Educ Stud Math

DOI 10.1007/s10649-011-9360-z

The effects of using drawings in developing young

children’s mathematical word problem solving: A design
experiment with third-grade Hungarian students

Csaba Csíkos & Judit Szitányi & Rita Kelemen

# Springer Science+Business Media B.V. 2011

Abstract The present study aims to investigate the effects of a design experiment
developed for third-grade students in the field of mathematics word problems. The main
focus of the program was developing students’ knowledge about word problem solving
strategies with an emphasis on the role of visual representations in mathematical modeling.
The experiment involved five experimental and six control classes (N=106 and 138,
respectively) of third-grade students. The experiment comprised 20 lessons with 73 word
problems, providing a systematic overview of the basic word problem types. Teachers of
the experimental classes received a booklet containing lesson plans and overhead
transparencies with different types of visual representations attached to the word problems.
Students themselves were invited to make drawings for each task, and group work and
teacher-led discussion shaped their beliefs about the role of visual representations in word
problem solving. The effect sizes of the experiment were calculated from the results of two
tests: an arithmetic skill and a word problem test, and the unbiased estimates for Cohen’s d
proved to be 0.20 and 0.62. There were significant changes also in experimental group
students’ beliefs about mathematics. The experiment pointed to the possibility, feasibility,
and importance of learning about visual representations in mathematical word problem
solving as early as in grade 3 (around age 9–10).

Keywords Word problem . Visual representation . Design experiment . Elementary school

mathematics

C. Csíkos (*) : R. Kelemen

Institute of Education, University of Szeged, Szeged, Hungary
e-mail: [email protected]

J. Szitányi
Faculty of Elementary and Nursery School Teachers’ Training, Eötvös Loránd University, Budapest,
Hungary
 C. Csíkos et al.

1 Introduction

1.1 Intervention programs focusing on mathematical word problems

In the last decades, a number of educational intervention programs have been conducted in
the field of mathematical word problems. These programs can be grouped according to
several factors: students’ age, the extent to which instructional methods and tasks differed
from those of control groups’, and whether the characteristics of word problems served as
dependent or independent variables of the experiment. In the following, we present a
review of the literature concerning research into the teaching of word problems in order to
frame our study and warrant our chosen methods.
The importance of word problems in research on mathematics education originates in several
factors. As Lave (1992) pointed out, the importance of word problems lies not in the
mathematical structure they reflect but in activities of the school system, and—consequently—
in learning theories concerning those situated actions. Another important feature of word
problem research is the potential to comply with the norms of analytic scientific reasoning,
i.e., the potential to change and alternate characteristics of word problems as independent
variables of a laboratory-based or a classroom experiment (see Verschaffel, Greer & De Corte,
2000). Since general problem-solving skills are of central importance in school mathematics
achievement (see, e.g., English & Halford, 1995), a third justification might be given by the
fact that research on arithmetic word problems has shifted towards a general problem-solving
perspective (Verschaffel, Greer & Torbeyns, 2006).
The role of word problem solving is highlighted in the Hungarian National Core Curriculum
(Nemzeti alaptanterv, 2007) in that “the individual is able to apply basic mathematical
principles and processes in acquiring knowledge and in solving problems in daily life, at
home and at the workplace” (p. 9). The objectives for word problem solving (i.e., observing
situations elaborated in word problems, separating relevant and irrelevant information etc.) are
declared for school grades from 1 to 8, without detailing the steps or levels and without
attaching them to grades or age groups. (It is the task of the local curricula to elaborate
objectives for a given school grade.) In sum, the importance of word problems and problem
solving in general is well recognized and emphasized in the National Core Curriculum, and
what actually happens in classrooms depends on mainly the local curricula, on textbooks and
on teachers’ own view of the objectives (implemented curriculum).
As a consequence of these justifications, word problems play an important role in
educational experiments and surveys. Solving word problems—at least when paper–pencil
tests are used in an experiment—requires an adequate (or at least fledgling) level of reading
skills, therefore the earliest possible school grade for intervention programs built around
word problems should be grade 3. For those teachers and researchers, who may find it too
early for third-grade students to encounter word problem-based interventions either because
of the inappropriate level of reading skills or because of children struggling with arithmetic
skills, we would like to underline that children around age 9, even when considered as low-
achievers in mathematics “are capable of building sophisticated mathematical concepts”
(English, 1996, p. 108.). Consequently, they should have the chance to meet problems that
permit the use of various solution strategies. Indeed, there are experiments with third-grade
students in the field of word problems.
Sáenz-Ludlow and Walgamuth’s (1998) 1-year long experimental program addressed
third-grade children’s use and understanding of equality and the equality symbol. In their
experiment, the idea of children’s writing of their own word problems helped children
overcome the difficulty of understanding contextual meaning of word problems. Making
Drawings in developing mathematical word problem solving

sense of contextual meaning may be considered as one source of difficulties children

encounter when solving word problems.
In Selter’s (1998) teaching experiment, the number line and the heterogeneity of solution
strategies play a central role. It was a qualitative study that presented a case study sharing
components of an experiment conducted with 28 children, illustrating the uniqueness and
complexity of that study.
Several intervention programs with students from higher grades have characteristics that
can be transferred to elementary school programs. Beyond similarities in research design,
content variables of other studies can be transferred to possible experiments with third-
grade students. One such rationale for research into word problems is eliminating students’
“superficial coping strategies” (Verschaffel & De Corte, 1997a). By upper elementary
school, students possess a variety of such strategies as searching for figures in the text,
using keywords to decide which operation would be used or simply looking at the numbers
that might indicate which operation is necessary to compute.
Over the last 15 years, a number of intervention programs have been successfully conducted
with both upper elementary and secondary school students. Significantly, with respect to
warranting the research reported below, experiments with fifth graders (English, 1997;
Verschaffel & De Corte, 1997b; Verschaffel et al., 1999), interventions with seventh-grade
students (Mevarech & Kramarski, 1997; Kramarski, Mevarech & Arami, 2002; Kramarski,
Mevarech & Lieberman, 2001) all have emphasized realistic (or authentic) tasks, and higher
level components of problem solving. In the particular case of Verschaffel et al. (1999) study,
three major pillars of the learning environment were changed: (1) the use of carefully
designed realistic word problems, (2) the use of powerful instructional techniques, and (3)
creation of a stimulating classroom culture. A successful adaptation of the Flemish
intervention program was realized in Hungary among fourth-grade students (Csíkos, 2005).
Beside changes in the learning environment, an explicit formulation of learning goals
different from “traditional” aims was revealed in these experiments: the aim of developing
students’ metacognitive strategies and shaping their beliefs about mathematical problem
solving. There is plenty of evidence to support the existence of metacognitive strategies
during mathematics problem solving as early at the age of 10 (Verschaffel, De Corte &
Lasure, 1994), making the development of intervention programs for third-grade students
both feasible and desirable.
Further support of the idea that elementary school children may profit from learning by
constructing drawings came from a study with fourth- and sixth-grade students. Van Meter,
Aleksic, Schwartz and Garner’s (2006) exploratory study concerning the role of student-
generated drawings as a learning strategy can be considered important from the viewpoint
of mathematics as well. Their study focused on expository texts (i.e., texts that belong to a
given content area and are dense in information), nevertheless, the study was built around
the so-called “generative theory of drawing construction”. This theory focuses on the dual
nature of verbal and nonverbal information, and a recursive process is described, i.e.,
drawing may require text reinspection supported by self-monitoring processes.

