A Design Science Approach - Joan Van Aken PDF
A Design Science Approach - Joan Van Aken PDF
A Design Science Approach - Joan Van Aken PDF
140
141
142
We first describe the design science perspective and then discuss a key
component in the design process: the role of reflection-in-action. The chapter then
describes the general process of changing the actual into the preferred, and
subsequently compares and contrasts designed-based changes in the material and
social worlds. We then explore the nature of generic evidence-for-design, and
143
contrast it with the nature of evidence in the social world, setting the stage for a
discussion of EBMgt from a design science perspective. The chapter subsequently
illustrates the EBMgt approach to design in a study of creating university
spinoffs. Finally, we give recommendations for both practitioners and
organizational researchers in advancing EBMgt as a design process.
The Design Science Perspective
Designing a future is fundamentally different from describing and explaining the
present. Simons (1969; 1996) The Sciences of the Artificial has played an
important role in recognizing the centrality of design in applied disciplines, by
demonstrating the fundamental differences between describing and explaining
what is and designing and evaluating what can be. Scholars often emphasize
describing and explaining the present as the core research mission of any
academic discipline. By contrast, designing what can be is a core activity in
architecture and the arts, and also in research-based engineering. Yet, an
increasing number of disciplines now resonate with Simons recognition that the
human world is artificial and that design activity is essential in making this world.
The primary mission of academic disciplines like Medicine and
Engineering is the creation of preferred futures -- for example, restoring health in
case of pneumonia or creating a more fuel-efficient car. This mission is realized
through knowledge produced by research as well as through the actions of the
professionals trained in these disciplines. As design plays a central role in the
creation of preferred futures, one may call disciplines for which design is central
design sciences, as opposed to explanatory sciences. The mission of an
144
explanatory science, like Physics and Sociology, is to describe and explain the
world as it is. Explanatory research is largely driven by a quest for knowledge as
an end in itself. The iconic research product here is the causal model. A causal
model is assessed in terms of descriptive validity, that is, how well it accounts for
the observed world.
Mainstream research in a design science is driven by field or real-world
problems. Field problems are situations in reality that, according to influential
stakeholders, can or should be improved. Knowledge development here is
instrumental in solving field problems; students are thus trained to become
professionals, using the knowledge of their discipline to explore, design and
create preferred futures. Whereas explanatory research typically uses an
independent observer perspective, Design Science Research draws on a
participant-observer perspective that requires knowledge on interventionoutcome-context combinations to be able to make judgments on the (likely)
outcomes of interventions in specific cases.
With disciplines like Medicine and Engineering as inspiration one can
define Design Science Research (DSR) as a family of approaches to research that
are driven by field problems, use a participant-observer instead of the independent
observer perspective, and pursue a solution orientation. This implies design
science researchers are not satisfied with describing field problems and analyzing
their causes, but also develop alternative general solution concepts for these field
problems. In Medicine and Engineering the participant-observer perspective does
not pose specific problems because most researchers in these disciplines are, or
145
146
particular type of contract. The most powerful design proposition is the fieldtested and grounded one: the intervention is tested in its intended field of
application and is grounded in an understanding of the generic mechanisms -- that
is the cause-effect relations -- that produce the outcome (van Aken, 2004).
Therefore, in the example of the proposition regarding the type of contract (given
above), one would like to have it tested through actual applications and grounded
in a theoretical explanation of why this type of contract is superior in this
particular context. The logic of the field-tested and grounded solution concept is
called the CIMO-logic (Denyer, D., Tranfield, D. & Van Aken, J.E., 2008), a
combination of problem-in-Context, Intervention, generative Mechanisms
producing the outcome, and Outcome.
