts are majority
minority. Districts that adopt Rosetta Stone have the highest per pupil expenditure, the lowest
share of EBs, and are the most likely of the seven groups to be located in remote, rural areas.
This suggests that districts that have smaller numbers of EBs and are located in rural areas make
different decisions about ELD materials adoption compared to more urban districts that have
higher enrollments of EBs.
In sum, the descriptive statistics show that ELD adoption patterns vary by district
characteristic, suggesting that there is local influence over how ELD materials are chosen.
Specifically, ELD adoption patterns differ by district demographics, spending, test scores, and
neighborhood characteristics.
To what extent, if any, do ELD adoption patterns vary by district characteristics?
Table 2.8 summarizes the results from the multinomial logistic regression. Districts that do not
adopt any ELD materials serve as the reference group. Results from the logistic regression
indicate that district enrollment, percentage of EBs, per pupil expenditure, and share of
expenditures spent on instructional materials are predictive of ELD materials adoption. Larger
90
districts are more likely to purchase state-adopted ELD materials (odds ratio= 2.06, p <0.001)
than no ELD materials. Similarly, an increase in per pupil expenditure is associated with an
increase in the probability of adopting state-approved materials (odds ratio=1.80, p <0.05). An
increase in the share of district expenditures spent on instructional materials is associated with a
decrease in the likelihood of adopting off-list ELD materials (odds ratio, 0.92, p <0.05) relative
to non-adoption. There is evidence that an increase in the percentage of EBs is associated with an
increase in the likelihood of adopting state-approved materials (odds ratio, 1.25, p < 0.05). The
following district characteristics do not seem to be predictive of ELD adoption patterns: district
reading scores, percentage of white students, urbanicity, median income and share of the
population with a bachelors degree or higher.
91
Table 2. 8. Results from Multinomial Logistic Regression of ELD Adoption Patterns on District
Characteristics (Relative Risk Ratios)
State-Approved
Materials
Off-List Materials
District Log Enrollment 10-11 2.062*** 1.303
(0.220) (0.243)
District Percentage White Students 0.882 0.592
(0.537) (0.631)
District EB Percentage 1.247* 1.028
(0.123) (0.164)
District Log PPE 10-11 1.800* 1.931
(0.527) (0.830)
District Instructional Expenditures 10-11 0.995 0.924*
(0.026) (0.033)
City 0.635 0.431
(0.266) (0.496)
Town 0.727 0.675
(0.158) (0.298)
Suburban 1.069 0.719
(0.379) (0.612)
Standardized District Reading Scores 0.962 0.972
(0.105) (0.161)
Log Median Income 0.949 1.352
(0.381) (0.949)
Percentage with Bachelor’s Degree or higher 2.111 2.011
(2.374) (4.105)
Average teacher experience 1.070 1.059
(0.043) (0.067)
Percentage of white teachers 0.456 3.511
(0.280) (4.936)
_cons 0.000 0.000
(0.000) (0.000)
N 886 886
Note. Relative risk ratios shown. Standard errors in parentheses. Districts that did not adopt any
ELD materials serve as the reference group. Rural schools serve as the reference group for the
urbanicity variables.
*p <0.05. **p<0.01. ***p<0.001.
Overall, certain district characteristics such as size, expenditures, and EB composition are
predictive of ELD adoption. As mentioned earlier, prior research finds that districts that adopt
state-approved ELD curricula outperform districts that adopt no ELD curricula by approximately
0.29 standard deviations. The inequitable distribution of ELD resources has important policy
92
implications—particularly for those EBs located in smaller districts that also have smaller shares
of EBs. Based on the results of the logistic regression, EBs in large districts with higher shares of
EBs and greater per pupil expenditure have the greatest access to comprehensive ELD curricula.
These implications will be further outlined in the discussion section. Now, I turn to a discussion
of interview data to understand why ELD adoption patterns vary across districts.
Research Question 2: To what extent can the social constructions of EBs explain the
variance in ELD materials adoption?
How are EBs constructed? After multiple rounds of coding interview transcripts and
district documents, I find that there is great variation in how EBs are socially constructed. While
12 districts possess positive constructions of emergent bilinguals, 6 districts hold negative
constructions of emergent bilinguals and 10 districts held hybrid views. Although several
districts displayed both positive and negative constructions of EBs, I placed districts in the
category that best represents their overall perspectives on EBs. In the event that robust evidence
existed for both positive and negative constructions of EBs, then the district was placed in the
hybrid category, or in the middle of the continuum.
Positive construction. Twelve (43%) out of the 28 districts in the study held positive
constructions of EBs. Districts that held positive constructions of emergent bilinguals tended to
view EBs’ native language as an asset that needed to be developed. For instance, one of District
7’s goals is to maintain EBs’ “cultural and linguistic identity” and to ensure that students
graduate as biliterate and bilingual.
Furthermore, these district leaders believed that districts had a responsibility to provide
quality dual language programs so that EBs can maintain their native language. This group
generally understood how the linguistic needs of EBs differed from non-EBs. In addition, this
93
group had a deep understanding of the difference between English language development and
English language arts. Those who held pro-EB stances located the issue of EB underachievement
in systems instead of in individuals. For instance, District 7’s bilingual/ESL handbook states that
the district strives to provide “meaningful access to rigorous instruction, materials, and academic
choices within the school district.” In other words, the district recognizes that one of the
underlying causes of EB underachievement is the failure of schools to provide equitable access
to quality instruction.
To improve educational outcomes of EBs, these districts increased teacher capacity to
work with EBs and institutionalized policies to ensure ongoing professional development related
to EBs. For example, District 18 provided a $1,000 stipend for teachers who obtained the ESL
certification. These districts tended to adopt policies that removed financial barriers towards ESL
certification so that EBs may have access to better qualified teachers.
In sum, the districts in this group saw EB underachievement as a result of a cultural and
linguistic mismatch between schools and EBs, rather than seeing the problem arising from EBs
themselves. This group spoke positively about EBs’ home culture and of the need to honor
students’ heritage.
Negative construction. Six (21%) out of the 28 districts in the study negatively
constructed EBs. Generally, districts in this group placed more emphasis on English proficiency
than biliteracy, adopted subtractive bilingual programs, and did not implement district-wide
structures to help teachers become ESL-certified. Instead of recognizing the systemic barriers
that prevented the parents of EBs from becoming involved (e.g., linguistic barriers, cultural
mismatch), these districts tended to place blame on EBs and their parents. For example, District
94
Leader 17 believes that EBs “need to become more assimilated with society.” In addition, this
individual held a deficit view of EBs’ home life and academic abilities:
The parents should be a lot more involved…They need to be a part of their student
success because without them it’s not going to happen. They need to learn the English
language as well.
This district leader believes that the current level of EB parent involvement is inadequate and
that the barrier to parent involvement is the parent themselves. From this individual’s
perspective, one way to increase parent involvement is having parents learn English. However,
research shows that EB family engagement is increased when schools acknowledge the cultural
and linguistic capital possessed by parents of EBs and provide classes in parents’ native language
(Arias & Morillo-Campbell, 2008). Districts in this group tended to believe that the solution to
EB underachievement is to change EBs and their families.
Hybrid construction. Ten, or 36%, of the districts in the study had both positive and
negative constructions of EBs. District policies did not always align with district leaders’ beliefs
about second language acquisition. While district documents suggest that the district believes in
biliteracy, interviews with district leaders revealed a different narrative. Though this group was
the most difficult to understand, this group provided the most interesting insights about how
social constructions shape district practices. District 1 is a prime example of a hybrid
construction.
District 1’s decision to adopt both additive and subtractive bilingual programs contradicts
its bold mission statement of developing bilingual, bicultural, and biliterate students. District 1
offers both an early-exit transitional bilingual program along with a one-way dual language
program. Transitional bilingual programs, particularly early-exit programs, are not designed to
95
enhance students’ native language abilities as they are designed to transition students as quickly
as possible into an all-English program. District Leader 1 states that, in the last three years of the
district’s transitional bilingual program, the goal is “to have them proficient and completely
assimilated into the second language.” Though the district’s mission is to develop bicultural and
biliterate students, this district leader has assimilation as a competing goal.
Similarly, District 11’s mission statement conflicts with some of its EB-related policies.
District 11’s vision is to “respect linguistic diversity” and to help EBs “achieve bilingualism and
biliteracy.” However, the district has adopted an early-exit transitional bilingual program in
which the majority of instruction has transitioned into English by third grade. As previously
mentioned, it is difficult to achieve biliteracy with an early-exit transitional bilingual program
because students’ native language is gradually phased out of instruction. A dual language model
is more conducive to developing biliteracy because at least 50% of the day is spent in the
minority language. Furthermore, District Leader 11 also portrays inconsistent attitudes towards
the target population. On the one hand, District Leader 11 is aware of EBs’ developmental needs
and is currently adapting a software for EBs in high school to better address their academic
needs. On the other hand, she conflates ELD programs with programs for struggling readers.
Though some EBs may have difficulties with reading, the two terms are not synonymous.
How do social constructions shape ELD textbook adoption? After I determined the
social construction of EBs in each district, I looked to see if there was a relationship between
social construction of EBs and ELD adoption decisions. Table 2.9 outlines the social
construction of each district, the predicted ELD adoption pattern, and the actual adoption. In
addition, Figure 2.4 provides a visual display of the distribution of districts across the various
typologies. For 24 out of the 28 districts interviewed, social constructions correctly predicted the
96
ELD adoption pattern. Though I find that social constructions are linked to ELD adoption
patterns in most districts, a more complicated picture emerges in a few districts. In the following
paragraphs, I describe each target population—advantaged, contender, dependent, deviant—by
profiling at least one representative district. Where applicable, I discuss districts in which the
model does not correctly predict ELD adoption practices. I discuss districts with hybrid
constructions separately as they do not fit neatly into any one category.
Running head: ENGLISH LANGUAGE DEVELOPMENT CURRICULUM EFFECTS
97
97
Table 2. 9. Social Constructions of EBs and English Language Development Adoption Patterns
District Number
of EBs
District
Size
Power Construction Predicted ELD
adoption pattern
Actual ELD Instructional Materials
(as reported by district)
Correct
prediction?
