ABSTRACT 
 
Title of dissertation: THANK YOU FOR BEING A FRIEND: THE RELATIONS BETWEEN 
FRIENDSHIP NETWORK CENTRALITY, READING ACHIEVEMENT, 
AND EXECUTIVE FUNCTIONING IN ELEMENTARY SCHOOL 
STUDENTS 
 
 Casey Archer, Doctor of Philosophy, 2022 
 
Dissertation directed by: Professor David Blazar 
 Department of Teaching and Learning, Policy and Leadership 
 
Peer relationships form the bedrock for numerous developmental outcomes, 
including students? social-emotional wellness, their sense of belonging, their decision-
making, and the development of their sense of self. While educators acknowledge the 
importance of students? social-emotional well-being and their relationship building, these 
concepts are often thought of as secondary to developing students? academic achievement, 
particularly considering the oversized role of high-stakes testing in the US educational 
system. However, the divide between social interactions and academic achievement is not 
as stark as policymakers may make it seem. Indeed, research by developmental 
psychologists and education researchers has long documented that having strong peer 
relationships will allow children to thrive, including on academic development. This three-
paper dissertation aims to provide evidence demonstrating the relation between being 
connected to one?s peers, reading achievement, and the development of executive 
functioning skills for elementary-aged students, including on a sample of primarily English 
Learners. Throughout all three studies, data were collected as part of Project LEARN, a 
three-year longitudinal study that measured various components of reading development 
 
alongside executive functioning and other variables, ultimately aiming to compare reading 
trajectories for elementary-aged English Learners and English Monolinguals. 
 Paper 1, ?Peer Effects on Oral Language Comprehension in Elementary School: A 
Social Network Analysis? uses student friendship nominations across three semesters (? =
414) as well as students? oral comprehension scores. This paper asks, What is the relation 
between student centrality (i.e., being connected to one?s classroom peers) and change in 
academic achievement, as measured by oral language skills? After generating social 
networks using the friendship nominations, I calculated how central each student was within 
their classroom friendship network. Using a student fixed effects model comparing students? 
oral comprehension growth as a function of their classroom centrality, I find that students? 
predicted oral comprehension growth is significantly greater when students are more central 
within their classroom network, even after controlling for other reading variables in the 
model.  This paper supports the connection between peer relationships and academics 
(using oral language as a proxy), suggesting that reading interventions and pedagogy 
should leverage peer relationships as one way to improve student learning. 
 Paper 2, ?The Differential Relation between Friendship Centrality and Reading 
Outcomes for English Learners,? uses the same data as Paper 1, focusing on whether 
student centrality differentially predicts reading outcomes for English Learners and English 
Monolinguals. This paper asks, a) To what extent does classroom friendship centrality 
predict reading achievement gains for English Learners? and b) To what extent does the 
relationship between classroom friendship centrality and reading achievement gains differ 
between English Learners and English Monolinguals? I used a series of multiple linear 
regression models, with students nested in homeroom classrooms, to answer the two 
research questions. First, using a sample of only English Learners (? = 160), I find that 
English Learners are more likely to experience significant gains in oral language 
comprehension?but not reading comprehension?when they are more central within their 
 
classroom network. Using the full sample of students (? = 229), I find that English 
Monolinguals are significantly more likely to experience gains in reading comprehension 
when they are more central within their classroom friendship network, but there is no 
relation between friendship centrality and reading comprehension development for English 
Learners. This surprising finding, that friendship centrality predicts English Learners? oral 
language development but not their reading comprehension development, raises 
pedagogical questions about how best to support English Learners? reading outcomes. More 
research is needed, with particular attention focused on whether English Learners who are 
more connected within their classrooms experience similar levels of self-efficacy and sense 
of belonging as their connected English Monolingual peers.  
 Paper 3, ?Do Executive Functioning Skills Predict Reading Comprehension Growth?? 
uses students? executive functioning composites?consisting of inhibitory control, cognitive 
flexibility, and memory scores?and reading comprehension scores in two subsequent 
semesters. This paper asks, To what extent do students? growth in executive functioning 
predict their reading comprehension growth? Using two methods that aim to limit omitted 
variable bias?multiple regression with covariate adjustment and propensity score 
matching?I find that students? growth in executive functioning significantly predicts their 
reading comprehension growth. This study provides support that interventions targeting 
students? executive functioning may also contribute to their reading development.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
THANK YOU FOR BEING A FRIEND: 
THE RELATIONS BETWEEN FRIENDSHIP NETWORK CENTRALITY, READING ACHIEVEMENT,  
AND EXECUTIVE FUNCTIONING IN ELEMENTARY SCHOOL STUDENTS 
 
 
 
by 
 
 
Casey Archer 
 
 
 
 
Dissertation submitted to the faculty of the College of Education  
of the University of Maryland, College Park, 
in partial fulfillment of the requirements 
for the Degree of Doctor of Philosophy 
in Education Policy and Leadership  
2022 
 
 
Advisory Committee: 
 
Dissertation Chair: Professor David Blazar 
Professor Claudia Galindo 
Professor Ken Rubin 
Professor Tracy Sweet 
Professor Ana Taboada Barber 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
?2022 
Casey Archer 
All Rights Reserved 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
i i 
Dedication 
 
 Five years ago, in October 2016, I developed a severe case of endocarditis. This 
heart infection is life threatening, and the outlook was not great for me by the time the 
doctors figured out what was happening. Thankfully, Arlington has excellent heart surgeons 
and medical staff, and I progressed that semester from not being able to move or walk, to 
being able to continue my studies. I dedicate this dissertation to the medical workers who 
saved my life, as well as the following people who have gotten me across the finish line of 
this degree program. 
 First, I have the best, most supportive friends. In no particular order, I could not 
have survived without the encouragement and support of Sarah, Jamie, Grace, Gracie, 
Andrew, Theresa, Elaine, Phoebe, Sydney, Katrina, Aaron, Hillar, Liz, Marie, Shannon, 
Christina, Katherine, Alexis, Becca, Sara, Meredith, Mark, Gail, Jim, JP, and Erin. Whether 
you were texting me about school, or Big Brother, or memes about corgis, you kept me 
sane and reminded me that I could succeed. Thank you tremendously. 
 Second, I have been truly blessed with the most supportive coworkers and 
employment. AnLar has helped develop my skills as an evaluator since I started as an intern 
in summer 2018, and they have helped keep me motivated every step of the way. 
Specifically, I want to thank Sarah and Kathi for always listening and being supportive; 
Laura and Tate for encouraging me and reading countless drafts; Tara, Becki, Angelique, 
Ayaka, and Dina for being great friends; and countless coworkers for always reminding me 
that I could do this. It?s not often to find a family at work, but I?ve got that at AnLar. 
 This dissertation would absolutely not exist without the support of my colleagues. I 
spent countless hours writing with Pamela Callahan, Joel Miller, and Nick Stroup over Zoom, 
both before and throughout the pandemic. They have read every draft and have given me 
innumerable suggestions and pep talks. I can?t wait to sit with Drs. Callahan, Miller, and 
Stroup in the near future as we relive our battle wounds and celebrate that we made it. 
ii i 
 I also would not be here without the support of my advisor, Dr. David Blazar, and 
the rest of my dissertation committee?Drs. Claudia Galindo, Ken Rubin, Tracy Sweet, and 
Ana Taboada Barber. David, you read more drafts than you probably ever wanted, and I?m 
thankful for your encouragement and your feedback. You pushed me to think harder and 
write better, and I?m so grateful for you. The other committee members have all helped me 
in unique ways, ranging from supporting me when I was feeling anxious about the 
dissertation to trusting me to lead your lab, and I?m grateful for all of you. There are also a 
few professors who particularly challenged me throughout the years, and I am grateful for 
their support (Drs. Cartwright and Malen). 
 When I was in the hospital, all of my siblings?April, Ruthann, Michael, Christopher, 
and Gabby?showed up at a moment?s notice. I also have amazing aunts?Susie, Diane, and 
Linda?who have also been there with me throughout my life, and I?m thankful for each of 
you. I love all of you, and I can?t wait to celebrate with each of you in person.  
 These last two paragraphs can never convey how important the following people are 
to me, but I?m going to try. This dissertation is as much a success for my parents, Linwood 
and Lorraine Archer, as it is for me. They raised me to love reading, to be kind, to be 
inquisitive, and to never give up, and everything that I have is thanks to their love for me. I 
really do have the best parents I ever could have hoped for, and I would not trade them for 
anything in the world. They gave up everything to care for me when I was sick, and I know 
they would do it again. I look forward to calling them every day (usually multiple times per 
day), whether to talk about Survivor, or Lucie, or what new recipes we found online, and I 
just think the world of them. This dissertation is for both of you. 
 Last, I would never be here without the support of my wife, Shamika. She has gone 
above and beyond for me over the last 6.5 years while I was writing this dissertation, 
ranging from supporting me financially, caring for me when I was sick, picking up my slack 
around the house, and just being there with me every day. Being stuck at home doesn?t feel 
so bad when you have your best friend locked in the house with you, and I can?t imagine 
i v 
going through 2020 (etc.) without you by my side. I can?t wait for all of the adventures we 
have left, and maybe now I?ll actually get to relax with you rather than spending all my free 
time on school. I owe you everything. 
 Also, the biggest shout out in the world to my study partner/son/editor, Ricky, and 
his big sister, Lucie. 
 To these friends and family, and to those who are no longer with us (Nana, TR, 
Susan, Ashley, Jeffrey, Lola, Gigi, Mia), I appreciate you more than you ever could know. 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
 v 
Acknowledgment 
 
The research reported herein was supported by the Institute of Education Sciences, U.S. 
Department of Education, through Grant R305A160280 to the University 
of Maryland. The opinions expressed are those of the author and do not represent views of 
the Institute or the U.S. Department of Education. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
v i 
Table of Contents 
 
Dedication?????????????????????????????????????????????????..ii 
 
Acknowledgment?????????????????????????..????????????????????v 
 
Table of Contents????????????????????????????????????????.???....vi 
 
Introduction.??????????????????????????????????????????..?????.1 
 
Paper 1?????????????..??????????????????..????????????..???..??8 
 
Paper 2?????????????..??????????????????..????????????..?????57 
 
Paper 3?????????????..??????????????????..???????.?????..????102 
 
Conclusion???????????..??????????????????..?????????????????.127 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 1 
Introduction 
Research from multiple disciplinary perspectives has demonstrated that students? 
peers are associated with social and academic outcomes. Studies indicate, for example, that 
students who are isolated or rejected within their classroom are more likely to experience 
negative consequences, including increased aggression (i.e., Ladd & Troop-Gordon, 2003) 
and greater rates of depression or other internalizing behaviors (i.e., Lereya, Copeland, 
Costello, & Wolke, 2015). However, being more connected with peers and having supportive 
friendships can protect against the negative social effects associated with isolation or 
rejection (i.e, Bowker, Rubin, Burgess, Booth-LaForce, & Rose-Krasnor, 2006). Further, 
having friends can provide children with academic benefits including the likelihood to 
explore and understand academic concepts more deeply (i.e., Hartl et al., 2015). Studies 
also have documented that students imitate the behavior of well-liked peers, which may 
lead students to mimic study skills, motivation, and classroom participation (i.e., Cooc & 
Kim, 2016; Hoxby & Weingarth, 2005). Finally, while the social and psychological 
mechanisms underlying peer effects are debated (Harris, 2010), researchers have provided 
strong empirical evidence that classroom peers can affect both mathematics and reading 
test scores (i.e., Gottfried, 2014; Hanushek et al., 2003). 
While research has investigated the relation between peers and academic 
achievement, there are limitations to what is known based?in part?on the typical 
methodologies employed to date to understand the academic effects of peers. A key feature 
of this dissertation is that it brings together cross-disciplinary research tools and theoretical 
frameworks in order to examine components of the logic chain linking peers to increased 
social and academic outcomes. One line of research, traditionally taken up by economists, 
uses large-scale administrative datasets to connect characteristics of peers (e.g., percent of 
students in a classroom who are the same race or ethnicity) to students? own academic 
achievement (e.g., Angrist & Lang, 2004; Gottfried, 2014; Hanushek, Kain, Markman, & 
Rivkin, 2003). These studies linking peers to student outcomes have tended to focus on the 
 2 
average benefit of peers across classes or grades. By using large-scale datasets, economists 
have causally linked peers to various outcomes, yet there are tradeoffs: the assumptions 
that large-scale datasets incorporate provide less specific information about which students 
interact with whom, instead defining ?peers? broadly as students in the same class or even 
grade. These assumptions do not account for the fact that students are most likely to be 
affected by and associate with students to whom they are similar or share some identity 
marker, such as gender, race, or even important hobbies (Harris, 2020). Further, the 
literature exploring English Learners (ELs?) peer relationships?a growing student group that 
is increasingly of interest to educators and policymakers?is understudied in comparison to 
other student subgroups (Bellmore, Nishina, & Graham, 2011; Flores-Gonzalez, 2006; 
Graham & Echols, 2018), and research is needed to understand the ways that language, 
race, culture, socioeconomic status, or other variables intersect to differentially predict the 
effect of ELs? friendships on student achievement. 
Last, the relation between two constructs hypothesized to be connected with 
students? friendships?reading comprehension and executive functioning?is also crucial to 
understanding levers to improve child outcomes. There are students with below-average 
reading comprehension scores even though they exhibit average (or above average) word 
decoding abilities (Oakhill & Cain, 2017), leading researchers to investigate the connection 
between executive functioning and reading comprehension. These students with reading 
comprehension deficits demonstrate lower proficiency in various executive functioning skills, 
such as inhibition (i.e., Borella, Carretti, & Pelegrina, 2010), working memory (Cain, 2006), 
and cognitive flexibility (Cartwright et al., 2017). While researchers have investigated the 
connection between executive functioning and reading comprehension (i.e., Cartwright, 
2002; Cartwright et al., 2017; Dahlin, 2011; Garc?a-Madruga et al., 2013), the studies use 
evidence from interventions with small sample sizes that simultaneously develop students? 
executive functioning skills and their reading comprehension, leaving questions about which 
skill develops first or if the skills develop mutually. Strong evidence from a study that more 
 3 
strongly accounts for omitted variable bias is necessary to substantiate the theoretical claim 
that executive functioning leads to reading comprehension development. 
To fill these gaps in the literature, I draw on rich primary data collected as part of 
Project LEARN (Taboada Barber et al., 2016), a three-year longitudinal study that 
investigated reading development and executive functioning among elementary students?
particularly English Learners. For this dissertation, 2nd- through 5th-grade students (N=414) 
identified who were their closest friends within their homeroom classroom once per 
semester from Spring 2018 through Spring 2019. Using Social Network Analyses, I 
calculated how central each student was within their homeroom friendship network. These 
data were used to predict reading outcomes, informing the relation between student 
friendship patterns and reading development. The purpose of each dissertation paper is as 
follows: 
 
? In Paper 1, I create a student fixed-effects model to estimate the effect of being more 
central within a classroom friendship network on students? English oral language 
comprehension growth. 
 
? In Paper 2, I use multiple linear regression models to estimate the association of student 
friendship centrality with both English oral language comprehension growth and English 
reading comprehension growth. I further investigate the differential relation between 
friendship centrality and language development for English Learners and English 
Monolinguals, to uncover whether friendships are equally predictive of achievement 
regardless of language status.  
 
? In Paper 3, I use multiple linear regression as well as propensity score matching to 
estimate the relation between students? change in executive functioning and change in 
reading comprehension. 
 4 
This three-paper dissertation contributes to the research literature by using 
longitudinal student-generated friendship data to predict student reading achievement over 
time, including on a sample of students that consists largely of English Learners. Rather 
than employing assumptions about which students were friends, these student-generated 
friendship nominations allow for more specific understandings about how being connected to 
one?s classroom peers can predict student reading achievement over time. The combination 
of student friendship data collected over time with modeling techniques that aim to limit 
selection and omitted variables biases builds on the strengths of both the economics and 
social network analysis literatures on peers. The final paper in this dissertation further 
explores linkages between the two key student outcomes explored: children?s executive 
functioning skills (i.e., the skills associated with goal setting and planning ahead) and their 
reading development. Both are thought to relate to students? peer networks, either as 
predictors or outcomes, and substantiating the connection between executive functioning 
and reading comprehension will further guide intervention development associated with 
children?s relationships, reading development, and executive functioning skills. 
It has long been understood that friendships are important for students? social 
development (for a review, see Bukowski, Laursen, & Rubin, 2018). These papers provide 
evidence that friendships are also important for students? academic development, including 
for English Learners, a growing student group of interest to teachers and policy makers. 
Implications for instructional practice and policy development are discussed in each paper. 
Ultimately, I make the case throughout this dissertation that student friendships may 
support achievement, and further work should explore interventions that strengthen student 
friendship networks as one way of improving reading outcomes. 
 
 
 
 
 5 
References 
Angrist, J.D., & Lang, K. (2004). Does school integration generate peer effects? Evidence 
from Boston?s Metco program. American Economic Review, 94(5), 1613-1634.  
Bellmore, A.D., Nishina, A., Witkow, M.R., Graham, S., & Juvonen, J. (2007). The influence 
of classroom ethnic composition on same- and other-ethnicity peer nominations in 
middle school. Social Development, 16(4), 720-740. 
Borella, E., Carretti, B., & Pelegrina, S. (2010). The specific role of inhibition in reading 
comprehension in good and poor comprehenders. Journal of Learning 
disabilities, 43(6), 541-552. 
Bowker, J.C.W., Rubin, K.H., Burgess, K.B., Booth-Laforce, C., & Rose-Krasnor, L. (2006). 
Behavioral characteristics associated with stable and fluid best friendship patterns in 
middle childhood. Merrill-Palmer Quarterly, 52(4), 671-693. 
Bukowski, W.M., Laursen, B., & Rubin, K.H. (Eds.; 2018). Handbook of peer interactions, 
relationships, and groups. Second edition. New York City: The Guilford Press. 
Cain, K. (2006). Individual differences in children's memory and reading comprehension: An 
investigation of semantic and inhibitory deficits. Memory, 14(5), 553-569. 
Cartwright, K. B. (2002). Cognitive development and reading: The relation of reading-
specific multiple classification skill to reading comprehension in elementary school 
children. Journal of Educational Psychology, 94(1), 56. 
Cartwright, K.B., Coppage, E.A., Lane, A.B., Singleton, T., Marshall, T.R., & Bentivegna, T. 
(2017). Cognitive flexibility deficits in children with specific reading comprehension 
difficulties. Contemporary Educational Psychology, 50, 33-44. 
Cooc, N., & Kim, J. (2016). Peer influence on children?s reading skills: A social network 
analysis of elementary school classrooms. Journal of Educational Psychology, 
Forthcoming. 
Dahlin, K. I. (2011). Effects of working memory training on reading in children with special 
needs. Reading and Writing, 24(4), 479-491. 
 6 
Flores-Gonzalez, N. (2005). Popularity versus respect: School structure, peer groups, and 
Latino academic achievement. International Journal of Qualitative Studies in 
Education, 18, 625-642. 
Garc?a?Madruga, J. A., Elos?a, M. R., Gil, L., G?mez?Veiga, I., Vila, J. ?., Orjales, I., ... & 
Duque, G. (2013). Reading comprehension and working memory's executive 
processes: An intervention study in primary school students. Reading Research 
Quarterly, 48(2), 155-174. 
Gottfried, M.A. (2014). Peer effects in urban schools: Assessing the impact of classroom 
composition on student achievement. Educational Policy, 28(5), 607-647. 
Graham, S., & Echols, L. (2018). Race and ethnicity in peer relations research. In Bukowski, 
W.M., Laursen, B., & Rubin, K.H. (Eds.), Handbook of peer interactions, 
relationships, and groups. Second edition. New York City: Guilford Press. 
Hanushek, E.A., Kain, J.F., Markman, J.M., & Rivkin, S.G. (2003). Does peer ability affect 
student achievement? Journal of Applied Econometrics, 18(5), 527-544. 
Hartl, A.C., DeLay, D., Laursen, B., Denner, J., Werner, L., Campe, S., & Ortiz, E. (2015). 
Dyadic instruction for middle school students: Liking promotes learning. Learning & 
Individual Differences, 44, 33-39. 
Harris, D.N. (2010). How do school peers influence student educational outcomes? Theory 
and evidence from economics and other social sciences. Teachers College Record, 
112(4), 1163-1197. 
Hoxby, C.M., & Weingarth, G. (2005). Taking race out of the equation: School reassignment 
and the structure of peer effects (No. 7867). Working paper. 
Ladd, G.W., & Troop-Gordon, W. (2003). The role of chronic peer difficulties in the 
development of children?s psychological adjustment problems. Child Development, 
74(5), 1344-1367. 
 7 
Lereya, S.T., Copeland, W.E., Costello, E.J., & Wolke, D. (2015). Adult mental health 
consequences of peer bullying and maltreatment in childhood: Two cohorts in two 
countries. The Lancet Psychiatry, 2(6), 524-531. 
Oakhill, J., & Cain, K. (2017). Children with specific text comprehension problems. In K 
Cain, DL Compton, RK Parrila: Theories of Reading Development (pp. 359-378). John 
Benjamins Publishing Company. 
Taboada Barber, A. (PI), Cartwright, K., & Stapleton, L. (Co-PIs) (2016-2020). Cognitive 
and motivational contributors to reading comprehension in English Learners (ELs) 
and English Monolinguals (EMs): Different or similar growth patterns. English 
Learners Research Grants Competition, Institute of Education Sciences. 
 
 
 
  
 8 
 
 
 
 
 
 
 
 
 
 
Peer Effects on Oral Language Comprehension in Elementary School: 
A Social Network Analysis 
 
 
 
 
 
 
Paper 1 
 
 
 
 
 
 
 
 
 9 
Theoretical and empirical work from developmental frameworks describe specific 
avenues through which peers influence one?s own attitudes, behaviors, and beliefs. To what 
extent do these peer influences subsequently benefit academic performance?  
A sizeable body of research from multiple disciplinary perspectives has linked 
characteristics of students? peers to a variety of social and academic outcomes. For 
example, studies indicate that students who are isolated or rejected within their classroom?s 
social groups demonstrate an increased likelihood of negative short- and long-term 
outcomes, including increased aggression (e.g., Ladd & Troop-Gordon, 2003), increased 
misbehaviors in school (Snyder et al., 2004), and greater rates of depression or other 
internalizing behaviors later in life (e.g., Lereya et al., 2015). In contrast, having strong 
peer support systems and friendships can protect against negative social effects of isolation 
or rejection (Bowker et al., 2006), which in turn provides children with academic benefits 
including the opportunity to explore and understand academic concepts more deeply (Hartl 
et al., 2015). Economists have built on this work from developmental psychologists by 
showing how access to peers with a common identity marker (e.g., English language learner 
status, race/ethnicity, gender) can further affect both mathematics and reading test scores 
(Gottfried, 2014; Hanushek et al., 2003). At the same time, this latter work generally is 
unable to identify the social and psychological mechanisms driving these relationships 
(Harris, 2010).  
Together, the theoretical and empirical ?peer effects? literature suggest that peers 
can serve as a catalyst to behaviors and beliefs that can lead to academic success. At the 
same time, because the literature often has developed from distinct disciplinary frames and 
traditions ? namely from developmental psychologists and from economists ? that have 
conceptualized peers in different ways and also have tended to focus on different sets of 
outcomes, gaps in the literature remain. While the peer effects literature to date has made 
important discoveries about the effects of various components of schooling on student 
achievement, such as how the achievement of one?s classroom peers can affect one?s own 
 10 
learning, the studies by nature focus on classroom averages (e.g., average reading ability) 
and/or demographic percentages as predictors. These studies assume all students have 
equal effects on each other merely by being in the same classroom, which contradicts the 
finding that kids are most likely to be influenced by those with whom they relate and 
interact frequently (Harris, 2010). 
This study contributes to our understanding of the benefits of student peers by 
utilizing student-level data that were generated by the students themselves, rather than by 
making assumptions that students in the same classroom or grade interact frequently. The 
prior literature connecting peers to academic outcomes primarily uses large-scale datasets, 
with peers operationally defined as students in the same school, grade level, or classroom. 
The present study addresses this limitation by utilizing student friendship nominations, such 
that students? friends are identified and used to predict student outcomes. Additionally, this 
study utilizes student panel data collected across three semesters to connect exogenous 
changes in network centrality to changes in reading achievement. Furthermore, the existing 
studies connecting peer social networks to reading outcomes either utilize a pre- and post-
design to measure change, or rely on students? grades, which are not largely incomparable 
across contexts. Student friendship data, which are often challenging to collect due to 
concerns related to student privacy as well as the ethics of asking students who their friends 
are (and are not), provide a unique dataset with which to investigate connections among 
friendship networks and reading outcomes.  
 
Motivation and Theoretical Frameworks: 
Interdisciplinary Perspectives on Why Peers and Friends Matter 
This section presents empirical and theoretical work documenting connections 
between peers or friendships and academics. Additionally, this section contains a brief 
discussion of Social Network Analyses, the primary methodological tool used in this study.  
 11 
Prior work has explored the ways in which groups can impact classroom 
engagement, motivation, and academic success. Further, group processes as well as dyadic 
relationships are associated with language development, particularly for younger children. 
For this study, group dynamics, dyadic relationships/friendships, and language development 
form the theoretical framework connecting friendship networks to reading outcomes. 
Given the complexity of relationships and group dynamics, it is important to 
distinguish among a few key terms that are used throughout this paper. In line with 
previous literature, ?peers? refers to other children within the same schools, grades, or 
classrooms (see Kindermann & Gest, 2018). ?Peer? does not assume any close relationships 
between two children; one?s peers are merely the students in a similar setting, and it can be 
assumed that children interact with their peers regularly through school or classroom 
activities. Among one?s peers, children form ?friends,? and these are others with whom 
children develop mutual trusting and caring relationships (i.e., Bagwell & Bukowski, 2019; 
Bagwell & Schmidt, 2011; Laursen & Hartup, 2002). By definition, friendships are 
reciprocated relationships in with both children feel an affinity towards each other. However, 
it is possible for one child to believe a friendship exists while the other does not reciprocate 
these feelings. In this instance, one child perceives another child is a friend, while the other 
child merely considers them a peer. Given the egocentric (i.e., student-driven) data 
collection metrics, in which students identify who their friends are, these directed, non-
reciprocated friendships are possible and common in the data. From the perspective of the 
child declaring their friends, this friendship is a reality; thus, these non-reciprocated 
friendships are still considered ?friends? in this paper. Compiling all students into a social 
network, in which all students are linked to their identified friends, results in ?student 
friendship networks.? These networks consist of all children in a classroom, with friends (or 
perceived friends) linked by ties. Students who did not identify each other as friends are not 
linked in these student friendship networks, but by nature of being a part of the classroom 
roster, these children are co-located in the network as peers. These underlying student 
 12 
friendship networks are represented using sociograms in a 2-dimensional plane (see Figure 
1). 
 