A discipline is called a Design Science if DSR-approaches are firmly
positioned in its mainstream. This is the case in fields like Medicine, Architecture
and Engineering. But the importance of design has also been acknowledged in
other fields, for example, IT (March & Smith, 1995; Hevner, A.R., March, S.T. &
Park, J., 2004), accounting (Kasanen, E, Lukka, K. & Siitonen, A., 1993; Labro &
Tuomela, 2003) and education (Kelly, 2003; Collins, A., Joseph, D. & Bielaczyc,
K., 2004). Moreover, there is a growing interest in developing knowledge-fordesign among management researchers. For example, inspired by architect Frank
Gehrys approach to design, Boland & Collopy (2004a) brought together a group
of people from very different backgrounds to explore the potential of design for
management studies. The design science perspective to organization and
management studies was initially developed by, among others, Romme (2003)
147
and Van Aken (2004) and later also in special issues of the Journal of Applied
Behaviorial Science (Bate, 2007) and Organization Studies (Jelinek, Romme &
Boland, 2008).
Like EBMgt, Design Science Research can thus be regarded as a family of
approaches: driven by field problems, using a participant-observer perspective,
and pursuing a solution orientation.
Research-Based Designing and Reflection-in-Action
The core activity of designing consists of synthesis-evaluation iterations. A
possible solution to the design problem is created or synthesized, typically by
contextualizing an appropriate solution concept. The designed solution is then
evaluated on paper to determine how well it solves the problem. Evaluation is
the basis for choosing among alternative solutions to a design issue; evaluation
on paper means judging the extent to which a design meets its desired
specifications, before that design is actually implemented. Evaluation may also
take place by engaging in small experiments or pilot studies to try out one or more
alternative designs. If the result is not satisfactory, the solution is re-designed and
evaluated again, an iterative process that continues until a satisfactory design is
obtained.
Engineers, psychotherapists, managers and other design professionals
draw on their creativity, skills and repertoires of tacit and explicit knowledge.
They engage with a specific situation and treat their case as a unique one. At the
same time, in a creative process of reflection-in-action they draw on an extensive
repertoire of, among others, examples, models, theories and solution concepts in
148
order to make sense of their case and to design alternative solutions (Schn,
1983). In this respect, the designer draws on a repertoire as a kind of grab bag
of knowledge, sometimes consciously but more often largely unconsciously, to
synthesize and evaluate alternative designs.
The repertoires of junior professionals are largely the result of their initial
formal training. These initial repertoires consist of internalized formal theory and
tacit clinical experience, obtained through personal experiential learning.
Subsequently, professionals continually enrich their repertoires with further
personal experiential learning, usually combined with efforts to keep their formal
disciplinary knowledge up-to-date.
The use of explicit disciplinary knowledge demands creativity and
considerable expertise: general knowledge has to be translated towards the
specific context in question. DSR is not really intended for use by the layperson,
but rather by experienced and well-educated professionals. That is, design science
results are best used by professionals, having mastered the body-of-knowledge of
their discipline, having the ability to locate and obtain (new) knowledge that is
relevant in their work setting, the ability to contextualize explicit knowledge, and
the ability to develop intimate knowledge on the case under consideration and its
context. For example, physicians adapt their use of patient care protocols to
individual circumstances and needs, and professional experience and judgment
are key elements in designing context-specific solutions.
The degree to which explicit knowledge can support the design and
realization of preferred realities varies by profession. In Engineering, this role is
149
150
151
want, so interior decorators have to design interiors in close interaction with their
principals.
An intense dialogue may be needed with those that will produce and
implement the design, to produce designs that are easy or inexpensive to realize.
In a wide range of routine settings (e.g. designing and producing shoes), designfor-manufacturing can be readily accomplished using explicit knowledge about
manufacturing requirements. Otherwise, a focused dialogue between designers
and producers is necessary.
Changing the actual into the preferred involves creating an action net
(Lindberg & Czarniawska, 2006), a network of actors -- individuals, groups,
organizations -- working together to create something new. This action net can be
emergent in nature, occurring in an ad hoc, as-needed fashion. However, if an
action net does not emerge spontaneously, design professionals need to
deliberately develop it.