1 High Medium Strong Hybrid Conflicting
policies
National Geographic Reach Yes
2 Moderate Large Weak Positive Inadequate or
symbolic aid
Texas Treasure Chest Yes
3 Low Small Strong Hybrid Conflicting
policies
Rigby On Our Way to English (2-5) Yes
4 Low Small Weak Hybrid Conflicting
policies
Rosetta Stone Yes
5 High Large Strong Positive Multiple ELD
adoptions
National Geographic Reach;
Houghton Mifflin Harcourt’s
English 3D; Academic Vocabulary
Toolkit
Yes
6 Moderate Large Weak positive Inadequate or
symbolic aid
National Geographic Reach (K-5) Yes
7 Low Small Strong Hybrid Conflicting
policies
Texas Treasure Chest (K-5) No
8 Moderate Medium Weak Hybrid Conflicting
policies
Rigby On Our Way to English (K-
5); Rosetta Stone; Imagine
Learning
Yes
9 Low Small Weak Hybrid Conflicting
policies
Rigby On Our Way to English;
Rosetta Stone
Yes
10 Low Medium Strong Positive Multiple ELD
adoptions
Total Physical Response
Storytelling, Academic Toolkit
Yes
11 Moderate Medium Weak Negative No adoption Flash Forward; Rigby On Our Way
to English (4-8)
Yes
12 Moderate Medium Weak Negative No adoption Rosetta Stone; National
Geographic Reach (7-8)
Yes
13 Moderate Medium Strong Positive Multiple ELD
adoptions
Rigby On Our Way to English;
Rosetta Stone; Reading A-Z
No
98
14 High Large Weak Positive Inadequate or
symbolic aid
National Geographic Reach;
Imagine learning
Yes
15 Low Small Weak Negative No adoption None Yes
16 High Large Weak Hybrid Conflicting
policies
Rigby On Our Way to English (K-
5); Duolingo
Yes
17 Low Small Weak Negative No adoption Odyssey Compass Learning,
Rosetta Stone, TX Treasure Chest
4-5
Yes
18 Moderate Medium Strong Positive Multiple ELD
adoptions
None for K-4; NG for 5-8; Istation No
19 High Large Weak Hybrid Conflicting
policies
National Geographic Reach for 6-
8; Reading Quest; Fast Forward;
(Imagine Learning)*
Yes
20 Low Small Strong Negative Confusing
policies
Lexia, Rosetta Stone, Mind Play Yes
21 Low Small Weak Negative No adoption None Yes
22 Moderate Medium Weak Hybrid Conflicting
policies
Rosetta Stone; Imagine Learning
(piloted); Rigby 4-5; Pearson 6-8
Yes
23 High Large Weak Positive Inadequate or
symbolic aid
Imagine Learning; Rigby; SIOP Yes
24 High Large Weak Positive Inadequate or
symbolic aid
EL Achieve; Treasure Chest;
Avenues
Yes
25 Low Medium Weak Positive Inadequate or
symbolic aid
Rigby On Our Way to English (K-
1); Longman Cornerstone (5
th
);
Longman Keystone; LitConn
English Now
Yes
26 High Medium Strong Hybrid Conflicting
policies
National Geographic Reach (7-8) Yes
27 High Large Strong Positive Multiple ELD
adoptions
Imagine Learning for high needs
campuses; Inside and Edge
Yes
28 High Large Weak Positive Inadequate or
symbolic aid
None (ELD block) No
99
Note. The categories for number of emergent bilinguals are as follows: low (0-500), moderate (501-1000), and high (1000+). The
categories for district size are: small (0-2,000), medium (2,001-10,000), and large (10,001+).
*This district piloted Imagine Learning but doesn’t have the funds yet to implement the curricula district-wide.
Running head: ENGLISH LANGUAGE DEVELOPMENT CURRICULUM EFFECTS
100
100
Figure 2. 4. Distribution of Districts by Social Construction
Advantaged. There are five districts that are positively constructed and are composed of
at least 17% EBs. For three out of the five districts in this category, the model correctly predicts
that beneficial policies will be oversubscribed to this population. Schneider and Ingram (1993)
posit that policy tools for advantaged groups will emphasize capacity building, free information,
training, and technical assistance. Districts in this group provided free training for teachers to
become ESL certified and, in many cases, paid for the cost of the ESL certification test for
teachers. Generally, districts in this category spoke positively about students’ native language,
adopted additive dual language programs, and created systematic practices that would increase
the number of ESL-certified teachers in the district. In all five districts, at least one ELD
101
curricula is purchased for EBs. In three of the districts, at least two ELD curricula are purchased
for EBs. However, the ELD policies for District 13 are more confusing and will be discussed
further.
District 5. I now take a closer look at one of the districts in this group in which the model
correctly predicts the ELD adoption pattern. District 5, a large urban district comprised of 22%
EBs, has a positive construction of EBs which is evident in the district’s goals and the various
programs that the district has created specifically for EBs and their families. District 5’s vision
for its EBs is to improve English proficiency “without sacrificing EBs’ native cultural beliefs and
values that are part of a global society” (District 5, 2017). The district views students’ native
culture as an asset that can help students engage with the wider global community. District 5 has
two types of additive bilingual programs: one-way dual language and two-way dual language.
District 5 strives to build students’ proficiency in both their native language and in English.
Lastly, this district removes linguistic barriers to parent involvement by providing free adult ESL
classes “to strengthen the family and communicate with parents.” The district views parents as
assets to their children’s education and aims to increase parent involvement by providing parents
with the linguistic tools needed to engage with school officials.
Given the positive construction and the high share of EBs in District 5, the model predicts
that beneficial policy will be oversubscribed to EBs. I do find that the district has multiple ELD
programs. Though the district formally adopted a state-approved ELD textbook, National
Geographic Reach, the district also has other supplemental ELD curricula such as Houghton
Mifflin Harcourt’s English 3D and Dr. Kate Kinsella’s Academic Vocabulary Toolkit. While
other districts in the study might adopt an ELD text for grades K-2 and nothing for grades 3-8,
District 5 is more systematic in its adoption decisions by adopting National Geographic Reach
102
for grades K-8. It should also be noted that both of the supplemental curricula used by this
district are curricula that are designed specifically for EBs. Later in this section, I discuss how
districts claim to adopt ELD curricula, but the curricula are actually reading programs designed
for struggling readers and not specifically for EBs. While 60% of EB-serving districts in Texas
do not purchase any ELD instructional materials, District 5 purchased three ELD curricula.
District 10. A close look at District 10 reveals a similar story. District 10 is a charter
management organization with 22% of its total student population identifying as emergent
bilinguals. The positive construction of EBs is evidenced by the organization’s goals and
practices. The mission of District 10 is to help students become multilingual. This organization
believes in multilingualism so fervently that all enrolled students must learn at least one language
outside of English. In addition, the organization institutionalizes practices so that teachers are
prepared to work with EBs. The district provides training and also covers the cost of the ESL
certification test:
So part of our plan is to make sure we offer opportunities for [teachers] to get
certified...let them know when these educator prep courses are being offered by either the
education service center or—and also if there are any study guides or any material that
we come across we make sure we forward it to the teachers…then we let them know the
district offers a reimbursement opportunity…We can reimburse them for the fee that the
state charges for them to take the test which is about $130.
Thus, there are few financial barriers for teachers to receive the qualifications needed to work
with EBs. Though the link between ESL certification and EB achievement is unclear, increasing
the number of ESL-certified teachers, at a minimum, allows EBs more flexibility in classroom
placement. In the state of Texas, EBs in the elementary grades must be served by ESL-certified
103
staff (TEA, 2012). In schools with few ESL-certified teachers, EBs are clustered into a few
classrooms leading to a de facto tracking system. This phenomenon will be discussed in greater
detail later.
The high share of EBs and the positive construction of EBs would predict that ELD
resources are oversubscribed to EBs in District 10. I find that schools leaders have at least three
different ELD programs. First, the district has an ongoing relationship with Stephen Krashen, a
leading scholar on second language acquisition and has asked Dr. Krashen to come to the school
and lead professional development for teachers. Dr. Krashen has been teaching a technique
called Total Physical Response Storytelling to the teachers in District 10. Second, the district has
also invited Dr. Kate Kinsella, another leading ELD scholar, to help train teachers on how to
teach academic vocabulary. Lastly, the organization is striving to train all teachers in the
Sheltered Instruction Observation Protocol (SIOP), which is a tool designed to help teachers
tailor their lessons to EBs. All three ELD tools utilized by District 10 are specifically designed
for EBs. The district has a deep understanding of how the linguistic needs of EBs are different
than that of non-EBs. Similar to teachers in District 5, teachers in District 10 have access to
multiple ELD curricula that are specifically designed for EBs.
District 13. Not all districts in this category adopted ELD practices that aligned with their
social construction of the target populations. For example, District 13 adopted a dual language
program with the goal of “preserving all cultures and languages” as well as ensuring that
stakeholders have a “clear and true understanding of the dual language program.” Despite these
visionary goals, I find that District 13’s ELD practices are conflicting and, at times, symbolic.
Though the district did adopt Rigby On Our Way to English, a state-approved ELD text, the
district also uses Reading A-Z and Rosetta Stone for EBs. Reading A-Z is designed to be a
104
reading program not a comprehensive ELD program. Reading A-Z emphasizes reading skills
while an effective ELD curricula foregrounds listening and speaking skills (Saunders et al.,
2013). Similarly, Rosetta Stone was not designed to be a comprehensive ELD program. Rosetta
Stone was designed to develop everyday conversational skills (Rosetta Stone, n.d.). Thus,
Rosetta Stone may not be sufficient to develop the academic language that EBs will need to
thrive in school. To further complicate the story, District Leader 13 states:
There’s just no standardization. Nobody does the same thing…One teacher may use
Rosetta Stone everyday. And one teacher uses Reading A-Z. And another might do
something completely different…I’ve never seen any of them use [Rigby On Our Way to
English]…all of them use [Rosetta Stone] different. Some of them use it once a week.
Some of them use it a couple of times a week. Some of them use it way too much…As
you can see, we’re having some work to do.
Though the district went through a formal adoption process complete with a standardized rubric
and committee, these efforts seem to be symbolic. The formally chosen comprehensive ELD
curricula, Rigby On Our Way to English, does not appear to be used in the classroom. Instead,
the district relies on Rosetta Stone and Reading A-Z, neither of which are comprehensive ELD
curricula. However, ELD policies in this district seem to be changing, with more beneficial
policies being allocated to EBs now than before.
District 13 represents a case of a district “in transit,” moving from one position to another
(Ingram, Schneider, & deLeon, 2007). EBs in district 13 may be better characterized as an
emerging advantaged group as the district has only recently prescribed benefits for EBs. For
example, the district recently hired a new elementary ELL specialist whose duty is to work with
teachers to implement curricula for EBs. This individual is attempting to standardize the ELD
105
curricula across the district. When asked about why the district created this new position, District
Leader 13 responds:
The district realized that they weren’t doing enough to help their ESL…ELL kids
progress. They’re not doing well on state assessments. They’re not progressing through
TELPAS like they should be. The number of ELL students has grown. Since it’s a large
population that we have, maybe we should make that our focus. And really focusing on it.
It took them a while to get on board with that, but now they have.
Due to their rising numbers, EBs are taking a more prominent position on the policy agenda for
this district. The rising power of the group together with its positive perception has led the
district to distribute more benefits to this group. Instead of placing blame on EBs, the district
recognizes its role in perpetuating educational inequality and is currently increasing resources for
EBs.
Deviants. There are five districts that have a negative construction of EBs and are also
comprised of less than 17% EBs. Overall, these districts have a limited understanding of how the
linguistic needs of EBs differ from those of non-EBs. Only one district in this group has a
bilingual program and that district adopted a subtractive bilingual model—early-exit transitional
bilingual. I find that the theory correctly predicts ELD adoption patterns for districts in this
group. Generally, districts in this category did not adopt comprehensive ELD curricula. I take a
closer look at two of these districts—15 and 21.