Peers, engagement, and academics 
Peers have important effects on others? growth. Social Cognitive theory (Bandura, 
1986; Hinde & Stevenson-Hinde, 1987) posits that peers influence each other?s behaviors, 
attitudes, and beliefs after observing how others behave as well as the consequences of 
those particular behaviors. If peers notice that behaviors or beliefs are rewarded, then they 
become more likely to endorse or subscribe to those particular actions. Within a classroom, 
children may be more likely to model the behaviors of students who are popular or well 
liked in an attempt to gain social standing themselves (Cillessen & Rose, 2005; Rambaran 
et al., 2017). Others have added to that framework, noting that if peers perceive rewards 
for behaviors or beliefs, then they become more likely to hold those beliefs or mimic those 
actions, especially pertaining to students? self-efficacy and motivation (Chen & Pajares, 
2010; Grigg et al., 2018; Usher et al., 2019). 
 Popular peers are also more likely to influence the norms of behavior within a 
classroom (Bukowski, Brendgen, & Vitaro, 2007). This process, known as peer socialization, 
can affect the behaviors and attitudes that guide a classroom atmosphere, such as 
motivation, engagement, or even achievement. Peers at the top of the social hierarchy set 
expectations that others follow, possibly in an attempt to ascend in the hierarchy or to avoid 
rejection (Rodkin & Ryan, 2012).  Popular students who promote behaviors that are 
associated with academic success (such as participation in class, studiousness, etc.) may 
lead to different academic outcomes than in classes where the popular students promote 
behaviors that negatively impact academic engagement.  
Students? perceptions of belonging can also have impacts on academic outcomes. 
Students who feel that they belong at school are more likely to put effort into their 
schoolwork and enjoy being in the classroom (Goodenow, 1993). Additionally, feelings of 
 13 
belonging have been associated with student achievement for early adolescents (Roeser, 
Midgley, & Urdan, 1996). Among adolescents, participating in school activities is associated 
with increased feelings of school belonging (Dotterer, McHale, & Crouter, 2007), particularly 
for Latinx students (Brown & Evans, 2002). Additionally, developing feelings of school 
belonging significantly mediated the connection between having friends in school activities 
and increases in academic achievement, suggesting school belonging is an intermediary 
step that develops after establishing friends in activities (Knifsend et al., 2018). However, 
few studies have examined this relationship for students in the early elementary school 
grades, in which students experience the most growth in reading understanding. 
 Social isolation can have detrimental effects on students? socio-emotional 
development. Lereya et al. (2015) demonstrated that students who do not have friends over 
long periods are more likely to develop internalizing problems, such as depression or 
anxiety. These students may also be less engaged within the classroom, since internalizing 
problems can lead students to disassociate and not focus. In contrast, having friends is 
associated with greater academic engagement (Wentzel & Caldwell, 1997), which might 
explain why students who feel more connected within their classroom network experience 
greater academic growth over time.  Again, much of the extant literature focuses on 
students in secondary schools, at times when socialization is crucial to students? transition 
from childhood to adolescence and early adulthood. However, this dissertation would shed 
light on the extent to which these relationships exist for pre-adolescent students in the 
elementary grades. 
 
Language Development and Peers 
This study utilizes oral language comprehension as an indicator of academic success. 
Reading comprehension underlies achievement across the curriculum, including 
mathematics (Grimm, 2008), science (Reed et al., 2017), and social studies (Vaughn et al., 
2013). Oral language is a significant longitudinal predictor of reading achievement 
 14 
(Babayigit et al., 2020; Silverman et al., 2020). Thus, identifying contributors to oral 
language comprehension has implications for achievement across the curriculum. Further, 
oral language comprehension is associated with success on high-stakes testing, particularly 
for younger students (Nyman, 2013). This study utilizes one such understudied variable, 
student centrality within friendship networks, because of the role that peer interactions and 
friendships have for children?s oral language development (Kory-Westlund & Breazeal, 
2019; Swain et al., 2002).  
Research has demonstrated that peers can be leveraged to improve each other?s 
reading skills in the elementary grades. The Peer Assisted Learning Strategies (PALS) 
instructional technique involves pairing students together who can coach each other and 
develop their reading levels (Fuchs et al., 2000). Randomized control trials have 
demonstrated that PALS improves students? word decoding and oral language 
comprehension (Lemons et al., 2014). This intervention utilizes the social contagion model, 
whereby peers learn from each other and mimic each other?s reading behaviors (Cooc & 
Kim, 2016). Thus, interventions capitalizing on peer behaviors, support, and capacity to 
work together can lead to reading developments for elementary students. 
Given the research, I hypothesize that greater opportunities for peer interactions, as 
well as increased efficacy and sense of belonging developed through greater social 
connectedness with friends, will foster children?s oral language development. Specifically, 
students will show increased growth in oral comprehension in semesters when they are 
more central within their classroom friendship networks, after controlling for prior 
achievement. 
 
Peer Effects Methodology 
While researchers have examined peer effects using various lenses, the conclusions 
linking peers and academic outcomes are limited based on the quality of existing data. One 
reason for limited conclusions linking students? peers to academic performance is the way in 
 15 
which peers have been operationalized by researchers from various disciplines. One line of 
research, traditionally employed by economists, uses large-scale administrative datasets to 
connect characteristics of peers (e.g., the percent of students in a classroom who are the 
same race or ethnicity) to students? own academic achievement (e.g., Angrist & Lang, 2004; 
Gottfried, 2014; Hanushek et al., 2003). Although students in the same classroom or grade 
are peers in the broader sense, administrative data cannot capture which students are 
interacting or who considers themselves friends. Using large-scale datasets to reach 
conclusions about the effect of peers on various outcomes is limited given that students are 
most likely to affect others with whom they share an affinity or identity marker (Harris, 
2010). Children?or adults?do not weigh all ?peers? equally when making decisions about 
behaviors. While peers are broadly defined, these studies do not allow researchers to 
consider the influence of students? friendships on their achievement. 
A second methodological approach asks students to declare who their friends are or 
with whom they interact. This approach allows researchers to create ?social networks? that 
are thought to provide more valid representations of the underlying classroom or grade-
level relationships within a closed system of students (e.g., Carolan, 2014). Researchers can 
analyze these networks to provide clearer quantitative representations about the 
connections between friendships and academic outcomes. However, empirical analysis in 
education policy studies has focused on creating social networks, rather than connecting 
them to student outcomes.  
Developmental psychologists have utilized sociometric tools throughout the twentieth 
century to study group relationships and dynamics empirically (see Cillessen & Bukowski, 
2018 for a review). These tools allow researchers to have a primary focus on peer status 
and various social/ecological phenomena related to friendships and peer networks such as 
aggression (e.g., Parkhurst & Hopmeyer, 1998), rejection (e.g., Parker & Asher, 1987), or 
popularity (e.g., Cillessen & Mayeux, 2004). Researchers can use sociometric tools and 
social network analyses to quantify various components of the networks and incorporate the 
 16 
measures into traditional statistical models to predict outcomes. These methods provide 
specific information about which students are interacting with whom, which can strengthen 
claims made about the effect of one?s peers on outcomes, rather than defining peers 
broadly as those students who share a classroom or school (Daly, 2010). Further, these 
methods allow researchers to study the effect of friendships on identified outcomes, since 
students can identify which of their peers they consider friends. While sociometric tools 
allow researchers to collect fine-grained student-level data on who one?s friends are, 
children?s friendships (and their accompanying friendship nominations) are often in flux and 
can vary wildly over time. However, within elementary students? classroom networks, 
children often retain stable friendship networks (Cooc & Kim, 2016; Ryan & Shim, 2012). 
 
Social Network Analyses 
This study makes use of Social Network Analyses as a way to operationalize 
students? positions within their classroom friendship network. Social Network Analysis (SNA) 
is a statistical procedure used to quantify the underlying social networks of contained 
groups (Carolan, 2014; Daly, 2010). Researchers analyze a closed group of individuals and 
the social ties among them in order to reproduce a visual image of the network and analyze 
the network using various quantitative methods (Carolan, 2014). Social network analysis 
enables researchers to model the influence that social processes (i.e., relationships, group 
dynamics) have on other variables (Daly, 2010; Davison, 2019). For example, social 
network analyses would allow researchers to answer questions such as: is being more or 
less connected with others in the network associated with other outcomes? Do individual 
measures predict whether a pair of students will become friends? To what extent does the 
structure of a group network predict whether the group overall will demonstrate increases in 
other outcomes? While SNA is becoming more common across research fields, it is currently 
underutilized in education research (Daly, 2010).  
 17 
Within SNA, actors (i.e., students) are the individuals forming relationships. The ties 
(i.e., connections) among actors can vary; examples include interactions, relationships, 
advice-seeking, or any other sort of connection among the network participants (Carolan, 
2014). The actors can also vary among specific individuals, groups of individuals, or 
institutions (Carolan, 2014). Network information can be used in conjunction with peer 
effects models to give researchers a better understanding of the social processes taking 
place within classrooms or schools. Furthermore, these social relationships can be used to 
leverage student outcomes in educational systems. Therefore, understanding social network 
analyses may lead to improved educational outcomes for students when policy is informed 
by this research (Daly, 2010). 
In addition, social network models allow researchers to gain a broader perspective of 
how one?s social environment can shape one?s own attributes and outcomes. Whereas peer 
effects models can account for how dyadic (i.e., two-student) relationships may lend some 
predictive power to various educational outcomes, such as test scores or delinquent 
behaviors, incorporating social network information into statistical models can develop a 
greater understanding of the structure of a network and one?s own position within that 
social group (Burk et al., 2007). Creating social network models allows researchers to 
predict how relationships might change or evolve over time, as well as how other actors 
may influence one?s actions or character traits (Burk et al., 2007).  
To my knowledge, only two studies to date have used student social networks to 
predict academic achievement in the general population1 (Cooc & Kim, 2016; van Rijsewijk 
et al., 2018), focusing on short-term reading outcomes or classroom grades. Cooc and Kim 
(2016) asked second- and third-grade children to identify which classroom peers they read 
with, using those nominations to identify peers. The average reading level of students? 
identified reading peers was used as a predictor for students? own reading achievement. 
 
1 A third study (Elreda et al., 2016) uses social network analysis to predict achievement but 
focuses on English Learners. This study is discussed in Paper 2. 
 18 
Ultimately, peers? reading scores predicted one?s own reading scores; however, the 
predictor was nonsignificant when controlling for a student?s own reading score. Thus, the 
authors demonstrated homophily of peers, whereby peers predict one?s own reading 
achievement as a result of peers being similar to themselves (and having similar reading 
achievement). This study improved on prior peer effects studies by incorporating study-level 
peer nominations rather than assuming all students in the same classrooms were peers of 
each other. Additionally, the study used a classroom fixed effects model to control for non-
observed teaching differences. However, the primary focus of this study was on peers? 
reading levels, demonstrating students are likely to read with others who are of similar 
reading achievement. The study also did not collect peer nominations from multiple time 
points, instead assuming that students? reading peers would be consistent across 
semesters. The current study builds on their fixed effects model but also captures data 
across time points to leverage variation in students? position in the classroom friendship 
network over time. 
A second study (van Rijsewijk et al., 2018) demonstrated a connection in student 
network centrality and academic achievement. Using early adolescents (with a mean age of 
13 years) in a school in the Netherlands, participants identified which of their peers they 
were likely to ask for help if they needed it. Classrooms with less dense help networks were 
associated with decreased academic achievement. Further, individuals who were more 
central within the help network were more likely to be high-achieving individuals. This study 
demonstrated that classroom climate was an important predictor for classroom-level 
achievement, and that students with more access to help from peers were more likely to 
achieve. While the study is similar to the present study, it uses one time point to assess the 
correlation between student network centrality and achievement. The authors acknowledge 
that more factors may be at play than just students? access to social capital, and they claim 
that one limitation was their inability to have longitudinal data to compare changes in 
network status over time. 
 19 
More research garnered from student self-created friendship networks is necessary 
to understand the extent to which these social networks are associated with changes in 
various academic outcomes, and whether the relations among constructs are maintained 
over time. Additionally, the findings from the extant studies can be strengthened using 
longitudinal data, in which student centrality is utilized as the independent variable. That 
analytic model will allow for an understanding of how student changes in network centrality 
can predict student changes in reading achievement over time. Finally, and most 
importantly, neither study asks the children to identify their friends; rather, they identify 
with whom they read or to whom they seek guidance. The current study contributes to the 
literature by providing evidence about the impact of friendships (or perceived friendships) 
on academics. While research has documented various important effects of friendships on 
social-emotional or developmental outcomes, this study will supplement those findings by 
examining academic effects of friends for elementary-aged children. 
 
Present Study 
The hypothesized logic model connecting friendships with reading achievement relies 
on developmental processes associated with peer relationships as well as classroom 
dynamics supporting student learning and engagement. There are positive effects on 
students? well-being that are associated with being connected to peers, including an 
increased sense of belonging (e.g., Dotterer, McHale, & Crouter, 2007; Goodenow, 1993; 
Roeser, Midgley, & Urdan, 1996) and an increased perception of self-worth (Hiatt, Laursen, 
Mooney, & Rubin, 2015), as well as a decreased chance of depression or other internalizing 
behaviors (Lereya et al., 2015). Due to these benefits, children who are more connected to 
peers are better able to enjoy being in the classroom and to participate in classroom 
activities. In turn, these students are better able to engage with the instructional activities 
(Hartl et al., 2015) and therefore are more adept at learning the classroom content. Outside 
 20 
of the instructional activities, students who are more connected with peers are engaging 
more frequently with others, which involves increased opportunities to communicate and 
support their oral language skill development. Given the correlation between oral language 
comprehension and reading comprehension (e.g., Babayigit et al., 2020; Silverman et al., 
2020), these children are developing their propensity to both produce and understand the 
English language. Further, within classroom networks that are characterized by many dense 
social connections, students are most likely to repeat behaviors of those students who are 
considered well-liked or popular (e.g., Cillessen & Rose, 2005; Rambaran et al., 2017), and 
if these students are also more likely to participate in the classroom and display study skills, 
others who are connected to them will similarly display these behaviors and therefore 
develop their own language skills. Ultimately, friendships allow children to meaningfully 
engage with each other and the classroom content, bolstering their oral language skills over 
time. 
According to this theory of action, students? connectivity with peers in their 
homeroom classroom will lead to increased language outcomes. Specifically, having more 
friends will improve social-emotional outcomes, including sense of belonging and self-worth. 
Additionally, students with greater support within the classroom will be more able to focus 
on the classroom material, thereby increasing their aptitude for learning. Further, students 
who are more connected will communicate more frequently with others, both during and 
outside of instructional activities, which will provide these students with greater 
opportunities to develop their oral language skills.  A combination of the social-emotional 
components and the greater ability to develop language skills will allow students with 
greater connectivity with peers to experience greater oral language outcomes over time.  
To examine whether student friendship networks have predictive power for student 
achievement, this study examines the relationship between student-created friendship 
networks and changes in academic outcomes over three semesters. The following research 
 21 
question, focused on peer friendship networks among students nested in homeroom 
classrooms, will guide this study: 
What is the relation between student centrality (i.e., being connected to one?s 
classroom peers) and change in academic achievement, as measured by oral 
language skills? 
To answer this question, I leverage a longitudinal dataset collected as part of a larger 
study that investigated various cognitive and language predictors of reading comprehension.  
This study builds from the strengths of prior work by drawing on social network data 
collected over time, paired with student fixed effects modeling strategies aimed at limiting 
sources of omitted variables bias when linking peer networks to academic outcomes. To 
understand the extent to which being central to one?s network (i.e., well-connected, as 
opposed to residing on the periphery of the network with few to no friends) predicts 
academic achievement, I leverage data from a larger study investigating reading 
development among elementary students. Over three semesters, 2nd- through 5th-grade 
students indicate which students from within their homeroom classroom are their closest 
friends. I use these data to create classroom friendship networks and to empirically 
measure each student?s network centrality. I used a student fixed effects model to leverage 
exogenous changes in network centrality and the associated changes in students? oral 
comprehension reading scores.  
 
Method 
This study is part of a larger longitudinal study, Project LEARN (Taboada Barber et 
al., 2016), examining elementary-aged English Learners? and English Monolinguals? reading 
development that occurred from fall 2016 through spring 2019. The present study 
incorporates data from twenty-four 2nd- to 5th-grade classrooms in public schools in a mid-
Atlantic US state across the transition from their 2017-2018 to 2018-2019 school years. In 
 22 
the three semesters (spring 2018, fall 2018, and spring 2019), students were asked to 
name their top five friends within their classroom. I use Social Network Analyses to quantify 
students? centrality within their peer friendship network as a predictor for reading 
achievement over time, hypothesizing that greater centrality within peer friendship 
networks would be associated with greater cognitive, linguistic, and academic growth over 
time. Given that data were collected across three semesters, I use panel data to examine 
whether changes in friendship networks can explain differences in reading growth.  
Five public schools in the mid-Atlantic US were part of the larger study and 
comprised the current sample. These schools had higher-than-average populations of 
English Learners (ELs) compared to the overall state population, which allowed for more 
powered analyses of ELs? development over time. Parents or legal guardians of students 
provided annual informed consent for their student to participate in the study; students who 
did not receive informed consent across both years of the present study were removed from 
the sample. This study was approved by the university institutional review board, and 
participants were treated in accordance with the ethical principles of the American 
Psychological Association. 
 
Participants 
In spring 2018, participants were 2nd-through 5th-grade students from the five 
elementary schools. The full sample of students who participated in the whole-class 
assessments in Spring 2018 included 765 students. To be included in the present study, 
students had to have complete social network data across all three time points (n = 414). 
Appendix 1 compares the students in the analytic sample with the full sample of students in 
the longitudinal study in Spring 2018. Of note, the students who comprise the analytic 
sample are more likely to be Hispanic, an English Learner, and/or to receive free or 
 23 
reduced-price lunch than the full analytic sample. This differential affects the external 
validity of the results and is discussed in the limitations section. 
Students in this sample were evenly distributed across grades 2 through 5 in the 
spring of 2018, with an average age of 9.20 years. A slightly greater number of participants 
were male (52.9%) than female (44.7%), although gender data were missing for some 
students (2.4%). Most of the students were Hispanic (73.7%) or Black (21.3%). Most 
students were eligible for free or reduced-price lunch in either year of the study (87.0%), 
and most students were identified by the school district as English Learners in either year of 
the study (70.5%). Table 1 provides complete descriptive data on the full sample of 
students in the present study.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 24 
Table 1 
Descriptive Statistics for the Sample 
Variable n M (SD) or % 
Grade, SY 2017-2018   
     Second 102 24.6% 
     Third 109 26.3% 
     Fourth  106 25.6% 
     Fifth 97 23.4% 
Age, Fall 2017 396 9.20 (1.17) 
Gender   
     Male 219 52.9% 
     Female 185 44.7% 
     Missing 10 2.4% 
Race/Ethnicity   
     Hispanic 305 73.7% 
     Black / African American 88 21.3% 
     White 3 0.7% 
     Asian 2 0.5% 
     Native Hawaiian / Pacific Islander 1 0.2% 
     Multi-Racial 5 1.2% 
     Missing 10 2.4% 
 
SES, Fall 2017   
     Eligible for Free or Reduced-Price Lunch 360 87.0% 
     Not eligible 38 9.2% 
     Missing 16 3.9% 
English Learner, either SY 17-18 or SY 18-19   
     Yes 292 70.5% 
     No 121 29.2% 
     Missing 1 0.2% 
Student has IEP, either SY 17-18 or SY 18-19   
     Yes 22 5.3% 
     No 381 92.0% 
     Missing 11 2.7% 
Note: Table 1 displays descriptive data on the 414 students in the sample.  
 
Measures 
Students were assessed using a battery of assessments twice per semester?once 
was an individual assessment and once was in a whole-class setting. For individual 
assessments, a Research Assistant (RA) led the child to a quiet spot in the school for a one-
hour assessment protocol, consisting of a research battery of thirteen various assessments 
 25 
each semester. For whole-class assessments, one or two RAs administered three whole-
class assessments to the students participating in the study within their classrooms. Any 
students not participating were allowed to read or meet with their teacher during that one-
hour block. For both the individual and whole-class assessments, RAs were trained each 
semester on administering and scoring the assessments. RAs were required to pass fidelity 
checks with the PI, co-PIs, or other lab staff each semester.   
The outcome variable (i.e., oral language comprehension) was administered as part 
of the individual assessment battery, and the independent variable of interest (i.e., student 
friendship network centrality) was assessed as part of the whole-class assessment. Other 
covariates (e.g., word decoding, reading comprehension, and vocabulary) were collected as 
part of the individual assessments.  
 The following sections describe the operational definitions of each of the key 
variables used in this study.  
Friendship centrality. The primary independent variable for the present study was 
student centrality within classroom-based friendship networks, which aims to characterize 
how ?central?, or crucial, each child is within the constructed social network (Carolan, 2014). 
While there are many different ways to define centrality, this study utilizes two methods: 
closeness and in-degree. Closeness involves creating a sociogram that displays student 
connections, using the students? friendship nominations, and measuring how central each 
student is within that network. In contract, in-degree is a count of the number of friendship 
nominations each student received.  
The following subsections describe the two centrality measures in detail, including 
how they are calculated and what constructs they highlight. 
Closeness. Carolan (2014) defines closeness as a function of the average distance 
between a particular node and all of the other nodes in the network. In other words, how 
connected is each student, on average, to each of the other students in the closed network. 
Thus: 
 26 
1
?????????? = ?  ?(?? , ??)
where d is the distance between Node i and every other Node j in the model (Carolan, 
2014). Using closeness centrality, a node (i.e., student) that is more central within the 
network has higher closeness than other nodes.  
In the fall and spring of each school year, students were provided with a complete 
roster from their homeroom class. The assessor asked, ?Who are your best friends in your 
class? Circle up to five friends.? Students were reminded that they could circle zero, one, 
two, three, four, or five friends from their class. They were instructed not to circle 
themselves.  
 Based on students? responses, directed social networks were created for each 
classroom at each of the three time points. Using the igraph R package (version 1.2.4.2), 
sociograms mapped students? social position on a latent space. Students who were more 
connected to their classroom networks based on the friendship nominations they indicated 
and received are positioned towards the center of the sociogram, and their less-connected 
peers appear on the periphery. See Figure 1 for a sample sociogram displaying the 
friendship network generated for one of the classrooms. Following Cruz et al. (2012), Elreda 
et al. (2016), and Li & Stone (2018), who also asked participants about who their ?best 
friends? or ?friends? were, I refer to the underlying social network as the ?classroom 
friendship network? throughout this dissertation. While not all identified relationships were 
reciprocated friendships, the friendship nominations allow me to capture students? perceived 
friendship networks that exist in each classroom. 
 
 
 
 
 
 27 
Figure 1 
Class Sociogram demonstrating the Friendship Network for a Sample Class 
 
Note: Figure 1 displays a sociogram that maps friendship nominations among students in a 
particular classroom in spring 2018. Each student is represented by a node, with numbers 
indicating the student ID. Students closer to the center of the network, such as student 
794, have increased centrality scores because they are more connected to their classroom 
peers. Student 3009, who was only nominated by one peer as a friend (and who nominated 
no friends of their own) would have the smallest centrality score from among this classroom 
network. The sociogram was created using the nwcommands Stata package (Grund, 2015). 
 
Students? closeness at each time point was calculated using Carolan?s (2014) 
definition. Students who were more central to their class friendship network using the 
closeness definition had higher scores than their less-connected peers. Closeness centrality 
uses both friendship nominations that a student makes (i.e., who do they think are their 
friends?) as well as friendship nominations that a student receives (i.e., who thinks of me as 
their friend?). Therefore, closeness centrality incorporates information about a student?s 
 28 
perception of their standing within the social network (i.e., do I think that I have lots of 
friends in this class?) as well as how others perceive that particular student. Students with 
increased closeness scores believe they are connected to other central students in the 
classroom, and other students reciprocate those nominations. Students with lower closeness 
scores are less likely to name central students as their friends, and/or central students are 
less likely to nominate them. These students are less likely to be friends with well-
connected peers, or they are more likely to only be connected to peers at the periphery of 
the social network.  
 In-degree. In contrast, in-degree centrality only utilizes friendship nominations that 
each student receives, with the idea that students who receive more friendship nominations 
are more central to their network (e.g., Philip, 2010). A student?s own nominations for 
whom they believe are their friends do not affect their own in-degree, whereas closeness 
centrality by definition is affected by the nominations that a particular student makes. In 
essence, in-degree does not allow students to affect their own centrality score. This 
measure highlights how others perceive a student, and it does not allow one?s own 
perception (or lack thereof) of connectedness or comfortability with peers to affect one?s 
score. Because of the differences in the underlying constructs that in-degree and closeness 
measure, it is not expected that these measures would be highly correlated with each other.  
 As with other social network variables, there are different ways to operationalize the 
variable (Daly, 2010). The simplest involves using a simple count of the number of students 
who nominated a particular student; however, students from larger classes could have more 
students nominate them by nature of a larger pool of students. In order to prevent network 
size from affecting students? scores, in this study in-degree centrality was operationalized as 
the number of nominations a student received, divided by the possible number of students 
in the network who could have nominated that student (i.e., ? ? 1, where N is the number 
of students in the classroom). Like closeness centrality, larger values of in-degree centrality 
indicate a student is more central in their friendship network.   
 29 
Table 2 contains descriptive statistics for the primary variables in the study, and 
Table 3 presents the correlation matrix of the study variables. 
 
