Sandilos 2020

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Journal of School Psychology 79 (2020) 1–15

Contents lists available at ScienceDirect

Journal of School Psychology


journal homepage: www.elsevier.com/locate/jschpsyc

English learners' achievement in mathematics and science:


T
Examining the role of self-efficacy

Lia E. Sandilosa, , Alison E. Baroodyb, Sara E. Rimm-Kaufmanc, Eileen G. Merrittd
a
Temple University, United States of America
b
San Francisco State University, United States of America
c
University of Virginia, United States of America
d
Arizona State University, United States of America

A R T IC LE I N F O ABS TRA CT

Action Editor: Milena Keller-Margulis The goals of the current study were twofold. The first goal was to describe levels of mathematics
Keywords: and science self-efficacy and achievement among a sample of students with varying levels of
English learners English language proficiency. The second goal was to examine the extent to which students' self-
Self-efficacy efficacy explains the relation between their English proficiency level and mathematics and sci-
Mathematics ence achievement. The sample consisted of 332 fifth graders (mean age = 10.46 years,
Science SD = 0.38) and their 63 teachers in 20 schools. The student sample was linguistically diverse
Elementary school with parents reporting 22 different home languages. Based on district classification procedures,
each student was coded into one of three English language proficiency level categories: English
proficient-speaking students (English proficient), English Learner (EL) students who are reaching
proficiency, yet are still being monitored (reaching proficiency), and EL students who are re-
ceiving English for Speakers of Other Languages services (ESOL; limited English proficient).
Regression analyses indicated that students identified as limited English proficient consistently
demonstrated lower achievement and self-efficacy across the content areas of mathematics and
science as compared to their peers who were English proficient and reaching proficiency. In
addition, students' self-efficacy partially explained the relation between limited English profi-
ciency level and achievement for science, but not for mathematics. Results indicate that edu-
cators should consider variability in students' English proficiency levels as they select supports to
promote both science achievement and self-efficacy. Findings also suggest promise for practices
and programs that foster self-efficacy in addition to language and content skills.

Mathematics and science knowledge are crucial for the academic and professional success of students in the 21st century
(National Mathematics Advisory Panel, 2008; National Research Council, 2011). Within the U.S. education system, there is an
increasing focus on enhancing students' skill development in mathematics and science during the elementary school years due to the
growing need for adults in the work force who are skilled in these content domains. Concurrently, English Learners (ELs) are the
fastest growing population of students in U.S. classrooms. Over four million public school students in the U.S. speak a language other
than English at home (National Center for Education Statistics, 2017). However, students who are ELs in the U.S., on average, tend to
have lower scores on assessments of mathematics and science achievement compared to their English-speaking peers (Huang, 2000;
Marks, 2005; Polat, Zarecky-Hodge, & Schreiber, 2016). To better address the needs of this growing population of students, recent
federal education law, the Every Student Succeeds Act (ESSA, 2015), emphasized the importance of academic success of ELs across all


Corresponding author.
E-mail address: lia.sandilos@temple.edu (L.E. Sandilos).

https://doi.org/10.1016/j.jsp.2020.02.002
Received 13 June 2018; Received in revised form 20 September 2019; Accepted 14 February 2020
0022-4405/ © 2020 Society for the Study of School Psychology. Published by Elsevier Ltd. All rights reserved.
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

content areas in public school accountability systems (American Federation of Teachers, 2018). The emphasis on ELs' achievement in
ESSA, as well as the importance of supporting EL students to achieve their potential, requires an expansion of our understanding of
the psychological and learning processes that contribute to ELs' academic outcomes. To date, the knowledge base related to language
and literacy skills in non-native English speakers is larger than that related to mathematics and science achievement (Kim, Wang,
Ahn, & Bong, 2015; Maerten-Rivera, Ahn, Lanier, Diaz, & Lee, 2016); this salient issue led to the present study.
Understanding factors that can be leveraged to boost achievement in mathematics and science for EL students is critical for
developing instructional approaches that foster success in school and future careers (Aud et al., 2013; Francis & Stephens, 2018;
Hernandez, Denton, & Macartney, 2008; Kindler, 2002). Students' perceptions of their competence, often referred to as self-efficacy
(Bandura, 1977), may contribute to lower achievement in EL students. Low self-efficacy may emerge given that students face ad-
ditional obstacles as they strive to understand the language of instruction in these content domains. The aim of the current study is to
describe levels of self-efficacy in students who differ in English proficiency and examine pathways by which English proficiency is
associated with mathematics and science achievement. We hypothesize that self-efficacy plays a significant role in understanding
links between students' English proficiency level and their achievement in mathematics and science.

1. Self-efficacy, academic achievement, and second language acquisition

Self-efficacy is a widely studied social cognitive construct that represents an individual's “beliefs about their capabilities to
exercise control over their own level of functioning” (Bandura, 1993, p. 118). Feelings of self-efficacy influence an individual's effort
and persistence on tasks, as well as their coping behaviors, such as avoidance of tasks (Bandura, 1997). Students' self-efficacy beliefs
about their ability to master challenging academic content can influence their motivation to learn, their degree of interest and
achievement in that content domain, and their career aspirations (Bandura, Barbaranelli, Caprara, & Pastorelli, 1996; Kim et al.,
2015; Martin & Rimm-Kaufman, 2015; Schunk & Pajares, 2002; Skaalvik, Federici, & Klassen, 2015; Wang, Harrison, Cardullo, & Lin,
2018). Self-efficacy is related to student achievement in various academic content areas, including mathematics and science
(Areepattamannil, Freeman, & Klinger, 2011; Borman & Overman, 2004; Britner & Pajares, 2001, 2006; Lau & Roeser, 2002; Mercer,
Nellis, Martínez, & Kirk, 2011; Pajares & Graham, 1999).
In a seminal paper, Bandura et al. (1996) described self-efficacy and children's academic achievement and presented a causal
model in which background and familial factors, such as socio-economic status and expectations of caregiving adults, contribute to
children's academic self-efficacy, which in turn predicts their academic achievement. Indeed, research supports this model by de-
monstrating that academic self-efficacy plays a mediating role (i.e., helps to explain the relation) between key variables such as
student background characteristics, academic learning behaviors, or prior achievement and academic outcomes (Pajares, 1996;
Pajares & Kranzler, 1995; Skaalvik et al., 2015). As one example, feelings of self-efficacy partially explained the relation between
background variables (socio-economic status, prior achievement) and students' performance in a middle school mathematics class
(Cleary & Kitsantas, 2017).
Despite the preponderance of scientific literature on self-efficacy of adolescents who are majority-language speakers in the U.S.,
few studies have examined self-efficacy with EL students, and research is particularly sparse in mathematics and science. Key ele-
ments of the conceptual model proposed by Bandura et al. (1996) are especially salient to the schooling experiences of English
learners. Proficiency in a language is not only essential to understanding the language of instruction, but it is also closely tied to an
individual's sense of identity. Although being bilingual can be viewed as a strength, this belief is not universally held by educators and
school systems (Flores, 2016; Nieto, 2010). As a result, ELs may perceive negative attitudes or lower academic expectations on the
part of educators or peers as a result of their language learning differences (Karabenick & Noda, 2004; Lewis et al., 2012). These
experiences can shape their overall academic identity and lower their feelings of academic self-efficacy, which ultimately contributes
to lower academic learning. In contrast, if ELs perceive themselves positively and have a strong sense of their own efficacy, this self-
perception may buoy them toward academic success despite the challenges present due to lower English proficiency and possible low
expectations for academic performance by peers and teachers. Students' own sense of efficacy drives their approach to academic work
and their interpretation of successes and failures. Understanding how this plays out among ELs could inform efforts to bolster their
academic success.
Some research exploring links between ELs' academic self-perceptions and attitudes and the development of their language skills
has indicated that feelings of efficacy can positively influence ELs' achievement across content domains (Neugebauer & Howard,
2015). Lewis et al. (2012) established that feelings of efficacy fully explained the relation between teacher caring and mathematics
scores among students facing the greatest language barriers. Further, studies examining the self-efficacy of EL students have iden-
tified an association between higher levels of self-efficacy and greater use of language learning strategies (Kim et al., 2015; Magogwe
& Oliver, 2007). More specifically, EL students who believe they are capable of learning in a specific content area may be more likely
to employ cognitive and metacognitive strategies that enable them to process, store, and retain academic language (Kim et al., 2015).
It can take EL students many years of schooling before they become fully proficient in academic English (Hakuta, Butler, & Witt,
2000); understanding the psychological experiences that accompany this process may identify ways to best facilitate academic
learning.

