ZDM

https://doi.org/10.1007/s11858-018-1002-9

ORIGINAL ARTICLE

The effect of a conceptual model-based approach on ‘additive’ word

problem solving of elementary students struggling in mathematics
Yan Ping Xin1

Accepted: 22 September 2018

© FIZ Karlsruhe 2018

Abstract
Whole number arithmetic is the foundation of higher mathematics and a core part of elementary mathematics. Awareness
of pattern and underlying problem structure promote the learning of whole number arithmetic. A growing consensus has
emerged on the necessity to provide students with the opportunity to engage in algebraic reasoning earlier in their educa-
tion. In fact, the U.S. National Council of Teachers of Mathematics curriculum standards and the nationwide mathematics
reform call for algebra readiness in elementary mathematical learning. The purpose of this study was to evaluate the effect
of Conceptual Model-based Problem Solving (Xin, Conceptual model-based problem solving: Teach students with learning
difficulties to solve math problems. The Netherlands: Sense Publishers, 2012) intervention program, with the Singapore bar
model method (Kaur, The model method—A tool for representing and visualising relationships, In: Conference proceed-
ings of ICMI study 23: Primary mathematics study on whole numbers, pp 448–455, 2015) serving as the bridge towards
symbolic mathematical model equations, on enhancing additive word problem-solving performance of elementary students
who are struggling in mathematics. Findings from this study indicate the promise of this program. Through representing a
range of additive word problem situations in one cohesive mathematical model equation, students are prepared for general-
ized problem-solving skills.

1 Introduction Core State Standards for Mathematics emphasize concep-

While mathematics problem-solving skills are critical for
successful functioning in the twenty-first Century, many
students who are struggling in mathematics fail to acquire
these skills during their early school studies. According to
recent United States National Assessment of Educational
Progress data (NAEP 2015), about 60% of American 4th
graders performed below proficiency level in mathematics.
In fact, about 5–10% of school age children are identified
as having mathematics disabilities (Fuchs et al. 2007) and
students whose mathematics performance ranked at or below
the 35th percentile are often considered at risk for learning
disabilities or for having learning difficulties in mathematics
(Bryant et al. 2011).
At the same time, expectations for all students, includ-
ing those with learning difficulties, have been elevated in
today’s educational climate. In particular, the US Common
* Yan Ping Xin
yxin@purdue.edu
1 operation. For instance, students were taught that words
Purdue University, West Lafayette, IN, USA
tual understanding in problem solving, mathematical mod-
eling, algebra readiness, higher order thinking and reasoning
(Common Core State Standards Initiative 2012). All students
are now expected to understand and solve complicated word
problems and develop solid reasoning skills in mathematics.
There is a need to implement evidence-based practice to
address this new emphasis, in order to facilitate all students’
access to higher-order thinking and meeting the challenging
curriculum standards.
2 Theoretical background
Whole number arithmetic is the foundation of higher math-
ematics and a core part of elementary mathematics. As
for teaching elementary arithmetic word problem solving,
one of the distinctive features of the traditional instruction,
often observed in US elementary classrooms, is its focus on
the choice of operation. To determine the choice of opera-
tion, it is not uncommon to see that students rely on the
‘keyword’ strategy for making a decision on the choice of