Table 2 
Descriptive Statistics for the Study Variables 
Variable M SD Reliability2 
Centrality (Closeness) 0.0198 .01 N/A 
Centrality (In-Degree) 0.26 0.14 N/A 
Oral language comprehension 15.43 4.45 0.78?0.83 
Word decoding 47.58 11.19 0.94?0.98 
Reading comprehension 26.12 8.12 0.94?0.98 
Vocabulary 25.47 4.46 0.77?0.78 
Note: Table 2 displays the sample mean (M), standard deviation (SD), and reliability for 
each of the variables in the model. The units for all reading measures are the number of 
questions correct on the WJ-IV assessment. 
 
Oral language comprehension. Students? oral language comprehension was assessed 
using the Woodcock-Johnson-IV Tests of Achievement (WJ-IV; Schrank et al., 2014) Oral 
Comprehension subtest during individual assessment sessions. This test provides students 
with short, orally-presented passages, and students must complete the passages using an 
appropriate word or phrase (e.g., ?Without a doubt, his novels are more complex than the 
novels of many contemporary ______.? A correct response would be authors.). Students? 
scores are the correct number of items out of 33. Split-half reliability coefficients3 for 
children ages 6-10 range from 0.78 to 0.83 (McGrew et al., 2014).  
Letter-Word decoding. The individually administered Letter-Word Identification 
subtest of the WJ-IV (Schrank, Mather, & McGrew, 2014) measured students? abilities to 
 
2 The reliability for the sociometric variables is challenging to assess accurately due to the 
changing social context of groups (Maassen et al., 2004). In other words, researchers 
measuring the reliability or stability of measures struggle to ascertain whether differences in 
reporting are due to measurement error or to slight changes in the groups being studied 
(Maassen et al., 2004). Cillessen, Bukowski, and Haselager (2000) found that Cohen?s ? for 
sociometric measurements vary between .01 and .44, and sociometric variables are 
particularly consistent for popular or rejected children. 
3 Split-half reliability involves splitting an assessment into two equivalent halves, each of 
which measures the same construct or set of constructs. Then, the reliability in student 
scores on the first and second halves of the assessment are calculated. 
 30 
decode letters and words. This test required students to read a list of English letters or 
words presented in print. Students? scores are the correct number of items out of 78. Split-
half reliability coefficients for children ages 6-10 range from 0.94 to 0.98 (McGrew et al., 
2014).  
Reading comprehension. The individually administered WJ-IV Passage 
Comprehension subtest (Schrank et al., 2014) indicated reading comprehension. This 
subtest consists of three item types: identifying which rebus (picture symbol) matches an 
actual picture, identifying a picture that corresponds to 1-3 words, and silently reading short 
passages and completing blanks in the passages with appropriate words. Students? scores 
are the correct number of items out of 52. Split-half reliability coefficients for children ages 
6-10 range from 0.89 to 0.98 (McGrew et al., 2014).  
Vocabulary. Students? English vocabulary was measured using the WJ-IV Picture 
Vocabulary subtest (Schrank et al., 2014). Students were presented with a picture, and 
they were asked to name the picture using the appropriate English vocabulary word. 
Students? scores are the correct number of items out of 54. Split-half reliability coefficients 
for children ages 6-10 range from 0.77 to 0.78 (McGrew et al., 2014).  
 
 31 
Table 3 
Correlation Matrix of Study Variables 
 Centrality Centrality  Oral Reading Letter-Word 
Vocabulary 
(Closeness) (In-Degree) Comprehension Comprehension Decoding 
Centrality 
-----      
(Closeness) 
Centrality  
0.09** -----     
(In-Degree) 
Oral 
-0.06 0.06 -----    
Comprehension 
Reading 
-0.02 -0.00 0.47*** -----   
Comprehension 
Letter-Word 
-0.07* 0.10** 0.57*** 0.62*** -----  
Decoding 
Vocabulary -0.09** 0.00 0.74*** 0.48*** 0.59*** ----- 
Note: Table 3 displays pairwise correlations for each of the variables in the model.  
*? < .05 
**? < .01 
***? < .001 
 
 32 
Of note, the two measures of centrality have a weak correlation (? = 0.09), providing 
evidence that the measures are capturing different components of centrality. This 
correlation aligns with previous literature finding only a moderate correlation on average 
between in-degree and closeness for studies using symmetric networks (i.e., if either 
student nominates the other, a tie exists; Valenti et al., 2008). Given that this study uses 
directed social networks (i.e., one student can nominate another, but the tie may not be 
reciprocated) as opposed to symmetric networks, one would expect a correlation between 
in-degree and closeness that is much closer to 0. The lack of a relation between the two 
centrality measures indicates that one?s in-degree does not explain variation in one?s 
closeness; therefore, students with high closeness centrality may or may not be the same 
students who made and received lots of nominations themselves.  
The weak correlation between the variables strengthens claims that they are validly 
measuring different components related to friendship. Closeness centrality utilizes friendship 
nominations that students make as well as friendship nominations that students receive, 
measuring both the extent to which a student is connected to their peers and the extent to 
which a student feels they belong within the network. In contrast, in-degree centrality only 
indicates the proportion of students in one?s class who nominated a particular student as a 
friend. Students? perceptions of their own belonging do not factor into in-degree measures; 
instead, in-degree provides evidence about how many of one?s peers consider a particular 
child is a friend. The distinction between the two operational definitions of centrality will be 
important when understanding the model results, since they are illuminating different 
components of the friendship network centrality. 
Data Analysis 
This study examines links between students? friendship network centrality and oral 
language proficiency. A primary concern when estimating these links is that student network 
centrality is not exogenous; student centrality is related to many factors that students have 
control over (e.g., determining who to befriend) and demographic factors (e.g., gender, 
 33 
race, etc.). In addition to time invariant factors, such as gender (Grard et al., 2018), 
students? network centrality is a factor of changing variables, such as students? disposition 
or interests. These variables need to be properly incorporated into the models to avoid 
confounding estimates of student friendship network centrality with other factors. 
To address these concerns, I leverage the longitudinal data and specify a student 
fixed effects model. These models absorb the within-student variation in key independent 
and dependent variables that is time-invariant (Hanushek et al., 2003) ? such as innate 
ability to form friendship networks and student demographic factors related to friendship 
formation ? by using student panel data to measure how changes in the independent 
variable (i.e., friendship network centrality) are associated with changes in the dependent 
variable (i.e., oral language achievement). Using this statistical model, students serve as 
their own controls. The student fixed effects model uses the following equation: 
???? ????.???? = ?1?????????????? + ?? + ????, 
where the oral comprehension score of individual i in class j at school k at time t is 
predicted as a function of centrality measures4. ? is a vector of the unobserved time 
invariant student fixed effects, and u represents the error term. I account for nesting of 
students within classes (and schools) by clustering standard errors at the class level, 
thereby accounting for non-independence of observations within classrooms. This clustering 
method allows for unobserved similarities among students within the same classroom that 
would be related to both the student friendship networks and reading outcomes. For 
example, a classroom in which the teacher purposely utilizes student friendship networks 
when differentiating reading instruction may have less of a relationship between network 
centrality and reading growth since the teacher may be exploiting friendships to drive 
 
4 Appendices 1 and 2 also contain estimates for reading controls, including students? 
measures of English language vocabulary, reading comprehension, and letter-word 
decoding. These estimates have not been incorporated into the primary model since student 
fixed effects models inherently incorporate time variant measures, such as students? 
mastery of various components of the English language. 
 34 
academic achievement. The students in this class may have residuals that differ in 
magnitude from the full sample of students, and thus their residuals need to be clustered 
together to incorporate similarities among this smaller group of students. 
I fit three regression models using the two identified measures of centrality (e.g., 
closeness and in-degree) as the independent variables. Using two different measures allows 
comparisons to be made about whether and how centrality measures differ in predicting 
student achievement growth over time. In the student fixed effects models, I fit one model 
each, using either closeness or in-degree as the independent variable. Then, the final model 
incorporates both centrality measures. Each model uses data from the three different time 
points. 
To examine the effect of omitted variable bias, I ran additional student fixed effects 
models that incorporated other time-varying student-level variables, such as letter-word 
decoding, vocabulary, and reading comprehension (see Appendices 2 and 3). The purpose 
of these additional models is to ensure that additional controls do not result in instability of 
the coefficients of the primary independent variable (i.e., friendship network centrality). By 
adding additional covariates yet not greatly impacting the estimates for centrality, these 
models provide greater evidence that the student fixed effects models are properly specified 
and that no lurking variables are biasing the effect sizes.  
Complete social network data were collected for 414 students within 24 classrooms. 
Between 5 and 13 percent of students were missing reading variables at each time point 
(Taboada Barber et al., 2020), due to transferring schools or being absent from school the 
day the individual assessments were conducted. Missing data can be problematic for social 
network analysis (Neal, 2008; Rogers & Kincaid, 1982), since missing data can distort the 
shape of the predicted network. For example, if data are missing for a key student in a 
network, such as a student who connects or brokers various portions of the network, the 
data may be mapped as two completely exclusive cliques. When mapping networks, more 
complete data help to demonstrate accurately the connections that exist among students. 
 35 
Thus, efforts were taken to incorporate students in the dataset if there was enough 
evidence to represent those students? relationships, either through their own or other 
students? friendship nominations. Students were only excluded from their classroom 
friendship network if they met the following criteria: a) they were absent on the day their 
classroom data were collected, and b) fewer than two students nominated them as friends.  
Missing data on the covariates (i.e., reading comprehension, letter-word decoding, 
vocabulary) were imputed using multiple imputation by chained equations in Stata (see 
Azur et al., 2011). I conducted twenty imputations, which exceeds the guidance of using at 
least the number of imputations to match the percentage of missing data (i.e., 13 
imputations to cover the 13% missingness in this study; White, Royston, & Wood, 2011). 
Following the multiple imputations, sensitivity analyses were conducted to ensure the 
imputation was successful and did not introduce biases (Rezvan, Lee, & Simpson, 2015). 
 
Results 
 Two series of models were run to measure if students? changing friendship network 
centrality was connected with their oral reading comprehension growth. The main results 
are presented in Tables 4 and 5, which display the fitted student fixed effects models?
either using closeness (Table 4) or in-degree (Table 5). Statistically significant predictor 
variables are indicated. Each series includes a null model as well as two versions of a final 
model: one using the variables? original scales, and another using standard deviation units. 
Additionally, models incorporating students? other reading measures (i.e., reading 
comprehension, vocabulary, and word decoding) as covariates are presented in Appendices 
2 through 4.  
 
Closeness Centrality 
 The student fixed effects models include closeness as a measure of friendship 
network centrality, providing evidence about whether student centrality (measured using 
 36 
closeness) can predict academic achievement growth (see Table 4). The closeness-only 
model significantly predicts oral comprehension growth over time (?(1, 402) = 8.48, ? = .004). 
According to the model, a 1-SD increase in centrality was associated with a .08-SD increase 
in oral language comprehension. All standard deviation unit predictions are presented in the 
standardized version of the full model in Table 4. 
 Further, closeness centrality is a significant predictor even after controlling for in-
degree (see Full Model) and other reading covariates (see Appendix 4). Controlling for 
students? in-degree, a 1-SD increase in closeness is associated with a 0.09-SD increase in 
oral comprehension. In this full model that controls for in-degree, students who are more 
connected with their peers as a result of having increased out-degree (i.e., they nominate 
more students as friends) or being closer towards the center of the classroom friendship 
network are expected to increase their oral comprehension above what they experience 
when less-connected to their peers. This predicted increase in oral comprehension as a 
result of increased closeness (but not in-degree) is explored in the discussion section. 
 
 
 
 
 
 
 
 
  
 37 
Table 4 
Student Fixed Effects Model Predicting Oral Comprehension Growth using Closeness 
Centrality 
 Closeness only In-Degree only Full Model 
(Standardized) (Standardized) (Standardized) 
 Estimate t Estimate t Estimate t 
(s.e.) (s.e.) (s.e.) 
Centrality 0.08** 2.91 ? ? 0.09** 3.15 
(Closeness) (.03) (0.03) 
Centrality  ? ? -0.05  -1.39 ?0.06 ?1.70 
(In-Degree) (0.04) (0.04) 
       
Constant -0.04* -2.13 -0.04  -1.88 ?0.03 ?1.84 
(.02) (.02) (0.02) 
       
N 414 ? 407 ? 407 ? 
F 8.48** ? 1.93 ? 5.94** ? 
Note: Table 4 displays the student fixed effects models predicting students? oral language 
comprehension scores. The models present estimates in standard deviation units. Student 
fixed effects are absorbed in the models. Robust standard errors clustered at the classroom 
level are in parentheses.  
*p<.05 
**p<.01 
***p<.001 
 
In-Degree Centrality 
 The student fixed effect model also measured student centrality using in-degree. As 
demonstrated in Table 4, centrality in the in-degree-only model (using in-degree proportion) 
was not a significant predictor of oral comprehension (? = ?0.05, ? = .17). This relation was 
also demonstrated in the full model: a 1-SD increase in in-degree proportion was not a 
significant predictor of oral comprehension, after controlling for closeness (? = ?0.06, ? = .09. 
These estimates were not statistically significant. Overall, in-degree centrality did not 
significantly predict students? semester-over-semester changes in oral comprehension.  
 Implications for these models, and for the differences in the findings of the two 
measures of centrality, are discussed in the next section. 
 
 
 38 
Discussion 
 Research has demonstrated that peer relationships are both associated with and lead 
to improvements in a variety of outcomes that are associated with success in school; 
however, developing connections among children is largely considered a secondary goal in 
schools in comparison with improving academic outcomes. Being well-connected with peers 
can improve students? sense of belonging in the classroom, which is associated with putting 
more effort into schoolwork (Goodenow, 1993) and even increased academic success 
(Roeser, Midgley, & Urdan, 1996). Students? classroom and school peers, broadly defined as 
those students who share a teacher or who attend the same school, can predict student 
achievement (Gottfried, 2014; Hanushek, Kain, Markman, & Rivkin, 2003; Rapaso & 
Goncalves, 2020). However, homing in on the specific peers who students identify as friends 
can provide more relevant information about the extent to which teachers and schools can 
leverage student friendship networks to develop academic achievement. This study 
incorporated students? friendship nominations to develop classroom friendship networks and 
determine the extent to which students? friendship network centrality is associated with 
academic growth, in order to provide specific evidence that peer relationships are significant 
predictors of achievement. Because of the importance of language comprehension to all 
school content areas, I used language achievement as an outcome variable to explore 
whether social network analyses could link student centrality to achievement growth.  
Language comprehension?whether for written or oral language?is crucial for 
academic success, and this paper provides support that it is intricately connected to 
students? peer relationships. Students? ability to process and understand language provides 
the foundation for many fundamental skills (Einarsd?ttir et al., 2016). Oral language skills, 
in particular, can be important predictors of later reading outcomes (Catts et al., 2001; 
Hulme & Snowling, 2009). Additionally, oral language skills can be as effective as written 
reading comprehension tests in determining student performance on high-stakes tests 
(Whitley, 2019). Given the importance of language comprehension to high-stakes testing, 
 39 
districts are in constant need of strategies to strengthen students? achievement. Those 
strategies become more worthwhile for budget-constrained districts when they are both 
inexpensive and effective. This study examines one potential mechanism?student 
friendship networks?which are empirically connected with academic achievement growth in 
the student fixed effects models. 
 Students? friends may supplement student academic growth, particularly for oral 
comprehension. When elementary students were more connected to their classroom 
network, they were more likely to have increased oral comprehension test scores, above 
and beyond growth explained by their reading comprehension, letter-word decoding, and 
vocabulary understanding. While the effect size of this increase was small?a 1-standard 
deviation increase in centrality was associated with a 0.09 standard deviation increase in 
oral language comprehension?this finding suggests the importance that friends can have on 
students? language development, which positively impacts academic growth. Based merely 
on friendship connectedness, a student at the median of oral language comprehension but 
1-SD above the mean in terms of network centrality could be expected to perform at the 
54th percentile after a semester. These estimated effect sizes are moderate in comparison 
with other education interventions (see Kraft, 2020), and the estimated effect size is aligned 
with other peer effect studies that find small to moderate effect sizes of peers on academics 
(e.g., Angrist & Lang, 2004; Cooc & Kim, 2016; Lefgren, 2004). However, the estimated 
effect size in this study is smaller than other larger detected effect sizes using longitudinal 
data (e.g., Hanushek et al., 2001; Henderson et al., 1978; Hoxby, 2000). Paired with other 
instructional supports, such as high-quality teachers, effective curriculum, and other school 
supports, highly connected children could experience increased reading achievement over 
time. 
This growth works in conjunction with academic growth students experience in 
schools as a result of instruction. Understanding how teachers can strengthen students? 
friendship networks is outside of the scope of this study, but there are interventions to 
 40 
increase student connections in the classroom (e.g., Cranley Gallagher, 2013; Krone & Yu, 
2019), and these interventions could provide a relatively inexpensive approach with many 
positive effects that more dense friendship networks have on students, including increased 
engagement (Liem & Martin, 2011), self-efficacy (Graber et al., 2016), happiness / well-
being (Troop-Gordon et al., 2019), and self-esteem (Antonopoulou et al., 2019), as well as 
academic growth. 
There were discrepancies in the results of the two series of student fixed effects 
models. Specifically, students had significant increases in oral comprehension when they 
had increased network closeness, yet in-degree did not significantly predict growth in oral 
comprehension. The component that separates the two measures is one?s out-degree, or 
the number of friends that a student claims to have in their network. Because in-degree is 
not a significant predictor of academic growth and given that in-degree and closeness are 
only weakly correlated in this dataset, it appears that the difference in the two models is 
largely being driven by one?s out-degree. Thus, students who are nominating more friends 
have greater closeness measures, and those are the students who are experiencing larger 
academic gains. In that sense, closeness may be a proxy for students? beliefs of 
connectedness with their peers, as well as how supported and comfortable students feel in 
their homerooms. Future research should explore differences in belonging, motivation, and 
engagement for students who have different levels of out-degree, particularly when there 
are differences in the extent to which friendship nominations are reciprocated by their 
peers.  
 Additionally, future research can continue to explore the mechanisms explaining why 
friendship matters for both academic achievement and oral language comprehension, but 
prior theoretical and empirical work offer preliminary hypotheses. Students who are more 
central in their classroom friendship networks are the most likely to feel a sense of 
belonging at school (Osterman, 2000). This sense of belonging is particularly relevant for 
students who are nominating many students as friends, as they are demonstrating beliefs 
 41 
that they are connected with peers. Additionally, having friends can provide a protective 
buffer against negative academic consequences, such as school dropout (Wentzel et al., 
2018), or against internalizing (Burgess et al., 2006) or externalizing behaviors (Bowker & 
Spencer, 2010). Longitudinal studies can untangle which constructs are at play as students 
experience changes in their social connectedness, as well as changes in academic 
achievement. Further, classroom-level connectedness may be an important factor?for 
example, do classrooms with a well-connected network of students and few (or no) students 
on the periphery experience greater academic achievement gains, on average, than classes 
with greater variability in connectedness? Future research can explore these differences in 
classroom networks and their associated differences in outcomes. 
 These results provide promising evidence that friendships have important predictive 
power for language growth and academic achievement, but there are some limitations to 
this study. The student population is not representative of the general US public school 
population; English Learners were heavily oversampled in the study. Because peers are 
theorized to be a major resource for the academic and social development of ELs (see 
Bukowski, Laursen, & Rubin, 2018), it is possible that the same pattern may not be 
observed in students who are English Monolinguals (EM) in US schools. Additionally, these 
findings should not be interpreted as friendship networks having causal impact on language 
growth. While fixed effects models control for time-invariant predictors, and the panel data 
allow students to serve as their own comparisons, there are other changing factors that 
could explain the observed increases in oral language comprehension, such as student self-
efficacy or students? connectedness with family members. As social network analyses 
continue to develop, more rigorous causal frameworks can test the causal nature of friends 
on academic achievement.  
 Furthermore, friendship quality was not measured in this study, and the quality of 
relationships is important in explaining one?s sense of belonging. For example, friendship 
quality is correlated with students having increased self-worth (Hiatt, Laursen, Mooney, & 
 42 
Rubin, 2015), and friendship quality is also predictive of students having fewer internalizing 
problems (Rubin et al., 2004). Because not all friendships are equal in quality (e.g., Bagwell 
& Bukowski, 2018), future studies should incorporate quality when comparing students? 
connectedness over time, using measures such as Bukowski, Hoza, and Boivin?s (1994) 
Friendship Qualities Scales.  
 Last, it is important to note that children?s social networks are not limited to their 
homeroom classrooms. While elementary students spend a majority of their school time 
engaging with peers in their classrooms, students? networks consist of children from across 
classes, grades, and even schools. This study used classroom rosters to define social 
networks, but children may have more support from individuals outside of these classes. It 
is possible that their sense of belonging and self-worth may be driven by those connections 
in addition to the friendships formed within a classroom. 
 These results are exploratory, and future work should continue to investigate the 
explanatory power that friends and peer effects can have on academic achievement. 
However, the significant results provide preliminary evidence that friends are an important 
resource that schools can utilize to leverage student growth. I do not suggest that effective 
teachers, evidence-based curriculum, or supportive learning environments are unimportant 
in the American education system for student growth; however, peers have been 
overlooked as a pillar that can support increases in academic achievement. Fostering a 
sense of community and developing friendship connections among children, particularly in 
elementary school classrooms in which students spend most of the school day with the 
same group of children, can support student growth?developmentally and socially, but also 
academically. 
 
 
 