1.2 The role visualized models in mathematics problem solving

The term “visual imagery” was defined by Presmeg (1986) as a mental scheme that depicts
visual or spatial information either existing with the presence of the object being visualized
or without the presence of that object. According to Jonassen (2003, p. 269), “Successful
problem solving requires the comprehension of relevant textual information, the capacity to
visualize the data.” Since mathematical concepts and relations are often based on visual
 C. Csíkos et al.

mental representations attached to verbal information, the ability to generate, retain and
manipulate abstract images is obviously important in mathematical problem solving. Goldin
and Kaput (1996) analyzed the structure of internal mathematical representations, and
found that the imagistic system (nonverbal configurations of objects, relations and
transformations, including visual imagery and spatial representation) receives much less
attention from educators than other systems of mathematical representations.
It is clear from Goldin and Kaput’s (1996) argumentation that the ability to visualize data
(and their relations) in a mathematical problem may contribute to mathematics problem
solving. Indeed, Hegarty and Kozhevnikov (1999, p. 688) have found that “some visual-
spatial representations promote problem-solving success” among sixth grade students (there
were only boys participating in their experiment). Furthermore, it was revealed that it was
possible to teach students how to produce appropriate visual representations. That is,
representations focused on relevant data and relations, and not irrelevant iconic or pictorial
representations. Following the verbalizer–visualizer dichotomy, Kozhevnikov, Hegarty and
Mayer (2002, p. 47.) designate people who “rely primarily on imagery processes when
attempting to perform cognitive tasks” as visualizers. Visualizers belong to two groups:
spatial and iconic type visualizers. Kozhevnikov, Hegarty and Mayer bring evidence about
the relevance of making distinctions between two types of images, and therefore, between
two types of visualizers. One aspect is the neuropsychological and neuro-imaging
differences, secondly the working memory literature is relevant, helping to distinguish
between two types of images Presmeg (1986) also differentiated between what she called
“pattern images” and “concrete images” and the former were more productive in solving
mathematical problems. Furthermore, especially importantly from an educational point of
view, Hegarty and Kozhevnikov (1999) found that it was possible to differentiate reliably
between schematic and pictorial representations made by sixth grade children. It was also
found that “the use of schematic spatial representations was associated with success in
mathematics problem solving” (Kozhevnikov, Hegarty & Mayer, 2002, p. 51).
The idea that schematic representations are meaningful for word problem solving in
terms of containing relevant data and relations calls forth the question whether using
explanatory drawings generated by mathematics teachers and students in classroom
situations is a powerful method of visualization that will improve students’ problem
solving ability. This broad hypothesis has antecedents in the literature of many frameworks
including the field of worked examples and the cognitive load theory.
In Mwangi and Sweller’s (1998) experiment, third-grade children were presented with a
worked solution that contained a schematic visualization of the objects and quantities from
the text of a word problem. The word problems were of consistent language, and students
of both the experimental and control groups solved almost all tasks; it was the number of
incorrect attempts that made a difference between the two groups, favoring the
experimental group. The authors drew two important consequences. First, the results may
be interpreted within the cognitive load framework, i.e., worked examples reduced working
memory load. They point to the importance of coherence of the source materials children
receive. Secondly, the possibility of an intervention using worked examples with schematic
drawings as early as in the elementary school years has been demonstrated.
Elementary school children’s capability to match types of schematic diagrams with a
mathematics problem has been further evidenced by Diezmann (2005). In her study of
third- and fifth-grade students’ judgments on types of diagrams, she found that even 3rd
graders were able to match diagram types with word problems in proportions unlikely to be
due to chance. As she stated in her conclusion: “students need to be able to select the
appropriate diagram for a particular problem and adequately justify their selection” (p. 286).
Drawings in developing mathematical word problem solving

In order to justify the importance of schematic drawings, more relevant features of

mathematics word problems must be taken into account. Awidely recognized source of incorrect
solutions is the use of direct translation strategy (see Hegarty, Mayer & Monk, 1995) often
caused by inconsistent language, e.g., when “greater than” is matched with subtraction.
Successful problem solvers often tend to stop and return during the phase of reading the text of
a word problem (Mayer & Hegarty, 1996). Unsuccessful problem solvers, on the contrary,
use a straightforward (direct) solution strategy like matching the hastily read key word
(e.g., “less than” with an arithmetical operation (subtraction in this case)). Jonassen
(2003) aimed at providing theoretical analysis of “story” word problems that prevent
students from using the direct translation strategy. For this prevention, a classroom
environment is needed that “requires that students first identify the class of problem …
before they are allowed to transfer them [the components of the problem] to a quantitative
representation” (p. 292). He emphasizes the role of worked examples in light of the
cognitive load theory, similarly to the approach expressed in Mwangi and Sweller’s study.
With respect to how students’ visual representations can foster word problem solving
competence, Van Meter and Garner (2005) reviewed the literature on student-generated
drawings that were made to support learning goals. They express their hope that “the
potential of drawing as a strategy to improve learning and problem solving will be realized”
(p. 322). It has been revealed that adding diagrams to word problems per se does not foster
understanding in a straightforward way: success of word problem solving may depend on a
number of factors like the types of the problems and on different student characteristics like
visuospatial abilities (Pantziara, Gagatsis & Elia, 2009).
What types of drawings can improve the understanding of mathematics word problems?
Berends and van Lieshout (2009) used a system with four categories: (1) bare picture (e.g.,
symbols), (2) useless, (3) helpful, and (4) essential illustrations. The latter contain
numerical information that is essential part of the word problem. The most important
distinction between the useless and helpful drawings is that useless illustrations contain
irrelevant visual information, whereas helpful ones represent numerical information from
the problem text. In another taxonomy, Gagatsis and Elia (2004) coined the terms
“decorative” and “informational” for naming the second and fourth types in Berends and
van Lieshout’s system.
Since schematic drawings seem to play an important role in mathematics understanding, it is
relevant to ask whether third-grade students possess the capabilities necessary to generate and
evaluate schematic drawings for simple arithmetic problems. One archetypal branch of
schematic drawings is the number line. If students at the age of 9 are already capable of
appropriately representing the number line, any results suggesting that third-grade students
have relatively accurate mental number lines will support the claim for early intervention
programs using student-generated drawings for better understanding word problems.
Children’s developing visual representations of natural numbers and their relations
(especially comparison) have been examined in investigations concerning the number line
and its mental representations. In grade 3, the 1–100 part of the number line is of great
importance both from curricular and metal representation reasons. According to Opfer and
Siegler (2007), there is a shift in mental number line representation from kindergarten to
second grade, from a logarithmic to a linear scale, and from second grade to sixth grade
estimating the locations of numbers between 1 and 1,000 develops. Siegler and Opfer
(2003) revealed that by second grade, children have both types of representations of the
number line available. Further evidence was brought to the development of accuracy in
mental visual representations of the 1–100 part of the number line by Schneider et al.
(2008). In their experiment, children from first to third grade age added two-digit numbers
 C. Csíkos et al.

