Educational Psychology Review (2021) 33:1097–1124

https://doi.org/10.1007/s10648-020-09554-w

META-ANALYSIS

Investigating the Unique Predictors of Word-Problem

Solving Using Meta-Analytic Structural
Equation Modeling

Xin Lin 1

Published online: 1 September 2020

# Springer Science+Business Media, LLC, part of Springer Nature 2020

Abstract
The purpose of the present study is to clarify the contributions of cognitive skills
(nonverbal reasoning, language comprehension, working memory, attention, processing
speed) and academic skills (mathematics facts retrieval, mathematics computation, math-
ematics vocabulary, reading comprehension) in performing mathematics word problems
among elementary school students. With the two-stage meta-analytic structural equation
modeling approach, I synthesized 112 correlation matrices from 98 empirical studies
(N = 111,346) and fitted the hypothesized partial mediation model. Overall, path analysis
indicated that language comprehension, working memory, attention, mathematics vocab-
ulary, and mathematics computation were unique predictors of word-problem solving.
Subgroup analysis demonstrated different unique predictors for younger and older stu-
dents to perform word problems (K-2nd grades versus 3rd–5th grades). Implications,
limitations, and future directions are discussed.

Keywords Academic skills . Cognitiveskills . Meta-analyticstructural equationmodeling . Word-

problem solving

Developing mathematics proficiency is necessary because it has a strong association with a

student’s academic performance, college enrollment, and adulthood outcomes (National
Council of Teachers of Mathematics 2006). Recently, there has been an increase in research
exploring important factors associated with students’ mathematics performance (e.g., Fuchs
et al. 2015; Fung and Swanson 2017; Namkung et al. 2019; Peng and Lin 2019). Two major
categories of factors are identified essential for mathematics performance, including cognitive
skills (i.e., working memory, nonverbal reasoning, processing speed; e.g., Fuchs et al. 2006;

Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10648-020-

09554-w) contains supplementary material, which is available to authorized users.

* Xin Lin
[email protected]

1
Department of Special Education, The University of Texas at Austin, Austin, TX 78712, USA
1098 Educational Psychology Review (2021) 33:1097–1124

Fung and Swanson 2017) and academic skills (i.e., mathematics facts retrieval, mathematics
vocabulary, reading comprehension; e.g., Fuchs et al. 2018; Peng and Lin 2019). Incorporating
such factors in mathematics instruction and intervention could help facilitate students’ math-
ematics performances and remediate their mathematics difficulties. The present study focus on
word-problem solving (WPS) because student’s WPS performance is a strong school-age
predictor of their employment and wages in adulthood (e.g., Every Child a Chance Trust 2009;
Murnane et al. 2001), and WPS is often emphasized in key educational initiatives (e.g.,
Common Core State Standards Initiative 2010; National Mathematics Advisory Panel 2008).
Word problems (e.g., John has $35. He spent $17 at the bookstore and $6 at the post office.
How much money did he spend?) are defined as linguistically presented problems that require
students to identify important information within the text or visual (e.g., chart or graph) to
answer a prompt, construct equation(s), and perform calculation(s) to develop a problem
solution (Powell et al. 2019; Wang et al. 2016). Solving word problems proves challenging
for many students, particularly students with or at risk of mathematics difficulties (Powell et al.
2019; Sharpe et al. 2014) because numerous steps and skills are necessary to solve a word
problem (Fuchs et al. 2006, 2015, 2018). To support students to improve their WPS,
investigating the relative contributions of each skill in performing word problems is necessary.
Given that many cognitive and academic skills are intertwined in performing word problems,
previous studies frequently used structural equation modeling (SEM) to investigate unique
cognitive and academic skills essential for WPS (e.g., Björn et al. 2016; Fuchs et al. 2006,
2016; Singer et al. 2018; Wang et al. 2016).
The increase of empirical research findings based on SEM, however, is not enough for drawing
conclusions (Cheung and Chan 2005). First, previous research involved different cognitive and
academic predictors of WPS. For instance, besides the cognitive skills (i.e., working memory,
nonverbal reasoning, language comprehension) considered by Singer et al. (2018), Fuchs et al.
(2006) also involved attention and processing speed in their models. By including different
predicting skills in their path models, not surprisingly, prior studies have revealed mixed findings.
Second, among the empirical studies using SEM to investigate WPS predictors, some studies
considered that the effects of cognitive skills on WPS are being partially mediated by academic
skills (e.g., Fuchs et al. 2006, 2015; Fung and Swanson 2017; Peng and Lin 2019; Singer et al.
2018; Zheng et al. 2011), whereas some studies did not (e.g., Björn et al. 2016). Additionally,
prior studies also included different academic skills as mediators in their models including
mathematics facts retrieval and mathematics computation (e.g., Fuchs et al. 2006; Singer et al.
2018), mathematics vocabulary (Fuchs et al. 2015; Peng and Lin 2019), as well as reading and
mathematics computation (e.g., Fung and Swanson 2017; Zheng et al. 2011). Finally, the lack of
research, including a comprehensive set of academic mediators, further limits our capacity to
conclude from previous WPS literature. Moreover, a limitation of past meta-analyses investigat-
ing the cognitive correlates of WPS is that none of them investigated theoretically meaningful
academic mediators of these relationships (e.g., working memory (Peng et al. 2016) and nonver-
bal reasoning (Peng et al. 2019; Hembree 1992)).
Given the issues remained in the past empirical research using SEM and univariate meta-
analysis, it is necessary to investigate the unique predictors of WPS using meta-analytic SEM
(MASEM). On the one hand, this approach allows for the inclusion of a comprehensive set of
cognitive and academic predictors, as well as accounting for the mediation effects of academic
skills. On the other hand, this approach helps integrate a comprehensive set of cognitive and
academic skills into the analysis. In the following section, I describe the theoretical framework
and practical consideration of the current study.
Educational Psychology Review (2021) 33:1097–1124 1099

Theoretical Framework and Practice Consideration

Cognitive skills refer to those general mental capacities that make it possible for us to learn involving
perception, attention, working memory, and nonverbal reasoning (Gottfredson 1997; Jaeggi et al.
2008). As opposed to academic knowledge (the actual knowledge acquired during the learning
progression), cognitive skills have more to do with the mechanisms of how we learn. Many studies
over the past decades have demonstrated that cognitive skills are some of the determining factors of a
student’s learning ability (e.g., Cattell 1987; Sternberg et al. 2008). In other words, cognitive skills
are foundational for or causative of acquiring academic knowledge during the early stage.
In addition to the predicting effects of cognitive skills on students’ academic performance in
general, the dual-process theory of higher cognition further highlights the mediation roles of
academic skills (Evans and Stanovich 2013). In other words, students rely on both cognitive as
well as academic skills that they accumulated over time to perform complex problems. The
retrieval of such academic knowledge stored in long-term memory enables students to allocate
their limited cognitive resources to other unfamiliar problem solving processes flexibly.
Therefore, accounting for the mediation roles of academic skills is necessary to investigate
the relative contributions of cognitive skills on WPS.
To solve a word problem, which can be considered a higher-order mathematics task (Lai 2011),
students must rely on both cognitive and academic skills. Therefore, accounting for the relationships
between cognitive skills and academic skills is essential to explore the relative contributions of
cognitive skills for WPS. Put another way, the relative contributions of cognitive skills on WPS
might change when accounting for the mediation effects of academic skills. For example, students
tend to compensate for their lack of academic skills for WPS (i.e., mathematics facts, mathematics
vocabulary) or the ability to apply these skills fluently to solving word problems with cognitive
resources. In line with the dual-process theory, the present study proposes a partial mediation model
in which the effects of cognitive skills on WPS were partially mediated by a comprehensive set of
academic skills necessary for performing word problems. Investigating such a model could help
clarify the relative contributions of cognitive and academic predictors in the WPS literature. The
hypothesized partial mediation model is presented in Fig. 1.
Understanding the unique predictors of WPS is important not only theoretically but also
practically for instruction and intervention purposes. Given that WPS is a higher-order mathematics
task (Lai 2011), which involves multiple cognition and thinking processes, exhausting too many
cognitive resources on simple academic processes like mathematics facts retrieval would hinder
students from performing other complicated WPS procedures (i.e., correctly identifying the problem
types, selecting appropriate solution strategies). In contrast, if students can directly retrieve and
fluently use their academic skills while performing word problems, they would have increased
flexibility in allocating their cognitive resources. In other words, if the predicting effects of cognitive
skills on WPS are significantly mediated by the academic skills, intervention efforts should be
devoted to enhancing WPS-related academic knowledge. If, however, cognitive skills demonstrated
unique impacts on WPS while accounting for academic skills as mediators, intervention should
involve cognitive training to facilitate students’ WPS performances (e.g., Fuchs et al. 2020).

Cognitive and Academics Predictors of Word-Problem Solving

To solve a word problem, students need to understand the word problem statement to identify
missing information, construct the number sentence, and derive the computation problem
1100 Educational Psychology Review (2021) 33:1097–1124

Fig. 1 Hypothesized partial mediation model

(Fuchs et al. 2008). Not surprisingly, solving word problems requires the consolidation of a
variety of cognitive skills (i.e., nonverbal reasoning, working memory, language comprehen-
sion) and academic skills (i.e., reading comprehension, computation). Due to the complexity of
WPS and the involvement of multiple cognitive and academic skills, word problems become
more difficult for students to solve than de-contextualized computation problems (Daroczy
et al. 2015). Understanding the relative roles of cognitive skills and academic skills in
performing word problems is necessary. Table 1 contains the definitions and examples of
the included cognitive and academic predictors.

Cognitive Predictors

In terms of cognitive predictors, prior research findings provide the basis for hypothesizing
that working memory, attention, nonverbal reasoning, processing speed, and language com-
prehension are essential for WPS (e.g., Fuchs et al. 2006; Singer et al. 2018; Zheng et al.
2011). Among these cognitive skills, working memory, attention, and processing speed are
closely intertwined. Working memory refers to the capacity to simultaneously maintain
information and process an additional task (Daneman and Carpenter 1980). Theoretically,
WPS relies on working memory to direct attention to relevant information, coordinate multiple
cognitive resources, and inhibit irrelevant information in the word problem statement (Fuchs
et al. 2020; Peng et al. 2016). Previous studies usually reported a composite working memory
score (i.e., a composite score of listening recall and backward digit recall; Fuchs et al. 2006;
Spencer et al., 2020) or used a single working memory measure (e.g., Peng and Lin 2019).
Therefore, the present study does not further investigate whether components of working
memory (i.e., phonological working memory versus visual-spatial working memory; executive
component of working versus storage component of working memory) would moderate the
relation between working memory and WPS (e.g., Fung and Swanson 2017).
Educational Psychology Review (2021) 33:1097–1124 1101

