EJ1192397

714773
research-article2017
LDXXXX10.1177/0022219417714773Journal of Learning DisabilitiesNelson and Powell
Article
Journal of Learning Disabilities
A Systematic Review of Longitudinal

2018, Vol. 51(6) 523–539
© Hammill Institute on Disabilities 2017
Article reuse guidelines:
Studies of Mathematics Difficulty sagepub.com/journals-permissions
DOI: 10.1177/0022219417714773
https://doi.org/10.1177/0022219417714773
journaloflearningdisabilities.sagepub.com
Gena Nelson, MA1, and Sarah R. Powell, PhD2
Abstract
Some students may be diagnosed with a learning disability in mathematics or dyscalculia, whereas other students may
demonstrate below-grade-level mathematics performance without a disability diagnosis. In the literature, researchers
often identify students in both groups as experiencing math difficulty. To understand the performance of students with
math difficulty, we examined 35 studies that reported longitudinal results of mathematics achievement (i.e., mathematics
performance measured across at least a 12-month span). Our primary goal was to conduct a systematic review of these
studies and to understand whether the growth of students with math difficulty was comparable or stagnant when compared
with that of students without math difficulty. We also analyzed whether identification of math difficulty was predictive of
mathematics achievement in later grades and whether a diagnosis of math difficulty was stable across grade levels. Results
indicate that students with math difficulty demonstrate growth on mathematics measures, but this growth still leads to
lower performance than that of students without math difficulty. Identification of math difficulty is strongly related to math
performance in subsequent grades, and this diagnosis is often stable. Collectively, this literature indicates that students with
math difficulty continue to struggle with mathematics in later grades.
Keywords
mathematics, learning disability, longitudinal, predictive, learning trajectories
Infants demonstrate an understanding of number, and this besides speech impairment and visual impairment but not
understanding is related to mathematics performance in the on level with students without a disability. Wei and col-
toddler years (Starr, Libertus, & Brannon, 2013). Toddler leagues’ (2013) research indicates that math performance
and preschool mathematics experiences influence readiness follows a longitudinal path for students with disabilities.
for kindergarten (Magnuson, Ruhm, & Waldfogel, 2007; The disability category of specific learning disability, how-
Murray & Harrison, 2011). Subsequently, mathematics per- ever, can include students who experience severe difficulty
formance at the elementary grades predicts mathematics with reading, writing, spelling, or math (Moll, Kunze,
performance during middle and high school (Watts, Duncan, Neuhoff, Bruder, & Schulte-Körne, 2014; Scanlon, 2013).
Siegler, & Davis-Kean, 2014). High school mathematics For that reason, specific learning disability is a category
performance is related to college readiness, which in turn with tremendous variability, and it is difficult to draw con-
increases adulthood outcomes (Baum, Ma, & Payea, 2010). clusions about the effect of math disability or difficulty
It is clear from this cadre of longitudinal research that math- from this single category. In this systematic review, we syn-
ematics knowledge at an earlier age predicts mathematics thesized longitudinal outcomes for students with math dif-
performance at a later age. (Note that, for the rest of this ficulty to determine whether growth across years was
article, we abbreviate mathematics as math.) comparable to students without math difficulty, whether
Wei, Lenz, and Blackorby (2013) recently analyzed the identified math difficulty was predictive of later math
math performance of students across ages 7 to 17 and com- achievement, and whether the diagnosis of math difficulty
pared the calculation and problem-solving performance of was stable across years. As math is the greatest predictor of
students in 11 disability categories with that of students later school success (Claessens, Duncan, & Engel, 2009), it
without disability. Except at age 7, students without a dis-
ability outperformed students with a disability on both types 1
American Institutes for Research, Washington, DC, USA
of math assessments at every time point. As students aged, 2
University of Texas at Austin, USA
the gap widened, favoring students without a disability.
Corresponding Author:
Students in the largest disability category (i.e., 41% of stu- Gena Nelson, American Institutes for Research, 1050 Thomas Jefferson
dents with a diagnosed disability), specific learning disabil- St. NW, #3224, Washington, DC 20007
ity, performed better than students in all disability categories Email: nels8101@umn.edu
524 Journal of Learning Disabilities 51(6)
is necessary to understand the math trajectories of students such deficits (Geary, 2000). We review the results from stud-
who have difficulty with math, to plan for intervention and ies where the performance of students with math difficulty
assessment. was compared with that of students in at least one other
group (e.g., math and reading difficulty, typical perfor-
mance) to (a) understand whether the growth is differential
Math Disability or Difficulty for students with math difficulty and (b) assess the predictive
Approximately, 3% to 8% of school-age students have a nature and stability of math difficulty identification.
diagnosed disability related to math (Desoete, Roeyers, &
De Clercq, 2004; Shalev, Auerbach, Manor, & Gross-Tsur, Math Performance of Students With
2000). Math disability may be referred to as dyscalculia
Math Difficulty
(e.g., Ashkenazi & Henik, 2010). A greater number of stu-
dents, however, struggle with low math performance with- Researchers often provide a snapshot of the math perfor-
out a disability diagnosis. In the literature, math difficulty is mance of students with math difficulty and compare it with
the term used to represent students with low math perfor- the performance of students in other groups (e.g., math and
mance as well as students with a diagnosed math disability. reading difficulty, reading difficulty, typical performance;
This is likely to due to several reasons. First, specific learn- Andersson, 2008). In the elementary grades, students with
ing disability is often diagnosed in the later elementary math difficulty demonstrate lower performance on counting
grades (O’Connor, Bocian, Beach, Sanchez, & Flynn, tasks (Stock, Desoete, & Roeyers, 2010) as well as arithme-
2013), and researchers investigating the effects of early tic fluency and computation (Tolar, Fuchs, Fletcher, Fuchs,
identification and intervention must recruit participants & Hamlett, 2016). Students with math difficulty often
without the help of a school diagnosis (e.g., Dyson, Jordan, exhibit difficulty with comparing quantities, but some of
Beliakoff, & Hassinger-Das, 2015). Second, it may be dif- this may be due to symbolic (i.e., numerical) representa-
ficult to locate the necessary sample of participants with a tions of quantities (De Smedt & Gilmore, 2011; Driver &
math disability, so researchers employ methods for identify- Powell, 2015). Additionally, word problem solving is prob-
ing students who demonstrate characteristics of math dis- lematic for students with math difficulty (Fuchs et al., 2008;
ability (e.g., Fuchs et al., 2014). In this systematic review, Kingsdorf & Krawec, 2014). Many of these areas continue
we describe 35 studies related to the math performance of to cause problems for students with math difficulty in sec-
students with math disability or difficulty. For consistency, ondary settings (Calhoon, Emerson, Flores, & Houchins,
we refer to the students in these studies as experiencing 2007), which has implications for postsecondary options.
math difficulty. To eliminate confusion with the original Researchers have also determined that cognitive character-
research synthesized in this manuscript, we do not use any istics—such as those related to working memory (Swanson
acronym for math difficulty (e.g., MD, MMD, MLD). & Beebe-Frankenberger, 2004), processing speed (Cirino,
How researchers identify math difficulty is quite vari- Fuchs, Elias, Powell, & Schumacher, 2015), and phonologi-
able (Watson & Gable, 2012). When researchers use cutoff cal processing (Fuchs et al., 2006)—are distinctive for stu-
scores or percentiles, some cutoffs are quite stringent (<10th dents with math difficulty. Additionally, students with math
percentile; Geary, Hoard, Nugent, & Byrd-Craven, 2008), difficulty often have high math anxiety and low self-effi-
whereas others are more relaxed (≤40th percentile; Jitendra cacy (Rubinsten & Tannock, 2010).
et al., 2012). Other methods of identification include scor- Students with math difficulty often benefit from specific
ing below a standard score or SD (e.g., Swanson, 2012) or and targeted intervention designed to improve math out-
nonresponse to effective small-group math instruction (e.g., comes (e.g., Bryant et al., 2008; Fuchs et al., 2010; Mancl,
Bryant et al., 2016). Even though one assessment may not Miller, & Kennedy, 2012; Swanson, Moran, Lussier, &
adequately portray students with math difficulty (Branum- Fung, 2014). Some of these interventions can change the
Martin, Fletcher, & Stuebing, 2012; Mazzocco & Myers, learning trajectories of students with math difficulty.
2003), many research teams rely on this method for Without intervention, however, the outcomes may not be as
identification. promising.
To understand the math profiles of students with math
difficulty—in both assessment and intervention research— Purpose and Research Questions of
performance is often compared with that of students with
the Present Study
math difficulty comorbid with reading difficulty (e.g.,
Powell, Fuchs, Fuchs, Cirino, & Fletcher, 2009) and stu- In this systematic review, we explore the growth in math
dents without math difficulty (e.g., Clarke et al., 2011). performance for students with and without math difficulty,
Comparing the math performance of students with and with- to understand the pattern of performance across years when
out math difficulty allows for an understanding of the defi- intervention is not applied. We also analyze whether math
cits associated with math difficulty and the uniqueness of difficulty can be identified in earlier grades (i.e., prediction)
Nelson and Powell 525
and whether this identification remains constant across works. We identified four more articles that met inclusion
grade levels (i.e., stability). Specifically, this systematic criteria and included each of these in this review for a total
review addressed the following research questions: of 35 studies.
Research Question 1: What are the characteristics of lon- Inclusion Criteria

