Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Early Education and Development

ISSN: 1040-9289 (Print) 1556-6935 (Online) Journal homepage: https://www.tandfonline.com/loi/heed20

Promoting Effects of RtI-Based Mathematical Play


Training on Number Sense Growth among Low-SES
Preschool Children

Yuan Liang, Lijin Zhang, Yang Long, Qian Deng & Yujuan Liu

To cite this article: Yuan Liang, Lijin Zhang, Yang Long, Qian Deng & Yujuan Liu
(2019): Promoting Effects of RtI-Based Mathematical Play Training on Number Sense
Growth among Low-SES Preschool Children, Early Education and Development, DOI:
10.1080/10409289.2019.1664261

To link to this article: https://doi.org/10.1080/10409289.2019.1664261

View supplementary material

Published online: 15 Sep 2019.

Submit your article to this journal

Article views: 44

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=heed20
EARLY EDUCATION AND DEVELOPMENT
https://doi.org/10.1080/10409289.2019.1664261

Promoting Effects of RtI-Based Mathematical Play Training on


Number Sense Growth among Low-SES Preschool Children
a a
Yuan Liang , Lijin Zhang , Yang Longb, Qian Denga and Yujuan Liuc
a
School of Psychology, Shaanxi Key Laboratory of Behavior and Cognitive Neuroscience, Shaanxi Key Research
Center of Child Mental and Behavioral Health, Shaanxi Normal University; bSchool of Education, Ningxia University;
c
Research Center for Moral Education, Psychology and Special Education, National Institute of Education Sciences

ABSTRACT
Research Findings: Low-socioeconomic status (SES) (intervention group, n = 31;
control group, n = 32) and middle/high-SES (reference group, n = 33) 5-year-
old Chinese kindergartners were screened through SES and IQ test. Their
number sense was examined over five time points. The pretest was conducted
at the beginning of the fall semester, followed by three immediate posttests
monthly. Math play training was administrated to the intervention group and
story reading to the other two groups during this time. Lastly, a delayed
posttest was conducted two months later after the last immediate posttest.
All the interventions were carried out in small groups, 15–30 minutes sessions,
2 days per week for 12 weeks. The promotion effect of math play training was
monitored based on response to intervention (RtI) paradigm. Utilizing latent
growth modeling, we found that the intervention group got significant
improvement in number sense and almost caught up with the middle/high-
SES group after training, and the intervention effects persisted for two months.
The control group had the slowest rate of progress and significantly lower
scores than the other two groups. Practice or Policy: RtI-based mathematical
game intervention could improve the number sense level and growth rate of
preschoolers in rural areas.

Number sense is an intuitive understanding of what numbers mean and of numerical relationships
(Jordan, Kaplan, Oláh, & Locuniak, 2006). The main components of number sense in 3- to 6-year-
old children include counting, number knowledge, and number operations (Jordan, Glutting, &
Ramineni, 2008, 2010). Counting is guided by principles related to one-to-one correspondence, fixed
order, and cardinality, and enables children to identify numerical sets (Griffin, 2004). Number
knowledge is a mathematical ability that includes numerical identification, number comparison,
and number distance judgments (Jordan et al., 2008). Number operations involve abilities to trans-
form sets through addition and subtraction, and to calculate in verbal and nonverbal contexts or
with and without referents (physical or verbal) (Jordan et al., 2006). Number sense not only makes
a difference on identifying mathematical difficulties (Jordan et al., 2010; Locuniak & Jordan, 2008),
but also is an effective intervention indicator to prevent mathematical difficulties (Cowan & Powell,
2014; Dyson, Jordan, & Glutting, 2013). Insufficient number sense in early childhood is likely to lay
a hidden danger for future mathematics learning (Fuchs et al., 2013). However, limited educational
resources of low-socioeconomic status (SES) families at a poorer place in the United States cannot
provide children with necessary mathematical enlightenment and numerical training, so these
children lag far behind than their middle-income peers when entering primary school, and the
gap widens with grade (Jordan et al., 2006; Morgan, Farkas, & Wu, 2011). Thus, it is important to

CONTACT Lijin Zhang zhanglijin2013@snnu.edu.cn School of Psychology, Shaanxi Key Laboratory of Behavior and
Cognitive Neuroscience, Shaanxi Normal University, Xi’an 710062, China
Supplemental data for this article can be accessed publisher’s website.
© 2019 Taylor & Francis Group, LLC
2 Y. LIANG ET AL.

implement timely and appropriate mathematical interventions for low-income children to effectively
reduce the likelihood of math difficulties later. Although computer-based mathematical game
interventions were more specific, vivid, intuitive, and helpful to attract children’s attention and
enhance their perception of quantitative information, kindergartens in poor and underdeveloped
areas did not have enough money to guarantee long-term use of tablet PCs. Accordingly, in the
current study, we aimed to improve the number sense of low-SES children by (a) using local easy-to-
get materials to design mathematical play training for low-SES children, (b) conducting an inter-
vention based on the method of response to intervention (RtI) and the repeated process of “pretest–
intervention–posttest – … … – delayed posttest” dynamic testing, and (c) evaluating the effective-
ness of RtI-based intervention training for preschool children in poor areas. The present study was
conducted in a suburb of Yinchuan City in Ningxia Hui Autonomous Region, an economically
backward place in China. Considering the serious shortage of early childhood education and
learning resources in rural areas, we will make full use of the existing local resources to develop
economical and convenient number sense physical games.

The Reasons for Insufficient Number Sense of Children with low-SES


The number sense gap between low-income and high-income children is associated with pre-
vious experience and learning opportunities, as well as intrinsic abilities which mainly refer to
domain-general factors, including language, working memory, executive function and so on
(Jordan, Glutting, Dyson, Hassinger–Das, & Irwin, 2012). The most direct reason for the
insufficient number sense of low-SES children is their poor home learning environment related
to quantity and number and their parents’ little involvement. Parents of children in low-SES
families generally have a low educational level, and their limited knowledge may reduce their
engagement in guiding children. They almost never buy reference books for their children or
create a natural environment for learning math at home – “Mini Curriculum” (which refers to
the process that family members teach children number knowledge in their daily life) (Griffin,
2004). Moreover, the number sense development of preschool children is influenced by language
(both oral and written language skills were found to have distinct relations with young children’s
acquisition of numbers and arithmetic calculations, see Purpura & Napoli, 2015; Zhang & Lin,
2015), working memory, and executive function (Friso-Van Den Bos, Van der Ven, Kroesbergen,
& Van Luit, 2013; Szűcs, Devine, Soltesz, Nobes, & Gabriel, 2014). Children from low income
backgrounds are less likely to engage in cognitively enriching verbal (Hart & Risley, 1995) or
reading interactions ; Serpell, Baker, & Sonnenschein, 2005); and reading at home are associated
with children’s math achievement (Sonnenschein & Galindo, 2015). Long-term poverty has an
adverse effect on children’s working memory (Evans & Schamberg, 2009) and executive function
(Noble, Mccandliss, & Farah, 2010), and prevents these children from reaching the level of their
peers on school readiness, math skills, reading, and writing (Willoughby, Blair, Wirth, &
Greenberg, 2012). Additionally, an fMRI study showed that low-SES children have smaller
volumes of gray matter in bilateral hippocampi, middle temporal gyri, left fusiform, and right
inferior occipitotemporal gyri than their high-SES peers (Jednoróg et al., 2012), and these areas
of the brain play an important role in mathematical cognition (Meintjes et al., 2010; Rivera,
Reiss, Eckert, & Menon, 2005).

Dynamic Evaluation and Intervention Can Effectively Promote the Development of Number
Sense in low-SES Children
The disadvantaged impact of low SES on children’s initial number sense level will be gradually
amplified as they enter the formal education stage. These children tend to underperform on
basic reading, writing, and counting before accessing to formal school, as well as experience
serious academic problems even after receiving formal school education (Willoughby et al.,
EARLY EDUCATION & DEVELOPMENT 3

