International Electronic Journal of Elementary Education

January 2022, Volume 14, Issue 3, 335-352

Predicting Mathematical Learning

Difficulties Status: The Role of
Domain-Specific and Domain-
General Skills
Riikka Mononena,*, Markku Niemivirtab, Johan Korhonenc
Received : 16 November 2021 Abstract
Revised : 1 December 2021
Accepted : 28 January 2022 This study investigated which domain-specific and domain-
DOI : 10.26822/iejee.2022.248 general skills measured at grade 1 predict mathematical
learning difficulties (MLD) status at grade 3. We used different
a,*
Corresponding Author: Riikka Mononen, Teachers,
Teaching and Educational Communities, University cut-off criteria and measures of mathematics performance
of Oulu, Finland and Department of Special Needs for defining the MLD status. Norwegian children’s (N = 206)
Education, University of Oslo, Norway numeracy, cognitive, and language skills were measured
E-mail: [email protected], [email protected]
ORCID: https://orcid.org/0000-0003-2883-5854 at grade 1 and arithmetic fluency and curriculum-based
mathematics (CBM) at grade 3. Logistic regression analyses
b
Markku Niemivirta, School of Applied Educational showed that symbolic numerical magnitude processing,
Science and Teacher Education, University of Eastern
Finland, Finland verbal counting, and rapid automatized naming predicted
E-mail: [email protected] MLD25 status (performance ≤ 25th percentile) based on
ORCID: https://orcid.org/0000-0001-7152-5152 arithmetic fluency, whereas verbal counting skills and
c
Johan Korhonen, Faculty of Education and Welfare nonverbal reasoning predicted the status based on CBM.
Studies, Åbo Akademi University, Vaasa, Finland The same predictors were found for MLD10 status
E-mail: [email protected] (performance ≤ 10th percentile), and in addition, rapid
ORCID: https://orcid.org/0000-0001-9666-9400
automatized naming also predicted the status based on
CBM. Only symbolic numerical magnitude processing
and verbal counting predicted LOW status (performance
between 11–25th percentile) based on arithmetic fluency,
whereas nonverbal reasoning and working memory
predicted LOW status based on CBM. Different cut-off
scores and mathematics measures used for the definition
of MLD status are important to acknowledge, as these seem
to lead to relatively significant variation in which students
are identified as having MLD and which factors contribute
to the MLD status.

Keywords:
Arithmetic, Counting, Mathematical Learning Difficulties,
Nonverbal Reasoning, Rapid Automatized Naming

Introduction

Copyright ©
www.iejee.com
Individual differences in mathematics learning and
performance in the beginning of schooling are well-
acknowledged (Aunio & Niemivirta, 2010; Jordan et al.,
ISSN: 1307-9298
2009; ten Braak et al., 2022; Zhang et al., 2020). Children
© 2022 Published by KURA Education & Publishing.
in early grades are rarely formally diagnosed as having
This is an open access article under the CC BY- developmental learning disorder in mathematics, also
NC- ND license. (https://creativecommons.org/ called dyscalculia, until the effect of teaching has been
licenses/by/4.0/) taken into account (ICD-11; World Health Organization,

335
 January 2022, Volume 14, Issue 3, 335-352

2019). Nonetheless, we can reliably identify students On one hand, it has been suggested that deficit
who perform weaker in mathematics than their peers in either numerical magnitude processing (NMP)
and show difficulties in their mathematics learning. (Butterworth, 2005) or accessing the magnitude in
In research literature, different cut-off criteria for symbols (Rousselle & Noël, 2007) underlies the most
performance, ranging from 10th to 35th percentile, severe MLD (but see Mammarella et al., 2021 for no
have been used to identify students at risk for or having evidence for a core deficit in MLD). On the other hand,
mathematical learning difficulties (MLD) (e.g., Aunio et a persistent low performance in mathematics has
al., 2021; Kroesbergen & van Dijk, 2015; Lin et al., 2021; been suggested to stem from having weaknesses in
Mazzocco et al., 2013). A wide cut-off range used for domain-general cognitive skills (e.g., working memory)
the definition of MLD leads to an inclusion of a variety or lacking mathematics motivation (Geary, 2011; Price
of performance within MLD; the lower end showing & Ansari, 2013). By contrast, the double deficit model
more severe difficulties than the upper end. Not until suggests that weaknesses in both NMP and working
recently have the researchers started to show more memory would be associated with the most severe
consensus on the terms and cut-off criteria of MLD. MLD (Kroesbergen & van Dijk, 2015). While there are
Individuals performing at the lowest 10th percentile competing theories about the associations between
are commonly referred to as having mathematical domain-specific and domain-general skills and MLD,
disability/disorder or developmental dyscalculia, and whether these differ between the subtypes of
whereas those performing between the 11th and MLD (Geary et al., 2012; Huijsmans et al., 2022; Tolar
25th percentile are referred to as low-performing/ et al., 2016), the empirical evidence has mostly relied
achieving (Geary, 2011). The term mathematical on cross-sectional data, thus focusing only on the
learning difficulties, independent of the severity of concurrent relations between the predictors and
MLD, can then be used as an umbrella term for all MLD status (e.g., Cañizares et al., 2012; Passolunghi
those performing at or below the 25th percentile. & Mammarella, 2010; Tolar et al., 2016; Van Luit & Toll,
2018; Willburger et al., 2008). Longitudinal data would
In addition to the severity criteria, different types seem more accurate in identifying the most relevant
of mathematics measures are used in research to domain-specific and domain-general predictors of
identify students with MLD. Measures of arithmetic MLD status.
performance are typically applied (Cañizares et al.,
2012; Koponen, Aro, et al., 2018), as poor arithmetic Our study adds to the current research by investigating
fluency is one of the main characteristics of students which domain-specific and domain-general skills
having MLD (Geary, 2011; Gersten et al., 2005). measured in the first grade predict MLD status in the
Alternatively, broad mathematics performance third grade. The novelty of our study lies in taking
measures are used (Jordan et al., 2002). Sometimes into consideration different cut-off criteria and
the definition of MLD includes a persistence criterion, mathematics measures (i.e., arithmetic fluency and
meaning that the student needs to perform low in curriculum-based mathematics [CBM]) in defining
at least two consecutive time points (Mazzocco the MLD. Further, we include simultaneously several
et al., 2013; Stock et al., 2010). Currently, we have domain-specific (i.e., symbolic numerical magnitude
little knowledge of the extent to which the different processing [SNMP] and counting skills) and domain-
mathematics measures overlap in defining MLD. general skills (i.e., nonverbal reasoning, working
That is, would a brief arithmetic fluency measure memory, rapid automatized naming, and vocabulary)
identify the same students as a broader mathematics as predictors of MLD status. To examine if domain-
performance test? specific and domain-general skills predict MLD status
differently when using different cut-off criteria based
Researchers have also been curious about which on the sample-based percentiles, we first divide the
domain-specific and domain-general skills are related students into two groups: those who perform at or
to or may contribute to MLD, and whether these differ below the 25th percentile are referred to as having
between the subgroups of MLD (Geary et al., 2012; mathematical learning difficulties (MLD25), while rest
Huijsmans et al., 2022; Salihu & Räsänen, 2018). A set of of the students are referred to as typically performing
domain-specific skills, such as numerical magnitude (performance over the 25th percentile). Next, we
processing (Cañizares et al., 2012; De Smedt & Gilmore, divide the MLD25 group into those who are showing
2011) and counting skills (Koponen et al., 2019), and more severe difficulties, namely mathematical
domain-general skills, such as working memory learning disorder (performance at or below the 10th
(Menon, 2016; Passolunghi & Mammarella, 2010) and percentile, MLD10) and those who are low performing
rapid automatized naming (Van Luit & Toll, 2018), have (performance between the 11th to 25th percentile,
been found to be related to MLD. However, drawing LOW). The status of MLD is based separately on
conclusions about their predictive role becomes arithmetic fluency and on CBM. This enables us
complicated due to the different cut-off criteria for to examine the overlap of MLD statuses based on
MLD and various measures used for the identification arithmetic fluency and CBM (e.g., how many students
of students with MLD. are classified as MLD25 in both arithmetic fluency and

