COMPUTER-BASED CASE STUDIES IN
[AGEM! TION AND RESE,
Alan K. Graham
Peter M. Senge
John D. Sterman
E40-294, MIT School of Management,
71 Mass. Ave., Cambridge, MA 02139, USA
John D. W. Morecroft
London Business School
Sussex Place, Regent’s Park London NW1 4SA, UK
Abstract
There is growing interest in combining system dynamics models with conventional case studies in order to
create learning environments for management education. Such computer-based case studies promise im-
provement in strategic thinking skills and better integration of modeling in the policy and strategy area.
Moreover, these models-with-cases are tangible products with which to conduct research in computer-
based learning. This paper surveys the opportunities for using computer-based case studies in manage-
ment education and for conducting novel research on management learning.
The paper first examines how models-with-cases fit in the established areas of strategic management and
business policy. Which issues, in the broad range covered by strategic management, do models-with-
cases address? Next, two current examples of computer-based case studies are presented (People Express
Airlines and the Intecom PBXs) to show explicitly how cases and models are combined and used.
Finally, the paper explores research questions that arise in conjunction with such work: 1) how to teach
effective inquiry skills, 2) how to teach conceptualization skills, 3) how to enhance the ability to apply
learned theories to new situations, and 4) adapting measurement methods to evaluating the educational
effectiveness of computer-based cases in teaching these mental skills.
Introduction: Where do models fit in?
To understand where system dynamics models, especially models-with-cases, fit into management educa-
tion, consider first how business policy and strategy are traditionally taught. Case studies are the comer-
stone: The objective is to inculcate skills to “think strategically”, to “view the business as a whole” or
“adopt the perspective of the general manager”. A typical case study is 20-30 pages in length, and con-
tains around 10,000 words of text, several diagrams and several pages of numerical information (often fi-
nancial reports and market data). Usually case writers provide a brief history of the company, a descrip-
tion of its products or services and information on competitors. Then, depending on the case writer's
purpose, the case may describe operating details of manufacturing, marketing, and distribution, or delve
into human resource policy, systems of administration and control, organizational structure, company
traditions and values, management style, leadership or personalities. Case teachers use this descriptive
and numerical information to trigger classroom discussion about the business, its administration and its
strategic options. But how and what do students learn?
Case teachers do not teach answers to cases. They offer frameworks that guide class discussion and help
learners organize case information and form opinions. For example, Porter's (1980 and 1985) competitive
analysis framework is now widely used to guide discussion about business unit competitive strategy (see
the process in action in Porter 1988). Whether the subject is supercomputers, water meters, airlines or
fashion watches, the framework prompts critical thinking and debate about questions like: is this an at-
tractive industry to be in, what stops competitors from entering, what motivates customers to buy, are
suppliers or distributors in a position to siphon-away the firm’s profits? It encourages close and critical
318
examination of the value-added activities upon which a company builds and sustains a unique competitive
edge. The framework provides the case teacher with a checklist of thought-provoking questions, and
graphics with which to collect and select participants’ comments.
At the end ofa 3 hour case class, learners see several chalk boards-worth of writing, based mostly on their
own comments, with a visual layout controlled by the case teacher (Ghemawat 1986). What has been the
pedagogical value of the process? First, learners feel involved—their opinions and comments have re-
ceived attention. Second, the instructor has imposed discipline on the discussion through the use of the
framework, but not a heavy discipline. The framework provides broad brush questions and will accept a
range of comments. With repeated use, and a lecture on competitive analysis, learners begin to adopt the
framework. Third, the class as a whole has a shared focus for debate—the information displayed on the
board. (The same learning benefits occur when strategic frameworks are used'to organize management
team discussion of live business problems, as in Morecroft 1988b).
Business policy and strategy courses cover topics in several dimensions. The strategies in question range
from strategies for the business unit/single business firm to corporate strategy (strategy for a complex
multi-divisional, multi-business firm). The considerations can range from the competitive and economic
forces that shape strategic options and choices, to administrative and organizational constraints.
Different frameworks develop different dimensions of strategic thinking. For example, McKinsey’s “7
S” framework (a simple diagram showing 7 labelled and interlinked bubbles, Hax and Majluf 1984, pp.
94-96) helps trigger thinking about constraints on strategy—the so called “fit” between strategy and struc-
ture: can the existing organization’s people, administrative systems, shared values, skills, and or-
ganization chart, execute the strategy? Portfolio frameworks like the famous growth-share matrix of the
Boston Consulting Group (Hax and Majluf 1984, Chapter 7) can guide discussion about corporate portfo-
lio management—which businesses should a multi-business firm be in, how should corporate investment
funds be allocated, what criteria should the corporate management use to evaluate and compare business
unit performance, how should one segment the business for strategic planning purposes?