 
 43 
References 
Angrist, J.D., & Lang, K. (2004). Does school integration generate peer effects? Evidence 
from Boston?s Metco program. American Economic Review, 94(5), 1613-1634.  
Antonopoulou, K., Chaidemenou, A., & Kouvava, S. (2019). Peer acceptance and friendships 
among primary school pupils: Associations with loneliness, self-esteem and school 
engagement. Educational Psychology in Practice. 
Azur, M.J., Stuart, E.A., Frangakis, C., & Leaf, P.J. (2011). Multiple imputation by chained 
equations: What is it and how does it work? International Journal of Methods in 
Psychiatric Research, 20(1), 40-49. 
Babayigit, S., Roulstone, S., & Wren, Y. (2020). Linguistic comprehension and narrative 
skills predict reading ability: A 9-year longitudinal study. British Journal of 
Educational Psychology. 
Bagwell, C.L., & Bukowski, W.M. (2018). Friendship in childhood and adolescence. In 
Bukowski, W.M., Laursen, B., & Rubin, K.H. (Eds.), Handbook of peer interactions, 
relationships, and groups. Second edition. New York City: Guilford Press. 
Bagwell, C.L., & Schmidt, M.E. (2011). Friendships in childhood and adolescence. New York: 
Guilford Press. 
Bandura, A. (1986). Social foundation of thought and action: A social cognitive theory. 
Englewood Cliffs, NJ: Prentice Hall.  
Beron, K.J., & Farkas, G. (2004). Oral language and reading success: A structural equation 
modeling approach. Structural Equation Modeling, 11(1), 110-131. 
Bowker, J.C.W., Rubin, K.H., & Burgess, K.B. (2006). Behavioral characteristics associated 
with stable and fluid best friendship patterns in middle childhood. Merrill-Palmer 
Quarterly, 52(4), 671-693. 
Bowker, J.C., & Spencer, S.V. (2010). Friendship and adjustment: A focus on mixed-grade 
friendships. Journal of Youth & Adolescence, 39(11), 1318-1329. 
 44 
Bukowski, W.M., Hoza., B., & Boivin, M. (1994). Measuring friendship quality during pre- 
and early adolescence: The development and psychometric properties of the 
Friendship Qualities Scale. Journal of Social and Personal Relationships, 11, 471-484. 
Bukowski, W.M., Laursen, B., & Rubin, K.H. (Eds.; 2018). Handbook of peer interactions, 
relationships, and groups. Second edition. New York City: The Guilford Press. 
Burgess, K.B., Wojslawowicz, J.C., Rubin, K.H., Rose-Krasnor, L., & Booth-LaForce, C. 
(2006). Social information processing and coping strategies of shy/withdrawn and 
aggressive children: Does friendship matter? Child Development, 77(2), 371-383. 
Burk, W.J., Steglich, C.E.G., & Snijders, T.A.B. (2007). Beyond dyadic interdependence: 
Actor-oriented models for co-evolving social networks and individual behaviors. 
International Journal of Behavioral Development, 31(4), 397-404. 
Cain, K., & Oakhill, J. (2007). Children?s Comprehension Problems in Oral and Written 
Language: A Cognitive Perspective. Challenges in Language and Literacy. Guilford 
Publications. 
Carolan, B.V. (2014). Social network analysis and education: Theory, methods, & 
applications. Thousand Oaks, CA: Sage Publications. 
Catts, H.W., Fey, M.E., Tomblin, J.B., & Zhang, X. (2002). A longitudinal investigation of 
reading outcomes in children with language impairments. Journal of Speech, 
Language & Hearing Research, 45(6), 1142-1157. 
Chen, J.A., & Pajares, F. (2010). Implicit theories of ability of grade 6 science students: 
Relation to epistemological beliefs and academic motivation and achievement in 
science. Contemporary Educational Psychology, 35(1), 75-87. 
Cillessen, A.H.N., & Bukowski, W.M. (2018). Sociometric perspectives. In Bukowski, W.M., 
Laursen, B., & Rubin, K.H. (Eds.), Handbook of peer interactions, relationships, and 
groups. Second edition. New York City: Guilford Press. 
Cillessen, A.H.N., Bukowski, W.M., & Haselager, G.J.T. (2000). Stability of sociometric 
categories. New Directions for Child & Adolescent Development, 2000(88), 75-93. 
 45 
Cillessen, A.H.N., & Mayeux, L. (2004). From censure to reinforcement: Developmental 
changes in the association between aggression and social status. Child Development, 
75, 147-163.  
Cillessen, A.H.N., & Rose, A.J. (2005). Understanding popularity in the peer system. Current 
Directions in Psychological Science, 14, 102-105. 
Clarke, P., Crawford, C., Steele, F., & Vignoles, A. (2015). Education Economics, 23(3), 
259-277. 
Cooc, N., & Kim, J. (2016). Peer influence on children?s reading skills: A social network 
analysis of elementary school classrooms. Journal of Educational Psychology, 
Forthcoming. 
Cranley Gallagher, K. (2013). Guiding children?s friendship development. Young Children, 
68(5), 26-32. 
Cruz, J.E., Emery, R.E., & Turkheimer, E. (2012). Peer network drinking predicts increased 
alcohol use from adolescence to early adulthood after controlling for genetic and 
shared environmental selection. Developmental Psychology, 48(5), 1390-1402.  
Daly, A. J. (2010). Social network theory and educational change. Cambridge, MA: Harvard 
Education Press. 
Davison, K. (2019). Early modern social networks: Antecedents, opportunities, and 
challenges. American Historical Review, 124(2), 456-482. 
Davison, M.D., Hammer, C., & Lawrence, F.R. (2011). Associations between preschool 
language and first grade reading outcomes in bilingual children. Journal of 
Communication Disorders, 44(4), 444-458. 
Dotterer, A.M., McHale, S.M., & Crouter, A.C. (2007). Implications of out-of-school activities 
for school engagement in African American adolescents. Journal of Youth and 
Adolescence, 36, 391-401. 
Einardsd?ttir, J.T., Bj?rnsd?ttir, A., & S?monard?ttir, I. (2016). American Journal of Speech-
Language Pathology, 25(1), 67-79. 
 46 
Fuchs, D., Fuchs, L.S., & Burish, P. (2000). Peer-assisted learning strategies: An evidence-
based practice to promote reading achievement. Learning Disabilities Research & 
Practice, 15(2), 85-91. 
Goodenow, C. (1993). Classroom belonging among early adolescent students: Relationships 
to motivation and achievement. Journal of Early Adolescence, 13, 21-43.  
Gottfried, M.A. (2014). Peer effects in urban schools: Assessing the impact of classroom 
composition on student achievement. Educational Policy, 28(5), 607-647. 
Graber, R., Turner, R., & Madill, A. (2016). Best friends and better coping: Facilitating 
psychological resilience through boys? and girls? closest friendship. British Journal of 
Psychology, 107(2), 338-358. 
Grard, A., Kunst, A., Kuipers, M., Richter, M., Rimpela, A., Federico, B., & Lorant, V. (2018). 
Same-sex friendship, school gender composition, and substance use: A social 
network study of 50 European schools. Substance Use & Misuse, 53(6), 998-1007. 
Grigg, S., Perera, H.N., McIlveen, P., & Svetleff, Z. (2018). Relations among math self-
efficacy, interest, intentions, and achievement: A social cognitive perspective. 
Contemporary Educational Psychology, 53, 73-86.  
Grimm, K.J. (2008). Longitudinal associations between reading and mathematics 
achievement. Developmental Neuropsychology, 33(3), 410-426. 
Hanushek, E.A., Kain, J.F., Markman, J.M., & Rivkin, S.G. (2003). Does peer ability affect 
student achievement? Journal of Applied Econometrics, 18(5), 527-544. 
Harris, D.N. (2010). How do school peers influence student educational outcomes? Theory 
and evidence from economics and other social sciences. Teachers College Record, 
112(4), 1163-1197. 
Hartl, A.C., DeLay, D., Laursen, B., Denner, J., Werner, L., Campe, S., & Ortiz, E. (2015). 
Dyadic instruction for middle school students: Liking promotes learning. Learning & 
Individual Differences, 44, 33-39. 
 47 
Henderson, V., Mieszkowski, P., & Sauvageau, Y. (1978). Peer group effects and educational 
production functions. Journal of Public Economics, 10, 97-106. 
Hiatt, C., Laursen, B., Mooney, K.S., & Rubin, K.H. (2015). Forms of friendship: A person-
centered assessment of the quality, stability, and outcomes of different types of 
adolescent friends. Personality and Individual Differences, 77, 149-155. 
Hinde, R., & Stevenson-Hinde, J. (1987). Implications of a relationships approach for the 
study of gender differences. Infant Mental Health Journal, 8(3), 221-236. 
Hoxby, C.M. (2000). Peer effects in the classroom: Learning from gender and race variation. 
NBER working paper No. 7867. 
Hoxby, C.M., & Weingarth, G. (2005). Taking race out of the equation: School reassignment 
and the structure of peer effects (No. 7867). Working paper. 
Hulme, C., & Snowling, M.J. (2009). Developmental disorders of language learning and 
cognition. Wiley-Blackwell.  
Keenan, J.M., Betjemann, R.S., & Olson, R.K. (2008). Reading comprehension tests vary in 
the skills they assess: Differential dependence on decoding and oral comprehension. 
Scientific Studies of Reading, 12(3), 281-300. 
Kindermann, T.A., & Gest, S.D. (2018). The peer group: Linking conceptualizations, 
theories, and methods. In Bukowski, W.M., Laursen, B., & Rubin, K.H. (Eds.), 
Handbook of peer interactions, relationships, and groups. Second edition. New York 
City: Guilford Press. 
Kory-Westlund, J.M., & Breazeal, C. (2019). A long-term study of young children?s rapport, 
social emulation, and language learning with a peer-like robot playmate in preschool. 
Front. Robot. AI, 6(81). 
Kraft, M.A. (2020). Interpreting effect sizes of education interventions. Educational 
Researcher, 49(4), 241-253. 
Krone, M.W., & Yu, S. (2019). Promoting friendship development in inclusive early childhood 
classrooms: A literature review. 11(2), 183-193. 
 48 
Ladd, G.W., & Troop-Gordon, W. (2003). The role of chronic peer difficulties in the 
development of children?s psychological adjustment problems. Child Development, 
74(5), 1344-1367. 
Laursen, B., & Hartup, W.W. (2002). The origins of reciprocity and social exchange in 
friendships. In W.G. Graziano & B. Laursen (Eds.), Social exchange in development: 
New directions for child and adolescent development (No. 95, p. 27-40). San 
Francisco: Jossey-Bass. 
Lefgren, L. (2004). Educational peer effects and the Chicago public schools. Journal of 
Urban Economics, 56, 169-191. 
Lemons, C.J., Fuchs, D., Gilbert, J.K., & Fuchs, L.S. (2014). Evidence-based practices in a 
changing world: Reconsidering the counterfactual in educational research. 
Educational Researcher, 43(5), 242-252. 
Lereya, S.T., Copeland, W.E., Costello, E.J., & Wolke, D. (2015). Adult mental health 
consequences of peer bullying and maltreatment in childhood: Two cohorts in two 
countries. The Lancet Psychiatry, 2(6), 524-531. 
Liem, G.A.D., & Martin, M.J. (2011). Peer relationships and adolescents? academic and non-
academic outcomes: Same-sex and opposite-sex peer effects and the mediating role 
of school engagement. British Journal of Educational Psychology, 81(2), 183-206. 
Maassen, G.H., Steenbeek, H., & van Geert, P. (2004). Stability of three methods for two-
dimensional sociometric status determination based on the procedure of Asher, 
Singleton, Tinsley and Hymel. Social Behavior and Personality, 32(6), 535-550. 
McGrew, K.S., LaForte, E.M., & Schrank, F.A. (2014). Woodcock Johnson IV Technical 
Manual. Rolling Meadows, IL: Riverside.  
Neal, J.W. (2008). ?Kracking? the missing data problem: Applying Krackhardt?s cognitive 
social structures to school-based social networks. Sociology of Education, 81(2), 
140-162. 
 49 
Nyman, M.C. (2013). A quantitative, comparison study: Oral language development and 
high stakes testing (Order No. 3570888). Available from ProQuest Dissertations & 
Theses Global. (1417074645). Retrieved from 
https://www.proquest.com/docview/1417074645?accountid=14696 
Osterman, K.F. (2000). Students? need for belonging in the school community. Review of 
Educational Research, 70(3), 323-367. 
Parker, J.G., & Asher, S.R. (1987). Peer relations and later personal adjustment: Are low-
accepted children ?at risk?? Psychological Bulletin, 102, 357-389.  
Parkhurst, J.T., & Hopmeyer, A. (1998). Sociometric popularity and peer-perceived 
popularity: Two distinct dimensions of peer status. Journal of Adolescence, 18, 125-
144. 
Philip, D.N. (2010). Social network analysis to examine interaction patterns in knowledge 
building communities. Canadian Journal of Learning and Technology, 36(1). 
Rambaran, J.A., Hopmeyer, A., Schwartz, D., Steglich, C., Badaly, D., & Veenstra, R. 
(2017). Academic functioning and peer influences: A short-term longitudinal study of 
network-behavior dynamics in middle adolescence. Child Development, 88(2), 523-
543. 
Rapaso, I.P.d.A., & Goncalves, M.B.C. (2020). Peer effects and educational achievement: 
Evidence of casual effects using age at school entry as exogenous variation for peer 
quality. Economia (University of Brazil), 21(1), 18-37. 
Reed, D.K., Petscher, Y., & Truckenmiller, A.J. (2016). The contribution of general reading 
ability to science achievement. Reading Research Quarterly, 52(2), 253-266. 
Rezvan, P.H., Lee, K.J., & Simpson, J.A. (2015). The rise of multiple imputation: A review of 
the reporting and implementation of the method in medical research. BMC Medical 
Research Methodology, 15. 
Roeser, R.W., Midgley, C., & Urdan, T.C. (1996). Perceptions of the school psychological 
environment and early adolescents? psychological and behavioral functioning in 
 50 
school: The mediating role of goals and belonging. Journal of Educational 
Psychology, 88(3), 408-422.  
Rogers, E., & Kincaid, L. (1982). Communication in networks and convergence. Inter Media, 
10(1), 14-18. 
Rubin, K.H., Dwyer, K.M., Booth-LaForce, C., Kim, A.H., Burgess, K.B., & Rose-Krasnor, L. 
(2004). Attachment, friendship, and psychosocial functioning in early adolescence. 
Journal of Early Adolescence, 24, 326-356. 
Ryan, A.M., & Shim, S.S. (2012). Changing in help seeking from peers in early adolescence: 
Associations with changes in achievement and perceptions of teachers. Journal of 
Educational Psychology, 104(4), 1122-1134. 
Schrank, F.A., Mather, N., & McGrew, K.S. (2014). Woodcock-Johnson IV Tests of 
Achievement. Rolling Meadows, IL: Riverside. 
Silverman, R.D., Johnson, E., Keane, K., & Khanna, S. (2020). Beyond decoding: A meta-
analysis of the effect of language comprehension interventions on K-5 students? 
language and literacy outcomes. Reading Research Quarterly, 55(S1), S207-S233. 
Snyder, J., Prichard, J., & Schrepferman, L. (2004). Child impulsiveness-attention, early 
peer experiences, and the development of early onset conduct problems. (Journal of 
Abnormal Child Psychology, 32(6), 579-594. 
Swain, M., & Brooks, L.A. (2002). Peer-peer dialogue as means of second language 
learning. Annual Review of Applied Linguistics, 22, 171-185. 
Taboada Barber, A. (PI), Cartwright, K., & Stapleton, L. (Co-PIs) (2016-2020). Cognitive 
and motivational contributors to reading comprehension in English Learners (ELs) 
and English Monolinguals (EMs): Different or similar growth patterns. English 
Learners Research Grants Competition, Institute of Education Sciences.  
Taboada Barber, A., Cartwright, K.B., Stapleton, L.M., Lutz Klauda, S., Archer, C.J., & 
Smith, P. (2020). Direct and indirect effects of executive functions, reading 
engagement, and higher order strategic processes in the reading comprehension of 
 51 
Dual Language Learners and English Monolinguals. Contemporary Educational 
Psychology, 61, 1-17. 
Troop-Gordon, W., MacDonald, A.P., & Corbitt-Hall, D.J. (2019). Children?s peer beliefs, 
friendlessness, and friendship quality: Reciprocal influences and contributions to 
internalizing symptoms. Developmental Psychology, 55(11), 2428-2439. 
Usher, E.L., Ford, C.J., Li, C.R., & Weidner, B.L. (2019). Sources of math and science self-
efficacy in rural Appalachia: A convergent mixed methods study. Contemporary 
Educational Psychology, 57, 32-53. 
Valentini, A., Ricketts, J., Pye, R.E., Houston-Price, C. (2018). Listening while reading 
promotes word learning from stories. Journal of Experimental Child Psychology, 167, 
10-31. 
van Rijsewijk, L.G.M., Oldenburg, B., Snijders, T.A.B., Dijkstra, J.K., & Veenstra, R. (2018). 
A description of classroom help networks, individual network position, and their 
associations with academic achievement. PLoS ONE, 13(12), e0208173. 
https://doi.org/10.1371/journal.pone.0208173 
Vaughn, S., Swanson, E.A., Roberts, G., Wanzek, J., Stillman-Spisak, S.J., Solis, M., & 
Simmons, D. (2013). Improving reading comprehension and social studies 
knowledge in middle school. Reading Research Quarterly, 48(1), 77-93. 
Wentzel, K.R., Jablansky, S., & Scalise, N.R. (2018). Do friendships afford academic 
benefits? A meta-analytic study. Educational Psychology Review, 30(4), 1241-1267. 
White, I.R., Royston, P., & Wood, A.M. (2011). Multiple imputation using chained equations: 
Issues and guidance for practice. Statistics in medicine, 30(4), 377-399. 
Whitley, S. (2019). Oral reading fluency and maze selection for predicting 5th and 6th grade 
students? reading and math achievement on a high stakes summative assessment. 
Reading Improvement, 56(1), 24-35. 
 
 
 52 
Appendices 
Appendix 1 
Table comparing full Project Learn sample (in spring 2018) to analytic sample 
 Full Sample,  Analytic Sample P-value 
Spring 2018 
Grade, SY 2017-    
2018 24.3% 24.6%  
     Second 23.3% 26.3% .520 
     Third 29.2% 25.6% 
     Fourth  23.3% 23.4% 
     Fifth 
Age, Fall 2017 9.23 (1.17) 9.20 (1.17) .668 
Gender    
     Male 47.2% 52.9% With missing: 
     Female 43.8% 44.7% < .001 
     Missing 9.0% 2.4%  
Without missing: 
.454 
Ethnicity    
     Hispanic 60.8% 73.7% With missing: 
     Black / African     23.5% 21.3% < .001 
American 2.8% 0.7%  
     White 1.6% 0.5% Without missing: 
     Asian 0.1% 0.2% .008 
     Native   
Hawaiian /   
Pacific 2.2% 1.2% 
Islander 9.0% 2.4% 
     Multi-Racial 
     Missing 
SES, Fall 2017    
     Eligible for 76.9% 87.0% With missing: 
Free or   < .001 
Reduced-Price 14.1% 9.2%  
Lunch 9.0% 3.9% Without missing: 
     Not eligible .005 
     Missing 
English Learner,    
either SY 17-18 or   With missing: 
SY 18-19 54.5% 70.5% < .001 
     Yes 36.5% 29.2%  
     No 9.0% 0.2% Without missing: 
     Missing < .001 
 53 
Student has IEP,    
either SY 17-18 or   With missing: 
SY 18-19 3.9% 5.3% < .001 
     Yes 87.1% 92.0%  
     No 9.0% 2.7% Without missing: 
     Missing .387 
Note: Chi-square tests of homogeneity were conducted for each set of categorical variables 
to test whether the analytic sample varies from the full sample. For the continuous variable, 
age, a two-sample t-test for independent means compared the two groups. For the 
categorical variables, 9.0% of the students did not have demographic data due to students 
being absent when individual testing was conducted, or from missing district-level data. 
Statistical testing was conducted using ?missing? as a category; however, since these 
students would not be incorporated in any models (due to missing other variables), the 
statistical test that excluded missing values gives a better representation of the difference 
between the full and analytic samples.   
 
 
 
 
 
 
 
 
 
 
 
  
 54 
Appendix 2 
Student Fixed Effects Model with Reading Covariates using Closeness Centrality 
 Model with  Model with  
Covariates Covariates 
(Standardized) 
 Estimate t Estimate in SD t 
(s.e.) (s.e.) 
Centrality 30.41** 2.82 0.07** 2.82 
(Closeness) (10.78) (.03) 
Reading -0.00 -0.24 -.00 -0.24 
Comprehension (.01) (.02) 
Vocabulary 0.13* 2.29 0.13* 2.29 
(.06) (.05) 
Letter-Word 0.16*** 5.99 0.42*** 5.99 
Decoding (.03) (.07) 
     
Constant 4.18** 2.72 .03 1.46 
(1.54) (.02) 
     
N 414 ? 414 ? 
F 16.72*** ? 17.78*** ? 
Note: Appendix 2 displays the student fixed effects models predicting students? oral 
language comprehension scores, utilizing closeness centrality and reading covariates. The 
full standardized version presents estimates in standard deviation units. Student fixed 
effects are absorbed in the models. Robust standard errors clustered at the classroom level 
are in parentheses.  
*p<.05 
**p<.01 
***p<.001 
 
 
 
 
 
 
  
 55 
Appendix 3 
Student Fixed Effects Model with Reading Covariates using In-degree Centrality 
 Model with Covariates Model with Covariates 
(Standardized) 
 Estimate (s.e.) t Estimate in SD t 
(s.e.) 
Centrality -0.79 -0.75 -0.03 -0.75 
(In-degree) (1.06) (.03) 
Reading -0.00  -0.37 -0.01  -0.37 
Comprehension (.01) (.02) 
Vocabulary 0.16** 2.67 0.15** 2.67 
 (.06) (.06) 
Letter-Word 0.16*** 5.75 0.43*** 5.75 
Decoding  (.03) (.08) 
    
Constant 4.06* 2.42 0.03 1.62 
(1.68) (.02) 
    
N 407 ? 407 ? 
F 14.90*** ? 14.90*** ? 
Note: Appendix 3 displays the student fixed effects models predicting students? oral 
language comprehension scores, utilizing in-degree and reading covariates. The full 
standardized version presents estimates in standard deviation units. Student fixed effects 
are absorbed in the models. Robust standard errors clustered at the classroom level are in 
parentheses.  
*p<.05 
**p<.01 
***p<.001 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
 56 
Appendix 4 
Student Fixed Effects Model with Reading Covariates using Closeness and In-Degree 
Centrality 
 Full Model with  
Covariates 
(Standardized) 
 Estimate  t 
(s.e.) 
Centrality 0.09** 3.29 
(Closeness) (0.03) 
Centrality  ?0.04 ?1.19 
(In-Degree) (0.03)  
Reading ?0.01 ?0.52 
Comprehension (0.02)  
Vocabulary 0.14* 2.30 
(0.06) 
Letter-Word 0.47*** 6.14 
Decoding (0.08) 
   
Constant 0.04 1.62 
(0.02) 
   
N 406 ? 
F 15.14*** ? 
Note: Appendix 4 displays the student fixed effects models predicting students? oral 
language comprehension scores, utilizing closeness centrality, in-degree, and reading 
covariates. The full standardized version presents estimates in standard deviation units. 
Student fixed effects are absorbed in the model. Robust standard errors clustered at the 
classroom level are in parentheses.  
*p<.05 
**p<.01 
***p<.001 
 
 
 
 
 
 
 
  
 57 
 
 
 
 
 
 
 
 
Paper 2 
 
  
 
 
 
 
 
 
The Differential Relation between Friendship Centrality and Reading Outcomes for 
English Learners 
 
 
 
 
 
 
 
 
 58 
Longstanding lines of research highlight gaps in academic outcomes between English 
Learners (ELs) and English Monolinguals (EMs). As such, considerable attention has been 
paid to interventions that may close these gaps (e.g., Cardoza & Brown, 2020; Carlo et al., 
2008). Among other resources, leveraging peers is hypothesized to be an important way to 
improve ELs? social and academic skill development and close achievement gaps (Chen, Lee, 
& Chen, 2018; Graham & Echols, 2018), due in part to the large role that peer socialization 
plays in ELs? English language development and overall well-being. The interaction among 
language status, culture, and/or race and ethnicity may leave ELs particularly susceptible to 
exclusion or even victimization (e.g., Nansel et al., 2001). 
Research has begun to identify whether social relationships can specifically benefit 
ELs? learning. For example, within their classroom networks, ELs with a higher proportion of 
non-EL friends5 are more likely to experience year-over-year growth in English Language 
Arts standardized tests (Elreda et al., 2016). While there is preliminary support that 
friendships matter for ELs? academic achievement, research exploring peer relationships in 
US public schools for ELs, as well as various ethnic and racial groups, is relatively 
underexplored (Graham & Echols, 2018). Further, research is necessary to understand the 
predictive value of classroom social networks on achievement for ELs, and whether and how 
teachers can leverage social relationships to support student learning. This research can 
 
5 As in Paper 1, ?friends? refers to reciprocated relationships between two individuals, 
characterized by trust, caring, and respect (i.e., Bagwell & Bukowski, 2018; Laursen & 
Hartup, 2002). ?Peers? are others who are in the same classroom, grade, or school, 
regardless of whether each pair of individuals considers themselves friends (see Kindermann 
& Gest, 2018). While friends are reciprocated relationships, one individual can perceive 
another is a friend, and in the mind of that individual, the friendship exists. Because of the 
egocentric (i.e., student-driven) measures used to capture friendships, these perceived 
friendships are still part of the underlying friendship network because the perceived tie 
affects the structure of the network. If all friendship ties were mapped among a classroom 
of students, this would form the friendship social network among all peers in the class. 
Those children who are most connected to their peers in this friendship network would be 
considered the most ?central,? and would therefore have the greatest centrality within the 
network. 
 59 
provide evidence of the extent to which connecting to peers may be particularly important 
for ELs? academic outcomes. 
To examine the connection between students? social networks and academic growth, 
and whether that relation is connected to student language status, this study explores the 
following research questions:  
? To what extent does classroom friendship centrality predict reading achievement 
gains for English Learners? 
? To what extent does the relationship between classroom friendship centrality and 
reading achievement gains differ between English Learners and English 
Monolinguals? 
 To answer these questions, I use a statistical modeling technique, known as social 
network analyses, which allows researchers to determine the extent to which people within 
a social network are important, or central, to the network overall. Data come from five 
elementary schools in the Mideastern United States. These schools have high proportions of 
English Learners (the students are primarily Latinx and are first- through third-generation 
immigrants). Students were in 2nd- through 5th-grade when data collection began, and data 
were collected over the course of three academic semesters. Students were asked at each 
time point to name their best friends in their classroom, and these data were analyzed using 
social network analyses in conjunction with longitudinal reading data. 
 The following section details what is known about ELs, peer relationships, 
friendships, and their connections to language development.  
 
Literature Review 
Why Peers and Friends Matter for Learning 
 The relation between peer relationships in the classroom and learning is 
multifaceted. Social interactions affect the ways children think about themselves and their 
belonging within the classroom environment. The dynamic and complex social interactions 
 60 
may result in children becoming either isolated or rejected by their peers, and in turn, at 
increased risk of internalizing behaviors such as depression (Lereya et al., 2015). These 
children are also more likely to display aggressive behaviors (Ladd & Troop-Gordon, 2003) 
and to be impulsive or inattentive in school (Snyder et al., 2004). In contrast, those 
children who have strong relationships with their peers, including those who are ?central? 
(i.e. well connected with many other children) in their social network, may be protected 
against the negative outcomes associated with being isolated or rejected (Bowker et al., 
2006).  
Children with friends in their classroom environments may display behaviors and 
develop motivations that are associated with academic success. Having friends in general is 
associated with increased motivation and self-efficacy (Bagci, 2018; Nelson & DeBacker, 
2008), both of which are contributing factors to academic achievement (Niehaus et al., 
2012; Usher et al., 2019). Further, having strong friend groups is associated with students 
exploring academic concepts more deeply in the classroom and having a greater 
understanding of the content (Hartl et al., 2015).  
Popular or well-liked children may affect classroom norms in ways that can 
strengthen or inhibit learning. Children are more likely to assume the behaviors of these 
well-liked or popular children (Cillessen & Rose, 2005; Rambaran et al., 2017), particularly 
when these well-liked or popular children share common identity markers with themselves, 
such as race, gender, language status, or other factors (Harris, 2010). As a result, the 
behaviors of those children may be influential in establishing a classroom environment that 
sustains learning and inquiry. Peer effects have empirical support, both for developmental 
and academic outcomes (Bukowski et al., 2018; Harris, 2010), and researchers have used 
large-scale datasets to connect peers with increases in both mathematics and reading test 
scores (Gottfried, 2014; Hanushek et al., 2003).  
In addition to factors related to popularity or likeability, learning is associated with 
children?s access to resources. Specifically, children are more likely to succeed academically 
 61 
when resources are available to help them achieve, and friendships afford children the 
opportunity to access resources. When they form friendships, children build the capacity to 
increase their social capital, or the potential resources that an individual can access through 
their social connections, such as knowledge, physical capital, or other forms of support 
(Bourdieu, 1986), since their friends may have access to resources or other individuals. 
Indeed, states with increased social capital (as measured by civic engagement) have been 
linked to improved student outcomes, due in part to the social capital opportunities afforded 
to residents (Braatz & Putnam, 1996). Similarly, schools with increased parental 
involvement are associated with increased academic achievement, even after controlling for 
school socioeconomic status (Park et al., 2017). Friendships allow children to expand the 
resources available to them, including human capital available through friends? families. 
These connections may provide children with greater social or academic support, which can 
translate into increased opportunities to learn. 
 