District 15. A small district comprised of 5% EBs, district 15 has no district initiatives
that require teachers to become ESL-certified, perhaps due to the small size of the EB
population. In addition, there is no bilingual program at this district, which, again, may be
106
explained by the small number of EBs. According to District Leader 15, each grade level has one
ESL teacher and all EBs in a given grade level are in the same class:
[EBs] all in the same classroom. They are serviced by the same teacher. We have one per
grade level so all of our ESL students, if they attend here throughout their whole
elementary career, they’re pretty much in the same cohort each year as they progress
through school
EBs are tracked into a de facto ESL class in this district due to the small number of ESL certified
teachers. As a result, EBs in this district find themselves on the periphery of the richest academic
discourse, a common theme in prior research on EBs (Adger & Peyton, 1999; Katz, 1999).
Similarly, Valenzuela (1999) finds that EBs have little chance to interact with English-dominant
students outside of mainstream classrooms, and, in these environments, teachers’ expectations
tend to be low.
Given the negative construction of EBs and the small number of EBs in District 15, I
predict that resources will be undersubscribed. Indeed, this district did not purchase any ELD
materials. Furthermore, this district uses Reading A-Z as a resource for EBs. As mentioned
previously, Reading A-Z is a reading program, not an ELD program. District 15 demonstrates a
lack of understanding of how the educational needs of EBs are different from non-EBs; thus, the
district uses the same instructional materials for all students, regardless of students’ linguistic
needs. With regards to burdensome policy, the district inadvertently created a policy whereby all
EBs in a grade level are tracked into the same classroom. This policy could be perceived as a
burden for the ESL-certified teacher as it creates high-needs classrooms without compensating
the teacher for the additional time and expertise to work with this population. In addition, this de
facto tracking policy creates a burden on EBs as it becomes more difficult for EBs to interact
107
with non-EBs and gain access to different types of cultural and social capital that are needed for
future academic success (Coleman, 1988; Lareau, 1987).
District 21. Now, I turn to District 21 which has less than 1,000 students and less than
10% EBs. Though District 21 does not have a bilingual program, the district does require that
teachers become ESL-certified and provides a stipend for ESL-certified teachers. Though
district-wide initiatives that increase teachers’ capacity to service EBs is evidence of a positive
construction of EBs, the evidence for a negative construction is far greater. For example, when
asked how her district chose their ELD model, District Leader 21 responds:
We learned and saw that that immersion program is much quicker…If you don’t put them
in a situation where they have to learn [English], then they’ll always—seems like they’ll
always use that crutch of getting help or not putting all of their effort into learning the
second language… they all end up speaking their first language more than trying to
practice and speak English.
Instead of considering the social and systemic factors that make EBs feel more comfortable in
their native language, this district leader attributes reliance on native language to a lack of
motivation to learn English. Furthermore, this district does not provide a separate block for ELD
for fear of clustering Spanish-speaking students together which would allow them to converse in
their native language. As previously mentioned, research indicates that schools should devote a
daily, separate block to ELD instruction (O’Brien, 2007; Saunders et al., 2013).
District 21 did not purchase any ELD materials nor did it use any free online ELD
resources. District Leader 21 justifies the decision to not purchase ELD materials by saying:
what’s good for our English as a first language is also what the English as a second
language learners need…like I said, what helps ELL learners it helps the English as a first
108
language student and vice versa. And the ESL component can be used with the English
students as well—the picture cards and labels and pictures with words. And that’s good
with all of our students too
She does not see the need to purchase ELD materials because she believes that the linguistic
needs of non-EBs and EBs are the same. In other words, whatever works for monolingual
English-speakers is sufficient to address the academic and linguistic needs of emergent
bilinguals. This finding is not new as previous research has found that a “just good teaching”
approach is not adequate to meet the academic and linguistic needs of emergent bilinguals (De
Jong & Harper, 2005).
Dependents. Seven districts in the study have a positive construction of EBs but are
comprised of fewer than 17% EBs. These districts are predicted to allocate symbolic or
inadequate resources to EBs. I find that the districts in this category had highly variable ELD
adoption patterns, ranging from no adoption to the adoption of state-approved ELD materials.
Despite the large range of ELD practices, a pervasive finding in this category is the symbolic
nature of ELD practices. The model correctly predicts ELD adoption practices for six out of the
seven districts in this category. District 28 is a notable outlier in this category and will be
discussed in greater detail below.
District 2. District 2 is a prime example of the complex nature of decision-making for a
positively constructed target population that does not wield much power. Comprised of 4.2%
EBs, District 2 has adopted a dual language program, required all teachers to be ESL-certified,
and delegated a team to evaluate the effectiveness of their bilingual program. During the 2011-
2012 school year, District 2 adopted McGraw-Hill Texas Treasure Chest, a state-approved ELD
text, as its ELD curricula for grades K-5. Even though the district has invested a significant
109
amount of financial resources into a comprehensive ELD curricula, the bilingual/ESL
coordinator from District 2 states, “I have not seen one class use Treasure Chest.” However, her
experience should be interpreted with caution as she started this role in the past few years. Thus,
it is possible that this district leader does not have a representative view of ELD practices in the
district. Regardless, District Leader 2’s statement suggests that the ELD curricula is not being
used as intended. For future ELD adoptions, this district leader states that:
If we adopted a textbook just for language acquisition, I’m afraid that students might get
pulled out when they should be in those classrooms getting core content taught by the
classroom teacher. Then it would be your kids and not my kids. We work really hard to
have a more inclusive environment and to build teacher’s skillset so they can teach the
content and also develop language.
This district leader is hesitant to purchase ELD materials in the future because she fears that the
presence of an ELD curricula will justify the separation of EBs and non-EBs. Though District
Leader 2 demonstrates positive intent, the desire to create more inclusive environments, her
beliefs may translate to actions that have negative impacts on EBs. Research shows that
providing ELD instruction is better than not providing it and that the likelihood of sustaining an
effective ELD instructional program increases when districts make it a priority (Saunders,
Goldenberg & Marcelletti, 2013). Positive intent, in this case, has led to enactment of a symbolic
policy in which ELD materials are not used as intended. Symbolic policies are not sufficient in
giving EBs the ELD instruction they need to become English proficient; systematic policies and
practices that are supported at both the district and school level are also needed. In sum, District
2 is a representative case of a dependent target population that receives symbolic policies.
110
District 28. One notable outlier in this group is District 28 whose ELD policies and
practices are not merely symbolic. The district is composed of 7.2% EBs and provides both
additive and subtractive bilingual programs for EBs. Overall, the district has a positive
perception of EBs and invests in district-wide initiatives to improve educational outcomes of
EBs. In contrast to previous years in which the district integrated ELD into the core subjects, the
district is currently transitioning into a separate ELD block for EBs. Research supports the idea
that separate ELD blocks are needed to address the specific linguistic needs of EBs (Saunders,
Goldenberg & Marcelletti, 2013). For District 28, the idea of a separate ELD block emerged
from a school-university partnership. When asked about how the ELD program was created,
District Leader 28 responds:
We’re always collaborating with the professors at the university to help us determine the
latest research. This is what we need to be doing to help support our ELLs. This is what
an ELD lesson would look like. From that, we take all of those suggestions and samples
and then we get what we need.
While the vast majority of districts in the study use an integrated ELD model in which ELD is
integrated into the language arts block, District 28 is the only district that is moving towards a
separate ELD block, a practice supported by research (Saunders et al., 2013). Though District 28
is predicted to adopt symbolic policies, I find that the district is investing significant resources
into creating systematic ELD policies and practices for EBs. One possible explanation for this
anomalous finding is the presence of a school-university partnership. The partnership appears to
provide district leaders with free information and the technical assistance necessary to curate a
systematic ELD program. The university’s influence over district ELD decision-making may be
explained by the university’s high-status positioning in the education arena.
111
Generally, the theory correctly predicts that districts that have positive constructions but
low concentrations of EBs adopt symbolic or inadequate ELD practices.
Contenders. One district in the study is composed of greater than 17% EBs and is
characterized as having a negative construction of the target population. Schneider and Ingram’s
(1993) model predicts that policies for contenders will be complex, confusing, and vague.
Indeed, I find that ELD practices for EBs in this district are confusing.
District 20. District 20 is comprised of 17% EBs and has a total student population of less
than 3,000. District documents and interview data reveal that EBs are negatively constructed in
district 20. When asked about how the district came to adopt a dual language program, District
Leader 20 states, “Our children who come speaking broken English don’t come from strong,
verbal backgrounds so we had to do something.” Scholars contest the deficit framing of EBs’
linguistic abilities as it does not take into consideration the rich language structures that EBs can
flexibly use across different linguistic settings (McSwan, 2017). In addition, the district’s stated
goal of encouraging students to “continue to retain their home language” makes no mention of
developing students’ home languages. Helping a student to retain a language is different than
helping a student gain new skills in a language.
Furthermore, the district provides conflicting descriptions of their bilingual program.
District documents state that they offer a “dual language program” to students in grades pre-K to
third. District Leader 20 states that the dual language program is aligned with the Gómez and
Gómez model which divides language of instruction by content area as well as by time (Gómez,
Freeman & Freeman, 2005). However, the Gómez-Gómez model is an additive bilingual
program in which students are taught literacy and content in two languages and at least 50% of
the day is spent in the minority language. District Leader 20 states, “I don’t care what Gómez
112
wants. We have a little conflict there but we have to put our [EBs] in an English classroom.”
This district leader believes that the Gómez-Gómez model does not adequately prepare EBs for
the state test, so she increases the instructional time spent in English, which necessarily decreases
the instructional time in Spanish. By reducing the amount of time spent in the minority language,
the bilingual program, in practice, is a subtractive one though district documents state that the
district offers an additive bilingual program. The negative perception of EBs in this district has
translated into a change in policy, with the additive bilingual program transforming into a
subtractive one.
District Leader 20 is also confused about the purpose of ELD instruction and Spanish
language instruction. When asked if her district should consider adopting ELD materials in the
future, District Leader 20 responds, “I don’t think that we would adopt an ESL textbook in the
fact that we have the two different types of dual language. I don’t see the point.” A dual language
program is not the same as English language development. A dual language program will deliver
content in two languages, but it does not necessarily mean that ELD will be delivered in a
separate block in which language standards are foregrounded.
The confusion between English language development, Spanish language development,
and English language arts might explain why this district has adopted an eclectic mix of
materials for ELD: Lexia, Rosetta Stone, and Mind Play. None of the materials mentioned by
District Leader 20 are comprehensive ELD curricular materials. In addition, none of the
materials are created specifically for EBs in academic settings. Though Rosetta Stone was
created for language learners, their products were created to build “everyday conversational
skills” (Rosetta Stone, n.d.) which is only one component of effective ELD instruction
(Saunders, Goldenberg & Marcelletti, 2013). According to its website, Lexia “provides explicit,
113
systematic, personalized learning in the six areas of reading instruction.” Lexia focuses on
reading, which is only one dimension of language. Similarly, MindPlay is a “phonics educational
software” whereby “students experience a virtual one-on-one session with a reading specialist
each time they log in” (MindPlay, n.d.). Though developing reading skills for EBs is important,
ELD instruction should focus on listening and speaking skills (Saunders, Goldenberg &
Marcelletti, 2013). As predicted by the theory of social construction, hidden benefits are
allocated to contending groups because no policymaker wants to be seen as favoring an
undeserving group (Ingram, Schneider & deLeon, 2007). To avoid appearing as if it were aiding
an undeserving group, the district adopts a reading program that can be used with all learners, not
just EBs.