2012). If they are provided with a number sense (instead of mathematical knowledge) interven-
tion in early childhood, they will probably catch up with their non-poor counterparts (Gersten,
Jordan, & Flojo, 2005). The idea and paradigm of dynamic testing provides valuable reference
strategies and a theoretical foundation for intervening and evaluating the number sense level and
its growth in low-SES children.
Based on the “zone of proximal development” theory, dynamic testing proposes the testing
methods for exploring the potential level of cognitive development of children through dynamic
process evaluation (Gustafson, Svensson, & Fälth, 2014). The goal of dynamic testing is to detect the
highest level of children’s cognitive development fairly and objectively, especially for the develop-
ment of “disadvantaged children” (including children with learning difficulties, mental retardation,
mild mental disorders, autism; and children with low SES, new immigrants, and minority children of
non-mainstream culture, etc.) (Jitendra & Kammeenui, 1993). RtI is a typical dynamic testing
method and a framework in which interventions are implemented mostly in general education
classes to resolve academic difficulties and mitigate contextual variables (i.e., lack of instruction, SES,
cultural differences, etc.), as a supplementary explanation for academic failure (Hinton, Flores, &
Shippen, 2013). RtI models are multitiered service delivery systems in which schools provide layered
interventions that begin in general education and increase in intensity (e.g., increased time for
instruction to smaller groups of students) depending on children’s response to instruction (Fletcher
& Vaughn, 2010). Tier 1 represents the universal classroom instruction delivered to all students to
ensure that the effectiveness of specific intervention for most children (D. Fuchs & Fuchs, 2006).
Children who do not respond to Tier 1 receive more intensive group instruction in Tier 2. The
systematic and explicit instruction at this level can provide additional opportunities to practice target
skills (D. Fuchs & Fuchs, 2006). Those who are unresponsive to Tier 2 will require Tier 3 or more
Tiers’ intervention to make adequate progress. RtI models (a) screen all children for academic and
behavioral problems, (b) monitor the progress of children at risk for difficulties in these areas, and
(c) provide increasingly intense interventions based on the response to progress monitoring assess-
ments (Vaughn & Fuchs, 2003).
Some researchers have demonstrated the usefulness of the RtI-based intervention for improv-
ing the numerical knowledge of preschoolers from low-income backgrounds. For example, 5- to
7-year-old children were asked to repeat, organize, arrange, and learn to associate new informa-
tion with prior knowledge and experience, as well as to master mathematical learning skills.
After 20 hours of RtI-based intervention within 3 months, children in the intervention group
exhibited greater improvements in number comparisons, classifications, and counting skills (Van
Luit & Schopman, 2000). However, a delayed posttest was not used to examine the long-term
effect of the intervention in the above study. Another RtI-based paradigm in classroom teaching
was conducted to 3- to 4-year-old children at risk for mathematics difficulties, and the research-
ers found that children in the RtI’s class achieved moderate or high improvement in language,
literacy, and mathematical skills, even a greater progress in social skills (Crowell, Polanski,
Lesiak, McCarthy, & Ramey, 2011). But this study was an exploratory examination of RtI
program without a control group. In addition, researchers verified the effectiveness of RtI for
first graders with insufficient number sense. After 6 weeks of mathematical game intervention,
the performance of the intervention group was significantly better than that of the control group
on number operations, estimation, and combination. This dynamic game intervention improved
children’s ability to operate and solve specific problems according to intuition to magnitude (i.e.,
number sense) in real situations (Zhang, Bi, Liang, & Liu, 2016). However, few studies have
examined whether the RtI-based math game intervention would also improve the number sense
of kindergartners. Therefore, our main target in the current study was to test the effectiveness of
the RtI-based math game intervention for number sense growth levels and acceleration of
children in the low-SES intervention group, compared to children in the low-SES control
group and the middle/high-SES reference group.
4 Y. LIANG ET AL.

Various Games Could Improve the Number Sense Development of low-SES Children
Recently, computer-based games provided a good intervention for preschool children. Researchers
distributed iPads with the Bedtime Learning Together Apps that comprised short numerical story
problems to parents in low-income families, and asked them to use the App to teach their children
from kindergarten to first grade. A hierarchical linear model analysis found that the more often
children used the Math App, the higher their math scores were (Berkowitz et al., 2015). Another
computer program Playing with Numbers-2.0, which contained games pertaining to numerical
identification, comparison, combination, and number line estimation, was conducted to 5.5-year-
old children with low math scores. After about 11 weeks’ training, the experimental group exhibited
significantly better performance on math problem solving and mathematical reasoning (Mendizábal,
Villagrán, Guzmán, & Howell, 2017).
In addition, non-symbolic approximate arithmetic training improved preschool children’s math
performance (Park, Bermudez, Roberts, & Brannon, 2016; Szkudlarek & Brannon, 2018). A study
demonstrated that the approximate number system (ANS, the ability to rapidly and accurately
compare quantities presented non-symbolically) abilities for 2- to 5-year-old children got improved
by using the Preschool Number Learning Scheme including comparison and estimation tasks (Van,
Costa, & Passolunghi, 2017). Moreover, a large-scale game-based mathematics curriculum field
experiment lasting 4 months was implemented for kindergartners in the slums of Delhi in India.
Results found that children have made remarkable and lasting progress in spontaneously developing
numerical and spatial abilities (Dillon, Kannan, Dean, Spelke, & Duflo, 2017).
As a matter of fact, computer-based mathematics games are not suitable for low-SES children in
China due to their high cost. It is important to generate appropriate methods for helping low-income
preschoolers increase their number sense knowledge. According to Education and Development
Guidance for 3- to 6-year-old Children from the Ministry of Education of the People’s Republic of
China, which proposed “providing children with abundant and easy-to-use materials, tools, or
articles, and supporting children to do independent painting, handwork, singing, acting, and other
art activities” (MOE, 2012), we developed practical games that were easy to operate for teachers or
parents to improve rural children’s mathematical ability.

The Current Study


Although many researchers from developed countries have exploited several interesting multimedia
software games to train children’s math ability, children in backward and poverty-stricken regions
have very little chance to use tablet PCs. Furthermore, although researchers have examined the
delayed long-term effectiveness of an RtI-based mathematical game intervention on primary chil-
dren’s number sense, none have addressed the longitudinal effect for preschool children. Moreover,
preschool children were sensitive to visualized math intervention activities by triggering multiple
sensory channels to promote the number sense development significantly (Joshi, Dahlgren, &
Boulware–Gooden, 2002). Therefore, the main purpose of the present study was to examine the
effectiveness of the RtI-based math game intervention for number sense growth levels and accelera-
tion of low-SES preschool children. We developed practical intervention games by using self-made
number cards and small bamboo baskets, beans, pills, and so on, which could be obtained anywhere
in daily life. These games were easy for teachers and parents to operate and beneficial for preschool
children’s mathematical learning, thereby improving their math skills. Additionally, we adopted
a longitudinal “pretest–intervention–posttest–delayed posttest” design to conduct RtI-style, hier-
archical, and personalized interventions on counting, number knowledge, and number operations
skills, for low-income rural children aged 5 to 6 years. Besides the low-SES control group, we
included a middle/high-SES reference group to explore the change degree and intensity of low-SES
children’s number sense level and growth trend. We expected that our RtI-based intervention would
reduce the gap in early numerical skills between children with low-SES and middle/high-SES.
EARLY EDUCATION & DEVELOPMENT 5

Our research questions were as follows: (a) Whether playing RtI-based mathematical games that
target the counting, number knowledge and number operations would be effective in improving
number sense of preschool children in the intervention group or not? We hypothesized that the
intervention would be effective and that children in the intervention group would make more
significant progress in number sense and its subareas than their non-intervention peers. (b) To
what extent do our self-developed number sense intervention games improve the number sense level
and growth rate of children in the low-SES intervention group? We supposed that children in the
low-SES intervention group would grow significantly faster than children in the low-SES control
group over the intervention period, and they would catch up with children in the middle/high-SES
group after the intervention. (c) How about the long-term effect of playing RtI-based mathematical
games intervention that target the counting, number knowledge, and number operations? We
assumed that the improvement of the intervention group would be able to maintain until two
months after the intervention.

Methods
Participants
Participants were recruited from two kindergartens: Central Kindergarten at Wangyuan Village in
Xingqing District, and Huisiyuan Kindergarten in Xixia District. Both are suburbs of Yinchuan City
in Ningxia Hui Autonomous Region. Original participants were 164 kindergartners who came from
different SES backgrounds. Children whose parents did not submit the SES questionnaire or without
a normal IQ tested by McCarthy Scales of Children’s Abilities were excluded, and only 148 remained.
Using the mean score of SES (M = 0.03, SD = 0.53) as a cutoff point, children who scored below 0.03
were assigned to the low-SES group (n = 72; 31 females), and those who scored above 0.03 were
assigned to the middle/high-SES group (n = 76; 34 females). The participants from low-SES families
were categorized into two groups with the same proportion of males and females: Half was the
intervention group (n = 36), and the other half was the control group (n = 36). Thirty-six children
were randomly selected from the middle/high-SES category as a reference group to ensure that the
number of participants in the three groups were nearly equal. During the five surveys conducted
over six months, 6 children dropped out of school due to a bad cold, while 6 transferred to other
schools. Finally, a total of 961 children completed the whole experiment: Specifically, 31 (14 girls)
were in the intervention group, 32 (14 girls) were in the control group, and 33 (16 girls) were in the
reference group. The participants ranged in age from 5 years 1 month to 6 years 2 months
(M = 5 years 6 months, SD = 0.52 years). The means and standard deviations of verbal IQ for
participants in the low-SES intervention group, low-SES control group, and Middle/high-SES
reference group were 66.84 (1.99), 66.79 (3.88), and 69.07 (3.34), respectively; and those of memory
IQ for the three groups were 35.62 (1.88), 36.05 (2.06), and 38.68 (2.38), respectively. They were all
in the normal and average range. The demographic information of participants according to SES are
presented at Table A1 (supplementary material). Written informed consent was obtained from each
participant’s parents. Our study was approved by the Institutional Review Board of the School of
Education Ethics Committee at Ningxia University.

