See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/232829283

A Modeling Perspective on the Teaching and Learning of

Mathematical Problem Solving

Article  in  Mathematical Thinking and Learning · July 2008

DOI: 10.1080/10986060802218132

CITATIONS READS

48 1,229

3 authors, including:

Nicholas Mousoulides Constantinos Christou

University of Nicosia University of Cyprus
54 PUBLICATIONS   508 CITATIONS    154 PUBLICATIONS   1,418 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Cognition View project

mascil project View project

All content following this page was uploaded by Nicholas Mousoulides on 29 May 2014.

The user has requested enhancement of the downloaded file.

Mathematical Thinking and Learning, 10: 293–304, 2008
Copyright © Taylor & Francis Group, LLC
ISSN 1098-6065 print / 1532-7833 online
DOI: 10.1080/10986060802218132

1532-7833 Thinking and Learning

1098-6065
HMTL
Mathematical Learning, Vol. 10, No. 3, Jul 2008: pp. 0–0

SHORT REPORT

A Modeling Perspective on the Teaching and

Learning of Mathematical Problem Solving

Nicholas G. Mousoulides
Teaching And Learning
MOUSOULIDES ET AL.Mathematical Problem Solving

University of Cyprus

Constantinos Christou
University of Cyprus

Bharath Sriraman
The University of Montana

This study analyzed the processes used by students when engaged in modeling
activities and examined how students’ abilities to solve modeling problems
changed over time. Two student populations, one experimental and one control
group, participated in the study. To examine students’ modeling processes, the
experimental group participated in an intervention program consisting of a
sequence of six modeling activities. To examine students’ modeling abilities, the
experimental and control groups completed a modeling abilities test on three occa-
sions. Results showed that students’ models improved as they worked through the
sequence of problem activities and also revealed a number of factors, such as stu-
dents’ grade, experiences with modeling activities, and modeling abilities that
influenced their modeling processes. The study proposes a three-dimensional theo-
retical model for examining students’ modeling behavior, with ubsequent implica-
tions for the teaching and learning of mathematical problem solving.

Correspondence should be sent to Nicholas G. Mousoulides, University of Cyprus, P. O. Box

20537, Nicosia 1678, Cyprus. E-mail: n.mousoulides@ucy.ac.cy
294 MOUSOULIDES ET AL.

INTRODUCTION AND THEORETICAL FRAMEWORK

A number of recent research studies documented the importance of implementing

modeling activities at the elementary-school level (English, 2006; English &
Watters, 2005). This is important not only because elementary school students
are capable of working with modeling activities, but also because modeling
needs to be introduced early in the curriculum, particularly if we want to success-
fully implement modeling at all school levels (Blum & Niss, 1991; Doerr &
English, 2003).
Modeling activities differ from traditional problem solving in at least two
ways. First, in solving modeling problems students need to use and interconnect
mathematical concepts and operations (Lesh & Zawojewski, 2007). This can
result in opportunities for students to elicit their own mathematics as they work
the problems and to make sense of the realistic situations they need to mathema-
tize. Second, in modeling activities students are encouraged to create models that
are applicable to a range of similarly structured situations, and as a result, they
can generalize and extend their solutions (English, 2006; Doerr & English,
2003).
The primary focus of many research efforts in mathematical modeling has
been on designing and trialing modeling activities in teaching and applications,
without contributing to the research efforts toward improving our understanding
of mathematical modeling (Blum, 2002). Blum (2002) clarified that there is a
need to shift from focusing on practice (e.g., modeling examples) to focusing on
theory, and raised a number of critical questions for making this shift. The
present study aims to contribute to the current research on modeling, by describ-
ing and analyzing the process components of modeling and by identifying
student modeling abilities and how these abilities can be developed over time.
Modeling processes are the processes students develop and use during their
efforts to solve a real-world problem (Lesh & Doerr, 2003). These processes
include describing the problem, manipulating the problem and building a model,
connecting the mathematical model with the real problem, predicting the behav-
ior of the real problem, and verifying the solution in the context of the real prob-
lem. Student modeling abilities include structuring, mathematizing, interpreting,
solving real-world problems, and working with mathematical models. The last
component includes the ability to validate the model, to analyze it critically, to
assess the model and its results, and to communicate the model (Blum, 2002).
The purpose of this study was to propose a theoretical model for understand-
ing “students’ modeling behavior.” To this end, the aims of the study were to (a)
describe the process components of modeling by examining students’ models in a
sequence of problem-solving activities, (b) examine how students’ modeling
abilities changed over time through an intervention program, and (c) examine
other factors that might influence the construction of students’ models.
 TEACHING AND LEARNING MATHEMATICAL PROBLEM SOLVING 295