2. Importance of examining ELs' mathematics and science learning

EL students face unique learning demands in mathematics and science classes. They must learn new skills and concepts and try to
explain their thinking in a language in which they are not yet fully proficient (Lee, Quinn, & Valdés, 2013). Across both content areas,

2
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

ELs of varying English proficiency levels may have less experience with discourse related to academics (also known as Cognitive
Academic Language Proficiency [CALP]; Cummins, 1979), than with basic interpersonal communication skills (referred to as BICS;
Cummins, 1979). For example, there are a variety of specialized words and word meanings related to both mathematics and science
content (e.g., ‘equation’, ‘hypothesis’, etc.) that are unlikely to be acquired through basic social communication (Ron, 1999).
Ernst-Slavit and Slavit (2013) described mathematics as a language with specific vocabulary, syntax, and symbols that must be
mastered in order for students to engage with concepts accurately and efficiently. Although the number system often is regarded as a
“universal language,” ELs still need to become familiar with irregularities in English number names and may need to learn new forms
of measurement to engage in basic mathematical procedures (Ernst-Slavit & Slavit, 2013; Janzen, 2008). Additionally, the academic
language used by mathematics instructors to teach the content, as well as the language infused in mathematical text and word
problems, may not be familiar or comprehensible to students with lower levels of English proficiency (Schleppegrell, 2007). For
example, there are a variety of homonyms in mathematics, such as table (furniture vs. chart), mean (unkind vs. average), and round
(circular shape vs. estimate). Teachers' use of seemingly basic mathematical terminology without sufficient explanation of vocabulary
could impair ELs' comprehension of instruction and word problems (National Council of Teachers of Mathematics, 2011).
Similar concerns exist as ELs learn science (Janzen, 2008), with the potential for even greater language demands during scientific
discourse. For example, in late-elementary science instruction, teachers engage students in discourse specific to the scientific method,
such as generating hypotheses and describing scientific phenomena or experimentation (Townsend, 2015). Common Core Standards
and Next Generation Science Standards (NGSS) encourage this use of scientific discourse to promote conceptual understanding in
science classrooms (National Governors Association Center for Best Practices & Council of Chief State School Officers, 2010; Next
Generation Science Standards, 2012). The implementation of new standards is likely to provide opportunities that benefit students'
learning and conceptual understanding, while at the same time, pose additional challenges to students who are learning English (Lee
et al., 2013). These challenges stem, in part, from the fact that elementary and secondary teachers tend not to have sufficient training
in effective pedagogical strategies that integrate science content learning with language learning for linguistically diverse students
(Fradd, Lee, Sutman, & Saxton, 2001; Hart & Lee, 2003; Swars & Chestnutt, 2016).
If EL students lack proper instructional supports in mathematics and science, they may have difficulty understanding the language
of instruction and consequently struggle to learn the content, which in turn may lead to feelings of low competence or low self-
efficacy. One particularly understudied area is whether or not these challenges are present for EL students of differing proficiency
levels.

3. Defining English proficiency

EL students can vary widely in their level of English proficiency. It is estimated that students develop CALP within approximately
4–7 years and the numbers of years may increase depending on the age at which they are first exposed to English (Cummins, 1981;
Hakuta et al., 2000). As such, it is possible that ELs who are no longer eligible for formal language support services may still show
lower levels of proficiency than their monolingual peers at any point during elementary or secondary education (Slama, 2012).
Measuring language proficiency is complex, perhaps best exemplified by the fact that criteria for defining English proficiency and
allocating services differ between states and between districts within a single state (Francis & Stephens, 2018; Maxwell, 2013).
Although existing categorization methods fail to reflect the true nuances of language proficiency, one way that researchers often
classify students as EL is based on school services and records (Lewis et al., 2012). For example, students who are receiving English
for Speakers of Other Languages (ESOL) services are defined as EL, whereas students who are no longer receiving ESOL services or
have never received these services are defined as English proficient (Santau, Maerten-Rivera, & Corinne Huggins, 2011).
Research suggests that dichotomous classification (i.e., EL or non-EL) does not provide enough specificity when describing English
proficiency levels of English learners (Cook, Boals, & Lundberg, 2011; Lewis et al., 2012). Students who are grouped together as EL
demonstrate a great deal of heterogeneity because varying background experiences contribute to the time it takes to gain mastery of
academic English. Consequently, students labeled as ‘EL’ may require very different levels of service and support in classroom settings
(Cook et al., 2011; Hakuta et al., 2000; Slama, 2012) and may have vastly different feelings of competence. For instance, a student
identified as EL who has just moved to the United States with minimal formal schooling in English differs considerably from a student
who has had English instruction for several years, and both types of EL students will differ from their monolingual peers. In the
current study, we endeavored to take a small step toward expanding the school-based definition of ‘EL’ by exploring a “reaching
proficiency” classification with the current sample, whereby students who are no longer receiving ESOL services, but are still being
monitored, will be explored as a separate language proficiency level. This is an important distinction given that students who are
exited from ESOL services may be viewed as English proficient despite the fact that they are likely still developing academic language
proficiency.

4. Developmental significance

In the current study, we focus specifically on fifth grade students with a range of language proficiency skills. Fifth grade marks an
important period in late childhood because children are at the cusp of early adolescence. Students, at this age, are experiencing rapid
changes related to their development of personal identity and academic self-concept (e.g., view of oneself and how it relates to
academic achievement; Bong & Skaalvik, 2003; Eccles, 1999; McKown & Weinstein, 2003). At this stage of development, classroom
learning experiences, such as academic successes or failures, shape students' self-perceptions and feelings of efficacy in particular
content domains (Schunk & Meece, 2006). Students are reading to learn and academic content is increasingly complex, thus putting

3
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

more demands on language proficiency skill (Tong, Irby, Lara-Alecio, & Koch, 2014). During this period, students' feelings of self-
efficacy become more differentiated (e.g., specific to content areas) and students tend to experience lower self-efficacy as they
advance through the grade levels (Eccles, Wigfield, & Schiefele, 1998; Harter, 1998; Schunk & Pajares, 2002). This issue is concerning
given that EL students tend to report lower self-efficacy than their monolingual peers (Wang et al., 2018). The field of education
needs to cultivate a better understanding of the self-efficacy of ELs at varying levels of proficiency during this sensitive phase of
development before low self-efficacy becomes fully entrenched and interferes with students' use of effective language learning
strategies, effort and perseverance in assignments, selection of course electives, and choice of career paths.

5. Current study

The goals of this correlational study are twofold. First, we describe levels of mathematics and science self-efficacy and consider
differences across three groups of students: students who are English-proficient speakers (English proficient) and never received ESOL
services, EL students who are no longer receiving ESOL services, yet still are monitored (reaching proficiency), and students who are
currently ELs receiving ESOL services (limited proficiency). Next, we describe the pathway by which students' English proficiency level
is associated with achievement in mathematics and science via mathematics and science self-efficacy. We address two questions:

1) To what extent do fifth grade students differ on their perceptions of mathematics and science self-efficacy and their achievement
scores on mathematics and science assessments based on varying English proficiency levels?
2) Can the relation between English proficiency level and mathematics and science achievement scores be explained, in part, by
students' mathematics and science self-efficacy?

Our research questions contribute to a limited body of research on ELs and self-efficacy. The present study is unique in that it
explores differences across three categorizations of language proficiency in relation to a psychological aspect of learning (i.e., self-
efficacy) and achievement in mathematics and science. We draw upon existing literature (e.g., Kim et al., 2015; Magogwe & Oliver,
2007) to hypothesize that EL students with lower English proficiency levels will have lower mathematics and science self-efficacy and
achievement compared to students with higher English proficiency levels. Furthermore, we hypothesize that the association between
students' English proficiency levels and achievement scores in mathematics and science will be partially explained by mathematics
and science self-efficacy.
Additionally, to better understand the distinct contributions of language proficiency and self-efficacy to academic achievement,
we consider the influence of student characteristics in each set of analyses. Student characteristics such as gender, family income, and
cognitive skill (e.g., working memory) have been shown to be important contributors to self-efficacy and achievement. Although not
the main focus of the present study, we include these factors as covariates in analyses. Gender has been studied extensively in the
research related to self-efficacy, as well as mathematics and science achievement (Britner & Pajares, 2006; Joët, Usher, & Bressoux,
2011; Kiran & Sungur, 2012; Robinson & Lubienski, 2011; Wilkins, 2004). Family income level is an important consideration because
EL students are more likely to be economically disadvantaged than native English-speaking students (Garcia & Cuellar, 2006; Huang,
2000). We use eligibility for free or reduced-priced lunch (FRPL) as a proxy for income. With regard to cognitive skills, working
memory has been linked to achievement in mathematics and science (Engle, 2002; LeFevre, DeStefano, Coleman, & Shanahan, 2005).
Prior research also indicates that working memory plays a central role in the reading comprehension of adolescent EL students
(Lesaux & Kieffer, 2010). As such, working memory is included in the current study as the best available proxy for prior achievement
in mathematics and science.

6. Method

6.1. Participants

The present sample of teachers and students were participants in a larger social-emotional intervention study that took place
within one school district in a Mid-Atlantic state (Rimm-Kaufman, Baroody, Larsen, Curby, & Abry, 2015). A total of 63 fifth-grade
teachers (White/Caucasian = 76%; mean years of teaching experience = 12.49, SD = 8.83, min = 1, max = 38 years) consented to
participate in the study. Across consented classrooms, 387 students were selected to participate in the larger social-emotional in-
tervention. Due to incomplete records on English proficiency for 55 students, data analyses for the present study were estimated for
332 students who had complete language data (Girls = 55%; mean age = 10.46 years, SD = 0.38, min = 8.24, max = 11.77).
Among the student participants, 63% were classified as English proficient, 21% as reaching proficiency, and 16% as limited English
proficient. The sample was linguistically diverse; 37% of parents reported home languages other than English. Of those, 69% spe-
cified the language spoken, which totaled to 22 different languages. A large percentage of parents (41%) reported that Spanish was
the primary language spoken at home. Other languages included: Vietnamese (11%), Korean (6%), Chinese (6%), Amharic (5%), Twi
(3%), and other (28%; including mainly European, Middle Eastern, and Southeast Asian languages). Parents reported that students
were 43% White, 19% Asian American, 15% Latino, 12% African American, and 11% multiracial/other. Across the sample, 21% of
students (n = 71) qualified for free or reduced-price lunch. Of mothers who reported their highest level of educational attainment
(n = 282), 61% had a four-year college degree or higher. Descriptive statistics for student characteristics and key variables of interest
are reported in Table 1.