where the sum was not bigger than 100. It has been shown that eye movements reflect
children’s developing knowledge about natural numbers and their spatial representations. In
that experiment, an increasing accuracy of number line estimation was revealed from eye-
tracking data, with larger differences between first and second grade, and slighter
differences between second and third grade.
The usefulness of providing appropriately scaled number lines for a special type of word
problems (one-step additive problems) in an educational setting has been documented by
Elia, Gagatsis and Demetriou (2007). It has been revealed that achievement on this number
line-added type of word problems relevantly increases at the age of 7–8.

1.3 Teachers’ knowledge about and approach for visual representations in mathematics
problem solving

Teachers’ knowledge about word problems and the role of visual representation has been
emphasized in various investigations. On the one hand, preservice elementary teachers tend
to use word problem strategies that may lead to failures in more difficult problems (Van
Dooren, Verschaffel & Onghena 2003). On the other hand, as revealed by Chapman (2006),
it is the elementary school teachers (comparing to secondary school teachers) who tend to
use the so-called narrative mode of conceptualization (in a Brunerian framework), and
“create a classroom environment that was motivational for students to learn word
problems” (p. 225). In Chapman’s study, the narrative mode of conceptualization was
opposed by the paradigmatic view. The latter refers to a cognitive functioning mode where
the central question is how to know the truth. The narrative or humanistic mode centers on
the meaning of experience. Whatever teaching style or cognitive functioning strategy is
observable among teachers, it cannot be overemphasized that teachers and students may
have strikingly different interpretations about a seemingly straightforward word problem.
Cai and Lester (2005) found remarkable differences in how Chinese and American
teachers helped students to solve word problems. Chinese teachers preferred visual
representations over verbal ones in their facilitation of students’ understanding of word
problems. These visual representations were not a required part of the solutions themselves.
Cai and Lester give a typical example of how a Chinese teacher used a number line-like
diagram to illustrate quantitative relations. On the contrary, US teachers usually used
symbolic rather than visual representations in their classroom practice.
In another comparative study, Uesaka, Manalo and Ichikawa (2007) found that 13–15-
year-old Japanese students—as compared to their New Zealand peers—had lower scores on
the questionnaire item asking whether they are explicitly told or encouraged by their
teachers to use diagrams. One possible reason can be that students tend to consider
drawings as illustrations provided by the teacher in the process of problem solving, and
“[Japanese] teachers need to spend more time in class teaching students how to actually use
diagrams” (p. 333).

1.4 Characteristics of the intervention program

The characteristics of the current investigation can be summarized as follows:

– The most important characteristic of our intervention program is the use of visual
representations in word problem solving. This use has several facets: (1) shaping
students’ belief about the importance of making drawings when solving word
problems, (2) showing students drawings that illustrate possible ways of representing
Drawings in developing mathematical word problem solving

or modeling a problem, and (3) reassuring students in making drawings that befit their
mental images and their understanding of the word problem text. In most of the
lessons, both a schematic- and a pictorial-type drawing was shown by the teacher or
generated by the students.
– The program has a maximum possible level of ecological validity, i.e., the system of
tasks and visual transparencies will be usable in the future without requiring extra time
and human resources, and without much change in curricular content.
– The program lasts for 20 units, which is about 5 weeks duration (at a rate of four
lessons/week), and the program is similar in length to the usual duration of time
dedicated to word problems in grade 3. The length is restricted by the aim of keeping
the intervention intact and uniform in a relatively short time, avoiding the effects of
possible changes in classroom and out-of-school characteristics that might interfere
with experimental variables.
– Since, at least from a historical point of view (see Verschaffel, Greer & De Corte,
2000) mathematics word problems can be either the means for practicing arithmetic
skills or the means for modeling reality, it seems to be necessary to make decisions
about what the word problems of our program can be used for. We strove to select
word problems with familiar and realistic content, trying to avoid even the
appearance of mere drilling practice. However, we agree with Palm (2008, p. 55)
in that “to practice real life task solving, then dismissing important conditions of real
life … is not an efficient practice”. Consequently, we selected realistic tasks as
defined by Cooper and Harries (2002, p. 5.), i.e., “where the textual representation of
the problem contains either persons or non-mathematical objects from ‘everyday’
settings such as shopping or sports”, but we avoided selecting so-called ‘more
authentic’ tasks as defined by Palm (2008).
– The intervention program focuses on several types of word problems in a
systematically structured way: number of arithmetic operations to be computed (if
operation should be computed at all), and whether the text has consistent or
inconsistent language. This intention about covering a wide range of word problem
types can be characterized as “building gradually” from simpler to more complex word
problems. However, this feature of the program is not unique when compared with the
textbooks from which the control group students learn, and their teachers plan their
lessons

1.5 Aims and hypotheses

A number of hypotheses underpinned the research design. Firstly, that the

experimental group would demonstrate (1) better results on word problems, (2) equal
or better results on the arithmetic skill test, (3) changes in their beliefs concerning the
importance of making drawings when solving mathematics word problems, gaining
self-confidence about their mathematical knowledge, and possessing more positive
attitude towards mathematics lessons. (4) It was also hypothesized that no significant
sex differences would emerge on the word problem test. Although there may be
differences between boys and girls in certain fields of mathematical knowledge
(Geary, 1999) or in certain populations (Halpern & LaMay, 2000), the effect size of
students’ sex will be lower than participating in the intervention program. (5) Finally, it is
assumed that experimental group students’ drawing skills would have no effect on their
achievement in the intervention program.
 C. Csíkos et al.