METHOD

Four hundred and three students from eight elementary and secondary schools
participated in the study. Specifically, 104 sixth graders and 90 eighth graders
from four sixth- and four eighth-grade classes participated in the experimental
group. Similarly, 93 sixth graders and 116 eighth graders from four sixth- and
four eighth-grade classes participated in the control group. Students were ran-
domly assigned to the two groups. From each participating school one class was
assigned as the experimental group and one class was assigned as the control
group.
Experimental group students participated in an intervention program consist-
ing of six modeling activities for a period of three months. Experimental group
students worked on each modeling problem approximately every two weeks. The
implementation of each modeling activity took place in four, forty-minute ses-
sions. Modeling lessons were conducted by the first author and classroom teach-
ers in all participating experimental classes. Each modeling activity entailed: (a)
a warm-up task comprising a mathematically rich “newspaper article” designed
to familiarize the students with the context of the modeling activity, (b) “readi-
ness” questions to be answered about the article, and (c) the problem to be
solved, including tables of data, figures, and graphs.
Each activity started with a whole-class discussion on the warm-up task and
readiness questions. Then, students worked in groups of three or four on the
activities. After completing their work, each group presented their model to the
rest of the class for questioning, comparisons with other models, and constructive
feedback. Students then worked in their groups to revise and refine their models.
Finally, a whole-class discussion focused on the key mathematical ideas and pro-
cesses that were developed during the modeling activity.
At the same time, control group students worked with their regular mathematics
textbooks, which did not include any modeling activities. In the mathematics text-
books, problem solving is mostly conceptualized as the activity of solving traditional
word problems. These problems usually present simplified forms of a decontextual-
ized world based situations, with the purpose of exercising a specific type of mathe-
matical learning, such as addition or subtraction or other arithmetic operations.
The development of the modeling activities designed for the purposes of the
study followed the six design principles proposed by Kelly and Lesh (2000).
According to the Model Construction Principle the solution to the activity requires
the construction of an explicit explanation or procedure. The Reality Principle
requires the activity to be meaningful to students from their different levels of
mathematical ability and general knowledge. The Self-Assessment and Model
Documentation Principles ensure the inclusion of criteria that students themselves
can identify and use to test and revise their models and to create documentation that
will reveal explicitly how they solved the problem. The Construct Share-Ability
296 MOUSOULIDES ET AL.

and Re-Usability Principles require students to produce shareable and reusable

solutions, and the Effective Prototype Principle ensures that the modeling activity
will be as simple as possible yet still mathematically significant.
Three of the developed modeling activities constituted a sequence of modeling
problems related to statistical concepts such as average, ranking, weighting, and
aggregating. The second set was related to geometrical reasoning and specifically
to the concepts of surface area and perimeter. The activities were adopted from
those used in previous research studies (e.g., Doerr & English, 2003; Lesh & Doerr,
2003). A short description of the six modeling activities is presented in Table 1.
The data for the study were collected through a variety of sources, namely: (a)
videotapes of students’ responses during whole-class discussions; (b) audiotapes
of students’ work in their groups; (c) students’ final models, their worksheets,
and final reports detailing the processes used in developing their models; and (d)
researchers’ field notes. The analysis of the data was completed in the following