Tools
SES Measurement
The items of Socioeconomic Status measurement were selected by reference to the studies of the
PISA (OECD, 2003, 2009), as well as the studies of Fang, Fan, and Liu (2008) and Fan, Fang, Liu,
Lin, and Yuan (2012). SES criteria is based on educational level and occupational status of parents as
well as family resources, which together represent SES better than any of these alone (White, 1982).
The educational level of parents includes the following categories: (1) Primary education and below;
6 Y. LIANG ET AL.

(2) Lower secondary; (3) Vocational/upper secondary; (4) Vocational tertiary; (5) Bachelor’s degree
and above; (6) Not clear. The occupational status of parents includes: (1) Building worker; (2)
Salesperson; (3) Teacher; (4) Soldier; (5) Civil servants; (6) Government cadre; (7) Driver; (8)
Factory worker; (9) Company employee; (10) Pedlar; (11) Farmer; (12) Self-employed company
owner; (13) Catering service staff; (14) Laid-off or unemployed; (15) Others. Family resources are
used for indirect measurement of family income, including 20 items in total: Internet (3 items),
Electric appliance (5 items), Learning space and learning equipment (6 items), Kitchen and bath-
room (4 items), and Entertainment (2 items).
The calculation steps for socioeconomic status are as follows (OECD, 2003, 2009; also see Fan
et al., 2012): (i) Assign values to the variables including educational level of parents, occupational
status of parents and family resources: The educational level of parents is scored according to their
years of education (e.g., 5 points for primary school and 8 points for lower secondary school). The
occupational level of parents is scored according to the international socio-economic index of
occupational status (ISEI) (Ganzeboom & Treiman, 1996), and all occupations are scored between
16 and 90 points (e.g., 29 points for building workers, etc.) One family resource gets 1 point, or
zero. (ii) Filter or convert the above assigned variables: The higher-education or higher-class
occupation scores of the parents are the index scores of parental educational level or occupations.
At the same time, the item response theory was used to perform parameter estimation on family
resources and obtained the parameter estimation index. (iii) Deal with the missing value of
variables by the following method: If a participant has more than two missing variables, then he
or she will be regarded as the missing sample. If only one variable is missing, then his/her
corresponding values will be computed to substitute the missing value by regression analysis
with another two variables. (iv) Transform educational level, occupational status, and family
resources into standard scores, then conduct a main component analysis, and use the following
formula to get SES scores: SES = (β1 × Z education + β2 × Z occupation + β3 × Z family resources)/εf (β1, β2,
and β3 are the factor loadings, εƒ is the root of eigenvalue of the first factor). The mean score of
SES was 0.03 (SD = 0.53), that of low-SES children was – 1.43 (SD = 0.23), and that of middle/
high-SES children was 1.62 (SD = 0.84) in our study.

Streamline Number Sense Screening Tool


The Streamline Number Sense Screening Tool (SNSST) (Jordan et al., 2008) was used to assess the
preschool children’s counting, number knowledge, and number operations abilities. The tool was
originally developed for children aged 5 to 6 years in kindergarten. Specifically, the tool comprises
7 items on counting skills, which examine children’s ability to grasp counting principles and
cardinality, and to use these principles to solve problems related to number sequences. Eleven
items pertain to number knowledge, which assesses preschool children’s ability to identify the
same number, to distinguish different quantities, and to identify numerical values by using
number and digital string size comparison tasks (i.e., number comparison tasks). There are 15
items on number operations, including addition and subtraction calculations in verbal and
nonverbal problems, as well as story problem patterns. All 33 items are suitable for examining
the arithmetic ability of Chinese preschool children. With regarding to the presentation form of
these questions, 21 items are numerical recognition and arithmetic problems presented in digital
form. For example, “Which is bigger: 5 or 4?” “Which number is closer to 5, 6 or 2?” The
remaining 12 items are verbal problems. The semantic meaning of verbal problems is simple
and clear. Therefore, these items are still easy for Chinese preschool children to understand after
they are translated into Mandarin. We closely followed Jordan et al.’s standard procedures when
administering and scoring the test. The Cronbach’s α of the original test is .87, the retest reliability
is .81, and the test-retest reliability of every subarea is above .70 (ps < .05) (Jordan et al., 2008).
Similarly, there was a study using the SNSST to investigate Chinese preschooler’s number compe-
tence (Zhang, 2016), whose Cronbach’s α was .71. The Cronbach’s α is .76 in this study.
EARLY EDUCATION & DEVELOPMENT 7

The McCarthy Scales of Children’s Abilities


The verbal and memory subscales of the McCarthy Scale of Children’s Abilities (MSCA-CR) (McCarthy,
1972) revised by Chen and Li (1994) were used in the present study. The normal score of the verbal
subscale is 34–79 points, and that of the memory subscale is 20–59 points for Chinese children aged
5 years 1 months to 6 years 4 months. The Cronbach’s α is .87 and .75, respectively.

Procedure
All the tests were conducted individually by trained instructors who were undergraduate students of
psychology in a quiet, well-lit classroom in the two kindergartens. Participants were pre-tested on SES,
number sense, verbal, and memory skills in September. The RtI-based math intervention and storybook
reading and three immediate number sense posttests were conducted from October to January of the
next year. Then, a delayed posttest was completed two months later in March of the next year. Children
in the low-SES intervention group received a number sense game intervention, whereas those in the low-
SES control group and middle/high-SES reference group were provided storybook reading instruction by
instructors in the classroom. The number sense intervention game consisted of two 15–30 minutes
sessions conducted each week over three months for a total of 24 sessions, in small groups.
Children were divided into 5–6 groups equally according to their initial number sense levels. In
the process of playing, the team members were adjusted according to the requirements of the
experimental task. If the number of children in the last group was not enough to form a pair,
then the children who were not active in the previous interventions would been to choose to
participate in the new round of intervention again. The whole of intervention procedure included
three phases. In the first phase, we aimed to promote counting and number knowledge; in
the second phase, we aimed to improve children’s abilities in number knowledge and number
operations; and in the third phase, we further implemented a high level of interventions on counting,
number knowledge, and number operations. Each phase consisted of 8 sessions over one month. At
the end of each intervention phase, children participated individually in the SNSST test immediately.
According to the RtI-based intervention paradigm, the duration and sessions of each intervention
phase were adjusted at any time, depending on the response and performance of the participants.

Intervention Materials and Tasks


The number sense intervention included 10 games that had three difficulty levels. The difficulty of
these games was increased gradually across the three phases. The setting of these games was based on
the implementation principles of Tier 2 of the RtI which asked the experimenter to adjust the task
difficulty and the intervention progress according to the improvement and mastery level of the
participants in each task. Three games in the first phase, including “ordering digital card,” “splitting
beans,” and “selecting the pill,” were used to improve children’s counting abilities, such as recogniz-
ing numbers and one-to-one correspondence. The second three games, including “taking cards
according to rules,” “Loto transferring card,” and “Bingo filling in grids,” were used to upgrade
children’s number knowledge, such as comparing symbolic numerals and understanding numbers.
The last four games, including “sprinkling buttons,” “throwing dice,” “card double-click battling,”
and “refueling the plane,” were used to enhance children’s nonverbal calculation abilities, such as
problem-solving, number operations, and number combinations.
Details of the intervention games for children in the low-SES group are as follows:

Ordering Digital Card. This game aimed to train kindergartners to master the rules of counting, in
small groups of 4 to 6 children. A set of cards with numbers 1–10 were given to each child in
a random order. Children were told, “Please place the cards according to the certain rules: The first
rule is from small to large, the second is from large to small, and the third is at a certain interval (one
digit, two digits, or three digits).” Children who mastered the first rule could move on the second
stage, from easy to hard.
8 Y. LIANG ET AL.

Splitting Beans. This activity addressed children’s number sequence ability through counting. The
materials were toy Bear senior,2 toy Bear junior, small bamboo baskets, and some beans. There were
two story situations: (a) “Bear senior and Bear junior are arguing about who has picked more beans.
Can you tell who has more beans and why?” (b) “Bear senior could not get out to pick beans, because
he sprained his leg this morning. As his friend, Bear junior had to share his beans. Now could you
help Bear junior to count how many beans he should share with Bear senior, so that they have the
same amount of beans?” At the beginning of the intervention, children in the intervention group
were instructed to solve questions by using finger counting or counting aloud, and then by silent
thinking after they have mastered the counting rules.