gitudinal studies related to math difficulty that identified
predictors of math achievement? We used the following criteria to determine study
Research Question 2: What is the growth of students inclusion.
with and without math difficulty on math measures First, the study examined a longitudinal trajectory of
across at least a 12-month period? math achievement for students with math difficulty without
Research Question 3: Is math difficulty related to earlier testing the efficacy of a math program or intervention; stud-
or later math performance? ies that focused on the success of math interventions were
Research Question 4: Is math difficulty stable over time? excluded (e.g., Nunes et al., 2009).
Second, participants were school-age (i.e., preschool to
12th grade), and a subset of participants demonstrated math
Method difficulty. Acceptable methods of documentation for math
difficulty included an author-specified criterion on a math
Literature Review measure or a disability diagnosis in math. Because one of
We reviewed studies published from January 1985 to our research questions focused on identifying differences in
December 2016 that focused on math difficulty and longitu- author-specified criteria for identifying students with math
dinal math achievement. The review included a search of difficulty and growth in math according to predetermined
the literature via Academic Search Premier, Education math difficulty identification, we excluded studies that uti-
Source, Educational Resources Information Center, and lized latent growth curve analysis to retroactively identify
PsycINFO databases with combinations of the following groups of students who performed in separate classes in
search terms: math*, achievement, longitudinal, growth, mathematics (e.g., Geary, Bailey, Littlefield et al., 2009).
predict*, traject*, stability, retained, dyscalculia, “math* We also excluded studies that focused on special popula-
difficulty” and “learning disability*.” This search resulted tions who are typically considered at risk, but we did not
in the identification of 1760 studies, which we reviewed in screen participants for math difficulty, such as students with
two phases. In Phase 1, we reviewed titles, keywords, and genetic disorders or learning disabilities (e.g., Murphy &
abstracts to eliminate studies that were clearly outside the Mazzocco, 2009).
scope of this review. Many studies (n = 1634) were not con- Third, the study included at least one math measure
sidered for further review, because the study was on a topic administered at a minimum of two time points at least 12
irrelevant to education (39.8%) or was focused on growth in months apart. We selected 12 months as the minimum time
reading or reading disabilities (16.0%), the effectiveness of frame to indicate at least one grade-level change from first
an intervention (15.2%), or assessment development to last data collection. The math measure did not have to be
research (13.6%). Other studies were excluded because the same measure administered at different time points.
they were not longitudinal or they included only unrelated Fourth, the study included appropriate results (e.g.,
predictors of math achievement (4.8%) and for other rea- regression analyses, correlations, stability of math diffi-
sons (4.0%; e.g., commentaries, focus on teacher percep- culty) to determine the longitudinal predictors of math
tions). In Phase 2, we conducted a comprehensive review of achievement, growth, and stability of performance. We
7.2% of articles (n = 126), and 31 studies met inclusion cri- excluded studies that included a math outcome but included
teria. Most articles were excluded for not meeting more only unfitting predictors for the purpose of this study (e.g.,
than one inclusion criteria. For example, many studies behavior, reading, motion sensitivity; Boets, De Smedt, &
(70.0%) did not include participants who were at risk for Ghesquiere, 2011) or studies that included appropriate pre-
math difficulty (e.g., Jordan, Glutting, & Ramineni, 2010), dictors of math achievement but included only unsuitable
or participants were considered at risk but the authors did outcomes (e.g., confidence in correct problem solving; e.g.,
not explicitly determine if the participants had math diffi- Garrett, Mazzocco, & Baker, 2006).
culty (e.g., participants had learning disabilities, but scores Finally, the study was published in English in a peer-
on math measures were not available; Wei et al., 2013). reviewed journal.
Other reasons for exclusion included nonlongitudinal
design of at least 12 months (37.6%) and not reporting
Coding of Studies
appropriate results (21.5%; see next section for specific cri-
teria). Finally, in an attempt to conduct an exhaustive We coded the 35 studies that met inclusion criteria for the
search, we also searched relevant researchers’ previous following information: study characteristics, average age
(or grade) of participants at the start of the study, attrition, 2009). The median sample size was 181, with a range of 28
data collection time points, measures of math achievement to 7,892 participants. The majority of studies focused on
administered at each time point, reliability and validity early elementary participants. Regarding grade level of par-
coefficients for measures, and results. Regarding study and ticipants, on average, participants were in first grade at the
sample characteristics, we coded the year of publication, start of the study and near the end of third grade at the con-
location, and if the study utilized a national data set (e.g., clusion of the study. Only 20% of studies collected data
Early Childhood Longitudinal Study–Kindergarten beyond fifth grade; no studies collected data beyond ninth
[ECLS-K]). Finally, we recorded the authors’ method for grade. In the majority of studies (91%), researchers collected
determining participants’ at-risk status for math difficulty. data themselves; the other three studies used the aforemen-
Many studies included more than one math difficulty group tioned ECLS-K national database. Only 16 separate research
or more than one comparison group; thus, classification teams collected data for the 35 studies. Of these, two teams
methods and sample sizes for all groups were coded produced 2 studies each; two teams produced 3 studies each;
separately. two teams produced 5 studies each; and one team produced
Regarding data points, we coded each data point (e.g., 6 studies each. Therefore, seven distinct research teams gen-
Time 1, Time 2, Time 3, . . .) and, when available, the time erated almost 75% of the studies.
of year for data collection data (e.g., fall Grade 1). At each In terms of classification of math difficulty, researchers in
time point, we recorded which measures were administered, 32 of the studies categorized students into groups using per-
a short description of the measure and skills assessed (e.g., centiles as a cutoff. The most stringent percentile cutoff for
broad math, calculation), and the reported reliability and determination of math difficulty was math performance
validity coefficients for the measures. Finally, for each below the 10th percentile; the least stringent cutoff was at or
study, we recorded general findings, specific significant below the 35th percentile. In two studies, performance based
results, and significant levels. Specific results included on SD below the mean was the determinant for math diffi-
findings such as the percentage of students who retained the culty. Interestingly, these two studies represented research
same math difficulty classification over time (i.e., stability from Spain (Navarro et al., 2012) and Sweden (Andersson,
of math difficulty), correlations, and results from regression 2010). One of the national database studies identified the
and growth curve analyses. math difficulty group by using proficiency levels on the
ECLS-K math assessment (Claessens & Engel, 2013).
By default (i.e., study inclusion criteria), all 35 studies
Interrater Agreement had at least one group of students with math difficulty. For
The first author coded all studies, and the second author 26 studies, determination of math difficulty was based on
double-coded 33% of the 35 studies. We discussed all dis- assessments administered during one data collection wave;
crepancies and focused on ensuring that the codes were in the other 9 studies, researchers identified math difficulty
accurate for the classification of students with math diffi- across two or more data collection waves or years. In some
culty, results, and time points of data collection. Coding dis- instances, math difficulty was predictively identified,
crepancies were resolved to determine the final code used in whereas in other cases, it was retroactively identified. To
the analyses. The second author double-checked all infor- provide more detail about math difficulty, researchers in 21
mation provided in Table 1. studies identified a second group of students with math dif-
ficulty. In 12 of these cases, the second group was identified
through less stringent criteria for math difficulty. This group
Results was often described as low achieving in math, whereas the
first group of students with math difficulty was described as
Characteristics of Studies
experiencing a math learning disability. In seven cases, the
Table 1 provides a summary of the math difficulty determi- second group with math difficulty included students dem-
nation within each of the 35 studies included in this system- onstrating difficulty with math comorbid with reading. The
atic review. All studies were published between 2000 and other two studies with a second math difficulty group
2016. Data collection occurred in eight countries, with focused on a group where a diagnosis of math difficulty was
approximately 65% of data collected within the United not stable across data collection years. Six studies included
States. Other countries in which data were collected included a group of students with reading difficulty without math dif-
Belgium (n = 4), Canada (n = 3), Germany (n = 1), the ficulty. In terms of comparison groups, all but two studies
Netherlands, (n = 1), Spain, (n = 1), Sweden (n = 1), and the (Peng et al., 2016; Toll & Van Luit, 2014) included at least
United Kingdom (n = 1). The mean sample size was 609 one comparison group without math difficulty. The lowest
participants, but this high mean was due to three studies that cutoff percentile for identification of typically achieving
analyzed data from the ECLS-K data set (Bodovski & students (i.e., not having math difficulty) was performance
Farkas, 2007; Claessens & Engel, 2013; Morgan et al., >10th percentile. In 75% of cases, the cutoff for typical
Table 1. Categorization of Math Difficulty Within Studies.
Math difficulty 1 Math difficulty 2 Other difficulty Comparison 1 Comparison 2
Authors (year) Location N (category;a n) (category; n) (category; n) (category; n) (category; n) Measures for determining math difficulty
Andersson (2010) Sweden 249 Special education services Special education services in math Special education No special education Math screener (addition, subtraction,
in math and ≥1.5 SD and reading and ≥1.5 SD below services in reading services (control; n = 94) multiplication), reading screener
below mean on math mean on math and reading and ≥1.5 SD (cloze passage comprehension)
(MD-only; n = 39) (MD-RD; n = 80) below mean on
reading (RD-only;
n = 36)
Bodovski and Farkas US 3,206 <25th (LA; n = 267) ≥25th (n = 2,939)b ECLS-K Mathematics Test
(2007)
Chong and Siegel Canada 214 ≤10th (MLD; n = not 11th–25th (LA; n = not reported) ≥35th (TA; n = not WJ-III Calculation; WJ-III Math Fluency
(2008) reported) reported)
Claessens and Engel US 7,154 Not mastered Proficiency Mastered Proficiency Level ECLS-K Mathematics Test
(2013) Level 2 (low-achieving; 2 (main analytic; n =
n = 2,329) 7,154)c
Cowan and Powell UK 258 ≤10th (MLD single time 10th–25th (LA single time point; >25th (TA single time point; WIAT-II (UK)
(2014) point; n = 29; MLD n = 42; LA persistent subset; n = 187; TA persistent
persistent subset; n = 14) subset; n = 166)
n = 11)
Desoete et al. Belgium 395 ≤10th (MD; n = 16) 10th–25th (LA; n = 64) >25th (TA; n = 315) Arithmetic Number Facts Test and/or
(2012) Kortrijk Arithmetic Test; confirmation
by school psychologist
Desoete and Belgium 66 ≤15th (delay; n = 22) ≥50th (n = 44) Arithmetic Number Facts Test
Grégoire (2006);
Study 1
Geary, Bailey, and US 158 <15th (MLD; n = 45) 15th–30th (LA; n = 17) >30th (TA; n = 96) WIAT-II Numerical Operations
Hoard (2009)
Geary et al. (2000) US 84 <35th in math (MD; n <35th in math and reading <35th in reading >35th in math and reading <35th in math WIAT Mathematics Reasoning, WJ-R
= 12) (MDRD; n = 16) (RD; n = 14) (normal; n = 26) at only 1 Letter-Word Identification
grade level
(variable; n
= 16)
Geary et al. (2007) US 278 <15th, 2 years (MLD; n 23rd–39th, at least 1 year (LA; >50th, 2 years (TA; n = 46) WIAT-II Numerical Operations
= 15) n = 44)
Geary et al. (2012) US 177 <25th and “slow growth” >25th but “slow growth” across >25th and “average growth” WIAT-II Numerical Operations
across 5 years (MLD; 5 years (LA; n = 29) across 5 years (TA;
n = 16) n = 132)
Geary et al. (2008) US 261 <11th, 2 years (MLD; n 11th–25th, 2 years (LA; n = 43) 26th–74th, 2 years (TA; WIAT-II Numerical Operations
= 19) n = 50)
Hecht and Vagi US 181 ≤25th (MD; n = 55) >40th (TA; n = 126) WJ-III Calculation composite
(2010) (Calculation and Math Fluency)
(continued)
527
528
Table 1. (continued)
Jordan and Hanich US 74 ≤15th on math (MMD- ≤15th on math and reading ≤15th on reading >35th (TA; n = 47) WJ-R Broad Math (Calculation and
(2003) only; n = 11) (MMD/MRD; n = 8) (MRD-only; n = 8) Applied Problems), WJ-R Broad
Reading (Letter-Word Identification
and Passage Comprehension)
Jordan et al. (2003a) US 105 ≤25th (poor fact mastery; 51st–75th (good fact Forced retrieval of number facts test
n = 45) mastery; n = 60)
Jordan et al. US 180 ≤35th on math (MD only; ≤35th on math and reading (MD- ≤35th on reading >35th (TA; n = 47) WJ-R Mathematics Composite
(2003b) n = 46) RD; n = 42) (RD only; n = 45) (Calculation and Applied Problems),
WJ-R Reading Composite (Letter-
Word Identification and Passage
Comprehension)
Jordan et al. (2002) US 180 ≤35th in math (MD only; ≤35th in math and reading (MD- ≤35th in reading ≥40th in math and reading WJ-R Broad Math (Calculation and
n = 46) RD; n = 42) (RD only; n = 45) (n = 47) Applied Problems), WJ-R Broad
Reading (Letter-Word Identification
and Passage Comprehension)
Locuniak and Jordan US 198 ≤25th (at risk; n = 48) >25th (not-at risk; n = 158) Number Knowledge Test, Number
(2008) Combinations Test
Martin et al. (2013) US 144 <32nd on math; >40th on >40th on math and reading WRAT-3 Arithmetic or WJ-III Math
reading (MD; n = 83) (no LD; n = 61) Fluency; WRAT-3 Reading
Mazzocco and US 106 ≤10th (MLD; n = 12) 11th–25th (LA; n = 18) >25th (TA; n = 76) WJ-R Calculation
Devlin (2008)
Mazzocco et al. US 71 ≤10th (MLD; n = 10) 11th–25th (LA; n = 9) 25th–95th (TA; n = 37) >95th (high TEMA-2, WJ-R Calculation
(2011) achieving;
n = 15)
Mazzocco and US 209 <10th, 2 or more years <10th, at least 1 year (MD all; ≥10th (non-MD; n = 174) TEMA-2
Myers (2003) (MD-persistent; n = 22) n = 35)d
Mazzocco et al. US 122 <10th (MLD; n = 11) 11th–25th (LA; n = 18) >25th (TA; n = 93) WJ-R Calculation
(2013)
Mazzocco and US 209 ≤10th (MLD; n = 23) >10th (non-MLD; n = 186) TEMA-2, WJ-R Calculation
Thompson (2005)
Morgan et al. (2009) US 7,892 ≤10th (MD; n = 988) >10th (no MD; n = 6,904) ECLS-K Mathematics Test
Murphy et al. (2007) US 249 ≤10th, 2 or more years 11th–25th, 2 or more years Did not meet MLD criteria TEMA-2
(MLD; n = 22) (MLD; n = 42) for at least 2 years (non-
MLD; n = 146)
(continued)
Table 1. (continued)