TABLE 1
Description of the Six Modeling Activities

Title Problem Context Mathematical Concepts

Best Drug Students are asked to develop a procedure Statistical concepts such
Award Activity for ranking five drugs based on as average and
information about the number of minutes frequency.
each drug needs to act for 30 cases.
Where to Live Students are asked to develop a procedure Weighting and ranking.
Activity for helping people choose a city to which
to move. The selection was based on both
qualitative and quantitative information
about the number of schools, restaurants,
budget available, etc.
University Cafeteria Students are asked to select six out of nine Proportional reasoning
Activity employees who should be rehired, based (i.e., pounds per hour).
on information about the hours worked
and the money collected by the
employees.
Carpet Design Students are asked to develop a procedure Geometric reasoning (i.e.,
Activity for designing patterns of different area and perimeter of
2-dimensional (2D) shapes for a carpet. different shapes).
Car Painting Activity Students are asked to develop a procedure Geometric reasoning (area
for finding the amount of paint needed for and perimeter of 2D
a brand new car. Students are given shapes).
information about the outer and interior
dimensions of the car.
New House Activity Students are asked to design a model of a Geometric reasoning (area
new house. Students are given partial of different figures, area
information about the building plot and and volume of solids).
the different rooms in the house.
 TEACHING AND LEARNING MATHEMATICAL PROBLEM SOLVING 297

steps. First, the transcripts were reviewed several times to identify the ways in
which students interpreted the problems presented in the activities, their
approaches in using the provided data, the modeling processes demonstrated, and
the mathematization processes used. The transcripts were also analyzed to iden-
tify how students interacted in their groups and how discussions within the
groups resulted in their final models. Finally, all students’ worksheets and their
final documentation were analyzed to identify the modeling processes and the
mathematization processes used in model development. The criteria of inter-rater
reliability was met with the third author independently analyzing the transcripts
and comparing this analysis with that done by the first author.
A test1 for measuring students’ modeling abilities was also developed in this
study. The test consisted of nine tasks. Three tasks were related to decision-making
problems, three were retrieved from system analysis and design category, and three
were troubleshooting problems (Organization for Economic Co-Operation and
Development, 2003). The tasks reflecting the decision-making problems pre-
sented students with a situation requiring a decision and asked them to choose
among alternatives under a set of conditions constraining the situation. Problems
of system analysis and design category require students either to analyze a com-
plex situation in order to find out what is the cause of the problem to be solved or
to design a system that works and achieves certain goals. Trouble shooting
problems assess students’ actions when confronted with the need to specify the
conditions under which a system is running properly or when a system of a
mechanism is underperforming.
The first author administered the modeling-abilities test three times to both the
experimental and control groups. Each administration lasted 45 minutes and the
test conditions were the same for the two groups. The test was first administered to
all participants prior to the experimental group commencing the intervention pro-
gram. The test was again administered to all participants when the first part of the
intervention (after the completion of the three modeling activities) was completed.
The test was finally administered to all students when the intervention concluded.

RESULTS AND DISCUSSION

The results of the study are presented with reference to the purpose of the study:
first, we describe the modeling processes that were evident in students’ models,
and second, we examine the development of students’ modeling abilities through
the intervention program. The data used for describing the modeling processes
were collected through the experimental-group students’ models created in work-
ing the activities presented in Table 1. The modeling processes presented in the
next section were demonstrated by the majority of the students in both the sixth-
and eighth-grade experimental groups. All modeling processes were not, however,
298 MOUSOULIDES ET AL.

present in students’ models in all modeling activities. Students’ modeling abili-

ties were measured through the administration of the test to both the experimen-
tal and control groups.

Modeling Processes
Data analysis revealed that a number of modeling processes appeared in students’
efforts to solve the problems presented in the activities. The modeling processes for
each step of the modeling procedure (description, manipulation, prediction, and
verification) that have been identified in students’ work are presented in Figure 1.
The first modeling process apparent in students’ work in the activities was
understanding the main question of the problem. Understanding the core question
was not an easy process, especially for the younger students. In doing so,
students tried to perceive different and additional information about the problem
by observing the provided data and by connecting their observations with the
question of the problem. Students made these connections more easily for sim-
pler modeling problems (e.g., Best Drug Award). In more difficult problems
(e.g., University Cafeteria) these connections were not easy, nor obvious, since
students could not understand how the different number of hours worked and
money collected by each worker in different semesters and different time slots
should be used to find the best six among nine workers. Related to problem
understanding is problem specification and simplification, in which students
identify a number of conditions and assumptions of the real-world problem. The
identification of conditions and assumptions assisted students in excluding irrele-
vant or less important information and directed students into simplifying the

