Thesis Chiaka Drakes

MATHEMATICAL MODELLING:
FROM NOVICE TO EXPERT

by
Chiaka I. Drakes
M.Sc., University of Bath, 2005
B.Sc. (Hons.), University of the West Indies, 2000
a Thesis submitted in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
in the
Department of Mathematics
Faculty of Science
c Chiaka I. Drakes 2012
SIMON FRASER UNIVERSITY
Summer 2012
All rights reserved.
However, in accordance with the Copyright Act of Canada, this work may be
reproduced without authorization under the conditions for Fair Dealing.
Therefore, limited reproduction of this work for the purposes of private study,
research, criticism, review and news reporting is likely to be in accordance
with the law, particularly if cited appropriately.
APPROVAL
Name: Chiaka I. Drakes
Degree: Doctor of Philosophy
Title of Thesis: Mathematical Modelling: from Novice to Expert
Examining Committee: Dr. Mary-Catherine Kropinski, Associate Professor
Chair
Dr. JF Williams, Associate Professor
Senior Supervisor
Dr. Peter Liljedahl, Associate Professor, Education
Co-Supervisor
Dr. Nilima Nigam, Associate Professor
Supervisor
Dr. Malgorzata Dubiel, Senior Lecturer
SFU Examiner
Dr. Chad Topaz, External Examiner,
Associate Professor of Mathematics,
Macalester College
Date Approved:
ii