13
Vol.:(0123456789)
 Y. P. Xin
such as ‘altogether’ in the word problem cue an operation
of addition; words such as ‘take away’ indicate an operation
of subtraction; and words such as ‘time(s)’ signal an opera-
tion of multiplication, etc. The keyword strategy, which has
been the practice in the US for generations (Sowder 1988),
directs students’ attention toward isolated ‘cue’ words in the
problem. The keyword strategy might be a ‘quick and dirty’
way to ‘fix’ word problem solving; however, it is at odds
with contemporary approaches to word problem solving that
stress conceptual understanding of mathematical relations
in a problem before attempting to solve it with an operation
(Jonassen 2003). In particular, the keyword strategy does not
orient students’ attention to a problem’s underlying math-
ematical structure or relations and encourage mathematical
modeling that is emphasized by contemporary curriculum
standards. Further, applying the keyword strategy might con-
tribute to students being prone to reversal operation errors
when encountering the so-called ‘inconsistent language’
problems (e.g., Tara solved 21 problems. She solved three
times as many problems as Pat. How many problems did Pat
solve?), where students might mistakenly multiply, when
they need to divide, for solution due to the key word ‘times’
(Xin 2007; Xin et al. 2011a).
Other strategies commonly used in teaching word prob-
lem solving include ‘draw a picture’, with which students
engage in drawing pictures based on the story told in the
word problem. Students will draw a real picture of the
objects or use cubes to represent the actual objects described
in the word problem, and then they will use counting as
the primary strategy for solution. Another strategy is called
‘guess and check’. Greer (1992) portrayed such a strategy
as: “look at the numbers; they will tell you which opera-
tions to use. Try all the operations and choose the most rea-
sonable answer” (p. 28). It should be noted that when the
numbers in the problem are small, it might be manageable
to solve the problem correctly using the ‘guess and check’
or ‘draw a picture’ strategies. However, when the numbers
become large, such problem-solving processes may become
cumbersome or inefficient. Nonetheless, none of the above
strategies focuses on an analysis of underlying mathematical
structure or relations in the word problem and representing
such relations in mathematical models at the symbolic level
for solution.
2.1 Schema‑based instruction and arithmetic word
problem solving
Hord and Xin (2013) conducted a review of intervention
research for helping elementary school students with learn-
ing difficulties understand and solve mathematics word prob-
lems. One of the emerging intervention strategies during
the past decade is schema-based instruction. According to
Hord and Xin (2013), schema-based instruction emphasizes
13
semantic analysis of word problems and mapping of the
problems into schematic diagrams (Marshall 1995) that are
specific to different problem types (e.g., change, group, and
compare). The semantic analysis and categorization of word
problems employ the framework of Cognitively Guided
Instruction (CGI, Carpenter et al. 1999). In schema-based
instruction, students need to identify the problem types
based on the definition of each type (e.g., change, combine
or group, compare or difference, etc.). Then students repre-
sent the problem in the corresponding schematic diagram
(see Table 2, second column, for schema diagram represen-
tations). After that, students need to make a decision on the
choice of operation for solution. The findings from existing
intervention research studies indicate a considerable treat-
ment effect on researcher-designed word problem-solving
measures. The intervention, however, did not result in a sig-
nificant generalization effect on commercial tests of problem
solving (Hord and Xin 2013).
Recently, schema-based instruction has been included in
computer-assisted instruction. Using a randomized control
trial design, Fede et al. (2013) studied the effect of a com-
puter tutor program titled “GO solve word problems” (Sny-
der 2005) on word problem-solving performance of low per-
forming fifth grade students. The GO solve word problems
tutoring program used worked examples to teach students
to solve word problems using graphic organizers (similar
to schema diagrams). The control group received “state test
preparation review” in regular scheduled mathematics peri-
ods. After a 12-week intervention, results indicated that the
group receiving the computer tutor program showed better
gains on test items from state high stakes testing and better
growth on researcher designed bi-weekly tests. However,
there was no statistically significant difference between the
computer tutor program and the control group on a stand-
ardized test: the Process and Application Subtest from the
Group Mathematics Assessment and Diagnostic Evaluation
(Williams 2004).
While research studies have shown the promise of
schema-based instruction in helping students with math-
ematical learning difficulties to solve word problems, the
emphasis of schema-based instruction is on decoding prob-
lems into subtypes (e.g., combine, change, difference) based
on various story situations. Cohesive mathematical mode-
ling is not initiated to make explicit the connections between
problem subtypes (e.g., combine, change, and compare prob-
lem types). Following the schema diagram representation,
students need to generate the mathematical sentence or equa-
tion for solution based on the rules they need to remember.
For instance, students were taught rules such as “if the total
is not given—Add”; “if the total is given—Subtract!” How-
ever, rules such as these do not lend themselves to a precise
mathematical equation for accurate problem solving.
The effect of a conceptual model-based approach on ‘additive’ word problem solving of elementary…

Recently, Nunes et  al. (2016) questioned: why does
schema-based instruction often used surface features of
problems to classify schemas? According to Vergnaud
(1979), the schema was useful for solving for a quantity
by adding a quantity to a given quantity, or for those prob-
lems involving combination of the quantities into a single
whole, but not for problems involving missing beginning
amounts. With schema-based problem-solving strategies,
students need to make a mental leap from real-world situ-
ated problem schemata or iconic representations of real-
world situations (Bruner 1973) to symbolic mathematical
expressions/equations for solutions. In light of research
in mathematics education, many students have difficulties
in making the transition from a real situational model to a
mathematical model; it is a weak area in students’ math-
ematical understanding (Blomhøj 2004).
There is a critical need for a paradigm shift in math-
ematics instruction/intervention toward mathematical
model-based problem solving that emphasizes the over-
arching conceptualization of mathematical relations in
elementary arithmetic word problem solving. In fact,
model-based problem solving is expected, according to
the Common Core Standards for Mathematics (Common
Core State Standards Initiative 2012). Essential features
of the Common Core include mathematical modeling,
higher order thinking and reasoning, and algebra readiness
(National Governors Association Center for Best Practices
and Council of Chief State School Officers 2010).
2.2 The conceptual model‑based problem solving
(COMPS) approach
While semantic analysis, as in schema-based instruction, is
perhaps important for students to understand the story situa-
tion as the very first step in problem solving, the Conceptual
Model-based Problem Solving (COMPS) intervention pro-
gram (Xin 2012) aims to make connections across various
problem types (e.g., change, combine, compare) through
students’ learning and constructing a cohesive mathemati-
cal model. For instance, “Part + Part = Whole” is a cohesive
mathematical model that generalizes various additive word
problem situations. Table 1 presents the representation of
a range of additive word problems in the Part–Part-Whole
diagram equation.
Table 2 presents a comparison between schema-based
diagram representations (adapted from Jitendra 2002) and
COMPS model equations. As shown in Table 2, solving for
the unknown quantity ‘a’ in the COMPS model equation
gives the answer to the problem. In contrast, a schema dia-
gram representation does not necessarily specify the solution
plan.