Navarro et al. Spain 127 ≤1 SD below the mean 1 SD below the mean to ≥1 SD above Early Numeracy Test (Spanish version)
(2012) (LA; n = 46) 1 SD above the mean the mean
(middle achiever; n = 76) (high
achiever;
n = 26)
Peng et al. (2016) US 176 ≤25th math and below Teacher nominations, followed by
average factor score WRAT-4, Rapid Sound Naming,
on reading measures Phonemic Decoding Fluency
(MDRD; n = 176) (TOWRE)
Schwenck et al. Germany 166 ≤25th math (MD; n = 22 ) ≤25th math and reading (COM; ≤25th reading (RD; > 26th math and reading Wurzburg Silent Reading Test; spelling
(2015)e n = 35) n = 24 ) (TD; n = 28) test (DERET 1–2+); Heidelberg
arithmetic test
Swanson et al. US 353 ≤25th mean on 2 >26th mean on 2 measures WISC-III Arithmetic; CTOPP Digit
(2008) measures (SMD; n = (not at risk; n = 219) Naming Speed
134)
Stock et al. (2010) Belgium 362f ≤10th (AD; n = 16) 11th–25th (LA; n = 27) >25th (TA; n = 319) Arithmetic Number Facts Test and/or
Kortrijk Arithmetic Test
Toll and Van Luit Netherlands 199 <15th (very low Early Numeracy Test–Revised
(2014) numeracy; n = 199)
Vanbinst et al. Belgium 28 <25th, all 3 time points >35th, all 3 time points (TD; Standardized mathematics test from
(2014) (MLD-persistent; n n = 14) Flemish Student Monitoring System
= 14)
Vukovic (2012) Canada 203 <25th on math at least <25th on math at least twice in 4 ≥25th math and reading WRAT-3 Arithmetic, WJ-III Letter-
twice in 4 years;g ≥25th yearsg and reading at least once (TD; n = 165) Word Identification
on reading (MD-only; in Grade 1, 2, or 3 (MD+RD;
n = 19) n = 19)
Vukovic and Siegel Canada 99 <25th, at least 2 years <25th, 1 year (MD-transient; ≥40th, all 4 years (TA; WRAT-3 Arithmetic
(2010) (MD-persistent; n = 26) n = 27) n = 46)
Note. All cutoffs are reported as percentiles unless otherwise noted. Student risk categories in table are reported as in the studies by authors. MD/MLD/MMD/SMD = math difficulty/disability/serious math problem-solving
difficulties; AD = arithmetic difficulty; MDRD/COM = math and reading difficulty/disability/comorbid; RD/MRD = reading difficulty/disability; LD = learning difficulty/disability; LA = low achieving; TA/TD = typically achieving/
developing; CTOPP = Comprehensive Test of Phonological Processing; ECLS-K = Early Childhood Longitudinal Survey–Kindergarten; TEMA-2 = Test of Early Mathematics Achievement–Second Edition; TOWRE = Test of Word Reading
Efficiency; WIAT-II = Wechsler Individual Achievement Test–Second Edition; WISC-III = Wechsler Intelligence Scale for Children–Third Edition; WJ-III = Woodcock-Johnson–Third Edition; WJ-R = Woodcock-Johnson–Second Edition; WRAT-3 =
Wide Range Achievement Test–Third Edition.
a
Specific category name provided by authors. bAuthors presented information for 25th–50th, 51st–75th, and >75th; we collapsed them into one category for analysis. cMain analytic sample includes the low-achieving students.
d
Includes the 22 MD-persistent students. eStudy included 2 other comparison groups with below-average IQs; groups not coded for analyses. fCalculated by adding AD, LA, and TA participants due to a discrepancy in the
manuscript. gStudents could not be identified if the two time points were only at kindergarten and Grade 1.
529
performance was, at a minimum, >25th percentile. Finally, the study was >70%, compared with 20% for students with
three studies included a second comparison group of stu- math difficulty (i.e., students without math difficulty had
dents without math difficulty, and the majority of research- less opportunity for growth). Of importance, with both
ers described this group as high achieving. trends, students with math difficulty continued to perform
worse than peers.
Studies reported parallel growth for students with and
Summary of Math Performance
without math difficulty regardless of restrictiveness of the
We organize the following sections as they relate to the criteria used to determine math difficulty. Math difficulty
growth of math skills, the longitudinal predictors of math criteria for this group of studies ranged from performance 1
difficulty and achievement, and the longitudinal stability of SD below the mean to <25th percentile. One study also
math difficulty. First, we discuss the growth of math required that students have performance <25th percentile
achievement to illustrate differences of specific skills over and “slow growth” to be identified as having math difficulty
time between students who experience math difficulty and (Geary et al., 2012), and another required that students have
students who do not. Then, we discuss predictors of math performance <25th percentile for at least 2 years of the
difficulty to later math to evaluate if specific skills at an study (meaning that students identified as having math dif-
earlier time point can accurately predict if students will ficulty likely did not have fast growth; Vukovic, 2012). The
experience consistent and severe deficits in math. Finally, third study had more restrictive criteria for determining
we examine the stability of math difficulty as it relates to math difficulty (<15th percentile; Jordan & Hanich, 2003).
variables such as the restrictiveness of the cutoff. Therefore, the parallel growth observed across these three
studies might be a result of the criteria for math difficulty
Differences in growth. The majority of studies examined dif- aligning with the students who are least likely to have faster
ferences in initial achievement of math and rate of growth growth. Interestingly, the study that reported faster growth
over time. Most math skills fell into one of the following for students with math difficulty in early numeracy was the
categories: early numeracy skills, computation, rational only study that also used a comprehensive early numeracy
numbers, or broad math (i.e., math performance across con- screening measure to identify students with math difficulty
tent areas). We report results according to the skills mea- (Navarro et al., 2012).
sured, and we report growth rates for students with math Though not specific to growth rates, differences in the
difficulty as compared with typically achieving peers. early numeracy skill of counting were consistently observed
over time (Desoete & Grégoire, 2006; Geary et al., 2000;
Early numeracy skills. Across studies, many early numer- Geary, Hoard, Byrd-Craven, Nugent, & Numtee, 2007;
acy skills were measured, including magnitude comparison, Murphy et al., 2007). For example, students who scored
counting, number knowledge, and number combinations ≤15th percentile on number knowledge and mental arithme-
(Geary et al., 2012; Jordan & Hanich, 2003; Navarro et al., tic in first grade had significantly lower counting abilities in
2012; Vukovic, 2012). Of the four studies that measured kindergarten (Desoete & Grégoire, 2006), and students with
growth in early numeracy over time, three used screen- math difficulty in first grade had more difficulty recogniz-
ing measures to determine math difficulty that focused on ing correct counting or recognizing counting errors (Geary
numeracy (e.g., Early Numeracy Test) or at least included et al., 2000; Geary et al., 2007; Murphy et al., 2007).
items related to numeracy (e.g., Numerical Operations sub- Mazzocco, Feigenson, and Halberda (2011) reported num-
test). Therefore, there was an alignment between screening ber sense differences measured as late as ninth grade, with
measures for determination of math difficulty and the mea- performance on approximate number sense acuity differen-
sures used to observe growth. tiating students identified with math difficulty.
Three studies reported that after the initial achievement
gaps were identified in kindergarten and first grade, growth Computation. All studies that measured growth in com-
on skills was essentially parallel for the early numeracy putation over time used measures that included subtests of
skills of composing and decomposing (Geary et al., 2012); computation or measures that focused only on computation
counting, digit placement, and number identification (e.g., researcher-developed computation screener) to iden-
(Jordan & Hanich, 2003); and number sequences and num- tify students with math difficulty. Therefore, there was an
ber identification (Vukovic, 2012). Two studies reported alignment between the screening measures and the measures
faster growth for students with math difficulty for skills used to observe growth across studies. Generally, students
such as number line estimation (Geary et al., 2012) and with math difficulty were less accurate with computation
broad numeracy skills measured with the Early Numeracy and used inefficient strategies more often when compared
Test (Navarro et al., 2012). Slower growth by students with- with peers from kindergarten through elementary school
out math difficulty on a broad measure, however, may have (Andersson, 2010; Chong & Siegel, 2008; Jordan & Hanich,
been the result of ceiling effects, as accuracy at the start of 2003; Jordan, Hanich, & Kaplan, 2003a; Jordan, Kaplan, &
Hanich, 2002; Swanson et al., 2008; Vanbinst, Ghesquière, subtests from norm-referenced measures of math achieve-
& De Smedt, 2014). The majority of studies reported results ment that included computation and word problem solving
for computational fluency, but there were a few studies that to screen students for math difficulty. All studies measured
also reported growth for computation in word problem con- growth with measures that were read orally to students and
texts. Regarding computational fluency, results were gener- that required them to solve word problems. Similar to accu-
ally consistent across studies in terms of how computation racy for computational fluency, the results for word prob-
was measured. Authors reported parallel growth (Chong & lem solving were inconsistent. Three studies reported
Siegel, 2008 [timed fact fluency]; Jordan & Hanich, 2003; substantially similar growth for students with math diffi-
Swanson et al., 2008) and slightly faster growth (Chong & culty (Andersson, 2010; Jordan, Hanich, & Kaplan, 2003a,
Siegel, 2008 [untimed written calculation]; Jordan et al., 2003b), and other studies reported slower growth for stu-
2002) for students with math difficulty when measuring dents with math difficulty (Cowan & Powell, 2014;
written computation skills. Authors reported parallel growth Swanson et al., 2008). Jordan and Hanich (2003) reported
between students with math difficulty and those without that the average increase in word problems was faster for
math difficulty for approximate arithmetic (e.g., selecting students identified as having math difficulty only, while