PartiaI Copyright Licence

iii
Abstract
This study strives to understand how mathematical modelling is perceived by novice, inter-
mediate and expert modellers, through comparing and contrasting their understanding and
habits of modelling. The study adopted a qualitative methodology based on observations,
interviews and surveys of 78 participants. This included 14 experts who are professors, 11
intermediates consisting of graduate students and post-doctoral fellows, and 53 undergrad-
uates or novices. The study incorporated interviews of the professors and the post-graduate
participants, while questionnaires were utilized to understand the perspective of the un-
dergraduate students. The study revealed that the majority of expert participants see
modelling as a collaborative eort. There is a dichotomy among them regarding whether
mathematical modelling is the setting up of a mathematical model alone, which is deemed
an art, or if it includes the solving of the model, which is more a science. These dierences
have implications on how modelling is taught and how novices and intermediates in turn
will view the modelling process. Experts also vary in their opinion on whether models must
be veriable or not. One key feature of the experts approach is that they begin by assuming
that they do not understand the question asked and work to ensure that they do. This is
despite their superior ability to solve problems. Intermediate participants were more forth-
coming with their emotions on modelling than experts; they cited research as opposed to
collaboration as their primary means of dealing with barriers arising during the modelling
process, and gave credit to intuition as a skill needed for solving - something not mentioned
among the experts. Novices were the most descriptive about their feelings when modelling.
They conveyed a tendency to be more passive when encountering barriers, waiting for help
or giving up as opposed to actively working through the problems. Many of our results, in-
cluding those mentioned above, have implications for the teaching of eective mathematical
modelling.
iv
Acknowledgments
Thank God!
I am so happy to reach this stage of the PhD and for all of the people that have helped
me to get here. A PhD is a journey (some longer than others) and many people have helped
me along the way. I can only mention a few here, but a whole-hearted thanks to those of
you who have touched my life throughout the journey and helped me in numerous ways.
Sincere thanks to my supervisory committee. Thank you JF for being patient with me
while I decided which path to take and for all the discussions that helped me make this
idea a reality. Thank you Peter for helping me with all the non-mathematical aspects that I
knew nothing about. You have inspired me to be a better teacher as well as learner! Thanks
also Nilima for encouraging me to just start writing before I felt ready. The thesis would
not be written now without those words of encouragement.
A big thanks to my team of editors: Brian, Reynold and Dami, who have taken the time
to read my work, give me comments, to teach me some rules about commas and to help me
get these words coherently on paper. Each of you is more to me than an editor, you are
my family and I love you very much. An extra special thanks to Brian, for being the best
teammate a girl could ask for: looking forward to the next stage of our journey!
A special thanks to Cathy Trudeau who taught me to speak my truth and be compas-
sionate with myself, and who helped me to make the decision to take a path less travelled
on my journey along the way to the PhD.
Thank you to my cheerleading team: Dami, Karin, Donna, Joy, Ruky, Shansi, Mona-
Lisa, Marsha-Ann, Priye, Curtis, Dionne, Diana, Pat, Ted, the Bible study group at Cullo-
den church, Uncle Tony, Maria, Tricia and my head cheerleader Shernelle. You all believed
that this thesis would be written, even when I was 100% sure it wouldnt be. Thanks for
the prayers, the words of encouragement and allowing me to vent (and boy did I!)
v
Thanks to my online study group: Neil, Reynold and Thabiti, for all those weekly check-
ups, which motivated me when I was stuck. I dont think anyone in the world has a set of
brothers as awesome as you three! Looking forward to the next family reunion!
A heartfelt thank you to my parents Paulette and Roosevelt Drakes who nurtured in me
a love and respect for mathematics, and who would not accept me doing any less than my
best. Ever. You started me on my journey and I am forever grateful.
Finally, a sincere thank you to those who participated in this study and helped me
answer some of my most pressing questions about modelling.
vi
Contents
Approval ii
Partial Copyright License iii
Abstract iv
Acknowledgments v
Contents vii
List of Tables xi
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 The Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Literature Review 6
2.1 Mathematical Modelling Literature . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.1 Issues with Modelling: Denition . . . . . . . . . . . . . . . . . . . . . 7
2.1.2 Issues with Modelling: Teaching Approach . . . . . . . . . . . . . . . . 13
2.1.3 Issues with Modelling: from Novice to Expert . . . . . . . . . . . . . . 13
2.2 Problem Solving Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.1 Problem Solving: by Design . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.2 Problem Solving: Incorporating the Extra-Logical . . . . . . . . . . . 18
2.3 Psychology of Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.1 Math Anxiety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
vii
2.3.2 Motivation and the Type I Personality . . . . . . . . . . . . . . . . . . 23
2.3.3 Creativity and Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3 Methodology 28
3.1 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Overview of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 Group 1: Expert Modellers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.4 Group 2: Intermediate Modellers . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.5 Group 3: Complete Novices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.6 Group 4: Novice Modellers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.7 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4 Experts 38
4.1 Dening Mathematical Modelling . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 Initial Thoughts, Plans and Feelings . . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Dealing with Initial Barriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.4 Determining Relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.5 Skills Needed for Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.6 Dealing with Being Stuck . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.7 Heuristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.8 Verifying Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.9 Changing Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.10 Dicult vs Easy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.11 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5 Intermediates 76
5.1 Dening Mathematical Modelling . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.2 Initial Thoughts, Plans and Feelings . . . . . . . . . . . . . . . . . . . . . . . 78
5.3 Dealing with Initial Barriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.4 Determining Relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.5 Skills Needed for Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.6 Dealing with Being Stuck . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.7 Heuristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
viii
5.8 Verifying Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.9 Changing Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.10 Dicult vs Easy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.11 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6 Novices 127
6.1 Question 1: Dening Mathematical Modelling . . . . . . . . . . . . . . . . . . 128
6.2 Question 2: Initial Thoughts . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
6.3 Question 3: Initial Barriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
6.4 Question 4: Determining what is relevant to the model . . . . . . . . . . . . . 134
6.5 Question 5: Mathematics Used in Modelling . . . . . . . . . . . . . . . . . . . 136
6.6 Question 6: Dealing with Being Stuck . . . . . . . . . . . . . . . . . . . . . . 138
6.7 Question 7: Problem Solving Heuristics . . . . . . . . . . . . . . . . . . . . . 141
6.8 Question 8: Experiencing AHA! . . . . . . . . . . . . . . . . . . . . . . . . . . 143
6.9 Question 9: Learning to Do modelling by Doing it? . . . . . . . . . . . . . . . 145
6.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
7 Complete Novices 149
7.1 Question 1: Time needed to solve a word problem . . . . . . . . . . . . . . . 150
7.2 Question 2: Initial Thoughts and Feelings . . . . . . . . . . . . . . . . . . . . 152
7.3 Question 3: Initial Barriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
7.4 Question 4: Plan or Plunge? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
7.5 Question 5: Organisation of work . . . . . . . . . . . . . . . . . . . . . . . . . 166
7.6 Question 6: Dealing with Being Stuck . . . . . . . . . . . . . . . . . . . . . . 168
7.7 Question 7: Problem Solving Heuristics . . . . . . . . . . . . . . . . . . . . . 173
7.8 Question 8: Verication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
7.9 Question 9: Dicult vs Easy . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
7.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
8 Conclusions 187
8.1 Contribution to Mathematical Modelling . . . . . . . . . . . . . . . . . . . . . 187
8.2 Research Question 1: What IS Mathematical Modelling? . . . . . . . . . . . . 188
8.3 Research Question 2: How does one move from stuck to unstuck? . . . . . . . 189
ix
8.4 Research Question 3: What are the dierences between the novice and expert
modeller? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
8.4.1 Dealing with being stuck . . . . . . . . . . . . . . . . . . . . . . . . . 190
8.4.2 The Psychology of Modelling . . . . . . . . . . . . . . . . . . . . . . . 192
8.4.3 Dening Mathematical Modelling . . . . . . . . . . . . . . . . . . . . . 193
8.4.4 Starting a Modelling Problem . . . . . . . . . . . . . . . . . . . . . . . 194
8.4.5 Knowledge and Skills Valued . . . . . . . . . . . . . . . . . . . . . . . 196
8.4.6 Verifying the Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
8.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
9 A Look to the Future 205
9.1 Implications for Teaching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
9.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Bibliography 210
x
List of Tables
2.1 Otto & Days Box 2.1 - A seven step mathematical modelling cycle. . . . . . 12
2.2 Polyas description of the dierent stages of problem-solving. . . . . . . . . . 16
xi
Chapter 1
Introduction
This story begins in 2007 when I rst decided to model crowd ow as a thinking uid. I
had done some work on this in my Computational Fluid Dynamics class and was pursuing
it further as my thesis topic. I found the work dicult to say the least. All Ph.D. work
is probably daunting at the beginning, but I was completely overwhelmed. I had no idea
what modelling was! How to start? What factors to include? What to do? I was constantly
stuck and unable to get unstuck without the specic and constant help of my supervisor.
My background had had little or no real modelling in it. Yes, I had worked on many
equations that modelled situations in real life during my masters degree, but I had never
started from scratch.
After working on crowd ow for a year with little progress, and reaching a point where I
was considering quitting my Ph.D. completely, I entered into a discussion with my supervi-
sor. He is an excellent modeller himself and teaches courses at SFU on modelling. He had
attended and hosted several modelling camps and workshops and understood how to work
collaboratively with others to acquire desired results. I asked him what I thought then were
simple questions: How does one DO modelling well? What are the skills needed to be a
good modeller? I thought that perhaps if I could work on these skills then I too could gure
out how to do my then thesis work. This last line of questioning became the starting point
of an even greater discussion, as we could not pinpoint what the magic was.
I thought perhaps it was dicult for him to articulate exactly what his process was, as
it had become instinctive after doing it for so long. So I tried to tackle this from a dierent
perspective, asking him what his learning objectives were when teaching a modelling class.
This yielded another interesting discussion, this time about identifying modelling outcomes.
1
CHAPTER 1. INTRODUCTION 2
This was signicant as many outcomes were not simply the particular mathematical content
he was covering. How does one know at the end of a modelling class that students have
learned the skills required to do modelling well? How can we check that the students who do
well are not simply those that had already acquired the necessary mathematical modelling
skills before taking the class? In other words, if the nal exam was given on the rst day of
class, apart from their understanding of the content knowledge, would they do the same as
at the end of the course?
This led me to ask what I then considered the most basic question of all: What IS mod-
elling? Even this proved to be unclear since denitions varied in textbooks and modelling
literature alike. Modelling is a broad area and to try to dene it is dicult. And so began
my quest and a new thesis topic to understand this aspect of mathematics which I had little
experience with, but realized was a very important part of doing applied mathematics.
1.1 Motivation
Mathematical modelling is an important aspect of the applied mathematics curriculum. It
provides students, particularly graduate students, with the skills to succeed professionally
in industry. It gives these students the tools to analyze, understand and forecast based on
data that in this age is easily accessible to them. Modelling helps students to transfer the
knowledge that they have learnt in their less open-ended classes, to real-world problems.
This transfer of knowledge is a skill that students can go on to use in the workplace and
other areas of study. Since most mathematics and applied mathematics students will not
go on to be mathematicians, these skills had better prepare them for whatever they aspire
to do next. In the 1970s McLone [1] reported that mathematics graduates had diculty
when moving from the classroom to the workplace:
Good at solving problems, not so good at formulating them, the graduate has
a reasonable knowledge of mathematical literature and technique; he has some
ingenuity and is capable of seeking out further knowledge. On the other hand
the graduate is not particularly good at planning his work, nor at making a
critical evaluation of it when completed; and in any event he has to keep his
work to himself as he has apparently little idea of how to communicate it to
others. ( [1], p.33)
In 1995 this was still the case and PhD students reported not feeling able to tackle real
world problems in industry [2].
As many universities promise to prepare their students to engage the world, modelling is
one area that allows universities to fulll the promise of equipping students with the knowl-
edge, skills, and experiences that prepare them for life in an ever-changing and challenging
world [3]. To this end universities often have courses, undergraduate degrees and graduate
degrees that cater to the modelling aspect of applied mathematics. This is precisely because
mathematical modelling links mathematics to real life phenomena. It occupies a middle
ground between mathematics and most other sciences and engineering disciplines [4].
In order to analyze the entire world around us, mathematical modelling is key. Modelling
problems are found in physics, engineering, chemistry, computer science, biology and even
in such subjects as psychology and sociology [4]. Mathematical models provide insight
into how various forces act to change a cell, an organism, a population, or an assemblage of
species [5]. They help to address the complex questions arising in healthcare where simple
intuition is not enough, allowing us to develop solid, defendable, evidence-based answers
to those questions [6]. This is just the tip of the iceberg.
Industrialists also value mathematical modelling, as it is cheaper than doing full exper-
iments or simulations in many cases. Many modelling problems come from industry, where
the detailed mechanics of the problems are often not as well known as one might think due
to the expense of experimentation: when you see the operating conditions ferocious tem-
peratures, inaccessible or minute machinery, corrosive chemicals you realise how expensive
and dicult it would be to carry out detailed experimental investigations [7]. Not only is
it dicult and expensive to use experiments to understand some of the industrial problems,
but sometimes it is simply impossible. The use of mathematical models allows us to explore
the problem more extensively:
It is often easy to vary parameters in the mathematical model over wide ranges,
whereas this may be very time-consuming or expensive, if not impossible, in an
experimental setting. ( [8], p.50)
Mathematical modelling is clearly an important skill for mathematics students to learn
as well as those from other disciplines. Having identied the signicance of modelling to
universities, students, industrialists and those in other disciplines, the next step is to identify
the areas of concern when doing and teaching modelling.
1.2 The Problem
Mathematical modelling courses have several problems that a classic mathematics course
(such as an introductory dierential calculus course) would not have. One problem with
teaching mathematical modelling is that we want to use precise mathematics to t imprecise
problems for which there may be no well-dened solution at all!:
It is the nature of real-world problems that they are large, messy and often
rather vaguely stated. It is very rarely worth anybodys while to produce a
complete solution to a problem which is complicated and whose desired out-
come is not necessarily well specied (to a mathematician). Mathematicians
are usually most eective in analysing a relatively small clean subproblem for
which more broad-brush approaches run into diculty. ( [7], p.4).
In many areas of mathematics the information needed is given and clearly laid out. However,
in modelling oftentimes much of the information must be discarded in order to boil the
problem down to its essence. This is non-rigorous, and requires creativity. A subsequent
issue that arises is what the learning outcomes should be in a mathematical modelling course
or degree. How should the instructor of modelling approach the subject, and what aspects
should she focus on?
The preferred mode of teaching modeling is quite problematic: Does one
have the students gain experience by having them face real problems (usually
ill-dened) and thus learn by doing, or does one use a book or books presenting
them with worked-out approaches? Should they know a fair amount of math
rst, or should they learn what is necessary as they face new challenges? Does
one mix the areas of problems, the continuous or discrete setting, the level of
mathematical sophistication? There is a host of hypotheticals.
I have given modeling courses quite a number of times, but I dont have many
strong views on the answers to what I have posed above. I have varied what
I have done and never been completely satised. (And establishing a student
grade is quite a challenge!) [9].
The problem is made worse by the fact that some mathematics students believe from ele-
mentary school onwards that mathematics problems all have a unique closed-form solution
that the teacher knows the proverbial right answer. This is to be expected on a develop-
mental level as college students begin with a belief that there are right and wrong answers
that can be acquired from the authorities and are to be memorized, as explained by Perry
in his work on intellectual development [10, 11]. However, this belief is in contrast with the
way mathematical modelling works as we will see in the literature discussed in the following
chapter (Chapter 2).
1.3 Summary
This introduction illustrates that mathematical modelling is an area of importance in which
experts do well. However, due to the messy nature of the problems and the issues involved
when attempting to teach modelling, novices have diculty with it. The practices of novices
are signicantly dierent from those of the experts, and novices are more inclined to be
overwhelmed by the enormity of the task of modelling [12,13]. The literature addresses some
of the issues that novices have but additional work must to be done in order to understand
the transition from novice to expert in more detail [14]. In my eort to understand the true
nature of mathematical modelling, I rst attempted to clarify what modelling means as it
covers such a broad range of skills and topics. I also looked to identify some of the issues
associated with the transition from novice modeller to expert. This was accomplished by
working with modellers with a range of ability to identify the dierences in the skill-set,
behaviour, and attitude of the varying groups. In particular, I tried to understand what
skills the experts use to move from being stuck to becoming unstuck.
In Chapter 2, I discuss the literature pertaining to modelling including the transition
from novice to expert. This discussion highlights that there are still some decits in the
modelling literature. A review of problem solving literature as well as literature on the pro-
posed psychological contributions suggest a framework for analysing the research questions
that emerge from the literature review. To answer those aforementioned research questions,
I describe the methods employed for conducting this study, in Chapter 3. This includes a
discussion of the participants who range from novice to expert, alongside an explanation of
the data-gathering process. I also identify the limits of the chosen research methodology.
In Chapters 4 7, I present the results and analysis of the study, and discuss conclusions
in Chapter 8. Finally, in Chapter 9, I suggest some implications these results have on the
teaching of mathematical modelling at the tertiary level and directions for future work.
Chapter 2
Literature Review
Mathematical modelling has a short but rich history in the literature, which highlights
several of the issues of the mathematical modelling process. I studied the mathematical
modelling literature, focusing principally on modelling textbooks, as these are the primary
resource for teaching and learning modelling. I also looked at articles found in the In-
ternational Community of Teachers of Mathematical Modelling and Applications (Ictma)
journals, that discuss dierences between the novice and the expert modeller. This was
followed by a look at the problem-solving literature, highlighting the similarities between
problem-solving and modelling while attempting to gain some insight into the issues that
the modelling literature has left uncovered. Finally, I investigated the possible psychological
issues that may aect the modelling process.
2.1 Mathematical Modelling Literature
In order to start looking at the broad issue of what exactly mathematical modelling is, I ex-
amined the mathematical modelling literature to begin addressing the questions introduced
in Chapter 1. I looked at two dierent types of literature on modelling: textbooks and
Ictma journal articles. These two types of literature encompass two dierent perspectives:
those who do and teach modelling and those who research modelling education and culture.
I began with a look at mathematical modelling textbooks, as these are the primary tools
for teaching modelling and are often a students rst introduction to the eld.
There is a wide range of mathematical textbooks available, but I focused on two cat-
egories: textbooks written by mathematicians primarily for students of mathematics and
6
CHAPTER 2. LITERATURE REVIEW 7
textbooks written by scientists in dierent elds to cater to students in various scientic
disciplines. I felt that the varying backgrounds of both the authors and the intended end
users would provide a spectrum of approaches to teaching modelling.
In the rst textbook category, those written by mathematicians primarily for students of
mathematics, I examined four books: Mathematical Modelling: A Case Studies Approach,
by Illner et al. [4]; Practical Applied Mathematics: Modelling, Analysis, Approximation, by
Howison [7]; Mathematical Models in the Applied Sciences, by Fowler [15]; and Ants, Bikes
and Clocks: Problem Solving for Undergraduates, by Briggs [16]. Note here that although
the title of Briggs textbook includes problem-solving, he species in the introduction of
his work that this book highlights modelling or story problems ( [16], p.2).
In the second category of modelling textbooks, those written by scientists in dierent
elds to cater to students in various scientic disciplines, I examined three books: The Na-
ture of Mathematical Modeling, by Gershenfeld [17] (Physicist and Computing Scientist), A
Biologists Guide to Mathematical Modeling in Ecology and Evolution, by Otto (Zoologist)
& Day (Mathematical Biologist) [5], and Modelling in Healthcare, by SFUs Complex Sys-
tems Modelling Group or CSMG [6] (Mathematicians, Physicists, Epidemiologists). These
two categories of authors approach the denition and the teaching of modelling in dierent
ways.
2.1.1 Issues with Modelling: Denition
The rst thing of note was the variation in denition of mathematical modelling. Illner et
al. give a very broad denition of modelling:
Mathematical modelling is a subject without boundaries in every conceivable
sense. Wherever mathematics is applied to another science or sector of life, the
modelling process enters in a conscious or subconscious way. ( [4], p.xi)
This denition certainly covers the breadth of modelling but is not a working denition.
Consistent with this, Howison explains that modelling should not be precisely dened:
There is no point in trying to be too precise in dening the term mathematical
model: we all understand that it is some kind of mathematical statement about
a problem originally posed in non-mathematical terms. ( [7], p.4).
Howison goes on to explain that while all models do not explain cause and eect, all useful
models should be predictive. Illner et al. make no such claims; however, Howisons colleague
Fowler also raises the issue of predictive power in modelling:
A model is a representation of a process. Usually, a mathematical model takes
the form of a set of equations describing a number of variables [...] Applied
mathematicians have a procedure, almost a philosophy, that they apply when
building models. First there is a phenomenon of interest that one wants to
describe or, more importantly, explain. Observations of the phenomenon lead,
sometimes after a great deal of eort, to a hypothetical mechanism that can ex-
plain the phenomenon. The purpose of a model is then to formulate a description
of the mechanism in quantitative terms, and the analysis of the resulting model
leads to results that can be tested against the observations. Ideally the model
also leads to predictions, which, if veried, lend authenticity to the model. It
is important to realize that all models are idealizations and are limited in their
applicability. In fact one usually aims to over-simplify ( [15], p.3).
Fowlers denition of modelling is initially quite succinct, but he goes on to elaborate with a
description of a modelling cycle, which is the applied mathematicians philosophy of building
a model [15]. He acknowledges that going from the observation to the mechanism sometimes
requires signicant eort, and also raises the idea of simplication often being the aim.
Fowler also describes some of the issues that are associated with mathematical modelling,
in particular the teaching of it:
Mathematical modeling is a subject that is dicult to teach. It is what applied
mathematics (or to be precise, physical applied mathematics) is all about, and
yet there are few texts that approach the subject in a serious way. Partly, this is
because one learns it by practice: There are no set rules, and an understanding
of the right way to model can only be reached by familiarity with a wealth of
examples.( [15], p.3).
Fowlers view that modelling is learned by practice is a very common one (see Gershenfeld
and Otto & Day below). He also implies by his use of quotation marks that the right way
to model is somewhat subjective.
These three denitions show an increase in specicity. Briggs denition of modelling
appears along the same lines at rst glance:
Modeling or story problems are quantitative problems that are posed in a re-
alistic context. A key distinction of these problems is that they are not posed
explicitly as mathematical problems. For this reason, their solution requires
an essential preliminary step that may be the crux of the solution. That step,
often called modeling, is to transform the stated problem from words into math-
ematics. Having formulated the problem in mathematical terms, it must still be
solved! ( [16], p.2).
On the surface this seems to be re-iterating Fowlers description of going from observation
to mechanism, description in quantitative terms, followed by analysis; however, it is quali-
tatively dierent. Briggs identies modelling as the step from phenomenon to mathematics
only. What follows after is no longer modelling, but solving. Briggs also does not require the
question to actually come from reality, only that it be posed in a realistic context, and makes
no mention of predictive ability. Therefore it cannot be construed that mathematicians who
do modelling all have the same working denition.
Interestingly, neither Gershenfeld nor Otto & Day explicitly dene what a mathematical
model is. They describe issues or steps involved when building a model, but never specically
say what it is that they are building. Gershenfeld does not even describe a modelling cycle
but instead raises some issues involved in the modelling process:
To build a model there are many decisions that must be made, either explicitly
or more often, implicitly. Some of these are shown in Figure 1.1. Each of these
is a continuum rather than a discrete choice. This list is not exhaustive, but its
important to keep returning to it: many eorts fail because of an unintentional
attempt to describe either too much or too little.
These are meta-modeling questions. There are no rigorous ways to make these
choices, but once theyve been decided there are rigorous ways to use them.
Theres no single denition of a best model, although quasi-religious wars are
fought over the question. ( [17], pp.1-2)
Gershenfelds explanation of modelling decisions being made implicitly and without rigour
hints at the lack of precision in the modelling process. He also raises the idea of there being
no best model, which is tied to Fowlers statement of the right way to model.
Otto & Day also avoid dening mathematical modelling but raise several issues involved
in doing modelling. These issues touch more on the feelings that might be experienced when
modelling:
If you have seen mathematical models but never constructed one, it may appear
like an overwhelming task. Where do you start? What is the goal? How do
you know whether the model makes sense? This chapter outlines the typical
process of modeling and gives helpful hints and suggestions to break down the
overwhelming task into manageable bits. The most important piece of advice
is to start. Start thinking about problems that puzzle you. Grab a piece of
paper and start drawing a ow diagram illustrating various processes at work.
The biggest hurdle preventing most biologists from modeling is the paralysis
one feels in the face of mathematics; [...] Over time, you will learn more tools
and techniques that will allow you to avoid pitfalls and to get further with the
problems that interest you. Your intuition will develop to help you see when
something is wrong with your model and to help you interpret your results ( [5],
p.17)
Otto & Day not only acknowledge the diculties of modelling, but also suggest how to
deal with them (just start) and reassure the reader that in time their intuition will develop.
They go on to describe a seven-step modelling process (see Table 2.1), again acknowledging
some student concerns:
Box 2.1 describes, in seven steps, how to construct a dynamical model. This is
like describing how to ride a bike in a series of steps; obviously we can only give
an idea about how the process works. Mastering the steps requires practice, [...]
the rst step, coming up with a question, can be more dicult than it sounds. In
most biology classes, students are told what the questions are and what answers
have been found. Rarely are students asked to formulate scientic questions
for themselves. This is very unfortunate because, in any scientic enterprise
(modeling or otherwise), the process begins with a question. ( [5], pp.17-19)
Otto & Day address the fact that the initial step in modelling (coming up with a question)
is a dicult one. This reinforces both Fowlers statement that moving from phenomenon
to hypothetical mechanism takes a great deal of eort, as well as Gershenfelds mention
of implicitness. Otto & Day also explain that this comes from a decit in modelling in
students background; they suggest starting simply as a means of dealing with this initial
diculty, and that mastery comes with practice.
The third modelling textbook, written by the CSMG primarily for those in healthcare [6],
does venture a denition of modelling:
In this book the word model means a simplied representation of a real-world
situation used to help answer a specic question. ( [6], p.4).
Note that the authors clarify that this denition pertains to this book only, implying that
there may be other denitions of modelling. The authors in this case are mathematicians
who have come to interdisciplinary work later in their careers, and are cognizant of the
fact that common words have disparate meanings in dierent communities. This denition
includes the need of a model to answer a specic question (which was alluded to in the
previous denitions) without the emphasis on the question needing to be specic. The
authors go on to explain what makes a good model, re-iterating Gershenfelds idea that
there is no best model:
[A] good model has as low a complexity as possible while retaining the details
necessary to approach the specic question the model is designed to examine.
In general, models with a focused question and a limited number of conditions
are more likely to be useful [...] there is no such thing as a unique best model
for a given problem. In fact, in most cases, more than one model discussed in
this book is applicable in solving a single question. In these cases dierent mod-
elling methods are often complementary, with the best results obtained through
an approach that integrates multiple methods. In general, modelling is most
convincing when various dierent kinds of models lead to the same conclusion.
( [6], pp. 4-5).
In contrast to Howison and Fowler who value the predictive power of the model, the CSMG
speak of a comparison of methods to lend validity to the model. This book also provides a
owchart of the modelling process.
Box 2.1: Seven Steps to Modeling a Biological Problem
Step 1: Formulate the question
What do you want to know?
Describe the model in the form of a question.
Boil the question down!
Start with the simplest, biologically reasonable description of the prob.
Step 2: Determine the basic ingredients
Dene the variables in the model.
Describe any constraints on the variables.
Describe any interactions between variables.
Decide whether you will treat time as discrete or continuous.
Choose a time scale.
Dene the parameters of the model.
Describe any constraints on the parameters.
Step 3: Qualitatively describe the biological system
Draw a life-cycle diagram for discrete-time models involving multiple
events per time unit.
Draw a ow diagram to describe changes to the variables over time.
For models with many possible events, construct a table listing the out-
come of every event.
Step 4: Quantitatively describe the biological system
Using the diagrams and tables as a guide, write down the equations.
Perform checks. Are the constraints on the variables still met as time
passes? Make sure the units of theRHS equal those on the LHS.
Think about whether results from the model can address the question.
Step 5: Analyze equations
Start by using the equations to simulate and graph the changes to the
system over time.
Choose and perform appropriate analyses.
Make sure that the analyses can address the problem.
Step 6: Checks and balances
Check the results against data or any known special cases.
Determine how general the results are.
Consider alternatives to the simplest model.
Extend or simplify the model, as appropriate, and repeat steps 2-5.
Step7: Relate the results back to the question
Do the results answer the biological question?
Are the results counterintuitive? Why?
Interpret the results verbally, and describe conceptually any new insights
into the biological process.
Describe potential experiments.
Table 2.1: Otto & Days Box 2.1 - A seven step mathematical modelling cycle.
2.1.2 Issues with Modelling: Teaching Approach
Along with the variation in the denition of modelling, there is variation in the approach to
teaching it. The rst category of authors, aiming at the students of mathematics, approach
modelling by making use of case studies as a means of exposition. Illner and Howison
introduce case studies almost immediately. There is some overview of the modelling process
discussed in Fowlers text, before case studies are presented. Briggs goes further with a
discussion on problem-solving heuristics rst, followed by case studies that highlight dierent
heuristics or steps in the modelling cycle. In contrast to the case studies approach, the
authors aiming primarily at those outside of the mathematics eld address the teaching
of modelling by focusing on techniques or modelling by design. Gershenfeld discusses the
techniques used for solving models including analytic, numeric and observational modelling
techniques, thus only addressing one area in the modelling cycle (solving the model). After
a description of the overall modelling process Otto & Day also discuss several modelling
techniques for solving problems arising in Biology. The exception to this is the CSMG who,
like Briggs, make use of case studies while highlighting the dierent aspects of the modelling
process when looking at each case.
2.1.3 Issues with Modelling: from Novice to Expert
The textbook literature reveals a dichotomy of approach to modelling: case studies or
modelling by design. There are also varying denitions of modelling, and in some cases no
denition at all. This lack of precision when dening modelling is understandable but begs
the question, are all modellers describing and focusing on the same thing when they use
the term mathematical modelling? While in many cases these authors have addressed some
of the diculties of the student, what they have not addressed are the skills necessary to
move from novice to expert. It is therefore necessary to continue on to the Ictma modelling
articles to see how the spectrum from novice to expert is addressed.
Crouch and Haines outline several cognitive and meta-cognitive dierences between ex-
perts and non-experts [12,18,19]: three of them will be highlighted here. The rst dierence
between experts and novices is their approach to problems. Experts begin with analysis and
a plan. They constantly return to re-examine the problem and re-dene variables. Novices
on the other hand tend to plunge in, go straight to equations and stick to their original
thoughts regardless of where these thoughts lead [2022]. Experts tend to use forward
thinking more often and better than their novice counterparts, who are more likely to work
backwards from the solution [2224]. This has serious implications for modelling, as using
the results to create the model results in a biased model. The third and nal dierence
is a meta-cognitive one. Experts not only have better domain-specic knowledge, but this
knowledge is also better inter-connected. This superior knowledge causes experts to focus on
underlying principles. Novices on the other hand have knowledge that is loosely connected
and tend to focus on the surface features of the problem rather than the underlying princi-
ples [25, 26]. It must be noted that these skills of expertise described here take a relatively
long time to acquire [27].
This look at the textbooks and the literature on modelling education provides a general
idea of the modelling process and the dierences we expect between the novice and the
expert modeller on a cognitive and meta-cognitive level. However, looking at them together
still does not provide a novice modeller with enough information to move along the path to
expertise. There are various denitions of modelling available making it dicult to ensure
that experts and novices are thinking of the same process when talking about mathematical
modelling. The cognitive decits of the novice modeller are cited, but as these aspects of
expertise take time to develop, the novices cannot force themselves to be more expert at
organising their thoughts, for example. There is also no discussion evidenced of how to get
unstuck, except via simplication of the model. This creates problems in the case when it
is uncertain how to simplify the model, and also if the model already seems to be in its
simplest form.
To address the issue of being stuck I turned to the literature available on problem-solving.
Problem-solving is not identical to mathematical modelling. As will be seen in the discussion
of the literature in Section 2.2, there is a range of opinion about problem-solving problems
being problematic i.e. it is only a problem if it causes the solver to be stuck [28, 29]. Our
above look at the varying modelling denitions and ideas has not revealed that modelling
problems have to be problematic in nature. However, since modelling problems do tend to
be messy [7], and the formulation step is important and not always obvious [15, 16], then
modelling problems can have problematic aspects to them, causing the modeller to be stuck
in the modelling process.
The issue of moving from a state of being stuck to becoming unstuck has been one of the
main focuses of problem-solving literature in the last few decades. In particular, problem-
solving literature has pushed past initial descriptions of problem-solving by design, which
focuses on the cognitive and logical processes only, to identify the extra-logical processes
involved in mathematical modelling. Therefore, I looked at how problem-solving was ad-
dressed in the earlier literature, and how it is approached now, focusing on the issue of
moving forward from a state of being stuck.
2.2 Problem Solving Literature
There are two main approaches to problem-solving. The rst approach focuses on the cog-
nitive aspects of problem-solving, which are the mental operations that support peoples
acquisition and use of knowledge [30, 31], in other words, the way we think. The word
acquisition suggests that deliberate eort is used in achieving the knowledge. This rst
approach is similar to what we have seen in the literature on mathematical modelling in
Section 2.1.3. The second approach to problem-solving pushes past the logical (the use of
deliberate reasoning), to the extra-logical (the use of subconscious processes) in order to
deal with the issue of being stuck. The term extra-logical encompasses those mental pro-
cesses that are not achieved through deliberate conscious thought, which includes creativity,
intuition, and illumination also known as the AHA! experience [32, 33].
2.2.1 Problem Solving: by Design
Looking at Polyas How to Solve It [34], Polya gives a four-stage description of the problem-
solving process that is similar in many aspects to the mathematical modelling process (see
Table 2.2).
Many mathematicians will agree this is indeed what they do when solving problems.
However, Schoenfeld explains in his book Mathematical Problem Solving [35] that while
these steps are a description of the problem-solving process, they are not enough to use as
a prescription for how to do mathematical problem-solving. This is due to the fact that
there are not enough details nor instructions for the novice to be able to solve any given
mathematical problem:
1. Typical descriptions of heuristic strategies, for example, examining special
cases, are really labels for categories of closely related strategies. Many heuristic
labels subsume half a dozen detailed strategies or more. Each of these more
precisely dened strategies needs to be fully explicated before it can be used
First. UNDERSTANDING THE PROBLEM
You have to
understand the
problem.
What is the unknown? What are the data? What is the
condition? Is it possible to satisfy the condition? Is the
condition sucient to determine the unknown? Or is it in-
sucient? Or redundant? Or contradictory? Draw a gure.
Introduce suitable notation. Separate the various parts of
the condition. Can you write them down?
Second. DEVISING A PLAN
Find the con-
nection between
the data and the
unknown. You
may be obliged to
consider auxiliary
problems if an
immediate con-
nection cannot
be found. You
should obtain
eventually a plan
of the solution.
Have you seen it before? Or have you seen the same problem
in a slightly dierent form?
Do you know a related problem? Do you know a theorem
that could be useful?
Look at the unknown! And try to think of a familiar prob-
lem having the same or a similar unknown.
Here is a problem related to yours and solved before. Could
you use it? Could you use its result? Could you use its
method? Should you introduce some auxiliary element in
order to make its use possible?
Could you restate the problem? Could you restate it still
dierently? Go back to denitions.
If you cannot solve the proposed problem try to solve rst
some related problem. Could you imagine a more accessible
problem? A more general problem? A more special prob-
lem? An analogous problem? Could you solve a part of the
problem? Keep only a part of the condition, drop the other
part; how far is the unknown then determined, how can it
vary? Could you derive something useful from the data?
Could you think of other data appropriate to determine the
unknown? Could you change the unknown or the data, or
both if necessary, so that the new unknown and the new
data are nearer each other? Did you use all the data? Did
you use the whole condition? Have you taken into account
all essential notions involved in the problem?
Third. CARRYING OUT THE PLAN
Carry out your
plan.
Carrying out your plan of the solution, check each step. Can
you see clearly that the step is correct? Can you prove that
it is correct?
Fourth. LOOKING BACK
Examine the solu-
tion obtained.
Can you check the result? Can you check the argument?
Can you derive the result dierently? Can you see it at a
glance? Can you use the result, or the method, for some
other problem?
Table 2.2: Polyas description of the dierent stages of problem-solving.
reliably by students.
2. The implementation of heuristic strategies is far more complex than at rst
appears. Carrying out a strategy such as exploiting an easier, related problem,
for example, involves six or seven separate major phases, each of which is a
potential cause of diculty. Training in the use of the strategy must involve
training in all those phases, and the training must be given with at least as
much care and attention as is given to standard subject matter. In general
attempts to teach such strategies have not been adequately precise or rigorous.
3. Although heuristic strategies can serve as guides to relatively unfamiliar
domains, they do not replace subject matter knowledge or compensate easily for
its absence. Often the successful implementation of a heuristic strategy depends
heavily on a rm foundation of domain-specic resources. It is unrealistic to
expect too much of these strategies. ( [35], p.73)
This again is reminiscent of the issues already addressed in the discussion of the mathe-
matical modelling literature. Schoenfelds research is focused on doing, understanding and
teaching mathematical problem-solving. He shows how for a particular problem we would
have to break down the given heuristic in order to teach it. The intent of his work is to
develop a prescription as opposed to a description (such as Polyas work), for mathematical
problem-solving.
Schoenfeld addresses the dierences between novice and expert problem-solving: even
though novice problem-solvers have the necessary resources, often they do not know to access
them for a given problem. Schoenfeld explains that this is an issue of control: knowing
what resources to access and when, and that this is integral to being able to be a successful
problem-solver. This control, or ability to understand your own thought processes is referred
to as metacognition, rst dened by Flavell as ones knowledge concerning ones own
cognitive processes or anything related to them. [35, 36]. Schoenfeld also addresses the
fact that expert problem-solvers understand the underlying processes of the problem while
novices do not, as was discussed in the literature on the dierence between novice and expert
modellers. Schoenfeld concludes that:
A huge amount of work lies ahead. We have barely begun to scratch the surface
of any of the categories described in Part 1: resources, heuristics, control and
belief. Even in the best laboratory situations, these are complex and elusive.
( [35], p.375).
So how do the researchers of modelling go on to grasp the elusive aspects of modelling?
The rst step is to recognise that mathematical problem-solving is qualitatively similar to
mathematical creativity. The framework for mathematical invention has been referred to
throughout the literature on mathematical problem-solving. More recently in his work on the
AHA! experience, Liljedahl [33] notes that [m]athematical problem-solving is synonymous
with invention and discovery ( [33], p.15).
2.2.2 Problem Solving: Incorporating the Extra-Logical
In his work The Mathematicians Mind: The Psychology of Invention in the Mathematical
Field [32], Hadamard looks at how mathematicians create mathematics. He uses the works
of Poincare to explain that there are four stages to mathematical invention:
1. Preparation: in this stage the mathematician is using deliberate conscious
eort to come to a solution.
2. Incubation: where the mathematician, if stuck in his/her deliberate eorts,
stops working on the problem
3. Illumination: where the unconscious mind solves the problem and the answer
becomes available to the conscious mind (also referred to as the AHA! experi-
ence)
4. Verication: the mathematician then veries that the solution is correct by
deliberate means.
In later problem-solving literature, there is an extension of Polya and Schoenfelds frame-
works to include the creative aspects of problem-solving informed by Hadamards work. The
creative aspect applied particularly to when the problem-solver was stuck (see Hadamards
step 2 above). This amalgamation of the two frameworks of Polya and Hadamard are
evidenced in Mason et al.s work Thinking Mathematically [28].
Masons framework of modelling begins with the premise that everyone can start by spe-
cialising. This is similar to the initial simplication recommendation in the modelling litera-
ture. Generalising or drawing inferences from the specialised cases follows this step. Mason
also recommends making notes of thoughts feelings and ideas, which ties into metacognition,
the idea of monitoring your own thoughts. In particular, if you are stuck he recommends
acknowledging this and suggests ways to move from the state of being stuck to unstuck.
STUCK!
Whenever you realize that you are stuck, write down STUCK! This will help you
to proceed, by encouraging you to write down why you are stuck. For example:
I do not understand...
I do not know what to do about...
I cannot see how to...
I cannot see why...
AHA!
Whenever an idea comes to you or you think you see something, write it down.
That way you will know later what the idea was. Very often people have a good
idea, but lose it subsequently and cannot recall it. In any case it feels good to
write down AHA! Follow it with:
Try...
Maybe...
But why...
CHECK
Check any calculations or reasoning immediately
Check any insight on some examples (specializing)
Check that your resolution does in fact resolve the original question.
REFLECT
When you have done all that you can or wish to, take time to reect on what
happened. Even if you do not feel that you got very far, it helps to write up
what you have done so that you can return to it freshly and eciently at some
later date. It is also the case that the act of summarizing often releases the
blockage. There are several things worth noting particularly:
Write down key ideas
Write down key moments that stand out in your memory
Consider positively what you can learn from this experience. ( [28], p.19)
Mason considers the reection step the most important activity to carry out ( [28], p.131)
as it helps to develop your metacognition or internal monitor. These last two steps of
Masons framework are similar to the nal step in Polyas model. Mason additionally
acknowledges that the problem-solver will possibly get stuck and require some extra-logical
illumination or AHA! moment as seen in Hadamards model. He also incorporates feelings
in the nal step that is not evidenced in Polyas work.
Perkins work Archimedes Bathtub: The Art and Logic of Problem Solving also deals with
the problem-solving issue of being stuck and how to deal with it. In fact, Perkins explicitly
addresses only problems where you are stuck initially. He calls these problems unreason-
able and they are the problems that require breakthrough thinking or illumination to solve,
that is, you cannot approach them by deliberate logical eort:
The surface pattern of breakthroughs reects not underlying mental processes
but the underlying structure of the problems themselves. To put it roughly,
many problems are reasonable: They can be reasoned out step by step to home
in on the solutions. But certain problems are unreasonable: They do not lend
themselves to step-by-step thinking. One has to sneak up on them. ( [29], p.22)
Perkins deals with four dierent ways that one may feel stuck and the solution for becoming
unstuck. The pertinent solution depends on why you are stuck and includes: brainstorming,
detecting hidden clues by looking in a dierent place or more carefully for incongruous
features, reframing the situation, backing up to an earlier point and taking a dierent path
from there by bracketing o the current approach and trying something else [29].
This range of literature on problem-solving began with a framework very similar to that
currently seen in mathematical modelling, dealing with the cognitive and meta-cognitive
aspects of problem-solving. This can be described as problem-solving by design. Later work
by Mason and Perkins has moved on to look at aspects outside of the cognitive, to those
subconscious processes that help the problem-solver to become unstuck. If mathematical
modelling is synonymous with problem-solving, and hence creativity, then these extra-logical
processes are equally transferable to the modelling issue of being stuck. Even if problem-
solving and mathematical modelling are signicantly dierent, do expert modellers employ
these skills for dealing with being stuck? Or are the experts never stuck and travel along
the modelling cycle with relative ease? Finally, are the extra-logical processes of getting
unstuck the only non-cognitive aspects that hinder novices from progressing?
2.3 Psychology of Modelling
From the look at the modelling and problem-solving literature it is clear that mere cognitive
and logical processes are not enough to make one a successful modeller. If this were the
case then simply studying more mathematics and looking at a ow chart of the cognitive
processes required for modelling would be enough to gain expertise. This has not proven
to be the case as the modelling literature indicates that modelling is dicult to dene,
dicult to teach, and can be overwhelming. This led me to an investigation into possible
psychological issues that could aect the novice modeller.
To delve into some of the psychological issues that may be associated with mathematical
modelling. I examined the works of Tobias, Pink and Csikszentmihalyi who discuss the
themes of anxiety, motivation and ow. These themes are found to interconnect and provide
a backdrop on some of the non-cognitive issues that may arise when modelling.
2.3.1 Math Anxiety
Mathematical modelling has no closed form solution, no set solution framework and requires
tools from many areas of mathematics. These characteristics may lead modelling to trigger
math anxiety, especially among the novice modellers. Tobias in her work Overcoming Math
Anxiety claims that all people endure some mathematics anxiety, but it disables the less
powerful [...] more( [13], p.9).
Tobias explains that math anxiety is the feeling that one can not do a problem and would
never go any further in mathematics( [13], p.50). This rst feeling of failure is instant
and frightening and feels like a sudden death( [13], p.50). Math anxiety is usually caused
initially by some signicant, traumatic event in the math-anxious persons mathematics life.
Tobias [13] tells us that [m]ath-anxious adults can recall with appalling accuracy the exact
wording of a trick question or the day they had to stand at the blackboard alone, even if
these events took place thirty years before( [13], p.38). She explains that math anxiety
creates feelings of paranoia, fear of asking for help, shame, guilt and feelings of fraud.
Math anxiety stems from several issues including the belief that errors are shameful and
the attitude that mathematics ability is inborn. In contrast, positive psychology, focused
on well-being, contentment, and satisfaction (in the past); hope and optimism (for the
future); and ow and happiness (in the present) [37] is associated with mastery. Thus to
identify if math anxiety is an issue in mathematical modelling, there is a need to investigate
the feelings of the expert, intermediate and novice modeller.
There are several issues highlighted in Tobias work that create math anxiety. The rst
cause is ambiguity in mathematics. Tobias notes, Though mathematics is supposed to
have a precise language [...] mathematical terms are never wholly free of the connotations
we bring to words and these layers of meanings may get in the way. The problem is not
that there is something wrong with math; it is that we are not properly initiated into its
vocabulary and rules of grammar ( [13], p.54). Modelling problems have been described
as being messy and coming from the real world [4, 7, 15, 16], which creates a new diculty
with the language of the eld of the problem as well as those language issues raised by
Tobias. Combined with this is the fact that the denition of mathematical modelling itself
is ambiguous, thus we suspect that this aspect of modelling may create anxiety, especially
for students that have been successful with other elds of mathematics.
The second cause of math anxiety is dubbed the dropped stitch, which is the belief
that if we havent learned something so far, it is probably because we cant ( [13], p.62).
This ties in with the belief that understanding and ability is xed [13, 38]. The belief that
ability is xed and the feeling of not being able to do a problem or learn a skill suggests a
low perceived self-ecacy. Bandura denes perceived self-ecacy as peoples beliefs about
their capabilities to produce designated levels of performance [39]. He explains that since
people with low self-ecacy view insucient performance as decient aptitude it does not
require much failure for them to lose faith in their capabilities [39]. Weiner discusses this
same issue, attributions of failure to lack of ability [...] produce low expectancies of future
success (tied to the stability dimension of causality), low self-esteem (linked with the locus
dimension), and humiliation and shame (because these are perceived as uncontrollable) [40].
This lack of belief in self and belief in ones ability to learn has serious ramications in
terms of modelling since modelling covers a broad range of topics and subject areas. Lack
of prociency in an area, where one believes they can no longer learn it, makes modelling
extremely dicult if not impossible.
A third cause of math anxiety pertaining to modelling is a distrust of intuition. Tobias
explains math-anxious people seem to have little or no faith in their own intuition ( [13],
p.62). This is a factor if intuition plays a large role in mathematical modelling, as hinted
by Otto & Day. In this analysis, it is necessary to identify if this is the case.
Tobias also notes that students who expect to nd the right answer can feel anxious
when that right answer does not present itself. Nevertheless she states, This emphasis on
right answers has many psychological benets. It provides a way to do our own evaluation
on the spot and to be judged fairly whether or not the teacher likes us ( [13], p.67). For the
successful mathematics student, nding the exact answer may be a reassurance rather than
a cause of anxiety (which is why they are successful mathematics students). These students
may have the problem of the right formula [becoming] a substitute for thinking ( [13],
p.69) They may therefore feel anxiety when the problem does not yield a clear closed-form
solution, as is the case with mathematical modelling.
2.3.2 Motivation and the Type I Personality
Daniel Pink discusses motivation in his book Drive: The Surprising Truth About What
Motivates Us [41]. He establishes two distinct types of motivation and the people who tend
to each type. The Type X person is extrinsically motivated, concerning themselves less
with the inherent satisfaction of an activity and more with the external rewards to which
that activity leads( [41], p.77). In contrast, the Type I person is intrinsically motivated
and primarily concerns themselves with inherent satisfaction primarily. As well as having
intrinsic and extrinsic motivation, one can attribute success to internal or external factors,
which is what Weiner [42] calls the locus of control, whereby success is either controlled
internally (e.g skills, ability, dedication) or by external aspects (e.g. luck, circumstances,
diculty of the task) [42].
Pink explains that the most successful people are working hard and persisting through
diculties because of their internal desire to control their lives, learn about the world and
accomplish something that endures ( [41], p.79). This is connected to self-ecacy and
Banduras work, as he explains that [p]eople with high assurance in their capabilities
approach dicult tasks as challenges to be mastered rather than as threats to be avoided.
Such an ecacious outlook fosters intrinsic interest and deep engrossment in activities [39].
In other words, the Type I personality is what we expect of experts in modelling. In
examining the dierences between the expert and the novice we must look at motivation
as this plays a role in expertise. Pink further elaborates on three main elements of Type I
behaviour. We will look at all three of these elements in relation to our investigation of the
dierence between the novice and the expert.
The rst element of intrinsically motivated behaviour is autonomy. Pink quotes designer
Sagmeister as saying, Autonomy over what we do is most important. The biggest dierence
between working for other studios and running my own has been the fact that I can choose
what job we take on and what product, service or institution we promote ( [41], p.104).
How does autonomy aect the modeller at all levels of expertise? Do expert mathematical
modellers have autonomy? Can the same be said for their non-expert counterparts, and if
not does this aect their level of expertise? Note here that autonomy is not synonymous
with isolation. Autonomy is not the rugged go-it-alone, rely-on-nobody individualism of
the American cowboy. It means acting with choice which means we can be both autonomous
and happily interdependent with others.( [43]; [41], p.90). As autonomy is an element of