Table 1  Additive word Part-Part-Whole (PPW) Problem Types COMPS Model Equation

problem types along with the
corresponding model equations
(adapted from Xin 2012)
Part-Part-Whole
1. Amy has 3 books. Pat has 4 books. How many books do they have all
together? (Combine)
2. Amy had 3 books. Pat gave Amy 4 more books on her birthday. How many
books does Amy have now? (Change-join)
3. Amy had some books. She gave Pat 4 books, and found that she has only 3
books left. How many books did Amy have in the beginning? (Change-
separate)
4. Amy and Pat have found that together they have a total of 7 books. Pat
says that she has 4 books. How many books does Amy have? (Combine)
5. Amy had 7 books in her collection. She gave away some books. Now she
has only 4 books left. How many books did she gave away? (Change-
separate)
6. Amy had some books. Pat gave Amy 4 more books. Now Amy has 7
books. How many books did Amy have in the beginning? (Change-join)
7. Amy had 7 books. Then she lost 4 books. How many books does Amy
have now? (Change-Separate)
8. Amy and Pat have found that together they have a total of 7 books. Amy
says that she has 3 books. How many books does Pat have? (Combine)
9. Amy had 3 books. Then Pat gave her some books. Now Amy has 7 books.
How many books did Pat give Amy? (Change-join)
Additive Compare Problem Types Additive Compare
10. Amy has 3 books. Pat has 4 more books than Amy. How many books
does Pat have? (Compare-more)
11. Amy has 3 books. She has 4 fewer books than Pat. How many books does
Pat have? (Compare-less)
12. Pat has 7 books. She has 4 more books than Amy. How many books does
Amy have? (Compare-more)
13. Pat has 7 books. Amy has 4 fewer books than Pat. How many books does
Amy have? (Compare-less)
14. Pat has 7 books. Amy has 3 books. How many more books does Pat have
than Amy? (Compare-more)
15. Pat has 7 books. Amy has 3 books. How many fewer books does Amy
have than Pat? (Compare-less)

13
 Y. P. Xin

Table 2  Schema diagrams vs.
COMPS model equations
Sample Problems Sample Schema Diagrams COMPS Model Equation
Rachel had some flowers in a big
vase. Then, 19 of the flowers
wilted, so she took those ones out.
There were 29 flowers left in the
vase. How many flowers in the
vase in the beginning?
Amy and Pat have found that
together they have a total of 27
books. Amy says that she has 18
books. How many books does
Pat have?

Kaylin has 48 candies. She has

70 fewer candies than Melody,
how many candies does Melody
have?

2.3 Summary and purpose statement
Most of the strategies adopted by traditional US mathemat-
ics textbooks seem to focus on using concrete manipulates
(such as blocks, counters) or pictures to represent the prob-
lems and to teach for conceptual understanding (Sherin
2001). Symbolic representation or the use of abstract algo-
rithms was avoided in elementary mathematics due to histor-
ical assumptions about children’s developmental readiness
(Carraher et al. 2006). A growing consensus has emerged
on the necessity to provide students with the opportunity
to engage in algebraic reasoning earlier in their education
(Carpenter et al. 2005), because introducing symbolic rep-
resentation and algebraic thinking in earlier grades facili-
tates a smoother transition from elementary to higher-level
mathematics learning. According to Mulligan and Woolcott
(2016), mathematical performance can be enhanced through
instruction that focuses on “the development of abstract gen-
eralisations rather than focusing on WNA [Whole Number
Arithmetic]” (p. 222). On the other hand, awareness of pat-
tern and structure promotes the learning of whole number
arithmetic (Mulligan and Woolcott 2016).
The COMPS approach aims to connect elementary arith-
metic to algebra learning through integrating mathematical
model-driven reasoning in arithmetic word problem solving.
Chappell and Strutchens (2001) suggested two reasons why
students fail to connect arithmetic to future learning of alge-
bra: (a) a lack of exposure to algebraic ideas and thinking,
and (b) students’ tendency to learn algebra as mere symbol
manipulation. COMPS serves to extend students’ arithmetic
experiences.
The purpose of the study was to evaluate the effect of
the COMPS intervention approach, with the Singapore bar
model method (Kaur 2015) serving as the bridge towards
symbolic mathematical model equations, on enhancing the
additive word problem-solving performance of elementary
students who are struggling in mathematics problem solv-
ing. Specific research questions were as follows: (a) did
participating students improve their additive word problem-
solving performance after receiving the COMPS interven-
tion? And (b) how did participants’ word problem-solving
performance change before and after the intervention?
3 Method
3.1 Participants and setting
Participants (pseudonyms are used in this paper) included
three third grade students from an elementary school located
in the Midwestern US. Marco (age = 10.25 years.) was diag-
nosed as having specific learning disabilities. He spent 69%
of the school hours in general education classrooms and 31%
in a learning support classroom for mathematics and reading
intervention. Ted (age = 9.92 years.) was diagnosed as hav-
ing mild intellectual disabilities. He was included in general
education classroom activities for 47% of school hours; he
spent the rest of the school hours (53%) in a learning sup-
port classroom for mathematics and reading intervention.
Marie (age = 9.75 years.) was diagnosed as having specific
learning disabilities. She was included in general education
classrooms for 74% of school hours, and 26% of the school
hours in a learning support classroom for mathematics and
reading intervention. All three participating students failed
in high-stake testing in mathematics prior to the interven-
tion. The intervention took place in the participating school’s
computer lab.
3.2 Dependent measures
3.2.1 Word problem solving (WPS) test
The WPS test includes 14 one-step addition and subtrac-
tion word problems. The construction of each word prob-
lem was systematically varied in reference to the structure
of specific problem schemata and the unknown position in