Table 1 Definitions and examples of word-problem solving and predictor variables

Definition Examples

Word-problem Questions that involve the ability to WISC-Word Problem; Key-Math Problem
solving understand the problem narrative, focus on Solving; WJ-Applied Problems
relevant information, ignore irrelevant
information, construct a number sentence,
and solve for the missing number to find
the answer
Nonverbal Matrix reasoning: tasks that require students Raven’s matrices; Stanford-Binet matrices
reasoning to identify a rule underlying an incomplete
matrix of geometric figures and to select
the geometric figures that satisfy the rule
Nonmatrix reasoning: tasks that require Concept formation; picture analogies and
identifying a rule underlying a set of sequence; Stanford–Binet pattern analy-
pictures by inductive or deductive sis
reasoning or both
Language Vocabulary: tasks that require students to Peabody Picture Vocabulary Test;
comprehension explain what a word means or point to a Wechsler Abbreviated Scale of
picture corresponding to a word Intelligence-Vocabulary
Listening comprehension: tasks that require WAIS-Information; sentence repetition
students to comprehend a passage/sentence
in an oral format (oral comprehension)
Working memory Tasks that tap simultaneous process and Story retelling; backward digit span; visual
storage of verbal, numerical, and matrix mapping and directions
visual/spatial information
Attention Teachers’ rating of classroom attention SWAN attention rating scale; Sky Search
Processing speed The ability to automatically and fluently Visual matching
perform relatively easy or overlearned
cognitive tasks
Mathematics Tasks that measures students’ understanding Test of Mathematical Abilities-Vocabulary
vocabulary of words, phrases, and symbols routinely Subtest; WJ-III Quantitative Concepts
used in mathematics textbooks, Test-Concepts subtest
instructions, and assessments
Mathematics facts Solve single-digit arithmetic problems WJ-Math Fluency (e.g., 3 + 4 = 7)
retrieval
Mathematics Tasks that require students to add, subtract, WRAT-4-Math; WIAT Arithmetic
computation multiply, or divide whole numbers,
decimals, or fractions using algorithms and
arithmetic
Reading Tasks that require students to comprehend a Nelson Denny Reading Comprehension;
comprehension passage in written format Woodcock Reading Mastery
Tests-Reading Comprehension

It is not surprising that attention also represents a promising candidate given working
memory and attention are very closely related (Awh et al. 2006). According to Heitz et al.
(2005), the executive component of working memory is an attentional construct, which
preserves relevant information in an active state in the presence of interference. The critical
relevant theory views attention as a gatekeeper for working memory by biasing the encoding
of information toward the items that are most relevant to the present processing goals (Awh
et al. 2006). In line with previous studies, the current research indexes attention as teachers’
rating of classroom attention to avoid confusion with executive function’s attentional control
(e.g., Fuchs et al. 2006, 2010). Prior studies (Fuchs et al. 2005, 2006) suggested that teachers’
rating of attention was an essential determinant of first- and third-grade students’ WPS
performance.
1102 Educational Psychology Review (2021) 33:1097–1124

Processing speed refers to the efficiency of executing simple cognitive tasks, which dictates
how quickly numbers can be counted (Case 1985). Processing speed is closely associated with
students’ capacity of mathematics facts retrieval (Berg 2008; Hecht et al. 2001). For example,
Hecht et al. (2001) demonstrated the importance of processing speed as a correlate of
mathematics facts retrieval while controlling for vocabulary knowledge. Moreover, with a
slower speed of processing, the interval for deriving counted answers and for pairing a
problem stem with its answer in working memory increases (Fuchs et al. 2008). Not surpris-
ingly, researchers have documented that impaired processing speed is a primary determinant
for impaired working memory performance (Chiaravalloti and DeLuca 2008). This decreased
efficiency thus affects the ability to perform higher-order cognitive functions such as WPS (Lai
2011). Previous research suggested a slow speed of processing relates to WPS difficulty
because students struggling with WPS tend to use inefficient strategies while solving word
problems (Geary 1993).
Nonverbal reasoning refers to the capacity to reason and solve novel problems independent
of any knowledge from the past (Cattell 1963; Newton and McGrew 2010; Peng et al. 2019).
In the present study, nonverbal reasoning is operationalized as two kinds of tasks involving
matrix reasoning and nonmatrix reasoning. As Cooper and Sweller (1987) stated, categorizing
a word problem as a specific problem type, inferring information not evident in the problem
presentation, generalizing learned problems to novel problem situations, excluding irrelevant
information, and adopting appropriate solution strategies rely on nonverbal reasoning. Previ-
ous empirical research has identified nonverbal reasoning as a unique predictor of WPS while
controlling for a set of essential cognitive and academic skills (Fuchs et al. 2006).
Language comprehension is operationalized as listening comprehension (tasks that require
students to comprehend sentences or passage in oral format) and general vocabulary knowl-
edge (tasks that require students to explain what a word means or point to a picture
corresponding to a word). It is important to consider language comprehension for WPS, given
the apparent need to process linguistic information to build a problem representation. Also,
language is the principal medium of mathematics instruction in the school setting. Curriculum
standards not only emphasize using direct and explicit instruction to enhance conceptual
understanding through language (Gersten et al. 2009) but also encourage communication
and discussion among students in solving various mathematics problems (Moschkovich
et al. 2018). Moreover, language can help facilitate the retrieval of mathematics facts from
long-term memory, which can help allocate cognitive resources for more complex problem
solving processes. Therefore, not surprisingly, language comprehension demonstrates a unique
impact on WPS, even after controlling for other cognitive skills (Fuchs et al. 2006; Peng and
Lin 2019).

Academic Predictors

In terms of academic predictors, prior work and theoretical framework provide the basis for
hypothesizing that reading comprehension and mathematics vocabulary are essential for
problem representation, whereas mathematics facts retrieval and mathematics computation
are crucial for problem solution (Hegarty et al. 1995).
Mathematics facts retrieval (e.g., 2 + 5 = 7) refers to solving single-digit arithmetic prob-
lems, which typically rely on direct retrieval from long-term memory to solve. The capacity to
direct retrieve arithmetic facts frees up limited cognitive resources for other aspects of problem
solving (Geary et al. 1992). Mathematics computation (e.g., 253 + 125 = 378) is defined as
Educational Psychology Review (2021) 33:1097–1124 1103

using algorithms and arithmetic to add, subtract, multiply, or divide whole numbers, decimals,
or fractions. Mathematics computation is transparently a necessity for WPS because compu-
tation is the mathematical process through which the numerical information is manipulated to
produce a solution (Swanson and Beebe-Frankenberger 2004). Moreover, mathematics com-
putation, the skills in manipulating numbers procedurally, should enhance understanding about
numerical relationships, which, in turn, facilitates WPS (Fuchs et al. 2006). The current study
included both mathematics facts retrieval and mathematics computation because prior research
demonstrated their different predicting effects on WPS (Fuchs et al. 2006). Specifically, Fuchs
et al. (2006) suggested that as opposed to mathematics computation, mathematics facts
retrieval demonstrates a significant impact on WPS.
Mathematics vocabulary refers to students’ understanding of words and phrases routinely
used in mathematics instruction, textbooks, and assessments (e.g., greater than, above;
Monroe and Orme 2002; Moschkovich 2013). Recent empirical research demonstrated that
mathematics vocabulary was particularly crucial for WPS (Peng and Lin 2019). Specifically,
after controlling for cognitive skills, mathematics vocabulary still contributes to WPS. There is
research, however, questioned whether mathematics vocabulary is a distinct construct neces-
sary for mathematics performance, or it merely reflects students’ mathematical knowledge or
general language comprehension skills (Purpura et al. 2017). By including a comprehensive
set of cognitive skills and academic skills critical for WPS, the present study aimed to
investigate the relative role of mathematics vocabulary for WPS.
Reading comprehension refers to students’ capacity to comprehend a passage in written
format. Compared to a computation problem, students need to read and understand the word
problem statement to build a problem representation before the solution. Even when word
problems are read aloud to children, reading comprehension capacity is involved because it
provides continuous access to the written problem narrative after the tester completed reading
the problem (Fuchs et al. 2006). Potentially, this may reduce the load on working memory
resources and facilitate solution accuracy. According to Kintsch and Greeno (1985), the
general characteristics of the reading comprehension process apply across word problem
statements, stories, and informational passages. Furthermore, Fuchs et al. (2015) suggested
that word problem statements are distinct from other forms of reading comprehension by
requiring comprehension of mathematics vocabulary.

Grade Level May Affect the Unique Predictors of Word-Problem Solving

The unique predictors of WPS may differ across different grade levels (e.g., Swanson 2004).
First, as opposed to younger students (K-2nd graders), older students (3rd–5th graders) are
more likely to rely on the direct retrieval of academic knowledge (i.e., arithmetic facts) to solve
word problems. The direct retrieval of mathematics knowledge from long-term memory can
help students efficiently allocate their cognitive resources to more complex WPS processes
(i.e., inference making in WPS).
Second, solving word problems becomes increasingly complex and technical with formal
schooling. Therefore, the required skills for performing word problems may change accord-
ingly. For example, mathematics vocabulary used within word problems becomes accumula-
tively complicated through grade levels (Powell et al. 2017). In early grade levels, the word
problem is presented through general everyday language (e.g., Tom has two apples, and Lily
has three apples. How many apples do they have in all?). In contrast, word problems in the
1104 Educational Psychology Review (2021) 33:1097–1124

intermediate elementary grades are presented through more technical mathematics vocabulary
terms (e.g., All of the sides of Figure X are congruent. The length of one side of the figure is
3 cm. What is the perimeter of Figure X?). According to a review of the Partnership for
Assessment of Readiness for College and Careers (PARCC) 2015 Grade 5 Released Items,
only 21% of problems were written in straightforward, everyday language. Therefore, the
relative contributions of academic skills to WPS may change accordingly.
Moreover, the development of cognitive skills may lead to differences in the skills required
to solve the problem (Miller and Vernon 1997; Peng and Kievit 2020). For instance, the
concept of nonverbal reasoning has been considered a developmental one, which closely
relates to the chronological age (Wohlwill 1980). According to Peng and Kievit (2020), formal
education has a potential effect on the development of cognitive capacities (i.e., nonverbal
reasoning). Learning processes involved in many school activities support the formation of the
cognitive strategies needed for successful performance on cognitive ability tests (Glaser 1984;
Peng et al. 2019). According to my knowledge, Swanson (2004) is the only empirical study
examining the age-related differences in WPS accuracy. The present meta-analysis adds to the
literature by investigating the unique cognitive and academic predictors of WPS for younger
and older students separately.

Purpose

The present study seeks to clarify the relative contributions of cognitive and academic skills to
WPS with a potential partial mediation model among elementary school students. With a
multivariate approach, I first examine the unique cognitive and academic predictors of WPS.
Then, I investigate whether the unique predictors of WPS would change according to grade
levels (K-2nd grades versus 3rd–5th grades).

Method

Literature Search

I conducted a systematic search process to locate relevant studies for this meta-analysis. An
overview of search and screening procedures is presented in Fig. 2. First, I searched Education
Source, ERIC, and PsycINFO databases for articles published from 1922 to July 2019 (1922 is
the earliest publication date among records appearing in the initial search). Titles and abstracts
were searched for the combinations of the following search terms: (“word problem” or
“problem solving” or “story problem”) AND (“reasoning” or “intelligence” or “language” or
“listening comprehension” or “vocabulary” or “working memory” or “executive function” or
“attention” or “processing speed” or “text comprehension” or “reading comprehension” or
“calcul*” or “computation” or “arithmetic” or “mathematics facts retrieval” or “math fluen-
cy”). Meanwhile, the search was restricted to studies in English involving elementary students.
The search was not limited to peer-reviewed journal articles, but instead included research
reports, technical reports, dissertations, and conference proposals. The initial search yielded
2493 records. An additional eight studies were identified as eligible through other methods,
such as hand search of the references lists of the relevant reviews (Peng et al. 2019; Hembree
1992) and table of contents searches of the educational journals. I reviewed all studies by
Educational Psychology Review (2021) 33:1097–1124 1105

Fig. 2 Flowchart of the search and screening process

reading the titles and abstracts first, locating any promising studies based on the inclusion
criteria described below. This preliminary screening excluded eight studies that were not in
English, 41 studies for age (e.g., preschool, middle school and above), 1916 studies that were
not relevant (e.g., nonempirical, single-case studies, commentaries), and 178 duplicate studies.