the answer closest to the correct answer; Andersson, 2010; participants identified as having comorbid reading and
Jordan & Hanich, 2003) and exact calculation (e.g., Jordan math difficulty had slower rates of growth. Moreover, we
& Hanich, 2003; Jordan et al., 2003a). Regarding measures did not observe consistent trends regarding criteria for iden-
of forced retrieval (i.e., examiners read and/or displayed tifying math difficulty or age of participants regarding
basic facts, and students were required to give an answer growth rates. For example, one study that reported slower
within 3 s), two studies observed parallel growth between growth for students with math difficulty had a restrictive
groups (Andersson, 2010; Jordan & Hanich, 2003), while cutoff (<10th percentile), while another study that had a
another study observed slower growth for students with similar restrictive cutoff (<15th percentile) reported faster
math difficulty (Jordan et al., 2003a). Vanbinst et al. (2014) growth for students with math difficulty.
observed faster growth for students with math difficulty
when students were required to answer computation items Rational numbers. Only three studies (Hecht & Vagi,
in a similar format but were not forced to use retrieval (i.e., 2010; Mazzocco & Devlin, 2008; Mazzocco et al., 2013)
students were told to answer quickly but were allowed to examined more advanced math concepts, such as the
use strategies other than retrieval). understanding and use of fractions and decimals. Although
Studies reported parallel growth for students with and researchers evaluated different components of performance
without math difficulty regardless of restrictiveness of the with fractions and included participants from different
criteria used to determine math difficulty. Math difficulty grade levels, authors consistently reported that students
criteria for this group of studies ranged from 1.5 SD below with math difficulty performed worse than peers over time.
the mean to ≤35th percentile. Interestingly, the two studies For example, Hecht and Vagi (2010) examined emerging
that reported faster growth for students with math difficulty fraction skills from fourth to fifth grade, and students with
in computation had the most and least restrictive criteria for math difficulty showed consistently lower performance and
determining math difficulty, ≤10th and ≤35th percentiles, smaller gains in performance. Meanwhile, Mazzocco et al.
respectively (Chong & Siegel, 2008; Jordan et al., 2002). (2013) evaluated the fraction comparison performance of
Several studies examined strategy use for computation, participants in fourth through eighth grade; students with
and results related to strategy use were generally consistent math difficulty did not reach ceiling-level performance by
across studies. Students with math difficulty used retrieval the end of the study as typically achieving students did.
strategies less often initially when compared with peers Finally, Mazzocco and Devlin (2008) measured student
(Geary et al., 2000; Geary et al., 2012; Vanbinst et al., understanding of decimals and representing fractions with
2014). Two studies reported that students without math dif- pictures and numerals from sixth through eighth grade;
ficulty made greater gains in correct computation retrieval students who performed ≤10th percentile had significantly
from first to second grade (Geary et al., 2000) and from first lower scores than students who performed between the 11th
to fifth grade (Geary et al., 2012). Vanbinst et al. (2014) and 25th percentiles.
observed that students with math difficulty more often used Unlike studies that examined growth for skills related to
procedural skills than retrieval, and Geary et al. (2012) other areas of mathematics, all studies that examined growth
reported that over time, students with math difficulty used with fractions did not use screening measures aligned to
more procedural skills to correctly solve simple addition fraction concepts. Studies identified students with math dif-
problems but had less accuracy over time with decomposi- ficulty using norm-referenced measures of mathematics
tion strategies. achievement and measured growth with researcher-devel-
Regarding word problem solving, researchers used either oped tools. Due to the small number of studies that evalu-
a researcher-developed measure related to computation or ated fraction growth, we were not able to report on themes
related to age or the restrictiveness of criteria for determin- math and reading difficulty. Interestingly, the growth trajec-
ing math difficulty. tories of each group were similar, and the achievement gap
remained through third grade. Jordan and Hanich (2003)
Broad math achievement. Finally, researchers observed examined the performance of students with math difficulty
group differences over time on performance in broad only and comorbid math and reading difficulty from second
math. We defined broad math as assessments with a vari- through third grade. In the fall of second grade, the two
ety of math content, concepts, and procedures. All but one math difficulty groups did not significantly differ, but at the
study used measures of broad math achievement to iden- second, third, and fourth data collection points (through
tify students with math difficulty; one study used a “forced spring of third grade), the group with math difficulty only
retrieval of number facts” measure, as the purpose of the outperformed the group with comorbid math and reading
study was to identify students with poor fact mastery (Jor- difficulty. So, although both math difficulty groups per-
dan et al., 2003a). Of the six studies that measured growth formed lower than the students with only reading difficulty
in broad math over time, four used the same measure to and typically achieving students, the students with only
screen and measure growth over time. Generally, there was math difficulty who were good readers grew faster in math
alignment between screening measures and measures used than did the students with math difficulty who were poor
to observe growth across studies. Students with math dif- readers.
ficulty consistently displayed parallel or slower growth
over time and maintained achievement gaps. For example, Predictors of math difficulty. Several studies (Desoete et al.,
according to two studies based on the ECLS-K data set, stu- 2012; Geary et al., 2009; Locuniak & Jordan, 2008; Maz-
dents with math difficulty in the fall of kindergarten yielded zocco & Thompson, 2005; Stock et al., 2010; Vukovic &
lower overall gains through third grade (Bodovski & Far- Siegel, 2010) evaluated specific variables that predicted
kas, 2007), and students with math difficulty in the fall and group membership (math difficulty vs. no math difficulty).
spring of kindergarten displayed the lowest achievement Two studies used discriminant function analyses (Stock et
and slowest growth rates during the next 5 years when com- al., 2010; Vukovic & Siegel, 2010). Vukovic and Siegel
pared with other groups (Morgan et al., 2009). Three other (2010) reported that significant predictors of persistent
studies observed parallel growth between students with math difficulty (at least 2 years) compared with math diffi-
math difficulty and typically achieving students, also main- culty displayed during only 1 year and typically achieving
taining the initial achievement gap (Jordan & Hanich, 2003; students were math concepts, number series, and number
Jordan et al., 2003a; Vukovic, 2012). naming. Stock et al. (2010) observed that two of the stron-
Interestingly, trends emerged regarding the rate of gest predictors of group membership in kindergarten
growth on broad math measures and the restrictiveness of through second grade were early numeracy skills in magni-
the criteria used to determine math difficulty. Half of the tude comparison and seriation. Stock et al. also reported
studies reported parallel growth; however, the only study to that conceptual counting measured in kindergarten cor-
report that students with math difficulty had faster rates of rectly identified 31% of students with math difficulty in
growth utilized the least restrictive cutoff (<35th percen- first and second grades and that 44% of students identified
tile), while two of the three studies reporting that students with math difficulty in first and second grades had severe
with math difficulty had significantly slower growth than deficits in magnitude comparison.
peers used the most restrictive criteria (≤10th percentile). In contrast, Mazzocco and Thompson (2005) evaluated
The other study that reported slower growth used perfor- whether scores on broad math measures accurately predicted
mance <25th percentile as a cutoff. The three studies that math difficulty status at a later time. Kindergarten scores on
reported slower growth used the same measure at screening the KeyMath (Connolly, 2007) and Test of Early Mathematics
as was used to measure growth. These studies followed stu- Ability (Ginsburg & Baroody, 2003) accurately predicted
dents from kindergarten through third and fifth grades, math difficulty (≤10th percentile) in second and third grades
while most studies that reported parallel or faster rates of with 83.5% correct classification. Researchers examined the
growth followed students from only second to third grade. relationship between specific items and math difficulty, and
To understand the complexity of math difficulty comor- interestingly, they reported results similar to other studies.
bid with reading difficulty, several researchers included stu- Mazzocco and Thompson (2005) reported that early numer-
dents who exhibited math and reading difficulty, in addition acy items that addressed counting, reading one-digit numer-
to students who exhibited math difficulty only, as a way to als, addition of one-digit numerals, magnitude judgments,
compare growth in broad math achievement. For example, and numeral comparison were significantly related to math
Vukovic (2012) observed student performance beginning in difficulty. Two studies reported that specific number sense
kindergarten. Results indicated significant group differ- skills, such counting, number knowledge, and composing
ences in initial math achievement. That is, typically achiev- and decomposing, as measured in kindergarten and first
ing outperformed math difficulty, which in turn outperformed grade predicted difficulty with arithmetic in second grade
with 52% accuracy (Locuniak & Jordan, 2008) and pre- of performance <32nd percentile to identify third- and
dicted broad math difficulty in third grade with 66% accu- fourth-grade students with math difficulty, but 2 years later