13
The effect of a conceptual model-based approach on ‘additive’ word problem solving of elementary…
a problem so that a range of additive word problems were
represented including combine, change-join, change-sepa-
rate, compare involving ‘more than…’ or ‘less than…’ sit-
uations. The construction of the items in the WPS tests is
similar to the structure of the various additive word prob-
lem types as shown in Table 1. However, we included only
14 items, rather than the 15 items as shown in Table 1,
as item #4 and item #8, the two variations of the com-
bine problem type (see Table 1), are similar in nature, and
therefore we included only one such item in the WPS test
to reduce the burden of testing. It should be noted that the
story context as well as the numbers involved in the WPS
test were different from the ‘skeleton’ items presented in
Table 1. Specifically, the items in the WPS tests were in
line with the Common Core State Standards (2012) per-
tinent to addition and subtraction. That is, second grade
students should be able to “use addition and subtraction
within 100 to solve one- and two-step word problems
involving situations of adding to, taking from, putting
together, taking apart, and comparing, with unknowns in
all positions, e.g., by using drawings and equations with
a symbol for the unknown number to represent the prob-
lem” (CCSS.Mathematics Content 2.OA.A.1). We devel-
oped multiple alternate forms of WPS tests for use during
the baseline assessment, intervention and post intervention
assessment.
3.2.2 Standardized measure
A standardized measure was developed to validate the
intervention effects on the primary dependent measure
(i.e., the WPS test). It included relevant problem solving
items taken from the KeyMath-Revised-Normative Update
(KeyMath-R/NU) Problem Solving subtest. Excluded Key-
Math items entail multiplication/division problems, num-
ber sense problems (e.g., what number comes next?), story
making based on pictures presented, and other non-routine
problems such as finding out missing information in the
problem. The internal consistency of the KeyMath-R/NU
is 0.97 (Connolly 1998).
3.2.3 Pre‑algebra model expression (Algebra) test
The pre-algebra model expression (algebra) test was
designed to test students’ pre-algebra expression of math-
ematical relations or ideas. Twelve items (e.g., “Write
an expression or equation. Choose a variable for the
unknown. Shanti had some stamps. She gave 23 to Penny.
Shanti has 71 stamps left”) were included in the algebra
test. We took these items directly from a commercially
published mathematics textbook series adopted by the par-
ticipating school (Maletsky et al. 2004).
3.2.4 Scoring
The percentage of problems solved correctly was used as
the dependent measure and calculated as the total points
earned divided by the total possible points. Specifically, if
the correct answer was given to a problem, one point would
be earned. In addition to scoring the answer correct, we also
examined whether conceptual model-based representation
in the diagram equation was performed before providing
the answer. A graduate student who was naïve to the pur-
pose of the study scored all the probes using an answer key.
The author rescored 40% of the tests. Interrater reliability
was computed by dividing the number of agreements by
the number of agreements and disagreements and multiply-
ing by 100. Interrater reliability for scoring was 99% (range
92–100) across the independent raters.
3.3 Design
An adapted multiple-probe-design (Horner and Baer 1978)
across participants was employed to evaluate the potential
functional relationship between the intervention and partici-
pants’ word problem-solving performance. Single-subject
research design was chosen because the design provides a
methodological approach well suited to the investigation of
single cases or groups (Kazdin 1982), as in the case of this
study, specifically students with learning disabilities. Unlike
uncontrolled case studies, intervention effects in multiple-
probe-design can be demonstrated by introducing the inter-
vention to different baselines/participants at different points
in time. “If each baseline changed when the intervention
is introduced, the effects can be attributed to the interven-
tion rather than to extraneous events” (Kazdin 1982, p. 126).
With single-subject design, “selection biases usually do not
present problems” or threats to internal validity “because
inferences do not depend on comparisons of different per-
sons” (Kazdin 1982, p. 80). Attrition or missing data points
is usually not a threat to internal validity in single-subject
research if the effect is not evaluated using average scores of
a group for data analysis over time (Kazdin 1982).
In contrast to group design studies, single-subject
research design focuses on individual performance, and
holds a different philosophical assumption in addressing
generality. In particular, it believes that the interrelation-
ship between the behavior of an individual and the envi-
ronment “establishes generalizable relationships in a much
more fundamental ways than group research” studies (Kra-
tochwill and Williams 1988, p. 150). A potential problem in
group research has been generalizing from the performance
of groups to individual participants. In particular, “small
13
 Y. P. Xin
effects that are found on the average may have little bearing
on the individual subject.” (Kratochwill and Williams 1988,
p. 150). External validity in single-subject design studies can
be demonstrated through the extent to which the results or
intervention effects generalized or extended to other persons,
settings, tasks or behaviors (Kazdin 1982).
The design of this study included a baseline (Baseline:
the WPS tests, KeyMath item test, and the Algebra test),
Module A instruction (problem representation and solving
of combine, change-join, change-separate problem types)
followed by two WPS tests (Probe [P]1 and P 2), Module B
instruction (problem representation and solving of additive
comparison problems that involve “more than”) followed by
another two WPS tests (P3 and P4), and Module C instruc-
tion (additive comparison problems that involve “less than”)
followed by the posttests (Posttest: the WPS tests, KeyMath
item test, and Algebra test).
3.4 Procedure
All three participants completed one WPS test during the
baseline condition. Then one student (Ted) took another
two equivalent WPS tests as well as the Algebra test and
KeyMath item test. Following the baseline, the intervention
on Module A was first introduced to Marco. Once Marco’s
performance showed an improvement, Module A interven-
tion was introduced to Ted after he took two additional WPS
tests (as well as the other two tests) immediately before the
intervention. The same sequence was followed until all three
participants were introduced to Module A intervention. Fol-
lowing Module A instruction, two probes on the WPS test
were taken to monitor participating students’ progress before
Module B instruction took place. After Module B, another
two probes were taken before Module C instruction. Post-
tests were given following all modules’ instruction.
Fig. 1  Bar model as the Rachel had 48 flowers in a big vase. Then, 19 of the flowers wilted, so she took those ones
bridging to the PPW diagram out. Then there were 29 flowers left in the vase.
equation
13
3.5 COMPS intervention primed with bar model
Participating students worked with the teacher, a pre-service
Special Education teacher one-on-one four times a week,
with each session lasting about 25–30 min. Intervention was
carried out through two phases: Problem representation and
problem solving. During the problem representation, word
problem stories with no unknowns were used to help stu-
dents understand the problem structure and the mathematical
relations between the quantities. Bar models were used to
present the quantitative relationship between the three parts,
which serve as a transition to the PPW diagram equation
model (see Fig. 1 for an example). After representing all
three quantities in the diagram equation, students were able
to check the ‘balance’ of the equation to shape and reinforce
the concept of ‘equality’ and the meaning of an equal sign.
Problem representation instruction was followed by prob-
lem solving instruction. During problem solving instruction,
word problems with an unknown quantity were presented.
Students were required to map the information from the
word problem into the diagram equation without going
through the bar model representation. Nevertheless, if
needed, the bar model could be brought back as the bridging
to help students understand the mathematical relations in the
Part–Part-Whole diagram equation. During this phase, stu-
dents were reminded to use a letter (any letter they preferred)
to represent the unknown quantity. Overall, the instruction
was carried out with explicit strategy explanation and mod-
eling, dynamic teacher-student interaction, guided practice,
performance monitoring with corrective feedback, and inde-
pendent practice.
To illustrate, the excerpt presented below showcases
potential teacher-student interaction during the problem-
solving instruction phase:
The effect of a conceptual model-based approach on ‘additive’ word problem solving of elementary…