2 Methods

2.1 Subjects

The students involved were recruited from six schools located close to each other in
Budapest, Hungary. Eleven classes participated, five experimental and six control classes.
The experimental classes were randomly chosen from the pool of 11 classes. In each case,
the whole class was designated as either a control or an experimental group. All participants
were third-grade students whose mean age was 9 years in March, 2008. The experimental
classes comprised 106 students (53 boys and 53 girls), and the control classes 138 students
(63 boys and 75 girls).
According to recent legislation, there must be no relevant differences among schools
within the same town concerning the rate of low social-economic status students in a
school. Therefore, classes participating in the investigation either as experimental or control
classes can be considered as representative groups of the capitol’s inner city schools. From
nationwide and international system-level surveys (see, e.g., Báthory, 2003), it can be
inferred that this student population has above-average characteristics in academic
achievement.

2.2 Pre- and post-tests of the experiment

A pre-test–post-test–control (PPC) design was adopted. We used two tests as both pre- and
post-test, consequently the PPC design was realized. The first test was an arithmetic skill
test comprised 32 items covering National Core Curriculum aims. The second test
comprised six word problems, along with some questions concerning students’ attitudes
and learning aims.
Each student’s score on the arithmetic skill test was calculated as the number of correctly
solved items. The test contained 32 dichotomous items. The Cronbach-α reliability
coefficient of the pre-test was 0.84 (N=228). Using this arithmetic skill test as post-test, the
reliability coefficient proved to be 0.83 (N=237).
Scoring of the word problem test was carried out according to the scoring system of
Verschaffel, De Corte and Lasure (1994), i.e., the score of the test was the number of
realistic reactions. Realistic reactions may have different forms, namely, mathematically
correct solutions taking realistic considerations into account, or explicit notions of the task
being unsolvable or otherwise problematic. The reliability of this six item pre-test was 0.72
(N=230). Using the word problem test as post-test, the reliability proved to be 0.72 again
(N=232)
In order to control whether experimental group students’ drawing skill influenced how
they could profit from the experiment, the first task of the Clark Drawing Test (CDT) was
administered as a pre-test (Clark, 1989). According to the Hungarian standardization results
of CDT (Kárpáti, 2001), we decided to use only the first task of CDT (drawing an
interesting house as seen from the other side of the street). Since the test has a very strong
inner consistency as measured by the Cronbach-α coefficient (0.97), a shortened version
may save time without losing much diagnostic value about the overall level of drawing
skill. This subtest can be considered as a neutral measure of drawing skill, i.e., independent
of any mathematical content. Another modification was skipping the item of giving a title
to the picture because there are only 0.2–0.3 correlations between the quality of giving title
and other items of the task. Thus, the maximum score was 60 points, deriving from12 five-
point items. The reliability coefficient proved to be 0.88 (N=100).
Drawings in developing mathematical word problem solving

2.3 The intervention program

Brown’s (1992) seminal study emphasized the need for conducting educational
experiments where the complexity and systemic nature of learning can be studied in
view of interwoven variables. It has been clear in the previous decades that it is almost
impossible to conduct single-factor experiments where only one independent variable
is tested and the other factors are kept constant. According to Bell (2004), there is a
wide range of different types of design-based experiments. Taking examples from the
two furthest points, there are multivariate, multifactorial designs and there are
ethnographic observations, and both can be considered as design experiments. Our
current research is much closer to the multifactorial experimental design approach, but
since there are several possible interwoven independent variables, we regard it as
design experiment.
The intervention program in our current investigation has several characteristics
(see below), but none of these variables can be isolated from the others, so the
intervention program is considered as a complex intervention package. This
complexity enables (1) practical feasibility in dissemination (Brown, 1992), therefore
assuring ecological validity (Gravemeijer & Cobb, 2006), and (2) possible contribution to
theories of classroom learning. The current design experiment program created a learning
environment that provided ample opportunity for children to encounter various types of
word problems and various types of drawings that might help them to solve those
problems.

2.3.1 Characteristics of the intervention program

The program had several interwoven characteristics. The basic idea of the intervention was
built around the role of visual representations in word problem solving. For this aim, we
first provided students with a systematic review of the types of word problems. The aspects
of this review were: (1) whether the word problem can be solved by executing arithmetic
computations. This aspect may be important to eliminate possible misconceptions or beliefs
that lead to frequent searching for the appropriate operation. (2) The number of operations
to be computed for solving a word problem is usually one in grade 3 but at the end of the
intervention program, two-step problems are applied. (3) Since Mayer and Hegarty’s (1996)
work, the term “consistent language” has become widely accepted by the discourse
community. This term indicates that the keywords in the text of a word problem correspond
with the arithmetic operation to be computed. The structure of word problem types in the
intervention is shown in Table 1.
The program contained 73 word problems altogether. Each task was presented to each
student on a separate A4 sheet one at a time during lessons. Children had no booklet,
because we wanted to avoid their prereviewing any of the tasks.
A booklet was prepared for teachers of the experimental classes. This booklet contained
a short theoretical introduction about the nature of the experiment and its hypotheses and,
for each lesson, the aim of that lesson, the type of tasks to be solved, the tasks themselves,
instruction about the methods, and the use of supplementary materials.
Since one focus of the program was helping students become aware of the role and
importance of making drawings related to mathematics word problems, teachers were
invited to encourage students to make drawings even in the case of the simplest—or
seemingly simple—tasks. Teachers were invited by us to initiate conversations during the
task solving process allowing students to gradually become aware of the existence of
 C. Csíkos et al.