Classroom Social Networks and English Learners 
While friends are important for all children, research and theory suggest they may be 
particularly important for English Learners. Because some ELs may not speak English 
outside of the school setting, school is an important location for these children to develop 
their oral English skills. Furthermore, English oral language abilities are an important 
predictor of reading achievement and comprehension (Babayigit et al., 2020; Silverman et 
al., 2020). Students? interactions with their peers thus provide opportunities for language 
development (Kory-Westlund & Breazeal, 2019; Swain et al., 2002).   
Knowledge about ELs? friends may provide information about ELs? English language 
development and, ultimately, ELs? abilities to perform well on English language tests. 
Empirical evidence has demonstrated that interacting with English Monolinguals has more 
predictive power for ELs? English oral language proficiency than other variables associated 
with verbal skill development, such as time spent in the US and maternal education (Carhill-
 62 
Poza, 2015). Additionally, ELs who report a greater proportion of their friends are non-ELs 
were more likely than other ELs to demonstrate gains in middle school standardized tests 
(Elreda et al., 2016), potentially due to more opportunities to practice the language as well 
as the increased sense of belonging in the school. For Latinx students in particular, who 
represent a plurality of ELs in US public schools (NCES, 2020), feelings of belonging at 
school were directly related to the strength of their friendships (Delgado et al., 2016). In 
other words, Latinx students who believed they had stronger friendships also had 
heightened perceptions of feeling they belong in their school community, which is associated 
with academic success (Allen & Bowles, 2012; Juvonen, 2006). Friendships thus are 
associated with many developmental predictors for academic outcomes.  
Research supports the benefits of ELs having both similar and dissimilar friends, 
including other students who are also learning the English language, and other students of 
the same ethnic/racial groups. Among marginalized groups including ELs and immigrant 
students, students with stronger friendships?both same-race and inter-race?are less likely 
to drop out of school than other student groups (Delgado et al., 2016). Immigrant children 
with greater propensity to build social capital, acquired through building connections with 
any of their peers, are more likely to succeed academically, regardless of whether their 
friends are also immigrants (Ryabov, 2009). Ultimately, building relationships among peers 
can provide them with greater likelihood to succeed on a variety of outcomes, including 
academics.   
 
Learning and Policy Considerations and English Learners 
There are also unique policy concerns that differentially impact English Learners in 
US schools. ELs are responsible for learning the same content as their grade-level peers, 
but learning presents an increased challenge given that ELs must navigate the new content, 
usually presented in English, while also learning the English language. While statewide 
?English-only? education policies currently only exist in Arizona (Jacobson, 2020), ELs are 
 63 
often denied supplemental support that might come in the form of instruction or resources 
in their native language, even though native language literacy as well as access to content 
delivered in students? native language are associated with higher student achievement 
(Garc?a-Vazquez et al., 1997; Ramirez et al., 1991; Thomas & Collier, 2003).  
Additionally, structural components may present ELs with challenges in the schools. 
Districts and schools with high proportions of ELs are more likely to have relatively 
inexperienced teachers (Dabach, 2015), often who do not have specific training on how to 
tailor instruction to ELs. While some states with high-EL populations, such as California, 
have recently mandated that teachers receive training during their preparation programs on 
supporting ELs, these policies can have unintended consequences in relegating ELs to only 
new teachers (Dabach, 2015). States with newer waves of immigrant populations, such as 
North Carolina, do not have similar statewide policies aimed at supporting ELs, leaving 
teachers in these states feeling unprepared to meet ELs? specific learning needs (O?Neal et 
al., 2008). Moreover, even among teachers with specialized degrees or certificates, there is 
still a pervasive perception among teachers that they are unprepared to guarantee success 
for ELs (G?ndara et al., 2005). ELs are less likely to graduate than other students, due in 
part to state graduation requirements mandating students pass high-stakes assessments 
(Center on Education Policy, 2006). Districts with high-EL populations?both urban and 
rural?are more likely to report having insufficient materials or even school facilities 
necessary to teach students (G?ndara & Maxwell-Jolly, 2006). Given these challenges, much 
work remains to be done to determine ways to support ELs?academically and socially. 
 
Hypotheses and Directions for Research 
While the extant literature provides insight into how peers can serve as an important 
resource for ELs? social and academic development, there are several gaps. Like the broader 
peer effects literature, existing research connecting ELs? peers to test scores tends to use 
administrative datasets that, by the nature of the available data, make assumptions about 
 64 
who one?s peers are. For example, researchers have focused on the percentage of same-
race peers, or percentage of students who passed previous standardized tests, and linked 
these measures to students? own academic achievement (e.g., Angrist & Lang, 2004; 
Gottfried, 2014; Hanushek, Kain, Markman, & Rivkin, 2003). Those studies contribute 
important findings to the broad peer effects literature, by measuring the impact of class- or 
grade-level average characteristics on student achievement, but they do not incorporate 
data about which students actually are interacting with and influencing each other regularly, 
which can be measured using student-level friendship or relationship data within classrooms 
or grades. An alternative approach proposed by some (and used in this study) is to use 
student-created ?social networks,? which are thought to provide more valid representations 
of the underlying classroom networks (e.g., Carolan, 2014). A second limitation of the 
current literature base is that these sorts of rich social networks rarely are connected to 
educational outcomes. Research in this arena is sparse in general, and the narrower 
literature exploring links between ELs? peers and their outcomes is understudied in 
comparison to other student subgroups (Bellmore, Nishina, & Graham, 2011; Flores-
Gonzalez, 2005; Graham & Echols, 2018).  
At present, Elreda et al. (2016) is the only study that uses social network analyses to 
predict academic outcomes for ELs, yet the study only includes as a predictor variable the 
percentage of one?s nominated friends who are non-ELs. While that study explores whether 
ELs with higher percentages of English Monolingual friends are more likely to improve their 
own English language skills, its findings beg the question of whether overall student 
centrality and acceptance?rather than just acceptance by EMs?can similarly improve 
outcomes for ELs. In other words, are ELs who are well connected within their classroom 
networks still likely to improve language outcomes, even if their connections are primarily 
with other ELs? Even though theoretical research on children in general links student 
acceptance (one component of student social networks) and school belonging with academic 
progress (e.g., Allen, Kern, Vella-Brodrick, Hattie, & Waters, 2018; Kindermann, 2007; 
 65 
Kindermann, 2016), few studies have examined the explicit connection among these 
constructs for ELs. More research is necessary to investigate whether peer relationships 
factor significantly into ELs? academic outcomes.   
Prior literature offers a hypothesized theory of action explaining why friendships may 
affect ELs? academic outcomes. As discussed in Paper 1, there are many reasons why 
friendships matter for academic outcomes for children in general, including developing one?s 
sense of belonging (e.g., Dotterer, McHale, & Crouter, 2007; Goodenow, 1993; Roeser, 
Midgley, & Urdan, 1996), one?s ability to interact with and learn the content (Hartl et al., 
2015), and one?s oral language skills through increased opportunities to socialize with 
others. Beyond these reasons, however, ELs are particularly likely to benefit from having 
friends in the classroom, primarily because many ELs only interact in English when in the 
classroom or with their friends. As a result, peer and friend interactions develop ELs? oral 
language comprehension (Kory-Westlund & Breazeal, 2019; Swain et al., 2002). Having 
increased opportunities to interact with peers is associated with greater English language 
growth (Carhill-Poza, 2015), so it stands to reason that those students with greater 
connections with peers will demonstrated increased development. Given the increased 
predictive ability of friendships to protect against dropping out of school later in their career 
(Delgado et al., 2016), it seems likely that having friends within the earlier grades may 
particularly offer a propensity for ELs to succeed and develop their language skills.  
In sum, I hypothesize that the effect size of being well-connected within the 
classroom friendship network will be even greater for ELs than within the general student 
population. Those students who are most connected (i.e., those with greater friendship 
network centrality) will experience the benefits of greater self-worth, increased feelings of 
belonging, greater access to social capital, and more opportunities to practice the English 
language, all of which will translate into greater oral language development and, ultimately, 
greater reading comprehension development. For these reasons, the hypothesized 
 66 
estimated effect of friendship centrality for ELs will be larger than the estimated effect of 
friendship centrality in the general elementary student population.   
 
Methods 
This study is part of a larger longitudinal study examining elementary-aged English 
Learners? and English Speakers? reading development. The full project took place from Fall 
2016 to Spring 2019. The present study encompasses data collected from Spring 2018 
through the end of the project. In Spring 2018, participants were 2nd-through 5th-grade 
students from five public elementary schools. The full sample of students who participated 
in the whole-class assessments in Spring 2018 included 765 students. To be included in the 
present study, students had to have social network data in the full semester of the project 
(Spring 2019). The final sample consisted of 232 students. (See Table 1 for a comparison 
between the full analytic sample and the ELs-only sample used for Models 1 and 2.) 
Students were assessed twice per semester?once was an individual assessment and 
once was in a whole-class setting. For individual assessments, a Research Assistant (RA) led 
the child to a quiet spot in the school for a one-hour assessment battery, consisting of 
thirteen short standardized assessments each semester. For whole-class assessments, one 
or two RAs administered three whole-class assessments to the students participating in the 
study within their classrooms. Any students not participating were allowed to read or meet 
with their teacher during that one-hour block. For both the individual and whole-class 
assessments, RAs were trained each semester on administering and scoring the 
assessments. RAs were required to pass semesterly fidelity checks with the study authors.  
The outcome variables (i.e., oral language comprehension, reading comprehension) 
were administered as part of the individual assessment battery, and the independent 
variable of interest (i.e., student network centrality) was assessed as part of the whole-class 
assessment. Students? English language status and their gender were part of an 
administrative dataset collected by the school district. 
 67 
Participants 
 Five public schools in the mid-Atlantic US were part of the larger study and 
comprised the current sample. These schools had higher-than-average proportions of ELs 
compared to the overall state population, which allowed for greater statistical power to 
examine the reading and cognitive development of ELs over time. Parents or legal guardians 
of students provided annual informed consent for their child to participate in the study; 
students who did not receive informed consent across both years of the present study were 
removed from the sample. This study was approved by the university institutional review 
board, and participants were treated in accordance with the ethical principles of the 
American Psychological Association. 
Students in this analytic sample were evenly distributed across grades 3 through 5 in 
the spring of 2019, with an average age of 9.7 years. A slightly greater number of 
participants were male (50.9%) than female (47.8%), although gender data were missing 
for some students (1.3%). Most of the students were Hispanic (78.5%) or Black (17.8%). 
Most students were eligible for free or reduced-price lunch in either year of the study 
(88.4%), and most students were identified by the school district as English Learners in 
either year of the study (73.3%). Table 1 provides complete descriptive data on the full 
sample of students in the present study, including the subsamples that were limited to only 
ELs.  
 
  
 68 
Table 1 
Descriptive statistics for the sample 
Variable ELs-only sample Full analytic sample 
 n M (SD) or % n M (SD) or % 
Grade, SY 2018-2019     
     Third 60 35.3% 77 33.2% 
     Fourth 60 35.3% 76 32.8% 
     Fifth 50 29.4% 79 34.1% 
Age, Fall 2019 167 9.65 (0.90) 229 9.7 (0.89) 
Gender     
     Male 88 51.8% 118 50.9% 
     Female 82 48.2% 111 47.8% 
     Missing 0 0.0% 3 1.3% 
Ethnicity     
     Hispanic 170 100.0% 182 78.5% 
     Black / African American 0 0.0% 41 17.8% 
     White 0 0.0% 4 1.7% 
     Asian 0 0.0% 2 0.9% 
     Missing 0 0.0% 3 1.3% 
SES, Fall 2017     
     Eligible for Free or Reduced-Price 161 94.7% 205 88.4% 
Lunch     
     Not eligible 9 5.3% 24 10.3% 
     Missing 0 0.0% 3 1.3% 
English Learner, either SY 17-18 or SY     
18-19     
     Yes 170 100.0% 170 73.3% 
     No 0 0.0% 62 26.7% 
Student has IEP, either SY 17-18 or SY     
18-19     
     Yes 13 7.6% 16 6.9% 
     No 157 92.4% 216 93.1% 
 
Measures 
Social Network Analysis (SNA) is a statistical procedure used to quantify the 
underlying social networks of contained groups (Carolan, 2014; Daly, 2010). Researchers 
analyze a closed group of individuals and the social ties among them in order to reproduce a 
visual image of the network and analyze the network using various quantitative methods 
(Carolan, 2014). While SNA is becoming more common across research fields, it is currently 
underutilized in education research (Daly, 2010). Network information can be used in 
 69 
conjunction with peer effects models to give researchers a better understanding of the 
social processes taking place within classrooms or schools. Furthermore, these social 
relationships can be used to leverage student outcomes in educational systems. Therefore, 
understanding social network analyses may lead to improved educational outcomes for 
students when policy is informed by this research (Daly, 2010). 
Social network models allow researchers to gain a broader perspective of how one?s 
social environment can shape one?s own attributes and outcomes. Incorporating social 
network information into statistical models can develop a greater understanding of the 
structure of a network and one?s own position within that social group (Burk et al., 2007). 
Creating social network models allows researchers to predict how relationships might 
change or evolve over time, as well as how other actors may influence one?s actions or 
character traits (Burk et al., 2007). 
Friendship Centrality. The primary independent variable for the present study was 
a student?s friendship centrality, which aims to classify how important each actor, or child, 
is to the network as a whole (Carolan, 2014). While there are many ways to define 
centrality, this study utilizes closeness. Following other researchers who also asked students 
to identify their closest friends (i.e., Cruz et al., 2012; Elreda et al., 2016; Li & Stone, 
2018), I use the term ?friendship network? throughout this study to refer to the underlying 
social network connecting peers via their indicated friendships. While not all indicated 
relationships are reciprocated, these friendship networks contain both the actualized and 
perceived friendships among the classroom peers.  
Carolan (2014) defines centrality as a function of the average distance between a 
particular node and all of the other nodes in the network. Thus: 
1
??????????? = ?  ?(?? , ??)
 70 
where d is the distance between Node i and every other Node j in the model (Carolan, 
2014). This particular measurement of centrality is referred to as ?closeness,? and a node 
that is more central within the network has higher closeness than other nodes.  
In the whole-class assessment, students were provided with a complete roster from 
their homeroom class. The assessor asked, ?Who are your best friends in your class? Circle 
up to five friends.? Students were reminded that they could circle zero, one, two, three, 
four, or five friends from their class. They were instructed not to circle themselves. Children 
were constrained by which other children were on their class roster. Because the 
longitudinal study used purposive sampling aimed at ensuring a large proportion of ELs in 
the sample, most children were interacting with and therefore nominating ELs. The large 
proportion of ELs in the sample affects generalizability, as discussed in the Limitations 
section. 
 Based on students? responses, directed social networks were created for each 
classroom at each of the three time points. Using the igraph R package (version 1.2.4.2), 
sociograms mapped students? social position on a latent space. Students who were more 
connected to their classroom networks based on the friendship nominations they indicated 
and received are positioned towards the center of the sociogram, and their less-connected 
peers appear on the periphery. Centrality scores were created using the full sample of 
students utilized in Paper 1, regardless of whether the students remained in the analytic 
sample used in this paper. Keeping these calculations from Paper 1 ensured that social 
networks were created using the full group of peers in the classroom. See Figure 1 for a 
sample sociogram displaying one of the social networks from one of the classrooms.  
  
 71 
Figure 1 
Class sociogram demonstrating friendship network for a sample class, shaded by EL status 
 
Note: Figure 1 displays a sociogram that maps friendship nominations among students in a 
particular classroom at Time 1. Each student is represented by a node, with numbers 
indicating the student ID. Students closer to the center of the network, such as student 
794, have increased friendship centrality scores because they are more connected to their 
classroom peers. Student 3009, who was only nominated by one peer as a friend (and who 
nominated no friends of their own), would have the smallest friendship centrality score from 
among this classroom network. The sociogram was created using the nwcommands Stata 
package (Gorund, 2015). Nodes are shaded according to whether the student was identified 
as an English Learner or an English Monolingual. 
 
  
 72 
Students? centrality at each time point was calculated using Carolan?s (2014) 
operationalization of closeness. Students who were more central to their class had higher 
scores than their less-connected peers. Because the social network was focused on 
friendships within the classroom, the primary independent variable is referred to as 
friendship centrality, or friendship closeness, throughout this study.  
Oral language comprehension. Students? oral language comprehension was 
assessed using the Woodcock-Johnson-IV Tests of Achievement (WJ-IV; Schrank et al., 
2014) Oral Comprehension subtest during individual assessment sessions. This test provides 
students with short, orally presented passages, and students must complete the passages 
using an appropriate word or phrase (e.g., ?Without a doubt, his novels are more complex 
than the novels of many contemporary ______.? A correct response would be authors.). 
Students? scores are the correct number of items out of 33. The Woodcock-Johnson IV tests 
have been widely used in education research, and analyses of other datasets show strong 
reliability indices. Split-half reliability coefficients on the oral language comprehension score 
for children ages 6-10 range from 0.78 to 0.83 (McGrew et al., 2014).  
Reading comprehension. The individually administered WJ-IV Passage 
Comprehension subtest (Schrank et al., 2014) measured reading comprehension. This 
subtest consists of three item types: identifying which rebus (picture symbol) matches an 
actual picture, identifying a picture that corresponds to 1-3 words, and silently reading short 
passages and completing blanks in the passages with appropriate words. Students? scores 
are the correct number of items out of 52. In other datasets, split-half reliability coefficients 
for children ages 6-10 range from 0.89 to 0.98 (McGrew et al., 2014).  
 Demographic variables. Students? English Language status was determined by the 
school district through the WIDA assessment. Students who did not score above a particular 
threshold, determined by grade level, were identified by the district as an EL and were 
eligible for English as a Second Language support. For the present study, students were 
labeled as English Learners if they were classified as an EL during either of the two years 
 73 
from which data were used for this study. The school district also indicated students? gender 
through the administrative dataset. The gender values were indicated by parents when 
students enrolled the student in the district, unless the district updated any gender 
identities by parent or student request. 
 
Table 2 
Descriptive statistics for the study variables 
Variable M SD Min Max N 
Friendship centrality (Closeness) 0.02 0.01 0.003 0.053 232 
Oral language comprehension 
15.9 4.3 4 27 224 
(Time 6) 
Oral language comprehension 
15.4 4.2 3 28 232 
(Time 5) 
Reading comprehension (Time 6) 27.0 5.6 10 39 232 
Reading comprehension (Time 5) 25.4 5.2 9 39 232 
Centrality * EL Status Interaction 0.01 .01 0 .053 232 
Note: Table 2 displays the sample mean (M) and standard deviation (SD) for each of the 
variables in the model. The units for all of the reading measures are the number of 
questions correct on the WJ-IV. This table does not include gender or EL status, which were 
included in Table 1. 
 
 
Data Analysis 
Linear regression models were run to explore the various research questions. Each 
model uses a reading achievement outcome measured in Spring 2019. The primary 
independent variable for the models is student friendship centrality, measured using 
closeness during the same semester as the achievement outcome measure. Additionally, 
Models 3 and 4 include an interaction term between student friendship centrality and 
English language status; this interaction term indicates the extent to which there is a 
differential predictive value of friendship closeness for ELs versus EMs.  All estimated 
coefficients indicate the predicted increase in reading comprehension (oral or written) that is 
associated with a 1-unit change in the independent variables, after controlling for the other 
variables in the model. The models control for the students? Fall 2018 score on the same 
achievement measure as the model outcome variable, a dummy-coded variable for whether 
 74 
a student is female, and students? English language status (for Models 3 and 4). Controlling 
for prior achievement allows the model to account for student-level variables that might 
otherwise introduce bias into the coefficients. For example, by controlling for initial reading 
achievement, the models will account for student factors that are associated with individual 
achievement, such as motivation, socioeconomic status, parent education level, etc., 
allowing the equation to model student growth (see Ballou, Sanders, and Wright, 2004).  
The students were nested in 26 classrooms; the robust error terms for the models 
are clustered by class to account for class-level dependencies. All non-demographic 
variables were standardized by subtracting the grand mean (i.e., all students for whom data 
were collected) and dividing by the overall standard deviation. Standardization allows for 
comparisons across variables, as well as for ease of interpretation. The table below displays 
the variables used in each model, as well as the research questions that guide the models. 
The fitted regression equations follow Table 3. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 75 
Table 3 
Research questions associated with each model 
 Model 1 Model 2 Model 3 Model 4 
Research To what extent does classroom To what extent does the 
Question friendship centrality predict relationship between classroom 
reading achievement gains for network centrality and reading 
English Learners? achievement gains differ between 
 ELs and English Monolinguals? 
Outcome Low-stakes oral Low-stakes Low-stakes oral Low-stakes 
Variable comprehension reading comprehension reading 
test comprehension test comprehension 
test test 
Independent Network centrality Network centrality, and the 
Variables interaction between network 
centrality and English language 
status 
Covariates Prior oral Prior reading Prior oral Prior reading 
comprehension, comprehension, comprehension, comprehension, 
gender, gender, English English 
socioeconomic socioeconomic language language 
status, grade status, grade status, gender, status, gender, 
socioeconomic socioeconomic 
status, grade status, grade 
Sample Students identified as ELs All students in the sample 
Note: Gender is operationalized using a dummy-coded variable indicating whether a student 
identifies as female. Socioeconomic status is operationalized using a dummy-coded variable 
indicating whether a student qualifies for free- or reduced-price lunch. Grade level is 
operationalized using a series of dummy-coded variables indicating whether a student is in 
second, third, fourth, or fifth grade.  
 
 
 
 
  
 76 
Model 1: 
???? ?????????????????? ?
= ?1 ? ?????????? ??????????? + ?2 ? ???? ???????????????????? + ?3 ? ????????????? ?
+ ??? 
Model 2: 
??????? ?????????????????? ?
= ?4 ? ?????????? ??????????? + ?5 ? ??????? ???????????????????? + ?6 ? ????????????? ?
+ ??? 
Model 3: 
???? ?????????????????? ?
= ?7 ? ?????????? ??????????? + ?8 ? ?????????? ??????????? ? ?? ??????? + ?9
? ???? ???????????????????? + ?10 ? ????????????? + ?11 ? ?? ??????? + ?? ?? 
Model 4: 
??????? ?????????????????? ?
= ?12 ? ?????????? ??????????? + ?13 ? ?????????? ??????????? ? ?? ??????? + ?14
? ??????? ???????????????????? + ?15 ? ????????????? + ?16 ? ?? ??????? ? + ??? 
 
For each model, ?s represent the estimated coefficients for each indicated variable using 
student i. Each model includes a vector of student demographics variables, including a 
dummy-coded variable for whether a student is a female, whether the student is eligible for 
free or reduced-price lunch, and the students? grade. The coefficients for these models are 
not reported in the tables below and serve as control variables to ensure children are being 
compared to similar students. However, the coefficients are included for reference in 
Appendices 1 through 4. Each model?s error term, ?, is clustered by class j.  
 
Results 
Oral Language Comprehension, English Learners 
 I created models to determine whether the relation between centrality and 
achievement holds for English Learners. The first model predicted English oral language 
comprehension using students? measures of classroom friendship centrality (see Table 4).  
 77 
 This model provided further evidence that students? friendship centrality is 
associated with English oral language growth over time. In aggregate, the final model 
significantly predicts oral language growth (?(6, 24) = 112.06, ? < .0001). Controlling for the 
other variables, a 1-standard deviation (SD) increase in centrality was associated with a 
0.12-SD increase in oral language comprehension, after controlling for the other variables. 
Students? prior oral comprehension was significantly related to their Spring 2019 oral 
comprehension (?(26) = 20.60, ? < .001).  
 
Table 4 
Regression model predicting standardized oral comprehension growth using friendship 
centrality for English Learners 
Variable Estimate  t-value p-value 
(s.e.) 
Friendship Centrality 0.12 2.85 .009 
(Closeness) (0.04) 
 
Prior oral 0.85 20.60 < .0001 
comprehension (Fall (.04) 
2018)  
    
Constant ?0.10 ?0.45 0.655 
(0.22) 
 
    
N 160   
?(6, 24) 112.06  < .0001 
R-squared 0.658   
Note: Standard errors were adjusted for the 25 clusters. The outcome variable was 
students? oral comprehension in Spring 2019. All units are standardized units except 
demographic variables (not reported). 
 
Reading Comprehension, English Learners 
 Thus far, this dissertation has explored growth in oral language comprehension due 
to the demonstrated theoretical and empirical connection between elementary students? 
relationships/friendships and oral language. However, it is also plausible that reading 
comprehension is associated with students? social network centrality. Model 2 tests this 
 78 
theory, and it explores whether friendship centrality is associated with English Learners? 
reading comprehension growth. 
 As demonstrated by Model 2, students? friendship centrality does not predict reading 
comprehension growth over time (see Table 5). A 1-SD increase in friendship closeness is 
associated with a .03-SD decrease in reading comprehension, controlling for other variables 
(?(25) = ?1.57 ? = .128). This result is not statistically significant. As expected, students? prior 
reading comprehension significantly predicts their Spring 2019 reading comprehension 
(?(25) = 15.43, ? < .0001), after controlling for the other variables. UThis model provides 
evidence that, whereas friendship centrality is associated with oral language 
comprehension, it is not significantly related to students? reading comprehension growth. 
This finding is explored in the discussion section. 
 
Table 5 
Regression model predicting standardized reading comprehension growth using friendship 
centrality for English Learners 
Variable Estimate  t-value p-value 
(s.e.) 
Friendship Centrality ?0.03 ?1.57 .128 
(Closeness) (0.02) 
Prior reading 0.36 15.43 < .0001 
comprehension (Fall (0.02) 
2018) 
    
Constant ?0.23 ?2.49 .020 
(0.09) 
    
N 168   
?(6, 25) 194.11  < .0001 
R-squared 0.677   
Note: Standard errors were adjusted for the 26 clusters. The outcome variable was 
students? reading comprehension in Spring 2019. All units are standardized units except 
demographic variables (not reported). 
 
 
 
 79 
Oral Language Comprehension, English Learners vs. English Monolinguals 
 While the first two models explore the relation between reading achievement and 
closeness (friendship centrality) for English Learners, the second set of models investigates 
whether there is a differential relation between reading achievement and friendship 
centrality for English Learners versus English Monolinguals. The findings from these models 
can be used to query whether friendship centrality, and the broader construct of friendship, 
is connected to ELs? reading achievement and potentially can be utilized by teachers to 
improve ELs? reading outcomes.  
 Model 3 investigates the differential relation for ELs and EMs between oral language 
comprehension and friendship centrality. The interaction term between English language 
status and centrality is not statistically significant (?(24) = 0.51, ? = .613), indicating that there 
is no significant difference in the relation between centrality and oral language growth for 
ELs and EMs. In other words, the estimated relation between centrality and oral language 
growth is the same for ELs and EMs, and neither group of students benefits more or less 
from being more connected to their peers. 
 Similar to the previous models, the covariates were also important predictors for oral 
language comprehension in the full sample. Specifically, students? prior oral language 
comprehension was statistically associated with their Spring 2019 scores (?(24) = 26.24, ? <
.0001); girls? oral language comprehension was marginally associated with their Spring 2019 
scores (?(24) = ?1.77, ? = .090).  
 