Computer-based case studies retain a strong process flavor—they stimulate and guide discussion, without
stifling it. By direct analogy with Porter's framework, the model, along with the systems thinking con-
cepts that accompany it, shapes the case discussion. (Later sections will describe the elements of systems
thinking.) The Intecom case described later uses STELLA maps (Richmond et al. 1987) combined with
“chalk talk” to fill the need for shaping. In addition models-with-cases allow “what-ifs” and role-playing
ina closed feedback system, which add important new process dimensions that cannot be replicated in the
conventional case method. Learners are drawn into the case by taking the role of key decision-makers,
exercising choice and judgement, and seeing the consequences. The People Express case described below
shows what is possible with decision-making simulation and the simple what-ifs. (See Graham and
Senge 1988 for additional discussion of how models extend the case study method.)
System dynamics models are particularly suited to understanding the coordination of operating policies—
how to design a set of interlocking policies and programs that support rather than frustrate strategic objec-
tives. This strength of dynamic modelling positions models-with-cases clearly at the level of business unit
strategy rather than corporate strategy and portfolio management. In addition, at the business unit level,
models can illuminate the so-called administrative issues (goal formation, incentives, time allocation, pri-
ority setting, information deficiencies, behavior and motivation) rather than market and micro-economic
issues (as covered by Porter's framework). Models-with-cases, then provide an important link between
strategy formation and implementation at the operations management level. Thee is no claim, then, that
models-with-cases have across-the-board applicability to all strategic issues, but instead make unique and
much needed contributions to developing strategic thinking skills for operating policy design.
Case 1: People Express Airlines
The Rise and Fall of People Express. One of the most popular Harvard Business School cases in recent
years focuses on People Express (PE) Airlines (Whitestone 1983). PE went from startup in 1981 to the
fifth-largest US airline in less than five years. Behind its meteoric rise were deep discount prices that
made an airline ticket comparable to a bus ticket on many east coast routes, a host of innovative “new
management” policies, such as universal employee ownership and job rotation, and a charismatic founder,
319
Don Burr. But, the spectacular success of People Express’ startup was not sustained. In the first six
months of 1986 the fm lost $132 million, and, in September 1986, the firm was purchased by Texas Air.
Business school professors have used the People Express case to examine a broad range of issues in
growth management, industry deregulation, human resources, and executive hubris.
A Strategy Model, The PE simulation model builds on the generic “market growth” structure of inter-
actions between developing demand and satisfying demand (Forrester 1968 and Senge 1986). The model
includes a human resources sector with multiple experience levels, and effects on productivity of stock
ownership and profit-sharing, morale, and tumover. The model assumes a two-tier market with es-
tablished air travelers and “new” air travelers attracted by lower fares, with “market diffusion” driven by
advertising and word of mouth. Potential customers respond to availability of routes and scheduled flights,
fares, service quality, and “service scope” (the range of services offered). The model includes competitor
price response based on PE’s market share (Sterman 1988).
The Lessons, For the past two years, the System Dynamics Group at MIT has used PE as a computer-
based case to show how a firm can develop a set of policies that make sense individually but fail in con-
cert. At the heart of People Express’ demise were inherent contradictions between its innovative human
resource policies and its aggressive growth policies. Don Burr wanted to demonstrate “a new way to run
an airline,” to show that a non-hierarchical structure, democratic principles, trust, and shared economic
risks and rewards would produce a vastly more productive workforce. He also was intent on becoming a
major player in the airline business in a few years, to which end he expanded fleet capacity 100% per year,
marketed aggressively, and maintained exceedingly low fares. The result was inadequate time to develop
management skills, falling service quality, declining morale, and high turnover.
A Game and Workshop, For learning purposes, the model has been converted into a decision-making
simulation (called the “People Express Management Flight Simulator”), which allows leamers to decide
each quarter year how many planes to buy, how many service personnel to hire, and how much to spend
on marketing. The learner can also set price and service scope. For a day-long session, learners receive in
advance the Harvard Business School case writeup (Whitestone 1983), a fifteen-page briefing book that
summarizes the assumptions of the gaming model and gives instructions on how to use the game (Sterman
1988). The morning focuses on interleaving game play by teams of 2-3 learners per computer with in-
structor-lead discussions.
The afternoon concludes the workshop with the instructor showing simulation experiments and generaliz-
ing lessons (both on the business results on strategy and human resources and on general systems thinking
principles) illustrated by the case. The People Express Flight Simulator was used in Fall 1988 by all 200
incoming masters’ students at the MIT Sloan School of Management in a day-long orientation, and is be-
ing used regularly in courses on operations management in service industries, behavioral decision-making,
and system dynamics.
