You are on page 1of 3

A System Dynamics Model of the Student Learning Process1

Abstract: Based on our mental model of our students' learning process, we present a system dynamics
model that allows the simulation of student success in a given course. The major input parameters of
the simulation are the student academic potential at the beginning of the course (student learning
ability), the student expectation in the course (grade objective), and the course evaluation method. The
simulation monitors the weekly learning performance of the student as well as his weekly-cumulated
grade, and his morale.

Introduction
The System Dynamics Society [1] defines system dynamics as follow: System dynamics is a
methodology for studying and managing complex feedback systems, such as one finds in business and
other social systems. Modern system dynamics tools use the stock and flow modeling framework to
quickly create a model that is easily understandable by non-experts. Although the origin of system
dynamics can be traced back to 1910, it was officially born in 1961 with the work of Jay W. Forrester
[2]. Since then it has been applied in various fields (corporate planning and policy design, public
management and policy, biological and medical modeling, energy and the environment, theory
development in the natural and social sciences, dynamic decision making, complex nonlinear
dynamics, etc.). It is mentioned as an important project manager tool for software engineering in the
famous Tom De Marco Deadline novel [3].
Recent emphasis [4] on introducing metrics to evaluate the productivity of learning organizations,
including Universities, led us to reflect on our mental model of the student learning process in the
courses we offer (upper-level undergraduate computer science courses). By expliciting our mental
model in an easily understandable language (stock and flow), we aim to achieve the following
objectives:

To gain a better understanding of our mental model of the student learning process,

To create a common basis for discussions with colleagues,

To experiment with the impact of various parameters on the student success in a course,

To build a preliminary framework for modeling and simulating the student learning process,

To calibrate the model with our students' current achievements and detect any parameter not present
in our mental model,

To improve our mental model with explicit modeling that reflects quantitatively the reality of our
teachings,

To improve students' success by providing a rigorous assessment framework to the introduction of


new teaching methods including the definition and negotiation of success metrics.

The current model presented reflects the mental model of both authors about the learning process for
one undergraduate student in computer science during a one-term course. While others have applied
system dynamics, or at least causal loops, to the education environment [5,6,7], our model builds from
our current teaching experience and provides a common basis to discuss various mental models used by
other instructors.
1

Yves Lucet & Patricia Lasserre, ylucet | plasserre@ouc.bc.ca, Computer Science Department, Okanagan University
College, 3333 University Way, Kelowna, BC V1V 1V7

Model Foundation
The core concepts of knowledge, Morale and Grade make the foundation of our mental model. In our
model, Knowledge represents the accumulated knowledge, skills, know-how, etc. that a student
possesses. Intuitively, the more Knowledge a student has, the higher grade he should get.
Grade is the grade percentage the student obtains.
During the term, Grade reflects the student success at
the different evaluation methods (midterms, final,
laboratory assignments, etc.). It is calculated each
week and accumulated to provide the total grade for
the course at the end of the term. A high grade implies
a high Morale.

Causal Loop Diagram

Knowledge

Morale reflects the general spirit of the student as a


consequence of external factors (personal tragedy,
part-time work, other time-eating tasks, etc.) and
internal factors (5 or 6 course load in the term, other
course assignments, etc.). A high Morale implies a
greater success rate at midterm and final exams, so it
translates into higher grades. It also translates into a
better frame of mind to learn i.e. a higher weekly

Grade

Morale

learning rate.

Stock and Flow model


The Knowledge part of our model consists of a Total Knowledge to learn reservoir, that is drained
weekly at the assigned weekly workload rate into a Knowledge to learn reservoir. The later represents
the assignments and other tasks the student has to complete so far. It is a reflection of the knowledge
the student has to learn. The student initial learning ability as reflected by the Weekly Learning Time
constant, and the morale as reflected by the effect of morale on weekly learning rate variable, are
combined to compute the weekly learning rate that determines how the student learns during the
current week. The resulting knowledge is reflected into the Knowledge learnt reservoir.
0.00 kng/wk

assigned weekly
workload

0.00 kng

55.00 kng
5.00 kng/wk
75.00 kng

Total knowledge to learn

Knowledge to learn

weekly learning rate

WEEKLY LEARNING
TIME

Knowledge learnt

effect of morale on
weekly learning rate

Similar sub-models are included in the global model to incorporate the evaluation method, Grade, and
Morale. The above figure also shows the units used in the model. Knowledge has a knowledge unit
(kng), Morale has a morale unit, and the time unit is in week (wk).

Simulator
STUDENT SUCCESS SIMULATION

Knowledge

Assigned

Learnt

kng

kng

kng

100

100

100

50

50

50

STUDENT INFORMATION
(INPUT)
Grade

30
20

40 50 60

10

70

Excellent
Good
Average
Pass
Fail

80
90

0
0

Student learning ability

100

Non-commercial use only!

Grade objective

10

20 30

40 50 60

70 80

90 100 %

Non-commercial use only!


100

COURSE INFORMATION (INPUT)


Student Morale (morale)
GRADE OBJECTIVE (%)

Lab weight

Grade (%)

0.0 0.2 0.4 0.6 0.8 1.0

projected grade (%)

50

Non-commercial use only!

Midterm weight

0.0 0.2 0.4 0.6 0.8 1.0

Non-commercial use only!

0
2/4

3/9

4/14

Final weight
Non-commercial use only!

0.0

0.2

0.4

0.6

0.8

1.0

Non-commercial use only!

The above picture shows the simulator after running a simulation for a Pass student (50%) who has a
very low-grade objective (15%). The course evaluation consists of 10 labs, 2 midterms and a final with
respective weights of 20%, 20%, and 40%. The final grade obtained is 50%, since some knowledge
was not learnt due to the student limited learning ability. The graph shows an increasing morale
corresponding to the student grade outperforming his grade objective.

Conclusion
The current model achieves some of our objectives: it expresses our mental model, provides a common
language to discuss student-learning models, and allows some initial simulations. In the future, the
model needs to be validated with surveys.

References
1. System Dynamics Society, http://www.systemdynamics.org/
2. Jay W. Forrester, Industrial Dynamics, MIT Press, 1961
3. Tom De Marco, The Deadline: A Novel About Project Management, Dorset House Publishing Co.,
Inc., 1997
4. William Post, Building Minds, Not Widgets, IT Pro, Sep-Oct 2004.
5. Trent Miskelly, Interactive Student Modeling, 1998, ACM library
6. Jeffrey J. McConnell, Active Learning and its use in Computer Science, 1996, ACM library
7. Elaine Lizeo, Understanding the Learning Process in Work Groups, 2004

You might also like