A design can be defined as a model to be realized. In engineering design,
evaluating alternative designs on paper is typically done with help of
mathematics (e.g. analytically or through simulation modeling). As discussed
below, using mathematical models for evaluation purposes is enabled by the fact
that key mechanisms in the material world can be described in terms of universal
and invariant laws. However, even in the material world it is not always possible
to develop models of future entities that can be analyzed mathematically. If the
entity being designed is too complex to be adequately modeled in mathematical
terms, case-based reasoning is useful (e.g. Leake, 1996; Watson, 1997). In case-
152
153
question (in the given context). In applied research in the material world, this
prediction is enabled by universal and invariant mechanisms governing the
behavior of matter -- even if these are contingent on this world (Mitchell, 2000).
An electron does not have the freedom today to behave differently from
yesterday, nor can it act differently in Amsterdam than in New York. A machine
tested in Barcelona will perform similarly in Buenos Aires -- assuming that the
human operator cannot significantly influence the machines performance.
Because of these universal and invariant mechanisms, responses identified in
past/current tests can reliably be used to predict responses in the future.
Human agency creates the social world and is subject to interactions with
the behavior of others. People do have the freedom to act on the basis of
anticipated effects of their actions. Thus, no universal and invariant mechanisms
exist in the social world. However, human behavior has patterns and regularities
as a result of human nature and nurture. These patterns and regularities can be
uncovered and then used in predicting -- within certain ranges -- the outcomes of
interventions in the social world. In fact, predicting human behavior is an almost
universal human competence (almost universal: autistic people lack this
competence, showing how important and rather generic it is). Starting from the
day we are born, we learn over the years what we can do to get what we want
from others: experiential learning (see Kolb, 1984).
The repertoires of professionals (both in the material and in the social
world) contain the explicit knowledge of their discipline and the results of their
personal experiential learning in real world settings. We argue that for applied
154
disciplines in the social world, given the absence of universal and invariant laws,
knowledge production should be largely based on the strategy of experiential
learning. However, knowledge production in this case should result from
objectified experiential learning, based among other things on controlled
observations (following protocols and using methods like triangulation),
comparative case analyses, and validation of designed interventions by alpha-and
beta-testing (van Aken, 2004) in the intended field of application. Objectified
experiential learning in academic research can, for instance, be done through the
comparative case study, resulting in a rich understanding of context, intervention
and outcome. But other approaches to objectified experiential learning can also be
of value.
According to Bhaskar (1998) it is the nature of the object that determines
the form of its possible science. In engineering design, the available science is
such that one often can evaluate alternative designs by means of calculations of
their expected performance. But in social system design, the nature of the
designed system implies that the evaluation of alternative designs proceeds
through experience-based judgment of expected performance. To support this
judgment, objectified experiential learning is a powerful knowledge production
strategy.
Problems arising in organizations are not limited to purely social
dimensions. They can also refer to socio-technical systems, as in operations
management. The smaller the social component in such a system, the stronger are
the universal and invariant mechanisms governing system behavior, and thus the
155
stronger the potential for adequate mathematical modeling of the system (of an
existing as well as a newly designed one). Designing socio-technical systems with
a small social component is more similar to engineering design than to social
system design, making math-based evaluation tools more useful. For instance, the
design and evaluation of a scheduling or inventory control system can be largely
based on mathematical modeling and analysis if the human operator can be
expected to quite closely follow the instructions from the information system.
However, if the operator must make frequent, significant adaptations to the
schedules calculated by the system (e.g. because the dynamic complexity of the
context leads the operator to make frequent exceptions to the built-in rules), any
evaluation effort has to draw on experiential learning as well as mathematical
modeling. Thus, if the social component of the system is large, one has to largely
rely on experience-based prediction of the performance of the designed system; if
the social component is small, one may fairly predict this performance on the
basis of calculations.
EBMgt in Action: Operating in the Swamp of Practice
Schn (1983) argued there is the high ground of theory and there is the lowly
swamp of practice. The ambition EBMgt is to provide some firm ground in the
swamp through the use of valid evidence. This will not drain the swamp: The use
of this evidence still needs a substantial level of competence, experience and
creativity.
In the present section we discuss EBMgt in the swamp of organizational
change, particularly in design-based organizational change. This discussion will
156
157
by major stakeholders and the decision to start a major effort to deal with it. If one
decides to adopt an organized process, then a project definition step is initiated.