Table 1 Types of word problems in the intervention program

Lesson Type of word problems

Can be solved by arithmetic One definite solution Can be solved by one Consistent
computation arithmetic operation language

1 No Yes
2 No No
3 No Yes
4 No No
5 Yes Yes Yes
6 Yes Yes Yes
7 Yes Yes No
8 Yes Yes No
9 Yes Yes Yes/no
10 Yes Yes Yes/no
11 Explicit analysis of drawings for arithmetic word problems
12 No No
13 Yes Yes Yes
14 No No
15 Yes Yes No
16 Yes Yes Yes
17 Yes Yes Yes/no
18 Yes Yes Yes/no
19 Yes No Yes
20 Yes No No

different types of drawings, and albeit there might be individual difference in preference,
usually it is the schematic drawings that are useful to sketch.
Another focus of the program was of instructional methodological nature. Changing the
usual classroom practice had two main dimensions. First, teachers were asked to use the
think-aloud technique in some of the tasks in order to demonstrate the possible branches
and circles in the solution process. Second, in some of the lessons students worked in
heterogeneous groups consisting of five to six students, learning to tolerate and discuss
different solution plans.
We prepared colored transparencies for the majority of the tasks. (Overhead
projector transparencies meant the common available tool for the experimental classes
from an educational infrastructural aspect.) The aim of using transparencies was to
standardize visual aid students may receive when copying, comparing or analyzing
drawings created for a word problem. For example, word problem #16 in the sixth
lesson had the following text: this morning, three uniform bunches of carnations were
ordered. The gardener picked 21 carnations. How many carnations did each bunch
contain? After individual seat work, students compared their drawings to three
different types of drawings presented on the preprepared transparencies by the teacher
(see Fig. 1). The drawings on the transparencies were planned by the first and third
author of this study, and were prepared by a graphic artist. The lines and the use of colors
imitated many characteristics of children’s drawings except for being purposeful about
Drawings in developing mathematical word problem solving

Fig. 1 Illustrations provided for word problem #16 of the sixth lesson unit

the mathematical–structural content. The lesson plans gave detailed instruction when and
how to use the transparencies.

2.3.2 Teacher training and consultation, implementation of the intervention program

Teachers of the experimental classes were gathered for an oral discussion led by the first
and second authors of this study. During this talk, we presented the aims of the experiment,
our hypotheses, and the lesson plans were handed out. Later, during the weeks of the
 C. Csíkos et al.

experiment, teachers’ questions were received and answered via emails and telephone
conversations.
The pre-test phase involved administering three tests in the experimental group in three
consecutive lessons, and administering two tests in the control group. After the pre-tests,
the experimental group received 20 lessons, at a rate of four per week. (Four mathematics
lessons per week is a general characteristic in the third grade in Hungary.)
The lessons were taught by the regular classroom teachers who, as indicated above, were
familiar with the aims of the experiment. The intervention lessons did not require the full
45 min of a mathematics lesson; the rest of the time could be freely used to make progress
with the regular material and targets.
Teachers of the control groups received no information about the aims and
hypotheses of the experiment, but they were informed about the importance of their
contribution as control class teachers of an experiment. In the participating schools,
during the time interval of the intervention program, the regular classroom material
consisted of solving and practicing on word problems. These spring months, when the
experiment took place are usually, country-wide, the season for allocating much time
for drilling practice on word problems, i.e., solving dozens or hundreds of word
problems; many of them can be labeled as routine and pseudo-realistic (a term
borrowed from Boaler, 1994). The main difference between the experimental and control
class practices—from the viewpoint of control group teachers—was allocating much
more time for drilling practice on routine word problems. The post-tests were
administered immediately after the intervention program.

3 Results

Only those students who completed both the arithmetic skill and word problem tests, as
both pre- and post-tests, are taken into account. Since test administration took place in the
students’ classroom, a 5% rate of attrition, typically due to illness, seems reasonable and
understandable. As we have seen, the reliability coefficients could be computed from 228 to
230 students’ scores, and there were 216 students (97 from experimental and 119 from
control classes) who completed all four tests.

3.1 Descriptive statistics and comparisons

Table 2 shows the basic statistics about the pre- and post-test results of the experimental and
control groups. The data in Table 2 suggest that while there were differences between the

Table 2 Means and standard deviations of the experimental and control groups’ test achievement and post-
test—pre-test differences (PPD) on the pre- and post-tests

Experimental group (N=97) Control group (N=119)

pre-test post-test PPD pre-test post-test PPD

Arithmetic skill test Mean 25.31 27.00 1.69 26.74 27.56 1.56
SD 4.61 4.35 3.04 4.25 4.00 1.43
Word problem test Mean 2.57 4.12 0.82 3.51 4.06 0.55
SD 1.68 1.68 3.13 1.56 1.67 1.25
Drawings in developing mathematical word problem solving

experimental and control groups on the pre-test favoring the latter group, there were notably
fewer differences on the post-tests results.
Table 3 shows the t test comparisons between experimental and control group pre-test
achievement. The differences in pre-test achievement were significant in both tests. Table 4
shows similar data for the post-tests.
There were no significant differences between the experimental and control groups in
either of the post-tests. The paired-samples t test provides information about the
development of students’ achievement during the 6-week period of the intervention
program. On the arithmetic skill test, both the experimental and control groups significantly
outperformed their pre-test achievement in the post-test. Statistical values are t(118)=2.87,
p<0.01 for the control group and t(96)=5.48, p<0.001 for the experimental. Similarly, in
the case of the word problem test, t(118)=4.78, p<0.001 for the control group and t(96)=
10.73, p<0.001 for the experimental group indicate significant changes during the 6-week
intervention period.

3.2 Effect sizes

The PPC design of this experiment calls forth the use of the modified Cohen’s d effect size
(Morris, 2005). Since it has been proven that both the experimental and control groups had
significantly higher means at the end of the intervention period in both tests than they had at
the beginning, it is this effect size that may indicate how much bigger development rate
took place in the experimental group.
In case of the PPC design, Cohen’s d index is the standardized mean change for the
experimental and control groups, and is estimated by the following equation (see
Morris, 2005):
ðMpost;exp Mpre;exp ÞðMpost;control Mpre;control Þ
$¼ SDpre;pooled ; where
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðnexp 1ÞSD2pre;exp þðncontrol 1ÞSD2pre;control
SDpre;pooled ¼ nexp þncontrol 2

The final d index will contain a c constant for unbiased estimation: d=c Δ, where
3
c¼1
4ðnexp þ ncontrol  2Þ  1
The unbiased effect size d for the word problem test was 0.62, and the unbiased d for the
arithmetic test is 0.20. According to Cohen (1969), d=0.8 can be considered as large effect
size, d=0.5 is medium, and d=0.2 is small effect size. Following this, the intervention
program had a small effect on the arithmetic skills, and a notably perceptible (between
medium and large) effect size on the word problem test.