336
 Predicting Mathematical Learning Difficulties Status / Mononen, Niemivirta & Korhonen

CBM), and whether predictors of MLD status vary as a retrieve the answer from long-term memory (Ostad,
function of mathematics measure (arithmetic fluency 1998). Counting skills have been found to be associated
vs. CBM) and cut-off criteria (10th vs. 25th percentile). with (Lopez-Pedersen et al., 2021) and to predict later
arithmetic and mathematics performance (Aunio &
Domain-Specific Skills as Predictors of MLD Niemivirta, 2010; Koponen et al., 2019). Furthermore,
children performing low (Hassinger-Das et al., 2014; Toll
Domain-specific skills in mathematics context mean & Van Luit, 2014) and with severe MLD (Landerl et al.,
different mathematical skills. Here, we focus on two 2004) have shown weaker counting skills compared to
such skills, numerical magnitude processing (NMP) and their peers.
counting skills, which have shown to be related to or to
predict later mathematics performance, concerning Domain-General Skills as Predictors of MLD
both the typically performing students and students
with MLD. NMP is considered to be an innate ability, Several domain-general skills, such as working memory,
which enables individuals to approximately process executive functions, rapid automatized naming, and
numerical magnitudes (Dehaene et al., 1998). A language, have been found to be associated with
traditionally used task for measuring NMP is to mathematics performance and development (Chu
compare two sets of dots, and to quickly decide which et al., 2016; Friso-van den Bos et al., 2013; Koponen,
side has more dots. Even if some studies have shown Eklund, et al., 2018; Purpura & Ganley, 2014), and
that students with MLD often have poorer NMP skills with MLD (Mammarella et al., 2021; Passolunghi &
compared to their peers without MLD (Mazzocco et Mammarella, 2010; Purpura et al., 2017; Van Luit & Toll,
al., 2011; Mussolin et al., 2010), there is recent evidence 2018). Further, separate domain-general cognitive
showing that symbolic NMP (SNMP) would be even a profiles depending on the severity level of MLD has
better predictor of mathematics performance and been suggested (Geary et al., 2012). However, in a
more strongly associated with MLD than non-symbolic recent study, Hujsman et al. (2022) did not find support
NMP (Cañizares et al., 2012; De Smedt & Gilmore, 2011; for this. In their study, the severity of MLD did not result
Desoete et al., 2012; Nosworthy et al., 2013). For this in differences in the cognitive profiles, and further, the
reason, we focus in this study on the role of SNMP. The cognitive profiles for mathematics development from
tasks often used to measure SNMP are similar to NMP fourth to fifth grade were rather similar between the
comparison tasks, but 1- or 2-digit numbers are used students with MLD and typically performing students.
as stimuli instead of dots (Brankaer et al., 2017). While
comparing the numbers, and to choose the bigger Nonverbal reasoning (also called as nonverbal or
number, the student needs to access the magnitudes fluid intelligence) has been shown to be a consistent
of those numbers. A slow response indicates a deficit predictor of mathematics performance in different
in accessing and processing of the magnitudes age groups of students (Kyttälä & Lehto, 2008; Pina
(Rousselle & Noël, 2007). Concerning school beginners, et al., 2014). Good nonverbal reasoning skills are
Desoete et al. (2012) found that SNMP measured in an advantage in solving mathematical problems,
kindergarten (5–6 years old) was a good predictor because students need to be able to make logical
of procedural calculations (i.e., 34 + 21, or given in as decisions and to proceed systematically with the task
“6 more than 48 is…”) in the second grade. Further, (e.g., entertain their solutions and if proven false try a
they found that although children with MLD showed new solution) (Engle, 2018). Students with MLD typically
weakness in both non-symbolic and symbolic NMP in perform weaker in nonverbal reasoning, albeit within
kindergarten, it was only in SNMP they showed a deficit normal range, compared to their peers without MLD
in grade 2. Overall, the findings are still mixed when it (Huijsmans et al., 2022).
comes to the role of (S)NMP and severity level of MLD.
Some research supports that students with severe MLD Solving mathematics tasks requires working
are characterized with poor NMP (Mazzocco et al., memory capacity for storing and manipulating
2011), while some research has not found a difference information temporarily. Working memory and its
between the subgroups of MLD (i.e., MLD10 and LOW) different components (i.e., visuo-spatial sketchpad,
(Huijsmans et al., 2022) phonological loop, central executive) have all been
linked with mathematics performance (Friso-van den
Counting skills (i.e., verbal number sequence skills and Bos et al., 2013). Students with MLD have been reported
object counting) develop typically parallel with early to show weaker working memory performance than
arithmetic skills (i.e.., addition and subtraction), and can their peers (Kroesbergen & van Dijk, 2015; Passolunghi &
also be seen as prerequisites for learning arithmetic Mammarella, 2010). Huijsmans et al. (2022) found that
skills, because children often use these as their those with low performance were characterized by
strategies in solving arithmetic calculation problems difficulties in visual working memory, but interestingly,
(Koponen et al., 2019). Typically, children with MLD not those with severe MLD. However, working memory
use more immature, counting-based strategies, while did not explain any variance in the mathematics
their peers use retrieval strategies, that is, they quickly development from fourth to fifth grade (Huijsmans

337
 January 2022, Volume 14, Issue 3, 335-352

et al., 2022). Our study, instead, focuses on the disorder. In general, these students have consistently
central executive component of working memory shown weaker mathematics performance compared
(i.e., manipulation of information). Prior research has to their peers in mathematical tasks that require
shown that students with MLD have a deficit in their expressing or understanding of language (e.g., verbal
central executive functioning, and especially when number sequences, counting of objects, arithmetic)
central executive has been measured with numerical while they have shown similar performance to their
stimuli (Andersson & Lyxell, 2007; for a meta-analysis TYP peers in mathematics tasks with less demand on
see David, 2012). language (e.g., NMP, conceptual mathematics tasks)
(for a review see Cross et al., 2019). Not many studies
Rapid automatized naming (i.e., quickly naming have investigated the role of different components
familiar non-alphanumeric objects such as colors of language among students with MLD. However,
and figures, or alphanumeric objects such as letters recently, Chow et al. (2021) showed that students
or numbers) has been found to be more strongly with MLD (performance below 20th percentile on
related to arithmetic fact retrieval than to more arithmetic fluency) performed lower than their peers
general mathematics performance (for a meta- in receptive vocabulary, morphology, and syntax.
analysis see Koponen et al., 2017). Rapid automatized
naming (e.g., naming a color) and fact retrieval (e.g., Present Study
5 + 3) both require quick access to and retrieval of
phonological representations from long-term memory The present study expands on previous research by
(“blue” and “eight”, respectively) (Koponen et al., 2017). investigating how domain-specific and domain-
Concerning students with MLD, Donker et al. (2016) general skills measured in the first grade predict MLD
found that primary school-aged students had weaker status among third graders when different cut-off
performance in non-alphanumeric rapid automatized criteria and measures of mathematics performance
naming (e.g., colors) compared to their peers, but not in are used. Our research questions are as follows:
alphanumeric format (e.g., letters). Further, Mazzocco
and Grimm (2013) found in their longitudinal study from (RQ1) What is the overlap of MLD statuses based
kindergarten to grade 8 that those with LOW showed on arithmetic fluency and curriculum-based
slight delays (i.e., slower response times) in naming of mathematics?
colors compared to their TYP peers, whereas those
with severe MLD showed more persistent weakness (RQ2) How do domain-specific (i.e., symbolic
in rapid naming of letters and colors. Based on prior numerical magnitude processing and verbal counting
findings that rapid naming of colors can be a good skills) and domain-general (non-verbal reasoning,
early predictor of mathematics performance and rapid automatized naming, working memory, and
especially arithmetic fact retrieval (Koponen et al., vocabulary) skills predict MLD status when using the
2017), and associated with MLD (Donker et al., 2016; 25th percentile cut-off criterion (MLD25) based on
Mazzocco & Grimm, 2013), in our study, we chose either arithmetic fluency (RQ2.1) or curriculum-based
to have rapid naming of colors as a proxy for rapid mathematics (RQ2.2)?
automatized naming.
(RQ3) How do domain-specific and domain-general
In early childhood, language skills have been linked skills predict MLD status when further dividing
with mathematics performance (Aunio et al., 2019), the MLD25 into MLD10 (≤ 10th percentile) and low
and especially mathematics related language to performers (LOW; 11–25th percentile) based on
be a good predictor of low performance (Purpura et either arithmetic fluency (RQ3.1) or curriculum-based
al., 2017), as well as influencing the development of mathematics (RQ3.2)?
early mathematical skills of LOW (Toll & Van Luit, 2014).
In our study, we focus on vocabulary (expressive) in Although we have limited evidence available to
general, which has shown to play a role in children’s strongly guide our hypothesis for RQ1, we expect
mathematics learning (LeFevre et al., 2010). However, relatively high overlap of MLD statuses based on
also conflicting results have been reported among different mathematics measures, but still distinct
school beginners, that is, no connection between to a certain degree, as the mathematics content in
vocabulary (receptive) and mathematics performance arithmetic fluency is much more limited than in the
(Chow & Ekholm, 2019). Vocabulary is needed not only broad CBM measure (H1).
to understand mathematics teaching in general, but
also to communicate using different mathematics Based on prior research, we hypothesize that all
concepts (e.g., comparison words, number words, domain-specific and domain-general skills under
geometrical object, words for operations [e.g., plus, investigation are likely predictors of MLD25 (H2), as this
minus]. Support for the importance of language in group encompasses those with more severe learning
mathematics learning comes from studies, which difficulties (MLD10) and milder learning difficulties
have included students with developmental language (LOW). Regarding MLD25 based on arithmetic fluency,
we expect SNMP (Desoete et al., 2012), counting skills