4
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

Table 1
Descriptive statistics for student characteristics and key variables of interest.
Characteristic n Percent Mean SD Min-Max

Qualified for FRPL 71 21% – – –


Gender (female) 181 55% – – –
African American 41 12% – – –
Asian American 64 19% – – –
Latino 51 15% – – –
Multiracial/Other race 27 11% – – –
White 143 43%
English proficient 208 63% – – –
Reaching proficiency 69 21% – – –
Limited proficiency 55 16% – – –
Working memory 323 – 23.04 5.47 2.00–37.00
Mathematics self-efficacy 328 – 3.30 0.58 1.40–4.00
Science self-efficacy 328 – 3.31 0.59 1.40–4.00
Mathematics SOL 326 – 543.81 61.16 351.00–600.00
Science SOL 328 – 469.88 61.14 342.00–600.00

6.2. Measures

6.2.1. Mathematics and science self-efficacy


Students completed five items on mathematics self-efficacy (e.g., “I know I can learn the skills taught in math this year”) and five
items on science self-efficacy (e.g., “I'm certain that I can figure out how to do even the most difficult science work”). Items were
adapted from the Academic Efficacy subscale of the Patterns of Adaptive Learning Scales (PALS; Midgley et al., 2000) to be specific to
mathematics or science (i.e., the word “class” was replaced with either “math” or “science”). The adapted measures were pilot tested
and validated in a sample of 39 children representing similar linguistic diversity to the larger sample (math pilot α = 0.89; science
pilot α = 0.91). On each measure, students rated each item on a 4-point scale (1 = almost never; 2 = sometimes; 3 = most of the time;
4 = all the time). Items were averaged to create a mathematics self-efficacy score and a science self-efficacy score. Internal consistency
estimates were computed for each student proficiency group: English proficient (math α = 0.82; science α = 0.84), reaching
proficiency (math α = 0.80; science α = 0.81), and limited English proficient (math α = 0.83; science α = 0.77).

6.2.2. English proficiency level


The World-Class Instructional Design and Assessment (WIDA) ACCESS Test® (WIDA, 2011) is a standards-based, criterion re-
ference test that evaluates students' social and academic language (oral and written) through four key domains: listening, reading,
speaking, and writing (Fox & Fairbairn, 2011). The test is developed in grade-level clusters for kindergarten through 12th grade that
assess the WIDA English Language Proficiency Standards (WIDA Consortium, 2007) at each particular grade level. In the present study,
the WIDA ACCESS test was the federally approved assessment for the state in which the data were collected.
WIDA ACCESS demonstrates high overall classification accuracy (i.e., proportion of students whose proficiency level is correctly
identified) for Grade 5 (0.71; Center for Applied Linguistics, 2017). Additionally, differential item functioning (DIF) analyses indicate
that test items do not show bias against certain subgroups of ELs (Center for Applied Linguistics, 2017).
At the beginning of the academic year, the school district used the WIDA ACCESS for ELLs test to determine students' English
proficiency levels. District classification procedures were as follows: students received ESOL services for the current academic year if
they scored at or below a level 4 on the ACCESS for ELLs test (i.e., levels 1-limited, 2-beginning, 3-developing, or 4-expanding); students
from families who spoke a language other than English at home, but scored at levels 5 (bridging) and 6 (reaching) on the WIDA were
monitored but did not receive ESOL services that year.
Based on the WIDA-level classification conducted by the district, we coded students in three categories to describe their English
language proficiency. Students receiving ESOL services (at or below a level 4 on the WIDA) were considered “limited English pro-
ficient.” Students whose families spoke a language other than English at home, but were presently being monitored (that is, levels 5
and 6 on the WIDA) were coded as “reaching proficiency.” Students who had never received ESOL services were considered “English
proficient” speakers. The category of “English proficient” included some students (10%; 21 out of 208) whose families reported
speaking a language other than English at home but never qualified for ESOL services or monitoring.

6.2.3. Student achievement


Students' mathematics and science achievement was measured using scores from the fifth-grade Standards of Learning (SOL)
assessment, a state standardized test administered in May (Virginia Department of Education [VDOE], 2008). Scale scores were
continuous ranging from 200 to 600; a score of 400 or above indicated proficiency on the assessment. For both the mathematics and
science assessments, the Virginia Department of Education reported that items were developed by content experts, Virginia educa-
tors, and psychometricians and were calibrated using Rasch modeling techniques. Concurrent validity was established by examining
correlations with other nationally recognized assessments of similar content (VDOE, 2008).
The 5th grade mathematics SOL consisted of 50 multiple choice items (Grade 5, α = 0.88) assessing students' procedural

5
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

knowledge and conceptual understanding of computation and estimation (e.g., multistep computation, order of operations), mea-
surement (e.g., perimeter, volume), geometry (e.g., classifying angles and figures), and probability and statistics (e.g., mean, median,
mode). A ‘Plain English’ version of the Mathematics SOL was administered if students were not proficient in English. Six students
classified as “limited English proficient” in the current sample took the plain English version. This version was designed to assess the
same sets of mathematical skills and was scored in the same manner as the regular mathematics SOL, but the Plain English version of
the Mathematics SOL (Grade 5, α = 0.90) used simplified language (VDOE, 2009). Adaptations include splitting longer sentences
into shorter sentences, removing descriptive language deemed unnecessary for problem solving, avoiding the use of idioms and
colloquial expressions, or using simple present or past tense verbs (for further description of the Plain English Mathematics SOL,
review VDOE, 2009).
The 5th grade science SOL included 50 multiple-choice items (Grade 5, α = 0.86) requiring students to demonstrate under-
standing of a range of topics, including the scientific investigation process (e.g., making predictions using patterns, manipulating
variables); force, motion, and energy (e.g., transmission of sound and light); matter (e.g., solids, liquids, gasses); living systems (e.g.,
cell structures, kingdoms of living things); Earth/space systems (e.g., ecosystems, geological characteristics); and Earth patterns,
cycles, and change (e.g., plate tectonics, fossil evidence). A Plain English version of this assessment was not available for students.

6.2.4. Parent questionnaire


Parents of participating students completed a questionnaire at the start of the study. Items asked about parents' highest educa-
tional attainment, language spoken at home, child's age, and child's gender.

6.2.5. Working memory


Research assistants assessed students' working memory using the Auditory Working Memory subtest of the Woodcock Johnson III
(WJ-III) Test of Cognitive Abilities (Woodcock, McGrew, & Mather, 2001). The assessment was administered using the standardized
test procedures. The subtest consists of 21 items administered via audio recording in English. For each item, students listen to a set of
words containing both objects and numbers. Students then repeat the words in the order they heard them, first listing the objects and
then listing the numbers. As the test proceeds, sets of objects and numbers become increasingly difficult. Students receive one point
for correctly listing the objects and one point for correctly listing the numbers. Scores per item range from 0 to 2 with possible total
scores ranging from 0 to 42. The assessment is administered until students reach the ceiling (i.e., scored 0 on three consecutive sets of
words) or complete the subtest.
The WJ-III was normed on a nationally representative sample of participants that were geographically and culturally/linguisti-
cally diverse (Mather & Woodcock, 2001). Students were included in the standardization pool if they had at least one year of
experience in English-speaking classes, indicating that there was a range of English proficiency in the normative sample (McGrew &
Woodcock, 2001). High internal consistency was reported for students ages 8 to 11 years old (α = 0.86–0.90; McGrew & Woodcock,
2001). Additionally, in the present study, working memory demonstrated a moderate correlation with language proficiency
(r = 0.41) indicating that they were related but not overlapping constructs.

