A Taxonomy for Computer Simulations to Support Learning
about Socio-Economic Systems
Frank H. Maier" and Andreas Grépler™
What Are You Talking About? A Confusion in Naming
In literature as well as in discussions among scientists or practitioners one can find some
confusion when computer-based simulation tools to support learning processes are under
consideration. Besides the unsolved question of their efficacy, this confusion often is caused— or
at least increased—by problems connected with terminology: many words are in use that
symbolize the same object, or a single word is used for different objects. “Microworld”,
“management flight simulator’, “business simulator’, “business game”, “management
simulation”, “learning environment”, can sometimes be found to describe the same kind of
simulation tool. But sometimes two instruments both called “management simulator” are quite
different. Some authors distinguish between “business games” and “business simulators”, some
do not.
For scientific research a clear, unambiguous terminology seems more than important: it is the
basis of understanding other people’s work and gives the opportunity to criticize it. Furthermore,
this will help analyzing the effects of different types of these simulation tools (see Kluwe 1993)
and building a bridge between the many involved fields of science (e. g., management and
decision science, psychology, instructional design, computer science).
The confusion is caused by various reasons: different academic backgrounds of the people
involved, marketing aspects (some terms sell better than others), and a not reflected adoption of
terms originally used with other intended meanings. In order to clarify these issues, in a first step
this paper presents and criticizes commonly used terms for learning instruments. Then a list of
possible characteristics is listed and applied to some well known tools. Finally, in order to
establish our suggestions for a coherent terminology a proposal for naming conventions is shown.
A Critique and Explanation of Some Commonly Used Names
A reflected consideration of some already used terms for computer simulations to support
learning can help to build a coherent naming scheme for these instruments. Often “management
simulator” is used in the same way as “business simulator’. Sometimes this term is also in use for
multi-person computer simulations (e. g., Milling 1995). However, its use is problematic because
*
Visiting Scholar from the University of Mannheim at Massachusetts Institute of Technology,
Sloan School of Management, System Dynamics Group, 30 Memorial Drive, E60 365, Cambridge, MA
02142, USA, Phone: 617-258-6838, E-mail: fmaier@ is.bwl.uni-mannheim.de, for home address see below.
Industrieseminar der Universitat Mannheim, D - 68131 Mannheim, Germany,
Phone: (+49 621) 292-31 40, Fax: (+49 621) 292-52 59,
E-mail: agroe@is.bwl.uni-mannheim.de
ood
not management as a group of persons or as a function within a firm is simulated, but the
company and market environment (the business). This allows to play the role of management. Or,
in other words, players do (really) manage a (simulated) firm in a (simulated) business.
Quite often the term “management flight simulator’ (MFS) is used (for single-user as well as for
multi-user simulation games; see, for instance, Sterman 1992). This apparently fortunate analogy
found wide spread use because not only under marketing aspects it seems quite promising: “like a
pilot learns to fly, one can learn managing a company with the help of a MFS”. Nevertheless,
when having a closer look this term shows two disadvantages: Firstly, at least for many persons
whose mother language is not English, the American phrase of “flying a company” which stands
for managing it is unknown. Due to that it could be misunderstood in a way that its use has
something to do with flying (“the management learns to fly”). Secondly, business simulators do
not aim to cover reality in an as congruent and comprehensive way as possible like real flight
simulators do. While flight simulators try to be as realistic as possible (including almost every
detail) business simulators try to abstract from details allowing to focus on important structures
and behaviors. At its best, the term “management flight simulator” suggests that behavior is
trained, but not that insights into the relation between structure and behavior are mediated.
“Microworld” goes back to Papert (1980). Again, the use of this term for computer simulations to
support learning seems promising. However, Papert understands “microworlds” as learning tools
which— following an “extreme” constructivistic approach— enable children to construct their
knowledge themselves. What is more, such “microworlds” do not convey one or more leaming
goals: the user is free to define himself what he wants to learn. Considering this, the term
“microworld” should rather be used for modeling-oriented software packages, which make it
possible to construct models and which also contain some sort of formalism (programming
language) to express thoughts about specific or general systems (see Laurillard 1993, p. 138, for
differences between simulations and microworlds; p. 144, for differences between microworlds
and modeling tools).