Districts with hybrid constructions and high shares of EBs. Four districts in the study
have hybrid constructions of EBs and are composed of more than 17% EBs. I predict that ELD
adoption in these districts will be confusing and complex due to the existence of multiple social
constructions. While all districts in this category have purchased state-approved ELD curricula,
not all districts have adopted ELD curricula in a systematic way. I discuss District 3 and 7 in
greater detail as they represent examples of a correction prediction and an incorrect prediction.
District 3. Comprised of over 20% EBs, District 3 has mixed perceptions of EBs. On the
one hand, the district adopted a subtractive bilingual program in which the district transitions
Spanish-speaking EBs as quickly as possible into an all-English setting. On the other hand, the
district has taken steps to ensure that all primary
13
teachers are ESL-certified. Though not all
elementary teachers are ESL-certified, over 95% of them have this classification. Adding to the
positive construction of EBs, District Leader 3 has a deep understanding of the structural
13
In this district grades PK-1 are considered primary grades and grades 2-5 are considered to be elementary grades.
114
elements needed in an effective program for EBs. With regards to how ELD materials are
purchased, District Leader 3 states:
I wanted to make sure that they were culturally relevant. They had an affective as well as
an academic component. There were ideas for instruction in each one of the units for
beginning, intermediate, and those students who were approaching the advanced levels of
fluency.
District Leader 3’s attention to the cultural and affective aspects of EB programming is
supported by prior research which finds that effective programs for EBs attends to the social,
cultural and linguistic diversity of the EB population (Gil & Bardack, 2010). Furthermore,
District Leader 3 also recognizes the diversity in language proficiency within each grade level
and strives to find materials that can differentiate instruction for these various groups.
Given the mixed perception of EBs in District 3, the district is predicted to adopt
confusing ELD practices. Indeed, this district adopted a comprehensive ELD curricula for grades
2-5 but not for grades K-1. This practice is particularly confusing because there are more EBs in
the lower grades than in the upper elementary grades, which points to a greater need for ELD
materials in grades K-1. This district leader also mentions that she has not seen the adopted ELD
curricula used in the classroom. On the continuum of social constructions, District 3 skirts the
boundary between the advantaged and contender groups, opening up opportunities for a wide
range of policy designs. By prescribing benefits to some EBs but not all, the district can satisfy
the varying perceptions of EBs.
District 7. District goals, interview data, and district policies regarding EB all provide
different conceptions of EBs. The district adopted a subtractive bilingual program—early-exit
transitional—and believes that if they can exit EBs by the second grade then “the bilingual
115
program has done its job.” This district leader believes that the goal of a bilingual program is to
develop English proficiency—not to develop biliteracy. However, the stated goal of the
bilingual/ESL department in this district is to ensure that students are “biliterate and bilingual.”
While district goals conceive of EBs’ multilingualism as an asset to be further developed, district
policies tell a different story, one in which multilingualism is only fostered until the end of 2
nd
grade.
District Leader 7’s beliefs about the purpose of bilingual education also conflict with the
department’s stated goals. This district leader justifies the decision to adopt an early-exit
program by saying, “We don’t want a late-exit because if you do a late-exit then they cannot exit
until 5th grade.” From the perspective of this district leader, it is better to transition students out
of Spanish as quickly as possible. However, research does not substantiate this point of view. In
fact, dual language programs have similar, if not better, outcomes compared to transitional
bilingual programs (Valentino & Reardon, 2015). Lastly, District Leader 7 states, “Our biggest
accountability problems are with our ELL population.” Instead of viewing the issue as an
institutional one whereby the district is failing to provide adequate programming for EBs,
District Leader 7 believes that EBs are introducing extra burdens for the district.
Given District 7’s conflicting construction of EBs as both deserving and undeserving, I
predicted that their ELD adoption practices would be confusing. However, this is not the case.
District 7 adopted a comprehensive ELD curricula for all of its elementary grades, K-5. One
possible explanation for this anomalous finding is the district’s commitment to improving
student outcomes. District 7 and the other districts in the region were awarded a significant grant
by the federal government to improve teacher effectiveness, suggesting that the district is
committed to improving school quality for its students.
116
Though District 7 adopted a state-approved ELD curricula for grades kindergarten
through fifth, the other districts in this category only adopted ELD curricula for certain grades.
For example, District 26 adopted National Geographic Reach for grades 7-8, and District 3
adopted Rigby On Our Way to English for grades 2-5 (see Table 2.9).
Districts with hybrid constructions and low share of EBs. Six districts in the study are
comprised of fewer than 17% EBs and have hybrid constructions of the target population. The
districts in this category had widely different ELD adoption patterns. While three districts have
purchased comprehensive ELD curricula, the other three districts have purchased reading
programs, Rosetta Stone, or combination of various programs. Of the various typologies in the
model, this group has the widest variation in ELD adoption patterns.
District 19. One manifestation of District 19’s mixed perception of EBs is its use of
additive and subtractive bilingual programs. Additive bilingual programs positively construe
EBs’ native language as an asset that should be fostered. On the other hand, subtractive programs
construct EBs’ native language as an impediment to English proficiency. The perception of EBs
seems to differ based on position in the district. While members of the bilingual/ESL model
believe that EBs should be taking the state test in Spanish, principals believe otherwise. District
Leader 19 states:
If we are true to a Gómez and Gómez model…it is recommended that our ELL
population test in their L1 because they have received instruction faithfully in pre-K,
kinder, 1st, 2nd, and 3rd. But the buy-in is not there because of testing, STAAR testing.
My principals, or the leadership, doesn’t understand that the value of a passing Spanish
test is just as valuable as an English STAAR test. That buy-in is critical in showing that
the program that we call dual language or Gómez and Gómez is of value.
117
District Leader 19 tries to stay faithful to the dual language model by having students test in the
language of instruction, which is Spanish for grade 3. However, principals at the district are
pushing for students to test in English as early as possible, even when the majority of their
instruction has been in Spanish. The privileging of English testing over Spanish testing suggests
that native language proficiency is viewed less positively than English language proficiency.
While principals in the district seem to have more negative perceptions of EBs, the head of the
bilingual/ESL department shows deep respect for students’ native language. She tries to purchase
Imagine Learning for EBs in her district because of the way the curricula “respects the use of
L1.”
Given the mixed construction of EBs in District 19, the theory of social construction
predicts that they will adopt confusing ELD policies. Indeed, District 19 adopted a variety of
materials for ELD: National Geographic Reach (Grades 6-8), Reading Quest and Fast Forward.
Two out of the three curricula are reading curricula, which are not designed to adequately
address ELD standards. Reading Quest is designed to provide students with “comprehension and
content reading strategies,” and Fast Forward is marketed as a reading intervention program.
District 19’s complex ELD policies and practices can be seen as the product of competing views
on how to improve educational experiences for EBs. While the perception of EBs as struggling
readers has led to the adoption of Fast Forward, others with more positive perceptions of
bilingualism are pushing for the adoption of Imagine Learning, a research-based program that is
first-language supported.
Limitations
There are several limitations to the study design. First, instrumentation selectively
determines how much of the data is collected. The questions on the interview protocol and how
118
they are asked determine what kinds of data are collected. Informants themselves are selective as
well, determining how much information they would like to share. For instance, interviewees
may gloss over parts of the instructional materials adoption process, and the researchers may not
be aware of this. To better collect relevant information, I reviewed district documents and
websites before each interview. By entering the interview with prior knowledge about what ELD
practices and materials are used in the district, I am able to probe on areas that are hard to recall
for interviewees. In addition, I probe on potential discrepancies that arise during an interview.
Specifically, some district leaders cited using ELD materials that differed from the materials
cited on state documents. Since I was attuned to these discrepancies, I was able to ask questions
during the interview that helped me understand how these differences arose.
Second, interview data may not align with what actually happens in the districts. In their
ethnography of an elite boarding school, Khan and Jerolmack (2013) find that there is a
discrepancy between participants’ accounts and their actions. Similarly, it is possible that district
leaders’ accounts of ELD practices may differ from what occurs for a variety of reasons. For
example, interview questions required district leaders to recall elements of the ELD adoption
process which, for some individuals, took place over four years ago. It may be difficult for some
individuals to recall events that took place several years ago. Though I cannot fully address this
limitation of interview data, I gained a more robust view of district practices and policies by
triangulating interview data with district documents and district websites.
Relatedly, information from public information requests may not accurately capture the
ELD texts that are used in the district. In a few cases, districts required more time to fulfill the
public information request because they could not identify the individual in charge of this
information. Furthermore, a few districts were experiencing high turnover in district offices,
119
leading to newly employed district officials with limited information on district ELD
instructional practices. However, there is evidence to suggest that information from public
information requests are generally accurate. During my interview with district leaders, I asked
interviewees to list any ELD instructional materials used in the district. The information
provided by district leaders generally aligned with the information from state reports and from
public information requests. In the few cases where the information did not align, district leaders
could not recall the name of the ELD text used in the district.
Third, the operationalization of power may seem arbitrary. Power is a multidimensional
concept that can refer to a group’s ability to mobilize, access to decision points, wealth, and
intensity of beliefs, or size of the group (Schneider & Ingram, 2008). However, it was not
feasible to create a composite power variable consisting of all these various components of
power. Thus, I only consider the size of the group in operationalizing power. In addition, the
cutoff point of 17% may seem arbitrary. Districts near the cutoff could have easily been placed
into another category depending on district demographics for the year. For example, several
districts reported that their EB population was growing. In one year, the district may have been
considered to have a “low” share of EBs, but in the following year, the district might have been
labeled as having a high share of EBs. The change in percentage might not be reflective of a
change in the perceived power of the target population.
Fourth, 6% of districts in Texas did not report which, if any, ELD materials they used.
Districts who did not respond to the FOIL request (N=67) are smaller in size, have higher shares
of EBs, and have higher shares of white students. Missing data may compromise the
generalizability of estimates from the multinomial logistic regression since those who report
information seem to be different than those districts that choose not to report ELD information.
120
However, it seems unlikely that the estimates would be substantially biased as ELD textbook
information is missing for a small number of districts.
Fifth, the vast majority of district leaders who were asked to participate did not respond
or refused to participate. Specifically, 7.3% of district leaders who were contacted agreed to
participate in an interview. Those who agreed to participate may differ from those who refused
on both observable (e.g., gender, race, and years of experience) and nonobservable
characteristics (e.g., knowledge of second language acquisition, motivation). These differences
may affect the information that was gathered from the interview. However, multiple individuals
were interviewed within each cell, except for the cells for off-list adopters, which may help to
capture the variation of responses.
Finally, as I only interviewed one individual in each district, this individual’s views may
not be representative of those involved in ELD decision-making. To elaborate, larger districts
have a bilingual/ESL department composed of multiple individuals who work together to create
programs and policies for EBs. If members of this department have diverse social constructions
of EBs, the diversity of perspectives may not be captured by the interview data in this study.