Communicating
Results Understanding
and Specifying
Description Simplifying Identifying
variables
and
relationships
Validating
Manipulation

Results
Verification

Real Model
World World
Problem

Building a
Evaluating Model Mathematizing
Solution
Prediction Analyzing Combining

Predicting Interpreting
Behaviour Solution

FIGURE 1 Modeling Processes in Mathematical Problem Solving.

 TEACHING AND LEARNING MATHEMATICAL PROBLEM SOLVING 299

problem. In the University Cafeteria activity, for example, students decided to

work with the different time periods (busy, steady, and slow) instead of
constructing a model based on the different semesters (see Mousoulides, 2007).
According to their documentation, “being productive in both busy and slow peri-
ods is more important than being productive in autumn and spring semester. We
therefore, decided to work with time periods.” This helped the students to
exclude the unnecessary information and to simplify the problem presented in the
University Cafeteria modeling activity.
The main characteristic of student work during problem manipulation was the
process of mathematizing. In mathematization, students identified variables and
relationships within the mathematical entity and connected variables and relation-
ships with prior conditions and assumptions. The identification of the necessary
variables helped students to analyze and combine the required mathematical prop-
erties for constructing the mathematical model. In the Where to Live modeling
activity (see Table 1), for example, students identified the appropriate variables
(e.g., number of buildings, facilities, budget for each city) and combined the prop-
erties related to these variables for constructing a model for selecting the best
among a number of cities. The following snapshot from students’ work in the
Where to Live activity shows how finding the links between the different variables
resulted in a better model: “Fantanasia is also a good choice. It might not have
many schools and the road quality is average; however, the city has an increased
budget and authorities can use the extra money to improve city’s facilities.”
In prediction and solution verification process, students’ efforts first focused on
interpreting their solution (model) and on predicting the behaviour of the real
problem. Finally, students validated and communicated their results. Under com-
munication, a number of subprocesses were present, including the interchange of
ideas, information, and instructions about the mathematical model. Communicat-
ing helped students to explain the solution of the problem, to predict the behavior
of structurally similar problems, and to elaborate on and enrich their solution for
solving more complex problems. In the Where to Live modeling activity, for
example, students used the model they constructed (based on the varied impor-
tance of buildings, facilities, and budget) to predict how their models would deter-
mine the best city in terms of available buildings and facilities. In communicating
their solution, students explicitly documented how they used their model to rank
the different cities and how this ranking should be used for selecting the best city.

Students’ Modeling Abilities

Prior to conducting data analysis to examine the differences between the experi-
mental and the control group, an alpha coefficient of 0.924 was determined, dem-
onstrating the modeling abilities test’s reliability as a scale. An analysis of
variance test showed that there were no initial significant differences between the
300 MOUSOULIDES ET AL.

experimental and the control group, both for the sixth grade (F(1,195) = 0.427;
p > 0.05) and for the eighth grade (F(1,204) = 2.252; p > 0.05).
Latent growth modeling (LGM) analysis was conducted to analyze students’
modeling abilities over time and the impact of the intervention program. The
results of the LGM analysis showed that the impact of the intervention program
on students’ modeling abilities was significant. The analysis showed that experi-
mental group students outperformed their counterparts in the control group. Spe-
cifically, sixth-grade experimental group student modeling abilities’ rate of
change (slope=0.683) was positive and statistically significant and it was two and
a half times greater than the respective rate of change for the control group stu-
dents (slope=0.242). Similarly, eighth-grade experimental group student model-
ing abilities’ rate of change (slope=2.071) was three times greater than the
control group students’ rate of change (slope=0.647). The growth curve models,
the model fit indices, and the slopes for the four groups are presented in Table 2.
Students’ performances in the modeling abilities test as they were measured
during the three test administrations (Y1, Y2, and Y3) are presented in Figure 2.
Sixth graders’ modeling abilities results are presented in Figure 2a, and the corre-
sponding results for eighth graders are presented in Figure 2b.
A second finding showed a significant negative relationship between students’
initial achievement in the modeling abilities test and their rate of change, in both
sixth-grade (r = −0.449) and eighth-grade (r = −0.482) experimental groups. In
contrast, there were no significant relationships in the respective models for the
control groups. This indicates that the intervention program was effective for all
students, but more effective for the students with lower modeling abilities. This
validates the claims made by Lesh and Zawojewski (2007) on the effectiveness of
modeling activities with lower achieving students. On the contrary, for the control
groups, the results indicate that students with lower achievement scores were likely
to also have lower rate of change. Based on the above findings, it can be claimed
that the context of the modeling activities, social interactions between the students
and the teacher, and the absence of direct instruction created a safe environment