Type I behaviour, then we will identify whether experts exhibit this mix of autonomy and
interdependence.
Mastery is the second element of Type I behaviour. It is dened in this work as the
desire to get better and better at something that matters ( [41], p.111). Dweck describes
dierent types of motivation as adaptive or maladaptive, explaining that [t]he adaptive
(mastery-oriented) pattern is characterized by challenge seeking and high, eective per-
sistence in the face of obstacles [...] In contrast, the maladaptive (helpless) pattern is
characterized by challenge avoidance and low persistence in the face of diculty [44]. Mas-
tery is linked closely with the concept of ow where the relationship between what a person
had to do and what he could do was perfect. The challenge wasnt too easy nor was it too
dicult. It was a notch or two beyond his current abilities, which stretched the body and
mind in a way that made the eort itself the most delicious reward ( [41], p.115,). This
denition of ow comes from Csikszentmihalyi [45] who links ow to creativity. Pink goes on
to state that several companies have realized that creating ow-friendly environments that
help people move toward mastery can increase productivity and satisfaction at work( [41],
p.117). Our study attempts to describe the landscape from the novice modeller to the
expert. It is by denition an investigation of dierent levels of mastery.
Pink explains that mastery is a mindset, and is rarely achieved if you are extrinsically
motivated; is a pain and requires perseverance and determination; and is an asymptote,
and thus for the most part unattainable. Our study will use these features of mastery to
identify whether the experts, and non-experts exhibit these characteristics of mastery. We
also want to identify if ow emerges as a theme among modellers.
The nal element of the Type I behaviour is that of purpose. Pink claims the most
deeply motivated people not to mention those who are most productive and satised
hitch their desires to a cause larger than themselves ( [41], p.133). This statement prompts
us to consider if there is a higher purpose among the expert modellers.
2.3.3 Creativity and Flow
Our look at work from Tobias on anxiety and Pink on motivation address issues that we
expect to aect the novice modeller or help describe the landscape from novice to expert.
Finally we look at Csikszentmihalyis work on creativity, a trait that should primarily reside
with the experts if anywhere. In his work Creativity: Flow and the Psychology of Discovery
and Invention [45] Csikszentmihalyi denes creativity as any idea, act or product that
changes an existing domain, or that transforms an existing domain into a new one ( [45]
p.28). Here it is important that creativity has an impact on the cultural matrix. He speaks
of creativity being a marriage of the individuals novel work, the acceptance of the eld of
experts and the impact on the domain of the creative work. He distinguishes this denition
of creativity by calling it creative with a capital C. The rst question we must address on
this front is whether mathematical modellers are creative in the Csikszentmihalyi sense.
What aspects and personality traits aect creativity? These are numerous and would
require a complete discussion of Csikszentmihalyis work. Instead I highlight a few aspects
that I believe may be pertinent to the mathematical modeller. The rst of these is curiosity.
Csikszentmihalyi establishes that, without a good dose of curiosity, wonder and interest in
what things are like and in how they work, it is dicult to recognize an interesting problem
( [45], p.53). With this in mind I will be looking for evidence of curiosity among participants
of all levels of expertise.
The second aspect of interest to this work is the trait of adaptability. Modelling is varied
and broad and may require the ability to deal with many situations. Csikszentmihalyi cites
adaptability as being one characteristic of the creative personality: Creative individuals are
remarkable for their ability to adapt to almost any situation and to make do with whatever
is at hand to reach their goals ( [45], p.51). Do the successful mathematical modellers
exhibit these traits? Do they adapt well? Do they recognize the resources around them?
Do they make intelligent use of all these resources?
The nal aspect that I looked at was that of ow. This was mentioned earlier while
looking at Pinks work on motivation. Csikszentmihalyi explains that creative personalities
have a multiplicity of traits that are often contradicting, however he states in one respect
they are unanimous: They all love what they do. ( [45], p.107). He goes on to explain that
this love of what they do comes from the experience of ow in their respective domains when
working on a problem. Flow has several essential components to it: clear goals, immediate
feedback, a balance between challenges and skills, a situation where action and awareness
are merged, lack of distraction, no worry of failure, no consciousness of self, no awareness
of time and autotelic activity (the activity is its own reward [41]).
Pink explains in a nutshell why ow is important for any group: When what they must
do exceeds their capabilities, the result is anxiety. When what they must do falls short of
their capabilities, the result is boredom. ( [41], p.119). This brings together all three works
examined for the psychology of mathematical modelling.
Mathematical modelling deals with open-ended problems and is a broad and varied
area. It has the ability to be overwhelming as well as intriguing. This suggests that the
skills required are more varied than the simple ability to do mathematics well. What other
aspects go in to being a good modeller? What motivates the mathematical modeller? How
does the modeller feel when working through dicult problems? Does he/she experience
anxiety when approaching a modelling problem? And if so how is this dealt with?
2.4 Summary
This brief review of modelling literature identies that there are still several gaps in the
modelling literature. For instance, there is no agreed upon denition of mathematical mod-
elling in the literature. Additionally, there is a dichotomy in the approaches for teaching
modelling and it is not clear what the best way to do this should be. Finally, while the
cognitive and metacognitive issues are well-researched, there is less emphasis on those non-
cognitive areas (subconscious, emotional and psychological processes) that are associated
with modelling.
A look at the problem solving literature highlights a shift in focus from a look at cognitive
processes to extra-logical or subconscious processes. In particular, illumination or AHA!
experiences are identied as integral when dealing with being stuck. Although mathematical
modelling is not identical to problem solving there is some overlap, and this shift in focus
in the problem solving literature suggests an investigation into the necessary processes for
dealing with being stuck when doing mathematical modelling.
The dierences along the spectrum from novice to expert are not well-researched at
the non-cognitive level. However, an examination of some of the cognitive, developmental,
and motivational literature suggests that non-cognitive issues (particularly emotional and
psychological) have a role to play in developing cognitive ability. Anxiety, self-ecacy,
motivation and ow all are related to success and mastery. This combined with the above-
mentioned issues raises several questions of interest. In the following chapter, I will discuss
some of those pertinent questions and describe how I went about trying to seek answers to
them.
Chapter 3
Methodology
Having described the main literature on mathematical modelling to give a glimpse of how
modelling is viewed in the applied mathematics community, it is clear that there are several
questions outstanding.
3.1 Research Questions
Research Question 1
The mathematical modelling textbooks have highlighted that denitions of mathematical
modelling are varied and in some cases non-existent [47, 1517]. Approaches to exposition
on the subject are dichotomous, split between case studies and focusing on teaching solution
techniques. This leads is to our rst research question:
What IS mathematical modelling? What do the experts mean when they talk
about modelling, and is it the same as what the non-experts mean?
This has implications for doing and teaching modelling. If we have a working denition of
what were talking about when using the term modelling, then we can better understand
and teach it. It also helps us to communicate with others who use the term if we know the
dierent denitions of modelling they could be applying.
Research Question 2
An investigation of the problem solving literature highlights extra-logical mechanisms for
getting unstuck. Modelling experts in the eld may use these subconscious processes as
28
CHAPTER 3. METHODOLOGY 29
well, though they are not evidenced in the modelling literature. This leads us to the second
research question:
How do expert modellers move from a state of being stuck to becoming unstuck?
Are extra-logical processes used, or do expert modellers have other strategies for
doing so?
The answer to this question will aid in doing mathematical modelling. The novice modeller
may be able to employ the techniques used by the experts when they are stuck in one stage
of the modelling process.
Research Question 3
Crouch and Haines work [18] indicate cognitive and meta-cognitive dierences between
novices and experts but do nothing to address the non-cognitive issues, particularly how
to get unstuck and what psychological areas aect modelling. A look at the math anxiety
literature indicates a need to probe into the feelings, attitudes and beliefs of the participants
of the modelling study. In this way we can identify how math anxiety hinders the ability to do
modelling well. The literature on motivation highlights that motivational factors are directly
correlated to expertise. In the study we hope to see evidence of intrinsic versus extrinsic
motivation being an indicator of expertise. Qualities of autonomy, mastery and purpose
will aid in identifying intrinsic motivation in the participants. Finally a look at creativity
suggests that these psychological factors may be contributing factors to the dierences
between novices and experts. This leads us to our nal research question:
What are the dierences between the novice and the expert modeller? Particularly
what are the non-cognitive dierences?
This nal question has implications for the teaching of mathematical modelling. Many of
the known dierences between the expert and the novice modeller cannot be taught and
need time to be developed. Nonetheless it is possible that some dierences revealed could
be used to help the novice move along the spectrum towards becoming an expert.
3.2 Overview of the Study
In response to these questions, I tailored this study in order to target dierent groups of
expertise, and answer the three research questions that have emerged from this look at the
literature. In order to answer the questions addressed above, I conducted a qualitative study
with 78 people who do modelling at some level. A qualitative as opposed to quantitative
approach was used as I was trying to establish a fundamental understanding of mathematical
modelling and the people who partake in it. The point of this work is not to create a formula
for doing mathematical modelling, or to establish that 400 people believe modelling is one
particular thing. The aim of the project is to understand dierent variations in modelling
and to understand the nuances in the modelling world. This requires a qualitative approach.
The literature on problem solving informed my tools of data gathering. In his study to
understand novice problem solvers Schoenfeld [35] does a clinical study with his participants.
However Liljedahls [33] work informs us that a clinical trial does not always help us to
identify the internal aspects of the problem solving process. I felt that the internal aspects
were worthy of study, which meant that a clinical trial would not make sense here. Hadamard
[32] and Liljedahl both make use of questionnaires, while Schoenfeld and Liljedahl both use
observations to inform their studies. Therefore, observations, interviews and questionnaires
were my tools of data gathering.
In an eort to answer the research question regarding the dierences between the expert
and the novice modeller, I needed to understand the modelling process from the point of
view of the expert, the novice and those who are in-between. How do we identify the
dierent groups and nd out what modelling entails for them? In order to do this I needed
to determine who would qualify as an expert or a novice in the eld of modelling. This led
me to look at the Dreyfus model of expertise [46, 47] for a description of the mental skills
expected as expertise increases.
In the 1970s the Dreyfus brothers [46] looked at highly skilled experts from pilots to
chess grand masters and did seminal research on how people attain and master skills. They
found that one fundamentally changes how one perceives the world beyond just gaining
new skills or knowledge. They developed a theory which has been applied to diverse elds
including nursing, education, computer programming, sport and others. In particular, they
developed ve categories dened by universal characteristics:
Novice: has no skill; needs rules; has no self-condence; not interested in learning
just completing; progresses by relying only on rules.
Advanced Beginner: can start tasks on their own; little self-reection; can be
mentored; has no idea of big picture or overall principles; can learn via guided
experience.
Competent: can develop novel conceptual models; troubleshoots: Is this right?;
work based on experience and planning; is often overwhelmed by the choices
available; can learn by doing.
Practitioner: Needs the big picture; self-reects: Can I do it better the next
time?; has sucient experience to learn from discussion and case studies; can
learn by seeing and listening.
Expert: Intuits solutions; may have diculty explaining what he/she does as it
has become automatic.
These characteristics helped to identify which participants will be targeted to ll the role of
expert and novice and those between. While the expert category and novice category are
obvious from the outset (one group has had signicant experience and the other has had
none), traversing the other three groups is a little more subtle and we expect to see some
overlap in abilities from one level to the other. To place a person in any of the given middle
categories may require a prior assessment of their work, which is not feasible.
I therefore split the participants of the study into four major groups: expert, interme-
diate, novice and complete novice, each of which represents a dierent level of expertise.
In this chapter I will discuss each group explaining who they are, how they were accessed,
what data was acquired and how. I will then give an overview of how the data was analysed,
before going on to discuss the results of this study in the following chapters.
3.3 Group 1: Expert Modellers
The rst group that I chose to work with were experts in the eld. This was informed
by Hadamards and Liljedahls work [32, 33]. From the Dreyfus model discussed above, we
know that experts intuit solutions and much of what they do has become automatic [46].
Therefore the obvious choice of experts was professors of modelling. These professors must
be active mathematical modellers, preferably being well recognised in the eld, as this is
an indicator of their expertise. If they also taught mathematical modelling this would be
an added bonus, as they would have insight into their novice counterparts. The experts
that participated came from prestigious universities in Canada, the United States and the
United Kingdom, including UCLA, Oxford and Duke.
Several of these experts mention and regularly take part in annual modelling study
groups hosted by many dierent universities and societies. Preceding the study group there
is usually a camp for the graduate students where a mentor leads the students through
modelling problems. This is followed by a week-long study group in which faculty and stu-
dents work together. The format for these events is as follows: before the event organizers
work with members of industry to nd suitable problems, on the rst day the industrialists
present their problems, participants then choose their favourite problem and work in teams,
often parcelling out the work to have it run more eciently. While everyone in the group
is entitled to opinions and encouraged to voice them, all opinions must be defended math-
ematically. At the end of the week some analytical or numerical mathematical models and
solutions are presented to everyone attending, in answer to the given industrial problem.
I contacted my rst experts by attending the Graduate Student Mathematical Modelling
Camp (GSMMC), hosted by the Rensselaer Polytechnic Institute. I worked on one of the
modelling problems and I was able to see rst hand how the process worked. Here professors
mentored graduate students in the art of modelling over the course of a week, oering
support (or not) as needed throughout the camp. These mentors were my rst targets for
the study. I acquired interviews from the four mentors that assisted the students in the
GSMMC.
I then emailed a list of 13 experts I received from my supervisor who also works in the
eld of modelling. Of the 13 people emailed, 7 agreed to work with me on the study. These
7, along with the initial 4 from the GSMMC brought my total experts up to 11. Along with
these I was also able to interview 3 experts who visited SFU to give presentations on their
areas of modelling expertise. This brought my total up to 14. At this point I noticed no
new data emerging so I stopped interviewing experts.
In the case of the experts the data consisted of the responses to ten interview questions.
Interview questions seemed appropriate as they gave some exibility in being able to ask
follow-up questions. This is pertinent as Dreyfus & Dreyfus warn us that experts may
have diculty explaining what they do as it has become automatic [46]. The interview
questions were informed by Hadamards survey [32] but were adapted to address my own
research questions. The interview addresses dierent aspects of the modelling process, and
questions were ordered to mimic the order of the steps in the modelling process. I therefore
started with questions on denitions and rst thoughts upon seeing a problem and worked
through dierent areas of modelling up to and including checking the model. For this
reason the questions were always asked in the same order, but follow-up questions were
included if necessary to glean further information from the responses. Question eight was
adapted slightly as indicated by feedback from the initial interviewees. The responses to
these questionnaires will be discussed and analysed in Chapter 4.
3.4 Group 2: Intermediate Modellers
The Dreyfus model describes three distinct levels between the expert and the novice [46],
ranging from advanced beginner to competent. While the characteristics of the expert and
novice were extreme and therefore easy to identify, characteristics of the intermediate levels
were more subtle and not always easily identiable from the outset. I therefore diverged
from the Dreyfus model and dubbed all participants intermediates if they were neither
expert nor novice. The intermediate participants were made up of 11 of SFUs applied
mathematics graduate students and post-doctoral fellows. These participants had varied
backgrounds coming to SFU from six dierent countries including, Canada, the United
States, and China. They were also varied in their applied mathematics interests.
I requested interviews with the intermediates in person, as I had direct access to them.
Many of these students I had worked with in a voluntary, weekly problem-solving session
held by and for the applied mathematics graduates. Oftentimes after an intermediate was
interviewed they would suggest another possible participant for the study whose work they
were more familiar with than I. While several of my colleagues were willing to help a few
declined, explaining that they had not done any mathematical modelling. (It is unlikely
that they had no modelling experience at all, which makes this an interesting response. I
assume that their denition of modelling was not the same as mine, since we have seen that
modelling has varying denitions.) As several of these intermediates took part in the weekly
problem-solving session they had this as a common background.
This problem solving session occurred once a week, usually on Friday evenings, for the
applied mathematics graduate students at SFU. It was voluntary, but students were expected
to participate if possible. A mentor provided a problem (and sometimes the students did
as well), which was oftentimes a modelling problem that the group of graduate students
then attempted to solve. There was no set time limit, with some problems spanning several
weeks. While the session was mentored, it was not as structured as the modelling camps
described in the previous section. Students were expected to self-organise and decide what
they worked on. This often led to several groups working at dierent paces, on the same
aspects of the problem. This common experience gave me something to fall back on if the
intermediates interviewed needed examples to help prompt their responses.
Once again the data consists of the responses to the interview questions. Initially I tried
to include graduate students that participated in the GSMMC (see above Section 3.3) with
the use of questionnaires to gather data. This resulted in varying qualities of response, due
to the range of expertise in these graduate students. I realised that although intermediates
did not necessarily have the wealth of experience of the experts, I still needed to interview
them to get more in-depth responses about their various modelling experiences. I therefore
used the same interview questions used with the experts in the eld to gather data from
this group, which would allow me to compare and contrast responses from the two groups.
Question one was modied to get an idea of their modelling background, and for half of the
intermediate participants a list of heuristics was given to aid in answering question seven.
We will discuss the implications of this, as well as an analysis of all responses in Chapter 5.
3.5 Group 3: Complete Novices
To complete the spectrum from novice to expert I rst chose a set of participants who were
clear novices on the Dreyfus scale: having no skill, needing rules, lacking self-condence,
interested in completing as opposed to learning, and progressing by relying only on rules.
These students came from two separate FAN X99 classes held at SFU. The FAN X99 class
is a foundations of numeracy course. In other words, these participants had issues with all
aspects of mathematics, not only mathematical modelling, with many of them not doing high
school mathematics past grade 10. For these students, mathematical modelling consisted
of solving word problems. This is the most basic of modelling problems and is what Briggs
calls modelling or story problems [16]. The work I did with these complete novices was
inuenced by the work of Schoenfeld and Liljedahl, who both observed and worked with
novices in classroom settings.
Forty-ve students from the two classes agreed to complete questionnaires for the study.
These students had a wide variety of barriers to mathematics, but one common theme
among the majority was a fear or dislike of mathematics. This made them an ideal contrast
to experts when addressing the aective components of modelling.
The students were accommodating and many of them were willing to discuss their
thought processes with me to help me see how they went about solving their word problems.
This allowed me to develop a relationship with them, and also gave me a rst hand view of
the issues and techniques unique to this level of expertise.
The data for this group consisted of responses to questionnaires distributed at the end
of the semester. Students were again reminded that they were under no obligation to
participate. I chose questionnaires instead of interviews here because I was asking them to
relate a specic experience. This was in contrast with the experts and intermediates where
there was the need to probe into varied experiences which a questionnaire might not cover.
The use of questionnaires yielded a large number of responses (45 students responded).
The questionnaire contained some questions that paralleled the expert and intermediate
interview. Other questions were included to establish novice students preconceptions. We
will discuss and analyse these responses in Chapter 7.
3.6 Group 4: Novice Modellers
Although the FAN X99 students qualify as complete novices, they are not expected to
progress along the spectrum to modelling expertise in this course, as modelling is not the
primary focus of the course. (This is not to say that the data from the FAN X99 students
was useless, far from it, as we see in Chapter 7.) This led me to SFUs fourth year under-
graduate modelling course Math 461, in which students are encouraged to participate in the
Mathematics Contest in Modelling (MCM). Although this was a fourth year course, many
of the students had very little or no modelling experience. This made them an appropriate
group of novice modellers, as their lack of experience qualied them as novices, but their
decision to take the class and participate in the MCM showed a likelihood of moving along
the spectrum towards expertise in modelling.
Those who were participating in the MCM were asked via email to ll out a questionnaire
about this experience. The contest is a weekend long modelling competition, in which
students work together in groups of three on a given problem. Students have a choice of
two problems to work on and are not allowed to consult anyone outside of their group for
help. Eight of the MCM participants agreed to complete the questionnaire (after a second
reminder a month after the MCM), with one of them not being a member of the Math 461
class.
The data for this group of novice modellers are their responses to the questionnaires.
This questionnaire was dierent from that for the acute novice. Students were asked to
comment on what made their chosen problem a modelling problem, as well as being asked
several questions that paralleled the interview of experts and intermediates. Due to the
lengthy turnaround in responses some students explained that they could not remember
clearly all the details of the MCM. The responses collected will be discussed further in
Chapter 6.
3.7 Analysis
The data in this study are the interview and questionnaire responses. While no discourse
analysis was done, pauses and exclamations that highlight a particular point were included.
The observations were used as a backdrop to the analysis of the transcribed and question-
naire data. As the data is primarily spoken words, it was edited to allow the reader to follow
the train of thought of the speaker.
After the expert and intermediate interview data was recorded and transcribed, I trans-
ferred the data to an excel spreadsheet. This allowed comparison by question as well as
by person. Individual responses to each question were then coded using line-by-line coding
informed by Grounded Theory [48]. To do so I looked at each line in a given response to
a question, and summarized it. I then looked at these lined summaries and identied any
themes of interest within them. I noted recurring themes as well as outliers and compared
the results across groups, contrasting the responses of the intermediates with those of the
experts. In many cases new codes had to be created specically for the intermediates.
While the coding was done using the principles found in Grounded Theory, the analysis
of the themes was not. Charmaz [48] describes Grounded Theory as developing theory as it
emerges from the data, thus the theory comes from the data as opposed to the data being
analysed using existing theories. In place of this method of analysis I made use of Pattons
[49] principle of analytic induction. Sriraman [50] explains that Pattons principle works
well when studying an extremely complex construct involving a wide range of interacting
behaviours. Since the literature had motivated my study, common themes that emerged
were compared to the existing literature using Pattons principles, as opposed to developing
a theory from the ground up as Grounded Theory suggests. This is because analytic
induction, in contrast to grounded theory, begins with an analysts deduced propositions or
theory-derived hypothesis and is a procedure for verifying theories and propositions based
on qualitative data. (Taylor and Bogdan, 1984, p.127) [50].
Questionnaire responses for both groups of novices were also transferred to spreadsheets
to allow comparison by question and by person. These responses were succinct and did
not require coding. Responses of the MCM student-novices were compared and contrasted
with those of the experts and intermediates. For the FAN students I focused more on their
preconceptions of mathematics. However I contrasted them with the experts to highlight
the dierences here, and in some cases, the similarities. In the following chapters we look
at the detailed analysis of responses for each group.
Chapter 4
Experts
The experts were the rst group of participants to be interviewed. These were professors,
all known in the eld of mathematical modelling, and many of them currently teaching
modelling courses. Many of these professors work at the some of most prestigious universi-
ties in the world including Oxford and UCLA. My primary objective in interviewing these
individuals was to understand how experts view modelling, specically trying to understand
their denition of modelling, how they feel when they approach a modelling problem and
their tendencies when becoming stuck while working on a model.
As explained in Section 3.3 many of these experts participate in week-long study groups
hosted by many dierent universities and societies. While in these study groups, the experts
worked with those from industry, colleagues and students in order to nd solutions to
industrial problems. These study groups were referenced in several responses to the interview
questions.
In this chapter I will discuss the results of the interviews of the experts, presenting a
few illustrative quotes highlighting the main themes raised by the experts. Note that the
names Bob, Ted, Tony and Mike refer throughout this study to experts who chose to remain
anonymous.
38
CHAPTER 4. EXPERTS 39
4.1 Dening Mathematical Modelling
Question 1: How do you dene Mathematical Modelling?
The denition of mathematical modelling is not a clear-cut entity. As expected, among the
experts in the eld we note several variations in their denitions; however, three common
themes of interest emerged. The experts generally agree that mathematical modelling is a
description of a real life situation using a mathematical framework:
Description of a physical problem, in terms of mathematical equations. (Brian
Wetton)
Something that takes things in the real world and describes some sort of math-
ematics to them. A precise mathematical problem. (Andrea Bertozzi)
It is essential to clearly state your assumptions and only once that is accom-
plished can one move forward to build a model that attempts to capture the
essence of a physical process. Were usually dealing with a physical problem,
but mathematical modelling is broader than this. Industrial mathematics focuses
on that branch of modelling that is primarily concerned with problems of either
a societal or commercial benet. One of the largest challenges in this eld is the
clarication of the problem. Typically researchers are presented with irrelevant
information and understanding the essence of a process is more valued than its
exact quantication. Abstracting the problem allows one to make connection be-
tween physical problems that on the surface seem completely unrelated but using
mathematics peels back this obscurity and allows for cross pollination of ideas.
This deepens the understanding and allows for innovation. (Sean Bohun)
For Sean Bohun understanding the problem is an important aspect of modelling. This theme
shall be addressed several more times throughout this chapter and beyond. For three other
experts, it was highlighted that the model would be a simplied or approximate version of
the physical system:
A process by which people take a situation in the real world and put it into the
simplest possible mathematics (at least at rst) that describes the process. And
by describes I mean captures the eects or experience. This is dierent from
analysis or verication. (Bob)
Its taking a real world problem coming out of some other scientic area and
trying to formulate mathematical equations to describe it to some level of ap-
proximation or other. (Thomas Witelski)
Four experts expand from here, expressing modelling as a process, thus solving, analyzing
and verifying the model are also parts of the denition. These experts then go on to talk
about rening the model for more accurate results or using the model to make predictions:
Mathematical modelling begins with a problem in the real world, which has prin-
ciples, structure and cause that we have an understanding of. Mathematical
modelling takes us from the physical world to a model via formulation, math-
ematization and idealization. If we could get a perfect model it would contain
every detail, but there is no perfect model. Instead we may get a hierarchy of
models where we take out more and more details for the sake of simplication.
Each model is self-contained. The model then goes through the process of solu-
tion/analysis/computation and we assess it to see if we identify model behaviours
that are analogous to real world behaviours. If not we need to go back to our model
and rene it. This process is what mathematical modelling is. (David Muraki)
This creates an interesting dichotomy on what mathematical modelling is. Is it exclusively
the setting up of the model that denes it or do the solution and use of the model need to
be considered as well? This dichotomy may aect the teaching of mathematical modelling:
one group might mainly focus on setting up the model from scratch, while another group
might focus on continuing past the initial set up to solving the model. Being able to set
up the model is considered more an art than a science, and in some ways more dicult to
teach as will be discussed later (see Sections 4.4 and 5.4).
A third and completely dierent response given by one expert is that mathematical
modelling does not need a denition, as it is no dierent from doing any type of problem
solving:
I dont think it needs a denition really. People just pretend its something
which is dierent. I dont really think its any dierent to anybody who works in
any particular subject you know? I mean you just do it. Everyone does it if they
have a problem. (Colin Atkinson)
Again we would expect that this view of modelling would have an impact on how it is
taught. Is modelling exactly the same as problem solving? I believe there are important
distinctions in the context of tertiary education, which will emerge from this study, and will
be discussed in subsequent chapters.
4.2 Initial Thoughts, Plans and Feelings
Question 2: Upon seeing a problem, what are your rst thoughts, feelings or
plans?
Feelings:
One of the things that is often debilitating to novices are the overwhelming negative feelings
associated with working on problems [5, 13]. For this reason, I was interested in knowing
what feelings the experts experience when rst faced with a modelling problem. The rst
point of interest is that almost all of the experts had to be prompted twice to answer this
part of the question. They were forthcoming with their initial thoughts and plans, but had
to be reminded about the fact that the question asked for feelings as well.
After prompting, several of the experts spoke of positive feelings: excitement, interest
and curiosity:
Ooh whats going on there? or I have no interest in this whatsoever (Bob)
Depends on the problem. Is it a good problem? Is it interesting? Do I think
the solution will make me a better person? Or the world a better place? Is it
a problem I can solve? (There are lots of problems that I like but I cant solve,
because Im not equipped to solve them.) (Lou Rossi)
Interviewer: Youve touched on when you see a problem what your rst thoughts
and plans would be but you havent said what your rst feelings would be when
you see a problem.
Brian Wetton: Well, feelings! (chuckles) well. So I dont know if Im answering
your question but there are certainly some things that get me more excited than
others. Theres all kinds of mathematically interesting questions for example in
math nance, but nance just doesnt do it for me. Ive worked in modelling
hydrogen fuel cells, so that kind of physics, its especially interesting because
its multiple phenomena that are coupled together in a complex way, thats the
kind of thing that gets me excited. So if I see something thats really interesting
my feeling is one of excitement trying to see how you would describe that in
mathematical terms. (Brian Wetton)
These experts illustrate the desire for the problem to be of interest to them. They are not
interested in all problems, and do not work on those they do not like. Lou Rossi further
highlights the need for the problem to provide him with purpose, supporting the work of
Pink [41]. These feelings of interest and excitement when modelling are expected, as their
expertise in the eld is correlated to positive feelings towards their eld of choice [45]. Even
more interesting, is that seven of the experts interviewed spoke of negative feelings: feeling
daunted, fear, panic, and wondering if they could solve the problem. Indeed there was some
overlap and three of them spoke of feeling worried and curious.
Oh my rst feelings! I usually feel very daunted, doing some thing new. Yes
its usually something new for me and I usually feel like: wow I have a whole lot
that I need to learn here. But I also feel excited I should say because its always
exciting to tackle a new problem. (Tony)
Feelings, its kind of a joke but, I hate this problem for being hard. I hate
myself for not being smart enough to know immediately how to write it down,
or some variation. (Thomas Witelski)
Well, feelings! That doesnt help! How am I going understand this? I have no
idea what the words mean. I dont know anything about this. (Sam Howison)
Three experts of these seven went on to explain that they were able to ignore, deal with or
suppress those negative emotions.
Ive learnt to suppress research anxiety years ago. (David Muraki)
I think youre worried and curious yeah. But you have to deal with it. You
have to begin things thats how you progress (Colin Atkinson)
Not knowing how to do it is a motivator. Im fearless. I dont mind not knowing
whats going on and Im willing to argue until understanding is agreed. There
are two types of students those who want to apply those things they know or those
who want to learn something new. (Sean Bohun)
The fact that they all rst ignored their negative feelings or recognized them but were able
to push past them is consistent with many of the ndings in psychology, both in positive
psychology [37] and work on mindset [38] where resilience is identied as a key to mastery.
Thoughts and Plans:
All but two of the experts explain that their rst plans were to explore the problem and do
research or gather data.
Explore it and see where it takes me. (Burt Tilley)
Oh you mean when I rst see a problem that I hope to model? My rst thought
is that before I even think about mathematics that I should try to learn as much
about the particular problem as I possibly can. If its a problem in biology I
try to read biology papers and talk to biologists or anyone else whos working
on the problem and really just nd out about the biology in the problem. And
I think thats crucial in order to be able to accurately narrow it down into a
mathematical description. So thats the rst thing I think, cause I dont even
think about math, I just think I had better learn about the background. (Tony)
Just learn about a subject. You have a problem you want to learn about it
right? Learn about it, try to make sensible decisions about it and if youre a
mathematician you apply the tools of your trade to it. (Colin Atkinson)
Additionally three other key themes were discussed: understanding the dominant process,
simplifying the problem and collaboration.
Understanding the dominant process:
Depending on the specics of the problem, the path to understanding is to rst
identify the broad strokes of what is happening, i.e., the dominant process. To
this end, are there similar problems I have worked on? Do I need to review
the underlying theory? Is a preliminary literature search necessary? These are
always questions that I ask myself. Occasionally I will choose a problem at a
workshop simply because I know the least about this material and I need to learn
more. The unforeseen benet is that by doing this you add a fresh unbiased
viewpoint. (Sean Bohun)
Mostly when I rst see a problem I try to think: what area of science do I need
to go to gure out how to write it down mathematically? Is it a physics problem?
Then I have Newtons laws. If its a chemistry reaction problem then I use the
chemical reaction kinetics equations, mass action. If its from some other area
can I use geometry? Or how can I break it down into simplest pieces that make
up the problem. (Thomas Witelski)
Simplifying the problem:
I suppose my rst thought would be: whats the simplest example of whatever
it is that someones showing me. If it looks really complicated, is there an easier
way to look at it, or, not even an easier way to look at it, is there an easier
problem? Maybe I can start with that one. The easier the better. (Mike)
Collaboration:
If you want to take modelling seriously, if you want to be able to come up with
a model in a given situation, then [...] you do need a repertoire of things, which
[...] you acquire with a bit of experience. Im a big advocate of going to meetings
like the [...] PIMS workshop and the Canadian study group meeting. Where
they have these open problems from industry [...]. A load of people from industry
come with open problems that they would like to see solved. And you have no
idea what youre going to get. And then a bunch of mathematicians sit around
and spend a week [working on them] (Sam Howison)
Well, you usually start with a discussion with someone whos an expert in the
application eld. Its usually in that discussion where they highlight what they
at least believe are the important phenomena and in that discussion you (well I
think what you should do, but which sometimes means the problem is not math-
ematically interesting) try and think of the simplest mathematical description of
what theyre saying. What, the simplest that it can possibly be [...] I think one
of the useful things a modeller can contribute to a scientic project is asking
the questions: is this important? What does this do? Because it really forces
the application people to think about really what is going on. They tend to get
sometimes sidetracked with particular descriptions of the problem and you can
come in and say,Well we could include that additional eect that you always
neglect, its not hard and maybe its important. (Brian Wetton)
These above-mentioned themes connect strongly with Polyas work on problem solving. For
Polya, step 1 is Understanding the Problem, which we see re-iterated by Lou Rossi:
I feel better if at the end I understand the problem better. I want to understand
it in a better way. (Lou Rossi)
The vast majority of experts interviewed identied understanding the problem as their rst
step, whether by research and exploration, discussion or simplication. The key observation
here is that all the experts begin with the assumption that they do not even understand the
problem. This, as we will see, is in direct contrast with the complete novices who are not
even aware that they do not understand the question asked, implying poor metacognition
[35, 51]. Lacking awareness of ones own inadequate understanding demonstrates the so-
called Dunning-Kruger eect where people [who] are incompetent in the strategies they
adopt to achieve success and satisfaction, [...] suer a dual burden: Not only do they reach
erroneous conclusions and make unfortunate choices, but their incompetence robs them of
the ability to realize it. [51].
4.3 Dealing with Initial Barriers
Question 3: If you have no idea how to start what do you do?
The most prominent theme to emerge was trying to better understand the problem before
starting. Eight of the fourteen experts do this by discussing or collaborating with others:
Talk to other people. Mathematical modelling is not a solitary activity. Ask
someone who you think knows [about it]. Read a [research] paper. (Bob)
Talk to other people. They may not know how to do it, or be able to tell you.
But just knocking around ideas. Its a reective process. (Lou Rossi)
If its a person Im talking to then I probably try to get more information from
them. If I dont know where to start then I might ask them, Is there a simpler
problem thats related? I dont really know how to do this one right away but,
what if there wasnt an x
2
term? Or what if there wasnt diusion? If Im
having a conversation then I would try to say, What can we do to make this
problem look like something that I might know how to begin? (Mike)
We notice here that in order to understand the problem better Mike raises the theme of
simplication. This may be through a discussion on how to simplify, but others speak of sim-
plication as a means of getting past being stuck initially, without necessarily collaborating
with others:
Try to solve a related problem, a simpler related problem is better. Solve the
trivial problems in dierent regimes. Try to sneak up on the problem. (Sean
Bohun)
[7 second pause] I try to take a small part, a really small, it could be just a
microscopic part of the problem. So usually a problem comes with a goal: the
person proposing the problem says, I want to nd the answer to this thing about
the system. And so if that seems overwhelmingly dicult then Ill start ignoring
his question and just try to answer something I think I can. And then maybe
using that as a rst step for getting to something more dicult. But its usually
just, rst build up your own condence that you understand whats happening
with at least parts of the problem. (Thomas Witelski)
[5 second pause] Not recently actually. Because really you should start with the
simplest thing and thats usually pretty straightforward. Now its certainly true
that that doesnt always work. So usually you get stumped later on when the
simplest thing doesnt work, doesnt describe whats going on. Theres something
more complicated because what youre doing just doesnt match the experimental
data thats in the literature. (Brian Wetton)
Seven of the responses indicated that research and looking at others previous work was
a method for getting unstuck initially. This again ties directly to Polyas rst heuristic
Understanding the Problem:
Try to see if someone else had done it. Collaboration. Make sure you have an
understanding of the resources at your disposal. This comes from experience and
practice (Burt Tilley)
If I have to solve it and I dont know where to start I look around and see if
someone else has solved it. (Lou Rossi)
Then you have to study more [laughs] (Ted)
Burt Tilleys comment about understanding resources points to Csikszentmihalyis state-
ment that creative individuals make do with whatever resources are at hand to reach their
goals [45]. Apart from this two experts advise taking a break or waiting: Wait it out
(David Muraki) or sleep over it (Reinhard Illner). Three others experts advise the op-
posite course of action, just starting: You have to start somewhere (Colin Atkinson),
just start writing (Lou Rossi) and youve got to start trying things (Sam Howison).
Sam Howison goes on to give a full description on how he would solve a problem he knows
nothing about:
So the problem might be something to do with data packets in a mobile phone
network, which is something I do not claim to know something about at all.
Though you could start saying, Where do they go and what do I need to know in
order to describe the system? Id need to know how many packets there were at
this place and how many there were at that place. Did I need to know how many
there were in between? Then Id ask the guy proposing the problem and see if
the guy says, Yes, you do need to know, or No, you dont need to know because
they go at the speed of light, and that kind of thing. And then what are the rules
whereby they get moved around from one place to another? Say someone takes
a look and says, No, that ones full, send it somewhere else, start writing down
little equations that say how these things are conserved, if they are conserved.
They may disappear. But assuming they are conserved (which I guess probably is
one of the aims of mobile phone people) you add up and you balance equations.
What goes in must come out again, or it must stay in the system, so you do all
of that kind of thing. You write down a great big discrete model or continuous
model if youre dealing with a continuous system. If youre going to do that
its very useful to have basic knowledge of the dierence between things like a
conservation law that says that mass or momentum or energy or something of
that sort is conserved, and a constitutive law which says how does a particular
material relate to types of stress or strain or temperature or thermal energy,
temperature gradient and heat ux all of those things. So that you have a good
idea what is allowed in any case and what you need to know in addition specic
to the system youre talking about. (Sam Howison)
One strikingly dierent response was the expert who explained:
Im not usually in that situation, usually Im working on a problem because I
have some ideas about it. Im not usually in a situation where somebody says,
You have to do this, because I wouldnt do that. I work on something where I
have some ideas. (Andrea Bertozzi)
This statement fully correlates with David Murakis views on why modelling is more stressful
for the student than the expert:
Of course you can imagine the situation where someone says, We know David
Muraki is an expert in this eld and weve been waiting for him to comment.
Now someone has set the bar for me. For students its implied that the per-
son asking has expectations about what they should know, so this is very stress-
inducing. Additionally theres a lack of condence novices have that they can
make intelligent responses, and this is near panic-inducing. (David Muraki)
Finally Colin Atkinson brings up an important point in his discussion below about why you
get stuck at the beginning of a problem.
The hard part is formulating it really. There are two aspects one is formulation
the other is solution. (Colin Atkinson)
Once again we see that modelling can be split into two steps as discussed in question 1
(Section 4.1), but here Colin Atkinson brings our attention to the fact that in starting a
mathematical modelling problem the hardest part is formulating it. This step is not at all
straightforward and leads us to our next question.
4.4 Determining Relevance
Question 4: How do you determine what information is relevant to the model?
For me this was a question to which the answer was not immediately apparent. In my
personal experience with modelling, trying to capture the entire phenomenon exactly was
very dicult, would often create unsolvable equations, and I was unsure how one decides
what information is relevant to the model. This diculty was echoed by Sam Howison and
Mike who refer to this process as being an art.
Of course, this question assumes that not all of the information available is necessarily
relevant to the model. We are assuming that we should try to include the fewest possible
phenomena that will describe it (Bob). Six other responses similarly address using the
simplest model possible, reinforcing the theme that was raised in question 3 above (Section
4.3):
If I were to say what were the precepts in modelling, one of them is: always
do the simplest problem you can rst, always solve the easiest, simplest problem
rst. (Sam Howison)
There are some simplied facts that are central to the model, so the crucial
modelling step is to be able to distil from the series of observations those relevant
central steps and throw away the things that you can add on later thats icing
on the cake. But thats dicult, and so the way we normally [proceed] now is
not to try to explain everything, every detail, but to have the simplest model that
explains the basic operation of what you observe. Its understood that a lot of
features that you wont be able to explain you need to ignore them for a while.
(Ted)
Because the type of modelling that I do, my broad interest is really in pattern
formation, Im really interested in understanding basic mechanisms of pattern
formation. So the models I tend to build are really quite minimal. For me the best
model is the simplest one which is real, which is reasonable and which reproduces
the phenomena of interest. So I will always start with the most simple models I
can dream of, before going to something more complicated. (Tony)
Notice that the simplied model does not need to replicate all results exactly or try to
explain everything. It is enough that the simple model reproduces the phenomena of
interest or captures trends qualitatively. But how exactly do we go about nding this
simplest model? Seven experts explained that the data or physical properties of the problem
help to determine relevance:
To build a model we need to mathematize the real world principles and structure.
Hence, we need to understand the world and have the appropriate mathematical
vocabulary. You have to know the reality youre talking about to talk about it
mathematically. (David Muraki)
Again this depends on the problem. Usually it is clear from the outset. Usually
the phenomenon is explained to you and they want an explanation. I try to
embed the problem into language and concepts I know, (e.g.: kinectic theory).
(Reinhard Illner)
Ok. How do I determine what information is relevant? Well I think modelling
is always a process of iterative renement. So if I take a few iterations I really
understand what is relevant. So I guess I create a model which gives some results.
We try to think about those and compare them to whatever data might be known
about the system that were trying to describe. And then the model might well
need to be changed. (Tony)
What can you nd out, in other words dont start making theories before you
have a good idea of what actually happened, is a very good idea its always very
easy to come along with your own particular branch of expertise, uid mechanics
as it were, and to rephrase the whole problem in terms of what you know how
to do. But rst you need to look at the data and just see whats going on. So if
people dont do that then they will end up in trouble, thats for sure. So thats
what, if I was going to have cardinal rules in modelling that would certainly be
one of them. (Sam Howison)
The theme of the language of mathematical modelling has been raised here by David Muraki
and Reinhard Illner, as understanding the language is important for understanding the
problem itself. A more prevailing theme presented itself of using the data to help determine
what is relevant, be it experimental or eld data. Brian Wetton also observed that seeing
how the experiment is done can give insight into what would be relevant to the model. Sam
Howison goes on to introduce the concept of using the conservation laws, which also depend
on the physical properties of the problem, as we are trying to conserve those properties.
This ties in to the four responses on understanding, or seeing what is relevant. For
Reinhard Illner, the way to help understand the problem is to rephrase it in his area of
expertise. Sam Howison warns that while rephrasing it in your area of expertise is easy you
must be careful to look at the data and conrm what is actually happening.
Five experts speak about using the mathematical technique of non-dimensionalising the
problem and looking for the signicant parameters this way:
But people that are starting, you dont always know how to [simplify]. I mean
you wont have that experience so you should keep everything to begin with. And
then you can, by doing dimensional analysis and taking ratios of the dierent
eects youll get numbers that are either big or small. And those parameters,
those ratios will tell you that this eect is much more important than that eect
so I can neglect that term compared to this term. Dimensional analysis is [...]
a very simple sounding mathematical approach but its one of the most useful
guides in modelling for knowing which things to consider. (Thomas Witelski)
Which of course requires that you have an idea what simplication is and that
means you do have to have some notion of such things as asymptotics and the
ideas of scaling and non-dimensionalisation so that you can have a guess at what
are the big mechanisms and what are the small ones in a given situation. (Sam
Howison)
Hopefully we have access to an expert and are interacting with someone whos
not a mathematician. Working with them you can deduce whats important. Usu-
ally you can reduce things down. Reducing non-dimensional quantities (which
is basically what applied mathematics was in the 80s). You have to be careful
though, not to always assume a parameter is not important because it never has
been in previous models. (Lou Rossi)
Lou Rossi cautions us here about prejudging a parameter as unimportant based on previous
experience.
Four experts re-introduce the theme of talking to others and collaborating. Four others
explain that this skill of determining relevance is one that develops with experience:
Develop a spider sense by having a go. (Bob)
So thats something that improves with experience. The people that Ive seen
that are even more experienced than me can do that very quickly almost o the
top of their head because theyve solved like dozens of a similar kind of problem
before. (Thomas Witelski)
Two experts here also explain that sometimes you start with a more complicated model,
and then move to the simpler one from there, as opposed to starting with the simplest one
to begin with:
By getting as much information as you can. Make sure you know the whole
problem. The equations will tell you what to get rid of. Two ways to model are
start with the simple model and add things, or start with the complicated model
and remove things, by using the physical properties. You can also use a quick
and dirty solution, but dont lose the big picture. (Sean Bohun)
But all of that assumes that rst youve written down the biggest nastiest pos-
sible version of the problem that has everything included in it. Thats always the
safe starting point, if youre not sure put it in the equation and well cross out
the things that will turn out to be small later. You cant start with something
simpler [as] its harder to kind of put back things unless youve started with the
whole thing in the beginning. (Thomas Witelski)
In many ways, these responses are similar to the previous ones except that they perform
some of the rst details on paper rather than mentally, perhaps even subconsciously.
4.5 Skills Needed for Modelling
Question 5: What knowledge, skills, and techniques do you think help you to
do Mathematical Modelling successfully?
This question was split into two parts. I explained to the interviewees that while I believed
there were certainly mathematical content skills needed to model, I also was sure that there
were other skills, life skills or characteristics of personality that aided in successful modelling
well. This line of questioning oered a rich set of responses.
The experts were able to list various topics in mathematics, which aided them in mod-
elling. A majority (nine of fourteen) explained that calculus, and partial or ordinary dier-
ential equations were useful to them: a consequence of the fact that these experts generally
work in continuum modelling. Apart from calculus, six experts each mentioned probabil-
ity and statistics, and computing or numerical analysis lending to their ability to model.
Several other topics were mentioned including: linear algebra, abstract algebra, analysis,
data analysis, queuing theory, graph theory, discrete mathematics, calculus of variations,
and optimization.
Even more interesting were the responses given to the non-mathematical category. Once
again the theme of simplication was mentioned, which is not surprising as it is ubiquitous
in the literature. A majority of experts also raised a couple new themes. Eight of the experts
explained that breadth as opposed to specic math knowledge was important:
They must also be able to recognize things (the mathematics) in dierent con-
texts. (Burt Tilley)
Being able to couch problems in many dierent ways. Usually the mathematics
is not hard, nding the right mathematics is! (Lou Rossi)
I think the ability to recognize the aordances and limitations of dierent math-
ematical modelling frameworks rst of all. I think you have to have a good knowl-
edge of what you put in and what you get out of dierent broad classes of models
so that you can choose between those things. (Tony)
The other side that I think is really crucial, is to have a fairly broad based
understanding of mathematics and I think this is something that comes from
experience. The more experience that one has in dierent types of problems,
and dierent mathematical techniques, the more success one might have with the
problem. Experience is really important. (Andrea Bertozzi)
Not specic content knowledge but I think [...] how an expert organises [his/her]
content knowledge is going to be pretty crucial. [...] So Im thinking specically
about when I learn new things Im trying to sort of t them into what I already
know as opposed to [thinking] its just another random fact. Its like tting some
kind of structure. I think thats an important thing, not just what one knows but
how one organises it. (Mike)
These experts agree that while some specic content knowledge is helpful, a breadth of
knowledge is more useful in developing models. The ability to recognize the mathematics in
several contexts is additionally identies as an indispensable tool. Mike reinforces the need
for meta-cognitive skills seen in the literature [18], by explaining that all the knowledge
needs to be organized in such a way that it is easily accessible. This ties in with the
seven responses endorsing relevant knowledge and the six responses that explained that
a background in sciences, particularly the area from which the problem comes, is highly
advantageous:
The ability to read a wide range of scientic literature. Because like I said,
to do a good model of something in biology or chemistry, you have to actually
understand something about biology or chemistry. Denitely [you have to have]
the ability to be able to read a range of scientic literature. (Tony)
Let me not answer for myself but for someone who [would like to be] successful
in modelling. That person needs to know that discipline well. A common pitfall,
for people doing mathematical modelling on the mathematical side, is to assume
that since we know all the advanced mathematics we know more than people that
are in the eld, and if I just apply my dynamical systems to the biology Ill be
able to explain everything. That usually is not true. Someone has said that a
mathematical modeller or applied mathematician working on a particular problem
in an area should be indistinguishable from the person and the practitioners of
that application area. You need to know everything and understand all the data.
But you have all these mathematical tools and you have an ability to distil, to
generalize, and thats what you apply to what you see in the eld. (Ted)
The assumption above is that the modeller will be working with experts from the eld,
which explains why the modeller must understand the background of the problem. This
also implies that the modeller must be able to communicate and collaborate. For eight of
the experts interviewed, the theme of collaboration and communication reappears here:
Also I think the ability to communicate verbally with scientists from elds of
application. Not just reading the papers but actually talking to them and having
interchange with them. Both of those things are crucial. (Tony)