Study Selection Criteria

Studies were eligible for this meta-analysis if they met all the following criteria. First,
because the present research is interested in informing the WPS instruction in elementary
school, studies had to include samples from kindergarten through fifth grade. For studies
that did not clearly state the grade level of the included students, I used the minimum age
for kindergarten entrance (4.58 years old) and the approximal graduation age for ele-
mentary school (11 years old) as cutoff points. Thus, I excluded studies involving only
preschool or middle school and above samples. Second, studies had to include at least
one quantitative measure of WPS and at least one quantitative measure of cognitive (e.g.,
nonverbal reasoning, working memory, attention, processing speed) or academic skills
(mathematics vocabulary, mathematics facts retrieval, mathematics computation, and
reading comprehension). Definitions and examples of these skills are presented in
Table 1. Third, studies had to report at least one direct bivariate correlation (r; not
partial correlation) between WPS and any measures of cognitive and academic skills.
Fourth, studies had to meet the adapted quality indicators for correlational research
developed by Thompson et al. (2005).
1106 Educational Psychology Review (2021) 33:1097–1124

Study Quality

I coded the study quality on six indicators adapted from the quality indicators for correlational
research developed by Thompson et al. (2005). The indicators included the following: (a)
report of reliability coefficients for all variables of interest, (b) inclusion of all effect sizes for
each outcome related to this meta-analysis, (c) appropriate interpretation of structure coeffi-
cients (e.g., beta weights, factor pattern coefficients, discriminant function coefficients), (d) use
of multivariate techniques for multiple outcome variables, (e) use of the highest available scale
of data (interval scale is not converted to nominal scale) except if such choices are justified and
thoughtfully considered, and (f) presentation of evidence that statistical assumptions are
sufficiently met (e.g., normal distribution, measures of central tendency, homogeneity of
variance).
Finally, all studies reported effect sizes relevant to this meta-analysis and used the highest
available scale of data. One study was excluded because reliability coefficients were not
reported for measures of interest. One study was excluded because it used univariate methods
in the presence of multiple outcome variables. Three studies were excluded to inappropriately
interpret the weights of the general linear model. Two studies were excluded because they did
not present evidence that statistical assumptions are met.

Included Studies

Full articles of the resulting 358 studies were obtained and reviewed carefully for eligibility. Of
358 studies, I excluded 153 studies without a WPS measure. Additionally, 54 studies were
excluded because they did not have any measure of cognitive (nonverbal reasoning, language
comprehension, working memory, processing speed, attention) and academic (reading com-
prehension, mathematics vocabulary, mathematics computation, mathematics facts retrieval)
skills. A total of 46 studies, which had measures of WPS as well as cognitive or academic
skills but did not report a direct bivariate correlation, were excluded. An additional seven
studies were excluded because they did not meet all the six quality indicators adapted from the
quality indicator for correlational research. In total, I identified k = 98 studies involving 112
independent samples.

Coding and Interrater Reliability

I followed the data-coding procedure specified by Cheung (2015) to facilitate the two-stage
MASEM analysis. In a spreadsheet using Microsoft Excel, each study was coded for the
following information: study name, year, publication type, socioeconomic status (SES),
sample disability status, measurement methods of word problems, sample size, grade level,
study quality, and available correlations between WPS and the nine predictor variables
(nonverbal reasoning, language comprehension, working memory, attention, processing speed,
mathematics facts retrieval, mathematics computation, mathematics vocabulary, and reading
comprehension).
Specifically, the publication type is coded as peer-reviewed manuscripts and others (dis-
sertation, report, and chapter). SES is coded as middle class or above, below middle class, and
unknown based on direct (e.g., middle class) and indirect (e.g., parental education level, free-
reduced lunch rate, income level) reports of SES from the original studies. Sample disability
status is coded as typically developing or atypically developing. For the measurement methods
Educational Psychology Review (2021) 33:1097–1124 1107

of word problems, I coded whether the word problem measure was in the oral model (tester
read the problem to students) or in the written model (students read the problem). I am
interested in the measurement methods of word problems because of their potential impact
on the predicting effects of involved predictors on WPS (Fuchs et al. 2006, 2018). Compared
to listening to word problem statements alone and solve the problem in their head, allowing the
child to use pencil and paper provides continuing access to the written problem narrative,
which transparently reduce the load on working memory. I coded grade level as early (K-2nd
grades) or intermediate (3rd–5th grades). When studies involved a mix of early and interme-
diate elementary graders, I determined whether they should be coded as early or intermediate
by examining the sample size for each grade level. Where correlations were missing from a
study, the cell was left blank. The two-stage SEM approach handles the missing correlations
efficiently with maximum likelihood estimation under the assumption of missing at random
(Cheung and Hong 2017).
I served as the first coder for all codes. I trained the second coder, a doctoral student in
educational psychology, to use the coding rubric, and resolved disagreements through discus-
sion for training purpose. Before coding the articles included in this study, the second coder
needs to reach 90% agreement on practice articles. I used the articles that did not meet
inclusion criteria for training purposes (e.g., articles included middle school samples). Fol-
lowing training, the second coder coded all studies for reliability purposes. Interrater agree-
ment calculating as [agreements (agreements + disagreements) × 100] was 95%. The mean
kappa was 0.95 (0.87–1.0) for the correlative matrices, 1.00 for publication type, 0.99 for SES,
0.97 for sample disability status, 0.93 for measurement methods of word problems, 0.99 for
sample size, 0.98 for grade level, and 0.94 for study quality. The discrepancies between the
two coders were solved through discussions regarding the original studies.

Methods of Analysis

The MASEM approach combines meta-analysis and SEM by testing hypothesized structural
equation models on the pooled correlation matrix (e.g., Viswesvaran and Ones 1995). How-
ever, most previous applications of MASEM have some issues (see Cheung and Chan 2005).
Based on Cheung and Chan (2005), the two-stage MASEM method is the first attempt to
incorporate meta-analytic techniques under the general framework of SEM. The two-stage
MASEM approach addresses issues by using the weighted least squares estimation to fit the
proposed models in stage 2 analysis. The present study adopted the two-stage MASEM
approach using the metaSEM package in R (Cheung 2015). Conducting the analysis involves
two stages. Stage 1 analysis is pooling correlation matrices together. I used fixed-effects model
for stage 1 analysis because of the small number of correlation coefficients related to some of
the comparisons. Stage 2 analysis is using the pooled correlation matrix to fit the hypothesized
structural equation models. For the completeness purpose, I also analyzed the same data with
one-stage MASEM technique for the full sample using age as a continuous moderator of the
regression coefficients.

Examined Partial Mediation Model

I examined the relative contributions of cognitive and academic predictors with a saturated
partial mediation model. Because in a saturated model, the number of observed variables is
equal to the number of parameters, and the saturated model has zero degrees of freedom and
1108 Educational Psychology Review (2021) 33:1097–1124

reproduces the observed correlation/variance-covariance matrix perfectly. Therefore, the mea-

sures of model fit cannot be applied.
In the saturated partial mediation model, I regressed WPS on cognitive skills (i.e., nonver-
bal reasoning, language comprehension, working memory, attention, processing speed) and
academic skills (i.e., reading comprehension, mathematics vocabulary, mathematics compu-
tation, mathematics facts retrieval). Additionally, the impacts of cognitive skills on WPS were
partially mediated by the involved academic skills.

Moderator Analyses Because the present study used the two-stage MASEM approach,
subgroup analysis is the only option to evaluate moderator effects (Jak and Cheung 2018).
The procedure for subgroup analysis involves separate stage 1 analyses for the subgroups. In
the second stage, the partial mediation model was then fit separately to these correlation
matrices from these separate stage 1 analyses. As noted by Jak and Cheung (2018), using
subgroup analysis to test moderation has its limitations, such as creating subgroups that may
lead to relatively small numbers of studies within each subgroup. Also, the moderator variables
have to be categorical or split the continuous variable into groups. It is also difficult to involve
more than one moderator in the one-stage MASEM (Jak and Cheung 2019). I only included
age as a continuous moderator in the one-stage MASEM analysis because it relates to my
research question.

Results

The present meta-analysis included a total of N = 111,084 participants from k = 98 studies. The
included samples have a median age of 8.29 years old. Most samples have an age that is
between 6.90 years old (25th percentile) and 9.77 years old (75th percentile), but some sample
has an age that is as low as 4.67 years old and as high as 11.73 years old. Table 2 shows the
results of the first stage of the two-stage SEM approach. Specifically, correlation coefficients
are below the diagonal, and the numbers of independent correlation coefficients are above the
diagonal. Correlations between WPS and predictor variables ranged from low to moderate
(0.22–0.65).
To investigate the unique predictors of performing word problems, I estimated the
hypothesized partial mediation model. The partial mediation model (see Fig. 1) assumed
that the effects of cognitive skills on WPS have been partially mediated by academic
skills. The results of the overall sample are presented in Fig. 3. See Appendix A for
direct and indirect effects. Among academic skills, mathematics computation (β = 0.22)
and mathematics vocabulary (β = 0.20) yielded significant paths to WPS. More impor-
tantly, several cognitive skills involving language comprehension, working memory, and
attention made unique impacts on WPS with the mediation effects of academic skills.
The standardized regression pathways from language comprehension (β = 0.14), working
memory (β = 0.17), and attention (β = 0.29) to WPS were significant and positive. With
other variables holding constant, for every unit increase in language comprehension,
working memory, and attention, WPS increased by 0.14, 0.17, and 0.29 SD units,
respectively. See Appendix D for one-stage MASEM results with no moderator and
with age as a continuous moderator. The omnibus test of the moderation effects was
statistically significant (p < 0.05).
Educational Psychology Review (2021) 33:1097–1124 1109

Table 2 Correlations and number of independent correlation coefficients for the full sample

Construct 1 2 3 4 5 6 7 8 9 10

1. WPS – 30 60 46 13 57 46 13 14 57
2. MFR 0.48 – 21 14 8 23 16 10 7 16
3. MC 0.61 0.58 – 25 9 38 27 11 10 31
4. RC 0.65 0.34 0.48 – 5 16 8 3 5 20
5. MV 0.64 0.52 0.56 0.51 – 7 7 3 6 4
6. NR 0.42 0.31 0.34 0.39 0.48 – 31 10 9 28
7. WM 0.34 0.29 0.31 0.37 0.34 0.33 – 10 7 21
8. PS 0.33 0.36 0.35 0.28 0.26 0.30 0.26 – 6 10
9. ATT 0.45 0.38 0.43 0.39 0.37 0.31 0.32 0.35 – 9
10. LC 0.50 0.22 0.38 0.55 0.48 0.37 0.34 0.26 .41 –

Correlations are below the diagonal; the number of independent correlation coefficients is above the diagonal
WPS word-problem solving, MFR mathematics facts retrieval, MC mathematics computation, RC reading
comprehension, MV mathematics vocabulary, NR nonverbal reasoning, WM working memory, PS processing
speed, ATT attention, LC language comprehension

Subgroup Analyses

I used grade level as a moderator during the fitting of the hypothesized partial mediation
model. Two groups were created: a group of younger students involving samples from early
elementary grade levels (K-2nd grades, average age = 6.52 years) and a group of older students
including students from intermediate elementary grades (3rd–5th grades, average age =
9.58 years). Additionally, I have coded publication type, SES, sample disability status, and
measurement methods of word problems, but I was not able to conduct further subgroup
analyses due to the lack of sufficient data for each subgroup.