338
 Predicting Mathematical Learning Difficulties Status / Mononen, Niemivirta & Korhonen

(Hassinger-Das et al., 2014; Landerl et al., 2004) and A curriculum-based mathematics (CBM) test was
rapid automatized naming (Koponen et al., 2017; developed in the project (Mononen, 2021) to measure
Mazzocco & Grimm, 2013) to be significant predictors children’s overall performance in mathematics
(H2.1). Since the tasks in CBM have more variety and taught in grade 3. This paper-pencil group-based
complexity regarding their mathematics content test includes 49 items from the topics of numbers
and procedures, thus requiring logical reasoning (number sequences, comparison of multi-digit
(Engle, 2018), executive functioning (David, 2012), numbers), measurement (volume, length, money),
and understanding task related vocabulary (Chow calculations (multiplication facts, addition and
et al., 2021), we anticipate MLD25 based on CBM to subtraction algorithms) and fractions, and follows
be predicted by counting skills (Hassinger-Das et al., the learning outcomes set for the third grade in the
2014), nonverbal reasoning, working memory, and national mathematics curriculum (ref.). Each task
vocabulary (H2.2). was instructed for the children and children worked
with the tasks independently for 20–25 minutes. Each
Similar to MLD25 based on arithmetic fluency, we correctly solved item gave one point.
expect the two domain-specific skills and rapid
automatized naming to predict both MLD10 and First-grade domain-specific numeracy skills
LOW status (H3.1). As to the status based on CBM, we
presume domain-general skills, especially concerning Symbolic numerical magnitude processing was
working memory (David, 2012; Huijsmans et al., measured using the 1-digit subtest of the SYMP test
2022), to exhibit different predictions on MLD10 and (Brankaer et al., 2017). In this paper-pencil test the child
LOW (David, 2012; Huijsmans et al., 2022). Further, we has 30 seconds to compare as many 1-digit number
expect the significant predictors to include counting pairs as possible out of 60, by choosing the bigger
skills (Hassinger-Das et al., 2014; Landerl et al., 2004), number. One point is given for a correct answer, thus
nonverbal reasoning (Huijsmans et al., 2022), and the maximum score being 60.
vocabulary (Chow et al., 2021) (H3.2).
Verbal counting skills were measured using a normed
Method Finnish LukiMat subscale (Salminen & Koponen, 2011),
which was translated into Norwegian. The child was
Participants asked to orally count number sequences forwards and
backwards, in steps of 1, 2, 5, and 10. Each correctly
The current study is part of a research project that given number sequence gave one point, the possible
follows Norwegian children’s numeracy development maximum total score being 29 points.
from first to third grade. Here, we use data from its
first (grade 1, t1) and last (grade 3, t2) measurement First-grade domain-general skills
time points. The final sample of participants was 206
children (Mage = 6 y. 9 m., SD = 3.4 m., girls 49%), from Nonverbal reasoning skills were measured using
four schools located in the Oslo region, and who had Raven’s progressive colored matrices (Raven et al.,
data available from both time points. Due to Covid-19 1990). The child chooses one of the six alternative
restrictions in schools in spring 2021, 27 children from pieces that fits the picture. One point was given for
the initial sample of 265 were not able to participate each correct answer, the maximum possible score
in t2, and 32 children had either moved away or were being 34, as two first items were practice items.
absent from school on the data collection day. An
ethical approval was given by the Norwegian Centre Working memory was measured using a digit span
for Research Data before the data collection started, backwards subtest from WISC-V (Wechsler, 2017).
and consents for the participation were given by Digit span backwards captures the central executive
children’s legal guardians. component of WM, as modelled by Baddeley
(Baddeley & Logie, 1999). Each digit span, ranging
Measures from 2–8 numbers, had two tasks, except for 2- and
3-digit spans having 4 tasks each. Each digit span was
Third-grade mathematics performance presented orally to the child forwards and the child
needed to repeat the digits backwards. Following the
Arithmetic fluency was measured using a standardized test guidelines, the test was stopped if the child could
arithmetic test Regnefaktaprøven (Klausen & Reikerås, not give a correct answer for both tasks with the same
2016). Children have 2 minutes to solve as many number of digits. The maximum total score was 18
addition problems as possible out of 45 on one sheet, points.
and same for subtraction. As a proxy of arithmetic
fluency, we combined the sum scores of each subtest, Rapid automatised naming was measured using a
thus the maximum possible score being 90 points. colors subtest from the Clinical Evaluation of Language
Fundamentals (CELF-4) (Semel et al., 2003). The child
needs to name 36 colored dots (including colors “gul”

339
 January 2022, Volume 14, Issue 3, 335-352

[yellow], “grønn” [green], “blå” [blue], “rød” [red]) as and those performing typically (TYP). The second
accurately and fast as possible. For the purpose of grouping variable for arithmetic fluency included
statistical analyses, we created a composite score, MLD10, LOW and TYP. Similar grouping variables were
in which the correct number of named colors was created for CBM. To answer for RQ1, we tested with
divided by used time, and multiplied by ten. a chi-square test what is the overlap of MLD statuses
based on arithmetic fluency and curriculum-based
Vocabulary was measured using a subtest mathematics (e.g., children having a status of LOW
‘Ordforstålse’ of WISC-V (Wechsler, 2017) targeting in both arithmetic fluency and CBM). To answer for
expressive vocabulary. A child needs to either name RQ2 and RQ3, which first-grade domain-specific
a picture (first 4 items, 1 point for a correct answer), or and domain-general skills predict MLD status, we
explain the meaning of a word (25 items, giving either conducted two binary logistic regression analyses,
1 or 2 points depending on the correctness of the one for arithmetic fluency and one for CBM, when
definition based on the test guidelines). A maximum having two status groups (MLD25 and TYP), and two
score for the task was 54 points. multinomial logistic regression analyses when having
three status groups (MLD10, LOW and TYP). Jamovi
Procedure 2.2.2.0 software (The jamovi project, 2021) was used for
statistical data analyses.
The first data collection (t1) took place in spring 2019.
Children came in groups of 8–10 students to the Results
data collection site for half a school day. During this
“adventure day” the children completed a set of Descriptive statistics of and correlations between
measures individually and in small groups depending the variables are reported in Table 1. SNMP showed
on the test format, together with trained research a stronger relation with arithmetic fluency (r = .52)
assistants. Small breaks were held between the than with CBM (r = .30), whereas counting skills had
sessions. The data collection in the third grade (t2) a moderate relation with both (r = .55 and r = .49,
took place in spring 2021. The Covid-19 measures respectively). The associations between domain-
set by the Norwegian government restricted the general skills and arithmetic fluency were weak
data collection so that we were not allowed to visit with correlations ranging from r = .23 to r = .43,
the schools. Instead, one research assistant gave all rapid automatized naming showing the strongest
test instructions online via Teams to one classroom relation, and vocabulary the weakest. Regarding the
at a time, and the students in the classroom were relations with CBM, the strongest association was
overlooked by their classroom teacher. The testing with nonverbal reasoning, r = . 43, and weakest with
was done in two sessions of around 45 minutes each, rapid automatized naming, r = .21. Multicollinearity
during one day. The research assistants had a video was unlikely as all correlations between the predictors
and audio connections to the classroom. No such were moderate at best.
technical issues were reported that would have
violated the testing situation and validity of the data. Means and standard deviations on arithmetic fluency
Test booklets were delivered to the schools a few days and CBM by each status group are reported in Table 2.
before the data collection took place and collected As to RQ1, the chi-square test of the crosstabulation of
after the data collection was completed. MLD25 and TYP based on arithmetic fluency and CBM
was significant, χ2 (1) = 59.40, p <.001. Sixty-five percent
The data was coded by trained research assistants, of children were observed as MLD25 in both arithmetic
and data from three randomly chosen students per fluency and CBM, while 88% of children as TYP. Similarly,
classroom (13%) were double coded by the first author. the crosstabulation of three groups, MLD10, LOW, and
The correlations between the first and second coding TYP, based on arithmetic fluency and CBM turned out
resulted in correlations of sum scores ranging from r to be significant, χ2 (4) = 104.01, p < .001. Sixty-seven
= .944–1.00, with coding errors connected to some percent of the children were observed as MLD10 in
children having few items in a test with non-matching both arithmetic fluency and CBM, whereas only 39%
values. When needed, the test papers compared of LOW, and 88% of TYP. These results confirmed that
to the punched values, and the final data matrix a series of separate logistic regression analyses for
corrected accordingly. arithmetic fluency and CBM would be justified.

Data analysis Predictors of MLD25 Status

First, descriptive statistics of and correlations between Arithmetic fluency (RQ2.1)

all variables were calculated. Second, grouping
variables for the MLD status based on sample- A binary logistic regression analysis was performed
based percentiles both in third grade arithmetic to ascertain the effects of domain-specific skills
fluency and CBM were created. The first grouping (i.e., SNMP and counting skills) and domain-general
variable for arithmetic fluency included MLD25 status skills (i.e., nonverbal reasoning, working memory,

340
 Predicting Mathematical Learning Difficulties Status / Mononen, Niemivirta & Korhonen

Table 1
Descriptive Statistics of and Correlations between the Variables
1. 2. 3. 4. 5. 6. 7. 8.
1. SNMP t1 —
2. Counting t1 0.46*** —
3. Nonverbal reasoning t1 0.25*** 0.28*** —
4. Working memory t1 0.25*** 0.41*** 0.31*** —
5. Rapid naming t1 0.42*** 0.29*** 0.23** 0.33*** —
6. Vocabulary t1 0.13 0.28*** 0.21** 0.21** 0.13 —
7. Arithmetic fluency t2 0.52*** 0.55*** 0.28*** 0.39*** 0.43*** 0.23*** —
8. CBM t2 0.30*** 0.49*** 0.43*** 0.34*** 0.33*** 0.21** 0.61*** —
M 17.39 20.22 24.18 6.26 10.00 13.45 36.58 34.76
SD 4.13 5.80 5.16 1.74 2.47 3.23 16.90 9.22
Min-Max 7–29 0–29 8–34 0–11 2.09–18.95 2–24 2–89 5–48
Skewness 0.06 -0.89 -0.32 -0.03 0.08 -0.02 0.52 -1.05
Kurtosis –0.06 0.51 -0.39 0.53 1.32 1.09 0.33 0.89
Cronbach’s α .887 .914 .934 .692 .983 .747 .971 .918