I think its really important to be a good communicator and be able to talk with
people who work on the application side. Find out what it is they know, what
they dont know and to try to put what they know in the mathematics. You have
to be able to ask the right questions and understand the problem from dierent
points of view. (Andrea Bertozzi)
Maybe the most unusual thing I can mention is perseverance in terms of group
interaction, because usually these things happen with a group of people. And
sometimes you may have to have some political or diplomatic ability in working
with the group to convince them that your approach is worth pushing through.
(Thomas Witelski)
Oh what life skills? Well Ill tell you two of them right away. One of them
is persistence, and the other one is listening. You will never make a success-
ful modeller unless you listen to what the experimental people or the industrial
people or your collaborators [say]. Youre going to be making models in another
discipline. Its going to be in industry or its going to be in another scientic
discipline. And if you dont listen to what they say, if you arent prepared to go
the extra mile and learn what the words mean and whats going on, then you
havent got a hope. So the sort of person who walks in and says, Give me the
equations Im going to go away and analyse them mathematically, is not do-
ing mathematical modelling. They may be doing mathematics (and maybe good
mathematics) but its not going to be mathematical modelling. (Sam Howison)
This reinforces the idea that mathematical modelling is a group activity for many of them.
Many collaborative skills are mentioned here: the ability to ask questions, the ability to
listen, being able to get results via communication, and being diplomatic. This is one as-
pect that separates modelling from traditional problem solving which is taught as a solitary
activity. In particular Sam Howison explicitly states learn what the words mean. This
is something that novices do not understand: some words mean dierent things to dier-
ent people. This is one reason for missteps from the less experienced they think they
understand what is being said when they dont (Chapters 6 and 7).
A variety of personal qualities were listed as being useful to the modeller: maturity, con-
dence, passion, curiosity, exibility, stamina, persistence, hard work and patience, among
others:
They also need a mathematical maturity which helps them to really understand
the content knowledge. (Burt Tilley)
Ok a third skill is patience. You need patience to know that your rst six models
may be thrown out for one reason or another. So you should be prepared to go
round the loop of: make a model, analyze it as far as possible, compute it, look
at the results, compare the results with the experiment or whatever information
you have, nd they dont t, try to work out what made them not t, and then
do it all over again. You may have to do that many, many times, so patience is
certainly a virtue. I suppose also tolerance of your colleagues, because modelling
on the whole is a team activity rather than a solo one. So you do need to be able
to get on with people and have a beer with them afterwards. (Sam Howison)
As for sort of other skills that I would draw upon, I suppose sometimes just [...]
the ability to go out on a limb and not really worry about getting it wrong. And
maybe its stupid but well gure that out. And if it is then, Whatever, I tried
it. So I think a condence to try things even though I might get it wrong, Im
still going to throw it out there and try it. And maybe someone laughs, maybe
its wrong, maybe I feel like an idiot but hey, I tried it. (Mike)
These responses illustrate that many dierent personal characteristics and content knowl-
edge outside of mathematics, help a modeller become successful.
4.6 Dealing with Being Stuck
Question 6: What do you do when you are stuck partway through a modelling
problem?
This question was one of the most important: as a student and a teacher, how do we help
ourselves and our students become unstuck? I did not assume that as experts they would
claim never to be stuck, but instead that their expertise would allow them to move from
this state often enough that modelling was not a frustrating experience for them.
This assumption was supported by Lou Rossi and Andrea Bertozzi explicitly stating that
being stuck is a regular occurrence for them. Of particular interest is the fact that Andrea
Bertozzi explained in response to question 3 (Section 4.3) that she is never in the situation
of being stuck at the beginning of a problem, because she would not choose to work on
such a problem. This illustrates that even though an expert has the ability to decide which
problem he/she is working on initially, this does not mean that they will never be stuck
when working on the problem. Unfortunately, this is something that students typically do
not see in class. Through example, it appears to novices that experts always know what to
do.
So what do the experts do when stuck? The theme of communication reappears here.
Six experts recommend talking to others and either: go back to those who brought the
question, talk to others with whom they are working, or sometimes talk to people who are
not involved in the project at all! What is important is that collaborating often is a useful
skill when stuck on a problem:
Go back for more! More data, interrogate people more. (Lou Rossi)
Oh I usually talk to people. I mean thats usually the rst thing I try. So if I
really have been stuck and Im unable to unstick myself to the best of my ability,
then Ill try to talk to people about the model. And that could be talk to other
mathematicians if Im stuck somewhere mathematically or it could be to talk to
scientists from whom the data Im interested in comes from. Talk to them to see
if theres something Ive missed. (Tony)
Andrea Bertozzi: I would say a non-trivial percent of the time I go talk to,
other mathematicians. One of the things that I do if I have a problem where
were stuck, and I think that the problem may have something to do with the
math, I talk to people who may be experts in parts of the math that Im not.
That has been incredibly helpful on several occasions. If I need to do a piece that
I dont have the real expertise to do, I go and nd somebody who does, and its
good for me because I usually learn something as well.
Interviewer: you said you talk to other mathematicians; do you ever talk to the
people whove brought the problem to you?
Andrea Bertozzi: Of course, but thats part of the standard procedure I would
say. I wouldnt call that getting stuck. (Andrea Bertozzi)
The other thing which is very dicult is, so you are stuck on the math. Its
math you dont know. Your model has some structure that you want to get a
handle on. One thing that came up with us is that one of the hardest things
is [when] you dont even know what its called. Its a math eld, maybe. And
maybe theres hundreds of people working on it with beautiful results that you
could use, but you cant even google it because you dont actually know what its
called! Because you just came up with it from the application. I dont know what
you do then [laughs]. You know this is hard though right? Then (this is very old
fashioned): you talk about it in various places and you ask people in the audience
if theyve seen anything like it. But you give 3 or 4 talks a year, thats not the
speed of development that you get with google if you knew what the name was.
(Brian Wetton)
Brian Wetton raises an important point here. Research can be faster and can often point
you to those you need to have a discussion with; however, it is dicult to nd the right
people to talk to if you are not sure what mathematical eld your problem falls under. Two
responses point to research as opposed talking to others in order to gather more data:
I look at the literature, you know, what others have done. (Reinhard Illner)
This theme was not widely expressed and was outweighed by the number of responses on
talking to others instead.
Simplifying the problem is another theme that re-emerges. Experts recommended this
theme if one was stuck initially. Similarly, three experts explain that this approach is one
they would try when stuck in the middle. Note that Lou Rossi also recommends complicating
the model, as it may be oversimplied:
I look for simple examples, as simple as I can make them to learn more about
the structure. And often you discover new features about your problem that help
you along. (Reinhard Illner)
If its a mathematical hang-up that Im having, that tells me that maybe I need
to simplify the model. Maybe I need to let go of some of my modelling details
and simplify the model even further. (Tony)
Simplify it more. Or make it more complicated (maybe you dont have in enough
information to solve it). (Lou Rossi)
Many experts speak of doing something dierent when stuck in the middle of a modelling
problem. This includes: thinking dierently, trying a dierent approach, doing a numerical
exploration, reconsidering assumptions, changing the experiment or model and even starting
over:
[14 second pause] I guess it depends on how [...] you dene stuck or what led
you to be stuck. But often trying a dierent approach: if you were doing things
analytically, it might help to just say ok whatever equation you have, instead of
doing it on the board or on paper and pencil lets compute it. Lets put it onto
matlab or lets write a simple program, something quick and ecient to see:
[were] our calculations that we were doing analytically right? Or did we make a
mistake someplace in the middle of the board. So getting conrmation using a
slightly dierent approach can often help cut through things that may be serious
roadblocks because youve led yourself to a contradiction. Sometimes you need
to back up and then start with a dierent way to get to where you think youre
going. (Thomas Witelski)
And sometimes I play with the computer, I may use maple or matlab in order
to do some numerical exploration. (Reinhard Illner)
The problem [of ] getting stuck, [could be that] your original assumption is wrong
and your model does not explain the essential things. Or the prediction of your
model contradicts [...] some of the essential behaviours of what youre trying to
model. Then you should be prepared to change the model go back to the drawing
board and start all over again. (Ted)
Three experts go even further and explain that they wait and think or do something else
completely, which is qualitatively dierent from trying a dierent approach:
Do another problem for about a week. Let the ideas come to you as opposed to
searching for them. (Burt Tilley)
You do have to give yourself time, you have to let your brain do the work, which
means you have to be thinking about the problem, but you would have to let your
subconscious do the work. So you have a shower, you have a bath, go for a walk,
go swimming, beer whatever. And when you come back to it you may well nd
that things are coming, that your brain has sorted [it] out (Sam Howison)
This idea of your subconscious doing the work while you pursue another activity completely
speaks directly to Hadamards work on incubation and illumination in mathematical cre-
ativity. However, another expert explains:
Creating new mathematics is not part of what mathematical modellers do.
(Bob)
This appears to be in direct contrast to waiting for the inspiration of a new idea. Bobs
belief also explains why eight other experts seek help from other sources, whether it be from
people or literature, to see what has been done before.
The common theme here however, is that these experts all seem to assume they are
capable of solving the problem, or that a solution exists already. They are simply missing
some critical insight or perspective. Their experience has led them to discuss with others,
or change their thinking, or simplify in order to gain this insight. But in some cases, they
need to take the time to let their subconscious arrive at the solution. They also do not
simply say, keep trying as some novices do (see Section 7.6), but each have specic and
often multiple strategies for dealing with being stuck.
4.7 Heuristics
Question 7: What heuristics do you use most often? (e.g. draw a picture, work
backwards, exploiting a related/simpler problem etc.)
In his work Ants, Bikes and Clocks, Briggs speaks of some twelve dierent heuristics for
problem solving. Knowing that the experts would have their own rules of thumb for solving
problems, only a few of Briggs heuristics were mentioned as a method of providing context.
The two most popular heuristics used by the experts were drawing a picture and exploit-
ing a simpler problem. Many of them mentioned that they were visual people and needed
a picture of some kind in order to better understand the problem.
I draw a lot of pictures. I really think very geometrically usually. But this is a
very individual thing. I also like to teach that way, I teach with a lot of graphics.
(Reinhard Illner)
Drawing a picture and using the simplest problem to start with are good.
(Thomas Witelski)
Always draw a picture. Physical intuition is a bit of a two-edged sword because
it could be wrong, but drawing analogues with other situations. (Sam Howison)
Well denitely draw a picture. Probably I wouldnt even know what the persons
trying to tell me if theres not some picture. So if someone else isnt drawing it,
Im going draw it and say, Is this what you mean? Is it like this? Are you
looking at it from the side? (Mike)
The recurring theme of simplifying the problem and trying to solve the simplest model
should not make this heuristic a surprise:
I would say one thing I do try to do, I try to always boil it down to the simplest
mathematical model that captures the important parts of the problem. If you
put too many parameters and factors into the model it may be too much for the
model. So keeping it as simple as possible is really important. If we end up with a
model that we think is correct physically, but we dont know how to solve, another
thing we do is come up with a simpler model that has some of the features that
we can solve. Thats where knowing a lot of detailed mathematics helps, because
you can say, I dont know how to solve this really hard problem mathematically,
but I think I can solve this simpler version, and you go from there. You do it
in stages. (Andrea Bertozzi)
Apart from these two main heuristics, exploiting a related problem was mentioned four
times and the theme of communication or talking to others three times:
Solving a similar problem: the joy of maths is the ability for dierent circum-
stances to generate the same maths. (Bob)
Talk to application people. Write down some simple things. Look at experi-
ments. Both how theyre conducted and what comes out. (Brian Wetton)
And asking questions is the other big, big heuristic. If theres someone, an
experimental person, ask them some questions, because youll develop your own
intuition of whats going on. (Sam Howison)
One other technique mentioned by two experts was looking at limiting cases and under-
standing the behaviour there:
[T]his is something we commonly do in applied mathematics [...] I try to see
if limiting cases of my model make sense. So if I make some parameters in my
model very large or very small, or if I make my system size very large or very
small, do I get the behaviour that I expect to and does it jive with what would be
seen or what has been seen in nature? One heuristic I use is to look at limiting
behaviour. (Tony)
Ill tell you a really useful heuristic which is in the what if line: ask what if one
of the parameters becomes very large or very small what would happen? Get a
really simple limit where you can actually see what would happen, and then does
that tell you anything about what might be happening in the more complicated
limit? So dont just always say all the parameters are 1 and 2 and 3 and so on.
Say, What happens if I make that one large or that one very small? It may
not tell you anything but its always worth trying. Its broadening the physical
universe a bit to consider physical set-ups, which arent exactly the same as the
one that youre thinking about. So trying a few thought experiments is always a
good idea. (Sam Howison)
The experts highlighted two of the heuristics mentioned in the question: drawing a picture
and exploiting a related/ simpler problem. We have seen the theme of collaboration raised
before as well. Interestingly, a new theme was identied: using a limiting case. This is the
only one of Briggs twelve heuristics that was raised independently.
4.8 Verifying Solutions
Question 8a): How do you check that you are pursuing the correct solution?
This question underwent a change as early responses such as correct solution means?
(Bob) highlighted the fact that in modelling there is no correct solution (David Muraki).
This prompted me to start asking, How do you check that you are pursuing a consistent
or sensible solution? to which ten experts responded that comparing with the data or the
experiment was the way in which to do this. Four of these ten experts simply speak of
comparing results to data.
Asking the person that proposed the problem, Do you have any data that we
could use as a guide to validate or invalidate what weve done so far? Something
to check? (Thomas Witelski)
I picked up a yer somewhere in the bank or out of my mailbox where some
company had published monthly payments for a mortgage under certain terms.
Of course this is an example that didnt require any new research, because its
all available. But I basically set the task to myself and to the students to say:
now apply our theory and verify that this is correct. And it was really rather
interesting because while the rst part of the theory gave us the correct gure
within a few cents, it didnt give exactly the same gure. And it turned out that
I had overlooked a small legal detail that exists in Canada, that regulates what
kinds of published rates companies may use. Anyway after we took this into
account it came to perfection. So essentially the short answer is I try to check
with the data that is available. (Reinhard Illner)
Two of the experts did not specify what form this data might take, and another one gave
examples of the data being a graph, a simulation or something observed. Reinhard Illner
indicates above that his data is information that is readily available or came from a le and
possible experiment. He also highlights that the model should be internally consistent.
Four of these ten experts stated that their data comes from experiments. Specically
their work comes from modelling some given experiment, and so a comparison to the exper-
imental results helps verify the accuracy of the solution:
You always want to go back and compare to experiments. That might suggest
redesigning the experiment as well or looking at a certain aspect of the data.
When you develop the model that would be what the physicists would call theory,
then you always go back and compare your theory to the actual experiment.
(Andrea Bertozzi)
[Try] to deduce predictions and then see how they compare with experiment or
the intuition of the people that know whats going on (or your intuition about
whats going on if its something that you know about.) (Sam Howison)
Youre guided by experimental work and maybe you have to use some of the
experimental results to t some coecients. So then of course you always worry:
have you used the data too much [...]? So the standard thing that they do is,
even if you have all the data at the beginning, you try and separate it. So you
use some of it to t and you see if it matches the other. Thats a nice sign right?
When you have a model you need some coecients often to match whats going
on, but then it can match something else. If you cant do that, then you ask them
to do some other experiments afterwards. Maybe you have an idea that it will
make things go better (whatever better is). If they do something slightly dierent
than what theyre doing now, you get them to try it out. If it does do better, then
this is the utility test of the model, that youve predicted something new or youve
helped their optimisation process. I mean that is why people want to model, is
because its cheaper and safer than doing experimental work. (Brian Wetton)
Andrea Bertozzi explains that the experiment may have to be reproduced or redesigned.
Sam Howison suggests comparison of the predictions of the model with the experiment, and
is the only one who mentions intuition. Brian Wetton cautions that we have to be careful
with the experimental data: the data that you use to create the model should not be the
data that you use to test it.
Finally, two of the ten experts spoke of comparing results to reality or nature, as opposed
to results discovered in a lab:
All my problems are usually motivated by some observation in nature, [so] I
compare what Im getting to nature. (Tony)
This is due to the fact that their work is based on modelling phenomena occurring in nature
as opposed to modelling work done by experimentalists.
Alternate ways to check the faithfulness of the model (David Muraki) were mentioned.
Other ways include checking that the model makes sense, checking that it works as expected
or looking at the qualitative behaviour of the model. One could also see if something
fundamental can be explained. Three experts touched on looking at limiting cases which
was introduced in response to the previous question (Section 4.7):
Come up with a reduced problem. Simplify the problem initially, solve that
problem rst and then check if the complicated solution matches the simple one
with the correct parameter set to 0. (Burt Tilley)
Therere the obvious ones: is the answer positive when it should be? But thats
just to eliminate mathematical errors. Theres always the check that you ask: is
the solution doing what it should? And at any stage: are there any special cases?
Which is again what I was saying just before, let that parameter be 0 or innity
and then look at special cases. Are there any special cases where you can see
whats going on that means that your theory is not wrong? This doesnt mean
its right, just means its not wrong. (Sam Howison)
Four experts mentioned the accuracy of predictions, as another way to compare to the data.
This re-iterates the dichotomy observed in answer to question 1 (Section 4.1) that for some
experts, predictions are a part of the modelling process. Ted speaks of the simplest problem
that can make predictions that are veriable:
Theres this hypothesis called Occams Razor that you need to come up with the
simplest explanation for a problem. If your simplest explanation not only answers
the question of that particular problem but actually could make predictions on
problems that were not asked, but then once you ask them can be veried, then
you know that you probably are on the right track (Ted)
Two experts touched on what to do if there is no data with which to compare. In this case
they compare the solutions to two dierent methods to see if they are consistent:
Well, I always use numerical simulations or some independent approach to kind
of validate an analytical thing. Or Ive always looked at least two independent
ways to solve something one analytical one numerical. (Thomas Witelski)
I also always try to, I usually for all my projects have some numerical and some
analytical results. My work is usually a combination of those things. So I make
sure that those things are also in agreement with each other and in support of
each other. (Tony)
In summary, these responses show a willingness by experts to stretch the situation under
consideration beyond reality by assuming symmetry, limiting parameter values or problem
reduction to test the model. As we will see in Chapters 6 and 7 novices instead are typically
bound by the constraints of the given parameters and conguration. This means that they
tend to proceed with less condence.
4.9 Changing Strategies
Question 8b): What makes you change solution strategies?
If you checked and realized you werent getting a sensible solution, then switching strategies
would seem to be the right thing to do. With this in mind, creating this question 8b)
as opposed to a new question seemed appropriate. Two main themes emerged from this
question, clarifying my idea with a bit more detail. Changing solution strategies can happen
as a result of being stuck or getting unsatisfactory results.
Five experts explained that being stuck was a reason for changing solution strategies.
Hitting a wall would be one. But I think, at least if its a group eort, were
probably going to be trying a bunch of solution strategies at the same time any-
way, so if one of them isnt really progressing anywhere then I might sort of leave
it on hold and think about something else. (Mike)
Mike also explained that trying several solution strategies at the same time is the norm
when doing modelling in a group. This is not unique to Mike, as I also observed this
happening whenever working in modelling camps. The group tends to split tasks and
approach problems in several dierent ways concurrently, ensuring that all solutions are
coherent and compatible. While Mike likens being stuck to hitting a wall, Lou Rossi uses a
more poetic description of changing strategies in this situation:
Its a lot like water owing downhill. If a method is working really well and
going along nicely, we like that. When it slows down, tributaries form and I
follow the path of least resistance. (Lou Rossi)
As expected, poor results are the main reason for switching solution strategies, with eleven
experts speaking to this point. Ted re-iterates Brian Wettons caveat on using the same
data to create and test the model (Section 4.8):
Suppose you come up with a theory and equation, whose solution explains what
you are asked to model. As I mentioned the next step then is to make predictions
that were not part of the original problem that was given to you. You make
predictions and someone else goes out and measures those, and compares with
you, and if that contradicts what you have, then you need to re-examine what you
have. See thats the danger of having a theory that explains everything that was
given to you [if you] have no more observations for it to verify. So the verication
has to come from observations that you dont know and nobody knows and its
not part of the input to your model. Once you have the model and you set to
make predictions that will be independent verication. (Ted)
Other experts speak to poor results without going into such detail on what data was used
to test the model:
If things in reality arent being captured then the model needs to be rethought.
(David Muraki)
Well obviously if all the predictions were completely at variance with the evi-
dence. You would then have to go back and look rather carefully at everything
youd done. (Sam Howison)
Well, if I get poor results. [laughs] That happens a lot you know, that you make
a rst assumption, you try it and it doesnt do what you expect. And that [means]
you have probably made wrong modelling assumptions. (Reinhard Illner)
If Im trying to do analytical types of proofs or if Im doing parameter explo-
rations within a numerical simulation [and] I cannot reproduce the behaviour
that motivated the model building in the rst place, thats a big clue to me that
somethings wrong. (Tony)
Notice that these experts expect to get a working model that provides a solution; however
they also appear to be very aware that the solution may not match the data. This often
results in a review of their original modelling assumptions.
Finally Andrea Bertozzi speaks to both issues: being stuck and unsatisfactory results:
If it didnt agree with the data. You know the obvious things right? Or you
cant solve it. Nothing that would be a surprise there. (Andrea Bertozzi)
However apart from these obvious things, there are two reasons for switching strategies
that were each mentioned by one expert only. The rst is the idea that the complicatedness
of the problem would prompt Sam Howison to switch strategies.
Another thing thatd make you change your mind is if its clear that the problem
youre formulating is horribly dicult. You might then go back and say, Well
look, if you really want to do this then we are going to have to solve the Navier-
Stokes equations in this horrible, complicated geometry. Are you sure this is
what you really want us to do? So thats also a part of my philosophy of trying
to do the simplest problem rst. Which is a philosophy I inherited it from John
Ockendon who is one of the gurus of study groups. (Sam Howison)
The second unique reason is whether it is worth the eort: Brian Wetton explains that
switching strategies is not always worthwhile based on the results it will yield and the time
commitment required:
Oh well lets suppose you come to one of these things where you are not agreeing
with whats happening and you realise that the most likely thing is a term that
will completely change the structure of your model. Well, then I guess it depends
on your timeline, the right thing to do is then you say well, Start from scratch
and lets put that term in, which completely changes everything and lets see what
happens. This is all at the stage where the kind of models were talking about are
things you could code up in matlab over a week or so. So a week of your time is
a lot to lose, but its conceivable. When you get to the stage when its a year of
your time to make a change then, well then you have to think if its worth your
time to do. Worth it because the application is so important or because youre
so interested in it or someones paying you a lot of money or whatever. (Brian
Wetton)
These two responses point to a desire to do the simplest thing, whether it is in terms of the
complicatedness of the problem or the time and eort required in solving the model. Sam
Howison speaks of a philosophy of doing the simplest problem rst, which he accredits to
modelling group originator John Ockendon. This ties in with Teds discussion on Occams
Razor in response to the previous question (Section 4.8) and Lou Rossis description of
following the path of least resistance.
4.10 Dicult vs Easy
Question 9: What makes a modelling problem dicult/ easy?
My expectation for this question was that the experts would speak to the structure of the
problem being a primary indicator of its level of diculty. Once again the experts were able
to answer this question with enough variety as to show the richness of their understanding
of what it entails, raising four major themes: mathematical diculty, problem familiarity,
problem clarity and problem complexity.
The rst theme that emerges was that of the diculty of the mathematics. This was
sometimes specically mentioned in relation to students:
Hidden depth. A problem may appear easy on the surface. For students, lack of
math knowledge hampers. Also: lack of insight, lack of maturity and the inability
to think in a non-linear way makes a problem dicult. (Burt Tilley)
For students: complicated equations for those who dont want to use comput-
ers, or analysis for those who only want to use computers, makes the problem
dicult. (Bob)
In some instances, the experts spoke about the diculty of the mathematics for themselves:
It could be dicult because you understand exactly the problem and when you
go to try to model it you come up across some dicult mathematics. [...]And
similarly for easy [...] maybe even the problem is sort of hard to describe, but once
you put it down it turns into the heat equation. It turns out to be mathematically
easy to deal with the model itself. (Mike)
Any problem that appears not to oer any analytical simplication at all, it
has to be done numerically, are not my cup of tea, I dont go into that. (Sam
Howison)
Mike describes a contrast between the model appearing easy (because the phenomenon is
easily understood and explained) and the actual math being dicult and vice-versa. This
correlates with Bobs explanation that hidden depth can make problems dicult. Sam
Howison describes a problem that requires numerical analysis, which is not his cup of tea.
This suggests more an issue with the structure of the problem in that it does not oer any
analytical simplication at all, rather than the actual diculty of the mathematics. Brian
Wetton also discusses how the mathematics being easy can make the problem an easy one:
Well thats sort of tricky. Cause it can be easy in a number of ways. It can
just be very simple type of mathematical equation that describes the phenomenon
accurately. In which case you write your code over the weekend and then you t
some parameters and then youre done. It can actually be quite a complicated
phenomenon but very similar to something that there has been a lot of develop-
ment in. So if its anything that is really governed by some kind of uid ow
of a particular kind, that an industrial computational uid dynamics code could
solve for you then, even though that could be geometrically very complex and
very delicate solvers are needed, but those are all in place. So it could be really
hard but if it ts into one of these packages then you might be ok too. (Brian
Wetton)
For Brian Wetton the mathematics being easy is mentioned in conjunction with the fact that
the easy mathematical model accurately describes the phenomenon. Secondly, he discusses
the theme of familiarity with the problem, the subject area or the solution methods. Six
other experts also address this theme:
When you know the answer! That sounds silly but its what you know. If Im
asked to solve a problem in (for the sake of argument) graph theory, I know almost
nothing about graph theory. So I cant tell an easy problem when it walks in the
door! But if its a certain [type of problem then I think] Oh thats easy! I know
how to do that. Reformulate it, (a lot of applied math is about reformulating
things) and it becomes such and such problem with this novel twist. (Sam
Howison)
Also, I think its just important what is the sort of general state of knowledge of
the underlying systems. I mean when we model a uids problem sometimes the
modelling is relatively simple because the rules, the basic rules of uids are so well
understood and have been so well agreed upon for so long, that the mathematical
description is not controversial. In the eld I work in its less well agreed upon
what are the actual underlying rules that govern the system. Its hard to translate
something into mathematics when theres a lot of uncertainty about the thing
youre trying to translate. (Tony)
A problem is denitely easy if it ts into one of the standard areas of science.
Or it has a strong resemblance to previous problems. So thats the easiest kind
of thing you could hope for. But that makes it a little less interesting. (Thomas
Witelski)
Sam Howison explains that a lot of applied mathematics is reformulating, which is not a
trivial skill. As an expert he has developed the ability to see one problem and recognize
that it can be reformulated into a problem that he is already familiar with.
In terms of attitudes when doing mathematical modelling, we see that Tony chooses to
work with the more controversial type of problem rather than one that is well understood
and therefore easy. Thomas Witelski expresses here that the easier types of problems for
him are not as interesting and Reinhard Illner echoes this mentality:
Well I think its easy if tools that youre familiar with, that you nd readily in
the literature apply to it. Then its easy. Its usually hard if you have to develop
the tools yourself in order to make any progress. Tools are the most important
aspect of mathematics. Tools are much more important than results, because you
can often obtain a result by just using a bunch of recipes. But then the problem
is easy. A problem is interesting when the known recipes dont apply and then
you have to develop your own toolbox for it. (Reinhard Illner)
Here Reinhard Illner discusses having to develop your own toolbox or mathematical methods
for solving a problem. This is dicult, but also what he deems interesting. Andrea Bertozzi
is also familiar with having to solve problems using models or mathematics never done
before:
I think from the mathematical standpoint what makes a problem dicult, there
are a couple of things: one is if the problem is such that it demands new models
that have never been explored before, that would be a challenge. Another chal-
lenge is if the model problem involves mathematics that hasnt been done before.
(Andrea Bertozzi)
The dierences between some answers: creating new mathematics is not part of what math-
ematical modellers do (Bob) (see Section 4.6) and Bertozzis above statement: another
challenge is if the model problem involves mathematics that hasnt been done before, arises
due to both the extraordinary level of mathematical sophistication of Andrea Bertozzi and
the fact that Bob typically works on a very short time scale one week workshops rather
than year long projects. Andrea Bertozzi goes on to discuss how collaboration has helped
her to identify and learn new techniques for solving dicult modelling problems:
We were stuck on a problem and a colleague of mine was [helping us]. In one
case it was an expert in conservation laws and in another case it was an expert
in probability, statistical physics. And in both cases I actually learned new math
by working with them, which was great. (Andrea Bertozzi)
Once again we notice that the attitude adopted when dealing with the dicult problem is
not a negative one and is actually what these experts nd interesting. This is also seen
in the language that they use. They speak of a problem being challenging and interesting
when it is dicult, even if they do not know the tools needed to solve it. This is in contrast
to the complete novices who speak of frustration in the face of a dicult problem (Chapter
7).
Problem clarity was the third major theme raised in response to this question. Four
experts spoke to this theme:
A problem is dicult if it is very exible, open or the question is not clear.
(Sean Bohun)
Well I suppose that, one thing that makes it easy is very clear and well dened
experimental data and natural observations. If theres clean data, thats much
easier to model. When data is not clean I think its much harder to be able to
describe that. (Tony)
Well it could be dicult because it isnt well-posed, and I dont mean that in
the mathematical sense. I mean it could be dicult because whoevers presenting
it is either not at all familiar with modelling, or just hasnt really thought about
how to ask their problem. So if someones just describing everything. It could be
dicult because its sort of hidden in that sense [...]And similarly for easy, you
could walk in and someones already thought about the problem and they describe
it in exactly the terms that you can translate into a model. (Mike)
Here the diculty is not necessarily with the mathematics or the modellers lack of knowl-
edge, but rather with the person presenting the problem. The exception is Tonys discussion
about the data being clean. This comes back again to understanding the problem being the
rst step, as seen in response to question 2 (Section 4.2).
Finally, the theme of the complexity of the problem itself was raised.
There are some problems that are inherently obviously complicated. If someone
comes along and says, I want to make a model for the climate. Well obviously
you can write down some simple balance law models, energy budgets and so on
and so forth, but getting any sort of validation, or having any idea about whether
their predictions are sensible or not, is clearly going to be a very major task. And
if you want to make any realistic models, there are thousands and thousands of
variables you should be thinking about. (Sam Howison)
A problem is hard if it has many interacting pieces and the interacting pieces are
maybe from dierent areas of science. Or involve eects that I cant immediately
say this one is important and these three other things are not as important, so
I just need to consider that one. We need to keep all of the things that are
possibly there, then there may be like a parameter for each one of them so Ill
have an answer that depends on 20 dierent parameters. It will be complicated
to say anything about the answer except well, heres a program to compute it.
(Thomas Witelski)
Hard I think is when the structure of the problem is really dierent. And that
usually happens if you have a bunch of coupled types of physics that are happen-
ing, thats when you get mixes of equations and then the problems not standard
and then its hard to know what to do and what the math is like, and theres no
code you can use. (Brian Wetton)
All modelling problems nowadays are dicult, the simple problems have already
been done. The most dicult problems are the ones where there are more un-
known parameters in your model than the observations. And this is a problem for
example in say Greenhouse warming, the models that are used have more total
parameters than their observations so they always explain what they observed by
tuning and yet you dont know whether or not you have the right model. (Ted)
In each case we see the common theme being many variables or interacting pieces in the
model. This complicated type of phenomenon is dicult to model. Sam Howison and Ted
give a specic example of climate modelling having this complicated nature.
This discussion was particularly interesting as I expected problem complexity to be the
major theme mentioned: however we see that there are several other aspects that make a
modelling problem dicult for the experts.
4.11 Summary
For the experts there is a dichotomy on the denition of mathematical modelling. For some
experts mathematical modelling is a description of the the real world problem, that is, the
formulation of a real world problem into a mathematical framework. For others modelling
is a process encompassing not only the formulation of the model, but also the solution of
that model, verication of the solution, rening and predictions.
Upon rst encountering a mathematical model experts focus on understanding the prob-
lem, particularly if they are stuck initially. Experts collaborate with the person who brought
the problem as well as their colleagues in order to better understand the problem. Experts
stated simplication of the problem initially as one of their main heuristics. While prompt-
ing was often necessary to get the experts to begin discussing their feelings, several of them
spoke of experiencing excitement, curiosity and interest when rst faced with a modelling
problem. Others spoke of initially feeling worry or anxiety, but were able to move past
those feelings to tackle the problem. A question about what makes problems dicult or
easy revealed that the dicult problems were the interesting ones for the experts.
The experts interviewed revealed several aspects that go into successful modelling. They
have autonomy and choose problems that they are interested in. They deal with being
stuck in the middle of modelling by collaborating, simplifying and trying to understand the
problem better. They always check that their solution is sensible, usually by comparing it
to the data, but also by comparing it to solutions of other methods. They deem several
skills important to modelling including a breadth of knowledge and an understanding of
the background of the problem. Among the non-cognitive skills valued by the experts are
patience, collaborative skills, persistence, maturity and passion.
Chapter 5
Intermediates
The second group of participants I worked with was the intermediate modellers, consisting
of 8 graduate students and 3 post-doctoral fellows at SFU. The post-doctoral fellows were
of particular interest as modelling had been their focus for the majority of their graduate
work. While they were in the intermediate range, they were closer to being experts than
novices. In this work we shall refer to them as PDFs, when dealing with them specically
in the group of intermediates. This allows a rich landscape of responses and will hopefully
provide insight into the transition from intermediate to expert.
Although all of these SFU graduates went through the same graduate program, many
of them had completely dierent experiences due to their existing background knowledge.
This, coupled with the fact that most of them partook in the problem solving sessions
(see Section 3.4) organized for the applied mathematics graduate students, allowed me to
compare how their diverse backgrounds and interests helped them in a common setting.
Using the coded themes of the experts, I noted which of the intermediates identied
with those same themes, highlighting the other themes that emerged which were exclusive
to the intermediates. Below are their responses to the questions previously asked of the
experts. Each intermediate in this study is anonymous, denoted with an initialled I. to
mark intermediate status. A few illustrative comments are used to highlight themes of
interest.
76
CHAPTER 5. INTERMEDIATES 77
5.1 Dening Mathematical Modelling
Question 1: How do you dene Mathematical Modelling?
It is noticeable that this description is not as rich as that of the experts. This group of
intermediates described mathematical modelling as simply the use of mathematics to solve
a real world problem. What particularly stands out is the fact that there is no obvious
dichotomy regarding whether the solution is a part of the process of modelling or not. Most
intermediates simply assumed that this was the case:
Mathematical modelling, for me its the dening and solving of problems in the
natural or social science, in a mathematical setting, framework. (Frank I.)
All other themes introduced by the experts here were minimally addressed by the inter-
mediates. Only two intermediates speak of simplifying the problem, another two speak of
modelling as describing the real world problem. Two PDFs mention the need to compare
solutions with data. It should be noted here that the PDF responses were somewhat richer
in their denition of modelling:
I guess its a way of taking some real life problem and breaking it down into
component parts, simplifying it in some ways and then applying known mathe-
matical tools or relationships to try and understand the real world problem and
then explore relationships within it. (Tess I.)
I think mathematics is a language, allowing us to study quantitatively any pro-
cess in physics or medicine or whatever else. So essentially mathematical mod-
elling is a kind of playing with this grammar with the syntax and so on just to
explain better our world. And moreover, we must also check the results with
experiments. (Saul I.)
The themes of the language of mathematics, and simplication are seen here for the rst
time. Notably no intermediates spoke of models requiring renement or being used to make
predictions. This gives rise to the question: why do the intermediates do their modelling?
If the purpose of modelling is simply to solve the modelling problem, then the rst answer
arrived at may be accepted. However, if they are trying to actually solve a problem in real
life, then renement and/or predictions might become more evident to the intermediates as
being part of the process.
5.2 Initial Thoughts, Plans and Feelings
Question 2: Upon seeing a modelling problem, what are your rst thoughts,
feelings or plans?
As with the experts, the responses to this question are separated into two main sections:
feelings, and thoughts or plans. Of note is the fact that fewer intermediates needed prompt-
ing about their feelings than their expert counterparts. There are several similarities and
dierences that are highlighted below.
Feelings
Only four intermediates needed prompting to discuss their feelings, and two of these were
PDFs. This may be indicative of the fact that feelings play a less pertinent role in modelling
as we move from being intermediate to expert. This is not to say that the expert does not
experience these feelings, but that they are probably more used to or better at moving past
those feelings and applying the more cognitive aspects of modelling.
For two intermediates, while their feelings were addressed without prompting, the ques-
tion seemed surprising to them:
Right. Oh. Emotionally? Emotionally I like that problem-solving aspect.
(Virgil I.)
Felt like emotionally? I guess it depends on the problem right? And how good
of an intuition you have about the problem, because if its an area that youre
not sure about, or its hard to visualise the physical situation or whatever youre
modelling, then it would be a little intimidating I think to be asked to come up
with a mathematical model for that situation. (Ryan I.)
Other intermediates addressed the feelings aspect of the question right away without prompt-
ing. In many of these cases (eight out of the eleven interviewed) the feelings referred to were
negative: mainly a lack of condence or a feeling of being overwhelmed:
Most of the time my rst feeling is, Oh my God I need to work on this with
someone! Because I always second-guess [...] what my rst thoughts are. And if
someone else has the same rst thoughts as I do then I feel a little more condent
going forward. (Isabel I.)
It depends, sometimes it could feel a little overwhelming. It could seem like this
is way too complicated. There cant be a mathematical model for this. (Frank
I.)
My rst feelings are: I have no idea what to do! I guess that often I feel that
with starting from scratch. I doubt the possibility of actually coming up with a
machine in terms of mathematics, because often I see that mathematics tends to
be a lot simpler than whats going on in the real world. (Reuben I.)
This agrees with half the experts who reported negative emotions. However, the experts were
able to overcome those negative emotions. When intermediates spoke of being overwhelmed,
this emotion stemmed from the complexity of the real world problem. This explains why
simplication is the rst plan of three intermediates inundated by the complexity of real
world problems. This ties in with the theme of simplifying seen in the literature [19] and
utilized by the experts as one of their rst plans and also one of the main heuristics they
implement (Sections 4.2, 4.7). Interestingly none of the intermediates spoke of overcoming
this discomfort or ignoring the panic as some experts did.
In comparison to the eight intermediates that spoke of negative feelings upon rst seeing
a problem, two spoke of positive feelings. It is also noteworthy that one of them (Virgil
I. quoted above) seemed surprised about the question, and the other needed prompting in
order to express her feelings on the matter:
Interviewer: Ok. First things that you felt, you left that out.
Danny I.: First things I felt?
Interviewer: excited, terried, nothing?
Danny I.: I wasnt terried. Nothing really. Not, no, no. Maybe, ok. I wouldnt
say excited but you know, eager to get going on it, you know, condent that Id
be able to do something but not, not scared that I dont know, you know what
complicated model Im going to have to look up or anything like that.
The emotional aspect for Danny I. seemed dicult to pinpoint. Both Danny I. and Virgil
I. experience positive feelings but require prompting or have diculty articulating these
feelings. This is in direct contrast to those intermediates that spoke clearly of being over-
whelmed.
Only one intermediate spoke of indierence then leading possibly to positive or negative
emotions:
At the beginning Im kind of indierent and either I gain excitement for the
problem or I dont. Usually if I can reduce it to some kind of dierential equation
then I get excited about it. (Linus I.)
This response is separated from the others because no other intermediates seem so undecided
emotionally about doing mathematical modelling, having clear cut feelings which were either
negative or positive.
Thoughts and Plans
There was some overlap between the themes raised by intermediates and experts. However,
the intermediates sometimes diered from the experts in terms of viewpoint. Additionally,
sometimes the intermediates only briey touched on major areas of concern for the experts.
The theme of exploring and doing research was mentioned by ve of the intermediates
interviewed. But where these dier from the experts is that they are all speaking of research
strictly in terms of the model itself in an eort to see what others have done:
The rst plan is exploratory. And you try to cover the breadth, the most number
of paths you could take, just think random stu. Also research. Like, look up
what other people have done. Thats, I would say for problem solving, thats the
number one thing you can do is, you should dedicate the rst 10% of your time,
at least 10% to just reading. (Virgil I.)
First of all you would want to try to look at and see if this problem has been
sort of looked at before. Because theres no reason reproducing something that
people have already done, so doing a thorough search of it. (Linus I.)
Only one intermediate speaks about general exploration, in a way that was reminiscent
of the expert responses. The experts also referred to researching the area from which the
problem comes, to get up to speed on their background knowledge, which is similar to seeing
what people have done before.
Five intermediates raised the theme of understanding the big picture or the dominant
processes. Notice here that there are several dierent ways that they go about trying to
understand the dominant processes:
But I think the rst thing youd try to do is probably you know draw a picture
or something and help sort of organise in your mind what it looks like and what
are the factors youre going to have to consider when youre coming up with the
model of your whatever your problem is. (Ryan I.)
If you start [by] coding youre only, youre limiting yourself to one model and
then you become concerned with the details, those dont matter. Thats just
like trivial stu. Thats like if Im hiking from here to Seymour, instead of me
thinking about the overall best path, Im looking at the ground in front of me
thinking, Should I step over this log or should I step over that log. Thats just
implementation. (Virgil I.)
I try and basically nd what is the simplest and most important aspect of the
physical problem that Im trying to model. (Reuben I.)
Well I think you have to try and think about exactly what it is youre trying to
achieve out of doing the model in the rst place. So what am I trying to show?
I think that in some way informs your starting point. (Tess I.)
Identied above at the themes of drawing a picture, organising your thoughts, looking at
the overall goal, identifying the important variables, thinking about what you are trying to
show and nding what is simplest.
The theme of nding the simplest model or simplifying broached by Reuben I., is touched
on by four intermediates in total. We noted earlier in this Section that three of the inter-
mediates who spoke of simplication as a plan were overwhelmed by the complexity of
modelling real world problems:
Most of the time youre going to need to make some simplifying assumptions
(Frank I.)
The rst things would be drawing a picture. Drawing a picture of the thing
youre trying to model. Or if you cant actually draw a picture then, write down
with simple words: rst we take this thing, and then were going do this to it and
then were going to try to see what happens after some time. So the simplest
parts of the problem that need to be dealt with. (Danny I.)
With my MSc thesis I worked at an ecology centre. And when I went to work
with them, there were no previous models written down for that kind of problem.
So it really was starting from base zero. And I asked myself, Well what is it
were trying to do? What is it were trying to model? And that was very dicult
actually, because youve got a real world problem and youre trying to then break
it down with some mathematical model. If there isnt already a framework there
then thats a lot more dicult, so you start o I think very, very simply and
always go to whats the simplest case scenario. For example modelling an island
as a dot on a 1-D line. Although it seems ridiculous but youre trying to solve
something in the real world by going to something abstract. I think thats where
you have to start because thats where you best understand it. (Tess I.)
Frank I. speaks of making simplifying assumptions most of the time, but does not explain
what criteria would cause one not to do so. Danny I. discusses the theme of simplication
along with drawing a picture. Tess I.s response is more elaborate: simplication is used
when there are no previous models of the problem to apply to your situation. This ties in
with the idea that the rst plan is researching previous models however Tess I. explains that
this is not always possible and in those cases the simplest model is the best way to start.
Again, it was seen that only a few intermediates touched on the other themes raised by
the experts. Reuben I. mentioned the fear of not being able to solve the problem at all, and
Tess I. was the only other intermediate to address this:
I suppose youre apprehensive because given that you know you have to make
so many simplications when you rst start o, you wonder how relevant it will
be. You go through all those anxieties of: what am I hoping to achieve? Will I
be able to achieve that? And how will it be used? Already youre thinking about
possible limitations of any model that you come up with. So there are all these
things going through your mind. (Tess I.)
Only the two PDFs raised the issue of using their knowledge of similar problems or relying
on experience as their rst plan:
Often when you have a problem then you can adapt already existing models. So
with atmospheric science youre trying to predict future states of the atmosphere.
But you know there are existing models so you already have some sort of basic
starting point. (Tess I.)
In the above case Tess I. refers to using known existing models and adapting them, this
ties in with experts discussion on experience and exploiting similar problems (Section 4.2).
Two other intermediates raise the topics of deciding what mathematics to use and trying to
write equations as their rst plan. These two quotes highlight dierent aspects of applying
mathematics rst:
The rst thing is: is it something you want to approach analytically, numeri-
cally? Should we be thinking about how the code should be structured? Should we
be thinking about what equations need to be used depending on which situation
youre going to be in? (Isabel I.)
Mathematical modelling includes dening the problem mathematically, so once
you dene the problem mathematically it becomes much more tractable. At that
point I believe you separate the model from the actual thing youre studying, but
the idea is to stay as close as possible. (Frank I.)
For Isabel I. the point is to determine what mathematics is necessary to solve the model.
For Frank I. the aim of the mathematics is simply to set up the model, so that you can
then determine how to solve it. This links to the dichotomy experts in the denition of
modelling.
While Isabel I. in her quote about her rst feelings explains that she wants to work with
others, no other intermediates explicitly state that talking to others is a part of their plans
or thoughts. This is in contrast with approximately half of the experts who expressed this
as their rst plan of action.
5.3 Dealing with Initial Barriers
Question 3: If you have no idea how to start what do you do?
In response to this question the experts suggested numerous plans of action: collaboration,
research, looking at others previous work, simplifying, starting somewhere, using special
cases, drawing a picture and waiting for inspiration (Section 4.3). Apart from the latter
three themes, the intermediates agreed with the above suggestions.
Research
Among the experts, the most popular course of action when unsure how to start, is to
collaborate with others. When looking at the replies of the intermediates we note that this
is not the case. Nine out of the eleven intermediates interviewed mentioned research as a
solution to not being able to start, making this the most popular course of action for the
intermediates:
Oh yeah. You mean besides the research part? Once Ive researched I still dont
know where to start? (Liv I.)
Ive never had that, because I always have some kind of plan, be it Googling the
terms (Virgil I.)
These two quotes illustrate how the use of research seems invaluable and automatic for these
intermediates. Virgil I. comments that researching means that he always has an idea how
to start. For Liv I., research is such an obvious step that she assumes it is done even before
being stuck initially. Three of the intermediates, including Virgil I. above, use blind internet
searching. This is dierent from others who use books and papers in their research to help
them start modelling, as seen below:
I guess its time to start reading probably. So look for some references starting
with the most basic references I think you can nd. Cause if you start looking
at the ones that are quite specic youre probably gonna get lost in the details,
you know what I mean? So try to nd introductory level references that can
get you started on it. And then once you feel that you have enough background
information, you can start trying to put together a model on your own. But also
looking at sort of more advanced references I guess to see if what has been done
for that problem (Ryan I.)
Depending on, usually you would try to look up at resources and books and like
what, so if its an area that you havent ever studied before then youd probably,
youd rst want to get like a basis for it, so you would read through skim through
books regarding like the standard approaches for starting these things and then
take steps from there. (Linus I.)
These descriptions are more detailed. Rather than a blanket statement about researching,
both explain how this research is done. Ryan I. explains that the way to do this is to
start with introductory references. Linus I. suggests reading through books to ascertain the
standard approaches to your type of problem, as opposed to general background knowledge.
Others Previous Work
Three of the intermediates tie researching with the theme of looking at others previous
work:
You can look up things in terms of community like what other professors might
have done, Google or whatever. All these dierent resources you have can take
care of that. (Frank I.)
I look in the literature. I look for somebody elses ideas basically. (Reuben I.)
This theme of looking at what others have done was raised by six intermediates in total.
This is comparative to the four experts who also raised this as a means of dealing with being
stuck initially.
Collaboration
A total of six intermediates mentioned the theme of collaboration. It is notable that in
contrast to the experts the majority of them spoke of research rst and collaboration after.
One PDF speaks of looking at others previous work in conjunction with collaboration, and
is the only one who does not mention research as a means to start solving the problem:
When youre working on a project with other people you have the benet of
being able to ask your supervisor. If its an M.Sc. thesis or a Ph.D. then youve
got a supervisor, youve got other people whove worked on it that you can go
and ask. And they say, Well such a body did something similar and you might
want to look at their work. So youve always got that as a starting point but I
guess the higher up you go then there are fewer reference points or fewer people
to rely on. I guess always work in a team! (Tess I.)
Tess I. mentions using others previous work in the context of working with a mentor as well
as others who have already worked on the problem. Similarly two other intermediates also
speak of looking at previous work in conjunction with collaboration, specifying that they
were working in a group and talking to their colleagues about the problem:
If youre in a group then you start talking to other people because usually the
combined experience of the group is going to be far higher than just yours. And
everybodys bringing something in. So then you start talking about, you have to
sort of agree on some form of model to begin with. We can get an agreement on
a way of even looking at the problem, then from there, how to go about solving it.
You start talking in the group about what kind of similar problems theyve solved
and what type of approaches they used to solve those problems right? (Linus I.)
This is slightly dierent from two intermediates who specically refer to speaking to the ex-
pert who brought the problem as opposed to mentoring mathematicians or their colleagues:
We didnt completely know what we had to do, but we had a couple of people
from industry there, this one person in particular, who was sort of the chief