Discussion and Implications
Federal court cases such as Castaneda v. Pickard (1981) have established EBs’ rights to
theoretically sound and empirically tested educational programs. However, access to ELD
instructional materials, which is one part of an effective program for EBs, is highly variable
across districts. One reason for the high variability is that districts in Texas have the autonomy to
purchase or not purchase ELD materials for their EBs. In the absence of a state mandate on ELD
materials, I find that the majority of EB-serving districts in Texas do not purchase any ELD
curricula for their students. In addition, the way that EBs are socially constructed plays a role in
121
determining what types of resources district leaders direct towards EBs. The results from this
study have several implications for how EBs are served.
The findings suggest that ELD materials are inequitably distributed across districts.
District characteristics such as size, expenditures, and EB composition are predictive of ELD
adoption. Larger districts are more likely to purchase state-adopted ELD materials than no ELD
materials. Districts that have higher shares of EBs are also more likely to purchase state-
approved materials. These quantitative findings indicate that EBs in small districts or districts
with small shares of EBs are at a disadvantage because they have a lower likelihood of being
exposed to state-adopted ELD instructional materials. Indeed, qualitative data from interviews
reveal that district leaders do take into consideration the number of EBs when making ELD
adoption decisions. Given that Le and Polikoff (2017) find that districts that adopt
comprehensive ELD materials outperform districts that do not adopt any ELD materials on
measures of English language proficiency, it is crucial that EB-serving districts consider
implementing systematic ELD practices. It is possible that off-list ELD materials are also
effective in promoting English language proficiency; however, Le and Polikoff (2017) are not
able to test this association due to the small number of off-list adopters in their study. Local
education agencies may benefit from creating additional supports for districts with small shares
of EBs so that EBs in these districts receive access to ELD materials.
Texas State Bill 6 (2011) allows districts to use state funds to buy materials that are not
state-approved. Districts do not seem to take advantage of this added flexibility. Of districts that
purchase an ELD curricula, 85% of them purchase from the state-approved list. One possible
explanation for this finding is that institutional memory regarding adoption practices influences
the choices that district leaders make. Some ideas become so ingrained in organizational life that
122
they become hard to shake. Research in institutional theory finds that the allocation of resources
can institutionalize practices (Dillard, Rigsby & Goodman, 2004). In this case, the state
originally allocated state funds to be used only on state-approved books. The allocation of
financial resources for state-approved textbooks has led to the institutionalization of state-
approved texts.
As mentioned previously, Schneider and Ingram (1993) posit that court mandates make it
possible for beneficial policies to be prescribed to powerless or negatively constructed groups.
However, I find that court mandates such as Castañeda v. Pickard (1981) cannot completely
overcome the negative construction of EBs. Over 50% of EB-serving districts in Texas do not
purchase any ELD materials. This is concerning as research shows that EBs benefit from a
separate, daily ELD block and when schools make ELD a priority (Saunders, Goldenberg &
Marcelletti, 2013). Though it is possible for districts to create their own comprehensive ELD
curricula and implement it systematically across multiple grade levels, this is much harder to do.
Even the advantaged group of EBs in the study did not always receive beneficial policy. I find
that a few districts in the advantaged group undersubscribed beneficial policies to their EBs.
Indeed, some districts utilized confusing ELD practices in which reading intervention programs
were used as a substitute for ELD curriculum.
Data from interviews with district leaders reveal that there is confusion about ELD and
English language arts. This confusion may partially explain why the majority of EB-serving
districts in Texas do not purchase any ELD curricula. Some teachers believe that the instruction
given to monolingual English-speakers should be sufficient to meet the needs of EBs. For
example, some district leaders believe that instruction for EBs is “just good teaching” (JGT).
Thus, there is no need to purchase materials specifically for EBs. This finding is not new as
123
previous research has found that a “just good teaching” approach is not adequate to meet the
academic and linguistic needs of EBs (De Jong & Harper, 2005). A decade after De Jong and
Harper (2005) coined the term “just good teaching” to describe this phenomenon, I find that the
JGT view of teaching EBs is still prevalent.
Similarly, there appears to be confusion the purpose of ELD. Some districts in the study
stated that they did not purchase ELD materials because they already have plenty of materials in
Spanish. While Spanish language development is an important of Spanish-English bilingual
programs, it is not the same as English language development. Without a common understanding
of these various terms, a variety of conflicting ELD policies emerge that may be
counterproductive for EBs. For example, some districts adopt comprehensive ELD curricula for
some elementary grades but not for all. In this type of setting, EBs are not receiving consistent
ELD instruction which may hinder their acquisition of English. Not only is ELD conflated with
English language arts, some district leaders confuse EBs with students with individualized
education plans (IEP). Though it is possible for an EB to also require specialized educational
needs, an IEP does not necessarily outline the linguistic needs of EBs. These findings suggest
that some districts may benefit from professional development that distinguishes English
language development, English language arts, native language development, and special
education services. Specifically, English language proficiency should not be used as a proxy for
academic ability.
Districts that remove financial and systemic barriers towards bilingual/ESL certification
have a greater share of bilingual/ESL-certified teachers. When the number of ESL-certified
teachers increases in a district, the more spread out EBs can be. In other words, districts with
only one ESL-certified teacher per grade create a de facto tracking system whereby EBs are
124
concentrated in the same classroom year after year. There is strong evidence that bilingual
certification has a positive impact on EB academic achievement. When EBs in the elementary
grades at Miami-Dade County Public Schools have a bilingually certified teacher, math
achievement is predicted to increase by 0.18 standard deviation (Loeb, Soland, & Fox, 2014).
However, it is unclear the extent to which ESL certification improves EB academic achievement.
Regardless, the increase in share of ESL-certified teachers decreases the likelihood that EBs are
tracked into the same classrooms. To increase the number of ESL-certified teachers, districts
should consider covering the costs of ESL certification and providing the necessary training for
teachers to become ESL-certified.
Lastly, some districts conflate EBs with struggling readers. Though some EBs may also
experience reading difficulties, the two terms are not synonymous. The conflation of these two
words is further evidence of the negative perception of EBs’ academic abilities. As a result, the
negative perception translates into a policy design that does not appropriately address the needs
of EBs. Instead of prescribing a comprehensive ELD program to EBs, some schools purchase
reading intervention programs for EBs.
Conclusion
Since prior research (Le & Polikoff, 2017) finds that state-approved ELD curricula is
associated with a significant increase in English language proficiency scores, the variation in
ELD adoption patterns in Texas invariably points towards the inequitable distribution of
resources for EBs. Specifically, districts with higher enrollments, shares of EBs, and per pupil
expenditure are more likely to adopt state-approved ELD materials. The variation in ELD
adoption patterns can be partially explained by how districts perceive of EBs.
125
How districts socially construct EBs is linked to the ELD practices they adopt. For
example, districts in which EBs were positively construed and had high numbers had at least one
ELD curricula, and, in many cases, had access to multiple ELD curricula. For districts in which
EBs are contenders, ELD policies are indeed confusing and conflicting. ELD adoption patterns
were highly variable in the groups that had hybrid constructions of EBs. Districts in the
dependent group also had a multitude of ELD adoption patterns. However, a common finding
amongst the dependent group is that ELD policies seemed to be symbolic in nature. Lastly, ELD
materials were generally undersubscribed in districts that constructed EBs as deviants.
Though the social construction of EBs is associated with ELD adoption patterns, social
constructions are not fully predictive of district decision-making. For four districts in the study, I
incorrectly predicted the ELD adoption patterns. These incorrect predictions may be explained
by alternative schemas in the environment about how EBs should be taught. For example,
District 28 and 6 are both located near universities and were predicted to adopt confusing and/or
symbolic ELD practices. However, both districts have adopted systematic ELD practices which
may be explained by their relationship with the nearby university. This study adds to the current
research literature which finds high variability in the implementation of federal policies. In
addition, the study confirms previous research on the influence of social constructions on policy
design.
126
References
Abadie, A., & Imbens, G. W. (2008). On the failure of the bootstrap for matching
estimators. Econometrica, 76 (6), 1537-1557.
Adams, M., & Jones, K. M. (2006). Unmasking the myths of structured English immersion: Why
we still need bilingual educators, native language instruction, and incorporation of home
culture. The Radical Teacher, 75, 16-21.
Adger, C. T., & Peyton, J. K. (1999). Enhancing the education of immigrant students in
secondary school: Structural challenges and directions. In C. J. Faltis & P. M. Wolfe
(Eds.), So much to say: Adolescents, bilingualism and ESL in the secondary school (pp.
205–224). New York: Teachers College Press.
Agodini, R., Harris, B., Atkins-Burnett, S., Heaviside, S., & Novak, T. (2010). Achievement
effects of four early elementary school math curricula: Findings for first and second
graders. Washington, DC: Institute of Education Sciences, National Center for Education
Evaluation and Regional Assistance.
Aina, J. K., Ogundele, A. G., & Olanipekun, S. S. (2013). Students’ proficiency in English
language relationship with academic performance in science and technical
education. American Journal of Educational Research, 1(9), 355-358.
Alajmi, A. H. (2009). Addressing computational estimation in the Kuwaiti curriculum: teachers’
views. Journal of Mathematics Teacher Education, 12(4), 263-283.
Alajmi, A. H. (2012). How do elementary textbooks address fractions? A review of mathematics
textbooks in the USA, Japan, and Kuwait. Educational Studies in Mathematics, 79(2),
239-261.
127
Arias, M. B., & Morillo-Campbell, M. (2008). Promoting ELL parental involvement: Challenges
in contested times. Retrieved from http://epsl.asu.edu/epru/documents/EPSL-0801-250-
EPRU.pdf.
Arnold, J. (1999). Affect in language learning. Cambridge: Cambridge University Press.
Association of American Publishers. (2015). Instructional materials adoption. Retrieved from
http://publishers.org/our-markets/prek-12-learning/instructional-materials-adoption.
Augurzky, B., and C. Schmidt (2001): The propensity score: A means to an end (Working
Paper). Berlin, Germany: Rhine-Westphalia Institute for Economic Research.
August, D., & Shanahan, T. (2006). Developing literacy in second-language learners: Report of
the National Literacy Panel on Language-Minority Children and Youth. Mahwah, NJ:
Lawrence Erlbaum Associates, Publishers.
Austin, P. C. (2008). A critical appraisal of propensity ‐score matching in the medical literature
between 1996 and 2003. Statistics in medicine, 27(12), 2037-2049.
Baker, K. (1998). Structured English immersion. Phi Delta Kappan, 80(3), 199.
Ballantyne, K. G., Sanderman, A. R., & Levy, J. (2008). Educating English Language Learners:
Building Teacher Capacity. Roundtable Report. Washington, DC: National
Clearinghouse for English Language Acquisition & Language Instruction Educational
Programs.
Beckett, K. (1997). Making crime pay: The politics of law and order in the contemporary United
States. New York: Oxford University Press.