6.3. Procedures

Recruitment for the larger intervention study took place within one school district in a Mid-Atlantic state (see Rimm-Kaufman
et al., 2015 for more information about the larger study). Initially, schools were recruited through invitation letters and in-person
meetings between the research team and school principals and fifth grade teachers. Twenty schools consented to participate in the
study (83% response rate). Within those schools, research assistants brought packets to the schools for distribution to fifth grade
teachers with invitations to participate in the present study and an offer of $100 for their efforts. Once teachers consented to
participate (79% response rate), families who had students in classrooms of participating teachers were recruited through mailings.
Standard district practices were used for family communication and materials were translated into the seven most commonly spoken
languages. Parents of 479 students consented to participate in the study. All parents/guardians who consented to their fifth graders'
participation received a gift certificate of $20. However, given that the teacher-report measures for the larger study were extensive,
the research team chose to randomly sample approximately five consented students per classroom in order to reduce the burden on
classroom teachers. Students selected to participate in the assessments received a small gift (e.g., pencil).
In the fall, parents completed consent forms and a demographic questionnaire which had been translated into the seven standard
languages most prevalent in the district. Students completed questionnaires on self-efficacy in the fall of the school year. The regular
classroom teacher, or math teacher in schools with departmentalized instruction, administered all self-efficacy measures to the whole
class in the beginning of one math class using paper surveys and following a standard protocol. The teacher distributed ques-
tionnaires, read the instructions aloud, and completed two examples (not related to self-efficacy) aloud with the students. No ad-
ditional adaptations were included in the standard protocol for ELs taking the self-efficacy questionnaires. However, if teachers
needed to further modify their directions or read aloud items to assist ELs with comprehension, they were not prevented from doing
so during survey administration. Researchers procured students' FRPL eligibility, English proficiency level, and 5th grade mathe-
matics and science achievement scores on state tests from district databases.

6.4. Plan of analysis

Descriptive statistics and missing data patterns were analyzed using SPSS version 25. As noted previously, although the full

6
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

sample in the larger study included 387 students, 55 students were missing language proficiency data and achievement data. Of the
332 students with complete language proficiency data, an additional four students were missing achievement data and six students
were missing self-efficacy data. We examined missing data patterns in the full sample of 387 students. There were no statistically
significant differences across covariates (i.e., working memory, race/ethnicity, gender, FRPL) and key variables of interest (i.e.,
achievement outcomes, self-efficacy) for students with missing language proficiency data and those with complete data. There were
also no statistically significant differences on demographic variables between students with and without achievement data and self-
efficacy data. Logistic regressions were estimated with a dichotomous indicator for missing data (1 = yes, 0 = no) for each key
variable of interest as the outcome and the observed scores on other key variables of interest as predictors. For example, self-efficacy
and language proficiency data were entered as predictors of a dichotomous outcome variable representing missing data on SOL scores
(1 = yes, 0 = no). No significant associations between the observed variables of interest and missing data on variables of interest
were identified. Finally, we examined relations between indicators of missing data (1 = yes, 0 = no) on variables of interest using
chi-square tests and found no significant associations. Thus, all models were estimated on the sample of 332 students with complete
language proficiency data using Full Information Maximum Likelihood (FIML) to account for the additional missing self-efficacy and
achievement (outcome) data (Enders & Bandalos, 2001).
Based on an examination of correlations and extensive prior research (e.g., Britner & Pajares, 2006; Garcia & Cuellar, 2006;
Huang, 2000; Joët et al., 2011; Kiran & Sungur, 2012; Robinson & Lubienski, 2011; Wilkins, 2004), gender, FRPL status, and race
were included as covariates in all analyses. Variables were coded in the following manner: Gender (1 = girl, 0 = boy); race (White
[reference group], African American, Asian American, Latino, Multi-racial/Other); and FRPL status (1 = eligible, 0 = not eligible).
Although the study was part of a larger social-emotional learning intervention, treatment condition was dropped from the analyses
because it was not correlated with any of the primary variables of interest.
The state test scores (SOL) in mathematics and science were estimated in each model as a scaled continuous outcome variable.
Although controlling for prior mathematics and science achievement in their respective models is preferred, there are two reasons
why we were not able to control for prior achievement, which is a limitation of the present study. First, students in this district did not
take the science SOL in fourth grade. Therefore, there was no way to garner this information on the sample. Although prior year
mathematics SOL scores were available, these had limited utility. Specifically, due to the complexity of the multi-categorical med-
iation, the mathematics model would not converge when prior year mathematics achievement was included. Thus, in order to keep
the models uniform, students' cognitive skill, as measured by their score on the WJ-III auditory working memory subtest, was
included in analyses as a covariate proximal to academic achievement for both mathematics and science.
To address the first question, we conducted multiple regression analyses to examine whether each English language proficiency
subgroup predicted mathematics and science self-efficacy and achievement, controlling for the aforementioned student covariates.
The reference category for language proficiency (i.e., English proficient, reaching proficiency, limited proficiency) was rotated so that
tests comparing all three proficiency categories could be obtained. With regard to statistical assumptions for regression, an ex-
amination of skewness and kurtosis values indicated that SOL and self-efficacy scores were approximately normal for each language
proficiency group. Correlations indicated that there was no multicollinearity between variables of interest. A visual inspection of the
distributions of standardized residuals for each model indicated the presence of homoscedasticity. Levene's Tests revealed homo-
geneity of variances across groups for SOL scores and self-efficacy.
To address question two, we conducted path modeling in the form of multi-categorical mediation analyses1 (Hayes & Preacher,
2014) to examine the pathways among students' English proficiency level, self-efficacy, and achievement. We initially estimated path
analyses with two different analytic models, one with English proficiency level treated as a categorical variable (i.e., three proficiency
categories) and another with proficiency treated as a continuous variable. Both sets of analyses yielded similar findings with regard to
mathematics and science achievement. We chose to present the multi-categorical mediation analyses in the paper because those
findings shed light on differences across the categorical groupings used to make placement decisions within districts.
First, we examined direct effects for mathematics and science achievement models. Next, we entered self-efficacy into the models
to examine the indirect effect among students' English proficiency level (i.e., limited proficiency, reaching proficiency, English
proficient [reference group]), self-efficacy, and achievement. We included the direct effect of covariates (i.e., gender, race, FRPL
status) on the dependent variables (i.e., mathematics and science SOLs) as well as on the mathematics and science self-efficacy
predictor variables.
Intraclass correlations (ICCs) were examined to determine the need to account for nested structures present in the data. All
classroom level ICCs for self-efficacy were small (< 0.07), but classroom-level ICCs for science and mathematics SOL were elevated
(0.37 and 0.36 respectively), indicating the need to account for clustering at the classroom level. At the school level, all ICCs for
mathematics/science self-efficacy and mathematics SOL scores were low (0.03–0.07), with the exception of science SOL (0.25).
However, given that there were only 20 clusters at the school level and the design effects, which are a function of the size of the
intraclass correlations and average cluster size, were < 2.00 (Muthen & Satorra, 1995), we determined that it was most appropriate
to account for nesting at the classroom level. Thus, all analyses accounted for students nested in classrooms using ‘type = complex’
command, which adjusts standard errors (Muthén & Muthén, 1998-2010). If standard errors were not adjusted in the presence of
nested data, they would be too small and the significance values associated with these standard errors would be too lenient. This
approach was selected instead of traditional hierarchical linear modeling (HLM) given that the number of classroom clusters was

1
The use of self-efficacy as a mediator aligned with our theoretical rationale (Pajares, 1996; Pajares & Kranzler, 1995; Skaalvik et al., 2015). To
confirm this theoretical model, we explored the presence of moderation, in addition to mediation, and did not find significant moderation.

7
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

Table 2
Multiple regression analyses predicting math and science self-efficacy and SOL scores from language proficiency and student demographic char-
acteristics.
Math self-efficacy Science self-efficacy Math SOL Science SOL

β β β β

Qualified for FRPL −0.03 0.03 0.01 −0.20⁎⁎⁎


Gender (female) −0.10 −0.06 −0.16⁎⁎⁎ −0.20⁎⁎⁎
African American 0.06 −0.07 −0.04 −0.15⁎⁎
Asian American 0.11 −0.01 0.13⁎ −0.11
Latino 0.07 −0.08 −0.03 −0.15⁎
Multiracial/Other race 0.01 −0.02 −0.00 −0.10
Working memory 0.16⁎⁎ 0.01 0.30⁎⁎⁎ 0.24⁎⁎⁎
Reaching proficiency −0.02 −0.13 0.10 −0.05
(Ref. English Prof.)
Limited proficiency −0.23⁎ −0.28⁎⁎⁎ −0.16⁎ −0.21⁎⁎⁎
(Ref. English Prof.)
Limited proficiency −0.13 −0.16⁎ −0.25⁎⁎⁎ −0.16⁎⁎
(Ref. Reaching Prof.)
R2 0.06⁎ 0.09⁎⁎ 0.17⁎⁎⁎ −0.27⁎⁎⁎

Note. Ref. = reference group. White/Caucasian used as reference group for race variables.

fairly small (i.e., 63 classrooms), and it is also less sensitive than traditional HLM to small within-cluster sample sizes (McNeish,
Stapleton, & Silverman, 2017).