researcher engineering person with this company that handled road maintenance.
If we asked him, Ok whats the cost of this procedure versus this procedure? he
could tell us that. Or, How long is this one going to last compared to this other
treatment? So we were fortunate to have someone who was already sort of an
expert in the eld to help us with that. (Ryan I.)
The experts also recommended this theme of speaking to the person in the eld who provided
the question when stuck initially, but are also willing to talk to anyone. Isabel I. gives a
similar explanation of how having the help of many can help start o the modelling process
by sharing ideas. There is, however, no specication of who should be consulted:
I try and consult as many people as possible, because I gure the more input I
have the better start Im going have. (Isabel I.)
Just Do Something
In two cases the theme of research is mentioned alongside the idea of just trying to do
something:
I look in the literature, I look for somebody elses ideas basically. Or I just mess
around. I just do stu that its not obvious [...] is going to lead to a solution.
But Im just interested in this tangential thing and maybe this will give me some
insight into whats happening. So those are my two strate[gies], I maybe do a
little bit too much messing around [laughs] (Reuben I.)
As with the theme of collaboration, research is mentioned rst, giving the impression that
this would be the rst tactic to aid when not knowing where to start, before moving on to
just trying something.
Outliers
Virgil I. pronounces that hes never had that situation of not knowing where to start,
because he can always do research or something simple:
Ive never had that, because I always have some kind of plan, be it Googling the
terms, or just trying a small example, or trying out the problem manually doing
it. Ive never been at the point where I had to really seek external help. Yeah Ive
never been completely like, I have no idea. Thats never happened. (Virgil I.)
The dierence between Virgil I. and his colleagues who do experience being stuck initially,
is perception: they consider themselves stuck and research, he researches and thus does not
consider himself stuck. Danny I. also claims that this is not a situation with which she is
familiar:
I dont think so. I think its usually the opposite, therere too many dierent
ideas, and diculty with agreeing with which one to start with. (Danny I.)
Danny I. claries that here she is assuming that modelling is taking place in a group.
Although collaboration is not explicitly mentioned, it is implied, and notably as a possible
detractor to identifying the rst step. Finding the group dynamic a barrier is a theme we
will also see raised by the novice modellers in Chapter 6.
Intermediates deal with initial barriers rst by research. They also look at others
previous work, collaborate and just try something. Only one intermediate mentioned sim-
plication, which the modelling literature and the experts address repeatedly. Another
intermediate raises the point that collaboration can have a downside.
5.4 Determining Relevance
Question 4: How do you determine what information is relevant to the model?
For the experts, collaboration, experience, the physical properties of the problem, dimen-
sional analysis and understanding the problem were ve aspects that helped them determine
what information was relevant to the model. This question addressed the important theme
of acquiring the simplest model. Similarly dierent intermediates addressed these themes.
The Simplest Model
Determining what information was relevant to the model for the experts meant identifying
the simplest model that captured the eects of the real world problem. Five intermediates
agreed, with one of them claiming simplifying was obvious, another saying it was dicult,
and a third pausing a signicant time before being able to answer:
Decide which is important and which is to drop? [6 second pause] Well to me
its really obvious whats the simplest thing, its convincing people that certain
details are not important at the beginning, and trying to convince them that, Its
ok, we will add them later. But for now lets start with the main dynamics of
the problem (Danny I.)
When you start out you leave out almost everything that is extraneous to your
fundamental dynamics of your problem. I guess thats a very dicult question to
answer. You need the bare minimum of something thats going to give you some
dynamical properties of what youre studying. But you can leave out certain
things that may not change the dynamics that much, but may give you a realistic
answer. (Tess I.)
[19 second pause] I guess that comes down to intuition, I mean if youre trying
to model a physical system usually you have some idea about what the behaviour
is. So you basically want to come up with the simplest set of things that will
at least roughly replicate the behaviour that you kind of expect. Whether youre
getting to the quantitative comparison or not. (Reuben I.)
Both Danny I. and Reuben I. pause before answering this question, suggesting some diculty
in doing so, while Tess I. states explicitly that this is a dicult question. Reuben I. and
Tess I. address the theme of capturing the qualitative as opposed to quantitative behaviour.
Danny I. and Reuben I. speak about having some intuition or understanding from their
knowledge of the problem. The intermediates address this theme of intuition more than the
experts interviewed.
For Danny I. the simplest thing to do is obvious. She raises the theme of collaboration
once again, speaking of the diculties that can be encountered when other members of the
group do not want to begin with the simple model. This is reminiscent of Thomas Witelskis
claim that diplomacy in collaboration is invaluable.
Understanding the Main Dynamics of the Problem
This theme of understanding the dominant processes, which was a recurring theme among
the experts, was only explicitly referred to by one of the intermediates when determining
what variables are relevant to the model:
If Im given a set of information and I have a good grasp on the problem, I
have some understanding as to what I need. (Isabel I.)
However, we also see this theme being addressed implicitly. Reuben I. speaks of having
some idea about the behaviour. Tess I. speaks to understanding the main dynamics of
the problem. Danny I. also alludes to the fact that capturing the main dynamics is more
important, when she explains that all the little details in the coee modelling problem should
be dealt with later.
The theme of understanding is mentioned explicitly in a dierent context, when Virgil
I. speaks of understanding why youre good at determining what is relevant.
I think that even people who are good at it dont understand why theyre good
at it. (Virgil I.)
This again alludes to the fact that this question of relevance is one of the more dicult
aspects of the modelling process.
Collaboration
The experts interviewed recommend collaboration with those who brought the modelling
problem to help determine what factors are relevant to the model. Two intermediates also
mention this theme of collaboration. Frank I. mentions it in a way that is reminiscent of
the experts:
I think its dicult, cause I mean at that point it becomes more of a problem,
more of a question for the person who brought up the, the non-mathematical
question, you know what I mean? Cause someone at some point emitted a
question that you then decided to answer using mathematical modelling. But
depending on, on that persons needs and what they want, yknow youre gonna
be able to remove some details and then add them on later. (Frank I.)
In contrast Danny I. gives two scenarios, the rst one highlights collaboration with colleagues
as useful:
Somebody would say, Oh how about adding this? We would immediately agree
that thats an excellent idea [and] once we have a working model, we can add
that on. Then the person was happy with that and shut up and let us work. [...]
Somebody would go up on the board and say, Ok, lets, Ill draw the model and
lets see if we agree on it. And if somebody had something to say they would say
it, the guy would continue drawing. If somebody wanted to continue on with the
model, they would go up there, continue on, and nobody would interrupt them
until theyre done. And once theyre done they might say, Oh did you ever
consider maybe adding this? They didnt interrupt each other, they were patient
with each other, and they agreed that certain things could be added later that
were not immediately important. (Danny I.)
In this situation Danny I. was working in an industrial math group. The culture of working
together on industrial problems was already established and the members of the group were
of the same mind where working on the simplest model was concerned. More importantly,
their collaborative skills of listening respectively, and diplomacy were well-honed, which ties
in with Sam Howisons theme of listening and other collaborative skills when working in
groups.
In the second scenario, Danny I. describes how collaboration can sometimes cause dif-
culty in achieving the simplest model, as there may be dierences of opinion about what
should be modelled and whether to even deal with the simplest model rst:
They wanted to add uid dynamics models to how the cream mixes. First of
all is it cream or milk? I guess he didnt really specify. Somebody decided to go
out of their way to ask: is it milk or cream? [sighs] Should we assume that the
coecient of the cream is the same as the coee because you know cream is a
little bit thicker? You know things like that, that once we have a model with some
parameters then we can assign values to based on that information, but for now
lets not worry about it so much. And the uid dynamics, my immediate thought
would have been: lets just assume that as soon as you pour it, it is immediately
mixed. [laughs] But no, there was a big debate about that which lasted half an
hour. Which is very frustrating and it didnt get us anywhere. (Danny I.)
Danny I. is describing work with a group of graduate students in the problem solving session
(see Section 3.4). There is not a general understanding or agreement that the simplest model
should be the rst approach, which is in contrast to Teds comment that this is understood
in the modelling community (Section 4.4).
Experience
Experts and intermediates alike value experience in modelling, with four individuals from
each group mentioning this topic:
It seems a little bit, well I hope its a little bit more eective anyway, to take
some of the experiences that youve had and try and decide whats important
right o the bat. As opposed to just having like this amount of information and
going like, Okay, go! (Isabel I.)
Experience is necessary especially at this step in the modelling process because it is dicult:
Its not a well-dened process. Its a process that I think comes from experience.
I think, this is probably the trickiest part in the whole thing. I think this is the
crux of the problem actually, Ok you made a mathematical problem, what good
is it? We essentially have no rigorous rules to check or test it. And, I think
that even people who are good at it dont understand why theyre good at it.
(Virgil I.)
Virgil I. admits that trying to determine what is relevant to the model is not a straightfor-
ward endeavour. Intermediates describe it as dicult and tricky, and experts call it an
art.
Data or Physical Properties
Unlike the experts, the intermediates did not refer to using the data, but three of them did
explain that the physical properties of the system would help to illuminate what factors are
important:
I mean if youre trying to model a physical system usually you have some idea
about what the behaviour is, so you basically want to come up with the simplest
set of things that will at least roughly replicate the behaviour that you kind of
expect. (Reuben I.)
This lack of discussion on the use of actual experimental data suggests that either there is
a lack of understanding of the signicance of the data, or that intermediates tend not to be
working with problems posed by industrialists or experimentalists.
Dimensional Analysis
One means of determining relevance is via dimensional analysis, which is dierent from
simply varying parameters, as more analysis and less trial and error is used. Saul I. is
the only intermediate to mention dimensional analysis as a means of determining what is
relevant to the model:
I think technically what I would do is, perform dimensional analysis and try and
using some physical quantities I found in the literature try to, to guess which are
the terms or the processes that are less important. (Saul I.)
However four of the intermediates did speak about varying parameters to see the eects,
and sometimes explicitly referred to trial and error:
Once [you] have a basic model, you can try experimenting with varying some
parameters and seeing if it makes a big dierence or not. And if you start, you
can assess your results and assess, Ok if Im changing this one thing, trying it
with some dierent values, are my results changing by very much? And if not
then, Ok well thats maybe something that we can (assuming that weve modelled
its contribution correctly) then its something we can kind of ignore. (Ryan
I.)
This raises the question of why the intermediates mention trial and error when the experts
do not. However this is not so surprising. Trial and error helps to build understanding and
intuition for the problem, which the experts already have. While the experts do not speak
of intuition, again the intermediates raise this theme:
And how you make those determinations I have no idea. It just comes from
intuition. (Virgil I.)
I guess that comes down to intuition and kind of like, I mean if youre trying
to model a physical system usually you have some idea about what the behaviour
is (Reuben I.)
We once again must ask why this theme of intuition or making sense was addressed
by the intermediates but never raised by the experts. Intuition is dened as the ability
to understand something immediately without need for conscious reasoning. So we can
speculate that for the experts the sub-consciousness of their intuition makes it something
that they are not aware of enough to discuss. The intermediates are now developing their
intuition and are more aware of it. It is also possible that the experts do not accredit their
ability to determine what is relevant to the model to a subconscious thought process, but
to a very conscious one. They may believe that it is not intuition that helps them, but
deliberate conscious thought and an understanding of the problem.
5.5 Skills Needed for Modelling
Question 5: What knowledge, skills, and techniques do you think help you to
do Mathematical Modelling successfully?
As previously mentioned, this question was split into two parts. I was interested in the
mathematics that the interviewees found helpful, but also topics outside of mathematics
that may be useful to the modeller. For the experts, six main themes emerged. While they
did mention content knowledge, in particular calculus and PDEs, they also addressed having
a breadth of knowledge, relevant skills, a science background and the ability to collaborate.
Experts also mentioned advantageous personality traits they thought essential for modelling
successfully. We will look at all three categories of themes for the intermediates, to see how
they were addressed.
Math Content Knowledge
Among the eleven intermediates interviewed, seven of them spoke of calculus or dierential
equations being helpful in their modelling process, which is reective of the fact that these
students mainly work in continuum modelling. Another major topic of value to the inter-
mediates was numerical or programming skill. Other mathematics topics mentioned were:
discrete mathematics, uid dynamics, Fourier analysis, probability and statistics, image
processing, linear algebra and Lagrangian mechanics.
Non-mathematical Skills
In contrast with the mathematical skills, there were only three main non-mathematical
skills that the intermediates valued: breadth and relevant skills, background and scientic
knowledge, collaboration.
Breadth and Relevant Skills
The two themes of breadth and relevant skills are closely tied. For the intermediates it
is more dicult to separate their thoughts on these two topics. The topic of breadth was
mentioned by three intermediates explicitly:
Well I mean, I mean [5 sec pause] I mean Id stop short of saying theres a
specic like mathematical tool, because dierent problems are going to require
totally dierent tools (Reuben I.)
Tied to the concept of having a breadth of knowledge (as opposed to specic content knowl-
edge) is the theme of having relevant skills. As modelling is a broad area covering many
topics, it is important to be able to identify the relevant skills and utilize them. Another
three intermediates therefore discuss the use of relevant skills:
If youre aware of standard models that have been applied in certain areas, and
you can apply them, then that certainly helps. Because you look at it and you
can say, Oh I know what thats going to need, I can just model that with some
uids dynamics equations and you know what the uid dynamics equations are,
and you know whether theyll be relevant or not. I think that denitely helps.
(Tess I.)
What is needed is not just the actual content knowledge, but also the ability to gure out
what to do, and which skills and methods to apply. In order to do this breadth is required.
This is also tied strongly to having knowledge in the eld that the problem comes from.
Science/Background Knowledge
This theme was raised by seven of the eleven intermediates interviewed. For some it
was just a passing comment on the need to study other subjects, but for Danny I. knowing
enough background knowledge is important in order to research those things that you do
not know.
Quickly be able to know enough about the topic and what kinds of things might,
you might need on the technical side and be able to nd those online. (Danny
I.)
Tess I. gives a dierent insight on background knowledge. She notes here that dierent
experts in dierent elds use the same words to mean dierent things, or vice versa:
Given that a lot of modelling within applications involves working with indus-
try and often non-mathematicians, it sometimes helps if you have a background
not necessarily in mathematics, although obviously you do need the basic funda-
mentals of mathematics. Thats one of the biggest obstacles Ive found when Ive
worked with industry. They have their own terminology, they have their own way
of thinking about problems and its very dicult for people with a mathematical
background, who also have their own terminology and way of thinking, to adapt.
If you cant make those two meet and you dont have any way of understanding
the real fundamentals of the problem away from your mathematical terminology
then youre really in a lot of diculty. It helps a lot to be able to try and under-
stand how others look at a problem and how its dierent to the way you look at
a problem. (Tess I.)
This idea of understanding what the terminology in the area from which the problem comes
has already been mentioned by Sam Howison when he says, learn what the words mean.
It is important to be able to communicate with those people who have brought the problem
to you. To do this, you need to understand their background, their way of thinking and
speak their language. Tess I. conrms this with her above explanation of how this was an
obstacle for her when working in industry.
In contrast to the other six intermediates Livingston I. explains that knowledge outside of
mathematics is not useful to him. This comment may be a result of the fact that Livingston
I.s experience with modelling comes from university courses and the MCM.
Collaboration and Communication
Working and talking with others is a recurring theme that the experts deem important
when doing all aspects of mathematical modelling. Thus, when asked what skills are helpful
outside of actual math content knowledge, many experts re-iterated this theme. The in-
termediates however did not have the same overwhelming response, with only four of them
mentioning collaboration, or communication as an important modelling skill:
I think you have to be really willing to collaborate, because I think that you cant
possibly know everything there is to know about every subject. And so if youre
given a problem, like ok ne, you have a math background but you dont have
any background in where the problems coming from, then you cant, youre not
going to be very, you might be able to solve it mathematically but you wont know
if like what youre getting makes any sense because you dont have that intuition
from the other side of it. And so I think collaboration is huge you need to know
that youre not going to be able to do it on your own. And you know the bigger
the brain trust the faster it goes. (Isabel I.)
Isabel I. explains that collaboration is necessary because the eld of mathematical modelling
is so broad. Liv I. discusses how working with her supervisor often helps her to understand
things in greater depth.
Im always like constantly amazed when I go and see, meet with [my supervisor].
Like Ill come to him I got this problem I understand it Ive got pictures for you
we can move on. And just in like 20 minutes with me, hell [say], What about
this? What about this? Have you proven this? Can we look at it in this way?
It just opens up so much that I didnt I didnt even consider, I didnt even think
to look at, and ultimately I didnt even understand what I had gotten, [or] why I
got the results I had gotten. (Liv I.)
Reuben I. explains that collaboration can help when youre unsure whether modelling a
problem is possible, tying collaboration with condence. Frank I.s collaboration with his
supervisor is successful because they have diering areas of expertise: numeric versus ana-
lytic.
Personal Qualities of a Good Modeller
Looking at the traits raised by the experts, the main personality traits are maturity and
experience; condence and risk-taking; passion, interest and curiosity; and persistence, hard
work and patience. These traits were also valued by the intermediates, and so we will deal
with each area separately.
Experience
While several intermediates implied that experience was important in their answers,
two of the intermediates spoke extensively on this topic. Virgil I. speaks of experience in
mathematical modelling being like practicing a sport, and developing your brain muscle
memory.
I think experience is the biggest. And what I mean is if youre playing any sort
of sport, you cant really concentrate on the overall game unless [everything else]
is just muscle memory. So for me I just think about things a lot all the time.
Ill just be sitting on the bus and some random thought will come in and I begin
to explore it. I think about things that are just of interest [...] youre almost
training your brain muscle memory. So, you know that you have to check xyz
constraints each time you try something, and that keeps you honest [...] I nd
I just think about those types of things, Ill read about them. When Im on the
Internet Ill see an article, Ill read through it. Some new idea pops out and then
I include that idea so my list of, checklist gets rened and modied, and thats
why its kind of a softie science. Its more of an art than a science at this point.
(Virgil I.)
This answer highlights other characteristics along with experience. He speaks of practicing
constantly and of his interest in certain areas, helping him to develop the skills to model
those areas of mathematics well.
Liv I. notes the dierence between her own experience working with rst year calculus
compared to students she works with. She then contrasts that with her experience working
on her graduate studies compared to her supervisor
When Im in the applied calculus workshop, Ive had experience so I can look
at the problems and solve them in a couple minutes, because Ive seen the way
that these problems work out. You pick up on clues along the way. When you
have experience youve seen certain clues, certain things happen before, and you
know how to deal with them, youve had experience in dealing with these things.
Whereas if you see it for the rst time you have to go out and learn how to
deal with it [...] Im always constantly amazed when I go and meet with [my
supervisor]. Ill come to him, I got this problem. I understand it. Ive got
pictures for you we can move on. And just in 20 minutes with me, hell [say],
What about this? What about this? Have you proven this? Can we look at it
in this way? It just opens up so much that I didnt even consider, I didnt even
think to look at, and ultimately I didnt even understand what I had gotten [or]
why I got the results I had gotten. (Liv I.)
This answer touches on collaboration, and she also highlights the theme of seeing the bigger
picture because of your experience. This is something she is able to do when trying to
help students with rst year calculus. Similarly she notices that her supervisor has a much
broader view of the problems that she is working on as a graduate student.
Condence and Risk-taking
Four intermediates mention risk-taking from two dierent points of view: condence,
and fear:
Of course some kind of optimism in the in approaching the in approaching the
problem and also I think it could be useful, an approach like this, ok: I start
working now I dont I dont care for the moment if I arrive to a solution or not
I work, stop. (Saul I.)
I think one thing that holds a lot of people back is not being able to jump in.
When youre unsure of something, youre afraid to jump in. So I think one
characteristic would be people who are good modellers are not afraid to make
mistakes. And just try something, you have an idea try something. (Liv I.)
In conjunction with Saul I.s optimism is the ability to simply try even when youre unsure
of the answer, but this ties in with fear of failure. Liv I. explains that a good modeller
cannot be bound by this fear. This is especially true since modelling consists of constant
mistakes and failure:
Well you have to be ok with getting it wrong, a lot! For the rst little bit you
know? Because if you think that youre going to walk into this and youre always
going to be on the right track on your rst step, youre going to be thoroughly
disappointed and walk away. Guaranteed. (Isabel I.)
Passion, Interest and Curiosity
Several experts when talking about their feelings upon seeing a new problem speak of a
passion for the subject. Both Linus I. and Virgil I. speak of their passion or interest in the
subject aiding them in being able to solve modelling problems.
Im one of those people when I have get a problem in my head I cant get it out
for the life of me, so I succumb to it and I, I have to work on it. A lot of people
when they get a math assignment they procrastinate and wait till the last minute
to work on it, and Im one of those people where as soon as I get an assignment
I have to look at it. I have to start thinking about these things because in the
back of my mind I know I have to solve these problems. So its very compulsive
that I have to start looking at them right away. And in the long run that is very
benecial because then youre not trying to do things last minute. And you enjoy
things a lot more when youre not doing things last minute. (Linus I.)
For Linus I. there is a compulsion to get the work done. He describes himself later as being
obsessive. This is not a negative trait for him, as he will work on a problem until it is solved,
and hence is often successful. He also speaks of the enjoyment he receives from working on
a problem and not having to do it last minute.
Virgil I. does not describe himself as obsessive, but does say that he thinks about things
all the time. He also spends a lot of time thinking about the areas that he nds neat.
Perseverance and Hard Work
This was a theme raised by several experts. Sean Bohun spoke of stamina, while Thomas
Witelski called it perseverance and Sam Howison called it persistence, hard work and pa-
tience. All of these words describe the ability to keep going. For many mathematics students
the idea of working for a long time on a problem is daunting. We will see when dealing with
the complete novices that perseverance is not attached to solving mathematics for many of
them.
Virgil I.s analogy of sports when speaking about experience highlights that one of the
things that he deems important is constant work and practice. One other intermediate rec-
ognizes and states explicitly that modelling requires the ability to keep going and persisting
through failure:
Chances are youre going to go down the wrong path from some time or another,
and if you dont have any perseverance at all youre not going to succeed at all.
(Isabel I.)
The intermediates value several skills when modelling, from calculus to perseverance and
hard work. Collaboration is not valued as much by the intermediates as the experts. In
contrast intuition is valued more by the intermediates.
5.6 Dealing with Being Stuck
Question 6: What do you do when you are stuck partway through a modelling
problem?
Experts and intermediates alike nd themselves stuck when working on modelling problems.
Experts Lou Rossi and Andrea Bertozzi explained that being stuck is a regular part of
modelling. Two intermediates echo that sentiment.
Its amazing how much youre stuck. Always stuck. Like even I leave [my
supervisor], the meeting with [my supervisor] and Im like, Ok I know exactly
what to do. and I go home and like, literally 30 minutes into work Im like,
Oh no! Im stuck again! [laughs] (Liv I.)
Acknowledging that being stuck is the norm is one thing, but the question is what do you
do in this situation. All of the experts had dierent techniques for getting unstuck. Of
the methods mentioned by the experts, the three themes of starting over, simplifying and
numerical exploration were not echoed by the intermediates. These omissions are telling,
separating the intermediates from their expert counterparts. However the following ve
themes raised by the experts did re-emerge extensively upon interviewing the intermedi-
ates: collaboration; researching and gathering more data; perseverance; doing something
else; waiting and thinking. One or two of those interviewed also touched on other themes:
reconsidering assumptions; trying a dierent approach; solving analogous problems; and
changing the experiment or model. We will examine the ve main theme separately, ana-
lyzing the intermediates responses and contrasting them to those of the experts.
Communication and Collaboration
When the theme of communication or collaboration appear, there are several dierent people
with which this can happen: with colleagues, with superiors, with experts in the eld, and
with people who are not working on the problem at all. Three intermediates recommend
talking to others even if they are not working specically on your problem:
I would say talk it over with somebody rst and see. My supervisor [...] says
sometimes you work on a problem so closely that your nose is only this far away
from it and you cant see the whole picture right? So its good to step back. And
I think talking with somebody else, maybe someone whos not as familiar with
the problem as you are, can give you insight into maybe something youve missed
or some assumption youve made that seemed justied to you but maybe isnt.
(Ryan I.)
In the middle of my Ph.D. there was this certain part of a solve that I was doing
and it was giving this error. I didnt understand why or where this error was
coming from. This involved a Poisson solve, so I talked to people that worked
on Poisson solvers and they said, Oh, its known that sometimes you get this
error. Although they werent working directly on the problem I was working on,
talking to them allowed me to nd the source. And I was able to understand that
aspect of it [...] talking to people helps a lot even if they arent working directly
in your eld. (Tess I.)
Talking to someone even if they dont know what youre doing, trying to explain
it also helps. And for me I use this a lot, I dont know if many people do this,
but just rewriting out my notes. Like do I understand what Im saying, can I
write this, can I articulate this out for myself ? (Liv I.)
Talking with another person can help you gain a dierent perspective or they may be an
expert in a portion of your problem and help you overcome a particular barrier. Talking
with others also helps you to articulate your problem, which may in turn help you to solve
it. Of course, intermediates also have the option of talking to supervisors:
When I am stuck I do sometimes email [my supervisor]. But, sometimes I
dont even email him I just start writing an email to [him]. Sometimes I even
just start emailing [him], and I write him, Ive written him like 20 line emails
or whatever, and then at the end I just dont even send them. Just writing it out
or talking it out actually helps me think about the problem and then either I put
my nger on the problem, [...] or I know where to look for the problem. Because
Im imagining what [my supervisor] would say. Cause sometimes when youre
stuck you dont really take a step back and say, Why am I stuck? What could
possibly give this error or what could possibly give me this problem? (Frank
I.)
Liv I. is trying to articulate the problem for herself, and Frank I. for his supervisor, but
the results are the same. In trying to explain the problem they realize what it is and are
better equipped to solve it and become unstuck. This metacognitive step is exactly what
Mason suggests, as writing down why you are stuck helps you to proceed ( [28], p.19). A
good supervisor can also provide guidance while the student is still fully engaged:
Well, recently I was working with [my supervisor], and for about 3 weeks he
kept telling me that I had to use conditional probability on something. And
I conditioned on every possible thing before I conditioned on the right thing.
Which was really frustrating because Id come and hed be like, No, try again.
Because, hes very patient but he does want me to try and get it by myself.
(Isabel I.)
Finally intermediates can collaborate with their colleagues:
Well with the MCM its dierent because you have these four days. So its
limited time and also youre usually a team of 3 or 4 so there theres a dynamic
of: if youre stuck you can always go talk to the other people, you can sort of
bounce ideas o each other. (Frank I.)
The group dynamic and limited time mean that rather than just communicating or articu-
lating in your head, constant collaboration takes place. Here the collaboration takes place
among peers. But as Danny I. explains, lack of experience in a collaborative environment
can make this dicult.
Research
When stuck at the beginning of a modelling problem, nine out of the eleven intermediates
interviewed said that research was something they would do to help them get started. So it
is not surprising that ve of them return to the theme of research when stuck in the middle:
And maybe then, check back in the literature to nd some other things that could
be useful and moreover, check also your calculations, calculations and your code
your numerical code. (Saul I.)
Saul I. mentions research in conjunction with checking your calculations, which is mentioned
by two other intermediates. This theme of checking your work is one that we see in the
intermediate responses, but has not been raised by the experts. Checking your work is
automatic for the expert (Section 4.8), and so they did not articulate it.
Danny I. describes her experience when working in a modelling camp. Note that here
research by one of the members is what helps the group to even identify that they are stuck,
as well as to become unstuck:
We were working on something, it was working out, but then somebody, on the
second or third day came and said, Oh by the way, I did some research and this
thing that we did here that is not valid. But then we all, because we were a lot
of us, we actually took some quiet time to think about it and looking up stu
online and then somebody said, Ok this is what we need to add. (Danny I.)
This scenario incorporates many themes: collaboration as they are working together in a
group; research to identify the problem and to solve it; as well as waiting and thinking.
Finally Ryan I. explains that research is sometimes a dicult option:
I mean obviously theres literature as well you can try to look at that, but that
can be like looking for a needle in a haystack sometimes. (Ryan I.)
This correlates with Brian Wettons statement (Section 4.10) on the diculty of online
research if you do not know exactly what you are looking for.
Perseverance
Persistence was addressed by six intermediates. Isabel I. described (above) her collaborative
eort with her supervisor where his guidance caused her to try several dierent approaches
as she was working. Reuben I. explains that the ability to persist and keep trying several
dierent things is one of the reasons he never considers himself stuck:
I guess I never feel like Im stuck in the middle because when Im in the middle
Ive got enough things that I realise that I dont understand [...]. A few months
back I was working on this code and some of the tests that I was doing werent
quite working out. Basically I was spending a lot of time like trying to debug it,
and trying to gure out what was going on. From [my supervisor]s perspective
I was stuck, but from my perspective I was coming to it every day, I was doing
stu and I was guring stu out. (Reuben I.)
This description re-iterates the theme of checking your work that the intermediates have
raised repeatedly. Reuben I. has a dierent perspective on being stuck: rather than thinking
Im stuck, I should do these things, as his supervisor believed him to be, he viewed it as,
I still havent checked all these things yet, so Im not stuck. This is similar to Virgil I.s
assertion that hes never stuck initially because he can always do research (Section 5.3).
Tess I. gives a completely dierent example of how persistence can be useful. In this
case she also has a dierence of opinion with her supervisor about what he perceived to be
an error but was not:
My supervisor told me there was a bug in my code and I was adamant that it
wasnt a bug. So we had this long process where he kept telling me to nd the bug
and it turned out that it wasnt a bug. So sometimes, if you think youre right
you have to persevere with that. Because otherwise I would have been looking for
this erroneous bug forever. Although working with other people can help, thats
an instance of where you also have to trust your own judgement as well. (Tess
I.)
This is an example of collaboration as well as persistence. Tess I. makes the observation
that sometimes when collaborating with others it is important not to passively follow what
they suggest as it may be incorrect. Linus I. discusses understanding why you are stuck in
conjunction with perseverance. He explains why giving up on your model may sometimes
be a bad idea:
A lot of times when you get stuck, the solution isnt just to throw your hands in
the air and abandon everything. Sometimes there is a subtlety in it that if you
think about the problem and you understand why its not working, then you can
x it. And then it all of a sudden magically works. And at that moment when
youre stuck you have no idea that youre actually sort of an inch away from
solving the problem. So you need to persevere somewhat, you cant just give up.
(Linus I.)
In understanding why you are stuck, you get a better idea of whether your previous work is
useful. This will help you to realize if you only need to persevere or whether or not an entire
revamp is necessary. Virgil I. also speaks to the benets of just pushing through the tough
part of the problem. This is an interesting statement because for many complete novices
(see Section 7.6) this is the reason they are often not successful. They are simply unaware
that sometimes sheer hard work is necessary to get the understanding or the answer. The
lack of understanding why they are stuck, as highlighted by Linus I., compounds the issue.
They are not even aware when only a few more minutes of work will get them the necessary
breakthrough.
Wait and Think or Do Something Else
From the literature on problem solving we see that while pure hard work is necessary,
and persistence is an important skill, sometimes persistence is ineective. Some modelling
experts explain that when they are stuck, they opt to take a break, stop and think or go
and do something else completely (Section 4.6). Briggs in his book Ants, Bikes and Clocks
describes this strategy as Dont spin your wheels. However, stopping and thinking is
qualitatively dierent from doing something else completely, as one involves conscious eort
and the other involves the subconscious. Three intermediates recommend stopping and
thinking, or simply taking a step back:
And the same way you have to, as soon as you get stuck too you also have
to start [...] opening your mind to other possibilities. So there could be other
approaches that you havent looked at and so start thinking about what other
things that you could possibly do. (Linus I.)
Another aspect of not spinning your wheels is doing something else. Rather than stopping
to think consciously of the problem, doing something else allows the subconscious to work
on the problem:
I get pretty obsessive at rst. Youre stuck [...] you try to hammer through.
That almost never works. So you basically, spend two, three hours just doing
nothing really. I mean you think about the problem, maybe its helping but it
seems like its going nowhere. Then I usually think to myself (I mean when I
was young I wouldnt, I would just keep going, and that would be useless.)[...]the
more you do that the more you get tired and mentally tired. [...] And sometimes
Ill work on something else. Ill still be working, but Ill be working on another
problem and maybe an easy problem, maybe something Im more familiar with
that might help me change my thoughts. And very often that works. Its sur-
prising. Usually, Ill go to bed and the next morning Ill start working on the
problem: within the rst half hour I nd the bug or whatever there was. Some-
times its not that easy sometimes it takes weeks but the point is: I try to limit
how much time I spend on a particular problem when I get stuck. (Frank I.)
Frank I. describes almost exactly Hadamards process of mathematical creativity. In the
preparation stage, Hadamard describes the mathematicians work as deliberate conscious
eort. This is echoed in Frank I.s statement on trying to hammer through. Following
this is the incubation stage where the mathematician if stuck in his deliberate eorts he
stops working on the problem. This is what Frank I. describes throughout the passage,
explaining that he stops working on it and does something else. This leads to illumination
where the unconscious mind solves the problem and the answer becomes available to the
conscious mind. Frank I. describes this a little dierently, when he says, within the rst
half hour I nd the bug or whatever it was. Note that this illumination comes with surprise.
Other themes
One theme that was raised by an intermediate was that of solving analogous problems.
Expert Bob mentions this as a means of getting unstuck. Similarly intermediate Isabel I.
has found this to be useful.
Theres lots of stu done on [my area of mathematics] but not in the sense that
I am working in it. So I have to try and say, Ok this is their problem this is
my problem, how do they relate? (Isabel I.)
Isabel I.s eld is one where not a lot of work has been done previously. This makes it
dicult to nd others work in her area. Because of this she looks at similar problems to
see if they can help her solve her own.
Another theme that emerged was that of change, whether changing assumptions, chang-
ing your approach to the problem or changing the model altogether:
So its good to step back and I think talking with somebody else, maybe someone
whos not as familiar with the problem as you are, can give you insight into
maybe something youve missed or some assumption youve made that seemed
justied to you but maybe isnt. (Ryan I.)
Check back in the literature to nd some other things that could be useful and
moreover, check also your calculations and your numerical code. And maybe at
the end also check back your assumptions [laughs]. (Saul I.)
Ryan I. explains that your assumptions may seem justied to you but may not be to another
person. For Saul I. everything should be checked, not just your assumptions. Two other
intermediates discuss changing your approach to the problem:
So there could be other approaches that you havent looked at and so start think-
ing about what other things that you could possibly do. (Linus I.)
Other than changing the assumptions, or your approach to the problem, the entire model
may need to be changed. Experts Andrea Bertozzi, Brian Wetton and Ted all mention that
this may be necessary if you nd yourself stuck in the middle of the problem, however, no
intermediates explicitly raise this theme.
5.7 Heuristics
Question 7: What heuristics do you use most often? (eg. draw a picture, work
backwards, exploiting a related/simpler problem etc.)
For this question I began, as with the experts, by only listing a few heuristics as examples for
them to expand on. However I realized that while the experts in the eld had several tried
and true heuristics that they would use, this might not be the case for the intermediates. I
therefore started listing the twelve heuristics mentioned in Briggs Ants, Bikes and Clocks.
This gave the intermediates the opportunity to choose which ones they found themselves
using most often. The heuristics of drawing a picture, solving a similar/simpler problem
or working backwards were mentioned to all interviewed. The other nine heuristics were
only mentioned to six of the intermediates. One other heuristic of communication emerged
from the interviews. I will discuss the three main heuristics of the intermediates: drawing
a picture, solving a simpler/similar problem and taking a break.
Drawing a picture
Nine of the eleven intermediates interviewed mentioned drawing a picture as an important
heuristic:
In terms of when youre just starting out I guess drawing a picture and every-
thing. (Ryan I.)
I denitely use drawing pictures all the time. (Liv I.)
Some intermediates simply agreed that drawing a picture was a heuristic used when it
was mentioned in the examples. Others went on to elaborate on how drawing a picture can
help when modelling.
Pictures will always help too cause most people are visual learners, whether
they think they are or not. I think its always going to help because pictures are
universal as opposed to math or biology or whatever youre working on. Pictures
are more universal. Youre going to have a better idea. (Isabel I.)
Oh I think its indispensable. Im a very visual person, so for me I cant even
start without a picture. People can describe a problem to me in as many ways
or words as they want, but as soon as the represent it as a picture I think, Oh
ok now I understand what youre saying. It helps me understand the problem to
draw a picture, and if someones trying to describe something to me, represent-
ing something visually is really, really helpful. I understand people think very
dierently, not everybody is a visual person but for me personally it helps a lot.
(Tess I.)
Drawing pictures, Ive been using a lot of pictures. And, appropriate pictures,
[my supervisor] has taught me a lot that you can write a draw a picture you can
have a picture but it might not tell you that much. (Liv I.)
Liv I.s comment about the pictures being appropriate is very interesting. She explains that
you need the right picture for it to be a useful heuristic. Note also that the theme of being a
visual thinker or learner emerges. Isabel I. believes that all people are visual learners. This
is not a unique point of view. Tess I. adheres to the more traditional view that dierent
people have dierent styles of thinking. She explains that for her without a picture it is
almost impossible for her to understand the problem. This is in contrast to Reuben I. who
explains that a physically visual representation is not important for him.
Solving a Similar/Simpler Problem
For some intermediates, solving a simpler problem was considered the same as solving a
similar one. Nine of them mentioned solving a similar or simpler problem or both as being
a useful heuristic in mathematical modelling. In some cases this was mentioned simply as
yes when asked if this was a heuristic they found useful. In other cases this was mentioned
as the rst heuristic to try:
Simple similar problems, ok the rst one is to check a simple model, problems.
And also in all of these it could be useful, is to, ok you have your model, you
know that in some simpler case that should be, you know the behaviour of the
system in some simpler case then check your model in that simpler case. (Saul
I.)
Yeah! Ok, most denitely. So my Ph.D. was a perfect example of this because
what I solved was again approximations to atmospheric problems that I could
solve on my laptop really quite simply, whereas if I tried to solve the full problem,
with all these extra bits gured in, I wouldnt have got it done. So yes, always try
to solve the simplest problem you can. And also, even with your simplest model,
you still want to solve that model numerically in a simple way that will give you
a fairly accurate solution. (Tess I.)
This is again in contrast with Reuben I. who states that he would prefer to solve the full
problem rather than other problems similar to it:
I think that often that the gap between the simpler problem and the less simple
problem is theres a lot of work involved in bridging that. And there are certainly
situations in which its worth it, but often the work involved in bridging that
is as much work as it would be to understand the more complicated problem.
(Reuben I.)
Taking a Break
This heuristic only mentioned to six of the intermediates interviewed. What is interesting
about it is that three of those to whom taking a break was not mentioned explicitly, still
said it was an important one for them. They take a break, do something else and wait for
ideas to come for them. This heuristic has been discussed extensively in light of Hadamards
work. Here I highlight other ways that taking a break can help in the modelling process.
When asked this question, six of the intermediates referred to taking a break. Reuben
I. does not claim to take a break often but notices the value of it:
Oh Im really bad at that. If Im stuck Ill just spin my wheels until Im blue in
the face. (Reuben I.)
Isabel I. speaks about taking a small break. This is not leaving to do something else
completely, but a means of stopping herself plunging in before she is ready:
A lot of time I just jump in cause Im like, Oh I can do this. Nope, nope you
cant! Take a step back! So it would be a lot more eective for me if I didnt just
jump in with both feet right away. (Isabel I.)
While this is not the same as waiting for inspiration, the idea of even taking a small break
to survey the problem is one that is dicult for complete novices (Chapter 7).
Tess I. explains that taking a break can also stop you from becoming bogged down or
sidetracked. She recommends taking a break so that one can stay focused on the big picture:
Yeah I think, take a break cause its very hard to switch o, when youre involved
in something it is very hard to switch o and not get bogged down in [it]. And
sometimes you can get sidetracked into exploring things that are really not that
relevant. I mean a whole other area could open up and you become obsessed
with why is this like this, and in terms of your end aim of what youre trying
to achieve in this modelling process it might not be all of that relevant. But
you could spend a lot of time trying to explore this issue. Its very hard once
your brain gets focused on something to get back on to this bigger picture I think.
Thats dicult. It helps if you can keep in mind what youre trying to do. (Tess
I.)
To go back to the big picture, this is once again dierent from Hadamards incubation
theme. Here the break is not to allow inspiration. Instead this taking a break helps her to
break away from functional xedness [52]. She can stop herself from being stuck in a rut
and re-focus on what is important for solving the problem.
Other Heuristics
Two other heuristics of interest are collaboration and working backwards. Collaboration is
mentioned several times throughout this work but was only suggested by two of the inter-
mediates as one of their problem solving heuristics. Neither the experts nor intermediates
deemed working backwards a useful modelling heuristic, even though it was mentioned ex-
plicitly to them. All other heuristics mentioned to the intermediates were considered useful,
but only one or two intermediates elaborated on each.
5.8 Verifying Solutions
Question 8a): How do you check that you are pursuing (the correct) a sensible
solution?
The experts all agree that if a solution is found it must be veried, this can be done by:
comparing results with the data; looking at special cases; checking that the model makes
sense; checking the accuracy of predictions; intuition; comparing dierent methods. The two
main themes for the intermediates were: comparing results with the data and checking that
the model makes sense. They also mentioned convergence testing which was not raised by the
experts and shall be discussed in this section. Also mentioned by one or two intermediates
were the themes of special cases, intuition and comparing two methods.
Comparing with Data or a Known Solution
Ten of the fourteen experts interviewed raised the theme of comparing the solution from
the model to actual data, as a means of verifying your model. This was by far the most
popular method for checking the validity of the solution (Section 4.8). Similarly for the
intermediates this was the most popular means of checking if the solution was sensible.
Four of the intermediates do not speak about comparing the solution to physical data or
experimental results. They instead speak of checking their solution against answers they
know:
And yeah thats what I would do, just try out a couple things where I know what
the answer is and see if my method catches that. (Danny I.)
So you know there were there were some solutions out there and they all gener-
ally gave kind of similar kind of answers but varied in in to certain degrees so I
did have some kind of benchmark in knowing that well I should try to re-produce
all those results I know that its going to give me this. (Tess I.)
This suggests that the solutions which they are using to compare may be known theoretical
solutions as opposed to physical data. This is somewhat dierent from the experts and
understandably so. The intermediates are very often still doing classroom problems as
opposed to real-world problems, and have little access to raw data.
Two other intermediates recognize that physical, experimental data would be useful.
Frank I. explains that this keeps him focused on solving the real problem and not being
drawn o on a mathematical tangent:
This solution, does this make sense with respect to what Im modelling, what
has been physically observed in a lab? Oftentimes, you always have to try and
keep yourself linked to the problem at hand right? Cause its easy to spread out
into the mathematical problem and forget about what the actual thing was you
were modelling. (Frank I.)
Model Makes Sense
When asked how they check that they are pursuing a sensible solution, six of the interme-
diates interviewed said that they check that the model or solution makes sense. This theme
appears to be somewhat redundant but many of them go on to explain what makes sense
means to them.
Ryan I. explains that the solutions make sense if he can explain them logically:
You should have some intuition of whats a reasonable range of values. And
then when you start changing parameters in your model the way that it changes
should make sense. If it doesnt agree with your intuition that doesnt necessarily
mean that its wrong, but if you do get surprises like that, you should be able to
reason through and gure out whats going on. If it really doesnt make sense at
that point then you have to consider that maybe there is something wrong with
your model or that there could be a bug or something. (Ryan I.)
Being able to logically explain any behaviour helps Ryan I. to determine if the model is
sensible. Note that this theme is raised here in conjunction with the theme of intuition,
rather than data. This assumes then that the model being used must be well-understood,
so that rather than comparing with actual data, your intuition about the range of values is
what is referred to.
Virgil I.s denition of the model making sense is that it is robust, i.e. able to handle
any given input:
We created the original function, we solved the end function, so we could just
compare them. But I think that is kind of like a logical fallacy in a way. I didnt
like it, got in arguments all the time because, rst, it just felt wrong. I think we
needed more. And I just need a feeling that Ive made it robust. And that word
kind of captures my feeling, and what would qualify as robust to me is: it almost
exceeds expectation in a way. (Virgil I.)
Virgil I. explains that it felt wrong, echoing Ryan I.s statement about your intuition
guiding you. Virgil I.s point about creating data for his model being a logical fallacy
echoes Brian Wettons response to this question and Teds response to the following one
(see Sections 4.8 and 4.9). One has to be careful not to use data to create the model and
then use that same data to test it.
Tess I. does not speak about checking with the data per se. However by consulting with
the experts in the eld she assumes that they have some knowledge of how the data should
be behaving:
So you then go back to the people if youre working in the applied eld and
say, Well doing this, what does that physically seem to do? Does that seem
reasonable? If there are already no kind of models out there existing that give
you some benchmark solutions then you always go back to ask yourself, Does
that physically make sense? (Tess I.)
She uses the same term reasonable but explains that this relates to the experts in the
elds knowledge of the physical properties. On the other hand Reuben I. says that for him
it is dicult to look at his mathematical solution and be able to see if it is physically making
sense. He therefore checks to see if it makes sense mathematically.
If I have a set of equations that Im solving I make sure that, [for example] the
travelling wave solutions will propagate correctly. (Reuben I.)
While Reuben I. checks a known solution, like the direction of the travelling wave in his
model, Livingston I. does a more rigorous check of existence and uniqueness:
Yes of course we need to [...] so when we get the equation then from the more
pure, rigorous set of mathematics you should (probably you dont want to prove
it but) be aware that the equation should like be well-posed, it has a solution, its
unique and such things. (Livingston I.)
All of these intermediates have dierent ways to check if the model or the solutions make
sense. Whether they are using their intuition, the intuition of the experts, checking the
robustness of the model or checking the mathematics, they all identify ways that tell them
that the model and hence the solution, makes sense.
However, the experts explain that intuition for the problem may be wrong. Just because
a model is mathematically correct this does not mean it is solving the problem it is supposed
to be. Note here that while these methods are not as ecient as checking the solution against
new data from the physical system itself, they are useful if the data is not available, which
may be the case for several of the intermediates.
Intuition
The theme of intuition was only raised by one of the experts interviewed. However, it
appeared four times when interviewing the intermediates. Reuben I. who comes from a
physics background laments his lack of physical intuition:
Yeah I mean for somebody whos done a degree in physics I actually have a
surprisingly, like Im not the greatest at physical intuition just like understanding
like is this doing a reasonable thing? (Reuben I.)
The above statement gives the impression that for Reuben I. physical intuition would be
a useful asset in determining if the solution determined is a viable one. This is similar to
Danny I.s discussion about the use of intuition in determining what makes sense:
We did some simulations so, the simulations kind of agreed with the intuition
of what should be happening. And yeah thats what I would do, just try out a
couple things where I know what the answer is and see if my method catches
that. (Danny I.)
Danny I. couples her intuition about the problem with use of test cases, where she already
knows the answer, to determine if the solution makes sense. Ryan I. also uses intuition in
understanding what your solution should be doing. Ryan I. also acknowledges that intuition
may not be reliable, as the problem or solution may be counter-intuitive.
Intuition is dened as a non-sequential information-processing mode, which comprises
both cognitive and aective elements and results in direct knowing without any use of con-
scious reasoning [53]. It is interesting that the theme of intuition appears more frequently
than with the experts. A reason for this may be because the intermediates see the experts
at work, but not knowing all of the steps taken, may attribute the experts skill to their
unconscious understanding of the problem. However, the expert does not see it this way.
They understand all of the steps they take and know what experiences they are drawing
from. This experience allows them to go through the process very rapidly, however, and
may seem unconscious to those who are not fully aware of what they are doing. One of
the principal ways that experts process so rapidly is that they aggregate knowledge into
schema or organised conceptual structures, that guide how problems are represented and
understood ( [54], p.33). In other words, experts are able to combine isolated facts that
relate to each other, thinking of information in chunks as opposed to isolated facts. This
organisation of information into chunks or schema allows the experts to process more rapidly
than the non-expert who is sifting through the individual facts.
Convergence Tests
Convergence testing is used with numerical methods, to check the numerical scheme. The
theme of checking that the solution is sensible by convergence testing is one that was
broached by two of the intermediates but not raised by any of the experts. Both interme-
diates that spoke about the use of convergence testing have strong backgrounds in numeral
analysis, and partial dierential equations:
Yeah. I mean, in terms of PDEs you often have convergence studies, so I do
those at every step. So as I said, when I started my modelling, I didnt start with
the biggest PDE, the full PDE that I wanted, I started with the heat equation on
a sphere and then worked my way up to what Im doing now. But every step of
the way you do convergence studies on solutions that you know hopefully. And
if you dont know them you just rene your grid and hope that youre getting
closer and closer to a solution. (Frank I.)
The procedure of convergence testing is well known in the study of PDEs, and partic-
ularly the numerical solution of them. Why then have no experts raised this as a viable
method for checking if the solution is sensible? The two intermediates acknowledge that
they hope to converge to the correct solution. The experts also know that convergence
studies while important to check that the model is working, does not give us enough in-
formation about the solution. This is simply a part of modelling for the expert, while for
the intermediate, this is a step that holds some signicance. This does not mean that the
experts do not do convergence studies, but those studies are not what they use to check to
see if the solution is sensible.
5.9 Changing Strategies
What makes you change solution strategies?
This question elicited two main themes among the experts, of being stuck and getting
unsatisfactory results. These themes also emerged among the intermediates, although their
language for describing them was not as concise as those of the experts. Three new themes
were identied by the intermediates, changing strategies because: of input from others,
usually more expert than themselves; of a desire to improve the robustness or eciency of