Note. ** p < .01, *** p < .001. t1 = time point 1 (grade 1), t2 = time point 2 (grade 3), SNMP = Symbolic numerical magnitude
processing, CBM = Curriculum-based mathematics

Table 2
Means and Standard Deviations for Arithmetic Fluency and Curriculum-Based Mathematics (CBM) by Each
Status Group

TYP MLD25 LOW MLD10

M M M M
(SD) (SD) (SD) (SD)
n n n n
43.29 16.69 21.16 10.10
Arithmetic fluency t2 a
(13.78) (6.28) (2.05) (4.21)
(max. 90 p.)
154 52 31 21
39.14 21.81 26.60 15.27
CBM t2 b
(4.68) (6.85) (1.65) (5.72)
(max. 49 p.)
154 52 30 22

Note. TYP = typically performing (performance >25th percentile), MLD25 = mathematical learning difficulties (performance ≤
25th percentile), LOW = low-performing (performance between 11–25th percentile), MLD10 = mathematical learning disorder
(performance ≤ 10th percentile). a The participants in each group are based on their performance on Arithmetic fluency. b The
participants in each group are based on their performance on CBM. The LOW and MLD10 include the same participants as
the MLD25.

rapid automatized naming, and vocabulary) on the predicted the MLD25 status: SNMP (B = -.18, p = .005,
likelihood that participants have MLD25 status. The odds ratio = .83), counting skills (B = -.14, p = .001, odds
logistic regression model was statistically significant, χ2 ratio = .87), and rapid automatized naming (B = -.22,
(6) = 61.69, p < .001, and explained 42.0% (Nagelkerke p = .035, odds ratio = .80). Decreasing performance
R2) of the variance in MLD status (Table 3). The model in these three skills was associated with increasing
correctly classified 83.0% of cases. With a cutoff set likelihood of MLD25 status. These are illustrated in
at 0.5, the prediction for children with TYP status was Figure 1.
more accurate (94.9%) than those with MLD25 (50.0%).
Two domain-specific and one domain-general skill

341
 January 2022, Volume 14, Issue 3, 335-352

Table 3
Logistic Regression Analysis for MLD25 Status on Arithmetic Fluency

95% CI for OR

Predictor B SE Z p OR Lower Upper

Intercept 6.69 1.64 4.07 < .001 NA NA NA

SNMP t1 -0.18 0.06 -2.84 0.005 0.83 0.73 0.94

Counting t1 -0.14 0.04 -3.20 0.001 0.87 0.80 0.95

Nonverbal reasoning t1 0.02 0.04 0.46 0.643 1.02 0.94 1.11

Working memory t1 -0.11 0.13 -0.83 0.408 0.90 0.69 1.16

Rapid naming t1 -0.22 0.10 -2.11 0.035 0.80 0.66 0.98

Vocabulary t1 0.02 0.07 0.30 0.767 1.02 0.89 1.17

Model fit measures Deviance AIC R2N χ2 df p

149.54 163.54 0.42 61.68 6 < .001

Note. Estimates represent the log odds of MLD25 vs. TYP (reference group). MLD25 = mathematical learning difficulties
(performance ≤ 25th percentile), TYP = typically performing (performance >25th percentile). SNMP = symbolic numerical
magnitude processing, OR = odds ratio, CI = confidence interval, AIC = Akaike information criterion, R2N = Nagelkerke’s R2.

Figure 1
Predicted Probability with 95% Confidence Interval for a Status of MLD25 in Arithmetic Fluency versus a) Sym-
bolic Numerical Magnitude Processing (SNMP), b) Verbal Counting, and c) Rapid Automatized Naming

Curriculum-based mathematics (RQ2.2) Predictors of MLD10 and LOW Status

A similar binary logistic regression analysis was Arithmetic fluency (RQ3.1)

conducted for CBM as for arithmetic fluency. The
logistic regression model showed to be statistically A multinomial logistic regression analysis was run
significant, χ2 (6) = 34.89, p < .001, and explained 26.0% with a dependent variable of status consisting of
of the variance in MLD25 status (Table 4). The model three status groups (TYP, LOW, MLD10). The model
correctly classified 79.0% of cases. With a cutoff set was statistically significant, χ2 (12) = 65.01, p < .001, and
at 0.5, the prediction for children with TYP status was explained 28.0% of the variance in status (Table 5).
more accurate (94.2%) than those with MLD25 (34.8%). When MLD10 was compared to TYP, the results were
One domain-specific and one domain-general skill similar to the MLD25 status; SNMP, counting skills, and
predicted the MLD25 status: counting skills (B = -.09, p rapid automatized naming, predicted MLD10 status
= .026, odds ratio = .92) and nonverbal reasoning (B = (SNMP: B = -.22, p = .019, odds ratio = .81; counting skills: B
-.12, p = .003, odds ratio = .89). Decreasing performance = -.16, p = .008, odds ratio = .85; and rapid automatized
in these two skills was associated with a higher naming: B = -.28, p = .038, odds ratio = .75). When LOW
probability of MLD25 status, as illustrated in Figure 2. was compared to TYP, only domain-specific skills,
SNMP (B = -.17, p = .021, odds ratio = .85) and counting

342
 Predicting Mathematical Learning Difficulties Status / Mononen, Niemivirta & Korhonen

Table 4
Logistic Regression Analysis for MLD25 Status on Curriculum-Based Mathematics

95% CI for OR

Predictor B SE Z p OR Lower Upper

Intercept 4.67 1.31 3.57 < .001 NA NA NA

SNMP t1 0.02 0.06 0.36 0.715 1.02 0.91 1.14
Counting t1 -0.09 0.04 -2.23 0.026 0.92 0.85 0.99
Nonverbal reasoning t1 -0.12 0.04 -2.94 0.003 0.89 0.82 0.96
Working memory t1 -0.13 0.12 -1.06 0.287 0.88 0.69 1.12
Rapid naming t1 -0.11 0.09 -1.21 0.225 0.89 0.75 1.07
Vocabulary t1 0.02 0.06 0.27 0.788 1.02 0.90 1.15

Model fit measures Deviance AIC R2N χ2 df p

172.05 186.05 0.26 34.89 6 < .001

Note. Estimates represent the log odds of MLD25 vs. TYP (reference group). MLD25 = mathematical learning difficulties
(performance ≤ 25th percentile), TYP = typically performing (performance > 25th percentile). SNMP = symbolic numerical
magnitude processing, OR = odds ratio, CI = confidence interval, AIC = Akaike information criterion, R2N = Nagelkerke’s R2.

Figure 2
Predicted Probability with 95% Confidence Interval for a Status of MLD25 in Curriculum-Based Mathematics
versus a) Verbal Counting and b) Nonverbal Reasoning

skills (B = -.13, p = .009, odds ratio = .88) were significant (B = -.11, p = .046, odds ratio = .89). In addition, rapid
predictors for LOW status. None of the predictors were automatized naming (B = -.28, p = .033, odds ratio = .76)
significant when comparing MLD10 with LOW. Figure predicted MLD10 status. When LOW was compared to
3 illustrates how decreasing performance in SNMP TYP, again, nonverbal reasoning predicted the status
and counting skills are associated with an increased (B = -.13, p = .009, odds ratio = .88), but counting skills
likelihood of MLD10 and LOW status, and similarly for was no longer a significant predictor. Instead, working
rapid automatized naming for MLD10 status. memory predicted LOW status (B = -.33, p = .035, odds
ratio = .72). These differences were also visible when
Curriculum-based mathematics (RQ3.2) comparing MLD10 versus LOW. Decreasing working
memory skills was associated with a higher probability
A similar multinomial logistic regression analysis was of LOW status than MLD10, and vice versa for rapid
done for CBM. The model was statistically significant, automatized naming. Figure 4 illustrates the predicted
χ2 (12) = 47.12, p < .001, and explained 21% of the variance probabilities for the status of TYP, LOW, and MLD10
in status (Table 6). When MLD10 was compared to versus counting skills, nonverbal reasoning, working
TYP, the same two factors that had predicted MLD25 memory and rapid automatized naming.
predicted also MLD10 status: counting skills (B = -.15,
p = .006, odds ratio = .86) and nonverbal reasoning

343
 January 2022, Volume 14, Issue 3, 335-352

Table 5
Multinomial Logistic Regression Analysis for MLD10 and LOW Status on Arithmetic Fluency