Case 2: Intecom and the PBX Market
Background. In the early 1980s the AT&T operating companies (such as New England Telephone and
New York Telephone) were converting their highly profitable base of electromechanical telephone switch-
ing systems (known as PBXs) to new electronic PBXs, in an increasingly competitive market. A key is-
sue facing senior managers was how to retain a high share of the installed base of PBX systems while mi-
grating customers from old to new technology systems. The strategy for managing the migration required
executives to think about pricing, the size of the sales force, motivation and compensation, and the actions
of competitors. In the event, some of the operating companies lost 60 percent of their market share during
the migration while others lost only 20 percent!
The Model and Case. A management team from one operating company commissioned a system dynamics
project to help the team think more closely about the migration strategy. It is this model (Morecroft 1984)
which is combined with the Intecom case (Ghemawat 1985) in order to replicate for students the insights
gained from the project. The Intecom case deals with the entry of Intecom, a new company affiliated with
Exxon, into the top-end of the US PBX market. Most of the case is devoted to describing switching
products, PBX technology, customers, channels of distribution, installation, service and manufacturing.
There is also information on competitors and deregulation. The appendices contain industry level data on
320
the installed base, line shipments, market shares, manufacturing costs and corporate financial perfor-
mance.
The Process. The case provides a wealth of background information on the industry, thereby mimicking
the experience base of the AT&T management team. Students are asked to read the case in advance and
think about the following questions: what is a relevant measure of market share in the PBX market, how
do you set reasonable sales objectives for a migration strategy, what pricing options would you consider
for old and new systems?
The case discussion lasts for 3 hours and is organized around 3 STELLA maps (Richmond et al. 1987) of
increasing complexity that represent visually the migration strategy. The maps are displayed on a Macin-
tosh computer linked to a large-screen projection system. The first map shows a base of electromechanical
PBXs (represented as a single stock) being depleted by migration rates which flow into a base of electronic
PBXs shared between competitors and the operating company (represented as two stocks). The case
teacher can use this very simple model to clarify the near monopoly starting position of the operating com-
panies, to trigger discussion of the changing regulatory environment and to think about market share defi-
nitions (share of base or share of sales?). He can also introduce simulation by posing the question and
simulating the answer to—how will the base change if migration proceeds at a constant (specified) rate for
the operating company and for competitors?
‘The case teacher then sets-aside the computer and reverts to conventional chalk talk in order to make the
class think carefully about the customers—who are they, why should they migrate, what factors influence
their decisions, what are their motives and incentives? Here the behavioral underpinnings of system dy-
namics (Morecroft 1985 and 1988b, pp. 24-33; Sterman 1987a, b, and.c) provide a checklist of leading
questions to structure the discussion and record people's comments. The second STELLA map incorpo-
rates the main features of the chalk talk and shows explicitly how sales effort, prices and customer be-
havior influence migration.
Class discussion next turns to the sales force. What is it like to be a systems salesman, how might you
spend your time, how do you set priorities, what motivates you, how do you decide to allocate your
time?—a series of questions that makes the class probe the vital issue of what determines sales effort. The
final step is to talk about the competition. A crucial point here is to have the class recognize that the com-
petitors (as a group) are growing companies that feed on their own success. The structured discussion
lasts about two hours and leads into the third STELLA map which incorporates sales effort, prices, sales
time allocation, customer behavior and growing competitors.
The students are now divided into teams of five, and sent away for half an hour to devise their own
migration strategy. The “levers” they have at their disposal are price (for both old and new systems) and
sales force size—a subset of the real levers that were available to AT&T management. Each team is given
ahard copy of three STELLA graph functions on which they can draw time profiles for price and sales
force, The teams are asked to justify their choice of time profiles and predict the likely effect on market
share, competitor sales, and sales expense to revenue ratio. During the remaining class time as many
teams as possible are given the chance to explain their customer migration strategy and to simulate it on-
line in the classroom.
The Lessons. Simulations show a wide range of outcomes. Some teams lose market share rapidly and
also incur high sales expenses. Others maintain a much higher market share with low expense to revenue
ratio. This diversity resembles the diverse outcomes of the real AT & T operating companies. The suc-
cessful teams recognized that a high lease price for the old base can be very profitable (milking the old
base) but also creates vulnerability if competitors aggressively market the new technology. However, by
using some of the extra funds from the old base to finance sales force expansion, growth of competition is
restricted, and the expense to revenue ratio is reduced. The model-with-case also demonstrates that fea-
tures of behavioral decision-making (customers’ imperfect knowledge of products, technology and prices)
create both opportunities and pitfalls in the implementation of strategy.