the numerical scheme; or to verify that the rst method was correct.
Being Stuck or Getting Unsatisfactory Results
Four intermediates explain that if you are not getting a sensible solution, then you should
change strategies. Ryan I. couples the theme of being stuck with the idea of understanding
why one is stuck:
At some level you have an intuition if you get stuck somewhere: am I stuck
because Ive done something fundamentally wrong and I need to go back and
change my model? Or am I stuck just because theres something a little less
serious I just dont understand this one thing [...] I guess that intuition is just
coming from experience. (Ryan I.)
Ryan I. also brings up the theme of intuition or innate understanding, which is found
throughout his answers. Ryan I. accredits this ability to understand why you are stuck to
experience. Liv I. approaches the same theme slightly dierently. She does not clarify here
whether she means that she is completely stuck, but also thinks it important to understand
why one is stuck:
You try a way and if its not seeming to work very well then trying something
else. For me [that] came when I read some of the research you see people are
doing in a couple dierent ways. So you pick one, see how it works for your
problem, and then you can change if its not working for your problem. But I
think theres learning in that too if you can understand why it didnt work using
that method. (Liv I.)
Liv I. uses the literature to determine what method to try, but does not advocate using the
literature blindly. It is important to understand why a method is chosen in the literature
and whether it is appropriate for your particular problem. Tess I. makes the point that
unexpected solutions do not necessarily mean that you are incorrect, but can mean that
you have broached new territory in your area:
If you try something and youre clear that its not worked, you have to under-
stand why thats not worked [...] You have to say: why is it not giving me that
solution? They could all be wrong and what youre doing is correct so you have
to try to understand why what youve done is giving you the solution that its
giving you [...] its by looking at the result and saying well is that physically
reasonable or can I explain it in terms of the approach Ive taken. (Tess I.)
This idea that everyone else could be wrong is an important point, because if you are
changing your method because you are not getting the expected solution, you may never
advance the eld with new ideas or methods. It is not surprising that as a PDF she seems
to be the only one thinking this way.
But of course we can also get results that simply do not make physical sense and cannot
be explained. This is another indicator that we should change our approach to the problem:
You start in one direction, the one that you think is the most promising. But
then for example if you decided to use forward Euler to solve this PDE, and then
you see that youre getting gibberish, just junk, when you compare it to what you
think you should get [...] Then maybe you try another method and see if that will
give you better results. And then obviously if the other methods dont give you
results you have to go back and see ok, well is the PDE itself even solvable, or
well-posed? I mean this is me because Im a little impatient, but ideally once you
have the PDE the rst thing you want to do is show uniqueness and existence
and solvability and all those things. (Frank I.)
Frank I. touches on his own impatience and desire to get to a solution before checking
existence and uniqueness of the solution as he should. This is in contrast with what he
thinks he should do, but the fact that he knows this is incorrect indicates he has passed the
novice stage [51].
Ryan I. explains that if the results you get are sensible, maybe they do not answer the
question you are trying to investigate, and hence a change is required.
It depends on whether the way you have it right now is able to do everything or
investigate everything you want to investigate. It could be that you come up with
a model that answers some of the earlier questions that you have, but then you
want to dig a little deeper and then if your models no longer sucient then you
might have to go back and start o using something a little more sophisticated.
(Ryan I.)
Finally, Reuben I. explains that he does not worry about understanding why his problem
is not working, but is more interested in getting the model to work. This is very dierent
from the other intermediates that speak about the importance of understanding why you
are stuck.
Other Peoples Input
For two intermediates, collaboration with experts is important, deferring to the expertise of
those that supervise them and have done it before:
I think a lot of it would be other peoples input. We thought we were getting good
results for our poster presentation and then [the professor] pointed out that the
equation we were modelling we didnt break it down properly into its components
[...] so collaboration again will always help you. Make sure you check with
someone who has a little bit more information than you do on that subject maybe.
That will always help. (Isabel I.)
This response illustrates how modelling with others especially those who are more expert
in the area, can help. The intermediate is also not passive, as she goes on to explain that
she xed the problem even though the professor pointed out the error. We will see that this
is in contrast with the complete novices who exhibit learned helplessness (Section 7.6).
Improve Eciency
Changing strategies usually comes because there is something wrong with the original
method, whether it gives no results, bad results or is just too complicated. However there are
other reasons for change even if the results are correct. Linus I. explains that the eciency
of the model can also be improved:
Normally when youre trying to nd analytical solutions Im happy if I can nd
one way. However Ive been interested in nding multiple ways of doing the
numerics for that problem. Because a lot of [times] theres better ways of doing
it [...] if the program takes less time to compute the solution then that would be
considered better. Also if you get a more accurate solution that would also be
better. (Linus I.)
Linus I. explains that a more ecient model is not necessarily easier to write or code, but
gives a more accurate answer or is faster. This is similar to Virgil I.s desire to improve the
robustness of the model. Robustness is another quality that makes the numerical model
more ecient as dierent code is not necessary for dierent inputs:
Its like lightning striking its going take a lot of little paths, but you dont want
to be too close to the ground and then realise you have to go all the way back to
the top and try a dierent path. You just want to branch out a little bit and go,
Oh you know what? We should have written this part of the code dierently.
Instead of, Oh weve got to change the basic data structure were using. So, but
if you have to do that I would say, the main reason is because I couldnt make it
a robust solution. (Virgil I.)
Virgil I. suggests making changes as early as possible as opposed to nding out at the last
minute that the entire method needs revamping. This is the reverse of Frank I. saying that
he goes to the end of the problem and then works back to nd the problem. There is some
similarity here, since both want to avoid having to go through the entire process again.
This is something that the experts accept may have to be done: You should be prepared
to change the model go back to the drawing board and start all over again. (Ted)
Complicatedness of the Problem
Expert Sam Howison explains that if the problem youre formulating is horribly dicult
then he would change strategies. Danny I. echoes this sentiment when describing her work:
Let me think [10 second pause] well I guess in my thesis I did change some
stu in the middle because I had one model that was doing everything at once
and it was extremely dicult to choose parameters, in a way that made it really
undesirable and non-generalisable. So I decided to split it in 2 parts. Do one
stage rst where its very simple and you can choose your parameter. And then
another stage which takes over after the rst one. I didnt see anybody do that.
For my particular problem it made a lot of sense to do that but I didnt see it
anywhere else before. So you know simpler is better. (Danny I.)
The theme of simplication is a recurrent one throughout this work and here we see it again.
If the problem is too complicated then instead of attacking it head on, stopping to simplify
it makes more sense.
5.10 Dicult vs Easy
What makes a problem dicult/ easy?
My expectation upon asking this question was that the experts would speak of the diculty
with the problem itself while the novice would have issues with the equations or mathematics.
The theme that dominates the intermediate responses however is their familiarity with the
problem, and whether they have solved it before. Other themes mentioned were: problem
complexity, problem clarity and collaboration.
Problem familiarity
For the intermediates, a familiar problem, one that theyve seen and solved before, is by
denition easy. This is why Frank I. describes textbook problems as being easy:
In that case its no longer a question of solving a new problem its just a question
of: Ok to solve all these new problems youre going to need to know how to solve
this [...] I think the main thing that that helps you with is transferring the word
problem to a mathematical framework. In that case its a simple sort of algebraic
sort of thing. Youll have ax = b and youll have to solve that for x. Its pretty
easy but I mean thats the base of everything right? So even when Im solving
my PDEs Im solving an Ax = b, its just my A is a matrix, my x is a super
long vector and my b is a right hand side vector. Essentially thats what youre
always doing. (Frank I.)
Frank I. believes that the textbook questions are easy because they are not new. He explains
that this ease comes from the fact that the solution method is therefore known, and so to
solve the problems, you only need to apply one particular known technique. He recommends
transferring those base ideas to the more dicult concepts in order to get a handle on them.
Isabel I. also believes that homework problems are easy ones, but because of their structure:
An easy problem would be one where it was very structured, it would be like
a homework problem, where theyve laid it out to work out nicely. Whereas a
hard problem would be one that maybe no ones looked at before, and you dont
really know what to expect out of it. Cause homework problems you know 95%
of the time theyre going to work out, because someones thought about them
enough and used them enough that you know what your solution should look like.
Whereas if youre venturing on new ground then its not inherently a harder
problem, but its a lot harder to know that youre on the right track. There can
be hard homework problems too but I think just knowing that theres a solution
out there makes it feel easier in the long run, a solution that you can check.
(Isabel I.)
As Frank I. did, she touches on the fact that the homework problems are not new so you
know better if you are on the right track. Having enough experience allows you to transfer
that knowledge to unfamiliar problems. The unfamiliar problem then becomes easier to
solve because it is seen as similar to a problem solved before:
Simple also would be something that is very similar to an existing model, that
has a widely accepted and reliable solution and you just change a few things to
make it work for you. (Danny I.)
Here the problem is not just similar to one that the modeller has seen before but it is also
similar to existing well-known models. Having a known, standard, reliable solution makes
the problem like those in the textbook, as you can check your answer to make sure that you
are on the right track.
This familiarity with the problem and its solution is echoed by Liv I. Being familiar with
the problem means that you have a better understanding and helps you to identify if you
are on the right track with the solution. The familiarity of the problem also means that the
solution is more tractable, even if the solution method is not previously known. Liv I. also
mentions the ability to visualize the problem, and this sentiment is reinforced by Ryan I.s
comment below:
I think its how easy it is to visualize and how much intuition you have about
the problem. A problem that people have seen in their daily life and have some
intuition about, people are going to get a lot farther on it even if they dont have
a lot of background in it. At least you have some understanding of what is the
end goal and what youre really trying to model. Whereas if youre dealing with
something where its more abstract, just o the top of my head something like
particle physics or quantum physics or something like that, I would have much
more diculty coming up with a model for that. (Ryan I.)
Ryan I. attributes the ability to solve the familiar problem not only to the ability to visualize
but to your intuition with regards to the real life situation as well.
Problem Complexity
The complexity of the problem is something that makes it dicult but might not be obvious
to the novice as being the real diculty. A person bogged down with long calculations may
not appreciate that the problem is actually easy because it is clear (though tedious) how
to solve it. Some intermediates show an understanding of this as four of the 14 raise the
theme of problem complexity when discussing what makes a problem dicult or easy.
Frank I. rst speaks on being able to visualize the problem as his colleagues Liv I. and
Ryan I. do, and raises the theme of intuition. He then discusses how the complexity of the
problem structure itself can cause the problem to be dicult:
Lets say for example right now you told me go code up something that models
molecules bumping into each other. Im going to say, Oh well thats super easy,
I just make a bunch of molecules, which are easy to implement. And then I make
an interaction between them, and then run it and solve it. But the problem is
theres things that Im not going to think about. For example, if I have a thousand
of them or a million of these balls, I need to think about computing power and
stu like that ( thats where neighbourless comes in, youre not going to compute
the interaction between everyone, youre gonna compute the interaction between
you and your 20 neighbours). My point is, if you think the problems easy chances
are its going to surprise you, in terms of complications. If only as I said for
numerical reasons. (Frank I.)
This account shows by example how the structure of the problem can be complicated and
thus make it dicult to model. This can be due to limitations on computing power, or simply
the ability to keep track of what is going on. Danny I. captures this thought concisely.
Lets see. Well dicult would be if theres too many things going on [laughs]. If
the problem itself in nature involves too many things that change so you have to
model a lot of things at the same time, chances are its not going to be accurate
when you put all the pieces together. Simple well, simple would be the opposite
of that. (Danny I.)
The simpler the structure of the problem the easier it is. Sam Howison explains that this
is why some problems are notoriously complicated like atmospheric modelling. Two of the
intermediates agreed with this comment, acknowledging that atmospheric modelling can be
complicated due to the many inuences that aect the model. Their solution is to nd those
inuences that are most important or have the greatest eects and ignore the rest. Tess I.
transfers this idea and explains that it can be done for all problems. This brings us back to
question 4 (Section 5.4) and the importance of determining what terms should be discarded
from the model, in order to acquire a model that is solvable with the given resources.
Problem Clarity
Problem clarity relates not to the nature of the problem itself, but to the ability of the
experts in the eld, those who want the problem solved, to articulate exactly what they
want. Five experts including Sean Bohun and Lou Rossi raise this theme, as seen in the
previous chapter. Lack of problem clarity is also mentioned by intermediate Virgil I. as
something that makes problems dicult:
To me its just if they poorly dene what they want. If someone says, I want
to model a car. What? Does it have to run the engine in my model? Or
go straight? Does it have to be able to collide with someone? And I nd that
the biggest dierence when I get a question from someone who really understands
what theyre doing versus doesnt is how well dened they even make the question.
I nd this is both math understanding and a communication problem. Just not
knowing what they want. Theres lots of reasons you could get a poorly dened
question. And that could be poorly dening what they want to model or what
they want to get out of the model. (Virgil I.)
Having a really open question makes the problem more dicult to model, as we need to
know what exactly we are modelling and why in order to do it eectively.
Poor Collaboration Skills
The theme of collaboration has been raised several times throughout the interviews with
experts and intermediates alike. What is interesting is that no one except Danny I. below,
has mentioned it in relation to making a problem easy or dicult:
Dicult, would be working in a large group of people that dont know what
theyre doing [laughs], and that dont have that patience we talked about earlier?
I think working with a lot of people is the biggest problem. Depending on the
setting of those people [it] could be really, really great, or it could be a disaster if
you have to explain everything a hundred times over. (Danny I.)
The fact that collaboration is deemed a useful tool by so many of the experts and in-
termediates only helps to make this point more valid. Poor collaboration can make the
mathematical modelling process a nightmare. We will look at the implications of this very
important theme in the nal chapter of this work (Chapter 8).
5.11 Summary
This look at the group of intermediates showcases several dierences and similarities be-
tween the intermediates and the experts. The intermediates tend to provide more detailed
responses on many of the topics, whereas experts were more succinct. The dichotomy in
the denition of modelling is not evident as with the experts, with intermediates dening
modelling as the use of mathematics to solve a real world problem. There is a shift in focus
here from formulation or the entire modelling process to the solution step of modelling.
Intermediates are more forthcoming with their feelings on modelling than experts and
have more feelings of persistent self-doubt. Intermediates also discussed trying to understand
the problem initially, however they make use of research primarily to do this as opposed
to collaborating with others. Intermediates recognised that the complexity of the problem
often leads to it being dicult, as well as a lack of clarity and the openness of the problem.
Several of the themes mentioned by the experts re-emerged here. While intermediates
have less autonomy, they are still usually interested and motivated in their particular area
of study. They recommend asking questions when stuck highlighting that taking a step
back or articulating your diculties often helps you to overcome them. They named several
mathematical areas of knowledge that were seen in the expert responses and recognised
breadth as opposed to depth of knowledge as being important. Intermediates valued non-
cognitive skills of perseverance, good collaboration and taking a break.
Chapter 6
Novices
Apart from the interviews of SFU graduate students and post-doctoral fellows, I also gath-
ered data from undergraduate students from the modelling course Math 461. I obtained
questionnaire responses from the students of this class who also participated in the mathe-
matical contest in modelling (MCM). Competitors in the MCM choose their team members
(a maximum of three persons per team) before seeing the problems. There are two possible
problems given to work on and team members cannot seek help from anyone outside of their
group. The teams work on the problems for four days and are then required to submit their
solutions (which may be partial solutions). These undergraduates that participated in the
MCM were on the novice end of the spectrum, as this was their rst modelling class and
their rst modelling contest.
Questionnaires were lled out by most of the students a few months after the compe-
tition. The questionnaire had been given immediately after the MCM; however many of
them had several other commitments, and while agreeing to ll it out, they required some
reminding and prompting to get the results in. Because of this in some cases they could not
remember exactly what they had done.
Below is a discussion of their responses to each question. Note that these eight students
are anonymous and the initial N. after a name indicates novice status.
127
CHAPTER 6. NOVICES 128
6.1 Question 1: Dening Mathematical Modelling
In your opinion, what made this problem a modelling problem?
This question parallels the rst interview question on dening mathematical modelling.
While I did not believe that these novices would be able to give a complete denition of
mathematical modelling, I expected that they would have some framework in their minds
which helps them to identify modelling problems.
Similar to the intermediates interviewed these students do not describe modelling as the
experts did. Instead, they speak of modelling simply in terms of using mathematics to solve
real world problems. Thus as with those interviewed the most common themes to emerge
are: real-world problems and the use of mathematics, as seen by the three below quotes:
It asked for a mathematical representation to a real life problem. (Phil N.)
Setting up a physics-based problem. (Mark N.)
This was a modeling problem because it involved using math to represent the
problem and simplifying it with assumptions. (Jamie N.)
Mark N. is even more succinct than Phil N. in his description, forcing us to assume that
the physics based problem is a physical real life one, and it was set up using mathematics.
Jamie N. goes a little further raising the themes of assumptions and simplication that have
been identied in the two previous chapters. Others provide additional depth by giving
some details about their problem:
In my opinion, this is a problem of designing a half-pipe in order to fully explore
the players skating skills, it has its practical meaning. And there are many
variables that we have to control and those variables are aected by each other,
in that sense, this problem falls in the modelling category. (Karl N.)
Karl N. gives a denition by example, explaining how their particular problem was a real
life problem by dening it as practical. He also comments that there were many variables
which contributed to this being a modelling problem. From the responses below we notice
that the themes of an optimal solution, and having more than one solution method or
approach were also mentioned:
There exists an optimal value to be found and under dierent assumptions, this
value might be distinct but similar. The essential thing is the problem can be
solved by many dierent ways ... no exactly correct answer. (Harry N.)
I believe the openness of this problem allowed many possible ways to model this
problem mathematically. For instance, cross-sections of the half pipe could be
modeled by using parabolas; relevant information could be found by examining
arc length (these would reect the distance a snowboarder would have to travel) or
curvature (reecting steepness). There were many other ideas that were brought
up by group members too, and most of these ideas were unrelated to one another,
implying that this problem had many dierent approaches to it. Thus I believe
the exibility of this problem made it a modelling problem. (Joe N.)
This problem is a modelling problem because we had to go through lots of as-
sumptions to simplify the problem and reach a logical solution. Also, there was
no right or wrong answer especially since not enough information was given. We
had to go through some research, but due to the time, we couldnt possibly have
a clear image of what the question was asking. (Karen N.)
These students appear to be dening a modelling problem as any problem that is open,
exible and possibly ambiguous. In contrast, the experts identied these qualities as the
side eects of working with real life problems that have many variables and are therefore
impossible to model exactly. This is especially interesting as these three students were in
the same modelling class. It is quite possible that because the fact that modelling problems
often have no exact solution was highlighted in class, these students believed that this is
what dened a modelling problem.
6.2 Question 2: Initial Thoughts
a) When you rst saw this problem, did you think you could solve it?
b)Why / Why not?
c) How long did you think solving the problem would take you?
d)What aspects of this problem were familiar?
This question was designed to identify the rst thoughts upon seeing a problem. Did the
novices questioned feel condent or unsure upon rst viewing their problem? Five of the
responses showed that at least at rst these students were condent that they could solve
the problem. One novice was more neutral in his/her answer, while the nal two did not
feel condent they would be able to solve the problem.
Of those who felt sure they could solve the problem, two of them thought that the
problem could be solved within a short period of time as well:
a) Yes.
b) It seems like a physics problem and Im a physicist.
c) 3-5 hrs.
d) Mechanics. (Phil N.)
a) Yes, I thought I could.
b) We chose problem a...it appears like a high school physics problem.
c) Then I thought ... this must be easy... it could not take me more than 1 day
d) Energy conservation law, newtons law, friction. (Harry N.)
The problem that these two students each worked on in the MCM was a physics-related
problem. The perceived ease of the problem was based on the self-proclaimed expertise of
the rst student and the fact that the second student viewed the problem as a high school
problem.
Three other MCM competitors also viewed their problem as solvable, but expected it
would take the entire weekend allotted to solve it:
a) When I rst saw this problem, I thought I was going to be able to solve it.
b) The answer to the problem was based on common sense.
c) A few full days.
d) Setting up the physics and solving the PDEs. (Mark N.)
When I rst saw the problem I thought was that was solvable because it seemed
to be quite intuitive. I thought that it would take all the allotted time in order to
model it well. This problem required velocity, distance, displacement calculations
that were very closely related to the material in physics class. (Jamie N.)
The reasons that Mark N. and Jamie N. give for their perceived ease of the problem is the
fact that it was intuitive or based on common sense. The physics of the problem also
seemed to be familiar to them. Juliet N.s response is slightly dierent, as she notes that
her initial assessment of the problem was not accurate.
a) Yes.
b) Because at rst glance it seemed easy, after starting however, I realized how
complex it was going to be / how many simplications were needed.
c) The whole weekend.
d) The physics of snowboarding. (Juliet N.)
Juliet N. discovered that the problem had a hidden depth as was discussed by Bob (Section
4.10). What she believed to be easy turned out to have complexity she had not anticipated.
Karen N. does not state whether or not she thought she would be able to solve the
problem at rst glance. For her, the rst stage was to try to understand the problem. This
is in direct correlation with Polyas rst step in How to Solve It:
a) Some people in my group thought they had an idea of what type of question
it was, so the rst stage was to understand the question and the direction of how
to solve it.
b) We couldnt ask the professor, and we didnt have enough time to do enough
research.
c) It was hard to tell, especially that we didnt have any previous knowledge on
the subject.
d) We thought it had a relationship to some physical aspects, like forces at certain
angles, and thats where we started. (Karen N.)
From Karen N.s response we can infer that she was not completely sure that she would be
able to solve it, or if so how long it would take her to complete it. Her comments that there
was not enough time to do research and that she could not contact a mentor, suggest that
for her these are favoured heuristics. This is in keeping with nine intermediates interviewed
including Liv I., Isabel I. and Virgil I. among others (see Section 5.3).
Two other MCM participants explain that they did not believe they would be able to
solve the given problem. For Karl N., this was due to ambiguity in the problem statement
and a need for more time:
a) No, absolutely clueless, so did my teammates.
b) Because there exists ambiguity of the way the problem is presented, maybe it is
our problem that we cannot understand the problem well, plus everything seems
to be so implicit, that we have to think of everything.
c) Maybe a week would be proper for us, including writing the report
d) Physics. (Karl N.)
Karl N. appeared to be overwhelmed claiming he had to think of everything, which is
similar to Joe N.s concern about the problem. He explains that he had to start from
scratch below:
a) No.
b) Most of the math problems I have seen modeled in the past had to do with
observing trends, rather than to do with designing something physically. Fur-
thermore, for this problem, we had to start from scratch - most [of ] the problems
I have seen before had someone guiding us through.
c) I cant quite answer this question since solving this problem doesnt really
mean much in this case. Creating a half pipe isnt too dicult, but maximizing
height isnt that easy. There are always tweaks to increase the height by more.
d) Maximizing height was the most familiar aspect. Ive seen problems in the
past where we had to maximize or minimize certain things. (Joe N.)
The absence of understanding, familiarity and guidance all contributed to their lack of
condence in their ability to solve the problem even after the fact. These are all themes
that have appeared in the two previous chapters. The novices illustrate how not having
these aspects when they are deemed necessary can aect the students condence in their
ability to do the problem.
6.3 Question 3: Initial Barriers
a) Did you start right away?
b) Why? Why not?
The responses to this question were split, with ve of the participants saying they did start
immediately and the other three saying that they did not. Of those who did not start right
away, the main reason was to allow some time to understand the problem. MCM contestants
Karl N. and Karen N. explain below:
a) No, we thought about it for a bit, and did some research online that afternoon,
maybe we had a sound rest that night.
b) Because we have to fully understand what this question wants us to do, and
have to decide what approach we are going to use before we start. (Karl N.)
We couldnt start right away because not everyone in the group was available,
and more importantly we needed time to gure what the question was asking.
We also had to go through some research to have a better idea on the subject.
(Karen N.)
This desire to get comfortable, research and understand before attempting it is in line with
Polyas rst step: understand the problem [34]. Students often tend to rush into trying to
understand the problem without fully understanding exactly what they are supposed to be
doing.
Other reasons for not starting immediately include not being there and being over-
condent that the problem could be solved.
I joined the team afterward so I wasnt able to start on it right away. By the
time I joined my two other team members have already worked on it a bit and so
I was able to take a look at what they already have and expand on it. (Jamie
N.)
No... since I thought it was easy, I relaxed... (Harry N.)
One student claims to have postponed starting for one day due to time constraints:
Due to time constraints, I started the next day. (Mark N.)
This is notable because of those who said yes they started immediately, the most common
reason for doing so was the constraint of time.
a)Yes.
b) We only had about 3 or 4 days, so time was quite valuable. (Joe N.)
a) As soon as I got in contact with the rest of my team members I did.
b) Because we had limited time to meet. (Juliet N.)
For these students, the limited time appears to have overridden the need to understand the
problem rst. We also must take into account that for some students, trying to understand
the problem may be seen as starting.
Finally, Phil N. did not give such a clear reason for why he started immediately:
a) Yes.
b) Why not. (Phil N.)
This type of response is signicant in its lack of details. There are several reasons why
starting immediately might not be ideal, such as a lack of understanding, not having a clear
plan. By responding why not, it implies that Phil N. is not aware of reasons for waiting
to start solving the problem, or is unwilling to answer the question.
6.4 Question 4: Determining what is relevant to the model
What information did you decide was unimportant? Explain.
Determining what information is relevant to the model is very important in the modelling
process; the experts and intermediates interviewed also agree on the diculty and impor-
tance of this step. The theme of simplifying and always using the simplest model rst
re-occured time and again in the previous two chapters.
Interestingly for two of the MCM participants, there is no recollection of identifying
information that was not important:
All the information I recall was important. (Joe N.)
I didnt nd any unimportant information. (Jamie N.)
This is even more interesting since their colleagues who worked with them did recall ignoring
certain aspects of the problem in order to make it more tractable. In complete contrast,
MCM contestant Karl N. thought that there was a lot of information that needed to be
discarded, leading to constant discussion:
Sorry there are too many of them, we are discussing all the time. (Karl N.)
However, Karl N. did not give any details about what he determined was necessary to ignore
in order to develop a working model in the time frame allotted. Five other novices were
more descriptive. These details fell into two categories: assumptions added or simplications
made. Mark N., Phil N. and Karen N. made assumptions about the problem that were not
specied:
I assumed there was a symmetry, so I ignored the entire case. (Mark N.)
A persons speed, for the course should be optimized for all persons or assume
the person most skilful. (Phil N.)
For part A, we actually had very [little] information given, so we had to make
our own assumptions, and decide what shape might the tube look like. We also
started thinking why it wouldnt look like the standard shape that is used by
skaters. (Karen N.)
Karen N. explains this is because the problem was particularly open and enough details
were not given. This forced the group to make assumptions as deemed necessary. Two
other MCM contestants Harry N. and Juliet N. explain features that were removed to
simplify the model as opposed to assumptions that were made:
I think how the athlete pump was unimportant...actually... I could not model
it....so I simplied the model in another way that assume the athlete did not
pump but an high initial velocity was given before he entered into the transition
part... the high velocity is given that could guarantee he could y out of the
edge... (Harry N.)
The friction of snow and the pumping technique because we wanted to simplify
the problem by having gravity and the normal force as the only two forces acting
on the snowboarder. (Juliet N.)
Juliet N. explains that she wanted only two forces but did not explain why the other forces
were unimportant. Harry N.s reason for his simplication is not based on the eects that
the pump had on the problem as much as his inability to model it. This is at odds with
the responses of the experts and intermediates interviewed, who explain that removing the
variables that are having the least eect on the overall dynamic of the problem is critical
for simplication (Section 4.4).
6.5 Question 5: Mathematics Used in Modelling
a) What mathematics did you use or try?
b) Was there anything you thought would be useful that was not?
The reason for asking part a) of this question was to discover what mathematical content
knowledge was needed to participate in the MCM. Students used a broad and varied skill-set
of mathematics including PDE solutions, ODE solutions, numerics, calculus, Euler-Lagrange
equations, mechanics, physics and calculus, with partial dierential equations being the most
used here. This is in keeping with the mathematics that the experts recommend.
Part b) of this question stemmed from my own observations of work at mathematics
modelling camps. Bob also talks of problems being dicult due to hidden depth, and
intermediate Danny I. explains that once they were working on a problem and someone
discovered that the scale with which their problem was set had in fact made their method
incorrect.
I was therefore interested in seeing whether this was something that these novices recog-
nized for themselves as happening. Seven of the eight students answered negatively to this
question. Four of them could not remember or simply said that this did not happen:
a) PDE related information.
b) None. (Mark N.)
This may be simply a case of not remembering the situation, as Phil N., Juliet N. and Jamie
N. explain in their responses.
a) Euler-Lagrange equations and formulae of mechanics.
b) Cant remember. (Phil N.)
a) Mainly geometry pertaining to circular objects and physics concepts.
b) Cant remember.(Juliet N.)
We tried to solve it from physics aspect, using calculus. We thought that we
could calculate the total force on the snowboarder and then derive acceleration
then velocity from that. Then we could plot the trajectories of the jumps using
parabolas. Im sorry but I dont remember if there was anything that we thought
might be important but was not. (Jamie N.)
This lack of recall is possibly a result of the time lapse between partaking in the exercise and
lling out the questionnaire. There is also a possibility they did not consider anything that
they later discarded or that the parts that were unimportant were not signicant enough
for the participants to remember.
Two other MCM contestants explained that their reason for going with the initial plan
only, was due to a limited time frame:
a) We used so much physics and calculus, and ordinary dierential equations.
b) We hardly had enough time to think of a method and go with it, so we didnt
have enough time to change our opinion. (Karen N.)
a) Lots of physics was involved (which I left to my partners to solve). I at-
tempted using graphs (I guess today I would call it curve tting, but back then
I hadnt studied this yet) to model it. Furthermore, rates of changes were quite
important, so basic calculus was key.
b) Perhaps due to the time constraint, as soon as we thought of something that
sounded like it worked, we just went with it. (Joe N.)
This response to the pressure of time is a negative one, as it stops students from identifying
whether a solution is viable or not. Karl N. below makes the interesting point that everything
used was valuable in some way:
a) We tried both continuous and discrete approach, since we found that by using
continuous method, we ended up with a bunch of nasty dierential equations that
we could not solve by hand or software, then we switched to discrete method, but
we present both approaches in our report.
b) I cannot say anything we thought of was useless, they are all valuable in some
ways, we put everything we thought of on our paper. (Karl N.)
Although we cannot discount the possibility that the time lapse between the contest and
the responses to the questionnaires may have caused novices to forget what aspects they
discarded, these responses appear to reinforce what is seen in the literature: novices tend to
go ahead with their initial plan no matter what [20]. In contrast, MCM contestant Harry
N. shows some expert behaviour when he makes mention of his change of mind from the
use of inertia in the problem to not using it:
a) Integral.....however, it could not be integrated by hand.... I mean the work
done by friction... so, I used midpoint rule...to get more accurate, I used many
points to ensure the accuracy.
b) I rstly thought to think about inertia...but nally did not think it that way.(Harry
N.)
This tendency of the novices to stick with one solution method reinforces the fact that
although breadth is key, it requires the mathematical maturity to know when to use what
and how the dierent areas are interlinked.
6.6 Question 6: Dealing with Being Stuck
a) What barriers did you encounter?
b) How were they dealt with?
c) Did working in a group help or hinder your progress? Explain.
The question of being stuck and what to do when one is stuck has been dealt with in the
two previous chapters (see Sections 4.6, 5.6). From this questionnaire I wanted to gather a
rst hand report of when the participants were stuck and how they overcame it. Part c) of
the question was added to gauge if the students, who did not have a lot of group experience,
would favourably view the group dynamic. This decision was made after some analysis of
the expert interviews where the theme of collaboration was seen as a signicant factor in
mathematical modelling.
In the previous chapter (Chapter 5), Danny I. discussed her problematic experience when
working with others who do not have the required collaborative skills. At the intermediate
stage of expertise, the benets of collaborating with those more expert appear more valued
than collaborating with those who appear to be less expert. This led me to wonder how
the participants of the MCM found the collaborative aspect of the work especially due
to their limited experience collaborating on modelling problems, and knowing no expert
mentor was available to them. This is especially in light of how they overcame barriers,
since collaboration is the main way used by the experts to do so.
The rst response is not very helpful. We see that Mark N. was stuck when trying to
justify his result and never resolved this issue:
a) Justifying my nal result.
b) Left unjustied.
c) Not really. (Mark N.)
His comment not really to part c) of the question does not tell us if he thinks the group
aspect of the MCM was a help or a hindrance. Similarly Harry N.s answer to part c) below,
does not really answer the question:
a) How to calculate the work done by friction.
b) I used numerical way to a approximate the work...use matlab.
c) I rstly thought about it by myself... then communicated with them... they
said okay.. go [a]head...see what we could get...one of our members thought mine
method should not get an optimal answer due to the curve we cannot assume.
(Harry N.)
What is helpful here is that we see that he resolves his problem by using numerics. This is
similar to Karl N.s method for becoming unstuck. When numerics did not work for Karl
N. he chose an alternate method:
a) I might be the one who had most barriers in the team, since I am not in
the modelling class, and never learned ODEs and PDEs. the most unforgettable
barrier that our team encountered was solving those nasty equations.
b) We tried matlab and maple to solve them, but we didnt get anything from
them, so we just gave up doing that, and switched method.
c) No, I really learned lots of things from my teammates, and from this experi-
ence, that will for sure benet my future study or next MCM. (Karl N.)
Karl N. also commends his group members with helping him. This is echoed in Joe N.s
response, although Joe N. implies that the numerous ideas that came from working from a
group were a possible issue:
a) i) Some of the denitions presented in the problem werent quite clear.
ii) At times there were disagreements among group members.
iii) Some of the models we created lead to nowhere.
b) i) We had to make many assumptions.
ii) Nothing you can really do here.
iii) We had to immediately discard the model and try another way rather than
xing it (again perhaps due to the time constraint).
c) It denitely helped. I recall one group member was great with physics, while
another member had more programming experience. Also, more members meant
more viewpoints and ideas. (Joe N.)
It is interesting that Joe N. speaks of models that lead to nowhere and were discarded,
because in response to the previous question he claims to have used everything that was tried.
This perhaps claries that for Joe N. anything that was used at some point was useful, even
if it did not work and had to be discarded afterwards. Again, this might be attributable to
the time lapse between participating in the MCM and answering the questionnaire, causing
him not to remember what was discarded when trying to answer the previous question. Joe
N.s method for becoming unstuck was to try an alternate approach.
The remaining MCM students are ambivalent when it comes to deciding if working with
a group was a help or a hindrance:
Working in a team was helpful because instead of just me trying to gure things
out, I was able to talk to other team members and we would ll in holes in each
others reasoning. There are times when it hindered the progress because we
couldnt agree on how to best approach the problem, but overall it was better to
work with team members. (Jamie N.)
Some of the barriers were dealing with working with students that werent in
Math 461. Also, not having enough background knowledge and having limited
resources. At some stages, being in a group helped to counteract and arise a
discussion about why an idea wont work, or sometimes to build on a logical
idea. (Karen N.)
The physics part was a hinderance since my group members didnt know much
physics, but the later numerical analysis I wasnt much help. (Phil N.)
Both; hindered because everyone had such a wide range of ideas and it was
hard to decide on one approach, helped because each of us had our own set of
strengths. (Juliet N.)
Ways to become unstuck here were making more assumptions and possibly simplifying the
problem, as well as using scientic computing (also called numerics) to solve analytically
complicated problems. Interestingly, the other barrier that these students encountered was
that of working in the group. This was caused by several factors: the other students did
not have the same background knowledge, were unresponsive, or had contradictory ideas. It
is because of this many of these students had issues with the un-mentored group dynamic,
although they recognized there to be some value to it.
6.7 Question 7: Problem Solving Heuristics
a) How many dierent problem solving strategies did you try?
b) What were they?
c) If you switched strategies, what caused you to do so?
This question is an amalgamation of interview questions 7 and 8b). Part a) and b) mirror
question 7 of the interviews on heuristics. Here I used the term problem solving strategies
as opposed to heuristics. Part c) of this question is the same as interview question 8 b).
The experts provided several heuristics for mathematical modelling: drawing a picture,
exploiting a simpler problem, exploiting a similar problem, collaboration and looking at
limiting cases. The MCM students responses suggest that they did not see the question
as asking for problem solving techniques as opposed to specic mathematical techniques.
However, we can still glean information from their answers as to why they switched strate-
gies.
The theme of being stuck is raised by four of the MCM students as a reason for changing
strategies:
a) I recall I had a bunch of ideas at the beginning, but I didnt have a strong
enough physics background to back them up so most [of ] my big ideas were dis-
carded. Eventually, I was just programming for the most part.
b) To be honest, I dont really remember anymore.
c) Switching strategy was due to a shortcoming of the rst strategy. If we bumped
into a big problem, we often didnt spend time trying to x it. (Joe N.)
a) A lot; we re-started our thinking / design at least around 5 times.
b) Researching, simplifying, etc.
c) Because the other strategy was resulting in problems. (Juliet N.)
a) A few.
b) 2D, 3D, dierent surfaces.
c) Getting stuck. (Mark N.)
We probably only tried two strategies. We rst started by thinking of the actual
shape, and why wont any other shape work. Then we gured well start with the
standard shape that skaters use, and determine the angles that would produce the
highest jumps. It was hard to determine what the question was asking, and why
have people decided to build it the way its built. We switched the strategy because
we were getting stuck and running out of time, and we had to do something that
would convince the judges, even though we were still not convinced by our own
strategy by the end of the competition. (Karen N.)
As with the intermediates that were interviewed, all of the responses did not explicitly use
the term stuck. Some students describe this as the shortcomings of the rst strategy
or the other strategy resulting in problems. The theme of time pressure is also evident
here in Joe N. and Karen N.s responses. Karen N. also re-iterates a concern about the
ambiguity of the problem given. This is not surprising, as we know one of the big diculties
of working with mathematical modelling for students is the openness of the problem. The
experts deal with this by going back to the expert in the eld for more details, but this
option was not available to the MCM students.
A complicated problem was also a reason for expert Sam Howison to switch strategies.
This ties in with the recurring theme of experts and intermediates alike to use the simplest
method possible, and is echoed by three of the MCM contestants:
a) We tried a very complex physics way... and my numerical way.
b) The complex physics way is actually out of my knowledge.. one of our members
came up with it... he is physics major.. but he cannot solve it nally... he asked
me to programm a numerical way....HOWEVER, it was too complex, I could not
even help it out although I thought my numerical skills was good.
c) After failing his method, I came up with my method because of failing to
solve....I used a simpler model and numerical method... (Harry N.)
a) Two.
b) Continuous and discrete.
c) As I explained before, since we cant solve those complicated dierential equa-
tions. (Karl N.)
a) At least 3 I remember.
b) Using the Newtonian mechanics, and Langrange mechanics, and numerical.
c) Too complicated. (Phil N.)
Of note here is that for Sam Howison the complexity was embedded in the problem itself,
whereas the complexity of the mathematics or numerics was the issue for these novices. This
illustrates the point made by expert Burt Tilley, For students, lack of math knowledge
hampers. This is reinforced by Bobs statement about what makes a problem dicult:
Hidden depth. A problem may appear easy on the surface. For students: com-
plicated equations for those who dont want to use computers, or analysis for
those who only want to use computers, makes the problem dicult. This is why
we should be working together. (Bob)
One MCM student is the exception as he claims not to have changed strategies:
We didnt exactly try to switch strategies. It was our rst time modeling a
real-life problem so none of us had much idea on what to do. (Jamie N.)
6.8 Question 8: Experiencing AHA!
Did you have a breakthrough or AHA! moment that helped you to make
progress on this problem?
For the most part experts prescribe collaboration as a means of getting unstuck. However,
a few of them spoke of taking a walk or waiting for ideas to come to them (Section 4.6).
My own work at modelling camps consisted of a few AHA! moments that were signicant
and had a big impact on my understanding of the problem. From Liljedahls work [33] on
this we know that what might be an AHA! moment for one person may not be for another.
Thus I was interested in nding out if AHA!s play an important role in modelling for the
students participating in the MCM.
This question elicited a full range of answers. Students claimed to have AHA! moments
that were helpful, AHA! moments that were not helpful, and no AHA! moments at all. Five
of the MCM students spoke of having a helpful AHA! moment:
Yes, that occurs in the last 50 min before it was due. We got an answer from
the discrete approach. (Karl N.)
Yes there was such a breakthrough moment and we ended up using it for some
parts of the problem. (Joe N.)
There was a moment when I realized it was simply an ODE. (Mark N.)
What is interesting about these rst three positive responses about AHA! moments is that
even though the AHA!s were helpful they were not elaborated upon. These rst three
descriptions of AHA! moments have almost no details in them. No information is given
on how or why the students were stuck, what the big idea or breakthrough was, nor what
caused it.
The other two students who had helpful AHA! moments, elaborated to a greater degree:
Yes....I assumed the curve was a ellipse....and nally found out the solution....it
was very close to the fact which is a semi-square....my optimal ellipse is close to
a semi-square but not a perfect one. (Harry N.)
We had some AHA! moments while we were trying to gure out why one
strategy or one assumption wouldnt work. We rst tried to simplify it a lot like
determining if its 1D or 2D or 3D. We started by solving it 1D, then thought
we need more variables and moved on to 2D. (Karen N.)
Although more detailed in description than the rst three, Harry N.s description does
not give much more insight. He articulates that he had an idea about the solution shape
which he found to be erroneous, but it is not clear exactly what was the AHA! was for
him. Karen N.s description is similar in its vagueness. This may seem surprising because
AHA! moments are moments of clarity, so we might expect that the description of those
moments would also be clear. However, we know from Liljedahls work [33] that students
have diculty describing their AHA!s.
Two other responses suggest that although there were AHA! moments they were not
particularly helpful in arriving at a solution:
Sometimes, but then wed just run into another problem and eventually ran out
of time (Juliet N.)
There were multiple AHA!moments, but they ended up not being very helpful
in the big picture. We thought that nding out some of the information would
help us with modeling the problem but they ended up not being able to contribute
much. (Jamie N.)
For Juliet N. the AHA!s seem to lose signicance due to the fact that immediately she was
faced with other problems. Jamie N. says that the AHA!s did not contribute much.
Phil N. is unique in his response as he claims no AHA! moments. He explains that
everything went according to the plan:
No, everything went as planned. (Phil N.)
This is in contrast to his response to the previous question (Section 6.7) that he tried 3
dierent strategies, switching strategies due to complexity of the equations (Section 6.6).
While his previous remarks do not indicate that everything necessarily went as planned
initially, this response does suggest that plan was adapted as needed without a breakthrough
or AHA! moment.
6.9 Question 9: Learning to Do modelling by Doing it?
9. a) If you saw a similar problem, what would be the likelihood of you solving
it?
b) If you saw another unrelated modelling problem, do you think you are better
equipped to solve it?
This question is aimed at identifying if the students, having been through this modelling
experience, now feel better equipped to do more modelling. Experts agree that the way to
learn how to do mathematical modelling is to do it. This ethos is similar to that of how to
do problem solving. From Polya [34], to expert Sam Howison (see Chapter 4) the general
consensus is to learn to do it via experience. However, Schoenfelds [35] recurring theme
of metacognition informs us that there must also be a big picture view of what one has
done, and an eort to understand your own thinking process in order to improve it. This
correlates to Masons discussion that experience alone is not enough [...] REFLECTING
on the key ideas and key moments intensies the critical moments of an investigation and
helps to integrate their resolution into your thinking repertoire [28].
The students responses again went through the full range of answers, from complete
condence, to not being condent to say that a new problem would be solved. Phil N., and
Mark N. both respond with condence in their ability to solve any new problem, with very
concise answers.
a) Probably 100%.
b) Ya, even if it is completely unrelated, since any exercise in prob[lem] solving
helps solving problems. (Phil N.)
a) Probably I would be able to solve it.
b) I would denitely be mentally prepared for it. (Mark N.)
Phil N. echoes the premise that becoming a better problem solver comes from doing prob-
lems. Mark N.s response is of interest because he does not credit the process with increasing
his mathematical ability. Instead he explains that he would be mentally prepared for it.
Harry N., gives more detailed response to the question, which is a little o topic but is
still interesting.
a) For sure I would solve it...and I think my solution to this problem is quite
reasonable, but we do not have enough time to write the summary perfectly..
b) Yes... modeling is really fun for sure..I want to participate again..however,
Im going to graduate after this term. (Harry N.)
Harry N. gives the impression from his answer that he assumes that the modelling I am
referring to is only in the MCM. Thus his impending graduation, which will make him
ineligible for the MCM will stop him from doing any further modelling.
The other students interviewed give caveats on their ability to solve other problems. For
Juliet N. and Karl N. the constraint of time would have to be lifted.
a) Depends how much time I had
b) Maybe; simplifying the problem would be the key, but the more its simplied,
the more unrealistic it becomes (Juliet N.)
a) If giving us 2 weeks, maybe we can hand in a complete solution.
b) Maybe, because physics is not my comfortable zone, I am better at stochastic
process. (Karl N.)
Juliet N. indicates an understanding that simplication and capturing the qualitative be-
haviour are important themes in modelling. Karl N.s response implies that he would be
better at solving a problem that he was more familiar with. This ties in with the theme of
familiarity raised in the previous chapter (see Section 5.10).
Jamie N. Joe N. and Karen N. were all hesitant to say that they would denitely solve
a similar problem.
If I saw a similar problem I think I would have a better idea of how to go about
modeling it. I dont think I will be able to solve it easily but then I would know
how to start. If I saw a unrelated problem I think I would be better equipped to
solve it. Because I have seen a mathematical modeling problem and worked on
it with two other individuals, I have a better idea of how to go about solving it.
(Jamie N.)
a) I think I would do better the next time I tried a problem like this, but I
wouldnt be condent enough to at out say I can solve it.
b) Denitely. This helped a lot. (Joe N.)
We didnt actually try to solve it afterwards, or go into details to see where
we went wrong, so Ill probably have an idea of how to solve such problems, but
I still cant exactly solve it. However, Math 461 prepared us to solve a lot of
dierent real life problems, but theres still a lot to come. Every math problem
requires a dierent approach and dierent assumptions, but at least now I know
several dierent approaches to solve a Modelling problem. (Karen N.)
These students all seem to believe that the experience has better equipped them for solving
unfamiliar problems. However, being better equipped does not necessarily mean that they
believe they would be able to actually solve the problem. Jamie N. claims that she would
have a better idea of how to go about solving it. Karen N. claries that her ability to
solve the problem is problem dependent. This was also mentioned by a few of the experts
when asked how they deal with dierent aspects of mathematical modelling (see Chapter
4).
6.10 Summary
The novice modellers who participated in the MCM have a basic denition of mathematical
modelling. However, some of them assume that modelling problems must have more than
one solution method which, based on the expert responses, is not a necessary condition for
a problem to be a modelling problem. The novices, like the intermediates, used research
more than collaboration when stuck initially.
There was a wide range of emotions discussed upon seeing the modelling problems, with
some novices feeling condent and others feeling completely clueless. There was much less
autonomy than the experts on possible choices of the problems they were to work on. The
novices did not always have a realistic idea of the time that would be required to work on
the problem and did not have great collaboration skills, although they did recognise that
collaboration could be useful. They did however indicate in their responses that they were
willing to defer to others in the group and change strategies if they were stuck.
Many novices either stuck to one solution method regardless of the outcome, or switched
completely with no eort made to understand why they were experiencing diculty. They
quoted several mathematical topics that they deemed necessary to solve the MCM problem
they chose to work on. There was also no mention of checking that their solution was
correct, but this may be due to the time constraint of the MCM.
Chapter 7
Complete Novices
To complete the spectrum of modelling ability in this research, I observed, worked with and
acquired questionnaire responses from 45 students in 2 separate FAN X99 classes. The FAN
X99 course is structured so that solving mathematics problems is the main focus. Early
on in the course the students spend time working on basic number theory skills and their
understanding of patterns. However after the rst few weeks of this, the focus switches to
translating from worded problems to equations and graphs.
These students often struggled with mathematics as a whole, and any type of word
problem in particular. Here we are looking at the most basic type of modelling, where the
problem is worded in a real world context, but is nowhere as complex as those problems
encountered in industry. However the basic premise is the same: they are given a problem,
which they have to translate into some mathematical framework and solve, in order to
answer the question asked.
These students were given questionnaires at the end of their semester, with questions that
paralleled those given to the experts and intermediates in many ways. Some of the questions
however, were geared towards identifying any myths they might hold about mathematics
and solving word problems. As before, some illustrative responses will be given to highlight
the themes that emerge.
149
CHAPTER 7. COMPLETE NOVICES 150
7.1 Question 1: Time needed to solve a word problem
How much time do you think solving a word problem should take in general?
This question was intended to identify how many of these complete novices had preconceived
notions about the time required to solve problems. While the experts will speak of a years
work towards solving a problem, complete novices often believe all mathematics can be
solved quickly or not at all. This is seen in the responses to the questionnaire. Students
expect solving of problems to be on the order of minutes for the most part.
I sorted the responses in three main categories. The rst category consisted of ve-
minute intervals or less, or exact answers. Fourteen students believed the time to solve a
word problem was between 0 5 minutes, with four of them saying it would take exactly
or on average 5 minutes. Twelve of the responses give the time range from 5 10 minutes.
This includes one student who simply says: minimum of 5 minutes; another who claims:
maximum of 10 minutes; one who species a range of 7 10 minutes; and three others who
give 10 minutes as their exact answer. Three responses put the time range between 10 15
minutes.
The second response category was either a time interval of more than ten minutes, or
two separate time intervals for easy and hard questions. In the larger time intervals there