95% CI for OR

Predictor B SE Z p OR Lower Upper

MLD10–TYP
Intercept 7.30 2.12 3.44 < .001 NA NA NA
SNMP t1 -0.22 0.09 -2.35 0.019 0.81 0.67 0.96
Counting t1 -0.16 0.06 -2.65 0.008 0.85 0.76 0.96
Nonverbal reasoning t1 -0.01 0.06 -0.16 0.876 0.99 0.88 1.11
Working memory t1 -0.09 0.19 -0.49 0.622 0.91 0.63 1.32
Rapid naming t1 -0.28 0.14 -2.07 0.038 0.75 0.58 0.98
Vocabulary t1 0.05 0.10 0.49 0.625 1.05 0.86 1.28
LOW–TYP
Intercept 5.34 1.78 2.99 0.003 NA NA NA
SNMP t1 -0.17 0.07 -2.31 0.021 0.85 0.74 0.98
Counting t1 -0.13 0.05 -2.63 0.009 0.88 0.80 0.97
Nonverbal reasoning t1 0.03 0.05 0.72 0.471 1.04 0.94 1.14
Working memory t1 -0.12 0.15 -0.80 0.421 0.89 0.66 1.19
Rapid naming t1 -0.18 0.11 -1.62 0.105 0.83 0.66 1.04
Vocabulary t1 0.01 0.08 0.10 0.923 1.01 0.86 1.18
MLD10–LOW
Intercept 1.97 1.98 0.99 0.320 NA NA NA
SNMP t1 -0.05 0.10 -0.52 0.604 0.95 0.79 1.15
Counting t1 -0.03 0.06 -0.52 0.602 0.97 0.86 1.09
Nonverbal reasoning t1 -0.04 0.07 -0.67 0.501 0.96 0.84 1.09
Working memory t1 0.03 0.20 0.12 0.901 1.03 0.69 1.53
Rapid naming t1 -0.10 0.13 -0.73 0.468 0.91 0.70 1.18
Vocabulary t1 0.04 0.10 0.40 0.689 1.04 0.85 1.27
Model fit measures Deviance AIC R2N χ2 df p

210.64 238.65 0.28 65.01 12 < .001

Note. Estimates represent the log odds of MLD10 vs. TYP (reference group), LOW vs. TYP (reference group), and MLD10 vs.
LOW (reference group). MLD10 = mathematical learning disorder (performance ≤ 10th percentile), TYP = typically performing
(performance >25th percentile), LOW = low-performing (performance between 11–25th percentile). SNMP = symbolic numerical
magnitude processing, OR = odds ratio, CI = confidence interval, AIC = Akaike information criterion, R2N = Nagelkerke’s R2.

Figure 3
Predicted Probability for a Status of MLD10, LOW and TYP in Arithmetic Fluency versus a) Symbolic Numerical
Magnitude Processing (SNMP), b) Verbal Counting, and c) Rapid Automatized Naming

344
 Predicting Mathematical Learning Difficulties Status / Mononen, Niemivirta & Korhonen

Table 6
Multinomial Logistic Regression Analysis for MLD10 and LOW Status on Curriculum-Based Mathematics

95% CI for OR

Predictor B SE Z p OR Lower Upper

MLD10–TYP
Intercept 4.28 1.72 2.49 0.013 NA NA NA
SNMP t1 0.02 0.08 0.20 0.840 1.02 0.87 1.19
Counting t1 -0.15 0.06 -2.77 0.006 0.86 0.77 0.96
Nonverbal reasoning t1 -0.11 0.06 -2.00 0.046 0.89 0.80 1.00
Working memory t1 0.15 0.19 0.79 0.431 1.16 0.80 1.69
Rapid naming t1 -0.28 0.13 -2.13 0.033 0.76 0.59 0.98
Vocabulary t1 0.05 0.09 0.60 0.551 1.06 0.88 1.26
LOW–TYP
Intercept 3.49 1.58 2.20 0.028 NA NA NA
SNMP t1 0.02 0.07 0.36 0.719 1.03 0.90 1.17
Counting t1 -0.04 0.05 -0.78 0.435 0.96 0.88 1.06
Nonverbal reasoning t1 -0.13 0.05 -2.60 0.009 0.88 0.80 0.97
Working memory t1 -0.33 0.15 -2.11 0.035 0.72 0.53 0.98
Rapid naming t1 0.03 0.12 0.25 0.806 1.03 0.82 1.29
Vocabulary t1 -0.02 0.08 -0.23 0.818 0.98 0.85 1.14
MLD10–LOW
Intercept 0.79 1.99 0.40 0.690 NA NA NA
SNMP t1 -0.01 0.09 -0.09 0.927 0.99 0.82 1.19
Counting t1 -0.12 0.06 -1.82 0.069 0.89 0.79 1.01
Nonverbal reasoning t1 0.01 0.07 0.19 0.849 1.01 0.89 1.16
Working memory t1 0.48 0.23 2.09 0.036 1.61 1.03 2.52
Rapid naming t1 -0.31 0.16 -1.98 0.048 0.74 0.54 1.00
Vocabulary t1 0.07 0.10 0.69 0.490 1.07 0.88 1.32
Model fit measures Deviance AIC RN
2
χ 2
df p

222.80 250.80 0.21 47.12 12 < .001

Note. Estimates represent the log odds of MLD10 vs. TYP (reference group), LOW vs. TYP (reference group), and MLD10 vs.
LOW (reference group). MLD10 = mathematical learning disorder (performance ≤ 10th percentile), TYP = typically performing
(performance >25th percentile), LOW = low-performing (performance between 11–25th percentile). SNMP = symbolic numerical
magnitude processing, OR = odds ratio, CI = confidence interval, AIC = Akaike information criterion, R2N = Nagelkerke’s R2.

Figure 4
Predicted Probability for a Status of MLD10, LOW and TYP in Curriculum-Based Mathematics versus a) Verbal
Counting b) Nonverbal Reasoning, c) Working Memory, and d) Rapid Automatized Naming

345
 January 2022, Volume 14, Issue 3, 335-352

Discussion Confirming that both arithmetic fluency and CBM

measures are important to consider, we focused
This study investigated how domain-specific (i.e., next on how domain-specific and domain-general
SNMP and counting skills) and domain-general skills skills predict different MLD status based on arithmetic
(i.e., nonverbal reasoning, working memory, rapid fluency or curriculum-based mathematics.
automatized naming, and vocabulary) measured in
the first grade predict MLD status among third graders. Predictors of MLD based on Arithmetic Fluency
Several studies have found both domain-specific (e.g.,
Cañizares et al., 2012; Desoete et al., 2012; Hassinger- As hypothesized (H2.1), MLD25 status based on
Das et al., 2014; Landerl et al., 2004) and domain- arithmetic fluency was predicted by SNMP, verbal
general skills (e.g., David, 2012; Huijsmans et al., 2022; counting skills and rapid automatized naming. The
Mazzocco & Grimm, 2013) to be negatively associated lower was the first-grade performance in these skills,
with MLD. However, relatively few studies have the higher was the probability of showing weak
considered whether these relations are dependent arithmetic fluency (MLD25) in the third grade. A similar
on how MLD has been defined and operationalised pattern of predictions was found when the MLD25
in terms of the severity of MLD and measures of group was divided into MLD10 and LOW. As expected
mathematics performance. The novelty of this study (H3.1), SNMP and counting skills predicted both statuses,
is that it took into consideration both the different but rapid automatized naming predicted only MLD10.
cut-off criteria and the measures of mathematics That is, SNMP and counting skills predicted the status
performance (i.e., arithmetic fluency and curriculum- of MLD based on arithmetic fluency independent
based mathematics) in defining the MLD status, and of the severity level of MLD. Weakness in early rapid
included several early domain-specific and domain- automatized naming, instead, seemed to be more
general skills as predictors of MLD. Our findings suggest strongly associated with MLD10.
that both different cut-off criteria and mathematics
measures used for the definition of the MLD status are Prior research has shown the importance of
important to acknowledge, as these led to relatively early mathematical skills for later arithmetic and
significant variation in which students were identified mathematics performance (Aunio & Niemivirta, 2010;
as having MLD and which domain-specific and ten Braak et al., 2022), which our findings support. The
domain-general factors contributed to the MLD status. role of (S)NMP in mathematics learning and MLD has
been under debate due to mixed findings (Cañizares
Prior research has used various different mathematics et al., 2012; De Smedt et al., 2013; De Smedt & Gilmore,
measures for the identification of students with 2011; Desoete et al., 2012; Mammarella et al., 2021).
MLD. Typically, arithmetic fluency (Chow et al., 2021; Our findings give further support to SNMP being an
Koponen, Aro, et al., 2018) and broader mathematics important factor for later mathematics performance,
performance tests (Jordan et al., 2002) have been as we found that SNMP measured at grade 1 was a
applied. Little is known whether these different significant predictor of MLD status based on arithmetic
measures identify the same participants under the fluency at grade 3. The task measuring SNMP involves
same MLD status. Therefore, in our study, we first recognition of number symbols (1-digit numbers) and
examined the overlap of MLD statuses (MLD25, MLD10 understanding their related magnitude (Brankaer et
and LOW) based on arithmetic fluency and CBM (RQ1). al., 2017). It could be that this type of basic symbolic
For the MLD25 status, the overlap was 65 %, while for magnitude processing in the beginning school is
the MLD10 status, the share was 67 %, and for LOW, relevant especially for arithmetic learning at school
only 39 %. These results show that the use of only one (De Smedt et al., 2013; Nosworthy et al., 2013), and thus
type of the mathematics measure would have missed a good predictor of MLD based on arithmetic fluency.
a number of children struggling with either arithmetic Also, verbal counting skills (i.e., knowledge of number
fluency or CBM. Consequently, using a measure sequences) was found to predict MLD status based
reflecting one area of mathematics to define MLD on arithmetic fluency. Prior research has shown that
may lead to an exclusion of students with difficulties in verbal counting skills are important for learning basic
another equally relevant area of mathematics. It is thus addition and subtraction skills, and used also as a
important to consider which mathematics measures strategy for solving unknown addition and subtraction
to use for the identification of MLD, and whether to facts (Koponen et al., 2019; Ostad, 1998). Therefore,
rely on one or multiple measures. As we are still lacking difficulties in early verbal counting skills may slow
a globally applicable diagnostic measure of MLD, we down the learning of arithmetic facts. For these
would encourage researchers to carefully report both children, solving addition and subtraction facts may
the cut-off criteria and mathematics measures used become more error-prone due to making mistakes
for the identification of MLD and MLD status (e.g., LOW, in number sequences, and continuously getting
MLD10) for better comparability of research findings. incorrect answers between the fact and the answer
may thus interrupt memorizing the facts fluently.