Fig. 2  Three participants’ word problem-solving performance on the WPS test, algebra model expression test, and KeyMath item measure
before, during and after the COMPS intervention

13
 Y. P. Xin

Kaylin has 48 candies. She has 70 fewer candies than Melody.

How many candies does Melody have?

(Students read the problem together.)

Teacher: What is this problem about?

Students: This problem compares the number of candies Kaylin has to the number of candies that Melody has.

Teacher: Correct! It is a comparison problem that describes one quantity as more or less than the other quantity.

(Teacher presents the Additive Compare Word Problem Story Grammar Poster, see Fig. 3b)

Can you tell which sentence tells one quantity is more or less than the other?

Students: Kaylin has 70 fewer candies than Melody.

Teacher: That is correct. This comparison sentence tells us that the number of candies Kaylin has is less than the

number of candies Melody has. Let’s underline this comparison sentence!

Students underline the sentence in the problem.

Teacher: What is the difference between the number of candies Kaylin has and the number of candies Melody

has?

Students: Kaylin has 70 candies less… The difference between the two is “70.”

Teacher: Let’s write the difference amount 70 in the PPW diagram, the box labeled as “difference.”

Students write 70 in the box labeled as difference.

Teacher: From the comparison sentence: “Kaylin has 70 fewer candies than Melody,” can you tell me who has

more, and who has less?

Students: “Kaylin has 70 fewer ……,” so Kaylin has less, and Melody has more.

Teacher: Superb! Please help me name the box for “smaller” and name the box for “bigger” on the PPW

diagram.

Student name the box for “smaller” as Kaylin, and name the box for “bigger” as Melody. See below:

Teacher: What number will we write in the box for “Kaylin” in the PPW diagram equation?

Students: “Kaylin has 48 candies,” so we will write “48” in the box for Kaylin.

Teacher: Great! Now what number we will write in the box for “Melody”?

Students: …...

Teacher: Do we know the number of candies Melody has?

Students: No.

Teacher: Correct, that is the unknown quantity we are asked to find out.

Let’s write the letter ‘a’ to represent the unknown quantity in the box for Melody in the diagram equation.

Students write letter a in the box for Melody.

Teacher: Now we have completed the mapping of information onto the PPW diagram equation. (Teacher points

to diagram equation) “It tells that, the number of candies Kaylin has (48) PLUS the difference amount (70)

EQUALS the number of candies Melody has. That is, 48 + 70 = a”

[Upon completion of the representation part, the student will rewrite the mathematics equation (without boxes) as

“48 + 70 = a” or “a = 48 +70”, and solve for the unknown quantity a for solution (i.e., a = 118).]

13
The effect of a conceptual model-based approach on ‘additive’ word problem solving of elementary…

4 Results (median = 29%) correct for Ted, and 26% (median = 21%)

Figure 2 presents the performance of three participants
in solving word problems on the WPS test, Algebra test,
and KeyMath item test before, during (only WPS test),
and after the COMPS intervention. The results indicated
that students improved their performance not only on the
researcher-developed WPS test (involving a range of 14
additive word problems), but also on the algebra model-
expression test (taken from a commercial published text-
book). In addition, all participants improved their problem
solving performance on relevant items from KeyMath, a
standardized test measuring essential mathematical con-
cept and skills.
4.1 Baseline performance
During baseline data gathering, average performance on the
WPS tests was 36% (median = 36%) correct for Marco, 24%
correct for Maria. Across the three participants, the baseline
performance was relatively stable and low, indicating the
need and setting the stage for intervention.
4.2 Intervention effect
During the intervention phase, Marco increased his perfor-
mance from 36% correct to 75% and 83% correct following
PPW problem representation and solving instruction, and to
100% correct after additive comparison problem represen-
tation and solving instruction. Similarly, Ted increased his
performance from 24% correct to 67% correct following the
PPW problem representation/solving instruction, and then
to 75% correct following comparison problem solving that
involves “more than” story situations, and finally to 100%
after receiving instruction on solving comparison problems
that involve “less than” story situations. Finally, Maria
improved her performance from 26% correct to an average