Selecting the Pill. This game aimed to strengthen children’s mastery of number sequence. The
instructor and children together laid 10–15 white toy pills out in a row, and then the instructor explained
every sequence number (at the same time, encouraging children to speak out loud), “First, second,
third … ” Next, children were asked to help the sick baby bear select a pill on request. In the first task: The
instructor told the children, “The baby bear was too weak to get out of bed to take his medicine. Now he
needs to take the nth pill, please pick this pill out for him.” In the second task: The instructor placed the
red and yellow pills in intervals again, red pills being marked as odd numbers and yellow ones as even
numbers, and asked participants to pick out the nth red (or yellow) pill.

Taking Cards according to Rules. This task aimed to improve children’s ability to identify the same
numeral. Children were asked to play this game in pairs. They were given 40 cards in which there were
two sets of number cards with 1–10 digits and two sets of shape cards with 1–10 dots. These cards were
the same sizes. Number cards were placed face down in front of participants. The instructor guided the
children, “Let’s play a ‘card train’ game. Take turns lining your own cards on the table one by one. Please
take one at a time from the top of your cards. If the number on your card is the same as the number or
dots of any card on the train, then you could win all the cards between the two cards. The cards you win
should be put aside. The child who gets the most is the winner.”

Loto Transferring Card. This task addressed children’s numerical identification ability, in small
groups of four children. The game materials included 10 basic cards (1–10 squares or triangles on
each card) and 10 matching cards. For example, if the basic card was “□□□,” then its matching card
was “3□.” If the basic card was “△△△△△,” then its matching card was “5△.” First, the instructor
explained the meaning and relationship between the two types of cards to children, then distributed
a basic card to each child. Subsequently, the instructor placed the matching cards upside down
randomly and told the children, “Now, we will pick up these cards (i.e., the matching cards) on the
table, and you will get one card. You should pay attention to the card in your hand (i.e., the basic
card). If you captured a card with the same as yours, “It’s a match,” then you should say ‘I find it!’”
Children first played cards with 1–5, and then with 1–10.

Bingo Filling in Grids. This game aimed to strengthen children’s ability to identify the same
quantity, in small groups of 2 to 4 children. The game materials included two types of cards: One
was the basic card and the other was the matching card. For the basic card, different numbers were
written in each grid. Cards with low difficulty level were comprised of 4 grids with 1–5, and cards
with high difficulty level were comprised of 9 grids with 1–10. The other was 36 matching cards with
different dots within 1–10. Each child was given a basic card. The instructor put the matching cards
in a random order. Children were required to identify the matching card corresponding to the
number in the grid of the basic card and to place the matching card on the corresponding grid
quickly and accurately. The child who filled in all the grids first won the game.

Sprinkling Buttons. This game aimed to improve children’s ability to perform calculations with
reference materials in pairs. The instructor presented each pair with a square paper with a circle on it
EARLY EDUCATION & DEVELOPMENT 9

and gave some buttons to one of the children. The child sprinkled buttons on the paper and asked
his or her partner to guess how many buttons there were in and out of the circle. The two children
played the game in turns.

Throwing Dice. This game aimed to train young children to master nonverbal operations and
number combinations with 3 to 4 children in a group. The game materials consisted of four
bunny pieces with different colors, two dice with 1–3 dots, and four checkerboards with 11 squares.
Each square was marked with a number from 0–10. The instruction was, “Let’s play a dice game.
Each of us has a bunny chess piece, standing at the starting point (‘0ʹ). Now we roll the dice in turns,
and then move forward according to the points on the dice. When we go forward, we must speak out
loud the number of the grid we’re on. For example, if you are on the second grid and throw a ‘3,’ you
just move the piece and say, ‘3, 4, 5; I’m at 5!’ The game is over when all four of us move from 0 to
10 or more.” The experimenter could add one dice after children were more proficient.

Card Double-click Battling. This game aimed to strengthen kindergartners’ calculation ability. It was
conducted in pairs. The instructor provided each child with 10 cards: Half were numbers 1–5 and the
other half were 1–5 dots. Each child was asked to randomly divide his or her 10 cards into two piles and
placed them upside down on the table. The instruction was, “When I say ‘start,’ everyone should use your
two hands to turn over the first card of every pile at the same time, and then calculate the sum of the
numbers and dot on cards. The child with the biggest number can take away another kid’s cards.
Whoever has the most cards at the end of the game is the winner.” Children who had cards with larger
quantities could take away others’ cards with smaller quantities. They retained their own cards if both
had equal numbers. The score was measured as the total sum for each group.

Refueling the Plane. This game aimed to further strengthen children’s number combination ability.
The experimenter first placed a plane model in front of the participants and the fuel tank beside
them, then told the children, “Let’s play a game that oil a plane.” The experimenter then explained
the game rules, “Some large and small planes need refueling. They are parked on the airfield. Each
aircraft needs different amounts of fuel. The number of oil tanks a plane needs is written on it, such
as ‘2 + 4.’ You need to refuel the planes in front of you. You’d better fuel nicely and quickly.”
During the implementation of the intervention game, the numbers in each task were changed to
increase the difficulty according to the proficiency of the participants. If the participant made a mistake,
the instructor would ask him or her to conduct self-check. If a child came across difficulties in the
process, he or she could discuss them with his or her peers. If all the children in the same group could not
solve the problem correctly, the instructor would provide appropriate guidance.

Storybook Reading
The control group and the reference group listened to stories, which were from the journal Preschool
Pictorial, including a series of books published by Chinese children’s Press & Publishing Group, whereas the
math game interventions were implemented with the intervention group. Additionally, content including
the Red Kangaroo’s Self-preservation story and Good Habits stories was suitable for 3- to 6-year-old children.

Data Analysis Plan


We conducted a latent growth modeling to investigate whether the development of the number sense
and its subareas was linear and whether there were significant individual differences in children’s
initial number sense levels and growth rate at the first four time points. The growth differences
between the intervention group and the control group were examined by conducting a multiple-
group latent growth model, which was extended to examine differences in growth trajectories across
10 Y. LIANG ET AL.

multiple manifest (observed) groups (Kim, Mun, & Smith, 2014). Then the intervention effects were
examined by using univariate analysis of variance to compare the number sense performance of the
three groups at pretest, the third immediate posttest, and delayed posttest. The data analyzes were
performed by using Mplus 8.0 and SPSS 22.0. The maximum likelihood estimation was used to
conduct the latent growth modeling.

Results
The Cronbach’s α for the full number sense battery ranged from .76 to .81 across all five time points.
Alpha coefficients ranged from .58 to .69 for counting, from .59 to .69 for number knowledge, and
from .73 to .83 for number operations across all five time points (ps < .05).

The Growth Trend Differences of Number Sense in the Three Groups


The unconditional growth curve model was used to investigate whether the number sense of the
three groups increased linearly. As shown in Figure 1, the model consists of two levels: Level 1
calculates the number sense scores of every child. NSit = αi + βi λi + εit. NSit is the observed indicator
for an individual i at time point t. It is explained by two latent factors, αi and βi, where αi reflects and
represents the latent intercept (initial level of number sense) and βi represents the individual latent
slope (growth rate); λi specifies the time course, and εit represents a random error of individual i at
time t. Growth modeling does not require the same two measurement intervals to be the same
(Kline, 1998). In this study, the first test time score was set to 0. Because the development of the
number sense was linearly increased, the time scores of the three immediate posttests after the
intervention were set to equal intervals, which were 1, 2, and 3, respectively. Therefore, the intercept
of the model is the estimated mean of number sense scores at the first test, and the slope of the
model is the average growth score of the number sense during the first four continuous tests. Level 2
considers the intercept (I) and slope (S) as dependent variables: αi = μα + ξαi; βi = μβ + ξβi, where μα
and μβ represent the mean of intercept and slope, respectively; ξαi, ξβi are the residuals of the
intercept and slope of individual i, respectively.
Model fit was determined by using the Comparative Fit Index (CFI) and Tucker Lewis Index
(TLI) with critical values ≥.90, and the standardized root mean squared residual (SRMR), or root
mean squared error of approximation (RMSEA) close to .06 (Marsh, Hau, & Wen, 2004). The
unconditional growth model fitted the data better, the index of the intervention group was: χ2/