346
 Predicting Mathematical Learning Difficulties Status / Mononen, Niemivirta & Korhonen

Rapid automatized naming was found to be the only in relation to MLD10 and LOW status. Because the
domain-general predictor of MLD status based on tasks in CBM had more variety and complexity in their
arithmetic fluency, which is in line with prior research mathematics content and procedures, we expected
(Donker et al., 2016; Mazzocco & Grimm, 2013). While counting skills (Hassinger-Das et al., 2014), nonverbal
we found non-alphanumeric rapid automatized reasoning (Engle, 2018; Huijsmans et al., 2022), working
naming to predict MLD25 and MLD10 status, Donker memory (David, 2012), and vocabulary (Chow et al.,
et al. (2016) similarly reported students with MLD25 to 2021) to predict MLD25 status. In accordance with
have weaker non-alphanumeric rapid automatized our hypothesis (H2.2), counting skills and nonverbal
naming skills. Further, Mazzocco and Grimm (2013) reasoning, but not working memory, were found to
found students with MLD10 to have persistent predict MLD25 status based on CBM. When MLD25
weakness in rapid automatized naming of colors was divided into two, counting skills only predicted
compared to their peers, and also students with LOW the MLD10 status, but not LOW, as we would have
status to show slight delay in their development of expected (H2.3). In fact, no domain-specific skills
rapid naming of colors. As to why rapid automatized predicted LOW status based on CBM. Regarding
naming and arithmetic fact retrieval are related, domain-general skills, in contrast to our hypothesis,
in both tasks children need to access quickly and nonverbal reasoning and rapid automatized naming,
retrieve phonological representations from long-term but not working memory, predicted MLD10 status
memory (e.g., “blue” and “seven”). The role of early based on CBM. Instead, working memory together
non-alphanumeric rapid automatized naming skills with nonverbal reasoning predicted LOW status.
in later MLD based on arithmetic fluency seems to be
important to acknowledge. As elaborated above, early counting skills have shown
to be associated with later mathematics performance
Based on our findings, both weak SNMP and counting (ten Braak et al., 2022). Concerning MLD status, our
skills could be considered as risk factors for later results revealed that verbal counting skills predicted
difficulties in arithmetic fluency, independent of the especially MLD10 status based on CBM. Taken
level of severity, while rapid automatized naming together, these findings imply that verbal counting
seems to be specifically associated with MLD10 skills are good at predicting MLD10 status independent
students’ arithmetic fluency. As a practical implication of the mathematics measure used for identification.
for early schooling, SNMP and verbal counting skills This further puts emphasis on supporting children’s
should be regularly screened in classrooms (see e.g., early counting skills in early schooling as a preventive
Brankaer et al., 2017; Nosworthy et al., 2013; Salminen step for later severe MLD.
& Koponen, 2011). Those who struggle in comparing
1-digit numbers or in reciting number sequences Nonverbal reasoning was found to predict MLD status
forwards and backwards, should be provided based on CBM, independent of the severity level. CBM
with relevant intensified pedagogical support (i.e., test measured a broad range of mathematical subskills
intervention) as early as possible (see e.g., Ramani et with different types of tasks (e.g., “Lisa’s little finger is 4
al., 2017). Even if the role of early rapid automatized g/kg/cm/m long”; “One apple costs 3 krones, 4 apples
naming skills in MLD is important to acknowledge, cost __ krones.”, simple word problems for fractions,
training of domain-general skills, with many examples and addition and subtraction algorithms). Solving
from working memory training research, has shown these tasks thus required making logical decisions and
rather weak far transfer effects on mathematics proceeding systematically in the task (i.e., nonverbal
performance (Melby-Lervåg & Hulme, 2013). However, reasoning) (Engle, 2018), which differs from solving
recently Pecini et al. (2019) showed the training of rapid simple arithmetic facts. This result also resonates well
automatized naming using specific software to be with Hujsmans’ et al. (2022) findings showing students
effective in ameliorating reading accuracy and speed. with MLD to perform weaker in nonverbal reasoning
Until we have solid evidence of the effectiveness of compared to their peers without MLD.
training rapid automatized naming and its transfer on
improved mathematical skills, it might be more useful MLD10 and LOW based on CBM were separated by
to focus on training mathematical skills as a preventive the domain-general predictors of rapid automatized
step with children identified with weakness in early naming and working memory (i.e., central executive
rapid automatized naming skills. functioning). Overall, rapid automatized naming
turned out to be an important predictor of MLD10 status,
Predictors of MLD based on Curriculum-Based as it predicted the status based on both arithmetic
Mathematics fluency and CBM. Previously, rapid automatized
naming has been found to be related to broader
Our findings on the predictions of MLD based on CBM mathematics performance as well, although not as
differed from those of MLD based on arithmetic strongly as to arithmetic fluency (Koponen et al., 2017).
fluency. Here, the domain-general skills seemed to be Because the tasks in mathematics performance tests
better predictors than domain-specific skills, especially typically involve more processes than quick retrieval

347
 January 2022, Volume 14, Issue 3, 335-352

only, the relationship of rapid automatized naming Note, that because of practical reasons, our sample
with broader mathematics performance are weaker was restricted to only include children from the Oslo
than with arithmetic fluency. This was also evident in region in Norway, due to which caution should be
our study (r = .43 with arithmetic fluency, and r = .33 exercised in generalizing the findings to other contexts.
with CBM). Also, the Covid-19 pandemic complicated the final
stages of the data collection due to which one school
Interestingly, working memory, and more specifically, withdrew from the study and the data collection in
central executive, predicted only LOW status based spring 2021 needed to be organized online. However,
on CBM, although prior research has found its role to there are no indications of this causing any bias in our
be significant in mathematics learning (Friso-van den data.
Bos et al., 2013) and MLD. Students with MLD often
have a working memory deficit (Andersson & Lyxell, Conclusions
2007; David, 2012; Passolunghi & Mammarella, 2010).
Concerning the MLD status here, our results partly Our findings suggest that both the different cut-off
reflect the findings by Huijsman et al. (2022), who criteria and mathematics measures used for defining
found students identified as LOW to be weaker than and operationalizing the MLD status are important to
MLD10 in the visual component of working memory. acknowledge in studies, as these may lead to relatively
Even if we used a generally recognized backwards significant variation in which students are identified
digit span as a measure of central executive in our as having MLD and which factors contribute to the
study, it might be that the central executive measured MLD status. In relative terms, domain-specific skills
this way captured only part of the construct, and a appear to be more predictive when the MLD status
broader measure would have been needed to retain is based on arithmetic fluency, while domain-general
its predictive power and thus obtain a similar effect on skills seem more influential when the MLD status is
MLD as in previous studies. based on CBM. Counting skills and rapid automatized
naming, instead, appear to be robust predictors of
Limitations and Future Directions MLD status regardless of the mathematics measure
used. As a practical implication for the prevention of
The current study has some limitations that need MLD, we advocate focusing on screening children’s
to be noted. First, although we included many SNMP and verbal counting skills in early grades, and
domain-specific and domain-general skills based providing appropriate intervention in these for those
on prior research, we may have missed some other in need of educational support.
important factors as predictors, such as non-symbolic
numerical magnitude processing, subitizing, or object Acknowledgements
counting as domain-specific skills, or inclusion of other
components of working memory (i.e., phonological This work was supported by the Norwegian Research
loop and visuo-spatial working memory) and Council [Grant number: 283396], for the first author. We
language (e.g., receptive vocabulary or syntax). Their thank all the participating children, their teachers and
role as a predictor of MLD would be of future interest parents, as well as the research assistants involved in
to explore. the data collections.

Concerning the severity criteria for the MLD status, References

we based our grouping of children on sample-based
percentiles both in arithmetic fluency and CBM. It
Andersson, U., & Lyxell, B. (2007). Working memory
might be argued that it would have been better to
deficit in children with mathematical
use mathematics tests with norms for identifying
difficulties: A general or specific deficit? Journal
students with MLD. At the time of the study, neither
of Experimental Child Psychology, 96(3), 197–228.
standardized broad mathematics performance tests
https://doi.org/10.1016/j.jecp.2006.10.001
nor combined tests of addition and subtraction facts
were available for this age group in Norway. Therefore,
Aunio, P., Korhonen, J., Ragpot, L., Törmänen, M.,
we developed a new CBM test, and applied sample-
& Henning, E. (2021). An early numeracy
based percentiles in both CBM and arithmetic fluency
intervention for first-graders at risk for
(i.e., a combination of addition and subtraction
mathematical learning difficulties. Early
fluency subtests) for the identification of MLD. The lack
Childhood Research Quarterly, 55, 252–262.
of global standardized mathematics measures to be
https://doi.org/10.1016/j.ecresq.2020.12.002
used in MLD studies is problematic in terms of the
comparability of results, and should be addressed in
future research on MLD.