Bhatt, R., & Koedel, C. (2012). Large-scale evaluations of curricular effectiveness: The case of
elementary mathematics in Indiana. Educational Evaluation and Policy Analysis, 34(4),
391-412.
128
Bhatt, R., Koedel, C., & Lehmann, D. (2013). Is curriculum quality uniform? Evidence from
Florida. Economics of Education Review, 34, 107-121.
Black, D. A., & Smith, J. A. (2004). How robust is the evidence on the effects of college quality?
Evidence from matching. Journal of Econometrics, 121(1), 99-124.
Bowler, M. (1978). Textbook publishers try to please all, but first they woo the heart of
Texas. The Reading Teacher, 31(5), 514-518.
Brewer, J., & Hunter, A. (1989). Multimethod research: A synthesis of styles. Thousand Oaks,
CA: Sage Publications.
Brisk, M. E. (1998). Bilingual education: From compensatory to quality schooling. Mahwah,
New Jersey: Lawrence Erlbaum Associates.
Burke, A. M., Morita-Mullaney, T., & Singh, M. (2016). Indiana Emergent Bilingual Student
Time to Reclassification: A Survival Analysis. American Educational Research
Journal, 53(5), 1310-1342.
Caliendo, M., & Kopeinig, S. (2008). Some practical guidance for the implementation of
propensity score matching. Journal of economic surveys, 22(1), 31-72.
Campbell, S., & Polikoff, M. (2017). Factors associated with the selection of curricular
materials at the district level. Paper presented at the annual conference of the Association
for Education Finance and Policy, Washington, DC.
Capps, R., Fix, M., Murray, J., Ost, J., Passel, J. S., & Herwantoro, S. (2005). The new
demography of America's schools: Immigration and the No Child Left Behind
Act. Washington DC: Urban Institute.
129
Carlisle, J. F., Beeman, M., Davis, L. H., & Spharim, G. (1999). Relationship of metalinguistic
capabilities and reading achievement for children who are becoming bilingual. Applied
Psycholinguistics, 20(4), 459-478.
Castañeda v. Pickard. (1981). 648 F.2d 989 (5th Cir.).
Chamot, A. U., & O’Malley, J. M. (1994). The CALLA handbook: Implementing the academic
language learning approach. Reading, MA: Addison-Wesley.
Cheung, A. C., & Slavin, R. E. (2012). Effective reading programs for Spanish-dominant English
language learners (ELLs) in the elementary grades: A synthesis of research. Review of
Educational Research, 82(4), 351-395.
Chingos, M. M., & Whitehurst, G. J. (2012). Choosing blindly: Instructional materials, teacher
effectiveness and the Common Core. Washington, DC: Brown Center on Education
Policy at Brookings.
Clark, R. (1970). Crime in America: Observations on its nature, causes, prevention and control.
New York: Simon and Schuster.
Coburn, C. E. (2004). Beyond decoupling: Rethinking the relationship between the institutional
environment and the classroom. Sociology of Education, 77(3), 211-244
Coleman, J. S. (1988). Social capital in the creation of human capital. American Journal of
Sociology, 94, S95-S120.
Coleman, R., & Goldenberg, C. (2012). The Common Core Challenge for ELLs. Principal
Leadership, 12(5), 46-51.
Collier, V. P., & Thomas, W. P. (2004). The astounding effectiveness of dual language education
for all. NABE Journal of Research and Practice, 2(1), 1-20.
130
Common Core State Standards Initiative. (2015). Standards in your state. Retrieved from
http://www.corestandards.org/standards-in-your-state/.
Cook, T. D., & Steiner, P. M. (2010). Case matching and the reduction of selection bias in quasi-
experiments: The relative importance of pretest measures of outcome, of unreliable
measurement, and of mode of data analysis. Psychological methods, 15(1), 56.
Creswell, J. W., Plano Clark, V. L., Gutmann, M. L., & Hanson, W. E. (2003). Advanced mixed
methods research designs. In A. Tashakkori & C. Teddlie (Eds.), Handbook of mixed
methods in social and behavioral research (pp. 209–240). Thousand Oaks, CA: Sage.
Cummins, J. (1979). Linguistic interdependence and the educational development of bilingual
children. Review of Educational Research, 49(2), 222-251.
Cummins, J. (1994). The discourse of disinformation: The debate on bilingual education and
language rights in the United States. Contributions to the Sociology of Language, 67,
159-159.
Cummins, J. (2000). Language, power, and pedagogy: Bilingual children in the crossfire.
Tonawanda, NY: Multilingual Matters.
Darling-Hammond, L. (2000). Teacher quality and student achievement. Education Policy
Analysis Archives, 8, 1.
Darling-Hammond, L. (2010). Teacher education and the American future. Journal of Teacher
Education, 61(1-2), 35-47.
De Jong, E. J., & Harper, C. A. (2005). Preparing mainstream teachers for English-language
learners: Is being a good teacher good enough? Teacher Education Quarterly, 32(2), 101-
124.
131
Dehejia, R. H., & Wahba, S. (2002). Propensity score-matching methods for nonexperimental
causal studies. Review of Economics and statistics, 84(1), 151-161.
Education Week (2011). English language learners. Retrieved from
http://www.edweek.org/ew/issues/english-language-learners/.
Fiester, L. (2010). Early warning! Why reading by the end of third grade matters. Baltimore,
MD: Annie E. Casey Foundation.
Flores, B. B., Sheets, R. H., & Clark, E. R. (Eds.). (2011). Teacher preparation for bilingual
student populations: Educar para transformar. New York: Routledge.
Francis, D. J., Lesaux, N., & August, D. (2006). Language of instruction. In D. August & T.
Shanahan (Eds.), Developing literacy in second-language learners: Report of the National
Literacy Panel on language-minority children and youth (pp. 365-413). Mahwah, NJ:
Erlbaum.
Gándara, P., & Hopkins, M. (2010). Forbidden language: English learners and restrictive. New
York: Teachers College Press.
Gándara, P., & Orfield, G. (2010). A return to the" Mexican room": The segregation of Arizona's
English learners. Cambridge, MA: Civil Rights Project/Proyecto Derechos Civiles.
Gándara, P., Maxwell-Jolly, J., & Driscoll, A. (2005). Listening to teachers of English language
learners: A survey of California teachers' challenges, experiences, and professional
development needs. Stanford, CA: Policy Analysis for California Education.
Gándara, P., Maxwell-Jolly, J., Garcia, E., Asato, J., Gutierrez, K., Stritikus, T., & Curry, J.
(2000). The initial impact of Proposition 227 on the instruction of English learners.
Davis, CA: Linguistic Minority Research Institute
132
Gándara, P., Rumberger, R., Maxwell-Jolly, J., & Callahan, R. (2003). English learners in
California schools: Unequal resources, unequal outcomes. Education Policy Analysis
Archives, 11(36), 1-54.
García, O. (2009). Emergent bilinguals and TESOL: What's in a name? Tesol Quarterly, 43(2),
322-326.
Genesee, F., Lindholm-Leary, K., Saunders, W., & Christian, D. (2005). English language
learners in US schools: An overview of research findings. Journal of Education for
Students Placed at Risk, 10(4), 363-385.
Gersten, R., Baker, S. K., Shanahan, T., Linan-Thompson, S., Collins, P., & Scarcella, R. (2007).
Effective literacy and English language instruction for English learners in the elementary
grades: A practice guide. Washington, DC: National Center for Education Evaluation
and Regional Assistance, Institute of Education Sciences.
Gil, L., & Bardack, S. (2010). Common assumptions vs. the evidence: English language learners
in the United States--A reference guide. Washington, DC: American Institutes for
Research.
Glaser, B. G. (1978). Theoretical sensitivity: Advances in the methodology of grounded theory.
Mill Valley, CA: Sociology Press.
Glazerman, S., Levy, D. M., & Myers, D. (2003). Nonexperimental versus experimental
estimates of earnings impacts. The Annals of the American Academy of Political and
Social Science, 589(1), 63-93.
Goldenberg, C. (2008). Teaching English language learners: What the research does-and does
not-say. American Educator, 32(2), 8–44.
133
Gomez, L., Freeman, D., & Freeman, Y. (2010). Dual language education: A promising 50-50
model. Bilingual Research Journal, 29, 145-164.
Grant, R., Cook, H. G., & Phakiti, A. (2011). Relationships between language proficiency and
mathematics achievement. Wisconsin: WIDA Consortium.
Greene, J. (1998). A meta-analysis of the effectiveness of bilingual education. Austin, TX:
University of Texas, Thomas Rivera Policy Institute.
Guo, S., & Fraser, M. W. (2014). Propensity score analysis (Vol. 12). Thousand Oaks, CA: Sage
Publishing.
Heckman, J., Ichimura, H., & Todd, P. (1997). Matching as an econometric evaluation estimator:
Evidence from evaluating a job-training programme. Review of Economic Studies, 64(4),
261–294.
Heckman, J. J., Ichimura, H., & Todd, P. (1998). Matching as an econometric evaluation
estimator. The Review of Economic Studies, 65(2), 261-294.
Hernandez, D. J. (2011). Double jeopardy: How third-grade reading skills and poverty influence
high school graduation. Baltimore, MD: Annie E. Casey Foundation.
Ho, D. E., Imai, K., King, G., & Stuart, E. A. (2007). Matching as nonparametric preprocessing
for reducing model dependence in parametric causal inference. Political Analysis, 15(3),
199-236.
Horwitz, E. K. (1999). Cultural and situational influences on foreign language learners' beliefs
about language learning: A review of BALLI studies. System, 27(4), 557-576.
Ingram, H., & Schneider, A. L., & deLeon, P. (2007). Social construction and policy
design. Theories of the Policy Process, 2, 93-126.
134
Johnson, B., & Christensen, L. (2012). Educational research: Quantitative, qualitative, and
mixed approaches. Thousand Oaks, CA: Sage.
Johnson, R.B., & Turner, L.A. (2003). Data collection strategies in mixed methods research. In
A. Tashakkori & C. Teddlie (Eds.), Handbook of mixed methods in social and behavioral
research (pp. 297-319). Thousand Oaks, CA: Sage.
Katz, S. R. (1999). Teaching in tensions: Latino immigrant youth, their teachers and the
structures of schooling. Teachers College Record, 100, 809–840.
Keck, C., Iberri-Shea, G., Tracy-Ventura, N., & Wa-Mbaleka, S. (2006). Investigating the
empirical link between task-based interaction and acquisition. Synthesizing Research on
Language Learning and Teaching, 13, 91.
Khan, S., & Jerolmack, C. (2013). Saying meritocracy and doing privilege. The Sociological
Quarterly, 54(1), 9-19
Koedel, C., Li, D., Polikoff, M. S., Hardaway, T., & Wrabel, S. L. (2017). Mathematics
Curriculum Effects on Student Achievement in California. AERA Open, 3(1), 1-22.
Krashen, S. (1981). Second language acquisition and second language learning. Oxford, UK:
Pergamon.
Lareau, A. (1987). Social class differences in family-school relationships: The importance of
cultural capital. Sociology of Education, 60(2), 73-85.
Lau vs. Nichols. (1974). 414 US 563, 94 S
Le, Q.T. & Polikoff, M.S. (2017). English language development curriculum effects on English
proficiency in Texas. Manuscript submitted for publication.