Fig. 3 Partial mediation model for the full sample. All lines indicated significant paths (ps < 0.05). Dashed
pathway is not significant
1110 Educational Psychology Review (2021) 33:1097–1124

Younger Students

The correlation matrix for the younger students (N = 32,700; k = 47) is presented in Table 3.
Correlations ranged from low to moderate (0.20–0.62). The results of the partial mediation
model for younger students are presented in Fig. 4. First off, mathematics facts retrieval (β =
0.20) and mathematics computation (β = 0.08) demonstrated significant predicting paths to
WPS. Moreover, attention demonstrated to be a unique cognitive predictor of WPS with the
mediation of academic skills (β = 0.31). In other words, with other variables holding constant,
for every increase in attention, WPS increased by 0.31 SD units. See Appendix B for direct
and indirect effects.

Older Students

The correlation matrix for the older students (N = 78,384; k = 65) is contained in Table 4.
Correlations ranged from low to moderate (0.15–0.67). Based on the partial mediation model
for older students (see Fig. 5), mathematics vocabulary (β = 0.21) and mathematics computa-
tion (β = 0.24) yielded significant predicting paths to WPS. Moreover, nonverbal reasoning,
language comprehension, working memory, and attention demonstrated to be unique cognitive
predictors of WPS with the mediation effects of academic skills. With other variables holding
constant, for every increase in nonverbal reasoning, language comprehension, working mem-
ory, and attention, WPS increased by 0.15, 0.19, 0.31, and 0.30 SD units, respectively. See
Appendix C for direct and indirect effects.

Discussion

This study used a two-stage MASEM approach to investigate the unique predictors of WPS
with a hypothesized partial mediation model. Overall, language comprehension, working
memory, attention, mathematics vocabulary, and mathematics computation were unique
predictors of WPS. Subgroup analysis based on younger students suggested that attention
demonstrates a unique impact on WPS. In contrast, nonverbal reasoning, language

Table 3 Correlations and number of independent correlation coefficients for the younger students

Construct 1 2 3 4 5 6 7 8 9 10

1. WPS – 9 22 12 5 21 22 8 9 32
2. MFR 0.52 – 5 5 2 9 6 5 3 9
3. MC 0.62 0.55 – 5 3 16 13 6 6 17
4. RC 0.57 0.41 0.50 – 2 2 3 2 3 7
5. MV 0.57 0.44 0.40 0.50 – 2 5 2 3 2
6. NR 0.43 0.36 0.34 0.37 0.34 – 12 5 5 10
7. WM 0.35 0.39 0.30 0.37 0.30 0.33 – 6 5 12
8. PS 0.33 0.37 0.30 0.26 0.27 0.35 0.29 – 4 6
9. ATT 0.44 0.40 0.38 0.40 0.31 0.31 0.34 0.28 – 5
10. LC 0.51 0.20 0.37 0.50 0.40 0.26 0.30 0.28 0.39 –

Correlations are below the diagonal; the number of independent correlation coefficients is above the diagonal
WPS word-problem solving, MFR mathematics facts retrieval, MC mathematics computation, RC reading
comprehension, MV mathematics vocabulary, NR nonverbal reasoning, WM working memory, PS processing
speed, ATT attention, LC language comprehension
Educational Psychology Review (2021) 33:1097–1124 1111

Fig. 4 Partial mediation model for the younger students. All lines indicated significant paths (ps < 0.05). Dashed
pathway is not significant

comprehension, working memory, and attention have unique impacts on WPS among older
students. Moreover, as opposed to younger students, for whom mathematics facts retrieval and
mathematics computation were essential for WPS, older students rely on mathematics com-
putation and mathematics vocabulary to perform word problems. I first summarize the findings
for the cognitive and academic predictors separately and then consider implications, limita-
tions, and future directions.

Table 4 Correlations and number of independent correlation coefficients for the older students

Construct 1 2 3 4 5 6 7 8 9 10

1. WPS – 21 38 34 8 36 24 5 5 25
2. MFR 0.46 – 16 9 6 14 10 5 4 7
3. MC 0.61 0.58 – 20 6 22 14 5 4 14
4. RC 0.67 0.31 0.47 – 3 14 5 1 2 13
5. MV 0.66 0.55 0.61 0.50 – 5 2 1 3 2
6. NR 0.42 0.28 0.34 0.40 0.54 – 19 5 4 18
7. WM 0.34 0.23 0.31 0.38 0.45 0.33 – 4 2 9
8. PS 0.33 0.35 0.37 0.35 0.15 0.28 0.24 – 2 4
9. ATT 0.45 0.38 0.46 0.38 0.34 0.31 0.30 0.41 – 4
10. LC 0.50 0.24 0.39 0.59 0.59 0.41 0.37 0.24 0.42 –

Correlations are below the diagonal; the number of independent correlation coefficients is above the diagonal
WPS word-problem solving, MFR mathematics facts retrieval, MC mathematics computation, RC reading
comprehension, MV mathematics vocabulary, NR nonverbal reasoning, WM working memory, PS processing
speed, ATT attention, LC language comprehension
1112 Educational Psychology Review (2021) 33:1097–1124

Fig. 5 Partial mediation model for the older students. All lines indicated significant paths (ps < 0.05). Dashed
pathway is not significant

Unique Cognitive Predictors of Word-Problem Solving

Regarding cognitive skills, in addition to attention (a unique cognitive predictor for younger students
to solve word problems), nonverbal reasoning, language comprehension, and working memory
demonstrated unique predicting effects on WPS among older students. One possible explanation is
that as opposed to the one-step word problems at early grade levels that mostly require mathematics
computation skills for the solution, complicated multistep word problems at intermediate elementary
grades require problem comprehension before setting up numerical equations for the solution
(Kintsch and Greeno 1985). To solve word problems, older students need to understand the word
problem statement, exclude irrelevant information, infer information not evident in the problem
statement, categorize a word problem as a specific problem type, generalize learned problems to
novel problem situations, and adopt appropriate solution strategies, which rely on language
comprehension, working memory, as well as nonverbal reasoning (Fuchs et al. 2015, 2018).
Nonverbal reasoning is only revealed as a unique cognitive predictor for older students to
perform word problems, as opposed to the negligible impact of nonverbal reasoning for
younger students in WPS. There are several possible explanations. As word problems become
accumulatively complicated at intermediate elementary grade levels, nonverbal reasoning
becomes particularly essential for students to perform more advanced problem solving pro-
cesses, such as inferring information not evident in the problem statement and categorizing a
word problem as a specific problem type (Wang et al. 2016). Another possibility is that formal
education has a potential effect on the development of nonverbal reasoning. Specifically, the
constant use of nonverbal reasoning in performing increasingly complex mathematics tasks
(i.e., word problems) serves as a potential training of nonverbal reasoning (Martinez 2000). A
prior longitudinal study also suggested that nonverbal reasoning is a significant predictor of
Educational Psychology Review (2021) 33:1097–1124 1113

age-related differences in WPS accuracy controlling for mathematics computation, reading

comprehension, phonological processing, and working memory (Swanson 2004).
Concerning the role of language comprehension for performing word problems, the present
study extended the current literature by including a comprehensive set of academic mediators.
Although previous research included different academic mediators in their path models (e.g.,
Fuchs et al. 2006, 2015; Peng and Lin 2019; Singer et al. 2018), language comprehension has
been documented as a unique cognitive predictor of WPS. For example, Fuchs et al. (2006)
included mathematics computation and mathematics facts retrieval as academic mediators in
their path models and suggested that language comprehension demonstrates a unique impact
on WPS. With the inclusion of mathematics vocabulary as a mediator, Fuchs et al. (2015) and
Peng and Lin (2019) also revealed that language comprehension yielded significant predicting
effect on WPS.
Working memory is also shown as a unique cognitive predictor of performing word
problems overall and among older students. Swanson (2004) is the only empirical research,
according to my knowledge, that explored cognitive and academic predictors for the age-
related differences in WPS. He suggested that working memory was a significant predictor of
the age-related changes in problem solving performances controlling for other cognitive (i.e.,
nonverbal reasoning) and academic (i.e., mathematics computation, reading comprehension,
phonological processing) skills. The present study determined, however, that working memory
is a unique cognitive skill for performing word problems accounting for the relationships
among cognitive and academic skills. This finding, therefore, provides further evidence
showing that working memory does not merely being predictive of the age-related difference
in WPS performances but is also a unique cognitive predictor for performing word problems
among students in intermediate elementary grades.
Previous cross-sectional studies could also shed some light on the role of working memory
for WPS. Inconsistent findings remain in the past literature regarding the contribution of
working memory on WPS, which may highlight some critical issues regarding the role of
working memory in performing word problems. First, the mixed findings may be related to the
domains of working memory being measured. According to the domain-specific models of
working memory, the operation of working memory depends on the domain knowledge and,
therefore, is strongly affected by domain specificity (Ericsson and Kintsch 1995). For example,
Fuchs et al. (2018) differentiated types of working memory and suggested that verbal working
memory has a unique impact on WPS, whereas numerical working memory does not. Simply
put, working memory may be domain-specific in performing word problems. Only working
memory tasks involving verbal processing are critical for WPS. Additionally, given that some
word problem measures only require one-digit addition or subtraction, the mixed finding may
also relate to the solution complexity of the word problems. Following this logic, it is not
surprising that previous research (e.g., Fuchs et al. 2006; Peng and Lin 2019) revealed no
significant effect of working memory on WPS.
Moreover, compared to most prior studies that used specific measures to assess working
memory, Fung and Swanson (2017) suggested that types of working memory (storage
component versus executive component) would also influence the impact of working memory
on WPS. Specifically, as opposed to the unique effect of the storage component of working
memory (visuospatial sketchpad, phonological loop) on WPS, the effect of the executive
component (central executive) on WPS is fully mediated by reading, calculation, and nonver-
bal reasoning. Given that the majority of previous literature relied on a single task or a
composite score to measure working memory, the present study could not further investigate
1114 Educational Psychology Review (2021) 33:1097–1124

if the impacts of working memory on WPS would change according to working memory
components.
Attention was demonstrated as a unique predictor of WPS across younger and older
students. Prior studies involving attention in their model (Fuchs et al. 2006, 2016) also
demonstrated that attention was the most robust predictor of WPS with the mediation roles
of mathematics facts retrieval and mathematics computation. There are several possible
explanations. The ability to focus attention creates the opportunity to persevere with academic
tasks that require serial execution (i.e., WPS; Luria, 1980). Another possibility is that
classroom instruction may fail to address the needs of students with a weak mathematics
knowledge base, and this mismatch creates inattentive behavior, which makes it even hard for
struggling students to catch up with their peers. Additionally, prior intervention research also
demonstrated that responsiveness to mathematics intervention was associated with attentive
behavior across all mathematics outcomes (Fuchs et al. 2005). The present study further
illustrates the importance of several instructional ideas, which aim at increasing students’
attention within word problem interventions (i.e., self-monitoring, small group instruction;
Jitendra et al. 2013). Theoretically, small groups provide natural settings for interpersonal
monitoring and regulating students’ goal-directed behaviors (Linnenbrink-Garcia et al. 2011).