Table 3 T tests for comparisons between experimental and control group achievement on the pre-tests, and
on each of the six tasks of the word problem test

Levene’s statistic Two-sample t test

F p |t| p

Arithmetic skill test 1.24 0.27 2.37 0.02

Word problem test 0.81 0.37 4.29 <0.001
 C. Csíkos et al.

Table 4 T tests for comparisons between experimental and control group achievement on the post-tests, and
on the word problem test

Levene’s statistic Two-sample t-test or Welch test

F p |t| p

Arithmetic skill test 0.52 0.47 0.99 0.32

Word problem test 0.40 0.53 0.28 0.78

3.3 Connections with background variables

Beyond the two mathematics-related achievement tests, students from the experimental
group completed the shortened Clark Drawing test, and both the experimental and control
groups received additional questions about their attitudes towards and beliefs about
mathematics. For each background variable, the first aspect of analysis will be whether that
variable has a significant impact on the post-test results of the experiment.
The first task of the Clark Drawing test can be considered as a manifest variable of
drawing skills. Since the effect of the intervention may partly lean on students’ drawing
skills, we computed correlation coefficients between CDT total score and the mathematics
related post-tests of the experiment. Table 5 shows the Pearson correlation coefficients.
None of the values are significant.
The CDT scores were used to divide the experimental group into two equal subsamples
in number. Using a threshold score of 35 allows this; and eta-squared effect size can be
computed to determine the proportion of the variance on the post test scores that can be
attributed to drawing skills. For the arithmetic skill test, eta-squared proved to be 0.003 and
for the word problem test, eta-squared was 0.018 or 1.8%. This value belongs to the “small”
effect size category according to Cohen (1969).
Since there was no significant difference between the experimental and control groups
but their pre-test achievement differed significantly; in the following analyses, the
increment in achievement will serve as a measure of achievement gain. In this way, we
can analyze the multiple analysis of variance (ANOVA) effects of different background
variables.
To compare the effect of the intervention program and students’ sex, and their
possible interaction, a 2 (experiment or control)×2 (boy or girl) analysis of variance
was conducted, first on the increment in word problem test scores. The eta-squared
index of the experimental effect (explained variance) was 0.13 (p<0.001) for the
experimental condition, and the gender effect was not significant: eta-squared=0.001 (p=
0.62). Neither was the interaction of the two variables significant: eta-squared=0.01 (p=
0.09). As for the increment in the arithmetic skill test scores, the same tendencies could
be revealed. Eta squared and p values (in parentheses) for the experimental conditions, for
gender effect, and for interaction were 0.02 (0.04), 0.001 (0.60), and 0.001 (0.66),
respectively.

Table 5 Pearson correlation

coefficients between the drawing Correlation p
skill test score and the post-tests
of the experiment (N=94) Arithmetic skill test −0.04 0.73
Word problems test −0.16 0.13
Drawings in developing mathematical word problem solving

There were five questions about the beliefs and attitudes concerning mathematics
learning. Out of the five questions, three were involved in both the pre- and post-tests.
Table 6 shows the main data about these background variables.
Two-sample t tests show significant differences (p<0.05) in the first question (how are
you getting on during mathematics lessons?) for both the pre- and post-test sessions. Paired-
samples t tests shows that in both the experimental and control groups the judgment about
the difficulty of mathematics changed significantly (p<0.05). In case of the experimental
group, the mean change was 0.16 which is almost twice as much as the 0.09 change in the
control group.
There were two questions that appeared only in the post-test session. In case of the question
“To what extent has your mathematics knowledge developed during the last 2 months?” there
was no significant difference between the experiment and control groups: t(210)=0.11, p=0.91.
As for the question “Do you agree that having drawn a good drawing about a word problem,
you will find the solution easier?” there was significant difference between the experimental
and control groups: t(209)=2.18, p=0.03. These results on the last two background variables
suggest that although awareness about the development of mathematics knowledge did not
differ by the end of the intervention, an important declarative metacomponent of
mathematical knowledge had developed by the end of the intervention program.

4 Discussion and conclusion

The first hypothesis of the experiment stated that the experimental group would have better
results on the word problem post-test. Our results suggest that the intervention program
proved to be successful in terms of significant development in the experimental group
comparing to the control group. The control group had a significant advantage at the
beginning of the intervention period in both the arithmetic skill and the word problem test.
After the 6-week long intervention period, there were no significant differences on these
tests between the two groups, therefore the success of the experiment lies in the difference
of the experimental group gains. Consequently, the volume of gains in achievement was
assessed by computing the experimental effect size.
The second hypothesis stated that there would be equal or better results achieved on the
arithmetic skill test in the experimental group than in the control group. Similarly to the
results on the word problem test, there was significant difference on the pre-test, and there
was no significant difference on the post-test.

Table 6 Descriptive statistics for three background variables (1=highest or most positive option, 2=middle
or neutral, 3=most negative)

Pretest Post-test

Exp. Control Exp. Control

How are you getting on during Mathematics lessons? Mean 1.65 1.43 1.61 1.41
SD 0.70 0.59 0.69 0.63
How difficult do you find Mathematics? Mean 1.85 1.87 1.69 1.78
SD 0.51 0.45 0.51 0.46
How important is it for you to master Mathematics Mean 1.07 1.02 1.09 1.05
SD 0.30 0.13 0.36 0.26
 C. Csíkos et al.

To quantify the relative magnitude of achievement change, experimental effect sizes