TABLE 2
The Latent Growth Models for the Four Groups

2
x
Group Function df CFI RMSEA Slope (S)

6th Gr. Exper. 0, 1, 2 1.625 0.996 0.05 0.683

8th Gr. Exper. 0, 1, 2 1.993 0.994 0.06 2.071
6th Gr. Control 0, 1, 2 1.611 0.998 0.05 0.242
8th Gr. Control 0, 1, 2 1.751 0.995 0.05 0.647

Degrees of Freedom, CFI: Comparative Fit Index, RMSEA: Root Mean Square Error of
Approximation.
 TEACHING AND LEARNING MATHEMATICAL PROBLEM SOLVING 301

1
0,95
0,9
Experimental
0,85
Control
0,8
0,75
0,7
Y1 Y2 Y3
(a)

1,1
1
0,9
Experimental
0,8
Control
0,7
0,6
0,5
Y1 Y2 Y3
(b)

FIGURE 2 Sixth-Grade (a) and Eighth-Grade (b) Students’ Modeling Abilities.

for students with low modeling abilities to present their ideas, to discuss possible
solutions, and to finally improve their abilities in solving modeling problems.

Modeling Students’ Modeling Behavior

In synthesizing the results from the analysis of the students’ modeling processes
and their modeling abilities, this study proposes that students’ modeling behavior
can be described in three interconnected dimensions that are presented in Figure 3.
The first dimension of the proposed theoretical model presents the modeling
processes that appeared in students’ work in the modeling activities. The model-
ing cycle(s) is influenced by students’ modeling abilities and a number of factors
that influence their work here. With respect to the latter, the study identified the
following four factors:
1. Student Modeling Awareness. This comprises students’ prior modeling experi-
ences, their mastery of modeling and problem-solving skills, and their under-
standings of related mathematical concepts and processes. Students’ models and
solutions improved as students gained experience in solving similar problems.
2. Problem Context. Students’ models are shaped by the situation (context of the
problem) in which they are created. This was evident in a number of cases in this
302 MOUSOULIDES ET AL.

Modeling Processes

Description

Manipulation
Verification

Prediction

Modeling Abilities Influencing Factors

FIGURE 3 Students’ Modeling Behavior.

study. In straightforward activities, such as the Best Drug Award activity, stu-
dents (even individually) easily reached satisfactory solutions, but they did not
get involved in model refinement and improvement, as they did not see the need
to cooperate and communicate their results. On the contrary, students’ results
were much better when activities were complex enough to challenge the students
and not so complex to discourage them. In such activities, students constructed a
number of different models, improved and refined them through a number of
modeling cycles, and finally successfully solved the related problems.
3. Grade Level. Students from different grade levels investigated and tried to solve
the problems using quite different approaches. Results showed that eighth grad-
ers constructed more complex and refined models in comparison to sixth grad-
ers. Eighth graders also adopted and transformed existing models more readily
than sixth graders, engaged in a greater number of modeling cycles, more effec-
tively communicated their results, and reflected on their solutions.
Additionally, eighth-grade students used more sophisticated formulas and
approaches in constructing their models. Considering that eighth graders are
taught mathematics in a more formal and abstract way, focusing on formulas,
symbolic expressions and algorithms, it could be argued that this teaching
approach might be a possible reason that the eighth graders effectively used
such formulas and algorithms in solving the modeling problems. Eighth grad-
ers’ discussions were also more precise, with improvements in ideas and mod-
els evident through the peer interaction. Students managed to solve conflicts
that arose and overcame related problems.
4. Presence of Tools. In three modeling activities (University Cafeteria, New
House, and Car Painting activities) a spreadsheet and a dynamic geometry
 TEACHING AND LEARNING MATHEMATICAL PROBLEM SOLVING 303