racy (Geary et al., 2009). Desoete et al. (2012) reported that only 40% of students retained the same classification; this
proficiency with symbolic (numerals) and nonsymbolic decrease was significant (p < .001).
(dots) comparison as measured in kindergarten was signifi-
cantly related to group membership (math difficulty vs. no
Discussion
math difficulty) in first and second grades, as determined by
achievement in computation, with students with math diffi- We conducted this systematic review of longitudinal
culty underperforming as comparing with peers. In sum- research related to math difficulty to understand the growth
mary, studies that evaluated predictors of math difficulty of students with and without math difficulty on math mea-
regularly reported that early numeracy skills, such as count- sures, if math difficulty was related to earlier or later math
ing and magnitude comparison, predicted math difficulty; performance, and if math difficulty varied according to fac-
however, with only three studies reporting such results and tors such as participant age and method of classification.
with each study focusing on participants in fourth grade and The goal of our first research question was to describe the
younger, it is difficulty to generalize and compare the results characteristics of longitudinal studies related to math diffi-
of these studies with other skills and other grade levels. culty. Although we conducted an electronic search of the
data from the last 30 years, the majority of studies (77%)
Stability of math difficulty. A subset of studies (n = 11) reported were published in the last 10 years, and most were con-
the longitudinal stability of math difficulty for participants. ducted in the United States. The recency of the literature is
Generally, more restrictive cutoffs for math difficulty likely due to improved educational funding for research
resulted in higher proportions of students retaining the same projects and the increase in focus on math education, espe-
math difficulty status. Morgan et al. (2009) examined the cially at earlier grade levels. With the push for high-quality
stability of math difficulty as a function of performing ≤10th math instruction at earlier grade levels (National Council of
percentile at specific time points. Stability of math difficulty Teachers of Mathematics, 2006), it is not surprising that
from kindergarten to fifth grade ranged from 28% to 65%, researchers shifted their attention to evaluating longitudinal
with the higher proportion of students retaining math diffi- differences between students with and without math diffi-
culty status being those who performed ≤10th percentile culty at earlier ages. The majority of studies focused data
during both the fall and spring of kindergarten, as compared collection at the early elementary grade levels, with only a
with math difficulty during either the fall or spring of kinder- few studies collecting data beyond sixth grade.
garten. In the same study, the highest proportion of students We also examined how math difficulty was determined,
(70%) who retained the same classification were those who and we observed a lack of consistency across studies and
experienced math difficulty in both the fall and spring of within research teams regarding the cutoff for classification
kindergarten and math difficulty in the spring of third grade. of math difficulty. Interestingly, only one study (Andersson,
In two other studies, Geary and colleagues reported that 2010) used special education services in math to identify
of the students identified with math difficulty by perfor- those students with math difficulty. Of the remaining stud-
mance <15th percentile in first grade, approximately 68% of ies, the majority identified students with math difficulty by
students had performed similarly the year before in kinder- employing a less restrictive cutoff (performance <25th vs.
garten (Geary et al., 2007), and 75% retained math difficulty <10th percentile); furthermore, only a few studies (e.g.,
status in third grade (Geary, Bailey, & Hoard, 2009). Mazzocco & Myers, 2003; Vanbinst et al., 2014) required
Mazzocco and Myers (2003) reported similar retention rates that students meet criteria for math difficulty more than one
(63%) for math difficulty for students who performed <10th time during the study. A deviation from more restrictive cut-
percentile from kindergarten to third grade. Although, Jordan off values and the lack of requiring for more than one mea-
and Hanich (2003) reported that only 18% of students identi- surement of low performance to identify students with math
fied as experiencing math difficulty by performing ≤15th difficulty may have critical implications for determining
percentile in second grade retained the same status in third how students are selected to receive additional supports and
grade; however, they also reported that 88% of students interventions and for the effectiveness of those interven-
identified as experiencing comorbid math and reading diffi- tions. For example, previous intervention research indicates
culty retained the same classification. Schwenck et al. (2015) that even when the average effectiveness of an intervention
reported a similar pattern between second and fourth grade: is reported as moderate, the subgroup performances of stu-
18% of students with math difficulty retained the same sta- dents with more severe math difficulty may in fact be much
tus, while 43% of students with comorbid math and reading lower than both the average effect and the effect with less
difficulty maintained status. In contrast, studies with less severe math difficulty (Toll & Van Luit, 2012).
restrictive cutoffs to determine math difficulty reported In summary, in the 35 studies included, more than 15
smaller proportions of students who retained math difficulty methods of identification were used to identify students in
status. For example, Martin et al. (2013) employed a cutoff just the first category of math difficulty (i.e., the most
restrictive category). Research teams with works included studies included, very few examined growth rates related to
in this synthesis, as well as researchers conducting interven- word problem solving and fractions. As all students are
tions with students with math difficulty, have noted the expected to set up and solve word problems on high-stakes
absence of a “gold standard” for determining math diffi- standardized assessments and as understanding fractions is
culty (Fuchs et al., 2005; Mazzocco et al., 2013; Vanbinst instrumental to success in algebra, more research related to
et al., 2014). When researchers use different methods to problem solving and rational numbers is necessary.
identify students as having math difficulty, the irregularity Furthermore, because no studies in this systematic review
may make it difficult for practitioners to confidently identify measured achievement beyond ninth grade, we could not
students who may need more intensive academic supports. determine growth patterns for students with math difficulty
Moreover, inconsistent criteria and methods of identification related to other skills, such as algebra or geometry. Finally,
make it quite difficult to compare results across studies. For although the primary purpose of this synthesis did not focus
these reasons, practitioners and researchers would benefit on math language, only one study measured growth in
from a consensus on what defines math difficulty. understanding of specific math language (Toll & Van Luit,
With our second research question, we investigated the 2014). Because comorbidity rates of math and reading dif-
longitudinal growth of students with and without math difficulty are high, researchers may consider achievement and
ficulty on math measures, and although results were some- growth in understanding specific math language a primary
what inconsistent, students with math difficulty performed focus in future research related to longitudinal achievement
lower than typically achieving peers and generally had par- of students with math difficulty.
allel or slower growth on measures of early numeracy, com- Our third research question aimed to identify whether
putation, rational numbers, and broad math. Due to the math difficulty was related to earlier or later math perfor-
nature of identifying students with math difficulty, it was mance. Generally, studies reported that early numeracy skills,
expected that students with math difficulty would have such as counting, number naming, seriation, and magnitude
lower initial math achievement; however, it was discourag- comparison, were predictive of math difficulty later on, with
ing that students with math difficulty failed to catch up to other skills, such as computation (Geary et al., 2009;
peers without math difficulty. The results of this synthesis Locuniak & Jordan, 2008; Stock et al., 2010; Vukovic &
were consistent with what is known about students with Siegel, 2010). What is interesting is that these early numer-
math difficulty regarding difficulty with skills such as acy skills are typically learned informally prior to beginning
counting and computation (Geary, 2004). school or early in formal schooling upon kindergarten entry.
Many studies followed students longitudinally through Furthermore, these skills are significantly related to broad
early elementary school, and a few studies continued track- math performance through fifth grade (Powell, Nelson, &
ing students through fifth grade and into middle school. The Peng, 2017). Thus, the results of this review suggest that it is
results of a review of longitudinal predictors of math critical to address deficits for students who exhibit math dif-
achievement determined that early numeracy skills—such ficulty early, such as use of procedural skills over retrieval
as quantity comparison, counting, and understanding of the skills for simple arithmetic, inability to identify correct or
number line (as measured in kindergarten and first grade) incorrect counting, and deficits in magnitude comparison and
and computation (as measured in early elementary school)— number naming. Practitioners must be aware of the skills that
were instrumental to student performance in broad math in predict later difficulty, determine which students may be at
fifth and sixth grades (Powell, Nelson, & Peng, 2017). If risk for math difficulty in later grades, and intervene with tar-
broad math achievement in later grades can be predicted as geted instruction in those skills in hopes of curtailing future
early as kindergarten with early numeracy measures, then math deficits.