Unique Academic Predictors of Word-Problem Solving

Older and younger students also demonstrated different academic skills essential for solving
word problems. As opposed to younger students, who mainly rely on computation skills
(mathematics facts retrieval, mathematics computation) to perform word problems, older
students rely on mathematics vocabulary and mathematics computation to solve word
problems.
The present study illustrated the importance of mathematics facts retrieval for WPS for
younger students, which is in line with previous research (Fuchs et al. 2006). Moreover,
fluency with foundational academic knowledge (i.e., mathematics facts retrieval) could help
efficiently allocate cognitive resources to solve more complex problems (Geary et al. 2008;
Kim et al. 2011). Unfamiliarity with mathematics facts would lead to more cognitive resources
involved in the mathematics computation process, thus resulting in fewer cognitive resources
available for the thinking and planning processes for WPS (Geary et al. 2008; Kim et al. 2011).
Additionally, this finding corroborates the importance of intensive early intervention for
fluency with single-digit addition and subtraction (Burns 2005).
In terms of mathematics computation, in contrast to Fuchs et al. (2006), the present study
demonstrated that mathematics computation is an essential predictor of WPS across younger
and older students. One possible explanation is that Fuchs et al. (2006) used simple arithmetic
word problems involving sums or minuends of 9 or less, which transparently require less
mathematics computation skills. In contrast, the current study includes simple one-step as well
as complicated multistep word problems involving complex operations. Moreover, mathemat-
ics computation (skill in manipulating numbers procedurally) enhances understanding about
numerical relationships theoretically, and this, in turn, may facilitate WPS (e.g., Gilbert and
Fuchs 2017).
The present study demonstrated the importance of mathematics vocabulary for WPS among
older students. According to Powell et al. (2017), the complexity of mathematics vocabulary
increases as students progress in school. Therefore, it is not surprising that mathematics
vocabulary is a unique academic predictor for older students to perform word problems.
Educational Psychology Review (2021) 33:1097–1124 1115

Rather than a proxy measure for language comprehension and mathematical knowledge
(Purpura et al. 2017), mathematics vocabulary is a unique construct important for WPS. This
finding is in line with previous research (e.g., Fuchs et al. 2015; Peng and Lin 2019), which
suggested that mathematics vocabulary has a significant impact on WPS controlling for other
critical cognitive skills.
Moreover, it is worth noting that reading comprehension did not emerge as a unique
academic predictor for WPS. One possible explanation is that most studies involved in the
present meta-analysis used word problems presented through oral models to control for
individual differences in decoding skills (e.g., Boonen et al. 2013; Fuchs et al. 2006).
Admittedly, even when presented through oral models, reading comprehension capacity could
still benefit students because it provides students with continuous access to written problem
narrative after the tester completed reading the problem (Fuchs et al. 2006). Yet, students could
always require the tester to read the problem statements twice by raising their hand, which, in
term, may decrease the importance of reading comprehension.

Limitations

A primary limitation of the two-stage MASEM approach is the difficulty of controlling for
study characteristics (i.e., SES, age of the sample). Conducting subgroup analyses based on
study characteristics is the only method of moderation analysis or controlling study charac-
teristics (Cheung and Chan 2005; Jak and Cheung 2018). I coded the type of publication, SES,
sample disability status, and measurement methods of word problems for all studies. Still, I
was not able to include them as moderators in the MASEM analyses given the insufficient data
for each subgroup. For example, due to the lack of word problems presented through written
models, the present research cannot investigate whether the formats of word problems would
affect the unique predictors of word problems.
Moreover, the numbers of correlation coefficients for several relationships included in the
meta-analysis are small (e.g., the correlation between mathematics vocabulary and processing
speed in Table 2), which may influence the findings to some extent. Finally, the present study
was based upon a secondary analysis of a meta-analytic dataset. Therefore, the range of
variables included in past research on WPS would limit the scope of the present study.
Although prior research suggested that the effect of working memory on WPS may differ
by different working memory components (Fung and Swanson 2017), the insufficient number
of studies that separate working memory components limits the present study to investigate the
issue further.

Implications

These findings may have important implications for theory and, by extension, practice.
From a theoretical standpoint, the current study provides evidence that the unique
predictors of WPS may change according to students’ grade levels, with attention and
mathematics computation demonstrating importance across younger and older students.
These findings also shed some light on whether integrating cognitive training (i.e.,
working memory training) with WPS instruction can create synergy for promoting
stronger WPS performance. Given that higher-order cognitive skills (i.e., nonverbal
reasoning, working memory) become unique cognitive predictors of WPS among older
students, assisting them in reducing cognitive deficits becomes essential. Future word
1116 Educational Psychology Review (2021) 33:1097–1124

problem instruction, specially designed for the intermediate elementary grades, needs to
pay more attention to promote reasoning using schemas and scaffold language compre-
hension within WPS instruction (Powell and Fuchs 2018; Fuchs et al. 2020).
Practically, assisting students in accumulating academic knowledge is also critical for word
problem instruction. As learning experience accumulates, academic skills become consolidated in
long-term memory, and students develop to rely more on direct retrieval of academic knowledge
from long-term memory to perform WPS. Therefore, explicit instruction on academic skills (i.e.,
mathematics vocabulary, mathematics facts retrieval) facilitates flexibility in allocating cognitive
resources (Kearns and Fuchs 2013). Furthermore, the present findings demonstrated that the focus of
word problem instruction needs to switch with development as the specific academic skills essential
for performing word problems change with development. As opposed to younger students, for
whom mathematics facts retrieval is necessary, older students need to understand mathematics
vocabulary to solve word problems.

Directions for Future Research

The present study provides a potential framework to examine the unique predictors of WPS. Future
research examining the unique cognitive predictors of WPS needs to account for essential academic
mediators. Additionally, future research investigating the unique cognitive predictors of WPS needs
to consider the hierarchical nature of the cognitive predictors (e.g., Swanson and Sachse-Lee 2001).
In other words, future research could differentiate high-level (nonverbal reasoning, working mem-
ory) and low-level (processing speed) cognitive predictors, with low-level cognitive predictors as the
foundation for more advanced high-level cognitive predictors in the path model. Second, there is the
need to continue investigating the unique predictors of performing word problems presented through
different models (oral versus written), among different populations (students with and without
learning difficulties), as well as at different elementary grade levels. According to my knowledge,
there is no empirical research that directly examines the cognitive and academic skills required for
solving word problems when presented through different models (oral versus written).
Third, future empirical research could differentiate different components (executive versus
storage; Fung and Swanson 2017) or different types of working memory (numerical versus
verbal) while examining the predicting effect of working memory on WPS. Last, future
experimental research is needed because most empirical studies quantitatively summarized
in the present study are cross-sectional and, thus, do not allow inferences on causality.
Specifically, future WPS intervention research could manipulate the unique predictors of
WPS (i.e., attention) identified in the present study and explore its contribution to students’
responsiveness to intervention.

Acknowledgements The author would like to thank Sarah Powell at University of Texas at Austin and
Jiangang Zeng at Louisiana State University for helpful discussions.

Compliance with Ethical Standards

APA ethical standards were followed in the conduct of the study. With the submission of this manuscript, I
would like to undertake that this manuscript has not been published elsewhere, accepted for publication
elsewhere or under editorial review for publication elsewhere.
Educational Psychology Review (2021) 33:1097–1124 1117

Conflict of Interest The author declares that he has no conflict of interest.

References

References marked with an asterisk indicate studies included in the meta-analysis.

*Abrahamson, C. A. (1995). Reading disabilities and the third R: the relationship between reading and
arithmetic performance in reading disabled children (Publication No. 9504121) [Doctoral dissertation,
American International College]. ProQuest Dissertations and Theses Global.
Awh, E., Vogel, E. K., & Oh, S. H. (2006). Interactions between attention and working memory. Neuroscience,
139(1), 201–208. https://doi.org/10.1016/j.neuroscience.2005.08.023.
*Bae, Y. S., Chiang, H. M., & Hickson, L. (2015). Mathematical word problem solving ability of children with
autism spectrum disorder and their typically developing peers. Journal of Autism and Developmental
Disorders, 45(7), 2200–2208. https://doi.org/10.1007/s10803-015-2387-8.
*Becker, D. R., Miao, A., Duncan, R., & McClelland, M. M. (2014). Behavioral self-regulation and executive
function both predict visuomotor skills and early academic achievement. Early Childhood Research
Quarterly, 29(4), 411–424. https://doi.org/10.1016/j.ecresq.2014.04.014
Berg, D. H. (2008). Working memory and arithmetic calculation in children: The contributory roles of processing
speed, short-term memory, and reading. Journal of Experimental Child Psychology, 99(4), 288–308.
https://doi.org/10.1016/j.jecp.2007.12.002.
*Bjork, I. M., & Bowyer-Crane, C. (2013). Cognitive skills used to solve mathematical word problems and
numerical operations: a study of 6-to 7-year-old children. European Journal of Psychology of Education,
28(4), 1345–1360. https://doi.org/10.1007/s10212-012-0169-7.
Björn, P. M., Aunola, K., & Nurmi, J. E. (2016). Primary school text comprehension predicts mathematical word
problem-solving skills in secondary school. Educational Psychology, 36(2), 362–377. https://doi.
org/10.1080/01443410.2014.992392.
*Blums, A., Belsky, J., Grimm, K., & Chen, Z. (2017). Building links between early socioeconomic status,
cognitive ability, and math and science achievement. Journal of Cognition and Development, 18(1), 16–40.
https://doi.org/10.1080/15248372.2016.1228652.
Boonen, A. J., van der Schoot, M., van Wesel, F., de Vries, M. H., & Jolles, J. (2013). What underlies successful
word problem solving? A path analysis in sixth grade students. Contemporary Educational Psychology,
38(3), 271–279. https://doi.org/10.1016/j.cedpsych.2013.05.001.
*Brenner, R. A. (1981). Elementary school students’ abilities to read and solve arithmetic word problems: a
study of prerequisite skills (Unpublished doctoral dissertation). University of California.
*Brock, H. (1982). Factor structure of intellectual and achievement measures for learning disabled children.
Psychology in the Schools, 19(3), 297–304. https://doi.org/10.1002/1520-6807(198207)19:3%3C297::AID-
PITS2310190306%3E3.0.CO;2-R.
Burns, M. K. (2005). Using incremental rehearsal to increase fluency of single-digit multiplication facts with
children identified as learning disabled in mathematics computation. Education and Treatment of Children,
28(3), 237–249.
Case, R. (1985). Intellectual development: Birth to adulthood. New York: Academic Press.
Cattell, R. B. (1963). Theory of fluid and crystallized intelligence: A critical experiment. Journal of Educational
Psychology, 54(1), 1–22. https://doi.org/10.1037/h0046743.
Cattell, R. B. (1987). Intelligence: Its structure, growth and action. Amsterdam: North-Holland.
*Cavanaugh, A. L. (1992). Computers, self-efficacy, and primary children [Unpublished doctoral dissertation].
The University of Connecticut.
Cheung, M. W.-L. (2015). metaSEM: An R package for meta-analysis using structural equation modeling.
Frontiers in Psychology, 5, 1521. https://doi.org/10.3389/fpsyg.2014.01521.
Cheung, M. W. L., & Chan, W. (2005). Meta-analytic structural equation modeling: a two-stage approach.
Psychological Methods, 10(1), 40–64. https://psycnet.apa.org/doi/10.1037/1082-989X.10.1.40.
Cheung, M. W. L., & Hong, R. Y. (2017). Applications of meta-analytic structural equation modelling in health
psychology: examples, issues, and recommendations. Health Psychology Review, 11(3), 265–279.
https://doi.org/10.1080/17437199.2017.1343678.
*Chia, N.K.H., & Kho, C.P. (2011). An investigative study on the learning difficulties in mathematics
encountered by primary 4 children: in search of a cognitive equation for mathematics learning. Journal of
the American Academy of Special Education Professionals, Winter (1), 93–119.
Chiaravalloti, N. D., & DeLuca, J. (2008). Cognitive impairment in multiple sclerosis. The Lancet Neurology,
7(12), 11399–11151. https://doi.org/10.1016/S1474-4422(08)70259-X.
1118 Educational Psychology Review (2021) 33:1097–1124