Also, simulation tools to support learning are mixed up with Decision Support Systems: the latter
are built to increase performance in short term decision making. Usually, these instruments
provide possibilities to test the outcome of decisions which have to be made (see Maier 1995,
pp. 136-160 for a description of goals and characteristics of DSSs). They do not aim on long term
changes in the users’ mental models; they are not constructed to support learning processes.
Frequently, the term “game” is used instead of or parallel to “simulator” or “simulation” (e. g.
Jensen et al. 1997). This distinction mostly seems to be influenced by the scientific tradition of
the author. Some authors therefore combine both terms to “simulation games” (e. g., Domer
1992). We do not distinguish between “business game” and “business simulator” (in the same
way, for instance, Keys and Wolfe 1996, or Klein and Fleck 1990; in contrast Lane 1995, who
distinguishes between “simulation” and “game”).
The terms “learning laboratory” or “interactive leaning environment” (ILE) usually contain more
than a pure business simulator: one or more simulation models are embedded into a computer-
based learning environment which could also include modeling tools (see Paich and Sterman
1993, for an appeal aiming in that direction). Such computer-based learning environments can
also comprise background information, original source material, and working instructions
2
integrated into one computer application (note the more general definition for “leaming
environment” as described below, p. 3). Through the use of the extended term “system dynamics
based interactive leaming environment” (SDBILE), however, it is clear that still a simulation
model is a central part of such learning tools (with this meaning used, e. g., by Davidsen and
Spector 1997). One can conclude that an interactive leaning environment is a business simulator
with additional components which are supposed to be necessary for its effectiveness.
Distinguishing Criteria of Computer Simulations
The manifold of different names and concepts led us to think about what the tools to which they
are applied have in common and what makes them distinct. The starting point was our aim to
investigate the “effectiveness of computer-based simulation tools to support learning”. Rather
rapidly we realized that such a construct does not exist in a general way: “effectiveness” as well
as “computer-based simulation tool” needs some further definition and explanation. This paper
addresses the second issue. We found that a taxonomy could be helpful. A taxonomy is a division
of objects into ordered groups or categories. The categorization indicates that there is a common
criterion which “characterizes” the category and can be used to distinguish the objects in the
particular category. However, the division of the objects (which is based on the observed
characteristics of the objects under consideration) can never be made without bias: it depends on
the reasons and the goals a taxonomy is constructed for.
Model
Simulation User
tools Interface
Learning
environment
Figure 1: Three aspects of simulation tools
Figure 1 depicts our basic assumption that simulation tools are constituted by three aspects:
underlying model, user interface, and (leaming) environment, which contains, for instance, the
situation the tool is used in, additional source material, etc. Note that in a broader definition of
“learning environment” it comprises virtually everything connected with the learning process
(see, e. g., Strittmatter and Maul 1997, p. 52). However, system dynamicists hold a strong belief
that the model and its presentation are prominent factors when building simulation tools (for the
relationship between model and user interface see Machuca 1991, and his later works). Following
this idea, we consider it to be reasonable to separate these two factors from the rest of the leaming
environment.
Of course, computer-based simulation tools can be characterized by a lot of features. The
following list is an attempt to systematize these features according to the three main aspects of
simulation tools described above. In addition to these, a fourth aspect which regards the target
group, goals and objectives of the tool is added for reasons of completeness and their potential
impact for the research about the effectiveness of computer simulations as learning environments.
Table 1 shows the four different categories.