The products of this step should include:
The project plan that outlines the approach to analysis, design and
realization, and provides a time line for these activities
158
The core process of designing the solution and the change plan
Realization, that is, the actual changes in roles, role structures and
activities
Typically, the project will not follow such an undisturbed process, because
of delays or accelerations in the process, resulting from the interactions with the
daily operations in the organization. Other urgent issues may demand attention, or
the issue itself suddenly may increase in urgency.
The Core Process: Designing
The basic process steps for solution design and change planning are:
159
Usually the actual design process is not a linear one. Rather, the design
process will use iterations and explorations. In an iteration one may return to a
previous step, because new insight reveals that more information is needed. There
may even be a need to change the project objectives at such an iteration. In an
exploration, one jumps to a step further in the process, for example to understand
what information is needed to design and evaluate alternative solutions.
In organizational solution design, many people may be involved. There
may be a lead designer or main sponsor (e.g. the CEO or his/her delegate), but in
all cases a group of people will work together on the issue. The composition of
this group may change over time. In the remainder of this section, we will call this
group of tightly or loosely coupled people the design group.
For major issues, this design group generally should be multidisciplinary.
The quality of the designed solutions depends on the extent to which this group
uses its members expertise effectively and efficiently. The quality of the
designed solution strongly depends on the performance of the Transactive
Memory System (TSM) of this group (Wegner, 1986; Peltokorpi, 2008). That is,
in terms of Schns repertoire notion: the quality of the design depends to a large
extent on the quality of the design groups shared repertoire.
160
161
Assign the main effort to the outline design (start the time-and
resource-consuming detailing only after the key design dilemmas
are solved)
The architect needed discussions with the client to become familiar with
the design dilemmas he faced. But he was not yet ready to resolve these dilemmas
in an outline design, being still in the phase of creatively synthesizing alternative
solutions. In organizational problem solving in the design mode, participants
therefore need to take time to play with alternatives, and put a lot of effort into
making a sound outline design. In managerial settings, too often the design group
becomes entrapped in the first feasible solution found, and then shifts all its effort
in detailing it.
As said previously, one cannot deduce the design from the inputs; there
always is a creative leap from input to design. But, of course, the inputs to the
design process are of critical importance to the designs quality. The explicit
inputs include:
In the synthesis steps, the design group merges these explicit inputs with
their tacit and explicit knowledge, to subsequently synthesize alternative solutions
to design issues in a process of reflection-in-action. In the same way, tacit and
explicit knowledge is used to evaluate the alternative designs against the
specifications, the most important being that the solution should solve the
162
problem. Like in very complex situations in engineering design that are rather
difficult to model, case-based reasoning is an important alternative approach to
evaluate organizational designs on paper.
Design-Based Change in the Social World
The overall process of changing the actual into the preferred in organizations has
similarities to engineering design. Much of the process is comparable: problem
analysis, development of specifications, interactions with the various
stakeholders, building of an action net to realize the solution. Fundamental
differences also exist.
Foremost among the differences is the need to contextualize generic
knowledge. How teams, meetings, or policies are thought of and operate in one
firm differs from those in another. Thick, rich and detailed descriptions are
therefore needed for the various solution concepts, the contexts in which these
concepts have been tested, and the nature of the (intended and unintended)
outcomes. Furthermore, the evaluation of alternative designs is done differently.
Many engineering designs can be adequately modeled, so engineers can use their
impressive array of engineering mathematics to evaluate their designs on paper.
This typically is not the case in the social world. An experiential learning
approach, like case-based reasoning, therefore has to be used to evaluate designs.
For instance, to evaluate the design for an account management system in a sales
department, the design group can draw on data on the performance of
implemented account management systems in other sales departments (within or
outside the same company).
163
164
In the course of problem and context analysis, the design group gathers a
lot of case-specific information, with respect to both the organization and its
environment. EBMgt promotes a rigorous approach to collecting information,
calling attention to scientific evidence as well as local facts and experiences, and
adopting a critical attitude toward the information collected. EBMgt thus
165
166
167
168
spin-offs. Thus the research team faced the challenge of transforming an ongoing,
largely emergent design process into a more deliberate one. A further challenge
was, as always in evidence-based solution design, the need to contextualize any
generic findings of a systematic literature review to take into account the features
specific to TU/e.