software were available for student use. Students effectively employed the soft-
ware’s capabilities and functionality, not only to make calculations quickly, but
also to export their data using multiple forms of representations and connecting
these representations to construct a model for solving the problem.
The interaction between students’ modeling abilities and the process components
of modeling (modeling processes) is continuous and reciprocal. Students’ modeling
abilities influence the way students solve a modeling problem. Similarly, as students
work on modeling problems and move through the modeling cycles, their modeling
abilities are shaped and further developed. The proposed theoretical model also sug-
gests the existence of a reciprocal relationship between students’ modeling abilities
and the influencing factors. These factors affect students’ abilities and, at the same
time, students’ abilities might change the impact of these factors on the modeling
processes. The interconnection and the continuous interaction between the three
dimensions shape and formulate students’ “modeling behavior” in such activities.

EPILOGUE

The findings of the study have implications for the design of modeling activities for
elementary and middle school students, for teaching mathematical modeling as a
didactic means for mathematical problem solving, and for ultimately improving
students’ modeling abilities. The range of models constructed by students suggests
that teachers need to be aware that modeling activities will produce diverse strate-
gies and thinking on the part of the students. Finally, it is clear that there are several
areas of research related to students’ modeling behavior that need to be further
investigated—these include other applications of the proposed model for students’
modeling behavior, exploration of younger students’ work with modeling, and tri-
aling of various types of modeling activities that are meaningful to students.

NOTE

1. The Modeling Test and the Modeling Activities can be found in Mousoulides (2007).

REFERENCES

Blum, W. (2002). ICMI 14 study: Applications and modeling in mathematics education. A discussion
document. Educational Studies in Mathematics, 51, 149–171.
Blum, W., & Niss, M. (1991). Applied mathematical problem solving, modeling, applications, and
links to other subjects—state, trends, and issues in mathematics instruction. Educational Studies in
Mathematics, 22, 37–68.
304 MOUSOULIDES ET AL.

Doerr, H., & English, L. D. (2003). A modeling perspective on students’ mathematical reasoning
about data. Journal of Research in Mathematics Education, 34, 110–136.
English, L. D. (2006). Mathematical modeling in the primary school. Educational Studies in Mathe-
matics, 63, 303–323.
English, L., & Watters, J. (2005). Mathematical modeling in the early school years. Mathematics
Education Research Journal, 16, 59–80.
Kelly, A. E., & Lesh, R. A. (Eds.) (2000). Handbook of research design in mathematics and science
education. Mahwah, NJ: Lawrence Erlbaum.
Lesh, R., & Doerr, H. M. (Eds.) (2003). Beyond constructivism: A models and modeling perspec-
tive on mathematics problem solving, learning and teaching. Mahwah, NJ: Lawrence Erlbaum
Associates, Inc.
Lesh, R., & Zawojewski, J. S. (2007). Problem solving and modeling. In F. Lester (Ed.), Second
handbook of research on mathematics teaching and learning (pp. 763–804). Greenwich, CT: Infor-
mation Age Publishing.
Mousoulides, N. (2007). The modeling perspective in the teaching and learning of mathematical
problem solving. Unpublished Doctoral Dissertation, Nicosia, University of Cyprus.
Organization for Economic Co-Operation and Development (2004). Problem solving for tomorrow’s
world—First measures of cross curricular competencies from PISA 2003. Retrieved July 1, 2007,
from http://www.oecd.org/dataoecd/25/12/34009000.pdf.
View publication stats