7. Results

On average, students in the sample reported feeling efficacious in mathematics and science “most of the time” with fairly high
scores across mathematics self-efficacy (M = 3.30, min = 1.40, max = 4.00, SD = 0.58) and science self-efficacy (M = 3.31,
min = 1.40, max = 4.00, SD = 0.59). Students scored in the proficient range (≥400), on average, on mathematics and science SOLs
(Table 1). Analysis of the first research question revealed some group differences by English proficiency level for self-efficacy and
achievement.
All multiple regression models controlled for student characteristics (demographics and working memory). Findings (Table 2)
indicated that students categorized as English proficient and reaching proficiency did not differ significantly on mathematics self-
efficacy (β = −0.02, p = .81). Students with limited proficiency had significantly lower mathematics self-efficacy compared to
English proficient students (β = −0.23, p < .05), and marginally lower mathematics self-efficacy compared to reaching proficient
EL students (β = −0.13, p = .10). For science self-efficacy, analyses indicated that students categorized as English proficient and
reaching proficiency showed marginal differences (β = −0.13, p = .10) whereas students with limited English proficiency showed
significantly lower science self-efficacy scores than reaching proficient students (β = −0.16, p < .05) and English proficient
students (β = −0.28, p < .001).
For mathematics achievement (Table 2), English proficient students and students who were reaching proficiency showed marginal
differences in mathematics achievement scores (β = 0.10, p = .10). In contrast, limited English proficient students scored sig-
nificantly lower than students who were reaching proficiency (β = −0.25, p < .001) and students who were English proficient
(β = −0.16, p < .05). Similarly, for science achievement, English proficient students and students reaching proficiency did not
differ on the science SOL (β = −0.05, p = .41). Again, limited English proficient students scored significantly lower on the science
SOL than students who were reaching proficiency (β = −0.16, p < .01) and students who were English proficient (β = −0.21,
p < .001).
Analyses examining the second question, the direct and indirect effects of proficiency on achievement through self-efficacy, found
different results for mathematics and science (Tables 3 & 4; Figs. 1 & 2). All models controlled for demographic variables and working
memory. The direct models revealed that neither reaching proficiency (β = 0.11, p > .05) nor limited proficiency were (β = −0.10,
p > .05) were significant predictors of mathematics achievement, as compared to the English proficient group, when self-efficacy
was included in the model along with the student background variables. Direct models for science achievement showed that reaching
proficiency was not a significant predictor of achievement (β = −0.03, p > .05), whereas limited proficiency was significantly and
negatively related to science achievement (β = −0.15, p < .01) as compared to the English proficient group, once student
background variables and self-efficacy were included in the model. Next, indirect models tested whether the association between
proficiency level and achievement was explained, in part, by students' self-efficacy. The indirect model estimated for mathematics
(Table 3, Fig. 1) indicated that there was no indirect effect of reaching proficiency (β = −0.004, p > .05, CI [−0.03, 0.02]) or
limited proficiency (β = 0.03, p > .05, CI [−0.05, 0.00]) on mathematics achievement through mathematics self-efficacy, which
was expected given the lack of a direct effect of proficiency on mathematics achievement.
Similar to mathematics, the indirect effect of reaching proficiency on science achievement (Table 4, Fig. 2) via self-efficacy was
not significant (β = −0.02, p > .05, CI [−0.05, 0.01]). In contrast, there was a statistically significant indirect effect of limited
proficiency status on science achievement through science self-efficacy (β = −0.04, p < .05, CI [−0.08 to −0.01]; Fig. 2).

8
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

Table 3
Indirect and direct effects models examining math self-efficacy, language proficiency level, and math
achievement.
β 95% CI

2 ⁎⁎⁎
Direct effects on math achievement (R = 0.22 )
Working memory 0.27⁎⁎⁎ [0.16, 0.37]
FRPL status 0.00 [−0.09, 0.10]
Gender (female) −0.13⁎⁎ [−0.21, −0.06]
African American −0.06 [−0.17, 0.06]
Asian American 0.11⁎ [0.03, 0.20]
Latino −0.03 [−0.13, 0.07]
Multiracial/Other race −0.01 [−0.08, 0.06]
Math self-efficacy 0.22⁎⁎⁎ [0.13, 0.31]
Reaching proficiency 0.11 [0.01, 0.22]
Limited proficiency −0.10 [−0.21, 0.02]

Direct effects on math self-efficacy (R2 = 0.05⁎)


Working memory 0.16⁎⁎ [0.06, 0.27]
FRPL status −0.03 [−0.13, 0.07]
Gender (female) −0.09 [−0.17, 0.00]
African American 0.06 [−0.03, 0.16]
Asian American 0.10 [0.00, 0.19]
Latino 0.01 [−0.01, 0.16]
Multiracial/Other race 0.08 [−0.06, 0.08]
Reaching proficiency −0.02 [−0.13, 0.10]
Limited proficiency −0.13 [−0.24, −0.02]

Indirect effects on math achievement


Reaching proficiency −0.004 [−0.03, 0.02]
Limited proficiency −0.030 [−0.05, 0.00]

Note. β = Standardized betas. 95% CI = 95% confidence interval.


R2 value include all predictors and covariates.

p < .05.
⁎⁎
p < .01.
⁎⁎⁎
p < .001.

Specifically, relative to English proficient speakers, those who were limited English proficient had a 0.04 standard deviation decrease
in science achievement, which could be partially explained by the negative influence of limited English proficiency on science self-
efficacy. In other words, this finding indicated that limited proficiency level modestly influenced science performance via self-efficacy
for science.

8. Discussion

The first aim of this study was to examine differences in mathematics and science achievement and self-efficacy across three
groups of students with varying levels of English proficiency. As expected, students identified as limited English proficient con-
sistently demonstrated lower achievement and self-efficacy across content areas as compared to their English proficient peers. The
differences in academic achievement are consistent with mathematics achievement trends between EL and non-EL elementary school
age students reported from the nationally representative U.S. samples (Polat et al., 2016). The current findings contribute to this
knowledge base by examining differences between students across three classifications of proficiency (Cook et al., 2011; Lewis et al.,
2012) and exploring science achievement as an outcome (Maerten-Rivera et al., 2016).
Notably, in the current sample, self-efficacy ratings were relatively high across different language proficiency groups with stu-
dents, on average, indicating that they felt efficacious “some of the time” or “most of the time” in mathematics and science. This
finding is positive in that higher ratings of self-efficacy in ELs is a potential strength that can be leveraged to support academic
performance (Kim et al., 2015).
Results for the students included in the reaching proficiency category indicated that this group of students exhibited some
marginal differentiation from their English proficient peers on measures of self-efficacy and achievement across mathematics and
science, but those differences did not reach statistical significance. In contrast, students reaching proficiency were significantly
different from students with limited proficiency in science self-efficacy and mathematics and science achievement. These findings
indicate that students who are reaching proficiency may more closely resemble their English proficient peers, and thus, reducing the
intensity of ESOL services for those students is likely a valid educational decision.
The second aim of this study was to examine the pathways by which students' English proficiency levels contribute to achieve-
ment through self-efficacy. First, the direct effects of self-efficacy and proficiency were explored. Consistent with prior literature and
theory (e.g., Areepattamannil et al., 2011; Bandura et al., 1996; Borman & Overman, 2004), mathematics and science self-efficacy
were significant contributors to their corresponding areas of achievement. When exploring differences across language proficiency
groups, after accounting for self-efficacy, demographic variables (gender, race/ethnicity, FRPL), and working memory, students with

9
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

Table 4
Indirect and direct effects models examining science self-efficacy, language proficiency level, and
science achievement.
β 95% CI

2 ⁎⁎⁎
Direct effects on science achievement (R = 0.37 )
Working memory 0.23⁎⁎⁎ [0.16, 0.31]
FRPL status −0.19⁎⁎⁎ [0.27, −0.12]
Gender (female) −0.18⁎⁎⁎ [−0.25, −0.11]
African American −0.13⁎ [−0.21, −0.05]
Asian American 0.10⁎ [0.01, 0.18]
Latino −0.13 [−0.21, −0.04]
Multiracial/Other race −0.09⁎ [−0.18, −001]
Math self-efficacy 0.16⁎⁎ [0.07, 0.25]
Reaching proficiency −0.03 [−0.13, 0.08]
Limited proficiency −0.15⁎⁎ [−0.25, −0.06]

Direct effects on science self-efficacy (R2 = 0.10⁎⁎⁎)


Working memory 0.01 [−0.09, 0.11]
FRPL status 0.03 [−0.06, 0.12]
Gender (female) −0.05 [−0.14, 0.03]
African American −0.06 [−0.17, 0.04]
Asian American 0.01 [−0.08, 0.11]
Latino −0.08 [−0.18, 0.02]
Multiracial/Other race −0.02 [−0.08, 0.05]
Reaching proficiency −0.12 [−0.26, 0.01]
Limited proficiency −0.27⁎⁎⁎ [−0.38, −0.17]

Indirect effects on science achievement


Reaching proficiency −0.02 [−0.05, 0.01]
Limited proficiency −0.04⁎ [−0.08, −0.01]

Note. β = Standardized beta. 95% CI = 95% confidence interval.


R2 value include all predictors and covariates.

p < .05.
⁎⁎
p < .01.
⁎⁎⁎
p < .001.