 
 
  
 80 
Table 6 
Regression model predicting standardized oral language comprehension growth using 
friendship centrality and English language status, full sample 
Variable Estimate  t-value p-value 
(s.e.) 
Friendship Centrality 0.09 1.50 .146 
(Closeness) (0.06) 
English language ?0.15 ?1.89 .071 
status (0.08)  
Friendship Centrality 0.03 0.51 .613 
* English language (0.07) 
status 
Prior oral 0.84 23.30 < .0001 
comprehension (Fall (0.04) 
2018) 
    
Constant 0.17 1.51 .143 
(0.11) 
    
N 221   
?(8, 24) 151.22  < .0001 
R-squared 0.694   
Note: Standard errors were adjusted for the 25 clusters. The outcome variable was 
students? oral language comprehension in Spring 2019. All units are standardized units 
except demographic variables (not reported). 
 
 
 
 
 
 
 
 
 
 
 
 
 81 
Figure 2 
Predicted oral language comprehension using friendship centrality, by English language 
status 
 
0.4
0.3
0.2
0.1
ELs
0 EMs
-1.5 -1 -0.5 0 0.5 1 1.5
-0.1
-0.2
-0.3
Closeness (SD)
 
Note: The predicted outcome values were created using the fitted regression model from 
Table 6. Initial reading comprehension and gender are held constant in this graph. The 
predicted outcomes are for boys whose oral language comprehension was at the group 
mean for Fall 2018. 
 
Reading Comprehension, English Learners vs. English Monolinguals 
 While there was no differential relation between friendship closeness and oral 
language comprehension for ELs and EMs, Model 4 explored the differential relation using 
reading comprehension scores. In the ELs-only sample, reading comprehension was not 
statistically associated with friendship closeness, yet this model explores if the association is 
found in the full sample. 
 As displayed in Table 7, friendship closeness is statistically associated with reading 
comprehension in the full sample. After controlling for the other variables, a 1-SD increase 
in friendship closeness is associated with a 0.07-SD increase in reading comprehension 
(?(25) = 2.30, ? = .030). Students who are more connected within their homeroom friendship 
networks are more likely to improve reading comprehension scores, even after controlling 
for their score in the prior semester. A student who was at the 50th percentile in reading 
Predicted Oral Language Comprehension 
(SD)
 82 
comprehension in Fall 2018 but who was one standard deviation above the mean in terms 
of friendship centrality were predicted to be in the 53rd percentile for reading comprehension 
in the Spring 2019 semester.  
 Furthermore, there is a differential relation between friendship centrality and reading 
comprehension for ELs and EMs. The interaction term between the two variables was 
statistically significant (?(25) = ?2.62, ? = .015). While the predicted slope between friendship 
closeness and reading comprehension for English Monolinguals was positive, the predicted 
slope between the variables for ELs was not statistically different from zero in this model. Of 
note, the difference in the slopes between ELs and EMs was statistically significant (see 
Figure 3).  Whereas English Monolingual students with increased centrality in their 
classroom networks were more likely to experience reading comprehension growth, even 
after controlling for their prior reading comprehension scores, ELs experienced no significant 
relation in the two variables. The significant relation between friendship closeness and 
reading comprehension for EMs but not for ELs is explored in the discussion section. 
 
 
 
 
  
 83 
Table 7 
Regression model predicting standardized reading comprehension growth using friendship 
centrality and English language status, full sample 
Variable Estimate  t-value p-value 
(s.e.) 
Friendship Centrality 0.08 2.05 .051 
(Closeness) (0.04) 
English language ?0.10 ?2.06 .050 
status (0.05)  
 
Friendship Centrality ?0.09 ?2.52 .018 
* English language (0.03)  
status  
Prior oral 0.31 8.18 < .0001 
comprehension (Fall (0.04) 
2018) 
    
Constant 0.02 0.18 .859 
(0.09) 
    
N 229   
?(8, 25) 38.94  < .0001 
R-squared 0.633   
Note: Standard errors were adjusted for the 26 clusters. The outcome variable was 
students? oral language comprehension in Spring 2019. All units are standardized units 
except demographic variables (not reported). 
 
 
 
 
 
 
 
  
 84 
Figure 3 
Predicted reading comprehension using friendship centrality, by English language status 
 
0.15
0.1
0.05
0 ELs
-1.5 -1 -0.5 0 0.5 1 1.5 EMs
-0.05
-0.1
-0.15
Closeness (SD)
 
Note: The predicted outcome values were created using the fitted regression model from 
Table 7. Initial reading comprehension and gender are held constant in this graph. These 
predicted values were calculated for boys whose initial reading comprehension scores were 
at the mean.  
 
Limitations 
 This study provides evidence of the differential association between friendship 
centrality and reading achievement for ELs versus EMs, but there are some important 
limitations to the findings. The students who made up the analytic sample are not 
representative of the ?average? US public school population for elementary students; ELs 
made up a larger proportion of the sample than in the overall population. While this 
purposive sample allowed for greater power to detect trends in the EL population, the 
results may not generalize to other school contexts. Friendship centrality may have 
dissimilar predictive power for schools in which ELs make up a slimmer proportion of the 
population, due to ELs feeling more excluded from the general population. Because ELs 
composed nearly 50% of the sample, and due to the tendency of students to interact with 
and befriend similar students (i.e., homophily), the estimated effect of network centrality 
Predicted Reading Comprehension (SD)
 85 
may be unique to school settings in which there are large proportions of students identifying 
as ELs. These students may or may not be conversing in English; future research may 
explore the ways that communication patterns (i.e., language spoken) among friends 
moderate the relation between friendship centrality and oral language growth. More 
research in other contexts and with data collected on communication patterns is necessary 
to explore these questions.  
 The measures incorporated in this study may limit conclusions that one can make 
about friendship and academic achievement. The study did not include a randomized control 
trial, so causal claims are not warranted when discussing the findings. While friendship 
closeness may predict achievement growth for students, lurking variables may exist which 
explain the model results. It is possible that these lurking variables relate to social 
relationships, or that they are related to other causes. For example, students with more 
friends may also have more access to their friends? social capital, which grants them 
opportunities to exceed academically; or it is possible that students who claim to have more 
friends are also more confident in themselves, and this increased confidence may be 
correlated with increased effort and self-efficacy. Future research should explore the 
mechanisms underlying this study.  
Some variables that are often collected in research studies that examine immigrants 
or children of immigrants were not included in this study. For example, data were not 
collected about students? immigrant generation, their parents? social networks, or their 
family?s cultural heritage, and these factors may lend explanatory power as to why 
particular ELs succeed while others do not, and why being connected to friends only matters 
for some students. These additional measures were not available, but they may have 
strengthened the models by ensuring all variables that are related to both immigrant status 
and language comprehension were incorporated. Additional research is necessary to 
determine how these variables affect estimated friendship centrality effects. 
 86 
 Last, classroom social networks are not fully representative of students? social 
networks. Even though elementary students spend a majority of their time with their 
classroom peers, students have relationships with siblings, family members, and other 
children outside of their classroom. These relationships also factor into student outcomes, 
and a comprehensive case study might explore those connections and their relation to 
achievement over time. However, for elementary education, in which students are clustered 
together in their homeroom classroom for most (if not all) of their time at school, these 
classroom networks are a worthwhile unit of analysis to explore. 
 
Discussion and Conclusion 
 Students? classroom social networks can serve as important predictors for student 
growth, yet there are important differences in their predictive ability between English 
Learners and English Monolinguals. Children who are more connected to their classroom 
peers are more likely to display growth in their English oral comprehension scores. These 
results were confirmed for both ELs and EMs, demonstrating that children who had more 
friendship connections with the students in their classroom?regardless of whether they 
themselves were an English Learner?were more likely to develop their oral language 
comprehension. It seems clear that elementary children having friendships within classroom 
networks is associated with developing English oral language skills. 
This model provides preliminary evidence that friendships and the underlying 
friendship networks are important for elementary students? academic success. The relation 
among friendship centrality and oral language development, both among ELs and EMs, is 
unsurprising given that children?s relationships are developed by oral conversations and 
interactions among dyads and within the social groups. On average, those children with 
more opportunities to engage in these interactions experience greater oral language growth 
than students who are less connected to their classroom peers.  
 87 
While oral language comprehension predicts long-term reading comprehension 
(Babayigit et al., 2020; Silverman et al., 2020), the hypothesized relation between reading 
comprehension and friendship was not demonstrated universally in this study; reading 
comprehension growth was only related to friendship centrality for English Monolinguals, 
and not for English Learners. Specifically, English Monolinguals who are more central within 
their friendship network were more likely to experience growth in reading comprehension, 
but there was no relation between friendship centrality and reading comprehension growth 
for ELs.  Even though being more connected was associated with oral language growth for 
ELs, the relation did not generalize to reading comprehension skills. 
More research is needed to understand the mechanisms underlying the differential 
predictive power of friendship closeness by English language status, but current research 
does provide a few suggestions as to why greater connectivity is not associated with reading 
comprehension growth for ELs. More time may be necessary for oral comprehension gains 
to translate to reading comprehension gain for ELs; it is possible that the relationship for 
ELs would be uncovered over time, if the study continued to follow students? centrality 
within their classroom friendship network and their reading comprehension. This differential 
may also be symptomatic of teachers (on average) not being prepared to meet the unique 
language needs of ELs, at least not in general education classes. Students have access to 
English as a Second Language (ESL) supplemental teachers, but ELs may benefit from 
teachers with dedicated training in the ways to support ELs as well as native-language 
support in the content areas (Rumberger & G?ndara, 2000). These teachers may more 
effectively leverage peer connections to ensure meaningful reading growth for all students, 
and not just those whose native language is English. Developing the content expertise of 
general education teachers to support ELs is the key to inclusive and effective education for 
all students, regardless of language status. 
Additionally, it?s unclear whether strong friendship network connections are 
contributing to the same positive classroom climate effects that EMs experience. For 
 88 
example, while having friends is associated with increased self-efficacy and motivation, a 
moderation model can determine whether ELs similarly experience those increases when 
well-connected to friends in general education classrooms. The language barrier may negate 
some of the positive effects that a supportive classroom environment can have on student 
achievement, such as self-efficacy or feelings of belonging (LeClair et al., 2009). 
Interventions exploring social connectivity and its impact on academic achievement (e.g., 
Cranley Gallagher, 2013; Krone & Yu, 2019) should investigate differential impacts based on 
teacher qualities (e.g., experience, training, support, and effectiveness) and student 
language status, in order to determine whether these interventions can help teachers 
leverage social connectedness to improve student outcomes, particularly for ELs. 
 These findings have implications for the ways that elementary teachers teach 
students. English language pedagogy emphasizes giving students opportunities to interact 
with classmates (Helfrich & Bosh, 2001), and teachers should continue using student-to-
student interactions as an opportunity for students to meaningfully develop their oral 
language skills. Social interactions are intricately connected with other components of 
language development, including reading comprehension (Babayigit et al., 2020; Silverman 
et al., 2020). Further, elementary teachers may consider developing students? friendships 
as an opportunity to develop students? language skills (Helfrich & Bosh, 2011), as well as 
social-emotional outcomes that are associated with long-term benefits such as academic 
achievement (Jackson et al., 2021; MacDonnell et al., 2021) and increased perceptions of 
school climate among Latinx students (Jones et al., 2020), among others. 
 Classroom social networks matter, and they can be useful tools to predict student 
achievement. Specifically, elementary students? friendship centrality within their classroom 
social networks can predict semester-over-semester growth in reading and oral 
comprehension. However, the association between network centrality and achievement 
growth is different for English Learners and English Monolinguals. On average, students who 
are more connected to their peers are more likely to develop their oral comprehension skills, 
 89 
which aligns with theoretical underpinnings exploring oral language growth due to the 
importance of verbal communication for young children?s relationship development. 
However, the same is not true for reading comprehension. For EMs, being connected to 
peers is associated with gains in reading comprehension, but ELs do not experience the 
same predictive power. Further research is necessary to understand how to build strong 
classroom networks that can support ELs? growth?both developmental and academic?in 
the same way as their English Monolingual peers. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 90 
References 
Allen, K.A., & Bowles, T. (2012). Belonging as a guiding principle in the education of 
adolescents. Australian Journal of Educational & Developmental Psychology, 12, 108-
119. 
Allen, K., Kern, M.L., Vella-Brodrick, D., Hattie, J., & Waters, L. (2018). What schools need 
to know about fostering school belonging: A meta-analysis. Educational Psychology 
Review, 30(1), 1-34. 
Angrist, J.D., & Lang, K. (2004). Does school integration generate peer effects? Evidence 
from Boston?s Metco program. American Economic Review, 94(5), 1613-1634.  
Babayigit, S., Roulstone, S., & Wren, Y. (2020). Linguistic comprehension and narrative 
skills predict reading ability: A 9-year longitudinal study. British Journal of 
Educational Psychology. 
Bagci, S.C. (2018). Does everyone benefit equally from self-efficacy beliefs? The moderating 
role of perceived social support on motivation. Journal of Early Adolescence, 38(2), 
204-219. 
Bagwell, C.L., & Bukowski, W.M. (2018). Friendship in childhood and adolescence: Features, 
effects, and processes. In Bukowski, W.M., Laursen, B., & Rubin, K.H. (Eds.), 
Handbook of peer interactions, relationships, and groups. Second edition. New York 
City: Guilford Press. 
Ballou, D., Sanders, W., & Wright, P. (2004). Controlling for student background in value-
added assessment of teachers. Journal of Educational and Behavioral Statistics, 
29(1), 37-65. 
Bellmore, A.D., Nishina, A., Witkow, M.R., Graham, S., & Juvonen, J. (2007). The influence 
of classroom ethnic composition on same- and other-ethnicity peer nominations in 
middle school. Social Development, 16(4), 720-740. 
Bourdieu, P. (1986). The forms of capital. In Richardson, J.G. (Ed.), The Handbook of 
Theory: Research for the Sociology of Education. New York: Greenwood Press. 
 91 
Bowker, J.C.W., Rubin, K.H., Burgess, K.B., Booth-Laforce, C., & Rose-Krasnor, L. (2006). 
Behavioral characteristics associated with stable and fluid best friendship patterns in 
middle childhood. Merrill-Palmer Quarterly, 52(4), 671-693. 
Braatz, J., & Putnam, R.D. (1996). Families, communities, and education in America: 
Exploring the evidence. In Wehlage, G.G., & White, J.A. (Eds.), Rebuilding the 
village: Social capital and education in America. London: Falmer Press. Retrieved 
from https://files.eric.ed.gov/fulltext/ED412637.pdf 
Bukowski, W.M., Laursen, B., & Rubin, K.H. (Eds.; 2018). Handbook of peer interactions, 
relationships, and groups. Second edition. New York City: The Guilford Press. 
Burk, W.J., Steglich, C.E.G., & Snijders, T.A.B. (2007). Beyond dyadic interdependence: 
Actor-oriented models for co-evolving social networks and individual behaviors. 
International Journal of Behavioral Development, 31(4), 397-404. 
Cardoza, J.A., & Brown, K.M. (2020). Supporting newly arrived Latinx students in upper 
elementary mathematics through one-way dual-language approaches. TESOL 
Journal, 11(2), 487. 
Carolan, B.V. (2014). Social network analysis and education: Theory, methods, & 
applications. Thousand Oaks, CA: Sage Publications. 
Carhill-Poza, A. (2015). Opportunities and outcomes: The role of peers in developing the 
oral academic English proficiency of adolescent English learners. The Modern 
Language Journal, 99(4), 678-695. 
Carlo, M.S., August, D., Mclaughlin, B., Snow, C., Dressler, C., Lippman, D., Lively, T.J., & 
White, C.E. (2008). Closing the gap: Addressing the vocabulary needs of English-
language learners in bilingual and mainstream classrooms. Journal of Education, 
189(1), 57-76. 
Center on Education Policy. (2006). High school exit exams: Special problems affecting 
English language learners. Exit Exams policy brief. Washington, DC: Center on 
 92 
Education Policy. Retrieved from http://www.cep-
dc.org/index.cfm?fuseaction=pages.viewPage&pageId=567%nodeID=1 
Chen, X., Lee, J., & Chen, L. (2018). Culture and peer relationships. In Bukowski, W.M., 
Laursen, B., & Rubin, K.H. (Eds.), Handbook of peer interactions, relationships, and 
groups. Second edition. New York City: Guilford Press. 
Cillessen, A.H.N., & Rose, A.J. (2005). Understanding popularity in the peer system. Current 
Directions in Psychological Science, 14, 102-105. 
Cranley Gallagher, K. (2013). Guiding children?s friendship development. Young Children, 
68(5), 26-32. 
Dabach, D.B. (2015). Teacher placement into immigrant English Learner classrooms: 
Limiting access in comprehensive high schools. American Education Research 
Journal, 52(2), 243-274. 
Daly, A. J. (2010). Social network theory and educational change. Cambridge, MA: Harvard 
Education Press. 
Delgado, M., Ettekal, A., Simpkins, S., & Schaefer, D. (2016). How do my friends matter? 
Examining Latino adolescents? friendships, school belonging, and academic 
achievement. Journal of Youth & Adolescence, 45(6), 1110-1125. 
Dotterer, A.M., McHale, S.M., & Crouter, A.C. (2007). Implications of out-of-school activities 
for school engagement in African American adolescents. Journal of Youth and 
Adolescence, 36, 391-401. 
Elreda, L.M., Kibler, A., Futch Ehrlich, V.A., & Johnson, H. (2016). Learning in linguistically 
diverse middle school classrooms: The role of the classroom peer network. Society 
for Research on Educational Effectiveness.  
Flores-Gonzalez, N. (2005). Popularity versus respect: School structure, peer groups, and 
Latino academic achievement. International Journal of Qualitative Studies in 
Education, 18, 625-642. 
 93 
G?ndara, P., & Maxwell-Jolly, J. (2005). Critical issues in developing the teacher corps for 
English learners. In T?llez, K., & Waxman, H.C. (Eds.), Preparing quality educators 
for English language learners: Research, policies, and practices. Mahway, NJ: 
Erlbaum.  
G?ndara, 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. Santa Cruz, CA: Center for the Future of Teaching 
and Learning. 
Garc?a-Vazquez, E., Vazquez, L., L?pez, I., & Ward, W. (1997). Language proficiency and 
academic success: Relationships between proficiency in two languages and 
achievement among Mexican American students. Bilingual Research Journal, 21, 
333-347. 
Goodenow, C. (1993). Classroom belonging among early adolescent students: Relationships 
to motivation and achievement. Journal of Early Adolescence, 13, 21-43.  
Gorund. (2015). nwcommands. Stata package. 
Gottfried, M.A. (2014). Peer effects in urban schools: Assessing the impact of classroom 
composition on student achievement. Educational Policy, 28(5), 607-647. 
Graham, S., & Echols, L. (2018). Race and ethnicity in peer relations research. In Bukowski, 
W.M., Laursen, B., & Rubin, K.H. (Eds.), Handbook of peer interactions, 
relationships, and groups. Second edition. New York City: Guilford Press. 
Hanushek, E.A., Kain, J.F., & Rivkin, S.G. (2009). New evidence about Brown v. Board of 
Education: The complex effects of school racial composition on achievement. Journal 
of Labor Economics, 27(3), 349-383. 
Harris, D.N. (2010). How do school peers influence student educational outcomes? Theory 
and evidence from economics and other social sciences. Teachers College Record, 
112(4), 1163-1197. 
 94 
Hartl, A.C., DeLay, D., Laursen, B., Denner, J., Werner, L., Campe, S., & Ortiz, E. (2015). 
Dyadic instruction for middle school students: Liking promotes learning. Learning & 
Individual Differences, 44, 33-39. 
Helfrich, S.R., & Bosh, A.J. (2011). Teaching English language learners: Strategies for 
overcoming barriers. The Educational Forum, 75(3), 260-270. 
Jacobson, L. (2000). Arizona board approves more flexibility for ELs under English-only law. 
K-12 Dive. Retrieved from https://www.k12dive.com/news/arizona-board-approves-
more-flexibility-for-els-under-english-only-law/571185/ 
Jackson, C.K., Porter, S.C., Easton, J.Q., Blanchard, A., & Kiguel, S. (2021). Linking social-
emotional learning to long-term success. Education Next, 21(1), 64-71.  
Jones, T.M., Fleming, C., & Williford, A. (2020). Racial equity in academic success: The role 
of school climate and social emotional learning. Children of Youth Services Review, 
119. 
Juvonen, J. (2006). Sense of belonging, social bonds, and school functioning. Handbook of 
Educational Psychology, 2, 655-674. 
Kindermann, T.A. (2007). Effects of naturally existing peer groups on changes in academic 
engagement in a cohort of sixth graders. Child Development, 78(4), 1186-1203. 
Kindermann, T.A. (2016). Peer group influences on students? academic motivation. In 
Wentzel, K.R., & Ramani, G.B. (Eds.), Handbook of social influences in school 
contexts (p. 31-47). New York: Routledge. 
Kindermann, T.A., & Gest, S.D. (2018). The peer group: Linking conceptualizations, 
theories, and methods. In Bukowski, W.M., Laursen, B., & Rubin, K.H. (Eds.), 
Handbook of peer interactions, relationships, and groups. Second edition. New York 
City: Guilford Press. 
Kory-Westlund, J.M., & Breazeal, C. (2019). A long-term study of young children?s rapport, 
social emulation, and language learning with a peer-like robot playmate in preschool. 
Front. Robot. AI, 6(81). 
 95 
Krone, M.W., & Yu, S. (2019). Promoting friendship development in inclusive early childhood 
classrooms: A literature review. 11(2), 183-193. 
Ladd, G.W., & Troop-Gordon, W. (2003). The role of chronic peer difficulties in the 
development of children?s psychological adjustment problems. Child Development, 
74(5), 1344-1367. 
Laursen, B., & Hartup, W.W. (2002). The origins of reciprocity and social exchange in 
friendships. In W.G. Graziano & B. Laursen (Eds.), Social exchange in development: 
New directions for child and adolescent development (No. 95, p. 27-40). San 
Francisco: Jossey-Bass. 
LeClair, C., Doll, B., Osborn, A., & Jones, K. (2009). English language learners? and non-
English language learners? perceptions of the classroom environment. Psychology in 
the Schools, 46(6), 568-577. 
Lereya, S.T., Copeland, W.E., Costello, E.J., & Wolke, D. (2015). Adult mental health 
consequences of peer bullying and maltreatment in childhood: Two cohorts in two 
countries. The Lancet Psychiatry, 2(6), 524-531. 
MacDonnell, M., McClain, K., Ganguli, A., & Elias, M.J. (2021). It?s not all or nothing: 
Exploring the impact of a social-emotional and character development intervention in 
the middle grades. Research in Middle Level Education, 44(2). 
McGrew, K.S., LaForte, E.M., & Schrank, F.A. (2014). Woodcock Johnson IV Technical 
Manual. Rolling Meadows, IL: Riverside.  
Nansel, T.R., Overpeck, M., Pilla, R.S., Ruan, W.J., Simons-Morton, B., & Scheidt, P. (2001). 
Bullying behaviors among US youth: Prevalence and association with psychosocial 
adjustment. Journal of the American Medical Association, 285, 2094-2100. 
NCES (2020). English Language Learners in public schools. Retrieved from 
https://nces.ed.gov/programs/coe/indicator_cgf.asp 
Nelson, R.M., & DeBacker, T.K. (2008). Achievement motivation in adolescents: The role of 
peer climate and best friends. Journal of Experimental Education, 76(2), 170-189. 
 96 
Niehaus, K., Rudasill, K.M., & Adelson, J.L. (2012). Self-efficacy, intrinsic motivation, and 
academic outcomes among Latino middle school students participating in an after-
school program. Hispanic Journal of Behavioral Sciences, 34(1), 118-136. 
O?Neal, D.D., Ringler, M., & Rodriguez, D. (2008). Teachers? perceptions of their preparation 
for teaching linguistically and culturally diverse learners in rural Eastern North 
Carolina. The Rural Educator, 30(1). https://doi.org/10.35608/ruraled.v30i1.456 
Park, S., Stone, S.I., & Holloway, S.D. (2017). School-based parental involvement as a 
predictor of achievement and school learning environment: An elementary school-
level analysis. Children & Youth Services Review, 82, 195-206. 
Rambaran, J.A., Hopmeyer, A., Schwartz, D., Steglich, C., Badaly, D., & Veenstra, R. 
(2017). Academic functioning and peer influences: A short-term longitudinal study of 
network-behavior dynamics in middle adolescence. Child Development, 88(2), 523-
543. 
Ramirez, J.D., Yuen, S.D., & Ramey, D.R. (1991). Longitudinal study of structured English 
immersion strategy, early-exit, and late-exit transitional bilingual education 
programs for language-minority children. Washington, DC: U.S. Department of 
Education. 
Roeser, R.W., Midgley, C., & Urdan, T.C. (1996). Perceptions of the school psychological 
environment and early adolescents? psychological and behavioral functioning in 
school: The mediating role of goals and belonging. Journal of Educational 
Psychology, 88(3), 408-422.  
Rumberger, R.W., & G?ndara, P. (2000). The schooling of English Learners. UC Berkeley: 
University of California Linguistic Minority Research Institute. Retrieved from 
https://escholarship.org/uc/item/4k87d8th 
Ryabov, I. (2009). The role of peer social capital in educational assimilation of immigrant 
youths. Sociological Inquiry, 79(4), 453-480. 
 97 
Schrank, F.A., Mather, N., & McGrew, K.S. (2014). Woodcock-Johnson IV Tests of 
Achievement. Rolling Meadows, IL: Riverside. 
Silverman, R.D., Johnson, E., Keane, K., & Khanna, S. (2020). Beyond decoding: A meta-
analysis of the effect of language comprehension interventions on K-5 students? 
language and literacy outcomes. Reading Research Quarterly, 55(S1), S207-S233. 
Snyder, J., Prichard, J., & Schrepferman, L. (2004). Child impulsiveness-attention, early 
peer experiences, and the development of early onset conduct problems. Journal of 
Abnormal Child Psychology, 32(6), 579-594. 
Swain, M., & Brooks, L.A. (2002). Peer-peer dialogue as means of second language 
learning. Annual Review of Applied Linguistics, 22, 171-185. 
Thomas, W., & Collier, V. (2003). A national study of school effectiveness for language 
minority students? long-term academic achievement. Santa Cruz, CA: University of 
California, Center for Research on Education, Diversity, and Excellence. 
Usher, E.L., Li, C.R., Butz, A.R., & Rojas, J.P. (2019). Perseverant grit and self-efficacy: Are 
both essential for children?s academic success? Journal of Educational Psychology, 
111(5), 877-902. 
 