the results of this review suggest that students with math Our fourth research question addressed the stability of
difficulty have lower achievement in skills such as early math difficulty according to different factors. Unfortunately,
numeracy. Therefore, students with math difficulty may lag the number of studies that reported stability of math diffi-
behind peers throughout school unless targeted interven- culty did not allow for analyses to examine the potential
tions are implemented to address deficits in math early. relationships between stability and type of measure and ini-
Moreover, our results shed light on the range of math skills tial age of identification for participants. The results of this
in which students persistently have difficulty, which may review, however, report themes regarding the restrictive-
inform necessary next steps for instruction and intervention ness of cutoffs and the proportion of students who retained
in math across grade levels. Students with math difficulty the math difficulty classification. These findings have criti-
consistently displayed deficits in counting, computation, cal implications for how researchers and practitioners con-
use of retrieval strategies, fraction comparison and estima- tinue to identify students with math difficulty. As discussed,
tion, and applied problem solving. The results of this review there is a lack of consistency regarding the cutoff used to
also highlight the lack of research regarding growth for stu- determine math difficulty status; the results of this synthesis
dents with math difficulty in specific areas of math. Of the suggest that more restrictive cutoffs (e.g., <10th percentile)
result in higher proportions of the same students exhibiting with a handful of studies extending to fifth grade. We did
math difficulty in later grades. Therefore, practitioners may not review any studies that focused on preschool or high
consider employing stricter cutoffs to determine which stu- school populations. This limits the generalizability of our
dents are at the greatest risk for later math failure. As findings and the knowledge base regarding what we know
resources in schools are limited, it is critical to provide ser- about students with math difficulty across school-age years
vices to the students who have the highest probability of regarding differences in growth, predictability of difficulty,
math failure without extra support. Furthermore, the limited and stability of math difficulty over time. In addition,
research on the stability of math difficulty sheds light on a because more studies focused on early elementary grades,
gap in the research base in early intervention and special we reported many findings on early numeracy skills, com-
education. Of the 11 studies that reported on the stability of putation, and some findings on fractions. Unfortunately,
math difficulty over time, only one study (Martin et al., due to the restricted age of participants in this synthesis, we
2013) reported on stability for participants beyond fourth cannot draw conclusions about other skills, such as word
grade. Little is known regarding the stability of math diffi- problem solving, algebra, or geometry. Future longitudinal
culty over time beyond elementary school. More research research should include preschool participants and extend
related to the stability of math difficulty that spans other into the middle and high school years to gain knowledge
grade levels (e.g., middle school and beyond) and considers about math difficulty in students of all ages.
the retention of math difficulty status for more than 2 or 3
years will move the field toward a consensus on the criteria
for identifying students with math difficulty.
Implications for Practitioners
The results of this systematic review provide the foundation
for implications for practitioners. First, students with more
Limitations and Directions for Future Research severe and persistent math difficulty may have fundamental
As with any study, there are limitations. First, participants math differences from peers with math difficulty identified
from approximately 75% of the studies were likely repre- with less restrictive cutoffs or difficulty that is not persis-
sentative of the same geographic areas because seven tent. Students identified with math difficulty with a restric-
research teams conducted the 75% of the studies. This may tive cutoff have lower initial performance and slower
limit the generalizability of the results to students in other growth than students who are identified with math difficulty
areas of the United States and even in other countries, as with less restrictive criteria, and students with persistent
well as students from different cultural and ethnic back- math deficits typically perform lower than peers with incon-
grounds. Furthermore, the results of this synthesis may be sistent deficits. These results have implications for practi-
limited because the research teams that represented the tioners, as there may be distinct differences within a
majority of the work may have employed similar method- subgroup of students with math difficulty and the effective-
ologies across their studies. Future research should include ness of an intervention may be moderated by the degree to
more longitudinal efforts that include participants from which students exhibit math deficits. Practitioners may con-
other cultural and regional populations as a way to replicate sider results with caution, as students identified as experi-
and extend current research findings. encing math difficulty with varying degrees of restriction
Second, much of the data collected in these studies was may actually represent fundamentally different groups of
collected prior to the adoption of the Common Core State students; thus, interventions that are effective for some stu-
Standards (CCSS; National Governors Association Center dents with math difficulty may not be as effective for other
for Best Practices & Council of Chief State School Officers, students with math difficulty (Powell, Cirino, & Malone, in
2010). Many American states have since adopted the CCSS, press). Practitioners can be proactive by frequently moni-
which means that standards for math learning and instruc- toring the progress of students with math difficulty, as a
tion are in flux. Thus, the differences in growth of math method for determining at what rate they are growing in
skills across time between students with and without math relation to the peers, such that they can make immediate
difficulty may change as a result of alternation to math stan- changes to instruction.
dards. Researchers may want to replicate their work within Second, practitioners may consider evaluating the
the era of the CCSS. Moreover, this synthesis did not instructional methods that they use with students with math
include studies that examined the effectiveness of interven- difficulty. The results of this review consistently show that
tions, but future research may explore the differences students with math difficulty do not catch up to peers across
between growth for students with and without math diffi- math content areas. This raises questions regarding the
culty when students with math difficulty receive extra sup- effectiveness of the instruction and intervention methods
port that corresponds to national and state standards for that students with math difficulty are exposed to in the
math learning. classroom. Previous research has identified high-quality
Third, many of the studies included in this synthesis and evidence-based practices in mathematics that practitio-
focused on students in kindergarten through third grade, ners may use with students with learning difficulties, such
as explicit and systematic instruction (Gersten et al., 2009) Andersson, U. (2008). Mathematical competencies in children with
and use of representations (Jitendra, Nelson, Pulles, Kiss, & different types of learning difficulties. Journal of Educational
Houseworth, 2016). Districts and schools may consider the Psychology, 100, 48–66. doi:10.1037/0022-0663.100.1.48
importance of providing their practitioners with more *Andersson, U. (2010). Skill development in different components
of arithmetic and basic cognitive functions: Findings from a
opportunities for professional development and on-site
3-year longitudinal study of children with different types of
coaching related to evidence-based practices in math for
learning difficulties. Journal of Educational Psychology, 102,
students who struggle. Another consideration for schools 115–134. doi:10.1037/a0016838
and practitioners is to examine the implementation fidelity Ashkenazi, S., & Henik, A. (2010). Attentional networks in devel-
of instructional programs and the alignment of interven- opmental dyscalculia. Behavioral and Brain Functions, 6(2),
tions to specific student deficits. Perhaps students with 1–12. doi:10.1186/1744-9081-6-2
math difficulty who show slower rates of growth over time Baum, S., Ma, J., & Payea, K. (2010). Education pays 2010: The
do not receive interventions at the recommended dosage or benefits of higher education for individuals and society. New
do not receive intensive intervention that is aligned to the York, NY: The College Board.
specific skill deficit in math. Furthermore, researchers who *Bodovski, K., & Farkas, G. (2007). Mathematics growth in early
conduct longitudinal research may consider collecting data elementary school: The roles of beginning knowledge, stu-
dent engagement, and instruction. The Elementary School
regarding the types of instruction that students in their stud-
Journal, 108, 115–130. doi:10.1086/525550
ies are receiving, to provide more insight related to growth
Boets, B., De Smedt, B., & Ghesquiere, P. (2011). Coherent
rates and stability of math difficulty. motion sensitivity predicts individual differences in subtrac-
tion. Research in Developmental Disabilities, 32, 1075–1080.
Conclusion doi:10.1016/j.ridd.2011.01.024
Branum-Martin, L., Fletcher, J. M., & Stuebing, K. K. (2012).
We reviewed longitudinal studies to compare the growth of Classification and identification of reading and math disabili-
students with and without math difficulty, examined spe- ties: The special case of comorbidity. Journal of Learning
cific predictors of math difficulty, and evaluated the stabil- Disabilities, 46, 490–499. doi:10.1177/0022219412468767
ity of math difficulty regarding classification criteria. At all Bryant, D. P., Bryant, B. R., Gersten, R. M., Scammacca, N.,
grade levels and across skills, students with math difficulty Funk, C., Winter, A., . . . Pool, C. (2008). The effects of Tier