348
 Predicting Mathematical Learning Difficulties Status / Mononen, Niemivirta & Korhonen

Aunio, P., Korhonen, J., Ragpot, L., Törmänen, M., Cross, A. M., Joanisse, M. F., & Archibald, L. M. D.
Mononen, R., & Henning, E. (2019). Multi- (2019). Mathematical abilities in children with
factorial approach to early numeracy—The developmental language disorder. Language,
effects of cognitive skills, language factors and Speech, and Hearing Services in Schools,
kindergarten attendance on early numeracy 50(1), 150–163. https://doi.org/10.1044/2018_
performance of South African first graders. LSHSS-18-0041
International Journal of Educational Research,
97, 65–76. https://doi.org/10.1016/j.ijer.2019.06.011 David, C. V. (2012). Working memory deficits in
Math learning difficulties: A meta-analysis.
Aunio, P., & Niemivirta, M. (2010). Predicting children’s International Journal of Developmental
mathematical performance in grade one Disabilities, 58(2), 67–84. https://doi.org/10.1179/2
by early numeracy. Learning and Individual 047387711Y.0000000007
Differences, 20(5), 427–435. https://doi.
org/10.1016/j.lindif.2010.06.003 De Smedt, B., & Gilmore, C. K. (2011). Defective number
module or impaired access? Numerical
Baddeley, A. D., & Logie, R. H. (1999). Working memory: magnitude processing in first graders
The multiple-component model. In Models with mathematical difficulties. Journal of
of working memory: Mechanisms of active Experimental Child Psychology, 108(2), 278–292.
maintenance and executive control (pp. 28– https://doi.org/10.1016/j.jecp.2010.09.003
61). Cambridge University Press. https://doi.
org/10.1017/CBO9781139174909.005 De Smedt, B., Noël, M.-P., Gilmore, C., & Ansari, D. (2013).
How do symbolic and non-symbolic numerical
Brankaer, C., Ghesquière, P., & De Smedt, B. (2017). magnitude processing skills relate to individual
Symbolic magnitude processing in elementary differences in children’s mathematical skills? A
school children: A group administered paper- review of evidence from brain and behavior.
and-pencil measure (SYMP Test). Behavior Trends in Neuroscience and Education, 2(2), 48–
Research Methods, 49(4), 1361–1373. https://doi. 55. https://doi.org/10.1016/j.tine.2013.06.001
org/10.3758/s13428-016-0792-3
Dehaene, S., Dehaene-Lambertz, G., & Cohen, L.
Butterworth, B. (2005). Developmental Dyscalculia. (1998). Abstract representations of numbers
In Handbook of Mathematical Cognition (pp. in the animal and human brain. Trends in
455–467). Psychology Press. Neurosciences, 21(8), 355–361. https://doi.
org/10.1016/S0166-2236(98)01263-6
Cañizares, D. C., Crespo, V. R., & Alemañy, E. G. (2012).
Symbolic and non-symbolic number magnitude Desoete, A., Ceulemans, A., Weerdt, F. D., & Pieters, S.
processing in children with developmental (2012). Can we predict mathematical learning
dyscalculia. The Spanish Journal of Psychology, disabilities from symbolic and non-symbolic
15(3), 952–966. https://doi.org/10.1016/j. comparison tasks in kindergarten? Findings
jecp.2008.12.006 from a longitudinal study. British Journal of
Educational Psychology, 82(1), 64–81. https://doi.
Chow, J. C., & Ekholm, E. (2019). Language domains org/10.1348/2044-8279.002002
differentially predict mathematics performance
in young children. Early Childhood Research Donker, M., Kroesbergen, E., Slot, E., Van Viersen, S.,
Quarterly, 46, 179–186. https://doi.org/10.1016/j. & De Bree, E. (2016). Alphanumeric and non-
ecresq.2018.02.011 alphanumeric Rapid Automatized Naming in
children with reading and/or spelling difficulties
Chow, J. C., Majeika, C. E., & Sheaffer, A. W. (2021). and mathematical difficulties. Learning and
Language skills of children with and without Individual Differences, 47, 80–87. https://doi.
mathematics difficulty. Journal of Speech, org/10.1016/j.lindif.2015.12.011
Language, and Hearing Research, 64(9),
3571–3577. https://doi.org/10.1044/2021_ Engle, R. W. (2018). Working memory and executive
JSLHR-20-00378 attention: A revisit. Perspectives on
Psychological Science, 13(2), 190–193. https://doi.
Chu, F. W., vanMarle, K., & Geary, D. C. (2016). org/10.1177/1745691617720478
Predicting children’s reading and mathematics
achievement from early quantitative Friso-van den Bos, I., van der Ven, S. H. G., Kroesbergen,
knowledge and domain-general cognitive E. H., & van Luit, J. E. H. (2013). Working memory
abilities. Frontiers in Psychology, 7, 775. https:// and mathematics in primary school children:
doi.org/10.3389/fpsyg.2016.00775 A meta-analysis. Educational Research
Review, 10, 29–44. https://doi.org/10.1016/j.
edurev.2013.05.003
349
 January 2022, Volume 14, Issue 3, 335-352

Geary, D. C. (2011). Consequences, characteristics, Koponen, T., Aunola, K., & Nurmi, J.-E. (2019). Verbal
and causes of mathematical learning counting skill predicts later math performance
disabilities and persistent low achievement in and difficulties in middle school. Contemporary
mathematics. Journal of Developmental and Educational Psychology, 59, 101803. https://doi.
Behavioral Pediatrics: JDBP, 32(3), 250–263. org/10.1016/j.cedpsych.2019.101803
https://doi.org/10.1097/DBP.0b013e318209edef
Koponen, T., Eklund, K., & Salmi, P. (2018). Cognitive
Geary, D. C., Hoard, M. K., Nugent, L., & Bailey, D. H. predictors of counting skills. Journal of
(2012). Mathematical cognition deficits in Numerical Cognition, 4(2), 410–428. https://doi.
children with learning disabilities and persistent org/10.5964/jnc.v4i2.116
low achievement: A five-year prospective
study. Journal of Educational Psychology, 104(1), Koponen, T., Georgiou, G., Salmi, P., Leskinen, M., &
206–223. https://doi.org/10.1037/a0025398 Aro, M. (2017). A meta-analysis of the relation
between RAN and mathematics. Journal of
Gersten, R., Jordan, N. C., & Flojo, J. R. (2005). Early Educational Psychology, 109(7), 977–992. https://
identification and interventions for students doi.org/10.1037/edu0000182
with mathematics difficulties. Journal of
Learning Disabilities, 38(4), 293–304. https://doi. Kroesbergen, E. H., & van Dijk, M. (2015). Working
org/10.1177/00222194050380040301 memory and number sense as predictors
of mathematical (dis-)ability. Zeitschrift
Hassinger-Das, B., Jordan, N. C., Glutting, J., Irwin, C., Für Psychologie, 223(2), 102–109. https://doi.
& Dyson, N. (2014). Domain-general mediators org/10.1027/2151-2604/a000208
of the relation between kindergarten
number sense and first-grade mathematics Kyttälä, M., & Lehto, J. E. (2008). Some factors underlying
achievement. Journal of Experimental Child mathematical performance: The role of
Psychology, 118, 78–92. https://doi.org/10.1016/j. visuospatial working memory and non-verbal
jecp.2013.09.008 intelligence. European Journal of Psychology
of Education, 23(1), 77. https://doi.org/10.1007/
Huijsmans, M. D. E., Kleemans, T., & Kroesbergen, E. BF03173141
H. (2022). The cognitive profiles for different
samples of mathematical learning difficulties Landerl, K., Bevan, A., & Butterworth, B. (2004).
and their similarity to typical development: Developmental dyscalculia and basic
Evidence from a longitudinal study. Journal numerical capacities: A study of 8–9-year-old
of Experimental Child Psychology, 214, 105288. students. Cognition, 93(2), 99–125. https://doi.
https://doi.org/10.1016/j.jecp.2021.105288 org/10.1016/j.cognition.2003.11.004

Jordan, N. C., Kaplan, D., & Hanich, L. B. (2002). LeFevre, J.-A., Fast, L., Skwarchuk, S.-L., Smith-Chant, B.
Achievement growth in children with learning L., Bisanz, J., Kamawar, D., & Penner-Wilger, M.
difficulties in mathematics: Findings of a two- (2010). Pathways to mathematics: Longitudinal
year longitudinal study. Journal of Educational predictors of performance. Child Development,
Psychology, 94(3), 586–597. https://doi. 81(6), 1753–1767. https://doi.org/10.1111/j.1467-
org/10.1037/0022-0663.94.3.586 8624.2010.01508.x

Jordan, N. C., Kaplan, D., Ramineni, C., & Locuniak, Lin, X., Peng, P., & Luo, H. (2021). The deficit profile of
M. N. (2009). Early math matters: Kindergarten elementary students with computational
number competence and later mathematics difficulties versus word problem-solving
outcomes. Developmental Psychology, 45(3), difficulties. Learning Disability Quarterly, 44(2),
850–867. https://doi.org/10.1037/a0014939 110–122. https://doi.org/10.1177/0731948719865499