In view of these challenges, a design science-driven EBMgt approach was
adopted in which two key notions connected professional practices and research
findings (cf. Romme & Endenburg, 2006): design propositions and solution
concepts (as discussed in the section on the design science perspective). Design
propositions can be based on research, but can also be derived from successful
professional practice. The research-based propositions serve as tangible artifacts
that allow different groups of people to focus upon and create shared
understandings (Romme & Endenburg, 2006). That is, these explicit propositions
allowed participants from diverse backgrounds to focus on a common set of issues
in the design process (cf. boundary objects).
The research team developed design propositions for university spin-off
creation by separately developing propositions based on practice (practice-based
propositions) and propositions based on scholarly knowledge (research-based
propositions). The synthesis of these propositions subsequently resulted in design
propositions, which thus provides a body of knowledge that is grounded in
research as well as tested in practice (cf. Van Aken, 2004). Thus, the key steps in
developing design propositions were as follows (Van Burg et al., 2008):
169
170
171
172
The set of design propositions exposed blind spots and major opportunities
for improvement and development. Moreover, they helped to create a shared
repertoire of knowledge on university spin-offs for the various parties concerned.
Thus, the EBMgt project by Van Burg and co-authors served to define the lack of
an entrepreneurial university culture (cf. design proposition 5) at the TU/e as the
main weakness of its spin-off creation capability. Throughout the university, the
awareness of this deficiency grew and several new initiatives and projects were
used to expose the TU/e community to role models (e.g. successful entrepreneurs
among the TU/e alumni) and to motivate entrepreneurial behavior (e.g. by
creating attractive financial benefits for scholars whose patented technologies are
used in successful spin-offs). The universitys senior management and other key
stakeholders were well aware of the long-term effort required here, because
universities tend to be rather conservative in view of their long-standing
commitment to academic research and education.
This evidence-based project on spin-off creation, set up in 2005, is still
ongoing at this writing. The preliminary results of the project are as follows. First,
the TU/e increased its spin-off rate from zero in the late 1990s and about 5 per
year around 2005, to the current rate (in 2010-11) of about 15 new firms
exploiting IP of the university per year. The design framework developed by Van
Burg et al. (2008), in combination with this boost in the spin-off rate, motivated
the universitys top management to increase its ambitions -- to double the spin-off
rate in the next 5 years. Of course, it is not possible to determine precisely to what
extent the EBMgt project of Van Burg et al. (2008) contributed to the increase in
173
the spin-off rate at the TU/e. It is not unlikely that without this EBMgt project the
TU/e would also have increased its performance in creating spin-offs. Moreover,
any EBMgt project is embedded in a continuously evolving and ambiguous socioeconomic system. The spin-off creation project at the TU/e, however, does
illustrate that EBMgt serves to build a systematic, theory-driven understanding of
a rather complex managerial issue, which previously was dealt on the basis of
common wisdom and personal experiences.
Another, more academic, upshot of this project is that the design
propositions developed by Van Burg et al. (2008) served to reflect on the
comprehensiveness of (previous) research and theory development. In this
respect, Van Burg et al. (2008) observed that some of their design propositions
are not yet (firmly) incorporated in the university spin-off literature. For example,
the proposition referring to clear and supportive rules and procedures was not
previously identified, and as such not grounded in any theoretical frameworks. In
turn, this finding motivated a new study exploring the role of transparency and
fairness in university spin-off formation. Moreover, other research teams have
adopted and replicated the design framework developed by Van Burg et al.
(2008), for example Barr, Baker, Markham and Kingon (2009) in the context of
several US-based universities.
Discussion
The term Evidence-Based Management (EBMgt) may evoke a picture of rational
decision-making (like the Rational Manager of Keppner & Tregoe, 1965) with
academic research findings as its main input, replacing intuitive, experience-based
174
175
job done. On the other hand, experienced medical doctors sometimes need to be
seduced to acquire and use state-of-the-art explicit knowledge; they may feel that
their initial training in the medical discipline, complemented with their rich tacit
knowledge gained through experiential learning, is quite sufficient to operate as a
medical practitioner. The challenge here is to demonstrate the added value of
recently developed evidence to medical professionals.