321
Research questions on teaching systems thinking
The amount of research into management education pales in comparison to its importance in society. For
better or worse, the foundation of American management is taught in MBA programs, and European man-
agement education is moving toward a similar system. The performance of entire economies would seem
to rest on the quality of management education, but research on measuring and increasing the effectiveness
of management education is nearly nonexistent, at least by comparison to, say, economics, which is in
tum paltry by comparison to research in the physical sciences. The process of teaching what is known
about management has changed little from a century ago. Researchers write articles and books, and
teachers as individuals use the books to build curricula, with little or no evaluation of effectiveness or
comparison with alternatives, and few resources available for innovation. There is adequate research in
management, as evidenced by publications and the number of experts in business schools and corpora-
tions. But possible teaching technologies for passing that knowledge to younger generations are sorely
neglected.
In our view, perhaps more important than merely creating a new teaching technology is measuring its ef-
fectiveness in objective, scientific fashion, and understanding as well how and why it is effective. What
follows here is a taxonomy of what we believe to be the three facets of effective systems thinking as it is
brought to bear on strategic and policy issues, and the contributions we expect system dynamics and
computer-based case studies to make to the teaching of each facet of systems thinking skill.
What are the elements of systems thinking skills as they apply to strategy? Consider what happens in the
design of a strategy. Often, new strategy initiatives begin with someone, the CEO perhaps, stepping back
from day-to-day business operations and coming up with a new idea that in retrospect is obvious. There
seem to be at least three elements of the process. First, asking the right questions to discover the most im-
portant facts might be called investigation. Second, stepping back and ignoring some facts while ex-
pressing others in more general terms, recasting them into a new framework, is conceptualization Third,
if the “creative leap” recasts the situations so that an already-known theory of business applies to it, that is
theory application or transfer of learning to a new situation.
The stock-in-trade of professional system dynamics modelers, like effective executives, similarly involves
the same skills of investigation, conceptualization, and transfer. But unlike most executives, most model-
ers believe their skills to have been developed by a systematic process of education and apprenticeship.
The central research question here—the hypothesis to be tested—is that elements of systems thinking,
modeling, and case studies can enhance the skills of investigation, conceptualization, and transfer.
Investigation / Learning Strategy. Morecroft (1988) describes simulation models and model-based gam
as microworlds, which are simpler and create outcomes faster than reality. Senge (1989b) advocates ust ‘sing
models to learn from (rather than to convince by weight of authority). Unfortunately, research suggests
that a disappointingly small fraction of people have learning skills sufficient to discover how things work,
in the microworlds, and by extension, in real life. Wheatley, Homaday, & Hunt (1988) found that stu-
dents enjoy standard management games but argue that performance in the classroom or later life is not
enhanced by games. Senge (1989a) describes a workshop where participants played a game (based on a
dynamic model) much like they would a video-game, with fast play, little time out for discussion or
reflection, no explicit leaming strategy, and in the end, very little learning. Moreover, even “knowing the
ground rules” through extensive conceptualization of the model structure before playing game didn’t allow
the participants to discover how to change the behavior of the system.
The primary contribution of computer-based case studies is creation of an environment in which actual in-
ve tion can occur. A lecture about investigation is no substitute for actual investigation, any more than
being told about solving algebra problems is a substitute for actually solving them. An important side ef-
fect of doing actual investigation is fun. The leaming process incorporates discovery, enjoyment, play,
and sometimes competition. People can sustain interest in the subject matter, and the learning experience is
more powerful and enduring.
A second contribution the models-with-cases can make to investigative skills is to support using the scien-
tific method -- hypothesis formation, designing tests, evaluating results of tests, With a microworld, the
computer can either be used to efficiently explore a large number of meaningful experiments, or it can be
322
used to randomly search for winning strategies, with no idea as to why those strategies are effective.
Senge (1989a) suggests that the critical elements of effective learning strategy are a clear business strategy
with an explicit rationale and expectations of behavior for key variables before playing, coupled with care-
ful explanations of gaps between expected and actual behavior after playing. Indeed, participants the the
Sloan School's People Express exercise are asked to state explicit predictions (hypotheses) about the ef-
fects of policies, and discuss results afterward.
A third contribution of system dynamics to investigation is fumishing conceptual tools and results. Asa
profession, system dynamics has a body of knowledge about how problems in complex systems arise,
and a language for describing them. These aid conceptualization also, but to the extent that systems prin-
ciples tell one what to look for, they also increase effectiveness of investigation. For example, if a prob-
lem persists despite efforts that apparently should solve it, one should look for feedback loops that com-
pensate for the solutions and perpetuate the problem behavior. Many of Senge’s “System Principles for
Leadership” (1986) are in fact principles to aid investigation.