Fig. 1. Direct effects for math self-efficacy, language proficiency level, and math achievement. English proficient students used as reference group.
Covariates consist of students' working memory, race, gender, and FRPL status. Standardized beta weights are reported.

limited English proficiency showed significantly lower science achievement compared to the English proficient group, but this re-
lation was not statistically significant for mathematics. Reasons for the influence of proficiency on achievement in science, but not in
mathematics, may relate to language demands across content areas. Whereas upper elementary mathematics incorporates some tasks
with reduced linguistic load, such as basic computation and memorization of numerical formulas, science consistently requires
language to engage in scientific reading, writing, and discussion (Janzen, 2008). The high demand of CALP needed to understand
scientific vocabulary and engage in interactive classroom discourse (NGSS, 2012) is likely to make science a particularly challenging
academic content area for ELs (Lee et al., 2013). From a measurement perspective, another key explanation is that the adaptations to
the mathematics SOL (e.g., development of a plain English version) may have been quite beneficial to students with lower English

10
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

Fig. 2. Direct and indirect effects for science self-efficacy, language proficiency level, and science achievement. English proficient students used as
reference group. Covariates consist of students' working memory, race, gender, and FRPL status. Standardized beta weights are reported.

proficiency and would likely be a useful adaptation, when possible, for standardized science assessment as well.
An additional explanation may stem from school allocation of resources and assistance provided for EL students. Given ac-
countability standards, educators emphasize some subjects (e.g., Reading, Math) more than others (e.g., Science, Social Studies),
resulting in different teaching practices (i.e., ESOL teachers and mathematics specialists may “push in” for mathematics instruction
more often than for science instruction). Districts and schools tend to devote additional staff, programs, and instructional time to
mathematics more than science (Diamond & Spillane, 2004; National Institute of Child Health and Human Development-Early Child
Care Research Network, 2005). Additional resources in mathematics may be positively benefiting EL students.
Partially consistent with our hypotheses, examination of indirect models revealed that the relation between limited English
proficiency status and achievement in science was influenced by science self-efficacy. It is important to acknowledge that the indirect
effect of science self-efficacy for limited English proficient students was modest in magnitude and requires replication, with the
inclusion of prior science achievement, before firm conclusions can be drawn. However, this indirect effect, considered in combi-
nation with the direct influence of science self-efficacy on science achievement, has potential implications for classroom instruction in
science. Educators may be more effective in teaching ELs with limited proficiency if they integrate supports for science content
understanding and vocabulary knowledge (Parker & Gerber, 2000; Wenglinsky, 2000), while simultaneously monitoring and working
to strengthen feelings of efficacy and motivation within the content domain, such as using verbal feedback to highlight student
strengths. Teachers also can support EL students' efforts to participate in scientific discourse by recognizing and valuing all student
contributions, regardless of the precision of their language, as a means of developing their sense of competence. These supportive
teacher practices are particularly relevant given that EL students' perceptions of their teachers' beliefs about their academic ability
also play a key role in academic performance (Lewis et al., 2012).

9. Limitations and future directions

Several limitations warrant mention. First, whereas this study examined levels of English proficiency, it did not explore differ-
ences in the acculturation and generation status of families or the variations in specific types of languages spoken by students. Slama
(2012) found that U.S.-born ELs acquired academic language at a slightly different pace than foreign-born ELs indicating that
generation status and acculturation are important considerations when understanding the heterogeneity of ELs' academic perfor-
mance. Furthermore, considering differences (or similarities) among the languages students speak at home and in the classroom
environment may guide the modification of instructional supports to foster both self-efficacy and achievement. For instance, students
who speak a language in which there are many cognates (i.e., words derived from the same root) between the native language and
English may have increased comprehension if those cognates are pointed out by teachers or peers (August, Carlo, Dressler, & Snow,
2005). Additionally, students in a classroom with multiple peers that speak their native language may benefit from discussing
concepts with their classmates in their native language (Ballantyne, Sanderman, & Levy, 2008). Examining sociocultural differences
among students in relation to self-efficacy and achievement was beyond the scope of this study, but likely an important factor to
consider when providing learning opportunities for EL students given that teachers can leverage these factors to enhance students'
feelings of academic competence.
Second, this study examined self-efficacy as a self-report measure at a single fall time point without controlling for prior
achievement. Tracking EL students over time as their English proficiency level increases would provide a stronger case for links
between proficiency level, self-efficacy, and achievement. It would also allow for cross-lag models that could examine prior self-
efficacy and achievement in relation to current self-efficacy and achievement, untangling some of the factors that contribute to EL
students scoring lower than their English proficient peers. Relatedly, this study used English self-efficacy measures that were

11
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

administered with no standardized adjustments for supporting the comprehension of ELs. There are inherent limitations to the
accuracy of self-report measures particularly with younger students or students who have lower language skills (Fulmer & Frijters,
2009). One on-going challenge in the measurement of certain social-cognitive processes, such as self-efficacy, is that students
themselves are often the best reporters (Assor & Connell, 1992; Bandura, 1993). However, future research should consider in-
corporating standardized language modifications as well as teachers' appraisal of ability as another influential factor in the relations
among proficiency, efficacy, and achievement.
Third, our fairly modest sample size was limited to fifth grade students in one district, and our language proficiency categories
were based on the practices used in that particular district. These aspects of the study limit its generalizability. Given the variability in
the methods used by districts and states to classify EL students (Francis & Stephens, 2018; Maxwell, 2013), continued research should
use robust sample sizes to examine longitudinal outcomes across districts with varying methods for classifying ELs. Inclusion of a
larger number of schools and districts would also allow for more complex HLM analyses to be conducted.
Fourth, the current study relied on standardized tests for student outcomes. Although standardized tests are a policy-relevant
indicator of student achievement, they may not be particularly sensitive to associations between internal processes and achievement
(Ashcraft & Moore, 2009; Ma, 1999). Using varied measures, such as teacher and student ratings of engagement or curriculum-based
assessments, may provide a more nuanced understanding of the relation between English proficiency and achievement.
Finally, the present study did not have observational data capturing the quality academic language instruction occurring within
the mathematics and science classrooms. Understanding the pivotal role of teaching practices in the relations among language
proficiency, self-efficacy, and achievement is an important aspect of this research (Lewis et al., 2012).

10. Practical implications

The present study reveals differences and similarities in efficacy and achievement across three English proficiency groups.
Importantly, students in the reaching proficiency category scored similarly to their English proficient peers in mathematics and
science achievement and self-efficacy, although there were some marginal differences in scores. This positive finding indicates, from
a practical standpoint, that reducing ESL supports in mathematics and science based on this particular classification of WIDA ACCESS
for ELLs scores appeared to be an appropriate instructional decision for this sample of students. Analyses indicated that students
classified as limited English proficient would likely benefit from continued supports in both content areas, with particular attention
given to science learning. These findings underscore the importance of teachers' awareness of students who receive ESOL services and
their subsequent use of culturally informed practices. Providing adequate support to those students, given the language demands of
science, in particular, may influence students' feelings of competence and overall achievement.

11. Conclusions

These findings explore heterogeneity among different EL student groups and extend our understanding beyond the basic di-
chotomy of ELs versus non-ELs. Educators and researchers should continue to consider more nuanced definitions of English profi-
ciency. Careful consideration of proficiency differences among EL students and the linguistic demands of course content (e.g.,
mathematics, science) may lead to higher quality teaching strategies and interventions tailored to the specific needs of each EL
student for each class subject.
The results suggest that educators should pay attention to EL students' perceptions of competence in science, especially for
students who have limited proficiency skills. Among students with limited proficiency, high self-appraisal of their own efficacy may
serve as a strength and bolster their science performance. Although we cannot presume causal inferences, the present investigation
suggests the value of future studies of practices and programs that provide students with high quality instruction and verbal feedback
to enhance their self-efficacy in this content domain.
Overall, adopting a strengths-based approach to educating students who are learning English is essential in efforts to increase EL
students' self-efficacy across content areas. Pre-service and in-service teachers must be equipped with empirically-based knowledge
and tools regarding effective instructional practices with ELs, particularly those students with lower levels of proficiency. ELs' feelings
of competence, and ultimately achievement, are likely to increase through learning environments that provide appropriate levels of
support and encouragement and that convey a sense of value for linguistic diversity (Sakiz, Pape, & Hoy, 2012).

Acknowledgements

The research reported here was supported by the Institute of Education Sciences, U.S. Department of Education, through Grants
R305A070063 and R305B130013 to the University of Virginia. The opinions expressed are those of the authors and do not represent
views of the U.S. Department of Education.

References

American Federation of Teachers (2018). Every Student Succeeds Act: A new day in public education. Retrieved from https://www.aft.org/sites/default/files/essa_ells.
pdf.
Areepattamannil, S., Freeman, J. G., & Klinger, D. A. (2011). Influence of motivation, self-beliefs, and instructional practices on science achievement of adolescents in
Canada. Social Psychology of Education, 14, 233–259.