2 intervention on the mathematics performance of first-grade
consistently performed lower than students without math
students who are at risk for mathematics difficulties. Learning
difficulty and had similar or slower growth. Moreover, the
Disability Quarterly, 31, 47–63. doi:10.2307/20528817
likelihood of retaining math difficulty across years was Bryant, B. R., Bryant, D. P., Porterfield, J., Dennis, M. S.,
high. Thus, students with math difficulty did not catch up to Falcomata, T., Valentine, C., . . . Bell, K. (2016). The
their peers. Without targeted interventions and early deter- effects of a Tier 3 intervention on the mathematics perfor-
mination of difficulty with math, students as early as kin- mance of second grade students with severe mathematics
dergarten who display math difficulty may be at risk for difficulties. Journal of Learning Disabilities, 49, 176–188.
poor secondary and adulthood outcomes. doi:10.1177/0022219414538516
Calhoon, M. B., Emerson, R. W., Flores, M., & Houchins, D. E.
Authors’ Note (2007). Computational fluency performance profile of high
school students with mathematics disabilities. Remedial and
The content is solely the responsibility of the authors and does not Special Education, 28, 292–303. doi:10.1177/074193250702
necessarily represent the official views of the Institute of Education 80050401
Sciences or the U.S. Department of Education. *Chong, S. L., & Siegel, L. S. (2008). Stability of computational
deficits in math learning disability from second through
Declaration of Conflicting Interests fifth grades. Developmental Neuropsychology, 33, 300–317.
The author(s) declared no potential conflicts of interest with respect doi:10.1080/87565640801982387
to the research, authorship, and/or publication of this article. Cirino, P. T., Fuchs, L. S., Elias, J. T, Powell, S. R., & Schumacher,
R. F. (2015). Cognitive and mathematical profiles for dif-
Funding ferent forms of learning difficulties. Journal of Learning
Disabilities, 48, 156–175. doi:10.1177/0022219413494239
The author(s) disclosed receipt of the following financial support
Claessens, A., Duncan, G., & Engel, M. (2009). Kindergarten skills
for the research, authorship, and/or publication of this article: This
and fifth-grade achievement: Evidence from the ECLS-K.
research was supported in part by Grant R324A150078 from the
Economics of Education Review, 28, 415–427. doi:10.1016/j.
Institute of Education Sciences in the U.S. Department of
econedurev.2008.09.003
Education to the University of Texas at Austin. This research was
*Claessens, A., & Engel, M. (2013). How important is where you
also supported by the Greater Texas Foundation.
start? Early mathematics knowledge and later school success.
Teachers College Review, 115, 1–29.
References Clarke, B., Smolkowski, K., Baker, S. K., Fien, H., Doabler, C. T.,
*Asterisk denotes a study included in this synthesis. & Chard, D. J. (2011). The impact of a comprehensive Tier
1 core kindergarten program on the achievement of students on the mathematical problem solving of third-grade students
at risk in mathematics. The Elementary School Journal, 111, with math and reading difficulties. Exceptional Children, 74,
561–584. doi:10.1086/659033 155–173. doi:10.1177/001440290807400202
Connolly, A. J. (2007). KeyMath3. Bloomington, MN: Pearson. Garrett, A. J., Mazzocco, M. M. M., & Baker, L. (2006).
*Cowan, R., & Powell, D. (2014). The contributions of domain- Development of the metacognitive skills of prediction and
general and numerical factors to third-grade arithmetic skills evaluation in children with or without math disability.
and mathematical learning disability. Journal of Educational Learning Disabilities Research and Practice, 21, 77–88.
Psychology, 106, 214–229. doi:10.1037/a0034097 doi:10.1111/j.1540-5826.2006.00208.x
De Smedt, B., & Gilmore, C. K. (2011). Defective number mod- Geary, D. C. (2000). From infancy to adulthood: The develop-
ule or impaired access? Numerical magnitude processing ment of numerical abilities. European Child and Adolescent
in first grades with mathematical difficulties. Journal of Psychiatry, 9(Suppl. 2), S11. doi:10.1007/s007870070004
Experimental Child Psychology, 108, 278–292. doi:10.1016/j. Geary, D. C. (2004). Mathematics and learning disabilities.
jecp.2010.09.003 Journal of Learning Disabilities, 37, 4–15. doi:10.1177/002
*Desoete, A., Ceulemans, A., De Weerdt, F., & Pieters, S. (2012). 22194040370010201
Can we predict mathematical learning disabilities from *Geary, D. C., Bailey, D. H., & Hoard, M. K. (2009). Predicting
symbolic and non-symbolic comparison tasks in kindergar- mathematical achievement and mathematical learning disabil-
ten? Findings from a longitudinal study. British Journal of ity with a simple screening tool. Journal of Psychoeducational
Educational Psychology, 82, 64–81. doi:10.1348/2044- Assessment, 27, 265–279. doi:10.1177/0734282908330592
8279.002002 Geary, D. C., Bailey, D. H., Littlefield, A., Wood, P., Hoard,
*Desoete, A., & Grégoire, J. (2006). Numerical competence in M. K., & Nugent, L. (2009). First-grade predictors of math-
young children and in children with mathematics learning dis- ematical learning disability: A latent class trajectory analysis.
abilities. Learning and Individual Differences, 16, 351–367. Cognitive Development, 24, 411–429. doi:10.1016/j.cog-
doi:10.1016/j.lindif.2006.12.006 dev.2009.10.001
Desoete, A., Roeyers, H., & De Clercq, A. (2004). Children with *Geary, D. C., Hamson, C. O., & Hoard, M. K. (2000). Numerical
mathematics learning disabilities in Belgium. Journal of and arithmetical cognition: A longitudinal study of process
Learning Disabilities, 37, 50–61. doi:10.1177/00222194040 and concept deficits in children with learning disability.
370010601 Journal of Experimental Child Psychology, 77, 236–263.
Driver, M. K., & Powell, S. R. (2015). Symbolic and nonsymbolic doi:10.1006/jecp.2000.2561
equivalence tasks: The influence of symbols on students with *Geary, D. C., Hoard, M. K., Byrd-Craven, J., Nugent, L., &
mathematics difficulty. Learning Disabilities Research and Numtee, C. (2007). Cognitive mechanisms underlying
Practice, 30, 127–134. doi:10.1111/ldrp.12059 achievement deficits in children with mathematical learning
Dyson, N., Jordan, N. C., Beliakoff, A., & Hassinger-Das, B. disability. Child Development, 78, 1343–1359.
(2015). A kindergarten number-sense intervention with *Geary, D. C., Hoard, M. K., Nugent, L., & Bailey, D. H. (2012).
contrasting practice conditions for low-achieving children. Mathematical cognition deficits in children with learning dis-
Journal for Research in Mathematics Education, 46, 331– abilities and persistent low achievement: A five-year prospec-
370. doi:10.5951/jresematheduc.46.3.0331 tive study. Journal of Educational Psychology, 104, 206–223.
Fuchs, L. S., Compton, D. L., Fuchs, D., Paulsen, K., Bryant, doi:10.1037/a0025398
J. D., & Hamlett, C. L. (2005). The prevention, identifica- *Geary, D. C., Hoard, M. K., Nugent, L., & Byrd-
tion, and cognitive determinants of math difficulty. Journal Craven, J. (2008). Development of number line rep-
of Educational Psychology, 97, 493–513. doi:10.1037/0022- resentations in children with mathematical learning
0663.97.3.493 disability. Developmental Neuropsychology, 33, 277–299.
Fuchs, L. S., Fuchs, D., Compton, D. L., Powell, S. R., Seethaler, doi:10.1080/87565640801982361
P. M., Capizzi, A. M., . . . Fletcher, J. M. (2006). The cog- Gersten, R., Chard, D. J., Jayanthi, M., Baker, S. K., Morphy, P.,
nitive correlates of third-grade skill in arithmetic, algorith- & Flojo, J. (2009). Mathematics instruction for students with
mic computation, and arithmetic word problems. Journal learning disabilities: A meta-analysis of instructional com-
of Educational Psychology, 98, 29–43. doi:10.1037/0022- ponents. Review of Educational Research, 79, 1202–1242.
0663.98.1.29 doi:10.3102/0034654309334431
Fuchs, L. S., Powell, S. R., Seethaler, P. M., Cirino, P. T., Fletcher, Ginsburg, H. P., & Baroody, A. J. (2003). Test of Early
J. M., Fuchs, D., & Hamlett, C. L. (2010). The effects of strate- Mathematics Ability (3rd ed.). Austin, TX: Pro-Ed.
gic counting instruction, with and without deliberate practice, *Hecht, S. A., & Vagi, K. J. (2010). Sources of group and indi-
on number combination skill among students with mathemat- vidual differences in emerging fraction skills. Journal
ics difficulties. Learning and Individual Differences, 20, 89– of Educational Psychology, 102, 843–859. doi:10.1037/
100. doi:10.1016/j.lindif.2009.09.003 a0019824
Fuchs, L. S., Schumacher, R. F., Sterba, S. K., Long, J., Namkung, Jitendra, A. K., Nelson, G., Pulles, S. M., Kiss, A. J., &
J., Malone, A., . . . Changas, P. (2014). Does working memory Houseworth, J. (2016). Is mathematical representation of
moderate the effects of fraction intervention? An aptitude- problems an evidence-based strategy for students with
treatment interaction. Journal of Educational Psychology, mathematics difficulties? Exceptional Children, 83, 8–25.
106, 499–514. doi:10.1037/a0034341 doi:10.1177/0014402915625062
Fuchs, L. S., Seethaler, P. M., Powell, S. R., Fuchs, D., Hamlett, C. Jitendra, A. K., Rodriguez, M., Kanive, R., Huang, J.-P., Church,
L., & Fletcher, J. M. (2008). Effects of preventative tutoring C., Corroy, K. A., & Zaslofsky, A. (2012). Impact of
small-group tutoring interventions on the mathematical prob- the primary school-age years. Annals of Dyslexia, 53, 218–
lem solving and achievement of third-grade students with 253. doi:10.1007/s11881-003-0011-7
mathematics difficulties. Learning Disability Quarterly, 36, *Mazzocco, M. M. M., Myers, G. F., Lewis, K. E., Hanich, L.
21–35. doi:10.1177/0731948712457561 B., & Murphy, M. M. (2013). Limited knowledge of frac-
Jordan, N. C., Glutting, J., & Ramineni, C. (2010). The impor- tion representations differentiates middle school students
tance of number sense to mathematics achievement in first with mathematics learning disability (dyscalculia) versus low
and third grades. Learning and Individual Differences, 20, mathematics achievement. Journal of Experimental Child
82–88. doi:10.1016/j.lindif.2009.07.004 Psychology, 115, 371–387. doi:10.1016/j.jecp.2013.01.005
*Jordan, N. C., & Hanich, L. B. (2003). Characteristics of chil- *Mazzocco, M. M., & Thompson, R. E. (2005). Kindergarten
dren with moderate mathematics deficiencies: A longitudinal predictors of math learning disability. Learning Disabilities
perspective. Learning Disabilities Research and Practice, 18, Research & Practice, 20, 142–155. doi:10.1111/j.1540-
213–221. doi:10.1111/1540-5826.00076 5826.2005.00129.x