1. Environment of application
1.1. Number of users
1.1.1. Single person
1.1.2. Multi person
1.2. Degree of integration
1.2.1. Stand-alone simulation.
1.2.2. Integration in computer-based
environment
1.3. Main area of application
1.3.1. Modeling- oriented
1.3.2. Gaming-oriented
1.4. Use of teachers/facilitators/coaches
1.4.1. Totally self-controlled leaming
1.4.2. Support by teacher/facilitator/coach
Design elements of user interface
2.1. Chance of intervention while simulating
2.1.1. Discrete periods
2.1.2. Simulation in one run
2.2. Transparency of simulation model
2.2.1. Black-Box
2.2.2. Transparent- Box
2.3. Advancing of time in user interface
2.3.1. Self-Proceeding
2.3.2. User-driven
2.4. Characteristics of users’ decisions
2.4.1. Policy-oriented
2.4.2. Decision-oriented
3. Model
3.1. Real-world domain
3.1.1. Business
3.1.2. Other
3.2. Structure
3.2.1. Feedhack- oriented
3.2.2. Process-oriented (mostly without
feedback)
3.3. Behavior
3.3.1. Deterministic
3.3.2. Stochastic
3.4. Generality of model in regard to domain
3.4.1. Special area of interest
3.4.2. Whole domain
3.5. Proceeding of time in simulation engine
3.5.1. Discrete
3.5.2. Continuous
3.6. Role of simulation model
3.6.1. Active generation of decisions
3.6.2. Clearing device for users’ decisions
3.7. Influence of extemal data
3.7.1. With such influences
3.7.2. Without such influences
3.8. Domain of variables
3.8.1. Integers
3.8.2. Real numbers
Target groups, goals, objectives
4.1. Width of target group
4.1.1. Special target group (client specific)
4.1.2. Open target group
4.2. Goals regarding users
4.2.1. Judgement
4.2.1.1. Users are going to be tested
4.2.1.2. Users are not going to be tested
4.2.2. Change
4.2.2.1. In attitude towards specific issue
4.2.2.1.1. Users are going to be
motivated
4.2.2.1.2. Motivation not intended
4.2.2.1.3. Leaming about modeled
system Domain specific
knowledge
4,2.2.1.4, Domain independent
knowledge
4.2.2.2. Mediation of knowledge about
system’s control
4.2.2.2.1. Imparting of procedural
knowledge
4.2.2.2.2. No imparting of procedural
knowledge
Table 1: Criteria for categorization of computer simulations
A strict distinction cannot always be made between listed pairs of characteristics. Furthermore,
there are likely some characteristics missing. However, this list was made to show the literally
thousands of possibilities how a computer simulation tool can be designed. Furthermore, all those
characteristics possibly have an influence on the effectiveness of a simulation game and can
therefore be varied in experiments.
But note that some combinations of characteristics do not make any sense, for instance, can
modeling- oriented tools never be black-boxes. And, of course, can simulation tools with different
characteristics in one or the other feature be summarized under one term. This is dependent on the
goals which are aimed at with the taxonomy in question. Only these prerequisites allows to
construct a taxonomy at all.
Proposal for a Taxonomy of Computer-based Simulation Tools
to Support Learning
The criteria shown above can be applied to analyze the characteristics of existing computer-based
simulation tools. However, the criteria also can be helpful to determine which of the
characteristics a new simulation tool should have. Exemplary, for some popular computer-based
simulation tools Table 2 visualizes the differences of the criteria which are considered in these
applications. The categories and criteria correspond to those of Table 1. If a criterion is applied in
the particular simulation tool, the according field in the table is marked red. In some cases both
criteria of a category are applicable, for which then the field is marked gray. The visualization
also is the starting point for the proposed naming convention of computer-based simulations to
support learning processes in socio-economic systems as shown in the tree represented in
Figure 2.
Considering the criterion “main area of application” at the root of this tree the first differentiation
is made between modeling- and gaming-oriented instruments. Modeling-oriented tools then can
be further distinguished through the criteria “structure” and “proceeding of time in the simulation
engine” into feedback-oriented continuos simulation environments like V ensim, Powersim, Ithink
or DYNAMO and process-oriented discrete simulation environments like Taylor or Simple++.
The usefulness as well as the efficacy of modeling-oriented tools for learning, problem solving
and insight is virtually undoubted within the System Dynamics community (see, for instance,
Senge 1989). The aim of process-oriented discrete simulation environments is mainly to optimize
process layouts and visualize the behavior of the system processes under consideration. Their
main real world domain is business, especially, manufacturing and logistic processes.
The second branch of the tree shows the gaming-oriented simulation tools which are further
distinguished by the criterion “number of users” into single-user and multi-user applications.
Single-user applications are defined as “simulators”, whereas multi-user applications are
designated as “planning games”. In a simulator usually a single person “plays” with and against
the computer model, as opposed to the planning games where several groups of players compete
against each other and the computer only has the role of calculation the resulting outcomes of
each group’s decisions. The research about the effectiveness of computer-based simulation to
support learning is mostly done in the area of simulators since it concentrates on the learning of
individuals. Group dynamics which may have a strong influence on learning and decision making
processes and therefore affect the effectiveness can not appear in single-user simulators.