were estimated by means of the unbiased PPC design-modified version of Cohen’s d
statistics. The results suggest that the intervention program had a notable (above-medium
level) effect on students’ word problem solving, and a weak but positive effect on students’
arithmetic skills.
The third hypothesis concerned students’ math-related beliefs. The analysis revealed
that (1) the experimental group was significantly more inclined to accept that making
an appropriate drawing for a word problem will make it easier to solve that problem,
and (2) there was no significant difference in their perceptions of their own
mathematical development during the 2-month intervention period. These results
indicate that the intervention program was successful in that students of the
experimental group were more aware of the importance of making drawings for
mathematics word problems.
The fourth hypothesis concerned gender differences. According to the {gender}×
{experimental condition} two-way ANOVA results, students’ sex was not a significant
variable in the experiment, and the lack of interaction with the experimental condition
shows that girls and boys could equally benefit from the intervention program. In similar
vein (fifth hypothesis), the intervention program was equally successful for students with
higher or lower level drawing skills as shown by the {drawing skill}×{experimental
condition} two-way ANOVA results.
The results of our intervention program may allow the drawing of some inferences. The
program aimed at substituting the tasks and lesson plans of 20 consecutive mathematics
lessons with our tasks and lesson plans. The structure of the intervention program aimed at
reviewing several types of word problems, while helping students to become aware of the
importance of making drawings in mathematics problem solving. It was important for us to
develop an ecologically valid intervention program, i.e., provide lesson plans and
supplementary materials that can be easily embedded in current lesson plans and textbooks.
Our program was successful in fostering the development of not only word problem solving
but the development of arithmetic skills.
There is a debate among in-service teachers about the role of word problems in
mathematics teaching and learning. First, word problems can be a “dressed up”
version of simple mathematics operations, therefore the means for drilling practice in
order to master arithmetic skills. The second viewpoint stresses the importance of
modeling realistic mathematical situations by means of using word problems in the
classroom. Our results may strengthen the second viewpoint, since by using a
carefully selected pool of word problems with realistic content and relations; it is still
possible to help students develop their arithmetic skills, without solving too many
similar tasks. By saying farewell to the drilling practice in the world of word
problems, we may offer new resources for efficient classroom methods that help
students become aware of their mental processes and of the importance of using
appropriate visualization methods in solving word problems.

Acknowledgments This research was supported by a grant from the Hungarian Scientific Research Fund
(OTKA 63360) awarded to Csaba Csíkos and by the MTA-SZTE Research Group on the Development of
Competencies. Thanks are due to Gabriella Pataky for her help in the use of the Clark Drawing Test. We
would also like to thank Paul Andrews, Andrea Kárpáti, Julianna Szendrei, Malcolm Swan and Lieven
Verschaffel for their constructive comments on an earlier draft of this paper. Parts of the present study have
been presented at a conference (Csíkos, Szitányi & Kelemen, 2009). Also, part of the data have been
published in Hungarian in Magyar Pedagógia (Csíkos, Szitányi & Kelemen, 2010).
Drawings in developing mathematical word problem solving

References

Báthory, Z. (2003). Rendszerszintű pedagógiai felmérések. [System-level educational surveys]. Iskolakul-

túra, 13(8), 3–19.
Bell, P. (2004). On the theoretical breadth of design-based research in education. Educational Psychologist,
39, 243–253.
Berends, I. E., & van Lieshout, E. C. D. M. (2009). The effect of illustrations in arithmetic problem-solving:
Effects of increased cognitive load. Learning and Instruction, 19, 345–353.
Boaler, J. (1994). When do girls prefer football to fashion? A analysis of female underachievement in relation
to ‘realistic’ mathematics context. British Educational Research Journal, 20, 551–564.
Brown, A. L. (1992). Design experiments: Theoretical and methodological challenges in creating complex
interventions in classroom settings. The Journal of the Learning Sciences, 2, 141–178.
Cai, J., & Lester, F. K. (2005). Solution representations and pedagogical representations in Chinese and U.S.
classrooms. Journal of Mathematical Behavior, 24, 221–237.
Chapman, O. (2006). Classroom practices for context of mathematics word problems. Educational Studies in
Mathematics, 62, 211–230.
Clark, G. (1989). Screeining and identifying students talented in the visual arts. Clark’s Drawing Abilities
test. Gifted Child Quaterly, 33, 98–106.
Cohen, J. (1969). Statistical power analysis for the behavioral sciences. New York: Academic.
Cooper, B., & Harries, T. (2002). Children’s responses to contrasting ‘realistic’ mathematics problems: Just
how realistic are children ready to be? Educational Studies in Mathematics, 49, 1–23.
Csíkos, C. (2005). A metacogniton-based training program in grade 4 in the fields of mathematics and
reading. Paper presented at the 11th European Conference for Research on Learning and Instruction,
Nicosia, Cyprus, 23–27, August, 2005.
Csíkos, C., Szitányi, J., & Kelemen, R. (2009). Promoting 3rd grade children’s mathematical problem
solving through learning about the role of visual representations. Paper presented at the 13th
European Conference for Research on Learning and Instruction held in Amsterdam, The Netherlands,
August 25—August 29.
Csíkos, C., Szitányi, J., & Kelemen, R. (2010). Vizuális reprezentációk szerepe a matematikai problém-
amegoldásban. Egy 3. osztályos tanulók körében végzett fejlesztő kísérlet eredményei. [The role of
visual representations in mathematical problem solving: Results of an intervention program with 3rd
grade students]. Magyar Pedagógia, 110, 149–166.
Diezmann, C. M. (2005). Primary students’ knowledge of the properties of spatially-oriented
diagrams. In H. L. Chick & J. L. Vincent (Eds.), Proceedings of the International Group for the
Psychology of Mathematics Education (pp. 281–288). Melbourne: PME.
Elia, I., Gagatsis, A., & Demetriou, A. (2007). The effects of different mode of representation on the solution
of one-step additive problems. Learning and Instruction, 17, 658–672.
English, L. D. (1996). Children’ construction of mathematical knowledge in solving novel isomorphic
problems in concrete and written form. Journal of Mathematical Behavior, 15, 81–112.
English, L. D. (1997). The development of fifth-grade children’s problem solving abilities. Educational
Studies in Mathematics, 34, 183–217.
English, L. D., & Halford, G. S. (1995). Mathematics education: Models and processes. Mahwah, NJ:
Erlbaum.
Gagatsis, A., & Elia, I. (2004). The effects of different modes of representations on mathematical problem
solving. In M. Johnsen-Hoines & A. Berit-Fuglestad (Eds.), Proceedings of the 28th Conference of the
International Group for the Psychology of Mathematics Education: Vol. 2 (pp. 447–454). Bergen,
Norway: Bergen University College.
Geary, D. C. (1999). Sex differences in mathematical abilities: Commentary on the math-fact retrieval
hypothesis. Contemporary Educational Psychology, 24, 267–274.
Goldin, G. A., & Kaput, J. (1996). A joint perspective on the idea of representation in learning and doing
mathematics. In L. Steffe, P. Nesher, P. Cobb, G. Goldin, & B. Greer (Eds.), Theories of mathematical
learning (pp. 397–430). Hillsdale: Erlbaum.
Gravemeijer, K., & Cobb, P. (2006). Design research from a learning design perspective. In J. van den Akker,
K. Gravemeijer, S. McKenney, & N. Nieveen (Eds.), Educational design research (pp. 17–51). London:
Routledge.
Halpern, D. F., & LaMay, M. L. (2000). The smarter sex: A critical review of sex differences in intelligence.
Educational Psychology Review, 12, 229–246.
Hegarty, M., & Kozhevnikov, M. (1999). Types of visual-spatial representations and mathematical problem
solving. Journal of Educational Psychology, 91, 684–689.
 C. Csíkos et al.