A fourth contribution is computer support for reflection and hypothesis testing . Intellectual skills, unlike
physical skills, require more than just repetitive practice. People need time out to conceptualize what has
and leam from it at an abstract level suitable for application to other situations. Ina typical
many-leamers-one-instructor situation, it is easy for an individual to “spin his (or her) wheels” trying
strategies that aren’t working. In a one-on-one situation, an instructor can call time out and suggest
different strategies; eventually, computer software will detect repeated failures and offer suggestions.
Similarly, one-on-one instruction frequently uses explicit statements of hypotheses, experimentation, and
pointed questions that encourage the leamer to look at the results carefully, compare them to the original
hypothesis, and state what has been learned. Again, the same software that allows a leamer to experiment
with different policies should perform these additional pedagogical functions.
Conceptualization “Abstraction. Conceptualization complements investigation: once information is ac-
‘quired, some skill is required to use it effectively. It has been said that asking the right question is halfway
to finding the answer. This is particularly true in dealing with complex business situations, where creating
the right concepts makes things almost obvious. (Graham 1977 Chapter | provides detailed discussion on
the role of conceptual tools in the creative and analytic processes.)
The primary contribution system dynamics makes to conceptualization skills is a well-worked out hierar-
chy of formalisms for moving between detailed descriptions of a situation and abstract descriptions. The
most-discussed advantage of the case study method is the real-life detail. Dreyfus and Dreyfus (1986, pp.
16-19) cite experiments where real-life detail was critical to problem-solving ability, even though the
details were not logically necessary. But if the case materials have a truly realistic amount of detail, learn-
ing from the case becomes as difficult as learning from real life. Case studies, both written and computer-
based, can be made almost arbitrarily detailed or misleading. The level of detail with which learners must
deal, then, is an important dimension in the educational process, and models-with-cases offer carefully-
controlled treatment of level of detail.
The People Express case handles the conceptualization and abstraction processes through a series of tran-
sitions from the detailed to the abstract: It begins with the Harvard Business School case (rich in detail,
without exposure of much underlying theory), goes to the briefing book (almost as rich in detail, but
organized to show cause-and-effect structure more clearly), and then immerses learners in the simulation
game (much less extra detail about cause and effect, but detailed, quarter-by-quarter decision making, and
clearly dynamic behavior). The instructor then analyzes simulations (where decisions are replaced by de-
cision rules and examination of behavior at particular points in time is replaced by examination of curves
over time), and finally, discusses behavioral principles.
The Intecom PBX case similarly uses a series of transitions, beginning from the Harvard Business School
case, to alternating use of STELLA maps (that build up the cause and effect structure in small steps) and
board diagrams (that allow detail-rich discussion of customer, sales force and competitor behavior), to
small group discussion of migration strategy, to instructor-led consideration of simulations, and finally
discussion of strategic lessons (that ties simulated behavior to system structure). (Recent experiments to
involve management teams in conceptualization and model-based leaming have also made specific use of
transitions using conceptualization tools—starting from loosely structured discussion, to maps, to friendly
323
algebra and to partial model simulations, Morecroft, Lane and Viita 1989). Kim (1989) describes the use
of conceptualization tools in a learning setting; Gould’s (1989b) research will investigate their effective-
ness.
At some point, sufficiently elaborated software should support learners in the conceptualization and ab-
straction processes. Other business simulation games offer a very detailed environment, in terms of re-
ceiving memos on the current state of the (simulated) business, memos asking for decisions, newspaper
reports on the state of the (simulated) industry, and so on. At first, such detail is useful for orienting a
learner and giving a reassuring sense of familiarity. But later, such details become clutter which obscures
the underlying issues. Giving learners the ability to turn off such “realism” may ease the conceptualization
of the dynamic issues in more abstract terms. Likewise, the “management information system” of the
People Express game gives more detailed information than learners can usually use in playing the game.
Asa leamer’s understanding of the dynamic issues grows, it may be productive for the learner to tailor the
information system to display a succinct summary of the information normally used in decision making.
Finally, as leamers begin to see that decisions are created by decision rules, it would be useful to allow
learners to specify decision rules instead of continuing to plow through game play time step by time step.
A second contribution, tightly linked to the first, is simulating the modeling experience. Neuhauser (1976)
argues that the value of games comes from the process of building the model used in the game; thus the
game designers find them interesting and useful while the leamers who merely play the games may be-
come bored and disenchanted. Toval and Flores (1987) make the same claim, and argue it is the process
of model-building which generates deep knowledge of the systems under study. Certainly this is the
experience of professional model builders.