Intercept Slope ξβi


ξαi 0
1 1
1 2 3
1 1

NS T0 NS T1 NS T2 NS T3

ε1 ε2 ε3 ε4
Figure 1. The unconditional growth model of number sense development.
Note: NS means number sense. ε means the residual of subject at time T.
EARLY EDUCATION & DEVELOPMENT 11

df = 1.40, RMSEA = .07, CFI = .97, TLI = .97, SRMR = .08; the control group was: χ2/df = 0.76,
RMSEA = .07, CFI = .98, TLI = .92, SRMR = .03; and the reference group was: χ2/df = 1.09,
RMSEA = .05, CFI = .98, TLI = .97, SRMR = .06.
The growth trend of number sense total score and its subareas for the three groups are depicted in
Figure 2, and the parameters are in Table 1. As expected, the average number sense total scores of
the three groups were significantly greater than 0, showing significant linear growth trend through-
out the four continuous tests. Acceleration of the low-SES intervention group was significantly faster
than the low-SES control group, and almost caught up with the middle/high-SES group. The
intervention group increased by an average of 2.28 points per month, while the control group
increased by 1.96 points, and the reference group increased by 2.09 points. With respect to the three
subareas of number sense, the initial scores of the three groups were all significantly greater than 0
(intercept ps < .001), and the three subareas of the three groups increased linearly at the first four
continuous tests (all the average slopes were significantly larger than 0). The variations of the
intercept were all significantly greater than 0, indicating that significant individual differences existed
among the three groups in the initial scores of the three subareas. Specifically, on counting, the mean
score of the low-SES intervention group increased by an average of 0.34 points per month, and the
growth rate was significantly greater than 0 (p < .001). The intervention group increased steadily and
almost caught up with the middle/high-SES group, whereas the scores of the low-SES control group
did not increase significantly. On number knowledge, all the three groups showed a linear growth
trend, their initial levels and growth rates were all significantly greater than 0 (ps < .001), and the
growth rate of the low-SES intervention group was the largest. The variations of the slope were not
significantly greater than 0, indicating that there were no significant individual differences in the
growth rates among the three groups. With respect to the specific tasks of number knowledge,
compared with children in the middle/high-SES reference group, children in the low-SES control

30 Number Sense 8 Counting


25
6
20
means
means

15 4
10
2
5
0 0
T0 T1 T2 T3 T4 T0 T1 T2 T3 T4

Number Knowledge Number Operations


10 15
8
10
means

means

6
4
5
2
0 0
T0 T1 T2 T3 T4 T0 T1 T2 T3 T4

Low-SES intervention group Low-SES control group

Middle/High-SES reference group

Figure 2. Fitted trajectories of number sense and subareas for the three groups over five time points.
Note: T0 indicates pretest, T1-T3 indicates the immediate posttest after each round of intervention, T4 indicates
delayed posttest after 2 months. Error bars 1 SE.
12 Y. LIANG ET AL.

Table 1. Results of growth curve model of the total number sense and its subareas scores for the three groups over the first four
time points.
Number Number
Group Parameter Number Sense Counting Knowledge Operations
Low-SES intervention M (SE) intercept 14.81 (0.70)*** 4.76 (0.24)*** 4.41 (0.49)*** 6.03 (0.52)***
group slope 2.28 (0.18)*** 0.34 (0.08)*** 1.00 (0.16)*** 1.62 (0.16)***
(n = 31) Variance intercept 19.46 (6.94)** 1.18 (0.39)** 3.52 (1.52)* 5.48 (1.82)**
(SE) slope 0.34 (1.03) 0.10 (0.06) 0.06 (0.23) 0.24 (0.40)
Low-SES control M (SE) intercept 16.38 (0.84)*** 5.41 (0.14)*** 5.03 (0.47)*** 7.24 (0.58)***
group slope 1.96 (0.22)*** 0.09 (0.05) 0.54 (0.16)*** 0.79 (0.23)***
(n = 32) Variance intercept 12.35 (5.27)* 0.22 (0.24) 4.38 (1.05)*** 4.79 (1.79)**
(SE) slope 0.35 (0.50) –0.01 (0.04) 0.27 (0.14) 0.93 (0.45)*
Middle/high-SES M (SE) intercept 19.41 (0.89)*** 5.72 (0.18)*** 6.18 (0.30)*** 7.38 (0.62)***
reference group slope 2.09 (0.33)*** 0.10 (0.07) 0.64 (0.10)*** 1.40 (0.21)***
(n = 33) Variance intercept 15.62 (6.49)* 0.53 (0.20)** 2.48 (0.83)** 6.49 (2.93)*
(SE) slope 0.68 (0.57) 0.08 (0.07) 0.00 (0.16) 0.29 (0.41)
Note: ***p < .001, **p < .01, *p < .05.

group scored an average of 0.41 points lower on number identification but 0.37 points higher on
number comparison and number distance judgment. Overall, the low-SES control group had lower
score on number knowledge than the middle/high-SES reference group over the course of the study
(as can be clearly seen in Figure 2). On number operations, the three groups showed a linear growth
trend, and their initial levels and growth rates were all significantly greater than 0 (ps < .001). The
scores of the low-SES intervention group increased significantly after three months of intervention.
The score of children in low-SES intervention group increased by 1.62 points per month, and that of
the middle/high-SES group increased by 1.40 points, whereas that of the low-SES control group grew
by 0.79 points, which was half of the increased scores exhibited by the former two groups.
Specifically, low-SES children significantly lagged behind middle/high-SES children on number
combinations and story problems presented in verbal form. The variation of the slope for the low-
SES intervention group (0.24) and the middle/high-SES reference group (0.29) was not significantly
larger than 0, showing that their growth rate had no significant individual differences. However, the
variation of the slope for the low-SES control group was larger than 0 (p < .05), indicating
a significant individual difference.
Additionally, we used a multiple group growth model to further explore the growth differences
between the low-SES intervention group and the low-SES control group. First, intervention/non-
intervention was selected as an independent variable and gender and IQ as covariances. The coefficient
β1 was used as an independent variable of intercept in the intervention condition to determine the effect
of intervention on the initial number sense level (intercept), and β2 was used as an independent variable
of slope. The model fitted well: χ2/df = 9.06, RMSEA = .04, CFI = .98, SRMR = .07. Results (see Figure 3)
showed that the predictive coefficient for the intercept was 2.16, which meant that the effect of
intervention on the initial number sense score of the low SES intervention group was significantly
greater than 0. Additionally, after 3 months of the intervention, the predictive coefficient for the slope
became – 1.37, reaching a significant level (t = – 4.77, p < .001), indicating that the growth rate of children
in the intervention group increased successively during intervention. Moreover, the correlation between
the intercept and slope of children in the intervention group was not significant (r = – .04, p = .527),
which revealed that the growth rate of children’s number sense had nothing to do with their initial
number sense level during the intervention.

Differences of growth performance of number sense in the three groups and the delayed
effect of intervention
Table 2 provides the means and standard deviations (by groups) and their differences at pretest, the
third immediate posttest, and delayed posttest. The total scores of the three groups had significant
EARLY EDUCATION & DEVELOPMENT 13

Intervention

2.16 (0.87) –1.37 (0.29) ***


0.036

5.74 (1.84) Intercept Slope 0.03 (0.23)


0

1 1
1 2 3
1 1

NS T0 NS T1 NS T2 NS T3

Figure 3. The conditional growth model of children in the low-SES intervention group and control group.
Note: NS means number sense.

differences at the three time points. At pretest, children in the middle/high-SES group scored higher
than the intervention group and the control group on counting, F(2,93) = 4.88, p = .01, ηp2 = .10;
number knowledge, F(2,93) = 6.03, p = .003, ηp2 = .12; and total scores, F(2,93) = 5.55, p = .009, ηp2
= .10. But no significant differences appeared on number operations among the three groups. There
were no significant differences between the low-SES intervention group and the control group on
both the total and the three subareas. At the third immediate posttest after interventions, the number
sense total scores of the low-SES intervention group increased significantly and almost caught up
with the middle/high-SES reference group. At delayed posttest, no significant differences were found
in the scores of counting and number knowledge for the three groups, and the scores of number
operations for middle/high-SES children were significantly higher than those of the two low-SES
groups. However, the number sense total score of the low-SES intervention group (M = 24.13) was

Table 2. Number sense scores and differences by groups at pretest, the third immediate posttest, and delayed posttest M(SD).
Low-SES Low-SES Middle/high-SES
Number sense Test time intervention group control group reference group F(2, 93) ηp2
a a b
Counting T0 4.87 (1.41) 5.41 (0.91) 5.78 (1.17) 4.88** .10
(k = 7) T3 5.81 (0.60) 5.69 (0.82) 6.03 (0.81) 1.74 .04
T4 5.74 (0.82) 5.59 (0.95) 5.61 (1.09) 0.23 .01
a
Number Knowledge T0 3.90 (2.55) 5.03(2.55) a, b 5.97 (2.02) b
6.03** .12
a, b
(k = 11) T3 7.32 (1.74) 6.88 (2.28) a† 7.84 (2.06) b†
1.85 .04
T4 7.87 (2.13) 7.09 (2.01) 7.88 (1.99) 1.56 .03
Number Operations T0 6.03 (3.10) 6.84 (2.95) 7.03 (3.27) 0.92 .02
a, b
(k = 15) T3 10.84 (3.28) 9.53 (3.78) a 11.42 (2.99) b†
2.68 .06
a
T4 9.94 (3.58) 9.13 (3.96) a 11.03 (2.77) b
2.48 .05
a
Total scores T0 14.84 (5.43) 16.38 (3.66) a 18.80 (5.10) b
5.55** .10
a
T3 23.97 (4.31) 21.41 (5.30) b 25.30 (4.83) a
5.44** .07
a
T4 24.13 (5.23) 21.31 (5.33) b 24.52 (4.58) a
3.85* .05
Note: T0, T3, T4 indicates pretest, immediate posttest after the intervention, and delayed posttest, individually. k indicates the
number of items.
**p < .01, *p < .05, †p < .10.
a, b
represents the significant difference of the means between each of two groups in Bonferroni multiple comparisons.
14 Y. LIANG ET AL.