Lechner, M. (2000a). An evaluation of public sector sponsored continuous vocational training
programs in East Germany. Journal of Human Resources, 35(2), 347-375.
135
Lechner, M. (2000b). A note on the common support problem in applied evaluation studies
(Discussion Paper). St. Gallen, Switzerland: University of St. Gallen.
Lechner, M. (2001). Identification and estimation of causal effects of multiple treatments under
the conditional independence assumption. In M. Lechner & F. Pfeiffer (Eds.),
Econometric evaluation of labour market policies (pp.43-58). Heidelberg, Germany:
Physica.
Lincoln, Y.S. & Guba, E.G. (1985). Naturalistic Inquiry. Newbury Park, CA: Sage Publications.
Loeb, S., Soland, J., & Fox, L. (2014). Is a good teacher a good teacher for all? Comparing
value-added of teachers with their English learners and non-English learners. Educational
Evaluation and Policy Analysis, 36(4), 457-475.
Lucas, T., Villegas, A. M., & Freedson-Gonzalez, M. (2008). Linguistically responsive teacher
education: Preparing classroom teachers to teach English language learners. Journal of
Teacher Education, 59(4), 361-373.
Lunt, M. (2014). Selecting an appropriate caliper can be essential for achieving good balance
with propensity score matching. American Journal of Epidemiology, 179(2), 226-235.
Lyster, R. (2004). Research on form-focused instruction in immersion classrooms: Implications
for theory and practice. Journal of French Language Studies, 14(3), 321-341.
MacSwan, J. (2017). A multilingual perspective on translanguaging. American Educational
Research Journal, 54(1), 167-201.
McLaughlin, B. (1992). Myths and misconceptions about second language learning: What every
teacher needs to unlearn. Santa Cruz, CA: National Center for Research on Cultural
Diversity and Second Language Learning.
136
Menken, K., & Antunez, B. (2001). An overview of the preparation and certification of teachers
working with limited English proficient (LEP) students. Washington, DC: National
Clearinghouse for Bilingual Education.
Menken, K., & Solorza, C. (2014). No child left bilingual: Accountability and the elimination of
bilingual education programs in New York City schools. Educational Policy, 28(1), 96-
125.
Merriam, S. B., & Tisdell, E. J (2009). Qualitative research: A guide to design and
implementation. San Francisco, CA: Jossey-Bass.
Meyer, J. W., & Rowan, B. (1978). The structure of educational organizations. In J. W. Meyer &
Associates (Eds.), Environments and organizations (pp. 78-109). San Francisco, CA:
Jossey-Bass Publishers.
Michelmore, K., & Dynarski, S. (2017). The Gap Within the Gap: Using Longitudinal Data to
Understand Income Differences in Educational Outcomes. AERA Open, 3(1).
Miles, M. B., & Huberman, A. M. (1994). Qualitative data analysis: An expanded sourcebook.
Boston, MA: Sage.
MindPlay. (n.d.). Student programs. Retrieved from http://mindplay.com/student-
programs/virtual-reading-coach/.
Mora-Flores, E. (2011). Connecting content and language for English language learners.
Huntington Beach, CA: Shell Education.
Morita-Mullaney, T. (2017). Borrowing legitimacy as English learner (EL) leaders: Indiana’s 14-
year history with English language proficiency standards. Language Testing, 34(2), 241-
270.
137
Morris, L. (1994). Dangerous classes: The underclass and social citizenship. New York:
Routledge.
Morse, J.M. (2003). Principles of mixed methods and multimethod research design. In A.
Tashakkori & C. Teddlie (Eds.), Handbook of mixed methods in social and behavioral
research (pp. 189-208). Thousand Oaks, CA: Sage.
Namey, E., Guest, G., Thairu, L., & Johnson, L. (2008). Data reduction techniques for large
qualitative data sets. In G. Guest & K.M. McQueen (Eds.), Handbook for team-based
qualitative research (pp. 137-161). Lanham, MD: AltaMira Press.
National Center for Education Statistics. (2011). English language learners. Retrieved from
http://nces.ed.gov/programs/coe/indicator_cgf.asp.
National Center for Education Statistics. (2016). English language learners. Retrieved from
http://nces.ed.gov/programs/coe/indicator_cgf.asp.
National Clearinghouse for English Language Acquisition. (2008). English learner resources.
Retrieved from http://www.ncela.us/data/dataclearinghouse.
National Council of Teachers of English. (2008). English language learners: A policy research
brief. Urbana, IL: Author.
National Education Association. (2008). English language learners face unique challenges.
Washington, DC: Author.
Newton, L. (2005). It is not a question of being anti-immigration: Categories of deservedness in
immigration policy making. In A.L. Schneider & H.M. Ingram (Eds.), Deserved and
entitled: Social constructions and public policy (pp.139-167). Albany, NY: State
University of New York Press.
138
Nicholson-Crotty, S., & Meier, K. J. (2005). From perception to public policy: Translating social
constructions into policy designs. In A.L. Schneider & H.M. Ingram (Eds.), Deserved
and entitled: Social constructions and public policy (pp. 223-242). Albany, NY: State
University of New York Press
Nolle, K. L., Guerino, P., Dinkes, R., & Chandler, K. (2007). Crime, violence, discipline, and
safety in US public schools. Findings from the School Survey on Crime and Safety: 2005-
06. Washington, DC: National Center for Education Statistics.
Norris, J. M., & Ortega, L. (2000). Effectiveness of L2 instruction: A research synthesis and
quantitative meta ‐analysis. Language learning, 50(3), 417-528.
O’Brien, G. (2007). The instructional features across three different approaches to oral English
language development instruction. Unpublished doctoral dissertation.
O’Day, J. (2009). Good instruction is good for everyone—or is it? English language learners in a
balanced literacy approach. Journal of Education for Students Placed at Risk, 14(1), 97-
119.
Orfield, G. (2001). Schools more separate: Consequences of a decade of resegregation.
Cambridge, MA: Civil Rights Project at Harvard University.
Orfield, G., & Lee, C. (2006). Racial transformation and the changing nature of
segregation. Cambridge, MA: Civil Rights Project at Harvard University.
Parrish, T., Merickel, A., Perez, M., Linquanti, R., Socias, M., Spain, A., & Delancey, D.
(2006). Effects of the implementation of Proposition 227 on the education of English
learners, K-12: Findings from a five-year evaluation. Palo Alto, CA: American Institutes
for Research.
139
Porter, R. E., Fusarelli, L. D., & Fusarelli, B. C. (2015). Implementing the common core: How
educators interpret curriculum reform. Educational Policy, 29, 111–139.
Porter, R. P. (1990). Forked tongue: The politics of bilingual education. New York: Basic.
Raynor Jr, W. J. (1983). Caliper pair-matching on a continuous variable in case-control
studies. Communications in Statistics-Theory and Methods, 12(13), 1499-1509.
Rios-Aguilar, C., & Gándara, P. (2012). (Re) conceptualizing and (re) evaluating language
policies for English language learners: the case of Arizona. Language Policy, 11(1), 1-5.
Rios-Aguilar, C., & Gandara, P. (2012). Horne v. Flores and the future of language policy.
Teachers College Record, 114(9), 1-13.
Robitaille, D. F., & Travers, K. J. (1992). International studies of achievement in mathematics.
In D. A. Grouws (Ed.), Handbook of research in mathematics teaching and learning (pp.
687–709). New York, NY: Macmillan.
Rolstad, K., Mahoney, K., & Glass, G. V. (2005). The big picture: A meta-analysis of program
effectiveness research on English language learners. Educational Policy, 19(4), 572-594.
Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in
observational studies for causal effects. Biometrika, 70(1), 41–55.
Rosetta Stone. (n.d.). K-12 products. Retrieved from http://www.rosettastone.com/k12/home/.
Rossell, C. H., & Baker, K. (1996). The educational effectiveness of bilingual
education. Research in the Teaching of English, 7-74.
Rubin, D. B. (2001). Using propensity scores to help design observational studies: application to
the tobacco litigation. Health Services and Outcomes Research Methodology, 2(3), 169-
188.
Saldaña, J. (2013). The coding manual for qualitative research. Thousand Oak, CA: Sage.
140
Samson, J. F., & Collins, B. A. (2012). Preparing all teachers to meet the needs of English
language learners: Applying research to policy and practice for teacher
effectiveness. Washington, DC: Center for American Progress.
Saunders, W. & O’Brien, G. (2006) Oral language. In Genesee, F., Lindholm-Leary, K.,
Saunders, W., & Christian, D. (Eds.), Educating English language learners: A synthesis
of research. Cambridge, MA: Cambridge University Press.
Saunders, W. M., Foorman, B. R., & Carlson, C. D. (2006). Is a separate block of time for oral
English language development in programs for English learners needed? The Elementary
School Journal, 107(2), 181-198.
Saunders, W., Goldenberg, C., & Marcelletti, D. (2013). English Language Development:
Guidelines for Instruction. American Educator, 37(2), 13.
Scarcella, R. C. (2003). Accelerating academic English: A focus on English language learners.
Oakland, CA: Regents of the University of California.
Schleppegrell, M. J. (2004). The language of schooling: A functional linguistics perspective.
Mahwah, NJ: Lawrence Erlbaum.
Schneider, A. L., & Ingram, H. (2008). Social constructions in the study of public policy. In J.A.
Holstein & J.F. Gubrium (Eds.), Handbook of constructionist research (pp. 189-211).
New York, NY: The Guilford Press.
Schneider, A., & Ingram, H. (1993). Social construction of target populations: Implications for
politics and policy. American political science review, 87(2), 334-347.
Shadish, W. R., & Steiner, P. M. (2010). A primer on propensity score analysis. Newborn and
Infant Nursing Reviews, 10(1), 19-26.
141
Shadish, W.R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental
designs for generalized causal inference. Independence, KY: Wadsworth Cengage
learning.
Shenton, A. K. (2004). Strategies for ensuring trustworthiness in qualitative research
projects. Education for information, 22(2), 63-75.
Slama, R. B. (2014). Investigating whether and when English learners are reclassified into
mainstream classrooms in the United States: A discrete-time survival analysis. American
Educational Research Journal, 51(2), 220-252.
Slavin, R. E., Madden, N., Calderón, M., Chamberlain, A., & Hennessy, M. (2011). Reading and
language outcomes of a multiyear randomized evaluation of transitional bilingual
education. Educational Evaluation and Policy Analysis, 33(1), 47-58.
Smith, J. A., & Todd, P. E. (2005). Does matching overcome LaLonde's critique of
nonexperimental estimators? Journal of Econometrics, 125(1), 305-353.
Snow, M. A., & Katz, A. (2010). English language development: Foundations and
implementation in kindergarten through grade five. In California Department of
Education (Ed.), Improving education for English learners: Research-based approaches,
(pp. 83-148). Sacramento, CA: California Department of Education.
Spada, N., and P. M. Lightbown. (2008). Form-focused instruction: Isolated or integrated?
TESOL Quarterly, 42(2), 181–207.