As an approximation to the model-building experience, one computer-based case study uses explicit con-
ceptualization exercises prior to playing the computer simulation game (Kim 1989 and Moissis 1989). The
leamers use causal loop diagrams coupled with paragraph-long “mini-cases” to begin thinking about major
interactions in their particular case. As a result, they are thinking about important feedback processes
evident in their case, and be formulating their own initial hypotheses regarding causes of behavior and
possible remedies, all in relatively abstract terms consistent with a system dynamics model and game.
Their conceptualizations can then be compared to assumptions built into the formal model. This gives the
leamers knowledge of what's in “the black box.” It also helps to develop familiarity and skill with think-
ing how feedback process can explain behavior, and of course, conceptualization skill.
Theory Application/ Transfer Skills. The ability to apply knowledge to new situations is the fundamental
goal of all education. Even so, we have no directly-measured clinical experience with the reverse of the
transitions described above, where leamers begin from an abstract theory and apply it to another real
situation. Prior research suggests transfer of concepts from one situation to another is difficult, though
some investigators are rather more optimistic than others (Nisbett et al. 1987). A large literature suggests
the traditional case method is not well suited to teaching systemic thinking and the ability to transfer in-
sights from one problem to another (Kardes 1987, Nisbett et al. 1987). Kardes suggests that an important
reason is the relative weakness of theory teaching, compared to the example teaching in the case method.
He cites experiments where teaching by examples was more effective than purely theoretical teaching, but
teaching that combined theory and example was more effective than either pure case. Equal amounts of
time were spent for each trial .
The experiment cited by Kardes, then, suggests a major contribution available from models-with-cases:
An explicit, well-worked out theory relating directly to the case materials, and in a form amenable to trans-
fer to new situations. (As an example of the desirability of integrating theory and cases, consider the ma-
jor success of Porter’s framework for strategy analysis.) A model not only provides a game-playing envi-
ronment for experimentation, but it also provides a tested theory of why the system behaves as it does, and
why some policies succeed and others fail. Moreover, the theory is stated in relatively abstract terms that
facilitate application to new problems. Bakken’s (1989) and Gould’s (1989b) research will measure
ability to transfer learnings from one case to another and hopefully identify factors that aid or hinder that
transfer. Bakken will measure it by game performance, and by assessment of verbal protocols (running
commentary by learners as they play).
324
Asecond contribution comes from the ability to identify cases that deal with common dynamic issues. The
idea has existed within the system dynamics field for twenty years or more that there is a small number of
fundamental corporate problem behavior modes (perhaps twenty), and that an equal number of moder-
ately-sized models could describe the fundamental lessons needed to control them. This is the idea of the
generic model of a corporate problem—a model with “no brand name.” (Paich 1985) discusses the con-
cept at a general level; (Graham and comin 1985) describe a generic model of the problems encountered
in automating production; (Graham 1988
offers a candidate list of generic corporate problem modes.
For example, the People Express model embodies the same dynamics identified in (Forrester 1968) and
that have recurred with variations many times since in system dynamics studies.
From a somewhat different perspective, (Sterman 1987b and c) and (Diehl 1989) begin to identify generic
deficiencies in the way people normally perceive and act on problems embedded in complex feedback sys-
tems.
How should systems thinking skills be measured?
Throughout this paper it has been suggested that computer based case studies can enhance the quality of
learning. How can the effectiveness of a model-with-case be determined? There are a variety of possible
measures;
1.
2.
Self-evaluation. Let participants rate how much they leamed. This has the difficulty raised by Wheat-
ley, Hornaday, & Hunt's (1988) of reporting enjoyment without significant and lasting learning.
Game performance on the model of the case. This can either be simple scoring (did learners under
conditions A score higher than under conditions B?) or by mathematical analysis of the changes in im-
plicit decision rules being used (as in Sterman 1988b and c, Moissis 1989).
Protocol analysis of game play. Transcriptions of running commentary by game players coded (by in-
dependent coders) for systems thinking content.
Verbal or written examination on the case., he traditional grading scheme for case studies. To reach
scientific conclusions, however, a reliable and repeatable grading scheme is needed. And such
evaluations measure the ability to talk about issues rather than willingness or ability to act on them,
either in a game or on the job.
Game performance on a model of a related case. Unlike the measures above, this metric explicitly
measures the ability to transfer insights from the case to new situations.
Verbal or written examination / discussion of a related case.
Self-reporting on usefulness on the job. This measure rests heavily on the perceptiveness and in-
trospection of the learner, which may be taxed to identify changes in problem-solving approaches, es-
pecially if several people in the work group have had the same training: “We've always thought sys-
temically like this, haven’t we?”
. Protocol analysis of performance on-the-job. Transcriptions or notes taken by an observer of actual
business meetings are scored for systems thinking content (or specific lessons from the original case).