Fig. 3  a Conceptual model of A

Part–Part-Whole word problem Part-Part-Whole (PPW)
(Xin 2012, p. 47). b Conceptual A PPW problem describes multiple parts that make up the whole
model of additive compare word ------------------------------------------------------------------------------------------
problems (Xin 2012, p. 67)

PPW Word Problem Story Grammar Questions

Which sentence (or question) tells about the “whole” or “combined” amount? Write that quantity in
the bigger box on one side of the equation by itself.

Which sentence (or question) tells about one of the parts that make up the whole? Write that quantity
in the first small box on the other side of the equation.

Which sentence (or question) tells about the other part that makes up the whole? Write that quantity
in the 2nd small box (next to the first small box).

B
Additive Compare (AC)
An AC problem describes one quantity as “more” or “less” than the other quantity
------------------------------------------------------------------------------------------

AC Word Problem Story Grammar Questions

Which sentence (or question) describes one quantity as “more” or “less” than the other? Write the
difference amount in the diagram.
Who has more, or which quantity is the bigger one?
Who has less, or which quantity is the smaller one? Name the bigger box and smaller box.
Which sentence (or question) tells about the bigger quantity? Write that quantity in the bigger box on
one side of the equation by itself.
Which sentence (or question) tells about the smaller quantity? Write that quantity in the smaller box
next to the difference amount.