*Chung, K. K., Liu, H., McBride, C., Wong, A. M. Y., & Lo, J. C. (2017). How socioeconomic status, executive
functioning and verbal interactions contribute to early academic achievement in Chinese children.
Educational Psychology, 37(4), 4029–420. https://doi.org/10.1080/01443410.2016.1179264.
*Cirino, P. T., Child, A. E., & Macdonald, K. T. (2018). Longitudinal predictors of the overlap between reading
and math skills. Contemporary Educational Psychology, 54(2), 99–111. https://doi.org/10.1016/j.
cedpsych.2018.06.002.
*Collis, B., Ollila, L. O., & Yore, L. D. (1986). Predictive validity of the Canadian readiness test and selected
measures of cognitive development for grade one reading and mathematics achievement. Alberta Journal of
Educational Research, 32(1), 2–11.
Common Core State Standards Initiative. (2010). Common Core State Standards for English language arts &
literacy in history/social studies, science, and technical subjects. Washington, DC: CCSSO & National
Governors Association.
Cooper, G., & Sweller, J. (1987). Effects of schema acquisition and rule automation on mathematical problem-
solving transfer. Journal of Educational Psychology, 79(4), 347–362.
*Cowan, R., & Powell, D. (2014). The contributions of domain-general and numerical factors to third-grade
arithmetic skills and mathematical learning disability. Journal of Educational Psychology, 106(1), 214–229.
https://doi.org/10.1037/a0034097
Daneman, M., & Carpenter, P. A. (1980). Individual differences in working memory and reading. Journal of
Verbal Learning and Verbal Behavior, 19(4), 450–466. https://doi.org/10.1016/S0022-5371(80)90312-6.
Daroczy, G., Wolska, M., Meurers, W. D., & Nuerk, H. C. (2015). Word problems: a review of linguistic and
numerical factors contributing to their difficulty. Frontiers in Psychology, 6, 348. https://doi.org/10.3389
/fpsyg.2015.00348.
*Day, S. L., & Connor, C. M. (2017). Examining the relations between self-regulation and achievement in third-
grade students. Assessment for Effective Intervention, 42(2), 97–109. https://doi.org/10.1177
/1534508416670367.
*Diamantopoulou, S., Pina, V., Valero-Garcia, A. V., González-Salinas, C., & Fuentes, L. J. (2012). Validation
of the spanish version of the Woodcock-Johnson mathematics achievement tests for children aged 6 to 13.
Journal of Psychoeducational Assessment, 30(5), 466–477. 10.1177/0734282912437531.
*Dilworth-Bart, J. E. (2012). Does executive function mediate SES and home quality associations with academic
readiness? Early Childhood Research Quarterly, 27(3), 416–425. https://doi.org/10.1016/j.
ecresq.2012.02.002.
*Donker, M., Kroesbergen, E., Slot, E., Van Viersen, S., & De Bree, E. (2016). Alphanumeric and non-
alphanumeric rapid automatized naming in children with reading and/or spelling difficulties and mathemat-
ical difficulties. Learning and Individual Differences, 47(3), 80–87. https://doi.org/10.1016/j.
lindif.2015.12.011.
*Downer, J. T., & Pianta, R. C. (2006). Academic and cognitive functioning in first grade: associations with
earlier home and child care predictors and with concurrent home and classroom experiences. School
Psychology Review, 35(1), 11–30.
*Dupere, V., Leventhal, T., Crosnoe, R., & Dion, E. (2010). Understanding the positive role of neighborhood
socioeconomic advantage in achievement: the contribution of the home, child care, and school environ-
ments. Developmental Psychology, 46(5), 1227–1244. https://doi.org/10.1037/a0020211.
Ericsson, K. A., & Kintsch, W. (1995). Long-term working memory. Psychological Review, 102(2), 211–245.
https://doi.org/10.1037/0033-295X.102.2.211.
Evans, J. S. B., & Stanovich, K. E. (2013). Dual-process theories of higher cognition: Advancing the debate.
Perspectives on Psychological Science, 8(3), 223–241. https://doi.org/10.1177/2F1745691612460685.
Every Child a Chance Trust (2009). The long-term costs of numeracy difficulties. Retrieved. August 14, 2009.
http://www.everychildachancetrust.org/counts/index.cfm.
Fuchs, L. S., Compton, D. L., Fuchs, D., Paulsen, K., Bryant, J. D., & Hamlett, C. L. (2005). The prevention,
identification, and cognitive determinants of math difficulty. Journal of Educational Psychology, 97(3),
493–513. https://doi.org/10.1037/0022-0663.97.3.493.
Fuchs, L. S., Fuchs, D., Compton, D. L., Powell, S. R., Seethaler, P. M., Capizzi, A. M., et al. (2006). The
cognitive correlates of third-grade skill in arithmetic, algorithmic computation, and arithmetic word prob-
lems. Journal of Educational Psychology, 98(1), 29–43 http://psycnet.apa.org/doi/10.1037/0022-
0663.98.1.29.
Fuchs, L. S., Fuchs, D., Stuebing, K., Fletcher, J. M., Hamlett, C. L., & Lambert, W. (2008). Problem solving
and computational skill: Are they shared or distinct aspects of mathematical cognition? Journal of
Educational Psychology, 100(1), 30–47. http://psycnet.apa.org/doi/10.1037/0022-0663.100.1.30.
Fuchs, L. S., Geary, D. C., Compton, D. L., Fuchs, D., Hamlett, C. L., Seethaler, P. M., et al. (2010). Do different
types of school mathematics development depend on different constellations of numerical versus general
Educational Psychology Review (2021) 33:1097–1124 1119

cognitive abilities? Developmental Psychology, 46(6), 1731–1746. https://psycnet.apa.org/doi/10.1037

/a0020662.
*Fuchs, L. S., Compton, D. L., Fuchs, D., Powell, S. R., Schumacher, R. F., Hamlett, C. L., Vernier E.,
Namkung J. M. & Vukovic, R. K. (2012). Contributions of domain-general cognitive resources and different
forms of arithmetic development to pre-algebraic knowledge. Developmental Psychology, 48(5), 1315–
1326. https://doi.org/10.1037/0027475
Fuchs, L. S., Fuchs, D., Compton, D. L., Hamlett, C. L., & Wang, A. Y. (2015). Is word-problem solving a form
of text comprehension? Scientific Studies of Reading, 19(3), 204–223. https://doi.org/10.1080
/10888438.2015.1005745.
Fuchs, L. S., Geary, D. C., Fuchs, D., Compton, D. L., & Hamlett, C. L. (2016). Pathways to third-grade
calculation versus word-reading competence: Are they more alike or different? Child Development, 87(2),
558–567. https://doi.org/10.1111/cdev.12474.
Fuchs, L. S., Gilbert, J. K., Fuchs, D., Seethaler, P. M., Martin, N., & B. (2018). Text comprehension and oral
language as predictors of word-problem solving: Insights into word-problem solving as a form of text
comprehension. Scientific Studies of Reading, 22(2), 152–166. https://doi.org/10.1080
/10888438.2017.1398259.
Fuchs, L., Fuchs, D., Seethaler, P. M., & Barnes, M. A. (2020). Addressing the role of working memory in
mathematical word-problem solving when designing intervention for struggling learners. ZDM, 52(1), 87–
96. https://doi.org/10.1007/s11858-019-01070-8.
Fung, W., & Swanson, H. L. (2017). Working memory components that predict word problem solving: Is it
merely a function of reading, calculation, and fluid intelligence? Memory & Cognition, 45(5), 804–823.
https://doi.org/10.3758/s13421-017-0697-0.
*Fung, W. W., Swanson, H. L., & Orosco, M. J. (2014). Influence of reading and calculation on children at risk
and not at risk for word problem solving: is math motivation a mediator?. Learning and Individual
Differences, 36(12), 84–91. https://doi.org/10.1016/j.lindif.2014.10.011.
*Gardner-Neblett, N., DeCoster, J., & Hamre, B. K. (2014). Linking preschool language and sustained attention
with adolescent achievement through classroom self-reliance. Journal of Applied Developmental
Psychology, 35(6), 457–467. https://doi.org/10.1016/j.appdev.2014.09.003.
Geary, D. C. (1993). Mathematical disabilities: cognitive, neuropsychological, and genetic components.
Psychological Bulletin, 114(2), 345–362 https://psycnet.apa.org/doi/10.1037/0033-2909.114.2.345.
Geary, D. C., Bow-Thomas, C. C., & Yao, Y. (1992). Counting knowledge and skill in cognitive addition: A
comparison of normal and mathematically disabled children. Journal of Experimental Child Psychology,
54(3), 372–391. https://doi.org/10.1016/0022-0965(92)90026-3.
Geary, D. C., Boykin, A. W., Embretson, S., Reyna, V., Siegler, R., Berch, D. B., & Graban, J. (2008). Chapter
4: Report of the task group on learning processes. National Mathematics Advisory Panel, Foundations for
success: The final report of the National Mathematics Advisory Panel. Washington, DC: US Department of
Education.
Gersten, R., Chard, D. J., Jayanthi, M., Baker, S. K., Morphy, P., & Flojo, J. (2009). Mathematics instruction for
students with learning disabilities: A meta-analysis of instructional components. Review of Educational
Research, 79(3), 1202–1242. https://doi.org/10.3102/0034654309334431.
Gilbert, J. K., & Fuchs, L. S. (2017). Bivariate developmental relations between calculations and word problems:
A latent change approach. Contemporary Educational Psychology, 51(4), 83–98. https://doi.org/10.1016/j.
cedpsych.2017.06.008.
Glaser, R. (1984). Education and thinking: The role of knowledge. American Psychologist, 39(2), 93–104.
https://doi.org/10.1037/0003-066X.39.2.93.
Gottfredson, L. S. (1997). Why g matters: The complexity of everyday life. Intelligence, 24(1), 79–132.
https://doi.org/10.1016/S0160-2896(97)90014-3.
*Griffin, C. C., & Jitendra, A. K. (2009). Word problem-solving instruction in inclusive third-grade mathematics
classrooms. The Journal of Educational Research, 102(3), 187–202. https://doi.org/10.3200
/JOER.102.3.187-202.
*Grossman, F. M., & Johnson, K. M. (1983). Validity of the Slosson and Otis-Lennon in predicting achievement
of gifted students. Educational and Psychological Measurement, 43(2), 617–622. https://doi.org/10.1177
/001316448304300235.
*Guthmiller, M. L. (2010). Identifying factors predicting the math computation and math reasoning performance
levels of Latino elementary school students (Publication No. 3407501) [Doctoral dissertation, the University
of Utah]. ProQuest Dissertations and Theses Global.
*Hart, S. A., Petrill, S. A., Thompson, L. A., & Plomin, R. (2009). The ABCs of math: a genetic analysis of
mathematics and its links with reading ability and general cognitive ability. Journal of Educational
Psychology, 101(2), 388–402. https://doi.org/10.1037/a0015115.
1120 Educational Psychology Review (2021) 33:1097–1124