(Anderson 1977) used this procedure to measure the use of systems thinking concepts in a bureau-
cracy; (Argyris et al. 1985) make more general use of this investigative technique to measure the
dominant concepts people use in practice. Any such experiment has difficulty in controlling for e.g.
the background of the participants, the investigator, and even the presence of an investigator (i.e. the
Hawthorne effect.) Wolfe (1985) notes the extensiveness of such difficulties as a partial explanation
for the diversity of conclusions about the effectiveness of management games.
. Job performance. Improving job performance is the ultimate goal of management education, but the
effects of education on job performance is the most difficult.measure for which to maintain
experimental controls.
There is a spectrum of measurement methods; in general, the closer a measure is to measuring actual use-
fulness, the more logistical and control difficulties it presents. Because somewhat different characteristics
are measured by each, and because the effects of computer-based case learning on job performance is un-
likely to be measured directly, a multiplicity of the less direct measures are needed to form valid conclu-
sions about effectiveness of management education. For example, in one study in progress, managers at
325
the Hanover Insurance Company participate in a three-day “learning laboratory” in which a CBCS and
simulation game are used (Senge 1989). The effectiveness of the workshop is measured by (i) leamers’
subjective reports of the usefulness of the program, (ii) demand within the organization for the workshop,
(iii) subjects’ performance in the game relative to optimal performance, and (iv) performance on an
examination to be administered some time after the ‘workshop (Moissis 1989).
Work on measurement methods is progressing in several projects. As mentioned above, the experiments
described in (Moissis 1989), (Gould 1989b), and (Bakken 1989) are under way at MIT. The Educational
Testing Service of Princeton is currently attempting to define metrics for systems thinking so they can
evaluate the effectiveness of system dynamics in high school curricula including physics, chemistry, and
social see: Pilot programs are underway in Brattleboro Vermont and several other school districts in
the United States.
Conclusion
Effective management education is important to individuals, corporations, educational institutions, and
ultimately, society in general. The preceding discussion has strongly suggests that considerable im-
provement is available in management education, in certainty about how to teach, in formalizing instruction
in the area of systems thinking applied to strategic analysis and execution, and in ‘technologies available to
assist in teaching. Computer-based case studies—models-with-cases—seem to offer considerable
potential in all of these areas.
Moving forward with computer-based case studies will require collaboration among numerous stakehold-
ers: Corporations and managers to fund, to serve as test sites for, and in some instances, be final users of
models-with cases; Modeler/researchers to create the cases and do the painstaking measurement necessary
to establish findings with certainty; and universities to provide institutional support for both research into
management education and for integrating models-with-cases into the standard curricula. Such a complex:
collaboration will require considerable time and money from all concemed, but the payoff portends to be
orders of magnitude larger.
References
Andersen, D. F. 1977, “Mathematical Models and Decision Making in Bureaucracies: A Case Story Told from Three Points
of View.” Cambridge, Mass.: MIT School of Management Ph. D. Thesis.
Argyris, C., R. Putnam, and D. Smith. 1985. Action Science. San Francisco: Joffe-Bass.
1989. “Learning in Dynamic Simulation Games: Using Performance as a Measure.” this volume.
. W. 1989. “A Study of Human Control in Stock-Adjustment Tasks.” This volume. Also available as MIT Sys-
tem Dynamies £3 Group ‘Working Paper D-4030.
and S. E. Dreyfus. 1986. Mind over Machine. New York: MacMillan.
wr 1968 “Market Growth as Influenced by Capital Investment.” Industrial Management Review, (2): 83-105.
Mass: MIT ing and Computer
- 1988. “Generic Models as a Basis for Computer-Based Case Studies.” Proceedings ofthe 1988 Iterations System Dy-
namics Conference. La Jolla, California. Also available as MIT System Dynamics Group Working Paper D-3947.
-, and D. N. Scogin. 1985. “Generic Models as Educational Tools: Teaching about Managing Technology Conversions.”
Proceedings of the 1985 International System Dynamics Conference. Keystone, Colorado. Also available as MIT System
Dynamics Group Working Paper D-3736-1.
-, and P.M. Senge. 1988, “Computer-based Case Study and Leaming Laboratory Projects.” Cambridge, Mass.: MIT Sys-
tem Dynamics Group Working Paper D-4008
Ghemawat, P. 1986. “Intecom Teaching Note.” Case Teaching Note 5-386-109, distributed by HBS Case Services, Harvard
Business School, Boston MA 02163, U.S.A.
Gould, J. 19892. “Research Summary.” Cambridge, Mass.: MIT System Dynamics Group Working Paper D-4021.
- 1989b. “Transfer of System Dynamics Learning,” This volume.