A similar problem occurs for experienced managers and management
consultants: the added value of research-based knowledge is not always selfevident to them (and experienced managers may also have different views on
what counts as evidence; see Green & Potwoworski, this Handbook). The
community of management and organization researchers and educators must take
charge of this challenge, by building a strong case for EBMgt, by creating
attractive conditions for practitioners to join collaborative EBMgt projects and
especially by giving much attention to EBMgt in management training.
The term evidence suggests a decision mode; that is, the idea that the
main challenge is to choose between known alternatives. This implies that the
main contribution of evidence-based practice is to produce the evidence for
making a rational choice between these alternatives. As previously argued,
EBMgt will typically address major managerial issues. For such issues, we
advocate a design mode that puts substantial effort in the design of alternative
solutions or arrangements of solutions, thus crafting a high-quality outline design
before going into details. Academic research can provide significant support for
this type of design effort by providing a range of well-tested solution concepts.
176
177
178
179
References
Archer, M.S. (1995). Realist social theory: The morphogenetic approach.
Cambridge: Cambridge University Press.
Barr, S.H., Baker, T., Markham, S.K., & Kingon, A.I. (2009). Bridging the Valley
of Death: Lessons learned from 14 years of commercialization of
technology education. Academy of Management Learning and Education,
8, 370-388.
Bate, P. (2007). Editorial to the special issue Bringing the design sciences to
organization development and change management. Journal of Applied
Behavioral Science, 43, 8-11.
Bhaskar, R. (1998). The possibility of naturalism: A philosophical critique of the
contemporary human sciences (3rd ed.). New York, NY: Routledge.
Boland, R. & Collopy, F. (2004a) (Eds.). Managing as designing. Stanford, CA:
Stanford University Press.
Boland, R. & Collopy, F. (2004b). Design matters for management. In Boland,
R. & Collopy, F. (2004a) (Eds.), Managing as designing. Stanford, CA:
Stanford University Press.
Briner, R.B., Denyer, D. & Rousseau, D.M. (2009). Evidence-based management:
concept cleanup time? Academy of Management Perspectives, 23, 19-32.
Collins, A., Joseph, D. & Bielaczyc, K. (2004). Design research: Theoretical and
methodological Issues. Journal of the Learning Sciences, 13, 15-42.
Denyer, D. & Tranfield, D. (2006). Using qualitative research synthesis to build
an actionable knowledge base. Management Decision, 42, 213- 227.
180
Denyer, D., Tranfield, D. & Van Aken, J.E. (2008). Developing design
propositions through research synthesis. Organization Studies, 29, 249269.
Denzin, N.K. (2009). The elephant in the living room; or extending the
conversation about the politics of evidence. Qualitative Research, 9, 139160.
Dewey, J. (1929). The quest for certainty: A study of the relation of knowledge
and action. New York, NY: Minton, Balch and Company.
Goodman, J. (2010). Healthcare: Who knows best? The New York Times Review
of Books, 57(2), February 11.
Hevner, A.R., March, S.T. & Park, J. (2004). Design science in information
system research. MIS Quarterly, 28, 75-105.
James, W. (1907). Pragmatism, a new name for some old ways of thinking. New
York, NY: Longmans, Green and Company.
Jelinek, M., Romme, A.G.L & Boland, R.J. (2008). Introduction to the special
issue Organization Studies as a Science for Design: Creating
collaborative artifacts and research. Organization Studies, 29, 317-329.
Kasanen, E, Lukka, K. & Siitonen, A. (1993). The constructive approach in
management accounting research. Journal of Management Accounting
Research, 5, 243-263.
Kelly, A.E. (2003). Research as design. Educational Researcher, 32, 3-4.
Keppner, C.H. & Tregoe, B.J. (1965). The rational manager. New York, NY:
McGraw Hill.
181
182
183
Theories of Group
184