Hegarty, M., Mayer, R. E., & Monk, C. A. (1995). Comprehension of arithmetic word problems: A
comparison of successful and unsuccessful problem solvers. Journal of Educational Psychology,
87, 18–32.
Jonassen, D. H. (2003). Designing research-based instruction for story problems. Educational Psychology
Review, 15, 267–296.
Kárpáti, A. (2001). Firkák, formák, figurák. A vizuális nyelv fejlődése a kisgyermekkortól a serdülőkorig.
[Scribbles shapes and figures. The visual language of children]. Budapest Hungary: Dialóg Campus
Kiadó.
Kozhevnikov, M., Hegarty, M., & Mayer, R. E. (2002). Revising the visualizer–verbalizer dimension:
Evidence for two types of visualizers. Cognition and Instruction, 20, 47–77.
Kramarski, B., Mevarech, Z. R., & Arami, M. (2002). The effects of metacognitive instruction on solving
mathematical authentic tasks. Educational Studies in Mathematics, 49, 225–250.
Kramarski, B., Mevarech, Z. R., & Lieberman, A. (2001). Effects of multilevel versus unilevel metacognitive
training on mathematical reasoning. The Journal of Educational Research, 94, 292–300.
Lave, J. (1992). Word problems: A microcosm of theories of learning. In P. Light & G. Butterworth (Eds.),
Context and cognition. Ways of learning and knowing (pp. 74–92). Hillsdale, NJ: Erlbaum.
Mayer, R. E., & Hegarty, M. (1996). The process of understanding mathematical problems. In R. J. Sternberg
& T. Ben-Zeev (Eds.), The nature of mathematical thinking (pp. 29–53). Mahwah, NJ: Erlbaum.
Mevarech, Z. R., & Kramarski, B. (1997). IMPROVE: A multidimensional method for teaching mathematics
in heterogeneous classrooms. American Educational Research Journal, 2, 365–394.
Morris, S. B. (2005). Effect size estimation from prestest–posttest–control designs with heterogeneous
variances. Paper presented at the 20th Annual Conference of the Society for Industrial and
Organizational Psychology, Los Angeles, CA.
Mwangi, W., & Sweller, J. (1998). Learning to solve compare word problems: The effect of example format
and generating self-explanations. Cognition and Instruction, 16, 173–199.
Nemzeti Alaptanterv (2007) [National Core Curriculum] Available from: http://www.okm.gov.hu/letolt/
kozokt/nat_070926.pdf.
Opfer, J. E., & Siegler, R. S. (2007). Representational change and children’s numerical estimation. Cognitive
Psychology, 55, 169–195.
Palm, T. (2008). Impact of authenticity on sense making in word problem solving. Educational Studies in
Mathematics, 67, 37–58.
Pantziara, M., Gagatsis, A., & Elia, I. (2009). Using diagrams as tools for the solution of non-routine
mathematical problems. Educational Studies in Mathematics, 72, 39–60.
Presmeg, N. (1986). Visualisation in high school mathematics. For the Learning of Mathematics, 6, 42–46.
Sáenz-Ludlow, A., & Walgamuth, C. (1998). Third-graders interpretations of equality and the equal symbol.
Educational Studies in Mathematics, 35, 153–187.
Schneider, M., Heine, A., Thaler, V., Tornbeyns, J., De Smedt, B., Verschaffel, L., et al. (2008). A validation
of eye movements as a measure of elementary school children’s developing number sense. Cognitive
Development, 23, 409–422.
Selter, C. (1998). Building on children’s mathematics—a teaching experiment in grade three. Educational
Studies in Mathematics, 36, 1–27.
Siegler, R. S., & Opfer, J. E. (2003). The development of numerical estimation: Evidence for multiple
representations of numerical quantity. Psychological Science, 14, 237–243.
Uesaka, Y., Manalo, E., & Ichikawa, S. (2007). What kinds of perceptions and daily learning
behaviors promote students’ use of diagrams in mathematics problem solving? Learning and
Instruction, 17, 322–335.
Van Dooren, W., Verschaffel, L., & Onghena, P. (2003). Pre-service teachers’ preferred strategies for solving
arithmetic and algebra word problems. Journal of Mathematics Teacher Education, 6, 27–52.
Van Meter, P., Aleksic, M., Schwartz, A., & Garner, J. (2006). Learner-generated drawing as a strategy for
learning from content area text. Contemporary Edcuational Pschology, 31, 142–166.
Van Meter, P., & Garner, J. (2005). The promise and practice of learner-generated drawing: Literature review
and synthesis. Educational Psychology Review, 17, 285–325.
Verschaffel, L., & De Corte, E. (1997a). Word problems: A vehicle for promoting authentic mathematical
understanding and problem solving in the primary school? In T. Nunes & P. Bryant (Eds.), Learning and
teaching mathematics: An international perspective (pp. 69–97). Hove, UK: Psychology Press.
Verschaffel, L., & De Corte, E. (1997b). Teaching mathematical modeling and problem-solving in the
elementary school. A teaching experiment with fifth graders. Journal for Research in Mathematics
Education, 28, 577–601.
Verschaffel, L., De Corte, E., & Lasure, S. (1994). Realistic considerations in mathematical modelling of
school arithmetic word problems. Learning and Instruction, 7, 339–359.
Drawings in developing mathematical word problem solving

Verschaffel, L., De Corte, E., Lasure, S., Van Vaerenbergh, G., Bogaerts, H., & Ratinckx, E. (1999). Design
and evaluation of a learning environment for mathematical modeling and problem solving in upper
elementary school children. Mathematical Thinking and Learning, 1, 195–229.
Verschaffel, L., Greer, B., & De Corte, E. (2000). Making sense of word problems. Lisse, The Netherlands:
Swets & Zeitlinger.
Verschaffel, L., Greer, B. & Torbeyns, J. (2006). Numerical thinking. In Boero, P. & Gutiérrez, A.
(Eds.), Handbook of research on the psychology of mathematics education: Past, present and future
(pp. 51–82). Rotterdam, the Netherlands: Sense Publishers.