Klausen, T., & Reikerås, E. (2016). Regnefaktaprøven. Lopez-Pedersen, A., Mononen, R., Korhonen, J., Aunio,
Lesesenteret, University of Stavanger, Norway. P., & Melby-Lervåg, M. (2021). Validation of an
early numeracy screener for first graders.
Koponen, T., Aro, M., Poikkeus, A.-M., Niemi, P., Scandinavian Journal of Educational Research,
Lerkkanen, M.-K., Ahonen, T., & Nurmi, J.-E. (2018). 65(3), 404–424. https://doi.org/10.1080/00313831.
Comorbid fluency difficulties in reading and 2019.1705901
math: Longitudinal stability across early grades.
Exceptional Children, 84(3), 298–311. https://doi.
org/10.1177/0014402918756269

350
 Predicting Mathematical Learning Difficulties Status / Mononen, Niemivirta & Korhonen

Mammarella, I. C., Toffalini, E., Caviola, S., Colling, L., & Ostad, S. A. (1998). Developmental differences in
Szűcs, D. (2021). No evidence for a core deficit solving simple arithmetic word problems and
in developmental dyscalculia or mathematical simple number-fact problems: A comparison of
learning disabilities. Journal of Child Psychology mathematically normal and mathematically
and Psychiatry, 62(6), 704–714. https://doi. disabled children. Mathematical Cognition, 4(1),
org/10.1111/jcpp.13397 1–19. https://doi.org/10.1080/135467998387389

Mazzocco, M. M. M., Feigenson, L., & Halberda, J. (2011). Passolunghi, M. C., & Mammarella, I. C. (2010). Spatial
Impaired acuity of the approximate number and visual working memory ability in children
system underlies mathematical learning with difficulties in arithmetic word problem
disability (dyscalculia). Child Development, solving. European Journal of Cognitive
82(4), 1224–1237. https://doi.org/10.1111/j.1467- Psychology, 22(6), 944–963. https://doi.
8624.2011.01608.x org/10.1080/09541440903091127

Mazzocco, M. M. M., & Grimm, K. J. (2013). Growth in Pecini, C., Spoglianti, S., Bonetti, S., Di Lieto, M. C.,
Rapid Automatized Naming from Grades K to Guaran, F., Martinelli, A., Gasperini, F., Cristofani,
8 in children with math or reading disabilities. P., Casalini, C., Mazzotti, S., Salvadorini, R.,
Journal of Learning Disabilities, 46(6), 517–533. Bargagna, S., Palladino, P., Cismondo, D.,
https://doi.org/10.1177/0022219413477475 Verga, A., Zorzi, C., Brizzolara, D., Vio, C., &
Chilosi, A. M. (2019). Training RAN or reading?
Mazzocco, M. M. M., Myers, G. F., Lewis, K. E., Hanich, A telerehabilitation study on developmental
L. B., & Murphy, M. M. (2013). Limited knowledge dyslexia. Dyslexia, 25(3), 318–331. https://doi.
of fraction representations differentiates middle org/10.1002/dys.1619
school students with mathematics learning
disability (dyscalculia) versus low mathematics Pina, V., Fuentes, L. J., Castillo, A., & Diamantopoulou,
achievement. Journal of Experimental S. (2014). Disentangling the effects of working
Child Psychology, 115(2), 371–387. https://doi. memory, language, parental education,
org/10.1016/j.jecp.2013.01.005 and non-verbal intelligence on children’s
mathematical abilities. Frontiers in Psychology,
Melby-Lervåg, M., & Hulme, C. (2013). Is working 5, 415. https://doi.org/10.3389/fpsyg.2014.00415
memory training effective? A meta-analytic
review. Developmental Psychology, 49(2), 270– Price, G., & Ansari, D. (2013). Dyscalculia: Characteristics,
291. https://doi.org/10.1037/a0028228 causes, and treatments. Numeracy, 6(1). http://
dx.doi.org/10.5038/1936-4660.6.1.2
Menon, V. (2016). Working memory in children’s math
learning and its disruption in dyscalculia. Purpura, D. J., Day, E., Napoli, A. R., & Hart, S. A. (2017).
Current Opinion in Behavioral Sciences, 10, 125– Identifying domain-general and domain-
132. https://doi.org/10.1016/j.cobeha.2016.05.014 specific predictors of low mathematics
performance: A Classification and regression
Mononen, R. (2021). Matteoppdraget 3. Curriculum- tree analysis. Journal of Numerical Cognition,
based mathematics test for grade 3. 3(2), 365–399. https://doi.org/10.5964/jnc.v3i2.53
Unpublished. University of Oslo.
Purpura, D. J., & Ganley, C. M. (2014). Working memory
Mussolin, C., Mejias, S., & Noël, M.-P. (2010). Symbolic and language: Skill-specific or domain-
and nonsymbolic number comparison general relations to mathematics? Journal of
in children with and without dyscalculia. Experimental Child Psychology, 122, 104–121.
Cognition, 115(1), 10–25. https://doi.org/10.1016/j. https://doi.org/10.1016/j.jecp.2013.12.009
cognition.2009.10.006
Ramani, G. B., Jaeggi, S. M., Daubert, E. N., & Buschkuehl,
Nosworthy, N., Bugden, S., Archibald, L., Evans, B., & M. (2017). Domain-specific and domain-general
Ansari, D. (2013). A two-minute paper-and- training to improve kindergarten children’s
pencil test of symbolic and nonsymbolic mathematics. Journal of Numerical Cognition,
numerical magnitude processing explains 3(2), 468–495. https://doi.org/10.5964/jnc.v3i2.31
variability in primary school children’s arithmetic
competence. PLOS ONE, 8(7), e67918. https://doi. Raven, J. C., Court, J. H., & Raven, J. (1990). Manual for
org/10.1371/journal.pone.0067918 Raven’s progressive matrices and vocabulary
scales: Section 2, Coloured progressive matrices.
Oxford University Press.

351
 January 2022, Volume 14, Issue 3, 335-352

Rousselle, L., & Noël, M.-P. (2007). Basic numerical skills in Wechsler, D. (2017). WISC-V Wechsler intelligence
children with mathematics learning disabilities: Scale for Children – fifth edition. Norwegian
A comparison of symbolic vs non-symbolic version.
number magnitude processing. Cognition,
102(3), 361–395. https://doi.org/10.1016/j. Willburger, E., Fussenegger, B., Moll, K., Wood, G., &
cognition.2006.01.005 Landerl, K. (2008). Naming speed in dyslexia
and dyscalculia. Learning and Individual
Salihu, L., & Räsänen, P. (2018). Mathematics skills of Differences, 18(2), 224–236. https://doi.
Kosovar primary school children: A special org/10.1016/j.lindif.2008.01.003
view on children with mathematical learning
difficulties. International Electronic Journal of World Health Organization. (2019). International
Elementary Education, 10(4), 421–430. Retrieved Statistical Classification of Diseases and Related
from https://iejee.com/index.php/IEJEE/article/ Health Problems (11th ed.). https://icd.who.int/
view/409 en

Salminen, J., & Koponen, T. (2011). Oppimisen arviointi: Zhang, X., Räsänen, P., Koponen, T., Aunola, K.,
Matematiikan oppimisen seurannan välineet. Lerkkanen, M.-K., & Nurmi, J.-E. (2020). Early
Käsikirja [LukiMat – Assessment of learning: cognitive precursors of children’s mathematics
Progress monitoring of mathematics learning. learning disability and persistent low
Manual]. Niilo Mäki Institute. achievement: A 5-year longitudinal study. Child
Development, 91(1), 7–27. https://doi.org/10.1111/
Semel, E., Wiig, E., & Secord, W. (2003). Clinical cdev.13123
Evaluation of Language Fundamentals – Fourth
Edition (CELF-4). Norwegian version. Pearson.

Stock, P., Desoete, A., & Roeyers, H. (2010). Detecting

children with arithmetic disabilities from
kindergarten: Evidence from a 3-year
longitudinal study on the role of preparatory
arithmetic abilities. Journal of Learning
Disabilities, 43(3), 250–268. https://doi.
org/10.1177/0022219409345011

ten Braak, D., Lenes, R., Purpura, D. J., Schmitt, S. A., &
Størksen, I. (2022). Why do early mathematics
skills predict later mathematics and reading
achievement? The role of executive function.
Journal of Experimental Child Psychology, 214,
105306. https://doi.org/10.1016/j.jecp.2021.105306

The jamovi project. (2021). Jamovi (2.0) [Computer

software]. https://www.jamovi.org

Tolar, T. D., Fuchs, L., Fletcher, J. M., Fuchs, D., & Hamlett,
C. L. (2016). Cognitive profiles of mathematical
problem solving learning sisability for different
definitions of disability. Journal of Learning
Disabilities, 49(3), 240–256. https://doi.
org/10.1177/0022219414538520

Toll, S. W. M., & Van Luit, J. E. H. (2014). Explaining

numeracy development in weak performing
kindergartners. Journal of Experimental Child
Psychology, 124, 97–111. https://doi.org/10.1016/j.
jecp.2014.02.001

Van Luit, J. E. H., & Toll, S. W. M. (2018). Associative

cognitive factors of math problems in students
diagnosed with developmental dyscalculia.
Frontiers in Psychology, 9, 1907. https://doi.
org/10.3389/fpsyg.2018.01907
352