 
 
 
 
 
 
 
 
 
 
 98 
Appendices 
Appendix 1 
Regression model predicting standardized oral comprehension growth using friendship 
centrality for English Learners 
Variable Estimate  t-value p-value 
(s.e.) 
Friendship Centrality 0.12 2.85 .009 
(Closeness) (0.04) 
Prior oral 0.85 20.60 < .0001 
comprehension (Fall (.04) 
2018)  
Female ?0.10 ?1.03 .313 
(0.10)  
Free or reduced- 0.02 0.11 .912 
price lunch (0.18) 
Grade 3 0.09 0.68 .503 
(0.13) 
Grade 4 0.10 0.85 .401 
(0.12) 
    
Constant ?0.10 ?0.45 0.655 
(0.22) 
 
    
N 160   
?(6, 24) 112.06  < .0001 
R-squared 0.658   
Note: Standard errors were adjusted for the 25 clusters. The outcome variable was 
students? oral comprehension in Spring 2019. The comparison group for the dummy-coded 
grade variable is students in Grade 5. All units are standardized units except demographic 
variables. 
 
  
 99 
Appendix 2 
Regression model predicting standardized reading comprehension growth using friendship 
centrality for English Learners 
Variable Estimate  t-value p-value 
(s.e.) 
Friendship Centrality ?0.03 ?1.57 .128 
(Closeness) (0.02) 
Prior reading 0.36 15.43 < .0001 
comprehension (Fall (0.02) 
2018) 
Female ?0.07 ?2.05 .052 
(0.04)  
Free or reduced- 0.19 2.32 0.029 
price lunch (0.08) 
Grade 3 0.04 0.70 .488 
(0.06) 
Grade 4 0.01 0.34 .737 
(0.04) 
    
Constant ?0.23 ?2.49 .020 
(0.09) 
    
N 168   
?(6, 25) 194.11  < .0001 
R-squared 0.677   
Note: Standard errors were adjusted for the 26 clusters. The outcome variable was 
students? reading comprehension in Spring 2019. The comparison group for the dummy-
coded grade variable is students in Grade 5. All units are standardized units except 
demographic variables. 
 
 
 
  
 100 
Appendix 3 
Regression model predicting standardized oral language comprehension growth using 
friendship centrality and English language status, full sample 
Variable Estimate  t-value p-value 
(s.e.) 
Friendship Centrality 0.09 1.50 .146 
(Closeness) (0.06) 
English language ?0.15 ?1.89 .071 
status (0.08)  
Friendship Centrality 0.03 0.51 .613 
* English language (0.07) 
status 
Prior oral 0.84 23.30 < .0001 
comprehension (Fall (0.04) 
2018) 
Female ?0.13 ?1.61 .120 
(0.08)  
Free or reduced- ?0.12 ?1.48 .152 
price lunch (0.08)  
Grade 3 0.10 0.90 .378 
(0.11) 
Grade 4 0.16 1.46 .156 
(0.11) 
    
Constant 0.17 1.51 .143 
(0.11) 
    
N 221   
?(8, 24) 151.22  < .0001 
R-squared 0.694   
Note: Standard errors were adjusted for the 25 clusters. The outcome variable was 
students? oral language comprehension in Spring 2019. The comparison group for the 
dummy-coded grade variable is students in Grade 5. All units are standardized units except 
demographic variables. 
 
  
 101 
Appendix 4 
Regression model predicting standardized reading comprehension growth using friendship 
centrality and English language status, full sample 
Variable Estimate  t-value p-value 
(s.e.) 
Friendship Centrality 0.08 2.05 .051 
(Closeness) (0.04) 
English language ?0.10 ?2.06 .050 
status (0.05)  
 
Friendship Centrality ?0.09 ?2.52 .018 
* English language (0.03)  
status  
Prior oral 0.31 8.18 < .0001 
comprehension (Fall (0.04) 
2018) 
Female ?0.03 ?1.04 .307 
(0.03)  
Free or reduced- 0.02 0.34 .740 
price lunch (0.07) 
Grade 3 ?0.03 ?0.40 .689 
(0.08)  
Grade 4 ?0.02 ?0.30 .763 
(0.06)  
    
Constant 0.02 0.18 .859 
(0.09) 
    
N 229   
?(8, 25) 38.94  < .0001 
R-squared 0.633   
Note: Standard errors were adjusted for the 26 clusters. The outcome variable was 
students? oral language comprehension in Spring 2019. The comparison group for the 
dummy-coded grade variable is students in Grade 5. All units are standardized units except 
demographic variables. 
 
 
 
  
 102 
 
 
 
 
 
 
 
 
Do Executive Functioning Skills Predict Reading Comprehension Growth? 
 
 
 
 
 
 
 
 
Paper 3 
 
 
 
 
 
 
 
 
 103 
 Developmentally, children have cognitive processes that may facilitate their 
academic success. Executive functioning skills are one set of cognitive processes that have 
been a primary focus in the developmental psychology literature base because researchers 
have linked these skills to various academic outcomes, including one?s reading 
comprehension (Cartwright et al., 2017; Cartwright et al., 2020; Locascio et al., 2010). 
Understanding which skills are associated with and predict executive functioning 
development can support researchers as they devise interventions to bolster students? goal-
planning abilities, and, more importantly, the outcomes that are associated with increased 
executive functioning, such as increased numeracy and mathematics performance (e.g., 
Agostino, Johnson, & Pascual-Leone, 2010; Raghubar, Barnes, & Hecht, 2010), increased 
literacy (Monette, Bigras, & Guay, 2011), and social and emotional well-being (Best, Miller, 
& Jones, 2009).  
However, the causal claims in the research base are limited for a variety of reasons, 
including small sample sizes, short-term studies, and the possibility of bidirectionality 
between executive functioning and reading comprehension. Specifically, questions about 
whether executive functioning leads to reading comprehension growth, or vice versa, or 
even if the skills develop together, makes intervention creation challenging unless 
researchers know which skills should be targeted first to best improve student outcomes. 
Additionally, prior studies generally focus on one component of executive functioning (e.g., 
cognitive flexibility, working memory, or inhibition) rather than combining all components 
into a composite measure.  This study contributes to the research base by using longitudinal 
data from children in five elementary schools, by incorporating all components of executive 
functioning, and by using methodology that more rigorously supports causal claims than 
correlational analyses or pre-post differences.   
Recent work has examined the bidirectionality of reading comprehension and 
executive functioning, but largely using theory and correlational analyses (see Peng & 
Kievit, 2020 for a review). This line of work suggests that students? executive functioning 
 104 
and cognitive processes develop alongside their reading and mathematics skills, and that 
both likely influence the other through a theory known as mutualism (van der Maas et al., 
2006). However, the samples used to study this relation are generally typically developing 
children (Peng & Kievet, 2020), and the theoretical and correlational analyses do not 
definitively answer whether the constructs are causally linked or merely mutually 
developing. Studies using children experiencing developmental delays have not 
demonstrated this bidirectional relationship (Ferrer et al., 2010), instead supporting the 
traditional view that executive functioning leads to improvements in reading and 
mathematics. Furthermore, studies have rarely explored the relation using other groups 
who are often underserved by the education system, such as English Learners (Taboada 
Barber et al., 2020).  
This study explores the relation between reading comprehension and executive 
functioning using student data collected within the classroom over the course of an 
academic year. I provide evidence that executive functioning contributes to elementary 
students? reading development. I use ordinary least squares (OLS) multiple linear regression 
with a rich set of covariates, aiming to limit concerns of omitted variable bias and provide 
stronger evidence of the connection between the constructs than with correlation. This 
model contributes evidence that executive functioning skills in part lead students to develop 
their reading comprehension skills. This paper concludes with pedagogical implications from 
this set of findings, as well as limitations from the study?s methods. 
 
Literature Review 
Theoretical Underpinnings 
Reading comprehension [RC] is a complex skill that is critical for many life outcomes, 
such as knowledge acquisition, educational attainment, and employment (Ricketts, 
Sperring, & Nation, 2014; Smart et al., 2017). However, many children struggle to meet 
grade-level expectations on RC tests (NAEP, 2017). While word decoding skills significantly 
 105 
predict reading comprehension (Garc?a & Cain, 2014), some children display deficits in 
reading comprehension even though they exhibit age-appropriate abilities in word decoding 
(Oakhill & Cain, 2017). Further, students with above-average word decoding skills can 
demonstrate below-average reading comprehension (Applegate et al., 2009), which 
suggests that there is more to reading comprehension than just one?s ability to ?read? (i.e., 
decode) the words. This deficit has led researchers to explore phenomena beyond 
comprehension abilities that might explain this finding.  
Of note, research has demonstrated that children who display deficits in reading 
comprehension also have deficits in coordinating thought processes to achieve cognitive 
goals, also known as executive functioning [EF] skills (Miller & Wallis, 2009). Cartwright et 
al. (2017) define executive functioning as ?cognitive processes that enable individuals to 
manage and direct their thinking toward particular goals? (p. 34), and the ability to manage 
goals is particularly important for students? reading success. Researchers have identified 
three components of executive functioning: inhibition, working memory, and cognitive 
flexibility (Diamond, 2013). 
Inhibition is one?s ability to suppress information that one sees or hears, and instead 
follow directions to report other information (Pimperton & Nation, 2010). For example, if a 
teacher primes a student to always report an ?up? arrow as ?down,? the student is 
successful at inhibiting their typical response when they correctly call the up arrows ?down.? 
Inhibition is connected to reading because inhibition allows children to ignore surroundings 
and other stimuli, instead being able to focus on the text. These stimuli can include words 
or phrases in the text that are unimportant to the overall understanding of the story 
(Butterfuss & Kendeou, 2018). Additionally, children with stronger inhibition skills are better 
able to retain directions and reading strategies that teachers prime children with prior to 
engaging in the reading activity.  
The second executive functioning component, which is highly related to inhibition, is 
working memory. Working memory involves one?s ability to recall a limited number, or 
 106 
?bits?, of information (Cain, Oakhill, & Bryant, 2004). Studies suggest that students who can 
retain more pieces of information from working memory might be more successful at 
reading because they have a stronger ability to recall the words from the sentences while 
constructing, and remembering, a model of text meaning.  
The final core component of executive functioning is cognitive flexibility, which is 
one?s ability to switch back and forth between elements of tasks (Cartwright et al., 2017). 
For example, a student who can sort pictures by both color (red versus yellow) and type 
(fruits versus flowers) would have higher cognitive flexibility than a student who can only 
sort using one dimension at a time (Cartwright et al., 2017).  Cognitive flexibility is 
important because readers need to simultaneously navigate decoding the words and 
sentences while understanding the context of what they are reading (Cartwright et al., 
2010).  
Because reading relies, in part, on skills connected with EF, such as focusing, 
decoding, sorting, and memory (Garc?a & Cain, 2014), it is reasonable to test how closely 
tied the two constructs are. Given links between reading comprehension and executive 
functioning, some have hypothesized that a critical way to increase the former is to build 
skills in the latter.  Indeed, there is a robust body of literature that suggests that executive 
functioning skills are associated with reading comprehension (e.g., Cartwright et al., 2017; 
De Beni, Palladino, Pazzaglia, & Cornoldi, 1998; Oakhill, Hartt, & Samols, 2005; Yuill, 
Oakhill, & Parkin, 1989). For example, the literature base suggests that students with 
higher inhibition, working memory, or cognitive flexibility skills tend to be the same 
students who are strong comprehenders.  Numerous studies demonstrate the connection 
between executive functioning skills and reading comprehension (see Butterfuss & Kendeou, 
2018, for a review). 
 
 
 107 
Challenges Testing the Relation between Executive Functioning and Reading 
Comprehension 
While the connection between the constructs is clear, extant research has not been 
conclusive on the directionality between EFs and RC. There is a body of literature in the 
developmental psychology field that shows evidence for a directional relation between EF 
and RC (Dahlin, 2011; Garc?a-Madruga et al., 2013), yet a separate body of literature within 
the broader cognitive psychology field demonstrates bidirectionality between EFs and RC for 
students in early grades (Peng & Kievit, 2019). More evidence is needed to determine 
whether growth in one construct leads to changes in the other construct, or whether both 
are developing simultaneously.  
Researchers have primarily used two methods of investigating the relation between 
EF and RC, each with its own strengths and limitations. Longstanding lines of research have 
focused on conceptualizing and measuring these distinct constructs, and then examining the 
relation between them (i.e., Cain, Oakhill, & Bryant, 2004; Cain, 2006; Cutting & 
Scarborough, 2006). While the research has demonstrated the association between the 
constructs, most researchers have used correlation analysis that is limited in its ability to 
demonstrate that student growth in EF can lead to change in RC.  
A second line of research has examined the extent to which classroom or laboratory-
based interventions improve RC and also EF (i.e., Cartwright, 2002; Cartwright et al., 2017; 
Dahlin, 2011; Garc?a-Madruga et al., 2013). Intervention work, in which the treatment is 
randomly assigned, is useful in supporting causal conclusions about the effect of that 
intervention on RC and/or EF. However, to date, the interventions involve a component of 
reading instruction in order to support students? executive functioning, which prevent 
researchers from untangling which construct?reading comprehension or executive 
functioning?develops first, and whether the constructs lead to one another.  
 Omitted variable bias has been the largest threat to internal validity claims in the 
literature on executive functioning and reading comprehension abilities (Cartwright et al., 
 108 
2017). A typical method involves classifying participants as either high or low on RC, and 
then testing if the groups have significant EF differences.  For example, De Beni et al. 
(1998) used this method to distinguish among students with high versus low working 
memory capabilities, and then the researchers used the classification to test for significant 
reading comprehension abilities between the two groups. In this study, the researchers 
found that students with high working memory skills performed better on reading 
comprehension tasks, so they were confident that there is a connection between these two 
skills. However, they were unable to determine which construct developed first, or if there 
might be another trait that causes both high executive functioning and high reading 
comprehension. Without using methodology that can plausibly answer this question, one 
cannot rule out the possibility that the relation is spurious, or that the causal link actually 
exists in the direction that is opposite from what the researcher might expect (Murnane & 
Willet, 2010). For example, strong reading comprehension skills might cause a student to 
have higher-than-average working memory abilities, since the student has more practice at 
reading stories, and this skill could influence their ability to recall information. A causal 
framework could tease out this distinction and serve to illuminate the direction of the 
relation (Murnane & Willett, 2010). 
 Finding a directional relation between executive functioning skills and reading 
comprehension, using more robust methods, would be important because there are 
intervention methods that might allow schools to improve a students? executive functioning 
abilities in order to affect their reading comprehension. However, few studies demonstrate 
the effectiveness of executive functioning interventions to specifically target reading 
comprehension, and especially within an authentic classroom setting (Cartwright et al., 
2017). Those studies that do exist generally target one component of executive functioning, 
rather than executive functioning writ large as a construct. For example, Cartwright et al. 
(2017) tested this relation within a classroom setting and did find that teachers were able to 
improve reading comprehension scores after implementing a cognitive flexibility intervention 
 109 
in the classroom, so there is evidence that executive functioning interventions targeting 
cognitive flexibility can lead to increases in reading comprehension. However, the 
aforementioned study only examined one classroom, and it used only one component of 
EFs. The present study examines if Cartwright et al.?s (2017) findings generalized to a 
larger sample while using a causal framework. Furthermore, this study uses information 
from all three types of executive functioning skills (i.e., cognitive flexibility, working 
memory, and inhibition) to determine if the skillset overall improves reading 
comprehension. 
 
Methods 
In the current study, I explore the connection between EFs and RC using a sample of 
elementary-aged children. The goal of using covariate-adjusted regression is to limit biases 
in estimating the effect of EFs on RC, thereby producing a closer approximation of how 
growth in executive functioning skills can develop one?s reading comprehension abilities. 
This study contributes to the broader literature by exploring the directionality for 
elementary-aged students who are developing their reading capabilities. Understanding the 
extent to which EFs contribute to students? reading comprehension can help teachers and 
policymakers understand whether interventions that target EFs can supplement reading 
comprehension development instruction for students. 
 
Sample 
 This study is part of Project LEARN (Language, Executive skills, And Reading for 
eNgagement; Taboada Barber, Cartwright, & Stapleton, 2016). The larger study sought to 
explore how English Learners? reading comprehension develops alongside domain-general 
 110 
cognitive skills such as EFs, in English Monolingual and English Learners in the elementary 
grades.6    
 For the present study, data from the first year of Project LEARN are utilized. Table 1 
displays demographic information for the analytic sample, including the grade-level 
proportions and English Learner proportions. The analytic sample in the present study was 
limited to those students whose executive functioning skills were measured in both the fall 
and the spring school semesters, which means students who missed one observation point 
were excluded from the present sample. The total sample size was 672 students (out of the 
original 762 students who Project LEARN collected data from across the two semesters).  
 
Table 1  
Demographic Information. 
 Frequency Percentage 
English Learners 392 41.67 
English Monolinguals 280 58.33 
Grade 1 152 22.62 
Grade 2 168 25.00 
Grade 3 189 28.12 
Grade 4 163 24.26 
School 1 254 37.8 
School 2 267 39.73 
School 3 151 22.47 
Speaks Spanish at Home 426 63.39 
Speaks English at Home 555 82.59 
 
6 For example, there is an explicit focus on measuring traditional reading abilities, such as 
comprehension, decoding, inference making, vocabulary, and oral comprehension. The 
study also collects information on executive functioning skills, theory of mind, motivation, 
reading engagement, and social network data. The goal of Project LEARN is to determine if 
the predictors for reading development differ between English Learners and English 
Monolinguals so that interventionists and policymakers can determine the best way to 
support the two different groups. In year one of the study, the project collected data from 
three schools in a medium-sized urban area in the Mideastern United States. These schools 
were selected because they were located in high-proportion immigrant areas, meaning the 
schools had high percentages of English Learners in comparison to other schools in the 
state. All students in grades one through four who did not have special education 
accommodations formed the initial analytic sample (as long as consent was given), unless 
the student could not understand directions given to them in English.  
 111 
Speaks Other Lang. at Home 43 6.40 
Male 338 50.75 
Female 328 49.25 
Qualifies for Free/Reduced-
558 83.78 
Price Lunch 
Average Days Absent per 
3.94 ? 
Semester 
TOTAL STUDENTS 672 ? 
 
Measures 
 Various measures were collected over the course of the two semesters. A Project 
LEARN assessor individually administered the battery of assessments to each child in a one-
hour assessment block. Assessments took place during the school day, both in the fall and 
spring semesters, and the assessments generally occurred in an empty classroom, the 
hallway, or a book closet, depending on where the school could offer space to the 
assessment team. Assessors were trained on each measurement, and they had to pass two 
fidelity checks with one of the project leaders in order to administer tests to students.  
Following the hypothesis that executive functioning leads to changes in reading 
comprehension, the outcome measure in this study is the student?s reading comprehension 
score in the spring semester. Reading comprehension was measured using the Woodcock-
Johnson IV passage comprehension reading subtest (Schrank, Mather, & McGrew, 2014), a 
standardized reading assessment that is given nationally to people of all ages. This test was 
selected because it has both a low floor and a high ceiling, meaning it is sensitive enough to 
differentiate among levels of reading at all points on the reading comprehension spectrum. 
Additionally, it was a vertically scaled test, so scores across grade levels align with students? 
developmental trajectories. Possible scores range from 0 to 54.  Split-half reliability 
coefficients for children ages 6-10 range from 0.89 to 0.98 (McGrew et al., 2014).  
 The primary predictor variable was a student?s executive functioning abilities, so 
special care was dedicated to creating a variable that represented all aspects of executive 
functioning. To capture inhibition, Project LEARN used the Inhibition subtest from the 
 112 
NEPSY-II standardized measure. Students were asked to name the arrows/shapes as 
quickly as they could without making mistakes, and then they were asked to inhibit what 
they were seeing and report the opposite arrow/shape. The ?Naming? portion of the test 
establishes a baseline time and level of accuracy, and then the ?Inhibition? portion of the test 
allows assessors to see how the student?s score changes when asking students to inhibit 
what they are seeing. Inhibition scores were calculated by dividing the number of correct 
responses by their time (in seconds), meaning students with higher scores were more 
accurate and proceeded more quickly than students with lower scores.  At Time 1 (the fall 
semester), scores ranged from 0 to 1.55 (? = 0.78, ?? = 0.24), and at Time 2 (the spring 
semester), scores ranged from 0.10 to 1.74 (? = 0.90, ?? = 0.26). Internal consistency in the 
fall semester was 0.77 and 0.85 in the spring semester (Taboada Barber et al., 2020).  
 The second component of executive functioning is working memory. Project LEARN 
used the Letters Backward Subtest of the Test of Memory and Learning-2 (TOMAL-2) 
assessment to measure this variable. This assessment involves reading series of letters to 
students of increasing length, and then the student was asked to report the letters 
backwards. This measurement allows us to assess the number of bits a student can hold 
and transform (i.e., by reversing the order of the original list) in their working memory. A 
student?s final score was the total number of letters they could remember in order after 14 
different items. The Time 1 scores ranged from 0 to 53 (? = 10.65, ?? = 4.65) and the Time 2 
scores ranged from 0 to 53 (M = 11.67, SD = 5.12). Again, higher scores indicate higher 
levels of working memory. Internal consistency in the fall semester was 0.78 in the fall 
semester and 0.84 in the spring semester (Taboada Barber et al., 2020).  
 The last component of executive functioning is cognitive flexibility. This was 
measured using Cartwright et al.?s (2017; Cartwright, 2002; Col?, Duncan, & Blaye, 2014) 
task to assess cognitive flexibility, which asks students to sort pictures or words by two 
dimensions at the same time?for example, by type and by color. Like the inhibition 
measurement, a student?s score was composed of the number of correct responses they 
 113 
gave and the time it took the student to complete the task. Higher scores indicate that the 
student is able to sort correctly, explain how they sorted the cards, and complete the task 
quickly, which indicates higher cognitive flexibility abilities.  At Time 1, scores ranged from 0 
to 75 (? = 13.11, ?? = 11.82), and at Time 2, scores ranged from 0 to 30.86 (? = 9.00, ?? =
6.39). Internal consistency in the fall semester was 0.60 and 0.77 in the spring semester 
(Taboada Barber et al., 2020).  
 After creating scores for each of the three components of executive functioning, I 
standardized the scores so that a student?s score indicates how many standard deviations 
above/below the mean the student performed, within the total sample of students. Then, I 
averaged each student?s scores to form the composite executive functioning score per 
semester. I also calculated a student?s change in executive functioning from the fall to 
spring semester. The change in z-scores ranged from ?6.37 to 6.52 (? = 0.09, ?? = 1.52). 
 Other variables that form part of this study include the student?s school ID (a dummy 
variable to indicate which school they attended), their grade level, whether the student was 
identified by the district as an English Learner, and whether the student qualified for Free or 
Reduced-Price Lunch. 
 
 
 
Analytic Plan 
Covariate Adjustment. Because prior literature largely does not make use of 
methodology that limits concerns around omitted variable bias, I use two techniques that 
provide stronger evidence of the relation between executive functioning and reading growth. 
I start by conducting a multiple linear regression model with a robust set of covariates (see 
Levine & Painter, 2008). While in general, covariate adjustment has the weakest claim to 
causality (Murnane & Willet, 2010), using a rich set of covariates can strengthen claims that 
 114 
the participants are only being compared to other similar participants.7 This model will 
identify whether a bidirectional relationship exists among reading comprehension and EFs, 
in line with findings from the research base. Controlling for baseline reading comprehension 
scores and key demographic variables strengthens claims that participants are only being 
compared to similar students. 
 There are strengths and limitations to using covariate controls as a primary way of 
estimating effect size, in comparison to matching methods such as propensity score 
matching. The biggest strength is that the final model can incorporate all participants, 
unlike matching methods in which participants are unmatched and thereby left out of the 
model (see Leow, Wen, & Korfmacher, 2015). Increasing the number of participants 
incorporated into the model increases power and therefore the ability to make claims about 
differences between the two groups. These regression models are also more easily 
understood than propensity score matching since regression models are more common in 
the literature (NICHD Early Child Care Research Network & Duncan, 2003). However, the 
key limitation is that researchers must be certain that the covariates actually remove 
differences between the two comparison groups, or else they cannot expect the groups to 
be equal in expectation (Murnane, & Willett, 2010). Without this requirement satisfied, 
 
7 If the covariates can explain away any initial differences that exist between the two 
groups, then this set of variables minimizes the threat that selection bias poses, allowing for 
greater confidence that the groups are equal on expectation. Ideally, the covariables are the 
only differences between the two groups, allowing for the coefficient to represent the 
unbiased effect size that executive functioning skills have on reading comprehension 
abilities. However, it is challenging to identify and collect all such variables, particularly 
when confounding variables likely exist which may bias estimated effect sizes. Additionally, 
the final model must meet all regression model assumptions or else these models may 
introduce systematic bias into the estimated effect sizes (Zanutto, Lu, & Hornik, 2005). 
Researchers must determine theoretically driven covariates based on existing literature, and 
even then, the work is limited due to the possibility of lurking variables. Additional steps 
must be taken, such as verifying effect sizes using other methodologies, or else estimates 
may naively describe the relation among the constructs.  
 115 
conclusions may be invalid since differences other than the independent variable may drive 
the estimated effect size of the models.  
For these reasons, it is important to identify those variables that can strongly explain 
variation between participants across the spectrum of initial executive functioning skills. I 
include a set of reading variables to compare children whose reading skills would be 
expected to be similar, on average. These reading variables include initial English language 
vocabulary understanding and initial oral comprehension understanding. To test this set of 
variables, I ran a regression model to determine whether each variable significantly 
predicted one?s executive functioning; each variable was statistically significant, and the set 
of variables jointly were statistically significant, which provided evidence that this set of 
variables is associated with differences among children with differing levels of executive 
functioning abilities.  
 The regression model incorporates all specified covariates, and I use the change in 
executive functioning across the two semesters as the key independent variable. The 
regression model tested is: 
 
??????? ????_??????
= ?1??_??????? + ?2??_???????? + ?3??????? ????_???????? + ?4??????????_????????
+ ?5???? ????_???????? + ??? 
 
where ?1 through ?5 are the estimated coefficients in the model for student i in school j. The 
residuals (?) are clustered by school to account for nesting of students within schools. This 
model will indicate how growth in executive functioning is associated with growth in reading 
comprehension. In line with other studies that include initial scores and change scores on 
predictor variables (see Taylor et al., 2013), this model allows for an understanding of 
whether students? initial executive functioning or the amount of executive functioning 
growth in a semester are more predictive of reading comprehension growth. The goal in this 
 116 
model is to attempt to make the groups equal on expectation, meaning differences among 
students that are associated with both executive functioning and reading comprehension 
should remain in the final model. For that reason, the non-significant covariates remained in 
the final model.  
 