Table 3. Comparison of the growth scores of three groups of children at pretest, the third immediate posttest, and delayed
posttest M(SD).
Low-SES Low-SES Middle/high-SES
intervention group control group reference group F(2, 93) ηp2
T3-T0 9.13 (4.19) a 4.81 (4.32) b 6.52 (4.42) b 7.98*** .15
T4-T0 8.71 (4.91) a 4.53 (5.70) b 5.73 (4.38) b 5.77** .11
T4-T3 -0.42 (3.71) -0.28 (4.80) -0.79 (4.83) 0.11 .00
Note: **p < .01, ***p < .001.
a, b
represents the significant difference of the means between each of two groups in Bonferroni multiple comparisons.

the same as that of the middle/high-SES control group (M = 24.52), and these two groups performed
significantly better than the low-SES control group.
To further explore children’s growth specifically, the increase in the number sense scores from
pretest to the third immediate posttest (i.e., T3 –T0), and to delayed posttest (i.e., T4–T0) were
analyzed. As depicted in Table 3, the incremental difference of number sense between the three
groups at pretest and the third immediate posttest was significant, F(2, 93) = 7.98, p < .001, ηp2 = .15.
There were also significant differences in the growth scores at pretest and delayed posttest among the
three groups, F(2, 93) = 5.77, p = .004, ηp2 = .11. Post hoc tests showed that the low-SES intervention
group increased more than the control group and reference group on number sense at the third
immediate posttest (p = .045) and delayed posttest (p = .051). There was no significant difference in
growth scores from the third immediate posttest to the delayed posttest among the three groups of
children, indicating that the incremental scores of the total number sense among the three groups
were almost the same in two months after the intervention. However, the three subareas showed
different growth trends during this period. As depicted in Figure 2, the performance in counting
neither increased nor declined for the three groups. The performance in number knowledge
continued to increase for children in the low-SES intervention group, while showed little increase
for the other two groups. The growth trend in number knowledge of the intervention group
maintained till the delayed posttest, but that of the other two groups hold a parallel and slow
pace. Meanwhile the number operations presented a certain degree of decline for the three groups.

Discussion
Building on previous investigations (e.g., Dillon et al., 2017; Dyson et al., 2013), we demonstrated
that the key areas of number sense can be boosted for kindergartners from low-income families.
Although most children seemed to make progress from their regular mathematics curriculum, the
intervention provided them with added benefits in a relatively short period. As expected, children in
the intervention group made more significant progress than their non-intervention peers in number
sense, and the intervention effects remained two months later after the intervention. Given the
importance of kindergarten performance to long-term outcomes, the findings are especially impor-
tant and meaningful. In addition, compared with the previous study (Dyson et al., 2013) that only
included an intervention group and control group, we added a middle/high-SES reference group.
Therefore, we could not only compare the number sense of children with different SES levels during
kindergarten, but also explore the degree of promotion effect of math game interventions on low-
SES children.
Considering the development trends of number sense for children in the low-SES control group
and the middle/high-SES reference group, we found that the two groups showed no differences on
basic counting ability, a parallel equidistant growth pace on number knowledge, and a large gap on
number operations over the course of the four months in the present study. These results suggest
that the difficulty between counting, number knowledge, and number operations is different. In
counting skills tasks, according to studies of Jordan and his colleagues (Jordan et al., 2008, 2006),
EARLY EDUCATION & DEVELOPMENT 15

children were asked to complete two kinds of tests, one is counting the number of stars, and the
other is judging whether counting principles of the finger puppet are “OK” or “not OK”.
Discriminating or comparing the magnitudes of two or three numbers is the number knowledge
task. And in number operations tasks, children were asked to solve story problems (“‘Xiaodong had
2 pennies. Xiaoli gave him 3 more pennies. How many pennies does Xiaodong have now?’”) and
number combinations (“‘How much is 2 and 3?’”). Children gradually gain number knowledge
through learning and practicing their early obtained counting skills (Griffin, Case, & Siegler, 1994),
and know that numbers later in the counting sequence have larger quantities than earlier numbers.
Afterward, number knowledge helps children think about mathematical problems, perform verbally
presented calculation problems, and obtain calculation skills, which constitute the main components
of number operations (Jordan et al., 2006). Additionally, the difficulty values of counting
(0.89 ~ 0.99), number knowledge (0.73 ~ 0.87), and number operations (0.58 ~ 0.82) decreased
progressively in our study. These results accorded with the principle of continuous development and
increasing difficulty of the number sense subareas, which was also the basis for Jordan and his
colleagues to build the five-component theory model of number sense and to develop the SNSST
(Jordan et al., 2008, 2010, 2006). Our findings about the developmental trend of number operations
(story problems and number combinations, Figure 2) are consistent with Jordan et al.’s findings.
Children from low-income families developed slowly in terms of story problems and number
combination tasks, whereas those from middle/high income families developed relatively fast
(Jordan et al., 2006). A recent study also showed that individuals who received rich exposure to
mathematical material in their homes and whose mother had high educational level, might perform
well on intuitive, non-symbolic mathematical tasks as well as learned, symbolic mathematical tasks,
even if the abilities underlying these skills and tasks were not causally related (Dillon et al., 2017).
Children in the middle/high-SES group could acquire number operations skills from their families
and show a better performance on these skills, while children from low-income families could not.
With respect to the degree of promotion effect of the math intervention, children in the
intervention group grew most in the number knowledge and number operations areas and per-
formed as well as the high-SES reference group. One of the main reasons is that the math game
intervention on number sense was effective for children in the intervention group. Firstly, the games
matching physical representations of numerical information (e.g., a certain number of dots, triangles,
squares, etc.) with symbolic numbers helped children understand and judge numbers correctly by
using the previously acquired counting strategy. Moreover, our math game intervention materials
were made of the buttons, checkerboards, cards, and toys (aircraft model), which were readily
available everywhere for low-SES children and convenient for experimenters to guide children
using the counting strategy to solve story problems and to determine number combinations. For
example, in the “throwing dice” game, participants were asked to follow the starting point (e.g., 2) in
conjunction with moving chess pieces while speaking every number out loud (such as interpreting
“3” for 3 steps and saying “3, 4, and 5”). This specific and direct guidance activated the multisensory
channels of participants and helped them gradually realize that “3” is not only the cardinal value of
a number set, but also the starting value of the second addend. Thus, children could learn to use the
counting-up strategy to solve additive problems, and internalize the numerical knowledge that they
have learned to solve mathematical problems correctly by receiving reinforcement exercises from low
to high level. This constant practice and operation essentially change children’s existing cognitive
structure so that the organism conforms to the acquired experience, thus constructing their own low
schema into a new and more complex schema (Piaget & Inhelder, 1969). Secondly, a targeted multi-
level RtI-based game intervention in this study allowed the instructors to perform a hierarchical
intervention on the participants. The implementation of the intervention included three phases from
low to high levels, according to the principle of continuous development and gradient of increasing
in the difficulty of the number sense subareas (Jordan et al., 2010, 2006). For example, in the “Loto
transferring card” game, children first matched the cards with subitizing range numbers: 1–5, and
then counting range numbers: 6–10, and finally the numbers from 1 to 10. Only after they had
16 Y. LIANG ET AL.

mastered the first round could they play the next round of games. During the RtI-based intervention,
the instructors paid attention to the response of the participants, discovered problems in time,
incorporated appropriate feedback and training, and facilitated independent thinking. Such a step-by
-step and targeted systematic intervention significantly improved children’s mathematical thinking
ability and produced a good intervention effect (Grosche & Volpe, 2013).
Furthermore, a small-group intervention method provided an effective teaching research for
children in underdeveloped areas. Children in this study played mathematical intervention games
in small groups and got improved in number sense. The small-group intervention helped children
receive timely assistance and instruction from peers when they made mistakes or failed to answer
questions (Hinton et al., 2013) and then gained growth in early numerical skills (Jordan et al., 2012).
This design coincided with the core idea of Vygotsky’s “zone of proximal development,” which
asserts that children’s potential development level is achieved and improved under adults’ guidance
or in cooperation with their competent peers (Vygotsky, 1978). Children in the same group had the
same background and similar developmental level of number sense, and the members could
communicate with and learn from each other, share effective learning strategies, and strengthen
their learning motivation through game interaction and competition to gain learning experience and
ultimately achieve progress together.
Unexpectedly but reasonably, the performance of some children in the intervention group declined
on number operations tasks two months later. Maybe because the items in the number operations tasks
were relatively difficult and imposed high requirements in terms of mathematical thinking ability for
children aged 5 years. Although there was a significant increase at the immediate posttest (T3) after three
months of intervention, this change could be attributed to memory effect or level of mastery in
continuous practice rather than skilled proficiency of understanding. Previous studies have confirmed
that the RtI-based intervention paradigm could identify children at risk for mathematical difficulties
(Fletcher & Vaughn, 2010). Difficulties appearing in number combination and story problem solving
skills in early childhood are key predictive features of learning difficulties in primary school (Jordan,
Kaplan, & Hanich, 2002). Therefore, children who performed poorly on number operations tasks and
were not sensitive to intervention could be at risk for mathematical difficulties.