were four responses: 2 20 minutes, 2 30 minutes, a maximum of 20 minutes, and from
1 to more than 10 minutes. There were three responses that had dierent intervals for easy
and dicult problems as follows: 1 2 minutes for an easy problem, 5 10 minutes for a
hard one; 1 2 minutes for an easy problem, 4 5 minutes for a hard one; and nally 1
minute for an easy problem and 6 10 minutes for a hard one.
The nal category was answers with no numeric value given. For nine of the students it
was dicult for them to pin down an actual number for the amount of time they believed a
problem should take. These responses can be further separated. For two of these complete
novices, they did not believe it should take long to solve a word problem in general:
Not long if you know the questions method.
No more than a few minutes.
Because they have not given a time, we do not know exactly what is mean by not long
and a few minutes. However we can assume here that the time frame will be in the order
of minutes. This is in sync with the previous responses and makes sense for these students.
At this stage in their mathematical career they would not have seen anyone working on a
problem for longer than a few minutes. If they themselves are unable to solve the problem
in minutes then they usually give up or seek help.
What is interesting is that the rst response indicates a recognition that the amount of
time is really problem dependent. This is also seen in the ve following responses:
Depends on the length of the problem and how many steps are necessary. No
standard time.
It depends upon the complexity.
Depends on skills needed. Harder skills means more time.
I believe that it is unfair to set a given time for a word problem, as there can
be so much variation between diculty of each.
These students may still have an idea that the time it takes is in the order of minutes,
but realize that they can not put an actual time as this depends too much on the type of
problem and the skills required to solve it. The fth student quoted above not only refers
to the time being problem dependent but goes even further to explain that it is unfair to
set a given time.
The nal two students who responded in this manner both explained that the amount
of time is however long it takes until you solve it. For the rst student the idea is that the
context in which the problem is given will determine the amount of time one should spend
on it:
However long it requires, depending, though on reason for answering them. On
an exam, take time but dont take too long. If homework, do it until one is sure
that one has no idea at all how to do it.
Until you solve it. There are still unsolved math problem[s] which have been
around for 200 years. Maybe the semester should be that long. Just kidding.
The second student makes an important point that there are problems that have been
unsolved for centuries. This time scale is completely dierent from the other students; the
rst student provides no specic time frame, while the second student provides a time scale
in the order of years. Both students explain that the time frame is problem dependent as it
depends on the reason for solving the problem or whether the problem has been unsolved
for a number of years.
7.2 Question 2: Initial Thoughts and Feelings
What are your rst thoughts/feelings when you see a word problem?
This question parallels the second interview question given to intermediates and experts
alike. What was noticed was that the experts had negative and positive feelings in almost
equal measure to this question, while with the intermediates eight out of eleven of them
reported negative feelings. The rst thing to note with this question is that the complete
novices responses are very dierent from those interviewed. For the experts and intermedi-
ates, thoughts and plans are distinguished from the feelings. For the complete novices, the
thoughts are often a reection of their feelings. So we will separate this section into three
categories. First we will deal with positive thoughts and feelings, then negative thoughts
and feelings, and nally those who observe a transition in their reaction to rst seeing a
word problem.
Positive Thoughts and Feelings
While one student reported that his rst thought upon seeing a word problem was Yes!
They are fun, for the majority of students who had positive thoughts, they were more along
the lines of a plan of action for solving a given problem. Some of them asked questions:
What is the question I need to solve?, What do I have to use? or How do I solve this?
Others spoke of trying the problem, whether it is trying to nd familiar aspects, trying to
understand or trying your best:
I try to see what familiar aspects of the problem I know.
Understanding. Read through the question thoroughly and do your best to com-
prehend what is being asked.
Lets see if I can gure this out.
Try my best to solve it.
This desire to identify the familiar aspects of the problem mirrors expert Sam Howisons
rst thought upon seeing a problem: One of the things I always I suppose I always think
about is, what do I know thats a bit like this? What have I done thats related?
One student reports his realization of the diculty of the problem while still having a
plan on how to solve it:
This will probably be dicult for me, but I should start with one step at a time.
This comment is particularly intriguing coming from a complete novice as it reects expert
Colin Atkinsons view on dealing with problems: I think youre worried and curious yeah.
But you have to deal with it. You have to begin things thats how you progress.
Four out of the 45 students spoke of positive feelings explicitly. The experts expressed
excitement, interest and curiosity upon seeing modelling problems and we note very similar
feelings here. Students speak about feeling excited and curious as well as challenge or
intrigue. One student elaborates further on mathematics problems being compelling:
I am compelled to solve the problem, I end up feeling self-gratied and happy
that I am able to solve a problem.
These positive emotions associated with solving mathematics aect people of all levels of
expertise. Experts and complete novices alike feel the curiosity, excitement, the challenge of
the problem, compulsion to get to the end, and then that happiness and satisfaction when
the problem is solved.
Negative Thoughts and Feelings
From this study we have seen evidence that there are modellers of varying expertise (from
expert to complete novice) that have positive feelings towards modelling. Similarly negative
thoughts and feelings are evidenced across the spectrum. Experts spoke of wondering if they
could solve the problem, intermediates of being overwhelmed. As expected complete novices
have a range of negative thoughts associated with their mathematics word problems.
The rst theme we see here is that of confusion: This is hard,Im confused. This ties
in nicely with the second theme raised by the complete novices of a lack of understanding:
Is it one I understand? I hope it is...
Hope I understand how to do it.
Uh oh, how should I begin this...
My rst feelings are of sadness. My rst thoughts are of how I am going to do
this.
The use of the word hope in the rst two above quotes may be misleading, as it speaks to
despair and fear of not understanding more than it does to actual hope about the problem.
Not knowing how to begin mentioned in the third response above is also another big issue,
which is related to the third interview question (see Sections 4.3 and 5.3). If students are
unable to start then it is impossible for them to make any progress. We might also look
at the last quote as simply a plan of action, but coupled with the feeling of sadness gives a
strong negative feeling associated with the thought.
Other responses also convey a sense of dismay or dread upon rst seeing the word
problems:
Oh no, not this again. Oh well.
OH NO!!! WHY CANT WE JUST BE STRAIGHTFORWARD!
Argh. So much work and reading.
Yuck.
Oh crap.
The HORROR!
I hate math/ this is going to take forever.
This is going to be hard - depends on diculty or if I can see what to do right
away.
Expressed in these comments is the worry that the problem will be time consuming, that
its not straightforward, and that it is automatically hard, if the solution cannot be seen
immediately. These are perceived as negatives because the expectation is that a mathematics
problem is good, or a mathematics student is smart, if the problem seems straightforward
and obvious. Many students also expect problems to be solved in 12 minutes as we have
seen in the complete novices responses to question 1. This is contrary to what the experts
have demonstrated, where the interesting problems are those that are not straightforward
(see Section 4.10).
Coupled with these thoughts are the twelve responses on negative feelings. Students
speak of feeling everything from Panic to Not happy and scared. Students go on to
explain that they hate word problems, and feel confused and stressed or dreadful and
confused. Some students elaborate further on the cause of their negative feelings:
Annoyance, angst. Especially if I dont understand the question.
Nervous! Or I feel like I cant solve it. Word problems always seem impossible
to solve when I rst read them.
Panic, Im not a fan of math and word problems tend to make me more nervous
and confused, I begin to question myself.
Again this is interesting because we know that the expert mathematical modellers express
these exact sentiments:
Um, well, feelings! That doesnt help! You know, how am I gonna understand
this? I have no idea what the words mean. I dont know anything about this.
(Sam Howison)
Feelings, its kind of a joke but, I hate this problem for being hard, I hate my
self for not being smart enough to know immediately how to write it down, or
some variation. (Thomas Witelski)
Finally one student explains that he feels numb when dealing with word problems:
Numbness. There is no point in getting emotional. It must be done so do it.
This is reminiscent of expert David Murakis statement about his feelings: Ive learnt to
suppress research anxiety (Section 4.2).
Transition of Thoughts and Feelings
Eight of these FAN X99 students describe a transition in their feelings or thoughts towards
word problems. One student simply explains that it varies without any further clarica-
tion. Three others describe the change they go through when viewing a single problem:
Ahhh ... No, wait Oh, this is ne (now that I have read the question).
Annoyance/Panic. Then curiosity.
Eek. Give me some time [and] I can do this. Have I seen something like this
before?
These students describe a feeling of panic, which then turns into a more positive thought or
feeling, whether it be a feeling of curiosity or a more calm state, or a plan of action. This
is a bit dierent from the student who feels a dierence depending on where the question
comes from:
In FAN X99, Lets do this! In a dicult course Oh boy...
For the nal three students the transition has taken place throughout the course. They
explain how they felt discouraged at the beginning of the course when seeing problems, but
have now reached a stage where they are not as daunted:
I used to think Oh crap but after FAN X99 class I think Now I have to think
more.
Before FAN X99 = discouraged to try. After FAN X99 = Willing to get an
answer.
Depends on the question, for the rst half of the course it was shoot me and
now I understand & enjoy them.
7.3 Question 3: Initial Barriers
What do you do if you have no idea where to start?
This question is again a parallel to the corresponding interview question. Experts and
intermediates suggest several dierent methods of dealing with the non-starting situation
in modelling. These include trying to understand the problem better, talking to others,
researching, just doing something, and solving a simpler problem. Experts also discuss
taking a break and coming back to the problem when stuck.
For me it was unexpected that many of these complete novices expressed some of these
expert or intermediate behaviours. Themes of trying to understand the problem, simplifying,
and looking at related problems, just doing something and talking to others and taking a
break were all raised. Students also discussed the use of drawing a picture as a heuristic to
help them when stuck initially. A more in-depth look at each of these themes showed the
dierence in expertise in the way each of these themes was approached.
Understanding the Problem: Re-read and Write
Experts go about trying to understand the problem via research and discussion/collaboration.
We will deal with the theme of collaboration later, but it is clear that reviewing the liter-
ature, i.e. looking at research papers, books and searching online, cannot be practiced in
a classroom setting of complete novices. Four students instead spend time re-reading the
question:
Break question down slowly. Re-read question.
Read the question over multiple times.
I re-read the question, then think for a moment. Sometimes that works. If not
then I go by trial and error. If it still does not work I ask for clues.
Try to re read the question and write down what I know.
These are all good practices for understanding the problem. What we must note here is
that for many of these students re-reading does not necessarily guarantee understanding.
This explains why the question may need to be read multiple times. To avoid this, the
alternative strategy of listing the information given from the question is also used by six of
the complete novices:
Write down the objective and information I have.
Re-write the question in point form.
Listing out what I know (what the question tells me) and hopefully something
comes to me. Or work backwards (guessing).
Dene facts about the problem. Think.
Pull out the information.
I list everything I do know.
Note that guessing or trial and error is mentioned here, as a strategy if re-reading the
problem or listing what is known does not yield a plan of action. These come under the
heading of: just doing something, and will be addressed later in this section. Two students
also mention that they think. This is noteworthy, because of the tendency for students
to rush to try to solve as opposed to taking time with the problem to ensure understanding
(see Section 7.4).
The experts do not mention re-reading and writing out the given information. This does
not necessarily mean that they do not practice this, but that it is not signicant enough for
them to note it. An expert would not consider himself stuck at the beginning if he had not
rst fully absorbed the problem and gone through it so that he is aware of the information
given. This dierence in perspective shows how complete novices view being stuck. There
is an expectation of being able to understand exactly what to do at rst glance. A complete
novice can therefore consider himself stuck upon viewing the question once, if he is not
completely sure of what the question is, or what information is given.
Simplify
Another way to attack a problem is to simplify it. The theme of simplication was mentioned
extensively throughout this work and is raised by experts and intermediates alike. As seen
from the rst quote in the previous section, Break question down slowly. Re-read question,
some of the FAN X99 students also recognized that this could be helpful. This simplication
is described in two dierent ways. The rst way is to attack the problem a bit at the time
mentioned by two of the complete novices:
Break up the question into parts [and] understand each part slowly.
I try to break down the main things to create an equation.
Here by dealing with individual pieces of the problem these two complete novices can un-
derstand each part slowly or create an equation. The second way to simplify the problem
is by solving a simpler version of the problem which is suggested by three students:
Give up. But then I realise if I break it down or try to solve a simpler problem
it will help to get at least somewhere.
Write out all the information given in a simpler form.
I try drawing a picture or use smaller numbers to start.
This last quote is most illuminating here, trying smaller numbers makes the problem easier
for them to deal with or identify patterns or structure. Simplifying the problem shows a
mathematical maturity and it is therefore not too surprising that such a small proportion
of students mention it.
Draw a picture
This heuristic is taught to the students taking the FAN X99 course. This is in correla-
tion with the high proportion of expert mathematical modellers interviewed who mention
drawing a picture as their number one means of getting entry into a problem.
Draw a diagram.
I try drawing a picture or use smaller numbers to start.
Look at the numbers, draw a picture, put the numbers in dierent orders. Skip
the question and come back to it later.
Leave it come back or break it down or draw.
Draw diagrams, search notes for similarities, brainstorm w/ classmates
Try to draw a picture or write an equation.
Notice straightaway that in all but one of these quotes, drawing is not the only thing that
is attempted. Two students combine the theme of drawing with that of leaving the problem
or taking a break. One other student suggests drawing along with collaboration and nally
we see drawing as an alternative to writing an equation.
Even though this heuristic is taught, it does not seem to be one of the most used based
on the fact that only six complete novices suggested it. This could be because the complete
novices do not see drawing as a real option when they are stuck. This would also explain
why one student responds that when stuck at the beginning he/she would try and make
a let statement and then nd an equation. This course of action was surprising to as
it is unclear how one could do this if one was indeed stuck. But it again illustrates a
desire to have an equation, which may be viewed by complete novices as having done some
mathematics.
Just Do Something
Experts Colin Atkinson, Sam Howison and Lou Rossi mention starting somewhere, starting
writing and trying things, when stuck initially. Six FAN X99 students also talk about just
trying something if they are stuck at the beginning of a word problem:
I start by randomly using theories that I remember - if its fractions, I will draw
visuals, if its a linear equation, Ill draw a graph.
Keep trying random ideas until they mesh.
Start somewhere and gure your way through it.
Honestly I just start with a random equation and hopefully I gure something
out.
Trying out dierent things and maybe wait for what the correct way is.
There is a dierence between the experts who start trying things and the complete novice
who claims to start with a random equation or to keep trying random ideas. Experts are
generally not working at random. While they may start somewhere that does not guarantee
a solution, there is still a method and reason for their choices. This is not to say that
the complete novices do not gain some measure of success from their work, but it must be
rational or systematic to be considered expert behaviour.
Similar/Related Problems
Solving similar problems is a viable heuristic mentioned by experts and intermediates alike
that was also mentioned by four of the complete novices:
Think of similar questions I have previously done and try to apply those to the
question.
Think back to lecture, see what I learned, and gure out how they relate.
Try think of all the ways to solve the problem that I know and just meditate on
it a while.
Draw diagrams, search notes for similarities, brainstorm with classmates.
These similar problems are not always problems solved by the students themselves but may
be problems that they have seen their instructor do. Once again we note that the majority
of the FAN X99 students (41 out of the 45) did not choose this as a means of accessing the
problem.
Collaboration: Ask for Help
The theme of collaboration is slightly dierent in the case of the complete novice. Experts
often speak of collaborating with others to clarify areas where they are stuck, as opposed
to asking others to solve the problem for them:
Otherwise talk to other people. They may not know how to do it, or be able to
tell you. But just knocking around ideas. Its a reective process. (Lou Rossi)
Experts use language including terms such as collaboration, talking to others and calling a
friend. In contrast complete novices often speak in terms of asking for help, and waiting for
the correct answers:
If I have no idea where to begin I ask for assistance of some sort.
Ask for help. Attempt to highlight key points.
Ask teacher or classmate.
Ask someone.
Wait for the teacher to explain or ask a neighbour.
Try plugging in what I can, if any, and then wait until someone else gures it
out.
Skip it and ask someone later.
I ask for help but if I dont get it right away I do nothing.
Trying out dierent things and maybe wait for what the correct way is.
This request for assistance is more passive than the term collaboration implies. There is a
tendency to ask or wait for solutions methods from those they consider experts or having
the answer. There are of course exceptions to this seen among some responses given, but
these are a minority:
Draw diagrams, search notes for similarities, brainstorm with classmates.
It could be argued that experts are behaving the same as novices when they ask for help
from the person who brings the problem. And indeed to some extent this is true because
the mathematical modelling expert is oftentimes a novice in the eld of expertise of the
problem and so must get some insight into the problem by asking for help.
However, there is a dynamic and an unspoken contract between the instructor and
student that is dierent from that between the expert in the eld and the mathematical
modeller. In the classroom it is expected that the instructor, who brings the problem,
already knows the answer. Even if they claim not to know, they have no real need of the
students to get a solution. This is very dierent from the relationship between modeller and
expert in the eld. While the expert in the eld is knowledgeable about the terms and the
problem itself, they also view the modeller as an expert in being able to solve the problem.
Thus there is a mutual respect of each others abilities and the collaboration is more active.
Take a Break: Move On
Another suggestion by the experts if you are stuck initially, is to wait it out (David
Muraki) or sleep over it (Reinhard Illner). We have discussed before that this idea of
taking a break when stuck is modelled after Hadamards incubation step in the process
of mathematical creativity. However it must be preceded by the preparation stage, which
requires serious conscious eort. When complete novices speak of taking a break, do they
mean the same as the experts or are they simply giving up on the problem?
One student gives the impression that he/she makes an attempt to solve the problem,
before leaving it and coming back to it:
Look at the numbers, draw a picture, put the numbers in dierent orders. Skip
the question and come back to it later.
For two others, moving on from the problem seems to be a means of abating the anxiety
associated with the problem as opposed to allowing time for the subconscious to work on it:
Move on to the rest of the test and come back to it when Im not stressed.
Panic, then move on to the next question and come back.
From the rst quotation we see that the student is assuming that the context for solving
the word problem is in a test situation. It should be noted here that moving on from a hard
problem in a test situation and coming back to it was a test-taking strategy taught to the
students. Four other complete novices do not clearly indicate if any attempt was made to
try to solve the problem rst, or that upon re-visiting the problem if they typically have
any more success than the rst attempt:
Skip to the next question and go back to it later.
Skip the question & go back to it later.
Keep trying or skip it and come back when all others are done.
Leave it come back or break it down or draw.
While the above seven students say they will come back to the problem four others make
no such claims. Their way of dealing with being stuck is to move away from the question
altogether. The answer can be acquired through others or not at all:
Move on.
Re-read the question, give up, cheat.
Skip it & move on to the next problem.
Skip it and ask someone later.
For these last four students in particular it is clear that they have simply given up. The
theme of moving on or skipping the problem like the previous theme of asking for help is a
very passive way to deal with being stuck. It is therefore not surprising that these are the
two most popular means chosen by the complete novices with each being chosen by eleven
students. (Compare this to solving a simpler problem chosen by seven students, re-writing
the problem, drawing a picture and just doing something at random chosen by six students
each, and re-reading the question and exploiting a similar problem chosen by four students
each). The tendency to wait passively for help or simply give up therefore aids us in dening
the attitude of those who are inexperienced.
7.4 Question 4: Plan or Plunge?
If you know what to do, do you plan your solution or do you plunge into it?
In the FAN X99 classroom the questions given generally only contain the information rel-
evant for solving the problem. Thus there was no point in asking a parallel question to
interview question number 4 regarding how to determine what information was relevant to
the model. However, of interest is whether complete novices take the time to plan their
solution before putting it to paper.
Experts working on modelling generally are working in groups, and have to plan and
delegate what will be done. Intermediates talk about recognizing that they should plan, but
often opting to jump straight in rst any way:
Yeah when I started, it was just like, BOOM jump into the big problem cause
thats what I wanna solve (Liv I.)
A lot of time I just jump in cause Im like, Oh I can do this. Nope, nope you
cant! Take a step back! So it would be a lot more eective for me if I didnt just
jump in with both feet right away. (Isabel I.)
Liv I.s response indicates a level of excitement and a desire to solve the problem causing her
not to stop and plan. Both of these intermediates go on to speak about learning patience
with the problem as they have become experienced. This question about plunging into the
solution helps us to identify if complete novices tend to jump right in without thinking the
problem through, and their rationale for doing so.
Plunge In
Of the 45 FAN X99 students that responded to the questionnaire, a majority of 28 explained
that they plunged in once they knew what to do. Of those students many just responded
by saying plunge or plunge into it. But for many of them this was not enough. Some
students emphasized their answers with exclamation marks: Plunge! one by using capitals
only: PLUNGE and also, Plunge in baby! This shows a level of excitement in the
response, which can be interpreted that this action of plunging in and solving is pleasing to
do.
Other students go on to explain a little further about what they do when they plunge
into the problem:
If you know well enough, just plunge in.
I always seem to plunge into it if I know exactly what to do.
In these rst two responses the participants clarify that this action is taken because you
know what to do.
If I know what to do I do it as quickly and thoroughly as possible.
I plunge into it in an organised fashion! :)
1) plunge into it 2) step back, review 3) try again 4) cross reference.
I mostly just plunge into it, though I usually keep in mind the steps I need to
do.
These four responses indicate that the participants are aware that plunging in is not nec-
essarily the best way to go about solving the problem. There are clarications about the
thoroughness and organized way in which they do this, as well as speaking about re-
viewing and cross-referencing. The idea of stepping back, reviewing and cross-referencing,
is expert behaviour, seen in Polyas [34] heuristic framework.
I plunge into it by solving it out & seeing if it works
I plunge into it and if it works, it works and if it doesnt I look back and see
where I went wrong.
Plunge into it. Therefore I make stupid mistakes
These last three responses clearly indicate that the participants are aware that though they
may believe they know what to do, they may still make mistakes by plunging in. This
accounts for the stupid mistakes as well as the need to see if it works, and where it
went wrong.
Plan
Ten participants claim to plan even if they know exactly what to do. Here there was no use
of exclamation marks or uppercase letters to convey enthusiasm. Most of these responses
were simply: plan. However four students explained their reasons for not plunging in
without a plan:
I plan my solution so to allow straight thinking
It depends on the question, but planning a solution usually has the best results
Plan out my solutions to get appropriate marks
Plan/organize rst to avoid making mistakes
These students have realized that by plunging in they are more likely to make mistakes,
lose marks and not give a clear answer. However there is a caveat to always trying to avoid
mistakes which is summarized by Liv I.:
I think one thing that holds a lot of people back is not being able to jump in.
Like when youre, when youre unsure of something, youre afraid to jump in.
So I think one characteristic would be people who are good modellers are not
afraid to make mistakes. And try, like just try something, you have an idea try
something. (Liv I.)
Other
The remaining seven students did not give a clear answer as to whether they tend to plunge
in or plan. Two students claimed to do a bit of both. Two others explained they also did
both but had a preference for one or the other:
Both. Mostly Plunge
A little bit of both, I plan as I am starting
The nal three are more ambiguous with their answers and are probably the most accurate
in describing their process:
I dont really plan, but I try to write down all my steps as I go.
I do all the work to the side of the question and then write it in a more clear
format.
I plan in my head while I plunge into it.
What we do not see in response to this question is a description of planning, trying, evalu-
ating, rening, and then evaluating again, as we would expect the experts to do.
7.5 Question 5: Organisation of work
Do you feel your work on word problems is generally organized or disorganized?
This is a follow up question to the previous one. Do students who plan consider themselves
organized? Do the plungers think of themselves as disorganized? Here it is my aim to
identify any correlation between questions 4 and 5.
Twenty-eight of the students described themselves as organized, while eleven of them
said they were disorganized and six of them were unable to clearly state which word best
described their work. As it was a follow up question, I noted that sixteen of the 28 that claim
to be organized also said that they plunge into a problem as opposed to planning. Only
seven of them were self-proclaimed planners. There were also more of those who claimed to
plunge into problems in the group of disorganized students, with eight of them saying they
plunge and only one student claiming to plan his solution.
There are several reasons and clarications given on why the work is organized, including:
I try to keep it organized so that my thoughts are easy to sort through.
Organized if I know how to do it.
It is generally highly organized, even if it is wrong.
Similarly some of those who explained they were disorganized had further clarications to
make:
Its all over the place because I use mental math to retain information. There-
fore, my work will be everywhere.
Disorganized until answer is solved.
Disorganized but Im good at them.
There was also the group of students for whom it was not easy to categorize themselves on
group or the other:
Id say depending on my understanding of the question.
If Im running out of time = disorganized. Beginning of the problem = orga-
nized.
Depends if I know the answer. If it is a simple method, it is more organized.
It becomes messy when I dont know right away.
Id say its improved since this course.
Fairly organized, though sometimes scattered.
Somewhere in between.
The rst two of these are self-proclaimed planners. This seems to t with their descriptions
of organization, particularly for the second student in this group. At the beginning of the
problem he/she is organized and plans the solution. However from the data there seems
to be no real correlation between a student thinking that he/she is organized and whether
he/she plans a solution or plunges.
Upon further examination of the responses four interesting comments seemed worthy of
discussion:
My work I feel is generally organised (despite my sometimes messy handwrit-
ing).
The actual work is disorganized but I rewrite it with more clarity.
My thought processes are organised. Written work unorganized but that because
Im generally a slob.
Organized because my writing is good/big.
These comments about handwriting raise questions about the meaning the students ascribe
to the term organized. Which of them are referring to neatness of presentation, and
which of them are referring to organization of ideas? This ambiguity in these students
understanding of the question makes it dicult to know exactly what any of the students
mean when they say: organized or disorganized. This also highlights the fact that they may
not be answering the question which is being asked.
7.6 Question 6: Dealing with Being Stuck
What do you do if you are stuck in the middle of solving the problem? How
does this make you feel?
This question is designed to mirror the corresponding interview question 6 on being stuck.
In the interview question I did not ask about feelings as I had already dealt with the feelings
surrounding being stuck in the second question directly. However I thought that for the
complete novices several opportunities might be necessary to talk about feelings, this was
reinforced by the way feelings and plans were tied in the responses to the second question
above (see Section 7.2).
Experts and intermediates alike discussed several means of dealing with being stuck
in the middle of the problem. These were collaboration, research, simplication, applying
a dierent approach including starting over, taking a break as well as persistence. The
complete novices that responded from the FAN X99 class spoke to several of these themes,
which will be discussed below.
Feelings
The experts were not forthcoming with their feelings when asked. The intermediates re-
quired less prompting, but often seemed surprised at being asked about their feelings. It
should be no surprise then that the complete novices generally spoke about their feelings,
sometimes at the expense of explaining what they do when stuck in the middle of solving
a problem. There were however still 5 students who did not discuss feelings, only their
strategy for becoming unstuck.
The overall feeling associated with being stuck is one of frustration. Fifteen students said
that they were frustrated when stuck. Others explained that they felt annoyed, panicked,
bad, angry, low in condence about their mathematical ability, stupid, nervous, uneasy,
stressed, horrible and unsure, or to put it in the words of one student:
I dont like not knowing how to do a question
I will look at all other quotations about feelings in conjunction with their associated plans
for getting unstuck.
Taking a Break: Move on and come back or give up?
Sixteen students chose the option of moving on to another question, or taking a break. Some
of these students chose this option after trying to go over the problem:
Start over and if it still doesnt work I generally give up. Makes me feel SO
frustrated.
I double check my previous work, something must have gone wrong there. If
not, then I move on.
I get frustrated & possibly move onto the next question or start over & check
my work for mistakes.
But for sixteen of them the only option mentioned is to move on from the problem:
If I get stuck I move on. I feel bad because I dont like giving up or saying Ill
come back to it. I think that is an excuse.
Skip and go back to it later. Frustrated.
Come back after. Sometimes I over-think.
I go back to it later, it makes me feel stupid.
Its annoying, so I leave it and do something else for a bit.
It makes me feel horrible! I just leave it, go through the rest of the exam and
return to it in the end.
Note that for the last student quoted above (and possibly others), the context is again
assumed to be a test situation, where they have been taught to move on and come back as
a test-taking strategy. For these above students the intent is to come back to the problem.
But as discussed in Section 7.2 moving on can mean giving up on the question altogether,
at least for seven of these students:
It makes me feel frustrated and I want to give up.
Uneasy. Nervous. Solve a dierent problem.
Take a break and go back to it. Arg!
I usually move onto the next one and come back to it. It makes me slightly
worried.
I usually stop working on it. It makes me feel frustrated.
I feel frustrated when I hit a wall. I usually get upset & just stop working all
together.
I get frustrated and move to another problem.
This is a striking dierence between those with more expertise and the complete novices.
The option of simply giving up is never mentioned by the intermediates or experts. This can
be attributed to ability and experience, as well as the situation in which experts or novices
are dealing with the problem. For the experts in particular, any problem that they choose
to work in is one of interest. This is very dierent from the situation of these complete
novices who have no choice in which problems they solve.
However, intermediates may nd themselves in a similar situation of solving problems not
of their choosing or liking and they also did not discuss the possibility of giving up. This can
again be attributed to the dierence in their situation. Even though an intermediate may
not choose the problem, he/she is aware that doing the problem is in their best interest. For
the complete novice often the problem is dicult, not of interest, it is not of their choosing
and they do not see how it is useful for them to be able to do it. It is therefore viewed as a
futile exercise.
Go over the problem
One of the themes raised by the experts in Chapter 4 is that of starting over when stuck.
Fourteen of the FAN students described starting over as their means of trying to become
unstuck. For ve of them, this entails re-reading the question and looking through their
steps:
I go back to the beginning and re-read the problem to see if something will click.
Its frustrating.
Panic then look through my steps and re-read the question. I feel uncondent.
Re-read a question. Feel nervous.
I retrace my steps
If I cant gure it out after going back and taking a 2nd look, I usually go into
a rage. Very angry.
There is no indication here what the purpose of looking over the steps is or what re-reading
the question will do. One can assume that there is a hope that by re-reading the question
they will get new information. This is similar to researching, or going back for more data
as described by the experts, although it is not articulated here.
Those who claim to retrace their steps do not express that there is an expectation that
they are doing anything dierent. Other complete novices explain that they start over
completely:
Start over and if it still doesnt work I generally give up. Makes me feel SO
frustrated.
I start over, or I get myself unstuck.
If I get stuck I go back to the beginning, and work it through again. But I would
be annoyed.
This is more useful than simply retracing steps, as one may not make the same mistake
twice. Another way would be to go through the problem again looking for mistakes being
made:
Annoyed. I take a moment to think through my steps, both before and what
to do next. I also check to make sure I did the previous steps correctly. I also
re-read the question.
I get frustrated and then breathe. Afterwards I keep the work I have down and
look over it to see where I could have gone wrong.
I double check my previous work, something must have gone wrong there. If
not, then I move on.
I get frustrated & possibly move onto the next question or start over & check
my work for mistakes.
Start over to see if I made an error along the way. Frustrated.
However this does not account for the fact that the initial method may be completely wrong.
For the expert, starting over does not mean simply, solving the problem the same way a
second time, but re-thinking assumptions and solution method. This was only articulated
by one of the FAN students:
I go back to the beginning and try again. If it still doesnt work I try it again
in a dierent way. It bewilders me mostly.
Collaboration: Ask or wait for help
Collaboration was one of the main methods used by experts and intermediates alike when
stuck in the middle. Experts go back to the one bringing the problem, or simply collaborate
with their colleagues. Intermediates often talk to supervisors, or others within the group.
What is surprising here is that only four out of the forty-ve FAN X99 students oered this
as a solution for being stuck in the middle of the problem:
If Im in class, I will wait until we go over the problem, otherwise I ask a friend
or even sleep on it if I have time. It makes me feel like I am poor at math.
I ask. Makes me feel like I need to study more.
It annoys me! I usually ask for help from someone around me or the prof.
I try to keep going, if I cannot do much more, wait for the correct way. I feel
unsure.
This is also in contrast to the eleven who recommended this plan of action if stuck at the
beginning. Indeed, eight of those eleven students who would ask for help if stuck at the
beginning of the problem, do not choose this option if stuck in the middle. This contrast
highlights the fact that, at the point where experts deem collaboration the most useful,
complete novices seem to choose to avoid asking for help or even waiting for help.
Other
The FAN X99 complete novices raised a few other themes. The rst one that we will look
at is persistence:
I would keep trying no matter how frustrated I become.
Hadamard explains that continuous persistence is useful to a point [32]. The novice does
not explain if he/she would keep trying the same method that is not working, or several
dierent methods. The theme of trying something dierent is raised by two other students:
Try new methods. Frustrated.
Try something else. Like any other time you get stuck on something.
The second quote above reveals that the student recognizes that the skills required in math-
ematics class parallel those in many other situations.
Other students mention trying to identify where the mistake could be, working back-
wards, and refocusing, as strategies when stuck:
I try to think where I may have gone wrong and then proceed to slowly try and
x it. It makes me feel tense and at times angry at myself.
Ill work backwards [and] see where I went wrong. Flustered.
Happy of course. I love coming up short. Jokes. I usually get a bit frustrated
then refocus.
Having several methods for getting past barriers in an open mathematics problem is vital.
Experts and intermediates often mention several that they have found useful. These com-
plete novices more often report only one thing that they do when they are stuck. Most
importantly, at this stage of the modelling process, experts seek active collaboration with
each other more than any other strategy. This is lacking in the complete novice responses.
7.7 Question 7: Problem Solving Heuristics
What sorts of problem solving techniques do you use?
(eg: Draw a picture? Work backwards? Solve an easier problem?)
Use specic problems from the course to illustrate if you can.
This question was aimed at identifying the heuristics used by these complete novices on
word problems in class. I once again only gave three in order to avoid overwhelming the
students and to be more consistent with the other data gathering questions.
Draw a Picture
An overwhelming 36 of the FAN X99 students (80%) cite diagrams and pictures as one
of their preferred heuristics. This seems to be especially the case for, but not isolated to,
dealing with fractions:
I found drawing a picture to be very useful when dealing with percentage prob-
lems and nding fractions of fractions.
Drawing a picture makes the problem easier to understand [gives a drawing of
a pole with the caption: a pole is pushed some % into the ground how far...].
I always draw pictures (like the monkeys in the tree problem or the swimming
pool problem).
Percentages = charts/pictures = visual. Visual diagram works best.
I look at the numbers, get a feel for the range of the answer. Then draw a
picture and gure it out visually.
Whatever I need. Diagrams can be helpful. i.e. You cut 3 equal pieces o
a wooden plank 180 cm long. Develop a formula. Id draw a plank then do
something like [divides rectangle, marked 180, into 4 pieces where 3 are equal
size and labelled x] = 180 - 3x.
Drawing pictures really helps. Like graphs or fractions. Example [draws 2 cir-
cles divided into quarters with 5 quarters shaded].
First of interest is the fact that for 12 of these 36 students, drawing a diagram is the only
heuristic that they mention. Secondly, many of them are able to give specic problems
and make use of illustrations to show how they would use pictures and diagrams to solve
these problems. This shows as clearly as the number of students, how fundamental drawing
pictures is for them. This may seem at odds with the fact that only 6 of them mentioned
drawing a picture when stuck. However, this may be a technique not reserved for being
stuck, but used before being stuck on the problem. We also cannot discount the possibility
that of the three heuristics mentioned, this might simply be the preferred one.
Work Backwards
Six of the forty-ve FAN X99 students that participated spoke about working backwards.
This skill is a little more sophisticated, as it requires recognition that it can be used in a
particular situation. Note that working backwards is not mentioned by itself in any of these
situations, as it is very problem dependent:
Sometimes Ill draw a diagram or work backwards, it depends on the problem.
Pictures always help. Rewrite key points. Cross ref. Reverse order.
I draw a picture, and label. Sometimes I work backwards. Often I then try to
solve it algebraically if possible.
Draw a picture, try to work backwards. i.e. The problem about who did the
fencing in the garden I tried to map it out [illustration of a square divided up
several times].
Questions about mass and/or volume, I draw a picture. Questions about per-
centages, I work backwards (ex: starting with the result). Questions about
graphs/functions I draw a graph if I can or list values. In most other cases
I draw a basic diagram and/or solve an easier problem.
I use all sorts of techniques:
-illustrations,
-equations,
-working backwards.
However many of the experts did not cite working backwards as being a preferred skill
either. And one FAN X99 student explains that he rarely uses this heuristic:
I often draw a picture or solve an easier question. I hardly ever work back-
wards.
Solve a Simpler/Similar problem.
Simplication is a major theme revisited many times by experts and intermediates as well.
FAN X99 students also touched on solving an easier problem, as seen in the previous quote,
as a means of getting unstuck. Nine FAN X99 students speak to it as one of their problem
solving strategies:
I draw pictures, solve easier problems, do anything to avoid long division. More
but hard to remember unless faced with a problem.
I use all those techniques listed but I nd memorizing an easier problem helps
most because its easy to go back to the basics.
Draw pictures & sometimes solve easier problems.
Sometimes I use drawings to help but I do try & solve easier problems rst.
Picture [and] the easier problems within the big problem.
Once again notice that this heuristic is not isolated. Drawing a picture is often another
option used by these students. Another way to simplify the problem is breaking it down
into easier parts:
I can use diagrams sometimes, but I like to break down the main components
to create an equation.
I usually do the problem in pieces and then put it all together.
Break up the problem and try to solve.
Two other FAN X99 students spoke of looking at a similar problem as opposed to a simpler
one:
Look at past problems that are similar.
I draw, or draw from previous similar questions.
These students are using their experience in order to deal with future problems. This is
also not so straightforward as one has to recognize which problems are similar in solution
method. Often problems that require the same strategy look quite dierent in terms of the
language used. In order to truly understand if problems are similar, one must understand
the underlying structure of the problem. From the literature [18] we know that novices tend
to focus on surface features as opposed to underlying structure.
Other
Many other techniques were mentioned. For some of them algebra or equations were men-
tioned as problem solving techniques. This illustrates that these students do not realize
that algebra is a mathematical technique, but not a problem solving technique per se.
One student says that his/her problem solving strategy is mental math again illus-
trating his/her lack of understanding of the dierence between a problem solving technique,
and mathematical technique:
I have decent mental math so I use that more than anything. If Im really stuck
Ill draw a picture
Another student relates problems to real life in order to solve them:
Draw a picture, relate it to real life and put myself in the situation.
Five students claimed solving the problem step by step, systematically or using logic was
their method of solution:
I try to work in a very linear manner and go step by step. I rarely, if ever use
pictures.
Write the question down. Go step by step. Dene the variables. Write down
all relevant information.
It really depends on the question. If I can I like to solve things with an equation
or systematically. If I have no idea where to start, I draw a diagram.
Find an equation [and] solve algebraically. I dont nd picture as helpful, unless
Im double-checking my work.
Logic. Pictures.
To solve a problem systematically or logically is never mentioned by experts as these are a
given: this is an expected behaviour when dealing with a mathematical problem. We would
expect the experts to mention only the extra-logical as this would be worth mentioning to
them. Here we get an understanding that for these students, this is a signicant process
and thus worth mentioning.
Finally there are two students whose response gives no indication of any problem solving
techniques being used at all:
I dont use any techniques, If I know it then I know it.
I try to work with what I have.
7.8 Question 8: Verication
How do you check if your solution is correct?
This question parallels the corresponding question 8a) in the interviews about how to check
if you are pursuing a sensible solution. The experts mentioned several methods, including
predictions, intuition, comparing the solution with data, looking at special cases, checking
that the model makes sense, looking at the qualitative behaviour of the model, and com-
paring two methods. Intermediates also added use of convergence testing as a means of
checking results.
Here I was expecting that the FAN X99 students were more likely to nd out their work
was correct by going to their instructor. In this assumption I was incorrect as FAN X99
students mentioned substituting the answer into the original equation, checking that the
solution made sense, working backwards, re-calculating as well as checking with others.
Substitute the Solution into the Equation
This was by far the most popular method for checking whether the solution was correct,
with 25 students claiming that this was what they would do:
Sub it back into the question.
I plug numbers in x to make sure it can be solved. Look rationally at the question
and my solution.
By inputting it back into the question.
I input my answer into the original equation.
If it is a question like solve for x I will sub the number I got as x back into
the equation and see if it works.
I re-work everything into the original equation.
Based on the types of questions that these students encounter, this is one of the easiest ways
to check if they have the right answer, provided that the equation is correctly expressing
the problem. Note that this check does not guarantee that they have modelled the problem
correctly with their equation to begin with.
Check that the Solution Makes Sense
Experts and intermediates alike raised this theme. The model needs to be internally and
qualitatively consistent. Thus the fact that these complete novices also attend to whether
the solution makes sense is an improvement on simply re-substituting the solution into a
possibly erroneous equation:
Sub in numbers and check to see if the number makes sense.
I plug numbers in x to make sure it can be solved. Look rationally at the question
and my solution.
Plug answer into variable. Think logically Does this make sense.
See if it ts back into the equation or question and makes sense.
Sub my answers in to the question. Reread with my answers. Does it make
sense?
Some students combine checking if the solution makes sense with other techniques including
solving the problem more than one way and referring to the original picture:
I look at my picture or I check it algebraically. I also check to make sure my
answer makes sense to me, given the question.
See if it makes sense. Do the problem over using a dierent method.
Eleven FAN X99 students in all speak about the solution making sense; many of whom
do not clarify what they mean by the term make sense and simply state it:
Does it make sense?
I review it and see if it makes sense.
However, two students are more explanatory in their responses:
If I can plug in numbers I do so. Otherwise I check my work again to see if I
made any mistakes.
Run through the calculation again. Check positives and negatives.
One student is looking for mistakes in calculations. The other is looking at the qualitative
behaviour of the solutions: are the numbers negative or positive where they should be? This
is more sophisticated behaviour than his/her colleagues and is mentioned by expert Sam
Howison as something to look for in order to eliminate mathematical errors:
Therere the obvious ones like is the answer positive when it should be? But
thats just to eliminate mathematical errors. (Sam Howison)
Work Backwards
This technique was raised by nine of the FAN X99 students who said that this was one of
the ways they checked the accuracy of their solution. This technique is surprising in light
of the fact that the experts and intermediates tended not to use this to check their work at
all. However the literature [22] informs us that novices tend to work backward. This may
also be because this heuristic was raised in the previous question:
Depends on the question. Usually Ill work backwards, or put my information
into the original problem
Work backwards/ plug in original formula with the solution.
Work backwards. Pictures.
I work it backwards if possible.
Work backwards.
Cross reference and reverse order.
I work backwards from the answer.
Do the problem backwards.
Of note here is the fact that only three of these nine students mention working backwards in
conjunction with another checking technique. Also only two of the nine mentioned working
backwards in their discussion of problem solving techniques used, suggesting this is only
used to check answers not to solve them for the other seven.
Re-calculate
Another means of checking if the solution is correct is to re-do any calculations:
Run through the calculation again. Check positives and negatives.
Skim over each step I took. Or plug in my answer.
Plugging it back into the equation or just try it again later.
Re-do the problem, or sub in x value to check.
While this could help catch calculation errors, a misunderstood concept would simply be
repeated. This makes this method one that is not a reliable check if the solution is correct,
unless in the case of the last student quoted above, re-doing the problem involves more
than re-doing calculations. Another student acknowledges the need to go back to the original
problem to check the solution, in his response:
Going back to what the question is asking originally.
He checks initial problem, as opposed to simply looking at mathematical errors. This is
expert behaviour (see Section 4.9). Expert Thomas Witelski also suggests re-calculation
using an independent method, which was mentioned by one of the FAN X99 students:
See if it makes sense. Do the problem over using a dierent method.
This may be more helpful in identifying a fundamental mistake in one of the methods.
However, once again if the original interpretation of the question is incorrect then both
solutions may be incorrect.
Check with others
Only four of the forty-ve FAN X99 student participants suggested checking their answers
with others:
I would plug my answer in to my equation to check if my solution is correct
(and then refer to an answer key if applicable).
No real method, sometimes I see what other people answered.
I usually go back and plug my answer in to the question and when Im lazy I
wait until the prof does the answer.
Either check back (in the case of Algebra) or verify with an instructor or tutor.
Interestingly, only one student mentioned checking with others as their only means of ver-
ifying the accuracy of the solution. This is surprising as I expected this to be the most
popular means of verifying the solution at this level.
Other
Three of the FAN X99 students responses illustrate that they did not fully understand the
question:
Yes.
Depends how much the question is worth usually more than 3 marks I do.
Most of the time.
These responses to the question How do you check if your solution is correct? serve to
highlight one of the main dierences between the expert and the complete novice. The
complete novice will not always take the time to ensure that he/she fully understands the
question before answering.
Finally we note that two FAN X99 students explain that they do not check their answers:
I usually dont.
I never fully complete them... so I dont check.
While experts explain that you must check your answers, these complete novices show that
this is not something that they do.
7.9 Question 9: Dicult vs Easy
Give an example from the course of a word problem you found dicult and one
you found easy. (Can you explain why?)
When asking this question I had my preconceived ideas of what these complete novices would
nd easy and dicult about a problem. While with the experts I expected a discussion on
the complexity of the problem structure, with the complete novices I fully expected them
to complain about the length of the equations themselves and the mathematics involved.
The experts discussed more than the complexity in the structure of the problem. Themes
of mathematical diculty, familiarity with the problem and the clarity of the problem also
emerged. Intermediates also identied poor collaboration as an aspect that made it dicult
to solve a given problem.
For thirteen of the FAN X99 students there was no response to this question or they
said that they were unable to remember any examples. Another three students did give
examples but failed to indicate whether these were considered dicult or easy problems for
them. This again illustrates the likelihood of the complete novice to misunderstand or not
answer a given question in a meaningful way. However there were still several problems and
subject areas listed as being dicult or easy. We will look at each category separately.
Easy
The most popular topic considered easy by FAN X99 students is algebraic equations. This
includes systems of equations. Students also explain that skills or quiz questions (which
were also purely skills questions as opposed to word problems) were considered easy. Other
students preferred percentage problems, or fraction problems:
Simple problems for me include writing algebraic equations for word problems.
Ones with setting up an equation let x equal were generally all easy for me.
Generally, I found the problems where we needed to nd an equation easier.
Easy: system of equations.
I found problems with one or two unknowns easy.
Easy = skills testing, or plugging in equations algebraically, or fractions.
Eqns involving fractions (and percent) = Great.
Easy: Fractions [and] graphing.
Ones where pictures work are easy [includes a drawing of a picture of fractional
amounts].
Easy - percentages and money. I dont remember specically.
Almost all problems were easy.
These topics give us some insight into what topics students seem to enjoy but do not
explain why. Six students also explained why they preferred particular problems to help us
to understand some of their reasons for picking their chosen easy problems:
Quiz questions are always easy, very straightforward.
Easy would be those algebraic questions, as they are more rule/formula based,
though I understand the reasoning behind my answers.
Easy: percentages, systems of equations. Why? Personal skill.
An easy one was the Tweedle Dee and Dum weight equation because I like simple
2 question equation[s].
Easy - algebra (because I know it well).
Well there are the easier algebra problems. i.e. 56 = 8x because I know my
times tables very well.
The one about carbon emissions was easy (I like percentages :) ).
Reasons mentioned for describing these problems as easy are: personal skill, liking the
topic, the problems are straightforward, familiarity with the topic, and the problems are
formula based. These reasons mirror those of the experts. Straightforward or formula
based problems speak to problem clarity. The experts mention familiarity with the topic
explicitly. Personal skill also implies a familiarity with the problem. While having a liking
for the problem explains why it is seen as easy, we do not know why there is a liking for
these problems as opposed to others. While these students do not explicitly speak about
the structure of the problem here, we do note from the examples given that the problems
that they tend to like appear conceptually straightforward to deal with.
Dicult
The topics and problems considered dicult were more varied than those considered easy.
Students discussed topics including: functions, graphing, rates, questions without clear or
specic numbers, percentages, distance, area, and fractions. They also listed several specic
problems that they found dicult:
Dicult - A LOT!
Graphing anything is dicult.
Equations with graphs = NOT FUN.
Fraction problems were dicult.
Harder questions would be word problems with graphs and such.
One dicult problem included someone paddling upstream a certain distance at
a certain speed, and then downstream at a dierent speed. We had to nd how
far he paddled both ways to end up at the starting point in one hour. I couldnt
gure it out.
Dicult: Questions involving time/rate/pace of work with multiple people.
Hard: solving word problem, word question. Eg locker problem at the beginning
of the year.
There was this hard one to do with monkeys in a tree asking how many were
on the ground...I still dont understand it.
Hard: the amoeba problem and lily pond.
Dicult: adding consecutive numbers: 1, 2, 3, 4 (like the dart question or
shaking hands questions).
First note that in the dartboard problem there is no need to add consecutive numbers. But
also note that there is a greater number of specic problems that are mentioned here. It
was dicult to nd this many specic problems that students found easy. This implies that
the dicult problems have a greater impact and are remembered better than problems that
students considered easy. But exactly what about these problems make them dicult? Once
again these above-mentioned students have failed to explain why. However, seven students
did manage to give their reasons for why they found these particular problems dicult:
Functions are hard, hard to conceptualize.
The gold rings problem near the beginning. I didnt quite understand the process
of solving word problems.
A dicult one was nding the percentage of weight from the kilograms of a
tomato. I had trouble with conversion and it was a lot of equations.
Hard - the most recent thing were doing (because I get mixed up with domain
and range).
Dicult = Word problems (logical thinking).
Word problems in equations because I need to work on my skills.