*Jordan, N. C., Hanich, L. C., & Kaplan, D. (2003a). Arithmetic Moll, K., Kunze, S., Neuhoff, N., Bruder, J., & Schulte-Körne,
fact master in young children: A longitudinal investigation. G. (2014). Specific learning disorder: Prevalence and gender
Journal of Experimental Child Psychology, 85, 103–119. differences. PLoS ONE, 9(7), e103537. doi:10.1371/journal.
doi:10.1016/S0022-0965(03)00032-8 pone.0103537
*Jordan, N. C., Hanich, L. B., & Kaplan, D. (2003b). A longitudi- *Morgan, P. L., Farkas, G., & Wu, Q. (2009). Five-year growth
nal study of mathematical competencies in children with spe- trajectories of kindergarten children with learning difficulties
cific mathematics difficulties versus children with comorbid in mathematics. Journal of Learning Disabilities, 42, 306–
mathematics and reading difficulties. Child Development, 74, 321. doi:10.1177/0022219408331037
834–850. doi:10.1111/1467-8624.00571 Murphy, M. M., & Mazzocco, M. M. M. (2009). The trajectory of
*Jordan, N. C., Kaplan, D., & Hanich, L. B. (2002). Achievement mathematics skills and working memory thresholds in girls
growth in children with learning difficulties in mathemat- with fragile X syndrome. Cognitive Development, 24, 430–
ics: Findings of a two-year longitudinal study. Journal of 449. doi:10.1016/j.cogdev.2009.09.004
Educational Psychology, 94, 586–597. doi:10.1037//0022- *Murphy, M. M., Mazzocco, M. M., Hanich, L. B., & Early, M.
0663.94.3.586 C. (2007). Cognitive characteristics of children with math-
Kingsdorf, S., & Krawec, J. (2014). Error analysis of mathematics ematics learning disability (MLD) vary as a function of the
word problem solving across students with and without learn- cutoff criterion used to define MLD. Journal of Learning
ing disabilities. Learning Disabilities Research and Practice, Disabilities, 40, 458–478.
29, 66–74. doi:10.1111/ldrp.12029 Murray, E., & Harrison, L. J. (2011). The influence of being ready
*Locuniak, M. N., & Jordan, N. C. (2008). Using kindergar- to learn on children’s early school literacy and numeracy
ten number sense to predict calculation fluency in sec- achievement. Educational Psychology, 31, 529–545. doi:10.
ond grade. Journal of Learning Disabilities, 41, 451–459. 1080/01443410.2011.573771
doi:10.1177/0022219408321126 National Council of Teachers of Mathematics. (2006). Curriculum
Magnuson, K. A., Ruhm, C., & Waldfogel, J. (2007). Does pre- focal points for prekindergarten through Grade 8 mathemat-
kindergarten improve school preparation and performance? ics. Reston, VA: Author.
Economics of Education Review, 26, 33–51. doi:10.1016/j. National Governors Association Center for Best Practices &
econedurev.2005.09.008 Council of Chief State School Officers. (2010). Common Core
Mancl, D. B., Miller, S. P., & Kennedy, M. (2012). Using the State Standards mathematics. Washington, DC: Authors.
concrete-representational-abstract sequence with integrated *Navarro, J. I., Aguilar, M., Marchena, E., Ruiz, G., Menacho,
strategy instruction to teach subtraction with regrouping to I., & Van Luit, J. E. (2012). Longitudinal study of low and
students with learning disabilities. Learning Disabilities high achievers in early mathematics. British Journal of
Research and Practice, 27, 152–166. doi:10.1111/j.1540- Educational Psychology, 82, 28–41. doi:10.1111/j.2044-
5826.2012.00363.x 8279.2011.02043.x
*Martin, R. B., Cirino, P. T., Barnes, M. A., Ewing-Cobbs, L., Nunes, T., Bryant, P., Burman, D., Bell, D., Evans, D., & Hallett,
Fuchs, L. S., Stuebing, K. K., & Fletcher, J. M. (2013). D. (2009). Deaf children’s informal knowledge of multiplica-
Prediction and stability of mathematics skill and dif- tive reasoning. Journal of Deaf Studies and Deaf Education,
ficulty. Journal of Learning Disabilities, 46, 428–443. 14, 260–277. doi:10.1093/deafed/enn040
doi:10.1177/0022219411436214 O’Connor, R. E., Bocian, K. M., Beach, K. D., Sanchez, V., &
*Mazzocco, M. M. M., & Devlin, K. T. (2008). Parts and “holes”: Flynn, L. J. (2013). Special education in a 4-year response
Gaps in rational number sense among children with vs. to intervention (RtI) environment: Characteristics of students
without mathematical learning disabilities. Developmental with learning disability and grade of identification. Learning
Science, 11, 681–691. doi:10.1111/j.1467-7687.2008.00717.x Disability Research and Practice, 28, 98–112. doi:10.1111/
*Mazzocco, M. M., Feigenson, L., & Halberda, J. (2011). Impaired ldrp.12013
acuity of the approximate number system underlies mathe- *Peng, P., Namkung, J. M., Fuchs, D., Fuchs, L. S., Patton, S., Yen,
matical learning disability (dyscalculia). Child Development, L., . . . Hamlett, C. (2016). A longitudinal study on predictors
82, 1224–1237. doi:10.1111/j.1467-8624.2011.01608.x of early calculation development among young children at
*Mazzocco, M. M. M., & Myers, G. F. (2003). Complexities in risk for learning difficulties. Journal of Experimental Child
identifying and defining mathematics learning disability in Psychology, 152, 221–241. doi:10.1016/j.jecp.2016.07.017
Powell, S. R., Cirino, P. T., & Malone, A. S. (in press). Child-level risk and not at risk for serious math difficulties. Journal of
predictors of responsiveness to evidence-based mathematics Educational Psychology, 100, 343–379. doi:10.1037/0022-
intervention. Exceptional Children. 0663.100.2.343
Powell, S. R., Fuchs, L. S., Fuchs, D., Cirino, P. T., & Fletcher, Swanson, H. L., Moran, A., Lussier, C., & Fung, W. (2014).
J. M. (2009). Do word-problem features differentially affect The effect of explicit and direct generative strategy training
problem difficulty as a function of students’ mathematics diffi- and working memory on word problem-solving accuracy
culty with and without reading difficulty? Journal of Learning in children at risk for math difficulties. Learning Disability
Disabilities, 42, 99–110. doi:10.1177/0022219408326211 Quarterly, 37, 111–123. doi:10.1177/0731948713507264
Powell, S. R., Nelson, G., & Peng, P. (2017). Mathematics trajec- Tolar, T. D., Fuchs, L., Fletcher, J. M., Fuchs, D., & Hamlett,
tories from preschool to postsecondary: The predictive nature C. L. (2016). Cognitive profiles of mathematical problem
of mathematics performance. Manuscript submitted for pub- solving learning disability for different definitions of dis-
lication. ability. Journal of Learning Disabilities, 49, 240–256.
Rubinsten, O., & Tannock, R. (2010). Mathematics anxiety in doi:10.1177/0022219414538520
children with developmental dyscalculia. Behavioral and Toll, S. W. M., & Van Luit, J. E. H. (2012). Early numeracy inter-
Brain Functions, 6(46), 1–13. doi:10.1186/1744-9081-6-46 vention for low-performing kindergarteners. Journal of Early
Scanlon, D. (2013). Specific learning disability and its new- Intervention, 34, 243–264. doi:10.1177/1053815113477205
est definition: Which is comprehensive? And which is *Toll, S. W. M., & Van Luit, J. E. H. (2014). The developmental
insufficient? Journal of Learning Disabilities, 46, 26–33. relationship between language and low early numeracy skills
doi:10.1177/0022219412464342 throughout kindergarten. Exceptional Children, 81, 64–78.
*Schwenck, C., Dummert, F., Endlich, D., & Schneider, W. doi:10.1177/0014402914532233
(2015). Cognitive functioning in children with learning prob- *Vanbinst, K., Ghesquière, P., & De Smedt, B. (2014). Arithmetic
lems. European Journal of Psychology of Education, 30, strategy development and its domain-specific and domain-
349–367. doi:10.1007/s10212-014-0242-5 general cognitive correlates: A longitudinal study in children
Shalev, R. S., Auerbach, J., Manor, O., & Gross-Tsur, V. (2000). with persistent mathematical learning difficulties. Research
Developmental dyscalculia: Prevalence and prognosis. in Developmental Disabilities, 35, 3001–3013. doi:10.1016/j.
European Child and Adolescent Psychiatry, 9, II/58–II/64. ridd.2014.06.023
doi:10.1007/s007870070009 *Vukovic, R. K. (2012). Mathematics difficulty with and with-
Starr, A., Libertus, M. E., & Brannon, E. M. (2013). Number out reading difficulty: Findings and implications from a four-
sense in infancy predicts mathematical abilities in child- year longitudinal study. Exceptional Children, 78, 280–300.
hood. Proceedings of the National Academy of Sciences, 110, doi:10.1177/001440291207800302
18116–18120. doi:10.1073/pnas.1302751110 *Vukovic, R. K., & Siegel, L. S. (2010). Academic and cogni-
*Stock, P., Desoete, A., & Roeyers, H. (2010). Detecting children tive characteristics of persistent mathematics difficulty from
with arithmetic disabilities from kindergarten: evidence from first through fourth grade. Learning Disabilities Research and
a 3-year longitudinal study on the role of preparatory arithme- Practice, 25, 25–38. doi:10.1111/j.1540-5826.2009.00298.x
tic abilities. Journal of Learning Disabilities, 43, 250–268. Watson, S. M. R., & Gable, R. A. (2012). Unraveling the com-
doi:10.1177/0022219409345011 plex nature of mathematics learning disability: Implications
Swanson, H. L. (2012). Cognitive profile of adolescents with math for research and practice. Learning Disability Quarterly, 36,
disabilities: Are the profiles different from those with reading 178–187. doi:10.1177/0731948712461489
disabilities? Child Neuropsychology, 18, 125–143. doi:10.10 Watts, T. W., Duncan, G. J., Siegler, R. S., & Davis-Kean, P. E.
80/09297049.2011.589377 (2014). What’s past is prologue: Relations between early math-
Swanson, H. L., & Beebe-Frankenberger, M. (2004). The relation- ematics knowledge and high school achievement. Educational
ship between working memory and mathematical problem Researcher, 43, 352–360. doi:10.3102/0013189X14553660
solving in children at risk and not at risk for serious math Wei, X., Lenz, K. B., & Blackorby, J. (2013). Math growth tra-
difficulties. Journal of Educational Psychology, 96, 471–491. jectories of students with disabilities: Disability category,
doi:10.1037/0022-0663.96.3.471 gender, racial, and socioeconomic status differences from
*Swanson, H. L., Jerman, O., & Zheng, X. (2008). Growth in work- ages 7 to 17. Remedial and Special Education, 34, 154–165.
ing memory and mathematical problem solving in children at doi:10.1177/0741932512448253

EJ1192397

Uploaded by

Copyright:

Available Formats

EJ1192397

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

EJ1192397

Uploaded by

Copyright:

Available Formats

714773

A Systematic Review of Longitudinal

Gena Nelson, MA1, and Sarah R. Powell, PhD2

Research Question 1: What are the characteristics of lon- Inclusion Criteria

Math difficulty 1 Math difficulty 2 Other difficulty Comparison 1 Comparison 2

You might also like