Implications, Limitations, and Future Research


The results of this research indicate that the considerable easy-to-get materials of number sense
intervention for preschool children would be easily implemented in kindergarten classrooms. We
have provided a new perspective for teaching mathematics to children in underdeveloped areas.
Cultivating and training children’s number sense is beneficial not only to transferring obtained
mathematical knowledge to daily life, but also to laying a good foundation and preparing them for
their future life and work in a highly developed scientific and technological society (Anghileri, 2000/
2007). Our intervention methods are available for teachers, parents, and individuals from low-
income backgrounds. Materials cost little or nothing and could be used to create a manipulatable
mathematical game. In the future, we should further develop a systematic intervention program and
train rural preschool teachers to apply these games to daily teaching activities.
Additionally, the RtI-based intervention provides evidence for further implementation and
a valuable reference for the evidence-based RtI theory regarding the number sense of children in
rural areas. RtI models emphasize the importance of teachers in intervention; teachers’ response to
teaching could prevent students with learning difficulties from suffering more “waiting for failure”
(Fletcher & Vaughn, 2010). This distinctiveness of RtI models provides us with a way to think about
how to standardize the process of teacher ability training, improve teachers’ response to teaching,
and help teachers use more professional strategies to meet the unique educational demands of
students with learning disabilities and children in general.
A potential limitation of the study is that we did not screen out children at risk for mathematical
difficulties. The early identification of these children is highly necessary, because research has shown
EARLY EDUCATION & DEVELOPMENT 17

that early intervention could potentially prevent later learning difficulties (Aunio & Mononen, 2018).
Additionally, the intervention intensity and duration were possibly insufficient for the high order
math skills of low-SES preschool children. Therefore, in the future, we should further screen and
classify children with mathematical difficulties at different levels, and implement sustained and
targeted individual interventions to help these children catch up with their peers and adapt to
regular mathematics learning in primary schools.

Notes
1. A power analysis is conducted by using the Monte Carlo Simulation (Van Vleet & Thompson, 2011), the effect
size is 0.12, which means our N is enough to support estimation of growth curve model in the current study.
2. Bear senior and Bear junior are the characters in the cartoon, which Chinese children are familiar with.

Acknowledgments
We wish to warmly thank all the preschool children and their parents and schools for their participation.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the Ministry of Education of the PRC under Grant [number 18XJA190001]; National
Natural Science Foundation of China under Grant [number 31260239] awarded to Linjin Zhang; and Fundamental
Research Funds for the Central Universities under Grant [number GK201704020] awarded to Yuan Liang.

ORCID
Yuan Liang http://orcid.org/0000-0002-5747-4771
Lijin Zhang http://orcid.org/0000-0001-8172-2977

References
Anghileri, J. (Ed.). (2007). Teaching number sense (W. B. Xu, Trans). Beijing, PRC: Beijing Normal University Press.
(Original work published 2000).
Aunio, P., & Mononen, R. (2018). The effects of educational computer game on low-performing children’s early
numeracy skills–An intervention study in a preschool setting. European Journal of Special Needs Education, 33(5),
677–691. doi:10.1080/08856257.2017.1412640
Berkowitz, T., Schaeffer, M. W., Maloney, E. A., Peterson, L., Gregor, C., Levine, S. C., & Beilock, S. L. (2015). Math at
home adds up to achievement in school. Science, 350(6257), 196–198. doi:10.1126/science.aac7427
Chen, G. P., & Li, D. (1994). The revision of the Chinese norm of McCarthy scale of children’s abilities. Chinese
Journal of Clinical Psychology, 2(3), 135−140. doi:10.16128/j.cnki.1005-3611.1994.03.002
Cowan, R., & Powell, D. (2014). The contributions of domain-general and numerical factors to third-grade arithmetic skills
and mathematical learning disability. Journal of Educational Psychology, 106(1), 214–229. doi:10.1037/a0034097
Crowell, N. A., Polanski, C., Lesiak, M. A., McCarthy, J., & Ramey, C. T. (2011). An exploratory study of a
response-to-intervention approach with preschool children from low-income urban neighborhoods. Effective
Education, 3(2), 89–107. doi:10.1080/19415532.2011.665768
Dillon, M. R., Kannan, H., Dean, J. T., Spelke, E. S., & Duflo, E. (2017). Cognitive science in the field: A preschool
intervention durably enhances intuitive but not formal mathematics. Science, 357(6346), 47. doi:10.1126/science.aal4724
Dyson, N. I., Jordan, N. C., & Glutting, J. (2013). A number sense intervention for low-income kindergartners at risk
for mathematics difficulties. Journal of Learning Disabilities, 46(2), 166–181. doi:10.1177/0022219411410233
Evans, G. W., & Schamberg, M. A. (2009). Childhood poverty, chronic stress, and adult working memory. Proceedings
of the National Academy of Sciences of the United States of America, 106(16), 6545–6549. doi:10.1073/
pnas.0811910106
18 Y. LIANG ET AL.

Fan, X. H., Fang, X. Y., Liu, Y., Lin, X. Y., & Yuan, X. J. (2012). The effect of social support and social identity on the
relationship between perceived discrimination and socio-culture adjustment among Chinese migrant children. Acta
Psychologica Sinica, 44(5), 647−663. doi:10.3724/SP.J.1041.2012.00647
Fang, X. Y., Fan, X. H., & Liu, Y. (2008). Perceived discrimination and loneliness: Moderating effects of coping style
among migrant children. Psychological Development and Education, 4(11), 93−99. doi:10.16187/j.cnki.issn1001-
4918.2008.04.011
Fletcher, J. M., & Vaughn, S. (2010). Response to intervention: Preventing and remediating academic difficulties. Child
Development Perspectives, 3(1), 30−37. doi:10.1111/j.1750-8606.2008.00072.x
Friso-Van Den Bos, I., Van der Ven, S. H. G., Kroesbergen, E. H., & Van Luit, J. E. H. (2013). Working memory and
mathematics in primary school children: A meta-analysis. Educational Research Review, 10, 29–44. doi:10.1016/j.
edurev.2013.05.003
Fuchs, D., & Fuchs, L. S. (2006). Introduction to response to intervention: What, why, and how valid is it? Reading
Research Quarterly, 41(1), 93–99. doi:10.1598/RRQ.41.1.4
Fuchs, L. S., Geary, D. C., Compton, D. L., Fuchs, D., Schatschneider, C., Hamlett, C. L., … Changas, P. (2013). Effects
of first-grade number knowledge tutoring with contrasting forms of practice. Journal of Educational Psychology, 105
(1), 58–77. doi:10.1037/a0030127
Ganzeboom, H. B. G., & Treiman, D. J. (1996). Internationally comparable measures of occupational status for the
1988 international standard classification of occupations. Social Science Research, 25(3), 201–239. doi:10.7763/
LNSE.2015.V3.173
Gersten, R., Jordan, N. C., & Flojo, J. R. (2005). Early identification and interventions for students with mathematics
difficulties. Journal of Learning Disability, 38(4), 293–304. doi:10.1177/00222194050380040301
Griffin, S. (2004). Building number sense with number worlds: A mathematics program for young children. Early
Childhood Research Quarterly, 19(1), 173–180. doi:10.1016/j.ecresq.2004.01.012
Griffin, S. A., Case, R., & Siegler, R. S. (1994). Classroom lessons: Integrating cognitive theory and classroom practice.
In K. McGilly (Ed.), Rightstart: Providing the central conceptual prerequisites for first formal learning of arithmetic to
students at risk for school failure (pp. 25–49). Cambridge, MA: MIT Press.
Grosche, M., & Volpe, R. J. (2013). Response-to-intervention (RTI) as a model to facilitate inclusion for students with
learning and behaviour problems. European Journal of Special Needs Education, 28(3), 254–269. doi:10.1080/
08856257.2013.768452
Gustafson, S., Svensson, I., & Fälth, L. (2014). Response to intervention and dynamic assessment: Implementing
systematic, dynamic and individualised interventions in primary school. International Journal of Disability,
Development and Education, 61(1), 27–43. doi:10.1080/1034912X.2014.878538
Hart, B., & Risley, T. R. (1995). Meaningful differences in the everyday experience of young American children.
Baltimore, MD: Paul Brookes.
Hinton, V., Flores, M. M., & Shippen, M. (2013). Response to intervention and math instruction. International Journal
of Education in Mathematics, Science and Technology, 1(3), 190–201.
Jednoróg, K., Altarelli, I., Monzalvo, K., Fluss, J., Dubois, J., Billard, C., … Ramus, F. (2012). The influence of
socioeconomic status on children’s brain structure. PLoS One, 7(8), 1–9. doi:10.1371/journal.pone.0042486
Jitendra, A. K., & Kammeenui, E. J. (1993). Dynamic assessment as a compensatory assessment approach:
A description and analysis. Remedial and Special Education, 14(5), 6–18. doi:10.1177/074193259301400503
Jordan, N. C., Glutting, J., Dyson, N., Hassinger–Das, B., & Irwin, C. (2012). Building kindergartners’ number sense:
A randomized controlled study. Journal of Educational Psychology, 104(3), 647–660. doi:10.1037/a0029018
Jordan, N. C., Glutting, J., & Ramineni, C. (2008). A number sense assessment tool for identifying children at risk for
mathematical difficulties. In A. Dowker (Ed.), Mathematical difficulties: Psychology and intervention (pp. 45–58).
San Diego, CA: Academic Press. doi:10.1016/B978-012373629-1.50005-8
Jordan, N. C., Glutting, J., & Ramineni, C. (2010). The importance of number sense to mathematics achievement in
first and third grades. Learning & Individual Differences, 20(2), 82–88. doi:10.1016/j.lindif.2009.07.004
Jordan, N. C., Kaplan, D., & Hanich, L. B. (2002). Achievement growth in children with learning difficulties in
mathematics: Findings of two-year longitudinal study. Journal of Educational Psychology, 94(3), 586–597.
doi:10.1037/0022-0663.94.3.586
Jordan, N. C., Kaplan, D., Oláh, L. N., & Locuniak, M. N. (2006). Number sense growth in kindergarten:
A longitudinal investigation of children at risk for mathematics difficulties. Child Development, 77(1), 153–175.
doi:10.1111/j.1467-8624.2006.00862.x
Joshi, R. M., Dahlgren, M., & Boulware–Gooden, R. (2002). Teaching reading in an inner city school through
a multisensory teaching approach. Annals of Dyslexia, 52(1), 229–242. doi:10.1007/s11881-002-0014-9
Kim, S. Y., Mun, E. Y., & Smith, S. (2014). Using mixture models with known class membership to address incomplete
covariance structures in multiple-group growth models. British Journal of Mathematical and Statistical Psychology,
67(1), 94–116. doi:10.1111/bmsp.12008
Kline, R. B. (1998). Principles and practice of structural modeling (1st ed.). New York, NY: Guilford Press.
Locuniak, M. N., & Jordan, N. C. (2008). Using kindergarten number sense to predict calculation fluency in second
grade. Journal of Learning Disabilities, 41(5), 451–459. doi:10.1177/0022219408321126
EARLY EDUCATION & DEVELOPMENT 19