Stein, Carol Stuen, Douglas Carnine, Roger M. Long, M. (2001). Textbook evaluation and
adoption. Reading & Writing Quarterly, 17(1), 5-23.
142
Steiner, P. M., & Cook, D. (2013). Matching and propensity scores. In T.D. Little (Ed.), The
Oxford handbook of quantitative methods, (pp. 237-255). New York, NY: Oxford
Univeristy Press.
Stevens, R. A., Butler, F. A., & Castellon-Wellington, M. (2000). Academic language and
content assessment: Measuring the progress of ELLs (CRESST Tech. Rep. No. 552). Los
Angeles: University of California, National Center for Research on Evaluation,
Standards, and Student Testing.
Stuart, E. A. (2010). Matching methods for causal inference: A review and a look
forward. Statistical science: a review journal of the Institute of Mathematical
Statistics, 25(1), 1-21.
Texas Education Agency. (2008). 2008 adequately yearlypProgress (AYP) guide: For Texas
public school districts and campuses. Retrieved from
https://rptsvr1.tea.texas.gov/ayp/2008/guide.pdf
Texas Education Agency. (2011). Textbook funding and adoption process gets overhauled.
Retrieved from
http://www.tea.texas.gov/WorkArea/DownloadAsset.aspx?id=2147502085
Texas Education Agency. (2015). Interpreting assessment reports. Retrieved from
http://tea.texas.gov/student.assessment/interpguide/
Texas Education Agency. (2011). Reports and Data. Retrieved from
http://tea.texas.gov/Reports_and_Data/.
Texas Education Agency. (2012). Texas English language learners portal. Retrieved from
http://www.elltx.org/certification.html.
143
Texas Education Agency. (2013). School finance 101: Funding of Texas public schools.
Retrieved from file:///Users/qtienle/Downloads/SF101manual_Jan2013.pdf.
Texas Education Agency. (2014). Reports and data. Retrieved on October 1, 2015 from
http://tea.texas.gov/Reports_and_Data/.
Texas Education Agency. (2015). A Brief Overview of the Adoption Process. Retrieved from
http://tea.texas.gov/interiorpage.aspx?id=2147485614.
Texas Education Agency. (2017). Adequate yearly progress. Retrieved from
http://tea.texas.gov/ayp/.
Texas Senate Bill 6. (2011). 82
nd
Texas Legislature, 1
st
Session.
Tonry, M. (1993). Sentencing commissions and their guidelines. Crime and Justice, 17, 137-195.
Tyack, D. B. (1974). The one best system: A history of American urban education. Cambridge,
MA: Harvard University Press.
United States Department of Education. (2013). SY 2012-2013 CSPR part 1. Retrieved from
http://www2.ed.gov/admins/lead/account/consolidated/sy12-13part1/index.html.
United States Department of Education. (2016). Non-regulatory guidance: English learners and
Title III of the Elementary and Secondary Education Act (ESEA), as amended by the
Every Student Succeeds Act (ESSA). Retrieved from
https://www2.ed.gov/policy/elsec/leg/essa/essatitleiiiguidenglishlearners92016.pdf
Valentino, R. A., & Reardon, S. F. (2015). Effectiveness of four instructional programs designed
to serve English learners: Variation by ethnicity and initial English
proficiency. Educational Evaluation and Policy Analysis, 37(4), 612-637.
Valenzuela, A. (1999). Subtractive schooling: U.S.-Mexican youth and the politics of caring.
New York: State University of New York Press.
144
Webb, E. J., Campbell, D. T., Schwartz, R. D. D., Sechrest, L., & Grove, J. B. (1981).
Nonreactive measures in the social sciences (2nd ed.). Boston, MA: Houghton Mifflin.
Weick, K. E. (1976). Educational organizations as loosely coupled systems. Administrative
Science Quarterly, 21(1), 1-19.
Werts, A. B., & Brewer, C. A. (2015). Reframing the study of policy implementation: Lived
experience as politics. Educational Policy, 29(1), 206-229.
Willig, A. C. (1985). A meta-analysis of selected studies on the effectiveness of bilingual
education. Review of Educational Research, 55(3), 269-317.
Wilson, J., & Komba, S. C. (2012). The Link between English language proficiency and
academic performance: A pedagogical perspective in Tanzanian secondary
schools. World Journal of English Language, 2(4), 1.
Zeringue, J. K.,Spencer, D., Mark, J. & Schwinden, K. (2010). Influences on mathematics
textbook selection: What really matters? Paper presented at the NCTM Research Pre-
session, San Diego, CA.
145
Appendix
Interview Protocol for District Leaders
Title:___________________________________________________________________
District: ________________________________________________________________
Textbook:_______________________________________________________________
Date:___________________________________________________________________
Hello! Thank you for taking the time to participate in this interview. The purpose is
to allow us gain an improved understanding of district policies and practices pertaining to
instructional materials for English language learners.
Your answers will be kept anonymous, and your identity will remain confidential.
Before we begin, I’d like to ask for your permission to record this interview. Your
participation in this interview is voluntary, and you may stop at any time. You may also
choose to not answer specific questions. This interview will last no more than 45
minutes. Do you have any questions for me before we begin?
First, can you tell me about your role in the district and your experience in the
education field?
What services does your district provide for your ELLs? (Bilingual education, ESL,
pull-out?) How did your district go about adopting that particular program model? What
percentage in each program? What linguistic modifications might teachers use? How often
do ELLs receive focused language study? For how long?
What is the language background of the ELLs in your district?
Can you describe your district’s policy on ESL certification? (any stipends or
reimbursements?)
146
Next, I’m going to ask about the district’s curriculum adoption and development process.
1. Has your district adopted any ESL materials? Can you describe the process by which
________ was adopted?
a. What do you like about it? What do you not like?
b. How does ESL textbook adoption differ from math textbook adoption?
c. Who is involved in the selection of new curriculum materials? Is there a formal
selection committee?
d. What are your thoughts on the quality of the instructional materials your district
has for ELLs?
e. How would you define “high quality” when it comes to instructional materials for
ELLs?
f. How did you or your committee assess the quality of instructional materials for
ELLs? (consult others, online resources? Rubric or checklist?)
g. How does the number of ELLs play a role in textbook adoption?
h. What are the sources of information used in making adoption decisions?
i. How does cost factor in to the adoption decision, if at all?
147
j. What information do you wish you had in making adoption decisions? (Probe:
Do you have or wish you had data on textbook effectiveness or alignment?
k. Moving forward, would your district consider adopting ESL textbooks again?
What factors impact your decisions to adopt?
Now, I would like to talk about how textbooks are used in your district.
1. How are ESL textbooks used by teachers in your district? (Probe: How closely to
teachers follow the ESL textbook?)
a. Do teachers supplement textbooks with other materials? If so, what materials?
Teacher created materials? Lessons pulled from online or from other sources?
How did the district decide to purchase these materials?
TEKS vs. ELPS
How much autonomy do teachers have over implementing the TEKS vs. implementing the
ELPS?
Abstract (if available)
Abstract
Federal court cases such as Castañeda v. Pickard (1981) have attempted to standardize program quality for emergent bilinguals (EBs) by mandating that programs for emergent bilinguals be effective. However, little is known about how federal mandates for EBs are enacted at the local level and the extent to which ELD instructional programs are effective. This dissertation paper focuses on English language development (ELD) curricular materials because they are one important aspect of educational programming for EBs. To answer questions about the effectiveness of ELD materials and ELD adoption patterns, I use two studies. ❧ In the first study, I use local linear matching to assess the effectiveness of ELD instructional materials. I find that schools that do not purchase any ELD curricula have significantly lower English language proficiency scores relative to schools that purchase state-adopted ELD materials. These results are robust across various matching models—inverse probability weights with regression adjustment, kernel matching, and nearest neighbor matching. There is no significant difference between schools that adopt the two most popular ELD curricula—Rigby On Our Way to English and National Geographic Reach. This study suggests that EBs who attend schools that have instructional materials that explicitly foreground English language proficiency standards outperform schools that do not have such materials. ❧ To understand how federal mandates are implemented locally, I use a mixed-methods sequential explanatory design to investigate the adoption of ELD instructional materials in Texas. First, I use a multinomial logistic regression to determine whether or not EBs have equitable access to ELD materials. Second, I interview district leaders in Texas to understand the extent to which local influence explains the variation in ELD materials adoption. I find that certain district characteristics are predictive of ELD adoption patterns. In addition, I find that the social construction of EBs at the local level can explain some of the variance in adoption patterns.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Preparing English language learners to be college and career ready for the 21st century: the leadership role of secondary school principals in the support of English language learners
PDF
A case study on influences of mainstream teachers' instructional decisions and perceptions of English learners in Hawai'i public secondary education
PDF
Effective coaching of teachers to support learning for English language learners
PDF
Evaluating the efficacy of the High Point curriculum in the Coastline Unified School District using CST, CAHSEE, and CELDT data
PDF
Instructional proficiency strategies for middle school English language learners
PDF
A case study of the roles of principals in dual language immersion programs
PDF
An alternative capstone project: bridging the Latino English language learner academic achievement gap in elementary school
PDF
The role of music in the English language development of Latino prekindergarten English learners
PDF
No place like home: a three paper dissertation on K-12 student homelessness & housing affordability
PDF
Building academic vocabulary for English language learners through professional development: a gap analysis
PDF
A phenomenological study of the impact of English language learner support services on students’ identity development
PDF
Examining the relationship between knowledge, perception and principal leadership for standard English learners (SELs): a case study
PDF
The effects of open enrollment, curriculum alignment, and data-driven instruction on the test performance of English language learners (ELLS) and re-designated fluent English proficient students ...
PDF
Teach them to read: implementing high leverage pedagogical practices to increase English language learner academic achievement
PDF
Reflective practice and pre-service language teacher preparation
PDF
A community cultural capital approach: bilingual teachers' perceptions and impact in their dual language immersion classroom
PDF
The use of culturally relevant authentic materials and L1 in supporting second language literacy
PDF
Building networks for change: how ed-tech coaches broker information to lead instructional reform
PDF
Levels of interest: the effects of teachers' unions on Congress, statehouses, and schools
PDF
A comparative study of motivational orientation of elementary school English learners in a dual language immersion program and a transitional bilingual education program
Asset Metadata
Creator
Le, Quynh Tien
(author)
Core Title
English language development materials in Texas: a study of effectiveness and selection
School
Rossier School of Education
Degree
Doctor of Philosophy
Degree Program
Urban Education Policy
Publication Date
09/08/2019
Defense Date
08/10/2017
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
curriculum materials,English Language Development,English language learners,OAI-PMH Harvest,second language acquisition
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Polikoff, Morgan (
committee chair
), Marsh, Julie (
committee member
), Painter, Gary (
committee member
)
Creator Email
qtienle@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c40-426584
Unique identifier
UC11264473
Identifier
etd-LeQuynhTie-5711.pdf (filename),usctheses-c40-426584 (legacy record id)
Legacy Identifier
etd-LeQuynhTie-5711.pdf
Dmrecord
426584
Document Type
Dissertation
Rights
Le, Quynh Tien
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
curriculum materials
English Language Development
English language learners
second language acquisition