12
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

Ashcraft, M. H., & Moore, A. M. (2009). Mathematics anxiety and the affective drop in performance. Journal of Psychoeducational Assessment, 27, 197–205. https://doi.
org/10.1177/0734282908330580.
Assor, A., & Connell, J. P. (1992). The validity of students’ self-reports as measures of performance affecting self-appraisals. In D. H. Schunk, & J. L. Meece (Eds.).
Student perceptions in the classroom (pp. 25–47). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Aud, S., Wilkinson-Flicker, S., Kristapovich, P., Rathbun, A., Wang, X., & Zhang, J. (2013). The Condition of Education 2013 (NCES 2013-037). Washington, DC: U.S.
Department of Education, National Center for Education Statistics. Retrieved from http://nces.ed.gov/pubs2013/2013037.pdf.
August, D., Carlo, M., Dressler, C., & Snow, C. (2005). The critical role of vocabulary development for English language learners. Learning Disabilities Research &
Practice, 20, 50–57. https://doi.org/10.1111/j.1540-5826.2005.00120.x.
Ballantyne, K. G., Sanderman, A. R., & Levy, J. (2008). Educating English language learners: Building teacher capacity. Washington, DC: National Clearinghouse for English
Language Acquisition. Available at http://www.ncela.gwu.edu/practice/mainstream_teachers.htm.
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191–215. https://doi.org/10.1037/0033-295X.84.2.191.
Bandura, A. (1993). Perceived self-efficacy in cognitive development and functioning. Educational Psychologist, 28, 117–148.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York, NY: Freeman.
Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (1996). Multifaceted impact of self-efficacy beliefs on academic functioning. Child Development, 67,
1206–1222.
Bong, M., & Skaalvik, E. M. (2003). Self-concept and self-efficacy: How different are they really? Educational Psychology Review, 15, 1–40.
Borman, G. D., & Overman, L. T. (2004). Academic resilience in mathematics among poor and minority students. The Elementary School Journal, 104, 177–195. https://
doi.org/10.1086/499748.
Britner, S. L., & Pajares, F. (2001). Self-efficacy beliefs, motivation, race, and gender in middle school science. Journal of Women and Minorities in Science and
Engineering, 7, 271–285.
Britner, S. L., & Pajares, F. (2006). Sources of science self-efficacy beliefs of middle school students. Journal of Research in Science Teaching, 43, 485–499. https://doi.
org/10.1002/tea.20131.
Center for Applied Linguistics (2017). Annual technical report for ACCESS for ELLs 2.0 paper English language proficiency test, series 400, 2015–2016 administration.
Retrieved from https://www.ride.ri.gov/Portals/0/Uploads/Documents/Instruction-and-Assessment-World-Class-Standards/Assessment/ACCESS/ACCESS_
Technical_Report_2015%E2%80%932016_Paper.pdf.
Cleary, T. J., & Kitsantas, A. (2017). Motivation and self-regulated learning influences on middle school mathematics achievement. School Psychology Review, 46,
88–107.
Cook, H. G., Boals, T., & Lundberg, T. (2011). Academic achievement for English learners: What can we reasonably expect? Phi Delta Kappan, 93(3), 66–69.
Cummins, J. (1979). Linguistic interdependence and the educational development of bilingual children. Review of Educational Research, 49, 222–251.
Cummins, J. (1981). Age on arrival and immigrant second language learning in Canada: A reassessment. Applied Linguistics, 11, 132–149. https://doi.org/10.1093/
applin/2.2.132.
Diamond, J. B., & Spillane, J. P. (2004). High-stakes accountability in urban elementary schools: Challenging or reproducing inequality? Teachers College Record,
106(6), 1145–1176.
Eccles, J. S. (1999). The development of children ages 6 to 14. The Future of Children, 9, 30–44. Retrieved from http://www.jstor.org/stable/1602703.
Eccles, J. S., Wigfield, A., & Schiefele, U. (1998). Motivation to succeed. In W. Damon, & N. Eisenberg (Eds.). Handbook of child psychology: Social, emotional, and
personality development (pp. 1017–1095). Hoboken, NJ: John Wiley & Sons Inc.
Enders, C. K., & Bandalos, D. L. (2001). The relative performance of full information maximum likelihood estimation for missing data in structural equation models.
Structural Equation Modeling, 8, 430–457. https://doi.org/10.1207/S15328007SEM0803_5.
Engle, R. W. (2002). Working memory capacity as executive attention. Current Directions in Psychological Science, 11, 19–23. https://doi.org/10.1111/1467-8721.
00160.
Ernst-Slavit, G., & Slavit, D. (2013, March). Matheamatically speaking. Language Magazine, 12, 32–36.
ESSA (2015). Every Student Succeeds Act of 2015, Pub. L. No. 114-95 § 114 Stat. 1177 (2015–2016).
Flores, N. (2016, November). Deficit perspectives and bilingual education in a post-civil rights era. Journal of Language and Literacy Education: Scholars Speak Out.
Retrieved from http://jolle.coe.uga.edu/wp-content/uploads/2014/01/SSO_Flores_November_LCG.pdf.
Fox, J., & Fairbairn, S. (2011). Test review: ACCESS for ELLs. Language Testing, 28, 425–431.
Fradd, S. H., Lee, O., Sutman, F. X., & Saxton, M. K. (2001). Promoting science literacy with English language learners through instructional materials development: A
case study. Bilingual Research Journal, 25, 479–501.
Francis, D., & Stephens, A. (2018). English learners in STEM subjects: Transforming classrooms, schools, and lives. Washington, DC: National Academies Press.
Fulmer, S. M., & Frijters, J. C. (2009). A review of self-report and alternative approaches in the measurement of student motivation. Educational Psychology Review, 21,
219–246.
Garcia, E., & Cuellar, D. (2006). Who are these linguistically and culturally diverse students? The Teachers College Record, 108(11), 2220–2246. https://doi.org/10.
1111/j.1467-9620.2006.00780.x.
Hakuta, K., Butler, Y., & Witt, D. (2000). How long does it take English learners to attain proficiency? (Policy Report 2000-1). Stanford, CA: The University of California
Linguistic Minority Research Institute.
Hart, J. E., & Lee, O. (2003). Teacher professional development to improve the science and literacy achievement of English language learners. Bilingual Research
Journal, 27, 475–501.
Harter, S. (1998). The development of self-representations. In W. Damon, & N. Eisenberg (Eds.). Handbook of child psychology: Social, emotional, and personality
development (pp. 553–617). Hoboken, NJ: John Wiley & Sons Inc.
Hayes, A. F., & Preacher, K., J. (2014). Statistical mediation analysis with a multicategorical independent variable. British Journal of Mathematical and Statistical
Psychology, 67, 451–470.
Hernandez, D. J., Denton, N. A., & Macartney, S. E. (2008). Children in immigrant families: Looking to America’s future. Social Policy Report, 22, 3–22.
Huang, G. G. (2000). Mathematics achievement by immigrant children: A comparison of five English-speaking countries. Education Policy Analysis Archives, 8(25),
1–16.
Janzen, J. (2008). Teaching English language learners in the content areas. Review of Educational Research, 78, 1010–1038.
Joët, G., Usher, E. L., & Bressoux, P. (2011). Sources of self-efficacy: An investigation of elementary school students in France. Journal of Educational Psychology, 103,
649–663. https://doi.org/10.1037/a002404amrti.
Karabenick, S. A., & Noda, P. A. C. (2004). Professional development implications of teachers’ beliefs and attitudes toward English language learners. Bilingual Research
Journal, 28, 55–75.
Kim, D., Wang, C., Ahn, H. S., & Bong, M. (2015). English language learners’ self-efficacy profiles and relationship with self-regulated learning strategies. Learning and
Individual Differences, 38, 136–142.
Kindler, A. L. (2002). Survey of the states’ limited English proficient students and available educational programs and services, 2000–2001 summary report. Washington, DC:
National Clearinghouse for English Language Acquisition and Language Instruction Educational Programs.
Kiran, D., & Sungur, S. (2012). Middle school students’ science self-efficacy and its sources: Examination of gender difference. Journal of Science Education and
Technology, 21, 619–630.
Lau, S., & Roeser, R. W. (2002). Cognitive abilities and motivational processes in high school students’ situational engagement and achievement in science. Educational
Assessment, 8, 139–162. https://doi.org/10.1207/S15326977EA0802_04.
Lee, O., Quinn, H., & Valdés, G. (2013). Science and language for English language learners in relation to next generation science standards and with implications for
Common Core State Standards for English language arts and mathematics. Educational Researcher, 42, 223–233. https://doi.org/10.3102/0013189X13480524.
LeFevre, J.-A., DeStefano, D., Coleman, B., & Shanahan, T. (2005). Mathematical cognition and working memory. In J. I. D. Campbell (Ed.). Handbook of mathematical

13
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

cognition (pp. 361–378). New York, NY: Psychology Press.