Marsh, H. W., Hau, K.-T., & Wen, Z. (2004). In search of golden rules: Comment on hypothesis-testing approaches to
setting cutoff values for fit indexes and dangers in overgeneralizing Hu and Bentler’s (1999) findings. Structural
Equation Modeling, 11(3), 320–341. doi:10.1207/s15328007sem1103_2
McCarthy, D. (1972). Manual for the McCarthy scales of children’s abilities. New York, NY: The Psychological
Corporation.
Meintjes, E. M., Jacobson, S. W., Molteno, C. D., Gatenby, J. C., Warton, C., Cannistraci, C. J., … Jacobson, J. L.
(2010). An fMRI study of magnitude comparison and exact addition in children. Magnetic Resonance Imaging, 28
(3), 351–362. doi:10.1016/j.mri.2009.11.010
Mendizábal, E. A., Villagrán, M. A., Guzmán, J. I. N., & Howell, R. (2017). Improving number sense in kindergarten
children with low achievement in mathematics. Anales De Psicología, 33(2), 311–318. doi:10.6018/
analesps.33.2.239391
MOE. (2012). Education and development guidance for 3- to 6-year-old children. Beijing, PRC: The Ministry of
Education of the People’s Republic of China Publishing.
Morgan, P. L., Farkas, G., & Wu, Q. (2011). Kindergarten children’s growth trajectories in reading and mathematics:
Who falls increasingly behind? Journal of Learning Disabilities, 44(5), 472–488. doi:10.1177/0022219411414010
Noble, K. G., Mccandliss, B. D., & Farah, M. J. (2010). Socioeconomic gradients predict individual differences in
neurocognitive abilities. Developmental Science, 10(4), 464–480. doi:10.1111/j.1467-7687.2007.00600.x
OECD. (2003). The PISA 2003 technical report: Scaling procedures and construct validation of context questionnaire
data.Paris, France: OECD Publishing.
OECD. (2009). The PISA 2009 technical report: Scaling procedures and construct validation of context questionnaire
data. Paris, France: OECD Publishing.
Park, J., Bermudez, V., Roberts, R. C., & Brannon, E. M. (2016). Non-symbolic approximate arithmetic training
improves math performance in preschoolers. Journal of Experimental Child Psychology, 152, 278–293. doi:10.1016/j.
jecp.2016.07.011
Piaget, J., & Inhelder, B. (1969). The psychology of the child. New York, NY: Basic Books.
Purpura, D. J., & Napoli, A. R. (2015). Early numeracy and literacy: Untangling the relation between specific
components. Mathematical Thinking and Learning, 17(2–3), 197–218. doi:10.1080/10986065.2015.1016817
Rivera, S. M., Reiss, A. L., Eckert, M. A., & Menon, V. (2005). Developmental changes in mental arithmetic: Evidence
for increased functional specialization in the left inferior parietal cortex. Cerebral Cortex, 15(11), 1779–1790.
doi:10.1093/cercor/bhi055
Serpell, R., Baker, L., & Sonnenschein, S. (2005). Becoming literate in the city: The Baltimore early childhood project.
Cambridge, UK: Cambridge University Press.
Sonnenschein, S., & Galindo, C. (2015). Race/ethnicity and early mathematics skills: Relations between home, class-
room, and mathematics achievement. The Journal of Educational Research, 108(4), 261–277. doi:10.1080/
00220671.2014.880394
Szkudlarek, E., & Brannon, E. M. (2018). Approximate arithmetic training improves informal math performance in
low achieving preschoolers. Frontiers in Psychology, 9. doi:10.3389/fpsyg.2018.00606
Szűcs, D., Devine, A., Soltesz, F., Nobes, A., & Gabriel, F. (2014). Cognitive components of a mathematical processing
network in 9-year-old children. Developmental Science, 17(4), 506–524. doi:10.1111/desc.12144
Van, H. J., Costa, H. M., & Passolunghi, M. C. (2017). Improving approximate number sense abilities in preschoolers:
PLUS games. School Psychology Quarterly, 32(4), 497–508. doi:10.1037/spq0000191
Van Luit, J. E. H., & Schopman, E. A. M. (2000). Improving early numeracy of young children with special education
needs. Remedial and Special Education, 21(1), 27–40. doi:10.1177/074193250002100105
Van Vleet, B. L., & Thompson, M. S. (2011). An investigation of power analysis approaches for latent growth modeling
(Doctoral dissertation). Arizona State University. Retrieved from https://repository.asu.edu/items/9312
Vaughn, S., & Fuchs, L. S. (2003). Redefining learning disabilities as inadequate response to instruction: The promise
and potential problems. Learning Disabilities Research & Practice, 18(3), 137–146. doi:10.1111/1540-5826.00070
Vygotsky, L. S. (1978). Mind in society. Cambridge, MA: Harvard University Press.
White, K. R. (1982). The relation between socioeconomic status and academic achievement. Psychology Bulletin, 91(1),
461–481. doi:10.1037/0033-2909.91.3.461
Willoughby, M. T., Blair, C. B., Wirth, R. J., & Greenberg, M. (2012). The measurement of executive function at age 5:
Psychometric properties and relationship to academic achievement. Psychological Assessment, 24(1), 226–239.
doi:10.1037/a0025361
Zhang, L. J., Bi, Y., Liang, Y., & Liu, M. H. (2016). Early screening and dynamic intervention of children with
insufficient number sense. Acta Psychologica Sinica, 48(7), 804–817. doi:10.3724/SP.J.1041.2016.00804
Zhang, X. (2016). Linking language, visual-spatial, and executive function skills to number competence in very young
Chinese children. Early Childhood Research Quarterly, 36, 178–189. doi:10.1016/j.ecresq.2015.12.010
Zhang, X., & Lin, D. (2015). Pathways to arithmetic: The role of visual-spatial and language skills in written arithmetic,
arithmetic word problems, and nonsymbolic arithmetic. Contemporary Educational Psychology, 41, 188–197.
doi:10.1016/j.cedpsych.2015.01.005

You might also like