Results 
 In this section, I provide the results from the regression model predicting the 
estimated association between growth of executive functioning skills and reading 
comprehension development. I find that both models indicate a link between executive 
functioning and reading comprehension, and the match between the models strengthens 
claims to validity.  
The independent variable in this model is whether one?s change in executive 
functioning between the fall and spring semester can predict one?s reading comprehension, 
after controlling for initial reading variables. Does expanding one?s executive functioning 
skills lead a student to increase their reading comprehension? This question is explored in 
this model. 
  Table 2 contains the estimated regression model. As mentioned earlier, no non-
significant variables were omitted from this model to ensure the set of covariates was 
expansive enough to control for any omitted variables. Of course, it is not possible to ensure 
that there are no lurking variables, but I used the language literature base to identify any 
variable that I have access to and which theory suggests might predict reading 
comprehension. 
 117 
Table 2  
Multiple Regression Model Predicting Change in Reading Comprehension as a Function of 
Change in Executive Functioning Skills 
 Standardized Robust Standard T-statistic 
Coefficient Error 
Change in Executive 0.114 0.026 4.47*** 
Functioning Skills 
Initial Executive 0.102 0.016 6.38*** 
Functioning Skills 
Initial Reading 0.574 0.072 8.03*** 
Comprehension 
Initial Vocabulary 0.066 0.028 2.35** 
Initial Oral 0.112 0.040 2.84*** 
Comprehension 
Constant 0.020 0.020 0.97 
   ?(5,664) = 354.39 
? < .0001 
?2 = 0.74 
Note: The outcome variable is the student?s standardized reading comprehension score in 
the spring semester. School identification was absorbed in this model in order to cluster the 
residuals. N = 672 
*   ? <. 1 
** ? <. 05 
***? <. 01 
 
 
 This model indicates that increases in executive functioning are associated with 
increases in reading comprehension. Specifically, controlling for all covariates in the model, 
a 1-standard deviation increase in executive functioning skills from the fall to the spring 
semester is associated with an increase in reading comprehension of 0.114 standard 
deviations (? = 4.47, ? < .001). This predicted increase is associated with a child at the 50th 
percentile in reading comprehension in the fall semester moving to the 56th percentile in the 
spring semester. 
 This regression model also indicates that significant predictors of one?s change in 
reading comprehension include a student?s initial reading comprehension (? = 8.03, ? <. 001), 
a student?s initial English vocabulary knowledge (? = 2.35, ? = .02), and a student?s initial oral 
language comprehension (? = 2.84, ? =. 01). Additionally, students? initial executive 
 118 
functioning skills in the fall semester significantly predicted reading comprehension, even 
after controlling for the other variables (? = 3.67, ? <. 001).   
 The model provides evidence that executive functioning skills do matter in predicting 
changes in reading comprehension. Both initial executive functioning skills and change in 
executive functioning are associated with predicted increases in reading comprehension. Not 
only are those children with higher executive functioning skills experiencing greater 
increases in reading, but also those children who more strongly improve their executive 
functioning skills from the fall to the spring semester are expected to see increases in 
reading comprehension. Both indicators provide evidence that executive functioning is 
predictive of reading growth. 
 
Discussion 
 The regression model supports that students? cognitive functioning capabilities are 
statistically significantly associated with a student?s reading comprehension growth. The 
multiple regression model indicated that a 1-standard deviation change in executive 
functioning is associated with a student improving their reading skills by 0.114 standard 
deviations, controlling for all covariates. The model improves confidence in the finding that 
executive functioning skills do lead a student to improve their reading comprehension each 
semester. This finding has policy implications, which I discuss in this section. 
 Understanding the connection between EF and reading comprehension bolsters the 
idea that schools may need to rethink the way they conceptualize improving reading scores. 
?Drill and grill? test prep, especially for young kids, is clearly not the only way to increase 
test scores. Rather, school administrators can also focus on improving executive functioning 
skills, with the understanding that growth in these skills may lead students to improve their 
reading comprehension both immediately and over time.  There are interventions for both 
working memory and inhibition skills, and researchers are developing interventions to help 
students improve their cognitive flexibility skills (Cartwright et al., 2017). I do not suggest 
 119 
that district officials and school administrators should revamp reading curricula immediately, 
but these results lend credence to the possibility that executive functioning interventions 
should be further studied to determine their impact on reading comprehension. Future 
studies can investigate whether interventions targeting executive functioning can work in 
conjunction with reading instruction, thereby increasing student reading outcomes.  
 Furthermore, this study is part of a longitudinal study, so it will be important to see if 
this pattern holds as data are analyzed from the subsequent two years of the study. If 
students with higher executive functioning skills continue to improve their reading 
comprehension scores beyond the scores of their classmates with lower executive 
functioning skills, there would be further evidence that these skills matter for desired 
reading outcomes. It also may be possible that this trajectory is not linear over time; 
instead, the data may suggest that those students with higher executive functioning skills 
proceed along a parabolic or exponential trajectory, such that the students continuously 
exceed their peers by greater and greater amounts over time. If this pattern occurs, then 
there will be even stronger evidence that limited resources should be allocated to executive 
functioning skills for kids at younger ages, with the understanding that these interventions 
may lead to reading gains as the children get older. 
 Whereas other studies investigated connections among specific components of 
executive function, this study provides evidence that executive functioning overall matters 
for reading outcomes. Previous literature has suggested the importance of working memory 
and cognitive flexibility for reading outcomes (e.g., Cartwright et al., 2017; De Beni, 
Palladino, Pazzaglia, & Cornoldi, 1998; Oakhill, Hartt, & Samols, 2005; Yuill, Oakhill, & 
Parkin, 1989), but this study expands those findings to suggest that executive functioning 
writ large is connected with reading. Helping students to expand on their cognitive abilities 
may be one strategy that schools use to improve student outcomes, but future studies are 
necessary to support this claim.  
 120 
 Additionally, this study provides preliminary evidence that the relation between 
executive functioning and reading comprehension is unidirectional, at least for the 
population of interest. Using data from 2nd?5th-grade students, there is evidence that 
changes in executive functioning lead to increase in reading comprehension, at least in part. 
This finding aligns with research suggesting that executive functioning and reading 
comprehension are not bidirectional for particular populations of children (Ferrer et al., 
2010). Future work can continue to use more rigorous causal methods, such as 
experiments, to determine if gains in EF lead to gains in RC. 
 While this study does support the explanatory power of executive functioning on 
reading comprehension, it is important to note that the methods are not experimental in 
nature, and thus there is a possibility that the claims are overstated. The literature base 
would benefit from a ?gold-standard? experimental method, or even a more robust quasi-
experimental method, to ensure that these causal claims are justified. Since we cannot 
randomly assign executive functioning capabilities to students, we would need to stick to 
other rigorous methods to test the theory that executive functioning interventions improve 
reading comprehension. An intervention with multiple rounds of data collection, allowing for 
structural equation modeling to test the pathways among constructs, would strengthen 
causal claims and may improve internal validity claims from the present study.  
 There are other important limitations to this study. While there is evidence that 
executive functioning can lead to increases in reading comprehension, it would be negligent 
to claim that executive functioning on its own will improve children?s reading skills. 
Executive functioning can improve children?s planning skills, memory, desire and ability to 
reach goals, and other abilities that are correlated with academic successes, but there are 
other variables with which executive functioning likely works in tandem to predict reading 
outcomes. For example, strong reading teachers (e.g., Etim, Etim, & Blizard, 2020), as well 
as evidence-based literacy interventions (for a review, see Puzio, Colby, & Algeo-Nichols, 
2020), can both improve reading outcomes for students, and it is possible that executive 
 121 
functioning skills may moderate the relation between these school-based factors and 
reading outcomes. Future investigations would benefit from testing the extent to which 
children?s executive functioning skills play a role in mitigating children?s reading 
improvements. 
 Reading outcomes matter, both for test-based accountability systems and for helping 
students to become successful adults. The search to improve reading outcomes has led 
researchers and policymakers alike to find methods that can be used to strengthen practices 
in the classroom. One question that remained unclear from the literature was whether 
executive functioning led to increases in reading achievement, and I used two methods to 
demonstrate that there is predictive power for growth in executive functioning on reading 
comprehension growth. The present study can inform the debate about how best to serve 
students, since these executive functioning skills fulfill accountability needs of school 
administrators?raising test scores?and these skills also help students develop their 
thought processes and become critical, logical citizens. Policymakers should continue to 
investigate the role that executive functioning instruction can play in elementary schools so 
that schools can maximize the cognitive growth that students can expect as they proceed 
throughout their schooling. 
 
 
 
 
 
 
 
 
 
 
 122 
References 
Agostino, A., Johnson, J., & Pascual-Leone, J. (2010). Executive functions underlying 
multiplicative reasoning: Problem type matters. Journal of Experimental Child 
Psychology, 105, 286-305. 
Applegate, M.D., Applegate, A.J., & Modla, V. (2009). ?She?s my best reader; she just can?t 
comprehend?: Studying the relationship between fluency and comprehension. The 
Reading Teacher, 62, 512-521. 
Best, J.R., Miller, P.H., & Jones, L.L. (2009). Executive functions after age 5: Changes and 
correlates. Developmental Review, 29(3), 180-200.  
Bolli, T., Olivares, M., Bonaccorsi, A., Daraio, C., Aracil, A.G., & Lepori, B. (2016). The 
differential effects of competitive funding on the production frontier and the 
efficiency of universities. Economics of Education Review, 52, 91-104. 
Borella, E., Carretti, B., & Pelegrina, S. (2010). The specific role of inhibition in reading 
comprehension in good and poor comprehenders. Journal of Learning 
disabilities, 43(6), 541-552. 
Butterfuss, R., & Kendeou, P. (2018). The role of executive functions in reading 
comprehension. Educational Psychology Review, 30, 801-826. 
Cain, K. (2006). Individual differences in children's memory and reading comprehension: An 
investigation of semantic and inhibitory deficits. Memory, 14(5), 553-569. 
Cain, K., Oakhill, J., & Bryant, P. (2004). Children?s reading comprehension ability: 
Concurrent prediction by working memory, verbal ability, and component skills. 
Journal of Educational Psychology, 96, 31-42. 
Cartwright, K. B. (2002). Cognitive development and reading: The relation of reading-
specific multiple classification skill to reading comprehension in elementary school 
children. Journal of Educational Psychology, 94(1), 56. 
 123 
Cartwright, K.B., Coppage, E.A., Lane, A.B., Singleton, T., Marshall, T.R., & Bentivegna, T. 
(2017). Cognitive flexibility deficits in children with specific reading comprehension 
difficulties. Contemporary Educational Psychology, 50, 33-44. 
Cartwright, K.B., Marshall, T.R., Dandy, K., & Isaac, M.C. (2010). The development of 
graphnophonological-semantic cognitive flexibility and its contribution to reading 
comprehension in beginning readers. Journal of Cognition and Development, 11, 61-
85. 
Cartwright, K.B., Marshall, T.R., Huemer, C.M., & Payne, J.B. (2019). Cognitive flexibility 
supports reading fluency for typical readers and teacher-identified low-achieving 
readers. Research in Developmental Disabilities, 88, 42-52. 
Col?, P., Duncan, L.G., & Blaye, A. (2014). Cognitive flexibility predicts early reading skills. 
Frontiers in Psychology, 5, 565. 
Cutting, L.E., & Scarborough, H.S. (2006). Prediction of reading comprehension: Relative 
contributions of word recognition, language proficiency, and other cognitive skills can 
depend on how comprehension is measured. Scientific Studies of Reading, 10(3), 
277-299. 
Dahlin, K. I. (2011). Effects of working memory training on reading in children with special 
needs. Reading and Writing, 24(4), 479-491. 
De Beni, R., Palladino, P., Pazzaglia, F., & Cornoldi, C. (1998). Increases in intrusion errors 
and working memory deficit of poor comprehenders. The Quarterly Journal of 
Experimental Psychology. A. Human Experimental Psychology, 51, 305-320. 
Diamond, A. (2013). Executive functions. Annual Review of Psychology, 64, 135-168.  
Etim, J.S., Etim, A.S., & Blizard, Z.D. (2020). Teacher effects, student school attendance 
and student outcomes: Comparing low and high performing schools in North 
Carolina. Educational Research Quarterly, 44(2), 48-84. 
 124 
Garc?a, J.R., & Cain, K. (2014). Decoding and reading comprehension: A meta-analysis to 
identify which reader and assessment characteristics influence the strength of the 
relationship in English. Review of Educational Research, 84, 74-111.  
Garc?a?Madruga, J. A., Elos?a, M. R., Gil, L., G?mez?Veiga, I., Vila, J. ?., Orjales, I., ... & 
Duque, G. (2013). Reading comprehension and working memory's executive 
processes: An intervention study in primary school students. Reading Research 
Quarterly, 48(2), 155-174. 
King, G., & Nielsen, R. (2019). Why propensity scores should not be used for matching. 
Political Analysis, 27(4), 435-454. 
Leow, C., Wen, X., & Kormacher, J. (2015). Two-year versus one-year Head Star program 
impact: Addressing selection bias by comparing regression modeling with propensity 
score analysis. Applied Developmental Science, 19(1), 31-46. 
Levine, D.I., & Painter, G. (2008). Are measured school effects just sorting?: Causality and 
correlation in the National Education Longitudinal Survey. Economics of Education 
Review, 27(4), 460-470. 
Locascio, G., Mahone, E. M., Eason, S. H., & Cutting, L. E. (2010). Executive dysfunction 
among children with reading comprehension deficits. Journal of learning 
disabilities, 43(5), 441-454. https://doi.org/10.1177/0022219409355476 
McGrew, K.S., LaForte, E.M., & Schrank, F.A. (2014). Woodcock Johnson IV Technical 
Manual. Rolling Meadows, IL: Riverside.  
Miller, E.K., & Wallis, J.D. (2009). Executive function and higher-order cognition: definition 
and neural substrates. Encyclopedia of neuroscience, 4, 99-104. 
Monette, S., Bigras, M., & Guay, M.C. (2011). The role of the executive functions in school 
achievement at the end of Grade 1. Journal of Experimental Child Psychology, 
109(2), 158-173. 
Murnane, R.J., & Willett, J.B. (2010). Methods matter: Improving causal inference in 
educational and social science research. Oxford University Press. 
 125 
?NAEP reading report card.? (2018). The Nation?s Report Card. Retrieved from 
https://www.nationsreportcard.gov/reading_2017/nation/achievement?grade=4 
NICHD Early Child Care Research Network & Duncan, G. (2003). Modeling the impacts of 
child care quality on children?s preschool cognitive development. Child Development, 
74(5), 1454-1475. 
Oakhill, J., & Cain, K. (2017). Children with specific text comprehension problems. In K 
Cain, DL Compton, RK Parrila: Theories of Reading Development (pp. 359-378). John 
Benjamins Publishing Company. 
Oakhill, J., Hartt, J., & Samols, D. (2005). Levels of comprehension monitoring and working 
memory in good and poor comprehenders. Reading and Writing, 18, 657-686. 
Peng, P., & Kievit, R.A. (2020). The development of academic achievement and cognitive 
abilities: A bidirectional perspective. Child Development Perspectives, 14(1), 15-20. 
Pimperton, H., & Nation, K. (2010). Suppressing irrelevant information from working 
memory: Evidence for domain-specific deficits in poor comprehenders. Journal of 
Memory and Language, 62, 380-391. 
Puzio, K, Colby, G.T., & Algeo-Nichols, D. (2020). Differentiated literacy instruction: 
Boondoggle or best practice? Review of Educational Research, 90(4), 459-498. 
Raghubar, K.P., Barnes, M.A., & Hecht, S.A. (2010). Working memory and mathematics: A 
review of developmental, individual difference, and cognitive approaches. Learning 
and Individual Differences, 20, 110-122. 
Ricketts, J., Sperring, R., & Nation, K. (2014). Educational attainment in poor 
comprehenders. Frontiers in Psychology, 5, 445. 
Schrank, F.A., Mather, N., & McGrew, K.S. (2014). Woodcock-Johnson IV Tests of 
Achievement. Rolling Meadows, IL: Riverside. 
Shadish, W.R., Clark, M.H., & Steiner, P.M. (2008). Can nonrandomized experiments yield 
accurate answers? A randomized experiment comparing random and nonrandom 
assignments. Journal of the American Statistical Association, 103(484), 1334-1356. 
 126 
Smart, D., Youssef, G. J., Sanson, A., Prior, M., Toumbourou, J. W., & Olsson, C. A. (2017). 
Consequences of childhood reading difficulties and behaviour problems for 
educational achievement and employment in early adulthood. British Journal of 
Educational Psychology, 87(2), 288-308. 
Taboada Barber, A. (PI), Cartwright, K., & Stapleton, L. (Co-PIs) (2016-2020). Cognitive 
and motivational contributors to reading comprehension in English Learners (ELs) 
and English Monolinguals (EMs): Different or similar growth patterns. English 
Learners Research Grants Competition, Institute of Education Sciences. 
Taboada Barber, A., Cartwright, K., Stapleton, L.M., Lutz Klauda, S., Archer, C.J., & Smith, 
P. (2020). Direct and indirect effects of executive functions, reading engagement, 
and higher order strategic processes in the reading comprehension of Dual Language 
Learners and English Monolinguals. Contemporary Educational Psychology, 61. 
Taylor, L., Hiller, M., & Taylor, R.B. (2013). Personal factors and substance abuse treatment 
program retention among felony probationers: Theoretical relevance of initial vs. 
shifting scores on impulsivity/low self-control. Journal of Criminal Justice, 41(3), 
141-150. 
van der Maas, H.L.J., Dolan, C.V., Grasman, R.P.P.P., Wicherts, J.M., Huizenga, H.M., & 
Raijmakers, M.E.J. (2006). A dynamical model of general intelligence: The positive 
manifold of intelligence by mutualism. Psychological Review, 113, 842-861. 
Yuill, N., Oakhill, J., & Parkin, A. (1989). Working memory, comprehension ability and the 
resolution of text anomaly. British Journal of Psychology, 80(3), 351-361. 
Zanutto, E., Lu, B., & Hornik, R. (2005). Using propensity score subclassification for multiple 
treatment doses to evaluate a national antidrug media campaign. Journal of 
Educational and Behavioral Statistics, 30(1), 59-73. 
 
 
 127 
Conclusion 
 This dissertation explores the connections between students? centrality within their 
classroom friendship networks and reading development, as well as the relation between 
executive functioning and reading outcomes. This dissertation is one of the first studies to 
use Social Network Analyses to explicitly connect students? friendship networks to academic 
achievement growth, including for both the general student body and for English Learners. 
To my knowledge, only two other studies have utilized network data to predict student 
learning (Cooc & Kim, 2017; van Rijsewijk et al., 2018). It seems likely that interest in 
children?s social and emotional wellness, as well as their effects on students? academic 
achievement, will continue to grow in light of the recent momentum that social-emotional 
outcomes have had in schools and among policymakers. The research from this dissertation 
can affect policy and future research in a few ways. 
 First, this dissertation provides further evidence that friendships matter in the 
classroom. Students are significantly more likely to improve their oral language 
comprehension during semesters in which they are more central within their classroom 
friendship networks. In essence, students who are more connected with classroom peers 
have more opportunities to develop their language through meaningful interactions, and 
these students are more likely to thrive in the classroom setting when not threatened by 
peer rejection or isolation. Additionally, English Monolinguals are significantly more likely to 
improve their reading comprehension?an important skill that is foundational to English 
Language Arts high-stakes tests?when students are more central within their classroom 
friendship networks. These findings support that educators and researchers should explore 
whether friendships and social connectivity can support student reading development, 
particularly for elementary students. Interventions that foster friendships and dense 
classroom networks (i.e., classrooms in which students are well-connected to each other) 
may support positive reading outcomes for students while simultaneously supporting 
student social-emotional well-being. 
 128 
 Second, social connectivity does not uniformly predict reading outcomes for English 
Learners and English Monolinguals, so peer interventions aimed at increasing achievement 
need to be examined with an equity lens. Even though English Monolinguals were more 
likely to demonstrate reading comprehension growth when they were more connected 
within their classroom friendship networks, there was no relation among centrality and 
reading comprehension for English Learners. Elementary children who were simultaneously 
building their English language skills and learning content in their non-native language, 
while navigating the social landscape, experienced no detectable pattern between their 
centrality and their reading development. Regardless of the mechanisms at play, research 
exploring the potential of social interventions to improve reading outcomes must pay 
particular attention to whether English Learners?a substantial and growing population in US 
public schools?equally benefit. Future research may investigate how to support reading 
teachers of ELs and ensure that they have sufficient training and expertise to connect 
students regardless of students? language status. As more is known about the differential 
impact of interventions for various student groups, educators can continue to close 
achievement gaps and work towards equitable outcomes for all students. 
 While I hypothesized that the relation between friendship centrality and reading 
comprehension would be stronger for English Learners than English Monolinguals, due to 
the importance of friendships on ELs? learning of the English language, these findings 
support the opposite: classroom instruction currently does not foster ELs to capitalize on 
friendships in the same way that it may benefit EMs. This finding may be indicative of a few 
mechanisms, which would need to be explored further: a) friendships may take longer to 
impact ELs? reading outcomes (in which case  the relation between friendships and reading 
comprehension is mediated by oral comprehension growth); b) ELs? who are primarily 
communicating with other ELs are not experiencing the benefits on reading comprehension 
that EMs would receive due to speaking in the same language in which they are reading, or 
c) there are other factors at play which are preventing ELs from gaining social capital due to 
 129 
expanding friendship networks, such as being part of a network of other ELs and not gaining 
access to supports associated with reading achievement growth. Future work should 
compare centrality of ELs as well as controlling for the proportion of one?s friends who are 
EMs, to determine if there is a difference in predictive power for ELs whose networks include 
higher proportions of EMs. 
 In the third paper, I found additional evidence supporting the theory that executive 
functioning contributes to reading comprehension development. Whereas there are 
hypotheses that the two constructs may develop in tandem, I used a method that aimed to 
limit omitted variable bias in order to more validly estimate the effect size of executive 
functioning on reading comprehension. Developing students? reading comprehension is one 
of the primary goals of elementary education, and this dissertation supports that one 
approach may involve supporting students? executive functioning growth. By contributing to 
students? inhibition, memory, and cognitive functioning, students may be more likely to 
develop their propensity for understanding challenging texts. School environments, 
including positive teacher-student and student-student relationships, as well as supportive 
parenting can contribute to children?s executive functioning development (see Cumming et 
al., 2020), so more can be done to explore the ways that educators can leverage 
relationship-building to improve students? EF, as well as their reading comprehension. 
Future interventions that are developed to improve executive functioning vis a vis 
developing positive and supportive classroom relationships should be investigated for their 
predictive ability to improve students? reading comprehension. 
While this dissertation is exploratory in nature, it provides evidence that friendships 
are an overlooked resource which may have benefits that go well beyond preparing children 
for happy and meaningful social lives. Many educators strive to support students? well-
being, helping to develop students? relationships in order to strengthen children?s 
developmental processes and ensure children have support from their peers. However, this 
dissertation contributes to education research by suggesting that friendships may also be a 
 130 
building block to develop academic benefits for children, including supporting their reading 
development. Without evidence bolstering an intervention, I do not claim that friendships 
are the most important resource; decades of research support the importance of teachers, 
high-quality curricula, safe learning environments, and many other inputs that can improve 
children?s learning. Even though much more work is necessary to understand how to 
leverage peers to develop reading outcomes, research that identifies the value in developing 
students? friendships is paramount given that peers are a universal resource that is already 
available to all districts?rich and poor?throughout the country. All recent major federal 
policies (e.g., No Child Left Behind, Race to the Top, Every Student Succeeds Act) have 
utilized reading test scores in some capacity to measure student performance; thus, 
demonstrating that students? friendship networks are connected to achievement is 
important because it justifies the role that relationship building must continue to have in the 
schools. Comparing pedagogy is outside of the scope of this dissertation, but it is clear that 
friendships?which are important for so many aspects of children?s development?are also 
important when considering their connection to student achievement. It is my hope that this 
dissertation may provide another datapoint highlighting the importance of peer relationships 
in our education system as policymakers continue to debate what is the purpose of schools, 
and how might we best prepare our children for the future. 
 
 
 
 
 
 
 
 
 
 131 
References 
Cooc, N., & Kim, J. (2016). Peer influence on children?s reading skills: A social network 
analysis of elementary school classrooms. Journal of Educational Psychology, 
Forthcoming. 
Cumming, M.M., Bettini, E., Pham, A.V., & Park, J. (2020). School-, classroom-, and dyadic-
level experiences: A literature review of their relationship with students? executive 
functioning development. Review of Educational Research, 90(1), 47-94. 
Hamilton, L., & Gross, B. (2021). How has the pandemic affected students? social-emotional 
well-being? A review of the evidence to date. Center on Reinventing Public 
Education. Retrieved from 
https://www.crpe.org/sites/default/files/sel_report_2021_final_8_10.pdf 
van Rijsewijk, L.G.M., Oldenburg, B., Snijders, T.A.B., Dijkstra, J.K., & Veenstra, R. (2018). 
A description of classroom help networks, individual network position, and their 
associations with academic achievement. PLoS ONE, 13(12), e0208173. 
https://doi.org/10.1371/journal.pone.0208173 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 132