Hax, A.C. and N.S. Majluf. 1984, Strategic Management: An Integrated Perspective. Englewood Cliffs, N.J.: Prentice-
Hall.
Kardes, F. 1987. “The Case Method of Instruction and Managerial Decision Making,” Cambridge, Mass.: MIT Sloan
‘School of Management Working Paper.
Keys, B. and J. Wolfe. 1988. “Management Education and Development: Current Issues and Emerging Trends.” Journal of
‘Management 14(2): 205-229,
326
Kim, D. H. 1989. “Leaming Laboratories: Designing a Reflective Learning Environment.” This volume. Also available as
MIT System Dynamics Group Working Paper D-4026.
Moissis, A. 1989. “Decision Making in the Insurance Industry: A Dynamic Simulation Model and Experimental Results,”
Cambridge, Mass.: MIT Sloan School of Management MS Thesis.
Morecroft, J. D. W. 1988. “System Dynamics and Microworlds for Policymakers.” European Journal of Operational Re-
search, 35(5): 301-320.
- 1984, “Strategy Support Models.” Strategic Management Journal, 5(3): 215-229.
- 1985. “Rationality in the Analysis of Behavioral Simulation Models,” Management Science. 31(7): 900-916,
- 1988b. “Executive Knowledge, Models and Strategic Change.” London, U.K.: London Business School Discussion Paper
GS-24-87, September 1988.
-,D.C. Lane and PS. Vita. 1989. “Understanding and Modelling Growth Strategy in a Biotechnology Startup Firm,” Lon-
don, U.K.: London Business School Discussion Paper GS-32-89,.. Abridged version, this volume as “Modelling a
Biotechnology Startup Firm.
Neuhauser, J. 1976. “Business Games Have Failed.” Academy of Management Review, 1(4): 124-129.
Nisbett, R., et al 1987. “Teaching Reasoning,” Science, 238(30): 625-631.
Paich, M. 1985. “Generic Structures.” System Dynamics Review, 1: 126-132.
Porter, M.E. 1980, Competitive Strategy. New York, NY: The Free Press,
- 1985. Competitive Advantage. New York, NY: The Free Press.
- 1988. Michael Porter on Competitive Strategy. Harvard Business School video series, distributed by HBS case services,
Boston MA 02163 and by Nathan/Tyler productions, 535 Boylston Street, Boston MA 02116.
Richmond, B.M., P. Vescuso and S. Peterson. 1987. STELLA for Business. Distributed by High Performance Systems, 13
Dartmouth College Highway, Lyme NH 03768.
Senge, P. M. 1986. “Systems Principles for Leadership.” in Transforming Leadership. Alexandria, Va.: Miles River Press.
- 1989a. “Catalyzing Systems Thinking in Organizations.” forthcoming in F. Masaryk (ed,), Advances in Organization De-
velopment. Also MIT System Dynamics Group Working Paper D-3877-9.
- 1989b. “Organizational Learning: A New Challenge for System Dynamics.” This volume.
Sterman, J. D. 1987a. “Testing Behavioral Simulation Models by Direct Experiment.” Management Science, 33 (12): 1572-
1592,
- 1987. “Misperceptions of Feedback in Dynamic Decisionmaking,” Cambridge, Mass.: MIT Sloan School of Manage-
ment Working Paper WP-1899-87. Also available as MIT System Dynamics Group Working Paper D-3876-1.
- 1987c. “Modeling Managerial Behavior: Misperceptions of Feedback in a Dynamic Decisionmaking Experiment.” Cam-
bridge, Mass.: MIT Sloan School of Management Working Paper WP-1933-87. Also available as MIT System Dynamics
Group Working Paper D-3919.
~ 1988, “People Express Management Flight Simulator” (software and documentation) available from J. Sterman, E52-562,
MIT, 77 Mass. Ave., Cambridge, Mass. 02139,
Toval, A. and M. Flores. 1987. “Computer Systems Simulation in Education: Description of an Experience.” Computers
and Education. 11(4): 293-303.
Wheatley, W, R. Hornaday, and T. Hunt. 1988. “Developing Strategic Management Goal-Setting Skills.” Simulation and
Games. 19(2): 173-185.
Whitestone, D. 1983. “People Express (A).” Cambridge, Mass: HBS Case Services, Case 483-103.
Wolfe, J. and G. Guth. 1975. “The Case Approach versus Gaming in the Teaching of Business Policy: An Experimental
Evaluation.” Journal of Business. 48(3): 349-364,
Wolfe, J. 1976. “The Effects and Effectiveness of Simulations in Business Policy Teaching Applications.” Academy of
Management Review. 1(2): 47-56.
- 1985. “The Teaching Effectiveness of Games in Collegiate Business Courses.” Simulation and Games. 16(3): 251-288.