Lesaux, N. K., & Kieffer, M. J. (2010). Exploring sources of reading comprehension difficulties among language minority learners and their classmates in early
adolescence. American Educational Research Journal, 47, 596–632.
Lewis, J. L., Ream, R. K., Bocian, K. M., Cardullo, R. A., Hammond, K. A., & Fast, L. A. (2012). Con Carino: Teacher caring, math self-efficacy, and math achievement
among Hispanic English learners. Teachers College Record, 114(7), 1–42.
Ma, X. (1999). A meta-analysis of the relationship between anxiety toward mathematics and achievement in mathematics. Journal of Research in Mathematics Education,
30, 520–540. https://doi.org/10.2307/749772.
Maerten-Rivera, J., Ahn, S., Lanier, K., Diaz, J., & Lee, O. (2016). Effect of a multiyear intervention on science achievement of all students including English language
learners. The Elementary School Journal, 116, 600–624.
Magogwe, J. M., & Oliver, R. (2007). The relationship between language learning strategies, proficiency, age and self-efficacy beliefs: A study of language learners in
Botswana. System, 35, 338–352.
Marks, G. N. (2005). Accounting for immigrant non-immigrant differences in reading and mathematics in twenty countries. Ethnic and Racial Studies, 28, 925–946.
https://doi.org/10.1080/01419870500158943.
Martin, D. P., & Rimm-Kaufman, S. E. (2015). Do student self-efficacy and teacher-student interaction quality contribute to emotional and social engagement in fifth
grade math? Journal of School Psychology, 53, 359–373.
Mather, N., & Woodcock, R. W. (2001). Examiner’s manual. Itasca, IL: Riverside Publishing.
Maxwell, L. A. (2013). Push underway to standardized ELL definitions. Education Week, 32(22), 1,13. Retrieved from http://www.edweek.org/ew/toc/2013/02/27/
index.html.
McGrew, K. S., & Woodcock, R. W. (2001). Technical manual: Woodcock-Johnson III. Itasca, IL: Riverside Publishing.
McKown, C., & Weinstein, R. (2003). The development and consequences of stereotype consciousness in middle childhood. Child Development, 74, 498–515. https://
doi.org/10.1111/1467-8624.7402012.
McNeish, D., Stapleton, L. M., & Silverman, R. D. (2017). On the unnecessary ubiquity of hierarchical linear modeling. Psychological Methods, 22, 114–140.
Mercer, S. H., Nellis, L. M., Martínez, R. S., & Kirk, M. (2011). Supporting the students most in need: Academic self-efficacy and perceived teacher support in relation to
within-year academic growth. Journal of School Psychology, 49, 323–338. https://doi.org/10.1016/j.jsp.2011.03.006.
Midgley, C., Maehr, M. L., Hruda, L. Z., Anderman, E., Anderman, L., Freeman, K. E., ... Urdan, T. (2000). Manual for the patterns of adaptive learning scales. Ann Arbor,
MI: The University of Michigan.
Muthén, L. K., & Muthén, B. O. (1998–2010). Mplus user’s guide. Sixth edition [computer software and manual]. Los Angeles, CA: Author.
Muthen, B. O., & Satorra, A. (1995). Complex sample data in structural equation modeling. Sociological Methodology, 267–316.
National Center for Education Statistics (NCES) (2017). Fast facts: Back to school statistics. Washington, DC: Institute of Education Sciences, U.S. Department of
Education.
National Council of Teachers of Mathematics (2011). Mathematics homonyms and homophones. Retrieved from www.nctm.org.
National Governors Association Center for Best Practices & Council of Chief State School Officers (2010). Common Core State Standards for mathematics. Retrieved
from http://www.corestandards.org/Math/.
National Institute of Child Health and Human Development–Early Child Care Research Network (2005). A day in third grade: A large-scale study of classroom quality
and teacher and student behavior. Elementary School Journal, 105, 305–323. https://doi.org/10.1086/428746.
National Mathematics Advisory Panel (2008). Foundations for success: The final report of the National Mathematics Advisory Panel. Washington, DC: U.S. Department of
Education.
National Research Council (2011). Assessing 21st century skills: Summary of a workshop. J. A. Koenig, Rapporteur. Committee on the Assessment of 21st Century Skills. Board
on Testing and Assessment, Division of Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press.
Neugebauer, S. R., & Howard, E. R. (2015). Exploring associations among writing self-perceptions, writing abilities, and native language of English-Spanish two-way
immersion students. Bilingual Research Journal, 38, 313–335.
Next Generation Science Standards [NGSS]. (2012, May 14). Next generation science standards. Retrieved from http://www.nextgenscience.org/.
Nieto, S. (2010, August). Language, diversity, and learning: Lessons for education in the 21st century. CalDigest. Retrieved from https://eclass.upatras.gr/modules/
document/file.php/PDE1439/language-diversity-and-learning.pdf.
Pajares, F. (1996). Self-efficacy beliefs in academic settings. Review of Educational Research, 66, 543–578.
Pajares, F., & Graham, L. (1999). Self-efficacy, motivation constructs, and mathematics performance of entering middle school students. Contemporary Educational
Psychology, 24, 124–139. https://doi.org/10.1006/ceps.1998.0991.
Pajares, F., & Kranzler, J. (1995). Self-efficacy beliefs and general mental ability in mathematical problem-solving. Contemporary Educational Psychology, 20, 426–443.
Parker, V., & Gerber, B. (2000). Effects of a science intervention program on middle-grade student achievement and attitudes. School Science and Mathematics, 100,
236–242. https://doi.org/10.1111/j.1949-8594.2000.tb17263.x.
Polat, N., Zarecky-Hodge, A., & Schreiber, J. B. (2016). Academic growth trajectories of ELLs in NAEP data: The case of fourth-and eighth-grade ELLs and non-ELLs on
mathematics and reading tests. The Journal of Educational Research, 109, 541–553.
Rimm-Kaufman, S. E., Baroody, A. E., Larsen, R. A., Curby, T. W., & Abry, T. (2015). To what extent do teacher–student interaction quality and student gender
contribute to fifth graders’ engagement in mathematics learning? Journal of Educational Psychology, 107, 170–185.
Robinson, J. P., & Lubienski, S. T. (2011). The development of gender achievement gaps in mathematics and reading during elementary and middle school: Examining
direct cognitive assessments and teacher ratings. American Educational Research Journal, 48, 268–302.
Ron, P. (1999). Spanish–English language issues in the mathematics classroom. In L. Ortiz-Franco, N. G. Hernandez, & Y. De la Cruz (Eds.). Changing the faces of
mathematics: Perspectives on Latinos (pp. 23–34). Reston, VA: National Council of Teachers of Mathematics.
Sakiz, G., Pape, S. J., & Hoy, A. W. (2012). Does perceived teacher affective support matter for middle school students in mathematics classrooms? Journal of School
Psychology, 50, 235–255. https://doi.org/10.1016/j.jsp.2011.10.005.
Santau, A. O., Maerten-Rivera, J. L., & Corinne Huggins, A. (2011). Science achievement of English language learners in urban elementary schools: Fourth-grade
student achievement results from a professional development intervention. Science Education, 95, 771–793.
Schleppegrell, M. J. (2007). The linguistic challenges of mathematics learning and teaching: A research review. Reading & Writing Quarterly, 23, 139–159.
Schunk, D. H., & Meece, J. L. (2006). Self-efficacy development in adolescence. In F. Pajares, & T. Urdan (Eds.). Self-efficacy beliefs of adolescents (pp. 71–96).
Greenwich, CT: Information Age Publishing.
Schunk, D. H., & Pajares, F. (2002). The development of academic self-efficacy. In A. Wigfield, & J. S. Eccles (Eds.). Development of achievement motivation (pp. 15–31).
San Diego, CA: Academic Press.
Skaalvik, E. M., Federici, R. A., & Klassen, R. M. (2015). Mathematics achievement and self-efficacy: Relations with motivation for mathematics. International Journal
for Educational Research, 72, 129–136.
Slama, R. B. (2012). A longitudinal analysis of academic English proficiency outcomes for adolescent English language learners in the United States. Journal of
Educational Psychology, 104, 265–285.
Swars, S. L., & Chestnutt, C. (2016). Transitioning to the Common Core State Standards for Mathematics: A mixed methods study of elementary teachers’ experiences
and perspectives. School Science and Mathematics, 116, 212–224.
Tong, F. T., Irby, B. J., Lara-Alecio, R., & Koch, J. (2014). Integrating literacy and science for English language learners: From learning-to-read to reading-to-learn. The
Journal of Educational Research, 107, 410–426.
Townsend, D. (2015). Who’s using the language? Supporting middle school students with content area academic language. Journal of Adolescent & Adult Literacy, 58,
376–387.
Virginia Department of Education [VDOE] (2008). Virginia standards of learning technical report: 2008–2009 administration cycle. Retrieved from http://www.doe.
virginia.gov/testing/test_administration/technical_reports/sol_technical_report_2008-09_administration_cycle.pdf.

14
L.E. Sandilos, et al. Journal of School Psychology 79 (2020) 1–15

Virginia Department of Education [VDOE] (2009). 2009 mathematics Standards of Learning: Plain English mathematics test information. Retrieved from: http://
www.doe.virginia.gov/testing/sol/standards_docs/mathematics/plain_english_information.pdf.
Wang, C. H., Harrison, J., Cardullo, V., & Lin, X. (2018). Exploring the relationship among international students’ English self-efficacy, using English to learn self-
efficacy, and academic self-efficacy. Journal of International Students, 8, 233–250.
Wenglinsky, H. (2000). How teaching matters: Bringing the classroom back into discussions of teacher quality. Princeton, NY: Milken Family Foundation and Educational
Testing Service.
WIDA Consortium (2007). The WIDA English language proficiency standards, 2007 edition, preKindergarten through grade 5. Retreived from https://wida.wisc.edu/
sites/default/files/resource/2007-ELPS-PreK-5.pdf.
Wilkins, J. M. (2004). Mathematics and science self-concept: An international investigation. The Journal of Experimental Education, 72, 331–346. https://doi.org/10.
3200/JEXE.72.4.331-346.
Woodcock, R. W., McGrew, K. S., & Mather, N. (2001). Woodcock-Johnson III tests of cognitive abilities. Itasca, IL: Riverside Publication.
World-Class Instructional Design and Assessment (WIDA®) (2011). WIDA ACCESS placement test. Madison, WI: Board of Regents of the University of Wisconsin System.

15

You might also like