3 26payowjo uoRePEW| —_s6paymoun jenpasaid jo Guned.ul
s sue cemala|__ SeSraSin aimed GREG
: Pompow Togs Gunes 26peyow 2yDeds UELOG|
| s105n upsesa, sieoo
3| ‘anssy aysads spuemen| epuayu!30u uoaennon|
a senpme ul abueu> PALAU aq a} GuIOH axe saisn|
Z| Paysaq 8q 0) Bulob you ae suas,
8 quawa6pn |
3] paisa 8g 0) 6UNB aIe SI95n
3] dhaib 79616 edo
© inc16 326,14 50 npn
& dnoi6 ya6ie, Dds
soyDeor/SiCIEMDeS (peoaaoeupeyieysess Aquoddns
/ssayre@ Jo oddns| Gujwea] payjaquos-jes AIJeVO | |
‘3S
#5 voneaidde| PaUaHO-bULIED
S28] seeaeuen pict bil
8
8 =I PUIUOIIAS paseq-seIndwWOD UI LOREENA
EB] vones6ay1j0 aa6aq|
28 UuoRe UIs SUoIe PUES
a
uwosied aianva
saisno soquinN
wosied 261s
5 suospap ssesn Panepouoismeq
3 BES IEREENS PeWeUO-ADI04
3 5
AA aoepaquy sosn| UeAuP5N
g 8) ur own jo Sursuenpy| ES
=
&
G8) peavuneus Seepaovel
weer A 2st
yo fouasedsues 5
5 xoq>Pe18
3 Supers Uunt auo uy uoneruans
a ym uomansenul
jo.20uey3 spousd a833510
aimanas|
soyreyaa|
eou
samgetien jo urewoal
‘sre
‘uy6ua uonenuts snonuauo5
Blu suse cuporaia TaEaG
2 aa IEP [EIEXD Jo BIUENYUT HOUMA
Lams ateiiaiak | erp jewave yo aruanyur ant | | | | |
apna UoIsbep SI98n aip.46} aD1N9p BUUEDID
uoneintus aun 40 210 32 ‘uOISDEp Jo UOIERUSB onnDv|
TREE z 8 cISDOR UNA
capadsai tam S32
Paw jo Ayjeiaua5 ga g Jseuqul JO Pave jeads
a3é PHO
ulewop PHOM [eay eZee
3 8 8 ssouna
a4
S$ 2c
ess
a 3st g a
ie a rey ft
ii FEE eee EE
s 5 S| S|E
fa] Fed Se
Table 2: Exemplary application of the distinguishing criteria
In simulators as well as in the planning games the computer model often serves as a “clearing
device” for the decisions the user has entered into the computer in each decision period. Though,
several simulators demand one or more “computer-based players” which actively take part in the
game having their own decision rules. This is the case, e.g, in the business simulator LEARN!,
where one “real” player competes against three virtual players. Therefore, in this case for the
category “role of the simulation model” both criteria are marked red. In contrast to this, in multi-
user planning games the computer model just has the function of a clearing device for the inputs
of the users. Note that in the case of the simulators LEARN! and the planning game Lobster the
only difference is that the first is a single-user version with the decisions of the three competitors
mapped through different policies within the computer model. The latter is a multi-user game
where each of the four competitors are “real” groups which have to decide based on their mental
models. The underlying computer model is apart from that identical.
Computer-based
simulations to support leaming
in socio-economic systems
¥ v
Modeling- oriented Gaming-oriented
simulation tools simulation tools
eT [
| | Feedback-oriented continuos Simulators Planning Games
simulation environments
> Vensim Bi Cc ts 1
_| Business | | Comorat | ...-planning
simulators sein OS planning games games
[>} Powersim
People Vensim’s World Hah banks
Express MES. Dynamics Lobatse diab
>) Think
>| LEARN! Okopoly Topic | “—> _ ....
Lp
|» Boor sad SimCity Marga
Process-oriented discrete
b>]
simulation environments
[>] Copy Shop Lohhausen
+> Taylor
L»| CABS
}->) Simple ++
Le.
Ly}
Figure 2: Taxonomy of computer simulations to support learning processes
in socio-economic systems
Simulators and planning games can furthermore be distinguished by the real-world domain in
which the simulation is situated. Here we differentiate between business applications and others,
and consequently name the related tools as “business simulators” and “corporate planning
games”. Since there is a variety of different real-world domain and our focus lies in the business
area, there is no further naming convention for the other simulators and planning games. The
7
nodes contain our proposal for the terms to be used in future. In particular it is suggested to use
the term “business simulator” instead of management simulator or management flight simulator,
since neither management is simulated nor the aim of these simulators is to teach the management
to fly.