Too many word problems. Everything had challenges from method of math or
arrangement of words in a sentence.
Reasons given for nding the problem dicult are also varied: diculty conceptualizing,
having problems with the process of problem solving, conversion and the number of equa-
tions, unfamiliarity with the terms, lacking logical thinking, lacking other skills, and the
number of words in the actual problem. Issues with the number of equations, and the num-
ber of words in the problem, speak directly to the apparent diculty of the mathematics as
opposed to the structure of the problem as expected. Diculty with concepts, logic and an
understanding of the problem-solving process are less supercial issues. These complaints
illustrate that the students are lacking basic tools to be able to solve the problems given.
What makes the problems dicult is not only the diculty of the mathematics, but the
deciency in ability of the problem solvers.
7.10 Summary
The focus of the FAN X99 class is not mathematical modelling per se and so the students
were not asked a denition of mathematical modelling. When asked about the time frame
for solving modelling or word problems, most students stated that these problems should
be solved in the order of minutes. However, two students explained that the time taken
to solve the problems are problem dependent. Students were very forthcoming about their
feelings when solving word problems, for the most part expressing fear, dread, panic and
anxiety, although several of them also expressed feelings of interest.
It should be noted that these students have little to no autonomy. For many of them the
course is compulsory for their degree, and in class they do not get to choose which problems
they prefer to work on. Unlike the experts, these complete novices tend to nd dicult
questions frustrating as opposed to interesting, with the theme of frustration being evident
throughout most of the students responses.
The complete novice participants tend to plunge rather than plan their solutions, but
the majority of them thought of themselves as organised. The suggestion of giving up
completely when stuck was only mentioned by members of the complete novice participants.
Interestingly, several complete novices discussed expert heuristics such as simplifying the
problem and drawing a picture to get access into the problem. Also of interest was the
transition for some of the students, where they noticed that they became less anxious and
more willing to try to do problems by the end of the semester.
Chapter 8
Conclusions
Throughout this work I have summarized my ndings chapter by chapter. I now present an
overall conclusion of all ndings. In this chapter I rst review the ndings in the literature,
highlighting the issues that pertain to this work. Finally, I answer the research questions
that emerged from the lack of coverage in the literature, using the data from this study to
do so.
8.1 Contribution to Mathematical Modelling
From the literature on mathematical modelling it is apparent that there are certain gaps and
a need for extension in some areas. The literature describes the cognitive and metacognitive
aspects of mathematical modelling, but there is no evidence of the extra-logical processes
(creativity, intuition or illumination) or the psychological aspects. I therefore set out to
get a more concrete understanding of exactly what mathematical modelling entails. I was
particularly interested in the process of moving from being stuck to unstuck. I investigated
the landscape from novice to expert at the tertiary level, looking not only at cognitive
processes, but also exploring the feelings experienced when modelling, the psychology of
the participants and the extra-logical processes required. What follows is a discussion and
summary of these ndings.
187
CHAPTER 8. CONCLUSIONS 188
8.2 Research Question 1: What IS Mathematical Modelling?
We have seen that the literature describes modelling in terms of the cognitive processes, using
ow diagrams to describe the process of modelling. This is not enough as the experience
of modelling is not a simple traversal along a ow diagram. It is important that we have
a more thorough understanding of what is involved in the mathematical modelling process.
To paraphrase expert David Muraki, we have to know what we are talking about to talk
about it (Chapter 4). We also note that in the literature there are two dierent approaches
to teaching modelling: case studies and modelling by design [18]. However, no justication
is made for this dichotomy in approach. Further, the Dreyfus model [46] indicates that
learning via case studies is a skill that requires almost expert competence.
In Chapter 4 the rst feature that emerged from the interviews with the experts was a
dichotomy in the denition of modelling. Some experts saw modelling as a description: a
way to describe a real world situation using a simplied mathematical construct. Others
dened modelling as a process, which started with a description of the real world using
a mathematical framework, was followed by the solution of the mathematics, and nally
returned from mathematics to the real world. This dichotomy in denition may explain the
dichotomy in approach towards teaching mathematical modelling.
The second idea that emerged was that the problems in mathematical modelling and
problem-solving diered in reasonableness in the Perkins sense. Perkins speaks of unreason-
able problems being those that cannot be approached by deliberate conscious eort. This is
not a requirement in the denition of mathematical modelling as a modelling problem can be
straightforward from the beginning to end. However, in response to question 9 (see Section
4.10), the interesting problems for the experts are those for which are not straightforward:
A [modelling] problem is interesting when the known recipes dont apply and
then you have to develop your own toolbox for it. (Reinhard Illner)
Thus while problem-solving techniques are not a requirement for all modelling problems,
they are useful for dealing with those modelling problems that are interesting. As we
will see in the next section this is evidenced in the responses to being stuck, which include
extra-logical processes.
8.3 Research Question 2: How does one move from stuck to
unstuck?
The mathematical modelling literature does not give a satisfactory description of this com-
plex step. The term simplify assumes that the modeller will know how to do so, but does
not address the fact that the modeller may be stuck because they are unsure how to simplify
their model any further. Problem-solving literature reveals when someone is stuck because
the problem cannot be solved by reason (i.e. it is unreasonable in the Perkins sense) that
deliberate eort is no longer the key. This is the stage in mathematical creativity where one
must wait for illumination or an AHA! moment [32]. These moments can be occasioned [33]
but not forced. Considering we have already established that mathematical modelling prob-
lems do not by denition have to be unreasonable, this problem-solving approach to getting
unstuck may therefore not be a viable method for the mathematical modeller.
The modelling experts of this study recommended several strategies, key among them
are communication and collaboration whether stuck at the beginning or the middle of the
problem. This theme is not recognized in the literature as a strategy for getting unstuck.
Communication with the person who brought the problem, colleagues working on the prob-
lem alongside them, or even others who were not involved at all, were all recommended by
the experts. This theme was not isolated to experts as intermediates also discussed with
their supervisors and others around them when stuck. The theme of collaboration also
highlights the fact that for many of the expert participants, mathematical modelling is a
group exercise.
The theme of simplication raised in the literature was seen in the responses as well.
This included looking for simple examples, removing modelling details, solving a small part
of the problem, or a related trivial problem. Simplication was often used in context of try-
ing to understand the problem better. Other themes were raised as well to help transition
from being stuck to becoming unstuck. Experts spoke of looking at others previous work
to deal with being stuck at the beginning of the problem (see Chapter 4), and trying some-
thing dierent (dierent thoughts, dierent model or modelling assumptions, or a dierent
approach) when stuck in the middle of the problem (Chapter 4).
Experts also described taking a break, waiting it out, and going for a walk when stuck,
mirroring the discussion by Hadamard about illumination after incubation (taking a break)
[32]. This illustrates that although mathematical modelling is not identical to problem-
solving, the extra-logical aspects of problem-solving are useful when dealing with the issue
of being stuck.
Finally one of the most important themes that emerged is that of understanding why
you are stuck. Experts spoke about trying to understand the problem by discussion and
simplication. They often claried that it depended on why you were stuck, whether you
were stuck on the math or the modelling aspects and how you dened stuck (see Chapter
4), which determined if communication, trying a dierent approach or taking a break was
the way to go.
8.4 Research Question 3: What are the dierences between
the novice and expert modeller?
In the literature there is evidence of several dierences between experts and novices; however,
these dierences refer to the cognitive and metacognitive ability of the modeller (Chapter
2). I now look at a landscape of modelling ability from novice, through intermediate, to
expert, highlighting the key dierences in psychological and extra-logical processes as well
as looking at the cognitive and metacognitive aspects of modelling. I rst revisit the theme
of being stuck, while comparing the responses of the other groups to what we have already
seen from the experts. This is followed by a discussion of the psychology of the varying
levels of modellers gleaned from their responses to questions 2 and 9. I then embark on
a journey from the beginning of the model (responses to questions 2 and 4), through the
skills deemed important (responses to questions 5 and 7) and nally look at verication of
solution (responses to question 8a)).
8.4.1 Dealing with being stuck
In the previous section we have already discussed how experts move from being stuck to
becoming unstuck, using collaboration as a primary resource. Other themes raised in re-
sponse to this question were: looking at others previous work, trying something dierent,
and taking a break. For these experts, the key aspect of transitioning from stuck to unstuck
is understanding why they became stuck.
The three main themes mentioned by intermediates when stuck were research, collabo-
ration and taking a break. In contrast to the experts, research was the number one solution
chosen when stuck, with ten of the eleven intermediates citing this, especially at the be-
ginning of a problem. Seven intermediates also mentioned collaboration with colleagues or
supervisors, but this was often after research had been done (Section 5.3). The theme of
taking a break was mentioned as well by ve of the intermediates interviewed. One interme-
diate mentioned taking time to think, but this is qualitatively dierent from taking a break,
or the incubation stage, where conscious eort ceases. Trying to understand the problem
and being willing to change the model completely were mentioned by intermediates as well,
but unlike the experts these were outliers as opposed to major themes.
When novices were asked about the barriers they encountered in the MCM they men-
tioned several dierent issues, including working with others in a group (Section 6.6). Sev-
eral methods were attempted to move from being stuck including the use of numerics and
switching strategies. One novice mentioned changing modelling assumptions and and an-
other mentioned collaboration as a means of getting unstuck. What is noticeable is that
none of them spoke of overcoming the barriers successfully, nor did they discuss under-
standing why they were stuck. Students responses about whether they found working in
a group useful were split. Students often cited the group dynamic as being a part of the
problem although they acknowledged that it can be useful as well. This is in contrast to
the experts but not unexpected. These novice modellers have limited experience with work-
ing in groups, and therefore have not developed their mathematical or even professional
collaborative skills, something deemed important by the experts (Section 4.5).
For the complete novices, as one might expect, there is a marked dierence in how they
deal with becoming unstuck. The rst thing of note is that this group of participants bring
forth the new theme of giving up completely. This is not mentioned by any of the experts,
intermediates or MCM novices. Those complete novices who do not give up spoke of re-
reading and re-doing the question, often without a discourse on trying to understand exactly
why they were stuck. Another major theme raised by this group was waiting for the answer,
as opposed to collaborating with their colleagues. This theme of learned helplessness [55]
is unique in this study to the group of complete novices. Although other strategies were
mentioned including breaking down the problem, drawing a picture, trying something and
exploiting a similar problem, these were in the minority being mentioned by six or less
students each out of a total of 45 students.
8.4.2 The Psychology of Modelling
We have seen above that the experts collaborate when stuck, while the intermediates turned
to research. The novice modellers attempted several methods of getting unstuck that were
for the most part unsuccessful, and the complete novices tended to give up, wait for help or
re-do the problem, with no mention of changed assumptions. We know from the literature
that experts have a broader knowledge and better cognitive and metacognitive skills, but
what psychological factors aect expert and student success?
The rst psychological factor explained by Andrea Bertozzi is that experts make the
decision about what problems they work on, while students do not posses similar autonomy
of choice. As David Muraki explains, there are expectations (by both the instructor and the
student) that when given a problem, the student should be able to do it, whether or not the
student feels capable enough or interested enough to do so (Sections 4.2, 4.3). Autonomy is
one of the three main intrinsic motivators quoted by Pink [41] when discussing the work of
Deci and Ryan: Autonomous motivation involves behaving with a full sense of volition and
choice [43]. This lack of autonomy for the novices and to some extent for intermediates,
implies a lack of intrinsic motivation as well, which is one of the main elements Type I
behaviour and thus expertise (see Section 2.3.2).
The second psychological factor of successful modelling is condence and risk-taking.
Throughout the interview experts speak about being fearless, being willing to try something,
and not worrying about whether the initial guess is wrong as they can learn from it. Mike
explicitly states that one must be willing to go out on a limb and not worry about getting
it wrong (Section 4.5). Sam Howison also speaks about not being afraid to ask questions
(Section 4.7). This explains why collaboration when stuck is a big favourite among these
experts. Fear of asking questions or fear of failing makes it dicult to collaborate and indeed
to make any progress at all.
Tobias mentions fear of making mistakes as one of the reasons that students would expe-
rience anxiety and refuse to voice mathematical thoughts: One thing that may contribute
to a students passivity is the fear of making mistakes in mathematics [...] Successful math
students know better. They do not despise their errors.( [13], p.52). This attitude, preva-
lent among expert modellers, becomes less evident as we move down the spectrum. Thus
we see experts willing to discuss their problems with anyone, intermediates discussing with
colleagues or supervisors, novices having a hard time collaborating with each other, and
acute novices simply giving up or waiting passively.
The experts in this study described interest and curiosity for the most part when faced
with modelling problems, particularly describing dicult problems as being more interesting
(Section 4.10). In cases where worry or anxiety was present, the experts were able to push
past those feelings and make progress. When stuck, some experts spoke of waiting for
insight, implying that they expected the ideas would be subconsciously solved. Experts
described these dicult problems as the ones where they learned a lot. These qualities hint
at creativity in the Hadamard sense as well as from Csikszentmihalyis perspective. The
experts are accustomed to illumination as being part of their process, which Hadamards
research also supports. The experts also do not crave the easy problem, but prefer the
problem to be a bit out of their reach. This implies a search for ow where the problems
are not too easy, which in turn lead to satisfaction and mastery [41].
The complete novice experiences frustration as opposed to motivation when stuck or if
the problem is dicult (Chapter 7). There is evidence that the MCM novices feel some
interest in the face of mathematics as they volunteered to take part in the competition.
However their discussion about the barriers they encountered also suggests a great deal
of frustration (Chapter 6). Intermediates on the other hand experience a mix of interest
and worry when rst faced with a modelling problem, often explaining that these feelings
are problem dependent. Some intermediates described waiting for insight or illumination,
pointing to creativity in the Hadamard sense, though when describing dicult problems
there is no evidence that these are the problems they nd more interesting (Section 5.10).
The fact that intermediates speak to feeling overwhelmed (Section 5.2), suggests that they
are not experiencing the enjoyment of ow, but rather suer some anxiety.
These beliefs, attitudes and feelings towards mathematical modelling have a signicant
impact on motivation and condence, which in turn impact mastery when modelling. This
is particularly evident when stuck.
8.4.3 Dening Mathematical Modelling
The experts of this study dene mathematical modelling in two ways (Section 4.1). The
rst is considering modelling as a simplied description of a real world situation understood
using mathematical formulation. In the second denition modelling is seen as a process
starting with the creation of a mathematical framework to describe a real world problem,
followed by the solution and renement of the mathematical problem and a return to the
real world problem in order to explain or make predictions.
The intermediates did not have this dichotomy in their denition, describing mathemat-
ical modelling as breaking down a real-world problem and using mathematical language to
explore it (Section 5.1). In this group the solution of the modelling problem is incorporated
into the denition. None of these intermediates spoke of rening the model or using the
model to make predictions. This may be indicative of the fact that they are more likely
modelling toy problems than problems directly from industry.
For the novice MCM participants what makes a problem a modelling problem is even
more simplistic. This group of modellers describe modelling as a mathematical represen-
tation of a real life problem. Again, no description of rening the model or predictions is
evident. The dichotomy noticed here was that some students depicted a modelling prob-
lem as one that had more than one solution method. This is not a dening condition of
modelling for the experts or in the literature, although it tends to be the case in reality.
The group of complete novices have not been introduced to the concept of modelling in
their FAN X99 class, or possibly elsewhere. With this in mind this group was not asked to
discuss their denition of a modelling problem.
We have examined the psychology of modelling for the dierent groups, particularly
how each group deals with being stuck and denes modelling. We will now traverse the
modelling process, taking a look at how each group starts a modelling problem, what skills
they deem important, and nally how they verify their solution.
8.4.4 Starting a Modelling Problem
Many of the experts interviewed begin problems with excitement and curiosity, although
they needed additional prompting to discuss their feelings. Some of them described feelings
of fear or worry, but were able to push past those feelings of self-doubt (see Section 4.2).
These experts begin by ensuring they understand the problem, which corroborates ndings
in the literature [18, 34]. This study reveals that the experts do this through research,
exploration and most importantly collaboration. Their rst plan is generally to begin with
the simplest possible problem. Key to this approach is that it is acceptable if this simplest
problem is not accurate, as it is more important that it captures some quality of the problem
and gives some understanding of the overarching processes (Section 4.3). Experts focus on
the big picture to determine relevance: this again is veried in the literature, which attributes
this characteristic to a superior expertise [18]. What the literature does not tell us is that
experts focus on the areas of the problem that can be best captured by the mathematical
tools they are most familiar with (Section 4.4).
In comparison, the intermediates begin modelling with curiosity or persistent self-doubt,
and were more willing to discuss their feelings than the experts (Section 5.2). Their initial
strategy tends towards undirected independent research via the internet, textbooks or pa-
pers. Unlike the experts, no mention of collaboration with others was noted. There is a
distinction here between what the experts and the intermediates view as research. For the
intermediates, research is via literature and the written word. Alternatively for the experts,
research implies seeing what others have done. This may require a look at the literature
but can also be achieved by communicating with others.
Simplication was a heuristic mentioned, particularly by those intermediates who felt
overwhelmed with the initial complexity of the problem. As with the experts the theme of
trying to identify the dominant process of the problem was raised. Intermediates indicated
that determining what was relevant to the model was dicult for them, either in the length
of time it took them to respond, or by explicitly stating that this step was tricky (Section
5.4). Some of them attributed this skill to experience or intuition, without explaining further
what this entails.
The novice modellers of the MCM exhibited a range of condence with their modelling
problem, from viewing it as a piece of cake to feeling completely clueless about how
to go about solving it. The literature tells us that novices tend to be quick to start on
a problem whereas experts take the time to understand and analyse the problem [35].
Strangely, starting immediately was only reported by three of the eight novice modellers.
However, of the ve that did not start immediately, three of them did not because the team
was not assembled immediately, and one of them did not because he thought the problem
was easy. Only one student mentioned trying to understand the problem rst, and this was
done via blind research (i.e. Google search). In determining what terms were important to
the model, a variety of responses were noted, from everything was important to too many
things were unimportant. For the most part these students did not give justication for
why they thought a particular thing was unimportant to the model.
Approximately three quarters of the acute novices or FAN X99 students described neg-
ative emotions when rst faced with a word problem (Chapter 7). These students tended
to plunge into a solution as opposed to planning, with 28 plungers and 10 planners (Section
7.4). In direct contrast a majority described themselves as organised, with 28 claiming to be
organised and 11 claiming to be disorganised (Section 7.5). This does not correlate with the
literature [25] and may indicate that students are unaware of their own lack of organisation
(i.e. the Dunning-Kruger eect [51]), or that students consider their work organised due
to neatness and handwriting skills, as opposed to organisation of thought. These students
generally had a preconception that word problems should be solved in the order of minutes
(Section 7.1). There was no discussion of determining what aspects were relevant to the
model as the students had no experience with such problems.
8.4.5 Knowledge and Skills Valued
When I interviewed the experts and novices with respect to the skills they deemed important
for mathematical modelling (question 5), I split the question into two parts. I identied
that there were mathematics skills that were important, but I was also interested in non-
mathematical skills deemed useful for modelling. This yielded a range of rich responses
from experts and intermediates alike. Question 7 asked specically for heuristics in the
interview. This question was mirrored in the questionnaires given to both groups of novices.
I will therefore address all three sets of skills in this discussion: mathematical skills, non-
mathematical skills, and heuristics. Note that the complete novices were not asked what
mathematics they used.
Mathematics
A look at the mathematical skills deemed important by the experts shows a wide variety of
topics. As the group of modellers interviewed worked primarily with continuum modelling,
it is not surprising that Calculus, ODEs and PDEs come out on top. Several experts also
mentioned Statistics and Probability and Numerical Analysis (Section 4.5). Of course, an-
swers were not limited to these 5 subjects and went on to include nine other topics: Linear
Algebra, Abstract Algebra, Analysis, Data Analysis, Queueing Theory, Graph Theory, Dis-
crete mathematics, Calculus of Variations and Optimisation. Many experts mentioned that
the topics they listed were by no means exhaustive.
The majority of intermediates in this study also focused on continuum mechanics in
their studies. This yielded Calculus and DEs as the more popular subjects mentioned
(Section 5.5). This was followed by Discrete mathematics and Numerical Analysis. Other
topics mentioned were not as varied as with the experts: Probability and Statistics, Linear
Algebra, Lagrangian Mechanics, Fourier Analysis and Analysis.
When asked what mathematics was used in the MCM, students mentioned several dier-
ent areas, with a slight focus on Calculus and Physics. Other topics mentioned were varied:
Euler-Lagrange Equations, Mechanics, Geometry, ODEs, Curve tting, Discrete mathemat-
ics and software. This is particularly interesting because all of these MCM participants
worked on the same problem. This illustrates that one problem can appear to require many
dierent areas of mathematics. I say appear because none of the students spoke of a par-
ticular area being more successful than another, and so we cannot conclude that any or all
of the mathematics they used was indeed useful.
Experts report a wide range of mathematical topics as useful for a wide range of math-
ematical problems. Intermediates list a more restricted range of topics, which we expect
from the literature [18]. Intermediates in the study tended to mention topics they were
more familiar with (Section 5.5). When dealing with a single problem, the novices supply
an equally varied list as the intermediates, although they make no claims that all of these
methods were successful in nding a solution to the MCM problem. We also saw that the
novices often were not familiar with topics suggested by their colleagues (Chapter 6). This
suggests that the novices were not sure how to go about solving the problem and simply
used all the knowledge available that seemed useful.
Non-Mathematical skills
This ability to determine what mathematics to use is a metacognitive skill, as opposed to
a purely cognitive one. This brings us to an examination of the dierences from novice to
expert in non-mathematical skills. Experts rst discuss that breadth as opposed to specic
mathematical skills are essential. This is evident in the breadth of mathematics topics
that they were able to supply, but also in the knowledge that they brought from outside
of mathematics. Not surprisingly we note that half of the experts actually speak about
needing scientic knowledge related to the area that the problem is coming from.
Another very important skill is collaboration, a theme that has been raised throughout
this thesis. Collaborative skills including communication, diplomacy, listening and patience
with others were mentioned specically by eight of the experts. Since this theme recurs
throughout this study, it may be worthwhile to note here that there are several possible
reasons why non-experts do not collaborate. In particular for the intermediates, as graduate
students, needing to demonstrate independent thinking and research may cause them to shy
away from collaborating too extensively (they may even be told explicitly not to collaborate
by their advisors). We therefore distinguish between active collaboration (as seen by the
experts) and passively asking for help (as exhibited by some complete novices). There is
also a distinction within the dierent ways experts use collaboration. Collaboration is used
to help clarify understanding to allow, for example, the formulation phase of modelling.
Collaboration is also used as a means to achieve metacognitive goals i.e. to articulate ones
thinking allowing one to pinpoint exactly where the diculty is and thus move from a state
of being stuck to becoming unstuck.
Apart from these three major themes, one expert also mentioned being able to organise
ones thoughts, which ties in with the ability to determine what mathematics to use. This
metacognitive skill is one which the novices lacked and is also supported by the literature [18]
on the dierence between novices and experts. Two experts highlight the need to understand
what the words mean, as people of dierent disciplines use dierent terminology to describe
similar topics. The experts mentioned several other personality traits as being invaluable:
maturity, condence, passion, curiosity, exibility, stamina, persistence, hard work and
patience. These characteristics are all essential aspects for motivation and creativity [41,45].
The intermediates interviewed mentioned several of the themes raised by the experts.
Intermediates discussed themes of breadth and relevant skills, as well as scientic or back-
ground knowledge. Four intermediates raised the theme of collaboration, which was seen as
a major skill-set by the experts, however this was approached in a slightly dierent manner.
The experts here listed dierent aspects of collaboration that were useful, while interme-
diates spoke of a willingness to collaborate and not being afraid to ask questions (Section
5.5). Intermediates also tended to use collaboration in a metacognitive way (to pinpoint
the diculty when stuck) as opposed to using it as a means of clarifying the modelling
problem in the initial stages (opting to turn to the literature instead). Intermediates are
also inclined to talk to mentors, which in many cases is not an available option for experts.
One intermediate (who was also a PDF) raised the theme of knowing what the words mean.
This comes from work in the eld with people of dierent disciplines who describe the same
concept dierently. This must be highlighted in the classroom, so that students know to
rst clarify the meanings of terms being used.
One other theme evidenced by the intermediates was the metacognitive skill mentioned
earlier of knowing how to approach problems. Research was the number one tool for getting
unstuck among intermediates so there is no surprise that being able to search for what you
dont know was mentioned as a useful skill. Intermediates mentioned several personality
traits raised by the experts, but in each case the context or wording is slightly dierent.
Intermediates spoke of experience as opposed to maturity, and while passion, interest, cu-
riosity, perseverance and hard work were mentioned they were not major themes as with
the experts. Condence was mentioned, but half of the intermediates that spoke of it stated
it from a negative perspective, such as not being afraid of failure, as opposed to you need
condence.
There were no questions on the questionnaire for the group of novice modellers that
specically addressed the non-mathematical aspects of modelling. However, a look at their
answers to several dierent questions gave evidence to their deciency in the skills valued
by experts and intermediates alike.
These novice students valued collaboration, but often described it as not being helpful
(Section 6.6), and instead listed it as one of the barriers to nding a solution to the problem.
Similarly, they spoke of their lack of breadth of knowledge in response to barriers that were
encountered by the groups. The metacognitive skill of knowing what mathematics to use
was lacking, evidenced by the range of mathematics employed to solve the problem. One
novice explained, it was hard to determine what the question was asking (Section 6.7),
which also shows that understanding the problem was an issue. On the other hand novices
exhibited interest by volunteering for the modelling competition. There was also evidence
of condence as they were willing to try some strategies despite not knowing if they would
work. In some cases this condence was misplaced (Section 6.7).
The complete novices were also not asked specic questions about what skills they
deemed important for modelling, particularly because they have no real concept of mod-
elling. Still, a look at their responses indicates several non-mathematical issues that they
have that would hinder their modelling ability. Although some students spoke of a transition
to more positive feelings as the semester progressed, for the most part feelings described
were negative. Also noted is a tendency to give up or wait for help instead of actively partic-
ipating, in contrast to the perseverance and hard work mentioned by the experts. Finally,
these students tend to plunge rather than plan their solution, while experts recommend
patience and ensuring you understand the problem rst.
Heuristics and Problem Solving Strategies
The experts interviewed in this study cited ve main heuristics (Section 4.7). The rst of
these is drawing a picture. This heuristic was mentioned as a means to clarify understanding,
in order to help those working on the problem or those bringing the problem to see what is
happening.
This ties in with the second heuristic: understanding the problem. This supports
Polyas framework, where understanding the problem is the rst step. This includes an
understanding of what the words being used mean, which was mentioned as an important
non-mathematical skill. A look at non-mathematical skills of novice modellers show that
understanding the problem is a heuristic that is lacking.
The third heuristic of the experts is simplifying the problem, which is also connected
to understanding the problem. This heuristic is mentioned in the literature [18], but two
experts clarify further that this entails looking at limiting cases in particular as a means
of simplication in applied mathematics. For the experts looking at limiting cases gives
an insight into the problem and yields a better understanding of the more complex cases.
Limiting cases would fall under the category of special cases in Briggs work [16].
The fourth heuristic the experts mention is exploiting a related problem, which is also
mentioned by Briggs. This requires experience, a skill broached by the intermediates; and
the ability to recognise which problems are related, which is a metacognitive skill. Experts
describe an easy problem as one that is familiar, which encompasses this heuristic of ex-
ploiting a related problem, as well as the ability to recognise which problems are related.
Being able to reformulate a problem into one you know so that you can exploit the related
problem is a high level skill.
The nal heuristic mentioned by the experts is talking to others. This is no surprise as
experts throughout this work have mentioned the themes of collaboration and communica-
tion. This is not seen in any of the literature on modelling; however, this is observed in the
mathematical modelling camps and workshops. The breadth of knowledge required to solve
an industrial modelling problem makes it necessary for a collaborative eort. As one expert
succinctly put it, mathematical modelling is not a solitary activity (Bob).
Having looked at the heuristics of the experts we now turn to those employed by the
intermediate modellers. Recall that for six of the eleven intermediates, a full list of possible
heuristics were provided, making the responses somewhat skewed. I will therefore only
discuss the main heuristics noted.
The rst two main heuristics mentioned by the intermediates are drawing a picture,
and solving a simpler problem, with nine intermediates each mentioning these. These two
heuristics were similarly discussed by the experts. Three of the eleven intermediates also
mentioned the heuristic of collaboration. While not nearly as prominent a heuristic as
described by the experts, it is evident nonetheless.
Intermediates also mentioned a fourth heuristic of taking a break, which was not sug-
gested by experts in response to this question (Sections 4.7, 5.7). This heuristic is of note as
not only did all intermediates prompted agree, but three others not prompted also suggested
that taking a break was a useful heuristic.
The novice participants of the MCM, mentioned only two heuristics, when asked what
strategies they tried and why. The rst heuristic mentioned was research. This was evi-
denced in responses by the intermediates but was not a heuristic included by the expert. The
second heuristic was simplication. This was stated specically by one MCM participant,
and was also implied by another participant who started with the actual shape of a snow-
board before trying a more standard shape. Novices also mentioned in their problem-solving
strategies: use of 2D/3D surfaces, complex physics and complex numerics, continuous and
discrete mathematics, Newtonian mechanics and Lagrangian mechanics. These responses
suggest that the mathematics used and the problem-solving strategies employed were not
distinct for the novices.
A look at the problem-solving strategies of the complete novices reveals a similar nding.
Students mention mathematics skills such as algebra and mental math as their problem-
solving strategy. The main heuristic seen here was drawing a picture, with an overwhelming
number of 36 out of 45 students claiming to use this strategy. Students also mentioned
working backwards, exploiting a similar or simpler problem, relating the problem to real life
and using logic. Working backwards is not noted as a useful skill in modelling by either the
intermediates or the experts, but these students experienced some word problems for which
this made sense. Relating the problem to real life is a very useful heuristic to transfer to
modelling, as this is the very essence of modelling. While thinking logically is used by all
groups, it is mentioned by none of the others because for every group except the complete
novices, it goes without saying.
8.4.6 Verifying the Solution
We now look at the nal stage of the modelling process verifying the solution. The experts
immediately claried that in modelling there is no correct solution: the end of the modelling
process is not checking that you are right, but that your solution is sensible. Experts also
comment that verication of the solution is necessary. To do this experts suggest rst and
foremost a comparison with the data or experimental results. Arising from this theme is the
fact that one has to be careful that the data used to create the model is not the same as the
data used to test it. Other experts speak of making predictions, which would also require a
comparison with data to test if the predictions are meaningful. Looking at limiting cases is
mentioned again, here as a means of verifying the viability of the model or solution. Experts
suggest that you will know your solution is sensible by looking at the qualitative behaviour
of the model. Two experts also suggest comparing two dierent solution methods, especially
if no data was available for testing.
Intermediates were not given an opportunity to explain that there is no correct solution to
a modelling problem as I had already changed this question in the interview. Intermediates
mentioned themes raised by the experts including: comparing with the data, that the model
makes sense, looking at simple test cases, and comparing two solutions.
The intermediate participants also raised themes not mentioned by the experts. The
rst of these is that intuition will let you know if the solution is sensible or not. Only one
expert speaks to intuition, however the intermediates raised this theme on more than one
occasion. Intuition is dened as the ability to understand something immediately without
need for conscious thought. This ties in with the Dreyfus model of expertise when pertaining
to experts, who nd it hard to explain their processes as they have become automatic. This
suggests that the experts do not mention intuition because the process is too embedded
in their subconscious to make it obvious for discussion. An intermediate looking on will
observe this as an expert seemingly arriving at the solution without thinking.
The second theme raised by the intermediates not seen in the responses from the ex-
perts is convergence testing. Two intermediates describe convergence testing as a means of
checking if their solution is sensible. Convergence testing will verify if the numerical scheme
is working to the order it should and if the results are converging as expected, but does not
show if the solution is applicable to the real world problem itself. However, this is a viable
way to test the numerics of the problem. This concentration on checking the numerics is not
an expert behaviour that is noted in the literature or from this study. It is perhaps evidence
of the intermediates losing sight of the important goals and focusing on what they know
and can do best. Testing the numerical models is an important part of scientic computing,
but it neglects the fact that perfect algorithms give no insight when applied to incorrect
models.
The complete novices oered several ways of checking if their solution was correct. Notice
that there is no dispute about the correct solution as these students are used to one closed-
form solution being the answer. Strategies for verifying the solution include checking that
the answer makes sense by re-substituting, working backwards, and re-calculating. These
methods have the potential to verify arithmetic errors, but do not give feedback about
whether the initial equations (or model) are correct, or if those solutions answer the original
word problem. Another method mentioned is checking with the instructor. This ties in
with the novices tendency to be passive and wait for help. This also speaks of a belief
that others hold the keys to answers that students are not privy to. Finally, some complete
novices admit that they do not check their solution at all, in contrast to experts who deem
this to be a necessary step.
The novices that competed in the MCM were not asked about verifying their results.
However, we can infer from the numerous references to time constraints that verication
may not have taken place.
8.5 Summary
These conclusions have implications for the teaching of mathematical modelling, if the aim
is to help develop the novice modeller become an expert. While the established literature
addresses the cognitive and metacognitive dierences between the novice and the expert,
there are still several things lacking. These include a denition of modelling, an investigation
of how one moves from being stuck to becoming unstuck, and a discussion of the non-
cognitive dierences between the expert and the non-expert. This study has addressed
these three issues in particular.
We have seen in the modelling literature that there is no agreed upon denition of
modelling. Among the participants of this study there are also dierences. The experts
exhibit a dichotomy in their responses, with some viewing modelling as the formulation
of the model and others viewing modelling as the entire process including verifying and
rening the model. The intermediates focus on the solution step of the modelling process
but expressed a similar denition to that of the experts. On the other hand, some novices
misunderstood what modelling is, assuming that modelling problems are ambiguous by
denition, as opposed to being ambiguous as a consequence of coming from real-world
problems.
When dealing with being stuck, the experts tend to collaborate with others around them,
those who have brought the problem, colleagues, and even those who have not worked on
the problem at all. This is not seen in the modelling literature but was raised by almost
every expert interviewed. Most important for the experts is understanding the problem in
order to become unstuck. The intermediates turned primarily to the literature to increase
understanding as opposed to collaboration. The novices spoke of switching strategies when
stuck without discussing trying to understand why they are stuck. Complete novices were
the only group to mention giving up completely when stuck. They also tended towards more
passively asking for help or waiting as opposed to active collaboration.
There are several other dierences as we traverse the landscape from novice to expert.
There is an increase in autonomy as we move along the spectrum, with the complete novices
having little or no autonomy and the experts having almost complete autonomy. There is
also a decrease in persistent self-doubt or anxiety as we travel along the spectrum from novice
to expert. Experts do speak of feeling some anxiety, although they are able to distance
themselves from these emotions in order to address the modelling problem. Experts also
described dicult problems as interesting, while complete novices saw them as frustrating.
These results clearly indicate that simply teaching more mathematics is not enough,
especially as the literature and the ndings of this study suggest that psychological, as well
as metacognitive characteristics need to be developed alongside the cognitive in order to aid
with success in modelling. We will discuss these implications for teaching as well as future
work in the nal chapter.
Chapter 9
A Look to the Future
There is still much work to be done to fully understand all of the aspects of the modelling
process. This qualitative study has hinted at several of the dierent issues that need to be
dealt with when modelling and more work can be done to tease these out further. Having
started out with the aim of understanding how to learn modelling, and what the learning
outcomes of a modelling class would be, it is only logical that I look at the implications this
work has for the teaching of mathematical modelling. Therefore in this chapter I will look
at the implications for the teaching of modelling, followed by suggestions of what future
work must still be done.
9.1 Implications for Teaching
From the literature on cognitive and meta-cognitive processes [12, 18, 19], we know that
these as well as maturity and experience take time to develop. These are not skills that can
be taught, but are a result of motivated work in a particular area. However, from this work
there are denite skills that can be taught to help modellers on their path to mastery.
The rst implication of this work is that a working denition should be given to students.
While it is acknowledged that modelling is a vast topic, there are still some key features
of modelling that experts and literature agree upon. A discussion in the classroom of the
dichotomy between modelling as creating the model, and modelling as creating and solving
the model, should give students a clear idea of what the process entails. This would help
them to identify when they are faced with a modelling problem. It also aids in developing
learning outcomes: which aspect of the modelling process is the class focusing on, the
205
CHAPTER 9. A LOOK TO THE FUTURE 206
creation of models, the solution of models or both?
From the literature on problem solving and the responses of the experts, a focus on trying
to rst understand the problem should be a major part of the modelling classroom. The
literature on novices as well as evidence of this work indicates a tendency of novices to plunge
in without rst understanding what is happening. Experts on the other hand spend a large
proportion of the time trying to fully understand the problem including: understanding what
the words mean, understanding exactly what is being asked, understanding what aspects
are relevant by looking for dominant processes, and obtaining background knowledge in the
eld the problem originates from. Classroom practices can mirror this, taking the focus o
trying to solve the problem initially, and just trying to understand the problem. Explicitly
stating that understanding the problem is the objective, helps to re-direct many students
unspoken belief that the point of mathematics is getting a quick solution as opposed to a
veriable one.
A third implication of the study is that students in the study were unable to separate
the mathematics from the heuristic. Why is this important? If the point of using a certain
area in mathematics is to help simplify a particular problem, then students may tend to use
that area in mathematics at all times, whether it causes simplication or not. Separating
the mathematics from the underlying principle allows them to access a wider range of
mathematics in theory.
Another decit identied from the study is the non-experts inability to verify solutions,
focusing instead on checking numerics, checking arithmetic, or simply skipping this step.
This step should not be overlooked or glossed over in the classroom, as it is a key step in the
modelling process and deemed essential by the experts. A focus on verifying the solution
needs to be evident in the classroom if we want students in turn to focus on this aspect when
completing the modelling process. This means including the renement process as well as
comparison to data, and a discussion of how to verify a solution by looking at qualitative
behaviour when no data is available.
A major focus of this work was helping students move from a state of being stuck to
becoming unstuck. So far all of the areas discussed are directly derived from the modelling
process and may be more obvious to a teacher of modelling. However helping students move
from stuck to unstuck is more subtle. Since collaboration and communication are the main
means of doing this by the experts then this must be encouraged. Students must also be
encouraged to focus on why they are stuck. Intermediate responses are helpful here, as they
suggest that trying to articulate why you are stuck often helps the transition to becoming
unstuck. This is also seen in the problem-solving literature. This again leads us back to
collaboration and communication. In applied mathematics modelling classes a focus on the
use of limiting cases in an attempt to simplify is also useful. Students would also benet from
seeing real modelling done by experts. Seeing experts deal with being stuck is informative,
and helps change the belief that experts simply rely on intuition. Finally, participating in
modelling camps, competitions and workshops should be encouraged, as much of modelling
is learnt from actually experiencing modelling in this collaborative environment, where
students can see rst-hand how to use collaboration as a means of getting unstuck.
In the previous paragraph we saw the theme of collaboration emerge as a means of helping
students and experts alike move from stuck to unstuck. However, collaborative skills are
not automatic. This is especially true in mathematics classrooms where focus on individual
skills is usually the norm. Thus, collaborative skills must be developed. This means that
modelling classes cannot be solely lecture based to be ecient. Students need to work
on the ability to articulate their thoughts, the ability to be diplomatic, the ability to listen
respectfully and to have patience with each others ideas. The environment must also be one
in which students are willing to collaborate and not afraid to ask questions. Discussion with
the mentor or teacher should be encouraged not only as a means of clarifying understanding,
but also to help students break away from passivity.
The nal implication for the teaching of modelling that we shall discuss here is that of
autonomy. Experts choose the general area they work in, as well as the specic problems
that they want to work on. This autonomy leads to intrinsic motivation as they are invested
in understanding the problem and coming to a viable solution. Similarly, in modelling camps
student have some choice of whether to attend, and which problem to work on. In order to
help facilitate this intrinsic motivation, teachers must provide problems that mathematics
students are familiar and condent with, and give the option to work on dierent problems
if one is not of interest. Pink [41] explains that for non-routine problems intrinsic motivation
is vital. Mathematical modelling is not routine due to the breadth and variety of problems
it covers and the fact that those problems come from the real world and are messy. This
therefore implies that mathematical modelling cannot be mastered without self-motivation
and interest.
9.2 Future Work
The work that is discussed in this thesis is only a small portion of what needs to be done
to fully understand the modelling process (if it can ever be fully understood). In an eort
to get a clearer view of the issues several extensions of this study can be explored. The
rst of these is a more varied participant group. While the experts and intermediates came
from dierent backgrounds, they were for the most part from North America and Western
Europe. A look at how modelling is viewed across the world may give rise to an even wider
variation in denitions, skills, and beliefs of modelling. The novices in the study all came
from SFU and it would be interesting to see if these traits hold true across universities.
I would also be interested in noting the dierences between novice modellers in dierent
subject areas.
I chose 4 broad categories for the purpose of this study, due to the ability to easily identify
the two extremes. The intermediate group represented all persons that could not fall under
the expert or novice categories. However, the Dreyfus model describes ve dierent stages
from novice to expert. It would be interesting to try to classify the intermediate participants
based on the skills that they have for a more specic look at a range of modelling ability.
It would be especially interesting to see if the level of expertise automatically increases
based on the number of years spent modelling, and if not, why some intermediates are
more advanced than others with fewer years of experience. The goal orientation of the
modeller also appears to change with expertise (experts tend to be motivated to learn while
novices tend towards avoidance). This suggests the development of tools to assess the goal
orientation of modellers from particular groups.
In this study I made use of observations to provide a context for understanding the
modelling process, and being able to interpret the responses to interviews and questionnaires.
However using those observations to verify or contradict responses would reveal an even
deeper understanding of the modelling process for all three groups. In particular it would
be interesting to see how the novices responses match to their actual process, if it were
captured on video. I suspect that for the intermediate group this would be a challenge, as
at the intermediate stage there is more awareness of what is unknown than at the novice
stage, but not as much condence as at the expert stage. Thus I believe that intermediates
would have the least desire to have their actual process recorded.
Of course, one of the obvious directions for future work is the implementation of the sug-
gestions in relation to teaching, to see if they lead to a more successful modelling experience
for students. In order to keep track of the aective and extra-logical aspects of modelling
we could have students keep a modelling journal, describing their understanding of the pro-
cess, their struggles, and their triumphs. Journaling is not straightforward for mathematics
students and may require some illustration [33] but the combination of the journals with
classwork would give a clear picture of the logical, extra-logical, and psychological advances
students are making as they travel along the spectrum towards modelling expertise.
Bibliography
[1] R.R. McLone. The training of mathematicians: a research report. Social Science Re-
search Council, 1973.
[2] N. Challis, H. Gretton, K. Houston, and N. Neill. Developing transferable skills: prepa-
ration for employment. Eective learning and teaching in mathematics and its appli-
cations, Kogan Page, pages 7991, 2002.
[3] Andrew Petter. Sfu: The engaged university, 2012.
[4] R. Illner. Mathematical modelling: a case studies approach, volume 27. Amer Mathe-
matical Society, 2005.
[5] S.P. Otto and T. Day. A biologists guide to mathematical modeling in ecology and
evolution, volume 13. Princeton Univ Pr, 2007.
[6] Simon Fraser University. Complex Systems Modelling Group. Modelling in Healthcare.
Amer Mathematical Society, 2010.
[7] S. Howison. Practical applied mathematics: modelling, analysis, approximation, vol-
ume 38. Cambridge Univ Pr, 2005.
[8] W.E. Boyce, R.C. DiPrima, and D. Mitrea. Elementary dierential equations and
boundary value problems, volume 9. Wiley New York, 1992.
[9] K.K. Tung and E. Cumberbatch. Topics in mathematical modeling. SIAM review,
50(3):603, 2008.
[10] K. Bain. What the best college teachers do. Harvard Univ Pr, 2004.
[11] W.G. Perry. Forms of intellectual and ethical development. New York: Holt, Rinehart
et Winston, 1970.
[12] C. Haines and R. Crouch. Mathematical modelling and applications: Ability and
competence frameworks. Modelling and applications in mathematics education, pages
417424, 2007.
[13] S. Tobias. Overcoming math anxiety. WW Norton & Company, 1993.
210
BIBLIOGRAPHY 211
[14] R. Crouch and C. Haines. Exemplar models: expert-novice student behaviors. Math-
ematical Modelling: education, engineering and economics. Chichester: Horwood Pub-
lishing, pages 90100, 2007.
[15] A.C. Fowler. Mathematical models in the applied sciences, volume 17. Cambridge Univ
Pr, 1997.
[16] W.L. Briggs. Ants, bikes, & clocks: problem solving for undergraduates. Society for
Industrial Mathematics, 2005.
[17] N.A. Gershenfeld. The nature of mathematical modeling. Cambridge Univ Pr, 1999.
[18] C. Haines, R. Crouch, and A. Fitzharris. Deconstructing mathematical modelling:
Approaches to problem solving. Mathematical modelling in education and culture:
ICTMA10, pages 4153, 2003.
[19] C.R. Haines and R. Crouch. Remarks on a modeling cycle and interpreting behaviours.
Modeling Students Mathematical Modeling Competencies, pages 145154, 2010.
[20] A.H. Schoenfeld. Whats all the fuss about metacognition? Cognitive science and
mathematics education, page 189, 1987.
[21] PL Galbraith and G. Stillman. Assumptions and context: Pursuing their role in mod-
elling activity. Modelling and Mathematics Education: Applications in Science and
Technology, pages 300310, 2001.
[22] R.M. Heyworth. Procedural and conceptual knowledge of expert and novice students for
the solving of a basic problem in chemistry. International Journal of Science Education,
21(2):195211, 1999.
[23] AG Priest and RO Lindsay. New light on noviceexpert dierences in physics problem
solving. British journal of Psychology, 83(3):389405, 1992.
[24] V.L. Patel, D.R. Kaufman, S.A. Magder, et al. The acquisition of medical expertise
in complex dynamic environments. The road to excellence. The acquisition of expert
performance in the arts and sciences, sports, and games, pages 127163, 1996.
[25] R.J. Sternberg and J.A. Horvath. Cognitive conceptions of expertise and their relations
to giftedness. American Psychological Association, 1998.
[26] M.T.H. Chi, P.J. Feltovich, and R. Glaser. Categorization and representation of physics
problems by experts and novices. Cognitive science, 5(2):121152, 1981.
[27] R. Glaser. Changing the agency for learning: Acquiring expert performance. Lawrence
Erlbaum Associates, Inc, 1996.
[28] J. Mason, L. Burton, and K. Stacey. Thinking mathematically. Addison-Wesley London,
1982.
BIBLIOGRAPHY 212
[29] D.N. Perkins. Archimedes bathtub: The art and logic of breakthrough thinking. WW
Norton & Company, 2000.
[30] S.K. Reed. Cognition: Theory and applications. Wadsworth Pub Co, 2007.
[31] U. Neisser. Cognitive psychology. Appleton-Century-Crofts, 1967.
[32] J. Hadamard. The Mathematicians Mind: The Psychology of Invention in the Mathe-
matical Field. Paperback printing. Princeton, New Jersey: Princeton Science Library,
1945.
[33] P.G. Liljedahl. The AHA! experience: Mathematical contexts, pedagogical implications.
VDM Verlag, 2008.
[34] G. Polya. How to solve it: A new aspect of mathematical method. Princeton University
Press, 1957.
[35] A. Schoenfeld. Mathematical problem solving. Academic press, 1985.
[36] J. Garofalo and F.K. Lester Jr. Metacognition, cognitive monitoring, and mathematical
performance. Journal for Research in Mathematics Education, pages 163176, 1985.
[37] M.E.P. Seligman and M. Csikszentmihalyi. Positive psychology: An introduction.
American Psychologist; American Psychologist, 55(1):5, 2000.
[38] C.S. Dweck. Mindset: The new psychology of success. Ballantine Books, 2007.
[39] A. Bandura. Self-ecacy. Encyclopedia of human behavior, pages 7181, 1994.
[40] J. Siegel and M.F. Shaughnessy. An interview with bernard weiner. Educational Psy-
chology Review, 8(2):165174, 1996.
[41] D.H. Pink. Drive: The surprising truth about what motivates us. Canongate, 2010.
[42] B. Weiner. Achievement motivation and attribution theory. General Learning Press,
1974.
[43] E.L. Deci and R.M. Ryan. Facilitating optimal motivation and psychological well-being
across lifes domains. Canadian Psychology/Psychologie canadienne, 49(1):14, 2008.
[44] C.S. Dweck. Motivational processes aecting learning. American psychologist,
41(10):1040, 1986.
[45] M. Csikszentmihalyi. Creativity: Flow and the psychology of discovery and invention.
Harper perennial, 1997.
[46] S.E. Dreyfus and H.L. Dreyfus. A ve-stage model of the mental activities involved in
directed skill acquisition. Technical report, DTIC Document, 1980.
BIBLIOGRAPHY 213
[47] H.L. Dreyfus and S.E. Dreyfus. Peripheral vision: Expertise in real world contexts.
Organization Studies, 26(5):779792, 2005.
[48] K. Charmaz. Constructing grounded theory: A practical guide through qualitative anal-
ysis. Sage Publications Ltd, 2006.
[49] M.Q. Patton. Qualitative research and evaluation methods. Sage Publications, Inc,
2002.
[50] B. Sriraman. The characteristics of mathematical creativity. The Mathematics Educa-
tor, 14(1):1934, 2004.
[51] J. Kruger and D. Dunning. Unskilled and unaware of it: how diculties in recognizing
ones own incompetence lead to inated self-assessments. Journal of Personality and
Social Psychology; Journal of Personality and Social Psychology, 77(6):1121, 1999.
[52] K. Duncker and L.S. Lees. On problem-solving. Psychological monographs, 58(5):i,
1945.
[53] M. Sinclair and N.M. Ashkanasy. Intuition. Management Learning, 36(3):353370,
2005.
[54] J. Bransford. How people learn: Brain, mind, experience, and school. National
Academies Press, 2000.
[55] M.E.P. Seligman. Learned helplessness. Annual Review of Medicine, 23(1):407412,
1972.
[56] M.H. Ashcraft and M.W. Faust. Mathematics anxiety and mental arithmetic perfor-
mance: An exploratory investigation. Cognition & Emotion, 8(2):97125, 1994.

Thesis Chiaka Drakes

Uploaded by

Copyright:

Available Formats

Thesis Chiaka Drakes

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thesis Chiaka Drakes

Uploaded by

Copyright:

Available Formats

MATHEMATICAL MODELLING:

FROM NOVICE TO EXPERT

You might also like