Hecht, S. A., Torgesen, J. K., Wagner, R. K., & Rashotte, C. A. (2001). The relations between phonological
processing abilities and emerging individual differences in mathematical computation skills: A longitudinal
study from second to fifth grades. Journal of Experimental Child Psychology, 79(2), 192–227. https://doi.
org/10.1006/jecp.2000.2586.
Hegarty, M., Mayer, R. E., & Monk, C. A. (1995). Comprehension of arithmetic word problems: A comparison
of successful and unsuccessful problem solvers. Journal of Educational Psychology, 87(1), 18–32.
https://doi.org/10.1037/0022-0663.87.1.18.
Heitz, R. P., Unsworth, N., & Engle, R. W. (2005). Working memory capacity, attention control, and fluid
intelligence. In Handbook of understanding and measuring intelligence (pp. 61–77).
Hembree, R. (1992). Response to critique of meta-analysis. Journal for Research in Mathematics Education,
23(3), 284–289. https://doi.org/10.2307/749122.
Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Perrig, W. J. (2008). Improving fluid intelligence with training on
working memory. Proceedings of the National Academy of Sciences, 105(19), 6829–6833. https://doi.
org/10.1073/pnas.0801268105.
Jak, S., & Cheung, M. W. L. (2018). Testing moderator hypotheses in meta-analytic structural equation modeling
using subgroup analysis. Behavior Research Methods, 50(4), 1359–1373. https://doi.org/10.3758/s13428-
018-1046-3.
Jak, S., & Cheung, M. W. L. (2019). Meta-analytic structural equation modeling with moderating effects on SEM
parameters. Psychological Methods. Advance online publication. https://doi.org/10.1037/met0000245
*Jensen, A. R., & Reynolds, C. R. (1983). Sex differences on the WISC-R. Personality and Individual
Differences, 4(2), 223–226. https://doi.org/10.1016/0191-8869(83)90029-6
Jitendra, A. K., Rodriguez, M., Kanive, R., Huang, J. P., Church, C., Corroy, K. A., & Zaslofsky, A. (2013).
Impact of small-group tutoring interventions on the mathematical problem solving and achievement of third-
grade students with mathematics difficulties. Learning Disability Quarterly, 36(1), 21–35. https://doi.
org/10.1177/2F0731948712457561.
*Johns, S. K., Valiente, C., Eisenberg, N., Spinrad, T. L., Hernández, M. M., Southworth, J., Berger, R. H.,
Thompson, M. S., Silva, K. M., & Pina, A. A. (2019). Prediction of children’s early academic adjustment
from their temperament: the moderating role of peer temperament. Journal of Educational Psychology,
111(3), 542–555. https://doi.org/10.1037/edu0000288.
*Jordan, J. A., Wylie, J., & Mulhern, G. (2015). Mathematics and reading difficulty subtypes: minor phonolog-
ical influences on mathematics for 5–7-years-old. Frontiers in Psychology, 6, Article 221. https://doi.
org/10.3389/fpsyg.2015.00221.
*Kang, C. (1990). The nature and development of a cognitive screening battery for academically gifted second
and third grade students [Unpublished doctoral dissertation]. University of Wisconsin-Madison.
Kearns, D. M., & Fuchs, D. (2013). Does cognitively focused instruction improve the academic performance of
low-achieving students? Exceptional Children, 79(3), 263–290. https://doi.org/10.1177
/2F001440291307900200.
*Kempert, S., Saalbach, H., & Hardy, I. (2011). Cognitive benefits and costs of bilingualism in elementary
school students: the case of mathematical word problems. Journal of Educational Psychology, 103(3), 547–
561. https://doi.org/10.1037/a0023619
Kim, Y. S., Wagner, R. K., & Foster, E. (2011). Relations among oral reading fluency, silent reading fluency, and
reading comprehension: A latent variable study of first-grade readers. Scientific Studies of Reading, 15(4),
338–362. https://doi.org/10.1080/10888438.2010.493964.
Kintsch, W., & Greeno, J. G. (1985). Understanding and solving word arithmetic problems. Psychological
Review, 92(1), 109–129. https://doi.org/10.1037//0033-295x.92.1.109.
*Kovacic, C. B. (1998). Construct validity of a math performance assessment project (Publication No. 9911740)
[Doctoral dissertation, the University of New Mexico]. ProQuest Dissertations and Theses Global.
*Kyttälä, M., Aunio, P., Lepola, J., & Hautamäki, J. (2014). The role of the working memory and language skills
in the prediction of word problem solving in 4-to 7-year-old children. Educational Psychology, 34(6), 674–
696. https://doi.org/10.1080/01443410.2013.814192.
Lai, E. R. (2011). Critical thinking: A literature review. Retrieved from http://www.pearsonassessments.
com/hai/images/tmrs/CriticalThinkingReviewFINAL.pdf
*Lin, X., Peng, P., & Luo, H. (in press). The deficit profile of elementary students with computational difficulties
versus word problem-solving difficulties. Learning Disability Quarterly.
Linnenbrink-Garcia, L., Rogat, T. K., & Koskey, K. L. (2011). Affect and engagement during small group
instruction. Contemporary Educational Psychology, 36(1), 13–24. https://doi.org/10.1016/j.
cedpsych.2010.09.001.
*Majumder, S. (2004). Factors in mathematical word problem solving: the role of inhibition [Unpublished
doctoral dissertation]. York University.
Educational Psychology Review (2021) 33:1097–1124 1121

*Männamaa, M., Kikas, E., Peets, K., & Palu, A. (2012). Cognitive correlates of math skills in third-grade
students. Educational Psychology, 32(1), 21–44. https://doi.org/10.1080/01443410.2011.621713.
*Martin, R. B., Cirino, P. T., Sharp, C., & Barnes, M. (2014). Number and counting skills in kindergarten as
predictors of grade 1 mathematical skills. Learning and Individual Differences, 34(6), 12–23. https://doi.
org/10.1016/j.lindif.2014.05.006.
Martinez, M. E. (2000). Education as the cultivation of intelligence. Mahwah, NJ: Erlbaum.
*Matthews, J. S., Ponitz, C. C., & Morrison, F. J. (2009). Early gender differences in self-regulation and
academic achievement. Journal of Educational Psychology, 101(3), 689–704. https://doi.org/10.1037
/a0014240.
Miller, L. T., & Vernon, P. A. (1997). Developmental changes in speed of information processing in young
children. Developmental Psychology, 33(3), 549–554. https://doi.org/10.1037/0012-1649.33.3.549.
Monroe, E. E., & Orme, M. P. (2002). Developing mathematical vocabulary. Preventing School Failure, 46(3),
139–142. https://doi.org/10.1080/10459880209603359.
Moschkovich, J. (2013). Principles and guidelines for equitable mathematics teaching practices and materials for
English language learners. Journal of Urban Mathematics Education, 6(1), 45–57.
Moschkovich, J. N., Wagner, D., Bose, A., Rodrigues Mendes, J., & Schütte, M. (2018). Language and
communication in mathematics education : International perspectives. Cham: Springer. https://doi.
org/10.1007/978-3-319-75055-2.
*Mullins, J. E. (1999). A comparison of performance of students referred for gifted evaluation on the WISC-III
and Binet IV (Publication No. 3002989) [Doctoral dissertation, West Virginia University]. ProQuest
Dissertations and Theses Global.
Murnane, R. J., Willett, J. B., Braatz, M. J., & Duhaldeborde, Y. (2001). Do different dimensions of male high
school students' skills predict labor market success a decade later? Evidence from the NLSY. Economics of
Education Review, 20(4), 311–320. https://doi.org/10.1016/S0272-7757(00)00056-X.
*Murrihy, C., Bailey, M., & Roodenburg, J. (2017). Psychomotor ability and short-term memory, and reading
and mathematics achievement in children. Archives of Clinical Neuropsychology, 32(5), 618–630.
https://doi.org/10.1093/arclin/acx033.
Namkung, J. M., Peng, P., & Lin, X. (2019). The relation between mathematics anxiety and mathematics
performance among school-aged students: a meta-analysis. Review of Educational Research, 89(3), 459–
496. https://doi.org/10.3102/2F0034654319843494.
National Council of Teachers of Mathematics. (2006). Curriculum focal points for pre-kindergarten through
grade 8 mathematics. Reston, VA: Author.
National Mathematics Advisory Panel. (2008). Foundations for success: The final report of the National
Mathematics Advisory Panel. US Department of Education.
Newton, J. H., & McGrew, K. S. (2010). Introduction to the special issue: Current research in Cattell–Horn–
Carroll–based assessment. Psychology in the Schools, 47(7), 621–634. https://doi.org/10.1002/pits.20495.
*Pace, A., Alper, R., Burchinal, M. R., Golinkoff, R. M., & Hirsh-Pasek, K. (2019). Measuring success: within
and cross-domain predictors of academic and social trajectories in elementary school. Early Childhood
Research Quarterly, 46(1), 112–125. https://doi.org/10.1016/j.ecresq.2018.04.001.
Partnership for Assessment of Readiness for College and Careers. (2015). Math spring operational 2015 grade 5
end of year released items. Washington, D.C.: PARCC Assessment Consortia. Retrieved from https://prc.
parcconline.org/system/files/5th%20grade%20Math%20-%20EOY%20-%20Item%20Set_Sept.%202016.
pdfk.
*Pawk, M. L. (2003). A quantitative analysis of full-day kindergarten in rural elementary schools: a measure of
academic and social outcomes for children at-risk in the Armstrong school district (Publication No.
3089872) [Doctoral dissertation, Duquesne University]. ProQuest Dissertations and Theses Global.
Peng, P., & Kievit, R. A. (2020). The development of academic achievement and cognitive abilities: A
bidirectional perspective. Child Development Perspectives, 14(1), 15–20. https://doi.org/10.1111
/cdep.12352.
Peng, P., & Lin, X. (2019). The relation between mathematics vocabulary and mathematics performance among
fourth graders. Learning and Individual Differences, 69(1), 11–21. https://doi.org/10.1016/j.
lindif.2018.11.006.
Peng, P., Namkung, J., Barnes, M., & Sun, C. (2016). A meta-analysis of mathematics and working memory:
Moderating effects of working memory domain, type of mathematics skill, and sample characteristics.
Journal of Educational Psychology, 108(4), 455–473. https://doi.org/10.1037/edu0000079.
Peng, P., Wang, T., Wang, C., & Lin, X. (2019). A meta-analysis on the relation between fluid intelligence and
reading/mathematics: Effects of tasks, age, and social economics status. Psychological Bulletin, 145(2),
189–236. https://psycnet.apa.org/doi/10.1037/bul0000182.
*Petrill, S., Logan, J., Hart, S., Vincent, P., Thompson, L., Kovas, Y., & Plomin, R. (2012). Math fluency is
etiologically distinct from untimed math performance, decoding fluency, and untimed reading performance:
1122 Educational Psychology Review (2021) 33:1097–1124