13
 Y. P. Xin
of 92% correct following PPW problem solving instruction
and then 100% correct following additive compare prob-
lem representation and solving intervention. In summary,
towards the end of the intervention, all three participants
reached 100% correct. During the post intervention assess-
ment, the three participants (Marco, Ted, and Maria) kept
an average performance of 100% correct, 98% correct, and
91% correct respectively.
4.3 Effect on transfer measures
To assess potential transfer effect, participating students
were assessed on the standardized measure derived from the
KeyMath-R/NU as well as the Algebra test (pre-algebraic
model expression). As shown in Fig. 2, Marco improved
his performance on the KeyMath item measure from 25%
correct to 100% correct, Ted from 0 to 75% correct, and
Maria from 75 to 100% correct. On the Algebra test, Marco
improved his performance from 17% correct during the
baseline to 92% correct following the intervention. Ted
improved from 8 to 75% correct, and Maria improved from
8 to 100% correct.
5 Discussion and conclusions
5.1 The effect of the COMPS primed
with the bar model
Based on visual analysis of the data, across all three par-
ticipants, there is an immediate change in level of per-
formance once the intervention (Part–Part-Whole prob-
lem representation and solving) was introduced to each
of the participants following the baseline. The data path
during the intervention phase and post-intervention phase
showed an upward or elevated steady trend, which indi-
cates an enhanced performance. There is no overlapping
between the baseline data points and the treatment/post
treatment data points, which indicates a strong treatment
effect [percentage of non-overlapping data (PND) = 100%,
Scruggs and Mastropieri 1998]. Thus, it seems that there is
a functional relationship between the COMPS intervention
and participating students’ enhanced mathematics word
problem-solving performance. That is, students’ enhanced
word problem-solving performance is likely due to the
implementation of the intervention.
5.2 Theoretical and practical implications
In the beginning of instruction, it was found that the par-
ticipating students did not have the concept of the numbers
13
let alone quantitative relations. They just grabbed all the
numbers in the problem and added them together. During
the representation stage of the intervention, bar models
were used to help students understand the Part–Part-Whole
relation in the diagram equation. The instructor used the
bar model to help the students establish the concept of
composite unit (by drawing a bar with marks to separate
each individual unit) and an understanding of quantitative
relationship in the Part–Part-Whole problem structure by
putting two short bars together to form a long bar (i.e., Part
and Part makes up the Whole).
5.2.1 No more gambling on what operation to use!
Based on the field notes, the participants took a relatively
longer time to learn initially the bar model representation
and then connect it to the Part–Part-Whole (PPW) diagram
equation. The Singapore bar model method (Kaur 2015)
served as a visual prompt to demonstrate the Part–Part-
Whole quantitative relationship in the problem. Once stu-
dents understood the part–part-whole relationship, they
no longer drew the bar model. Rather they moved on to
representing the information from the word problem in the
PPW diagram equation directly, and then solving for the
unknown quantity in the equation for the solution.
Although exiting research shows that “word problem
involving comparisons are often more difficult than either
combine or change problems, it seems that with the PPW
diagram equation model, those challenging comparison
problems (in particular, those ‘inconsistent language’
problems—e.g., Emily has 5 marbles. Emily has 14 less
marbles than Jacob does. How many marbles does Jacob
have?) are no longer difficult to solve. As long as the stu-
dents understand which quantity is the bigger quantity and
map that quantity to one side of the equation by itself, the
other two quantities (i.e., the smaller quantity and the dif-
ference quantity) would be the two “Parts” and therefore
stay “side-by-side” on the other side of the equation. Once
a correct PPW model equation is in place, the operation for
solution is determined by the model equation—no more
gambling on what operation to use!
5.2.2 Students’ construction of the Part–Part‑Whole model
equation
During a total of 13 sessions of intervention, participants
engaged in (a) constructing the concept of “part and part
makes up the whole” through composing/decomposing num-
bers using bar models; (b) representing word problems in
one cohesive COMPS diagram equation, supported by the
prompting cards as needed (see Fig. 3a, b for the prompt-
ing cards); and (c) solving the problem through finding the
The effect of a conceptual model-based approach on ‘additive’ word problem solving of elementary…
unknown quantity in the equation. Through representing
a range of Part–Part-Whole problem situations (including
nine variations of combine, change-join, change-separate
problem situations as well as six variations of additive
compare problem situations involving ‘more than’ or ‘less
than’ relations) in a cohesive mathematical model (i.e.,
Part + Part = Whole), the connection among various problem
types are made explicit to participating students.
In summary, through building most fundamental concepts
pertinent to additive reasoning, the concept of composite
unit, and through representing a range of additive word
problem situations in one cohesive PPW diagram equation,
students are expected to construct the concept of ‘Part and
Part makes up the Whole.’ With the construction of the PPW
model equation, students are prepared for generalized prob-
lem-solving skills. This is reflected in participants’ improved
performance in this study. Symbolic expression of math-
ematical relations in equation as proposed in COMPS should
not be a mechanical process of translating text to equations.
Students’ mapping of information in the diagram equation
is based on their conceptual understanding of the three key
elements or quantities involved and relations among them.
That is, formations of symbolic expression or equation are
experienced as arising from an understanding of underlying
structure of the problem (Thompson 1989).
As Sherin (2001) argued, “we can strive for concep-
tual understanding while basing instruction on the use of
equations” (p. 529). In fact, “we do students a disservice
by treating conceptual understanding as separate from the
use of mathematical notations” (Sherin 2001, p. 482). The
use of symbolic expressions can involve significant under-
standing. Introducing symbolic representation and algebraic
thinking in earlier grades may facilitate a smoother transition
from elementary to higher-level mathematics learning and
improve middle and high school mathematics performance.
As Devlin (2012) has argued, mathematics is “more than
arithmetic,” generalization resulting from identification
and analysis of patterns is the key to mathematical thinking
(p. 1). Awareness of pattern and underlying problem struc-
ture promotes the learning of whole number arithmetic at a
deeper level (Mulligan and Woolcott 2016).
5.2.3 Extended evidence of the COMPS approach
To date, the COMPS strategy has been applied to solve
multiplication and division word problems. According to
existing research, both single-subject-design studies (e.g.,
Xin and Zhang 2009) and randomized control trial (RCT)
studies (e.g., Xin et al. 2011a, b) have shown promising