Each of the leaves of the tree gives an example of an according simulation tool. Of course, this
covers only the typical usage of these products: for example, there is always the chance that a
group of people works with a business simulator like LEARN! or People Express. However, in
this case the group can be seen as a single virtual user. Excellent examples for business
simulators are the People Express and Boom and Bust Enterprises from Sloan School or LEARN!
and Copy Shop developed at University of Mannheim in cooperation with Simcon GmbH.
Examples for simulators from other real-world domains are the World-3 simulator included in the
Vensim simulation environment, the famous computer game SimCity and Domer’s Lohhausen
simulator. As examples for corporate planning games serve Lobster, Topic, and Marga, a
planning game from other domains is, for example, the Fish Bank Game.
References
Davidsen, Pal I. and J. Michael Spector. 1997. Cognitive Complexity in System Dynamics Based
Leaming Environments, in: Yaman Barlas, Vedat G. Diker and Seckin Polat (eds.): 15th
International System Dynamics Conference: Systems Approach to Learning and Education
into the 21st Century, Istanbul: 757-760.
Domer, Dietrich. 1992. Die Logik des Miflingens. Strategisches Denken in komplexen
Situationen [The Logic of Failure. Strategic Thinking in Complex Situations], Reinbek bei
Hamburg.
Jensen, Kjeld et al. 1997. An Interactive Telecommunications Business Game for Strategic
Exploration, in: Yaman Barlas, Vedat G. Diker and Seckin Polat (eds.): 75th International
System Dynamics Conference: Systems Approach to Learning and Education into the 21st
Century, Istanbul: 491-494.
Keys, Bernard and Joseph Wolfe. 1996. The Role of Management Games and Simulations in
Education and Research, in: Yearly Review, Journal of Managment, Vol. 16, No. 2: 307-
336.
Klein, Ronald D. and Robert A. Fleck. 1990. International Business Simulation/Gaming: An
Assessment and Review, in: Simulation and Gaming, V ol. 21, No. 2, June 1990: 147-166.
Kluwe, Rainer H. 1993. Knowledge and Performance in Complex Problem Solving, in: Gerhard
Strube and K. F. Wender (eds.): The Cognitive Psychology of Knowledge, Amsterdam:
401-423.
Lane, David C. 1995. On a Resurgence of Management Simulations and Games, in: Journal of
the Operational Research Society, V ol. 46: 604-625.
Laurillard, Diana. 1993. Rethinking University Teaching: A Framework for the Effective Use of
Educational Technology, London and New Y ork.
Machuca, José A. D. 1991. When Creating and Using Games, Are We Neglecting the Essential of
System Dynamics?, in: Khalid Saeed, David Andersen and José Machuca (eds.): System
Dynamics °91, Bangkok: 329-335.
Maier, Frank. 1995. Die Integration wissens- und modellbasierter Konzepte zur
Entscheidungsunterstiitzung im Innovationsmanagement [Integration of Knowledge and
Model-Based Concepts for Decision Support in Innovation Management], Berlin.
Milling, Peter. 1995. Organisationales Lemen und seine Unterstiitzung durch
Managementsimulatoren [Organizational Leaming and Its Support by Management
Simulators], in: Z/B Ergédnzungsheft 3/95: Lernende Unternehmen: 93-112.
Paich, Mark and John D. Sterman. 1993. Boom, Bust, and Failures to Learn in Experimental
Markets, in: Management Science, V ol. 39, No. 12, December 1993: 1439-1458.
Papert, Seymour. 1980. Mindstorms. Children, Computers, and Powerful Ideas, New Y ork.
Senge, Peter M. 1989. Organizational Learning: A New Challenge for System Dynamics, in:
Peter M. Milling and Erich O. K. Zahn (eds.): Computer-Based Management of Complex
Systems: Collected Papers from the 1989 International System Dynamics Conference,
Berlin et al.: 229-236.
Sterman, John D. 1992. Teaching Takes Off - Flight Simulators for Management Education, in:
OR/MS Today, October 1992: 40-44.
Strittmatter, Peter and Dirk Mauel. 1997. Einzelmedium, Medienverbund und Multimedia [Single
Medium, Network of Media, and Multimedia], in: Ludwig J. Issing and Paul Klimsa:
Information und Lernen mit Multimedia, Weinheim: 46-61.