evidence from a twin study. Journal of Learning Disabilities, 45(4), 371–381. https://doi.org/10.1177
/0022219411407926.
*Pina, V., Fuentes, L. J., Castillo, A., & Diamantopoulou, S. (2014). Disentangling the effects of working
memory, language, parental education, and non-verbal intelligence on children’s mathematical abilities.
Frontiers in Psychology, 5, Article 415. https://doi.org/10.3389/fpsyg.2014.00415.
*Pina, V., Castillo, A., Cohen Kadosh, R., & Fuentes, L. J. (2015). Intentional and automatic numerical
processing as predictors of mathematical abilities in primary school children. Frontiers in Psychology, 6,
Article 375. https://doi.org/10.3389/fpsyg.2015.00375.
Powell, S. R., & Fuchs, L. S. (2018). Effective word-problem instruction: using schemas to facilitate mathemat-
ical reasoning. Teaching Exceptional Children, 51(1), 31–42. https://doi.org/10.1177
/2F0040059918777250.
Powell, S. R., Driver, M. K., Roberts, G., & Fall, A. M. (2017). An analysis of the mathematics vocabulary
knowledge of third-and fifth-grade students: Connections to general vocabulary and mathematics computa-
tion. Learning and Individual Differences, 57(5), 22–32. https://doi.org/10.1016/j.lindif.2017.05.011.
Powell, S. R., Stevens, E. A., & Berry, K. A. (2019). Effects of a word-problem intervention on word-problem
language features for third-grade students with mathematics difficulty. Learning Disabilities: A
Multidisciplinary Journal, 24(2), 1–14. https://doi.org/10.18666/LDMJ-2019-V24-I2-9835.
*Purpura, D. J., & Ganley, C. M. (2014). Working memory and language: skill-specific or domain-general
relations to mathematics?. Journal of Experimental Child Psychology, 122(6), 104–121. https://doi.
org/10.1016/j.jecp.2013.12.009.
Purpura, D. J., Logan, J. A., Hassinger-Das, B., & Napoli, A. R. (2017). Why do early mathematics skills predict
later reading? The role of mathematical language. Developmental Psychology, 53(9), 1633–1642. https://doi.
org/10.1037/dev0000375.
*Razza, R. A., Martin, A., & Brooks-Gunn, J. (2015). Are approaches to learning in kindergarten associated with
academic and social competence similarly?. In Child & Youth Care Forum (Vol. 44, No. 6, pp. 757–776).
NIH Public Access.
*Roberts, J. E., Burchinal, M. R., & Zeisel, S. A. (2002). Otitis media in early childhood in relation to children’s
school-age language and academic skills. Pediatrics, 110(4), 696–706. https://doi.org/10.1542
/peds.110.4.696.
*Schmerold, K., Bock, A., Peterson, M., Leaf, B., Vennergrund, K., & Pasnak, R. (2017). The relations between
patterning, executive function, and mathematics. The Journal of Psychology, 151(2), 207–228. https://doi.
org/10.1080/00223980.2016.1252708.
*Seethaler, P. M., Fuchs, L. S., Fuchs, D., & Compton, D. L. (2012). Predicting first graders’ development of
calculation versus word-problem performance: the role of dynamic assessment. Journal of Educational
Psychology, 104(1), 224–234. https://doi.org/10.1037/a0024968.
*Seethaler, P. M., Fuchs, L. S., Fuchs, D., & Compton, D. L. (2016). Does the value of dynamic assessment in
predicting end-of-first-grade mathematics performance differ as a function of English language proficiency?
The Elementary School Journal, 117(2), 171–191. https://doi.org/10.1086/688870.
Sharpe, S. T., Krawec, J., & Fults, R. (2014). The effects of a mathematical problem-solving intervention on the
errors made by middle school students with and without learning disabilities. Middle Grades Research
Journal, 9(3), 43–60.
*Shaw, R. R. (2007). The relationship between pediatric HIV infection, CD4 percentage, HAART, and WISC-III
performance (Publication No. 3249129) [Doctoral dissertation, Columbia University]. ProQuest
Dissertations and Theses Global.
*Simmons, F., Singleton, C., & Horne, J. (2008). Brief report—Phonological awareness and visual-spatial
sketchpad functioning predict early arithmetic attainment: evidence from a longitudinal study. European
Journal of Cognitive Psychology, 20(4), 711–722. https://doi.org/10.1080/09541440701614922.
Singer, V., Strasser, K., & Cuadro, A. (2018). Direct and indirect paths from linguistic skills to arithmetic school
performance. Journal of Educational Psychology, 111(3), 434–445. https://doi.org/10.1037/edu0000290.
*Sisco-Taylor, D., Fung, W., & Swanson, H. L. (2015). Do curriculum-based measures predict performance on
word-problem-solving measures?. Assessment for Effective Intervention, 40(3), 131–142.
10.1177%2F15345084145565040.
*Slate, J. R. (1995). Two investigations of the validity of the WISC–III. Psychological Reports, 76(1), 299–306.
https://doi.org/10.2466/pr0.1995.76.1.299.
*Slot, E. M., van Viersen, S., de Bree, E. H., & Kroesbergen, E. H. (2016). Shared and unique risk factors
underlying mathematical disability and reading and spelling disability. Frontiers in Psychology, 7, Article
803. https://doi.org/10.3389/fpsyg.2016.00803
*Smiley, D. F. (2002). Problem solving ability in elementary school age children with hearing impairment
(Publication No. 3054144) [Doctoral dissertation, the University of Tennessee]. ProQuest Dissertations and
Theses Global.
Educational Psychology Review (2021) 33:1097–1124 1123

*Sorhagen, N. S. (2013). Early teacher expectations disproportionately affect poor children’s high school
performance. Journal of Educational Psychology, 105(2), 465–477. https://doi.org/10.1037/a0031754.
*Soto-Calvo, E., Simmons, F. R., Willis, C., & Adams, A. M. (2015). Identifying the cognitive predictors of
early counting and calculation skills: evidence from a longitudinal study. Journal of Experimental Child
Psychology, 140(12), 16–37. https://doi.org/10.1016/j.jecp.2015.06.011.
*Søvik, N., Frostrad, P., & Heggberget, M. (1999). The relation between reading comprehension and task-
specific strategies used in arithmetical word problems. Scandinavian Journal of Educational Research,
43(4), 371–398. https://doi.org/10.1080/0031383990430403.
Spencer, M., Fuchs, L. S., & Fuchs, D. (in press). Language-related longitudinal predictors of arithmetic word
problem solving: A structural equation modeling approach. Contemporary Educational Psychology.
https://doi.org/10.1016/j.cedpsych.2019.101825.
Sternberg, R. J., Grigorenko, E., & Zhang, L. (2008). Styles of learning and thinking matter in instruction and
assessment. Perspectives on Psychological Science, 3(6), 486–506. https://doi.org/10.1111/2Fj.1745-
6924.2008.00095.x.
Swanson, H. L. (2004). Working memory and phonological processing as predictors of children’s mathematical
problem solving at different ages. Memory & Cognition, 32(4), 648–661. https://doi.org/10.3758
/BF03195856.
*Swanson, H. L. (2011). Working memory, attention, and mathematical problem solving: a longitudinal study of
elementary school children. Journal of Educational Psychology, 103(4), 821–837. https://doi.org/10.1037
/a0025114.
Swanson, H. L., & Beebe-Frankenberger, M. (2004). The relationship between working memory and mathe-
matical problem solving in children at risk and not at risk for serious math difficulties. Journal of
Educational Psychology, 96(3), 471–491. https://doi.org/10.1037/0022-0663.96.3.471.
*Swanson, H. L., & Fung, W. (2016). Working memory components and problem-solving accuracy: are there
multiple pathways? Journal of Educational Psychology, 108(8), 1153–1177. https://doi.org/10.1037
/edu0000116.
Swanson, H. L., & Sachse-Lee, C. (2001). Mathematical problem solving and working memory in children with
learning disabilities: Both executive and phonological processes are important. Journal of Experimental
Child Psychology, 79(3), 294–321. https://doi.org/10.1006/jecp.2000.2587.
Thompson, B., Diamond, K. E., McWilliam, R., Snyder, P., & Snyder, S. W. (2005). Evaluating the quality of
evidence from correlational research for evidence-based practice. Exceptional Children, 71(2), 181–194.
https://doi.org/10.1177/2F001440290507100204.
*Träff, U., & Passolunghi, M. C. (2015). Mathematical skills in children with dyslexia. Learning and Individual
Differences, 40(4), 108–114. https://doi.org/10.1016/j.lindif.2015.03.024.
*Tupa, D. J., Wright, M. O. D., & Fristad, M. A. (1997). Confirmatory factor analysis of the WISC–III with child
psychiatric inpatients. Psychological Assessment, 9(3), 302–306. https://doi.org/10.1037/1040-3590.9.3.302.
*Ulu, M. (2016). A structural equation model to explain the effect of fluent reading, literal comprehension and
inferential comprehension levels of elementary school 4th grade students on success in problem solving.
Education & Science/Egitim ve Bilim, 41(186), 93–117. https://doi.org/10.15390/EB.2016.6303.
*University of Iowa (2008). ITBS research guide. Retrieved from https://itp.education.uiowa.
edu/ia/documents/itbs-research-guide.pdf.
*Van Geuns-Miranda, E. J. (1987). A study of the relationships among perceptions of parents, teachers and
principals of bilingual education and the self-concepts and achievement of Hispanic students [Unpublished
doctoral dissertation]. New York University.
*Vilenius-Tuohimaa, P. M., Aunola, K., & Nurmi, J. E. (2008). The association between mathematical word
problems and reading comprehension. Educational Psychology, 28(4), 409–426. https://doi.org/10.1080
/01443410701708228.
Viswesvaran, C., & Ones, D. S. (1995). Theory testing: Combining psychometric meta-analysis and structural
equations modeling. Personnel Psychology, 48(4), 865–885. https://doi.org/10.1111/j.1744-6570.1995.
tb01784.x.
*Vukovic, R. K. (2006). The development of numeracy: a longitudinal study of children from first through fourth
grade [Unpublished doctoral dissertation]. The University of British Columbia.
*Vukovic, R. K., Lesaux, N. K., & Siegel, L. S. (2010). The mathematics skills of children with reading
difficulties. Learning and Individual Differences, 20(6), 639–643. https://doi.org/10.1016/j.
lindif.2010.08.004.
*Vukovic, R. K., Kieffer, M. J., Bailey, S. P., & Harari, R. R. (2013). Mathematics anxiety in young children:
concurrent and longitudinal associations with mathematical performance. Contemporary Educational
Psychology, 38(1), 1–10. https://doi.org/10.1016/j.cedpsych.2012.09.001.
1124 Educational Psychology Review (2021) 33:1097–1124

Wang, A. Y., Fuchs, L. S., & Fuchs, D. (2016). Cognitive and linguistic predictors of mathematical word
problems with and without irrelevant information. Learning and Individual Differences, 52(8), 79–87.
https://doi.org/10.1016/j.lindif.2016.10.015.
*Willoughby, M. T., Blair, C. B., Wirth, R. J., Greenberg, M., & The Family Life Project Investigators. (2012).
The measurement of executive function at age 5: psychometric properties and relationship to academic
achievement. Psychological Assessment, 24(1), 226–239. https://doi.org/10.1037/a0025361.
Wohlwill, J. F. (1980). Cognitive development in childhood. In O. Brim & J. Kagan (Eds.), Development
Psychology (pp. 359–444). New York: Plenum.
*Wong, T. T. Y. (2017). The unique and shared contributions of arithmetic operation understanding and
numerical magnitude representation to children’s mathematics achievement. Journal of Experimental
Child Psychology, 164(12), 68–86. https://doi.org/10.1016/j.jecp.2017.07.007.
*Wong, T. T.-Y., & Ho, C. S.-H. (2017). Component processes in arithmetic word-problem solving and their
correlates. Journal of Educational Psychology, 109(4), 520–531. https://doi.org/10.1037/edu0000149.
*Zhang, X., & Lin, D. (2015). Pathways to arithmetic: the role of visual-spatial and language skills in written
arithmetic, arithmetic word problems, and nonsymbolic arithmetic. Contemporary Educational Psychology,
41(2), 188–197. https://doi.org/10.1016/j.cedpsych.2015.01.005.
*Zhang, X., & Lin, D. (2018). Cognitive precursors of word reading versus arithmetic competencies in young
Chinese children. Early Childhood Research Quarterly, 42(1), 55–65. https://doi.org/10.1016/j.
ecresq.2017.08.006.
*Zhang, X., Räsänen, P., Koponen, T., Aunola, K., Lerkkanen, M., & Nurmi, J. (2017). Knowing, applying, and
reasoning about arithmetic: roles of domain-general and numerical skills in multiple domains of arithmetic
learning. Developmental Psychology, 53(12), 2304–2318. https://doi.org/10.1037/dev0000432.supp.
*Zhang, J., Cheung, S. K., Wu, C., & Meng, Y. (2018). Cognitive and affective correlates of Chinese children’s
mathematical word problem solving. Frontiers in Psychology, 9, Article 2357. https://doi.org/10.3389
/fpsyg.2018.02357.
Zheng, X., Swanson, H. L., & Marcoulides, G. A. (2011). Working memory components as predictors of
children’s mathematical word problem solving. Journal of Experimental Child Psychology, 110(4), 481–
498. https://doi.org/10.1016/j.jecp.2011.06.001.

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.