effects of the COMPS intervention program. In particular,
Xin et al. (2011a, b) compared the COMPS intervention pro-
gram with a business-as-usual (BAU) condition in teaching
multiplication and division word-problem solving to elemen-
tary students with learning difficulties. The results indicate
that only the COMPS group significantly improved, from
pretest to posttest, their performance on a multiplicative
word problem-solving test and a pre-algebra model expres-
sion test. The results of these studies suggest that elementary
students with learning difficulties can be expected to move
beyond concrete operations and to represent mathematical
relations algebraically in conceptual models that drive the
solution plan for accurate problem solving.
Acknowledgements  The study reported in this paper was assisted by
You Luo, a former graduate students at Purdue University. The author
would like to thank the administrators, teachers, and students at Lafay-
ette School Corporation who facilitated this study.
References
Blomhøj, M. (2004). Mathematical modelling—A theory for practice.
In B. Clarke et al. (Eds.), International perspectives on learn-
ing and teaching mathematics (pp. 145–159). Göteborg: National
Center for Mathematics Education.
Bruner, J. S. (1973). Beyond the information given: Studies in the psy-
chology of knowing. Oxford: W.W. Norton.
Bryant, D. P., Bryant, B. R., Roberts, G., Vaughn, S., Pfannenstiel, K.
H., Porterfield, J., & Gersten, R. (2011). Early numeracy interven-
tion program for first-grade students with mathematics difficulties.
Exceptional Children, 78(1), 7–23.
Carraher, D. W., Schliemann, A. D., Brizuela, B. M., & Earnest, D.
(2006). Arithmetic and algebra in early mathematics education.
Journal for Research in Mathematics Education, 37, 87–115.
Carpenter, T. P., Fennema, E., Franke, M., Levi, L., & Empson, S. B.
(1999). Children’s mathematics: Cognitively Guided Instruction.
Portsmouth, NH: Heinemann.
Carpenter, T. P., Levi, L., Franke, M., & Zeringue, J. (2005). Algebra
in the elementary school: Developing rational thinking. ZDM,
37(1), 53–59.
Chappell, M. F., & Strutchens, M. E. (2001). Creating connections:
Promoting algebraic thinking with concrete models. Mathematics
Teaching in the Middle School, 7(1), 20–25.
Common Core State Standards Initiative (2012). Common core state
standards for mathematics practice. http://www.cores​tanda​rds.
org. Accessed 15 Mar 2018
Connolly, A. J. (1998). Keymath revised/normative update. Circle
Pines: American Guidance Service.
Devlin, K. (2012). Introduction to mathematical thinking. Palo Alto,
CA: Keith Devlin.
Fede, J. j., Pierce, M. E., Matthews, W. J., & Wells, C. S. (2013). The
effects of a computer-assisted, schema-based instruction interven-
tion on word problem-solving skills of low-performing fifth grade
students. Journal of Special Education Technology, 28(1), 9–21.
Fuchs, L. S., Fuchs, D., & Hollenbeck, K. N. (2007). Expanding
responsiveness to intervention to mathematics at first and third
grade. Learning Disabilities Research and Practice, 22, 13–24.
Greer, B. (1992). Multiplication and division as models of situations. In
D. Grouws (Ed.), Handbook of research on mathematics teaching
and learning (pp. 276–295). New York: MacMillan.
Hord, C., & Xin, Y. P. (2013). Intervention research for helping ele-
mentary school students with math learning difficulties understand
13
 Y. P. Xin
and solve word problems: 1996–2010. Learning Disabilities: A
Multidisciplinary Journal, 19(1), 3–17.
Horner, R. D., & Baer, D. M. (1978). Multi-probe technique: A varia-
tion of the multiple baseline. Journal of Applied Behavior Analy-
sis, 11, 189–196.
Jitendra, A. (2002). Teaching students math problem-solving through
graphic representations. Teaching Exceptional Children, 34(4),
34–38.
Jonassen, D. H. (2003). Designing research-based instruction for story
problems. Educational Psychology Review, 15(3), 267–296.
Kaur, B. (2015). The model method—A tool for representing and visu-
alising relationships. In Sun, X., Kaur, B. & Novotna, J. (Eds.)
Conference proceedings of ICMI study 23: Primary mathemat-
ics study on whole numbers (pp. 448–455). Macau, Macao SAR:
University of Macau.
Kazdin, A. E. (1982). Single-case research designs: Methods for clini-
cal and applied settings. New York: Oxford.
Kratochwill, T. R., & Williams, B. L. (1988). Perspective on pitfalls
and hassles in single-subject research. Journal of the Association
for Persons with severe Handicaps, 13, 147–154.
Maletsky, E. M., et al. (2004). Harcourt math (Indiana edition). Chi-
cago: Harcourt.
Marshall, S. P. (1995). Schemas in problem solving. New York: Cam-
bridge University Press.
Mulligan, J., & Woolcott, G. (2016). What lies beneath? Conceptual
connectivity underlying whole number arithmetic. In Sun, X.,
Kaur, B., & Novotna, J. (Eds.) Conference proceedings of ICMI
study 23: Primary mathematics study on whole numbers (pp.
220–228). Macau, Macao SAR: University of Macau.
National Assessment of Educational Progress Result (NEAP) (2015).
http://www.natio​nsrep​ortca​rd.gov/readi​ng_math_2015/#mathe​
matic​s/acl?grade​=4. Accessed 15 Mar 2018
National Governors Association Center for Best Practices and Council
of Chief State School Officers (2010). Common core state stand-
ards for mathematics. Washington, DC: National Governors
Association Center for Best Practices and Council of Chief State
School Officers.
Nunes, T., Dorneles, B. V., Lin, P.-J., & Rathgeb-Schnierer, E. (2016).
Teaching and learning about whole numbers in primary school.
Dordrecht: Springer International Publishing.
13
Scruggs, T. E., & Mastropieri, M. A. (1998). Summarizing single sub-
ject research: Issues and applications. Behavior Modification, 22,
221–242.
Sherin, M. G. (2001). Developing a professional vision of classroom
events. In T. Wood, B. S. Nelson & J. Warfield (Eds.), Beyond
classical pedagogy: Teaching elementary school mathematics
(pp. 75–93). Hillsdale: Erlbaum.
Snyder, T. (2005). GO Solve word problems: Using graphic organ-
izers to understand and solve word problems. Position statement.
Watertown: Tom Snyder Productions.
Sowder, L. (1988). Children’s solutions of story problems. Journal of
Mathematical Behavior, 7, 227–238.
Thompson, P. W. (1989). Artificial intelligence, advanced technology,
and learning and teaching algebra. Research Issues in the Learn-
ing and Teaching of Algebra, 135–161.
Vergnaud, G. (1979). The acquisition of arithmetical concepts. Educa-
tional Studies in Mathematics, 10, 263–274.
Williams, K. T. (2004). GMADE Group mathematics assessment and
diagnostic evaluation. Assessment and diagnostic evaluation.
Technical manual. Circe Pines: AGS Publishing.
Xin, Y. P. (2007). Word-problem-solving tasks presented in textbooks
and their relation to student performance: A cross-curriculum
comparison case study. The Journal of Educational Research,
100, 347–359.
Xin, Y. P. (2012). Conceptual model-based problem solving: Teach
students with learning difficulties to solve math problems. The
Netherlands: Sense Publishers.
Xin, Y. P., Liu, J., & Zheng, X. (2011a). A cross-cultural lesson com-
parison on teaching the connection between multiplication and
division. School Science and Mathematics, 111(7), 354–367.
Xin, Y. P., & Zhang, D. (2009). Exploring a conceptual model-based
approach to teaching situated word problems. The Journal of Edu-
cational Research, 102(6), 427–441.
Xin, Y. P., Zhang, D., Park, J. Y., Tom, K., Whipple, A., & Si, L.
(2011b). A comparison of two mathematics problem-solving strat-
egies: Facilitate algebra-readiness. The Journal of Educational
Research, 104, 381–395.