Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

European Journal of Information Systems (2015) 24, 107–115

© 2015 Operational Research Society Ltd. All rights reserved 0960-085X/15


www.palgrave-journals.com/ejis/

ISSUES AND OPINION

Distinguishing and contrasting two strategies


for design science research

Juhani Iivari Abstract


This paper distinguishes and contrasts two design science research strategies in
Department of Information Processing Science, information systems. In the first strategy, a researcher constructs or builds an IT
University of Oulu, Oulu Yliopisto, Finland meta-artefact as a general solution concept to address a class of problem. In the
second strategy, a researcher attempts to solve a client’s specific problem by build-
Correspondence: Juhani Iivari, Department of ing a concrete IT artefact in that specific context and distils from that experience
Information Processing Science, University of
prescriptive knowledge to be packaged into a general solution concept to address a
Oulu, PO Box 3000, Oulu yliopisto, 90014,
class of problem. The two strategies are contrasted along 16 dimensions represent-
Finland.
Fax: +358 8 553 1890; ing the context, outcomes, process and resource requirements.
E-mail: juhani.iivari@oulu.fi European Journal of Information Systems (2015) 24(1), 107–115. doi:10.1057/ejis.2013.35;
published online 7 January 2014

Keywords: design science research; design research; design science research strategies

Introduction
There has been increasing interest in design science research (DSR) in the
information systems (IS) discipline (Nunamaker et al, 1990–1991; Walls et al,
1992; March & Smith, 1995; Hevner et al, 2004; Iivari 2007; Peffers et al,
2007–2008; Vaishnavi & Kuechler, 2008; Hevner & Chatterjee, 2010).
DSR attempts to provide new innovative artefacts that address hereto-
fore unsolved problems or address them more effectively/efficiently than
previous attempts (Hevner et al, 2004).
The scientific discourse on DSR has suffered from some confusion
(Baskerville, 2008; Kuechler & Vaishnavi, 2008; Winter, 2008). The purpose
of this paper is to clarify the discourse by distinguishing two DSR strategies.
In the first strategy, a researcher constructs an IT meta-artefact (Iivari, 2003)
as a general solution concept (van Aken, 2004) possibly to be instantiated
(March & Smith, 1995) into a specific solution concept (van Aken, 2004) or a
concrete IT artefact (application) to be adopted and used in a specific
context. An alternative strategy is that a researcher attempts to solve a
client’s specific problem by building a concrete IT artefact (application) in
that specific context and distils from it knowledge to be generalized into a
general solution concept. After introducing these strategies they will be
contrasted along 16 dimensions.
To the author’s knowledge the two DSR strategies have not been separated in
the prior IS literature, and still less contrasted with each other. The present
essay proposes that DSR covers both. At the same time it is important to be
aware that they differ considerably from each other in several respects.

Two strategies of DSR

Received: 25 June 2013 Introduction


Revised: 11 October 2013 The scientific discourse on DSR is still in a state of conceptual confusion.
Accepted: 25 November 2013 Kuechler & Vaishnavi (2008) note that we still lack ‘a common
108 Distinguishing and contrasting two strategies for DSR Juhani Iivari

understanding of the definition and scope of design Similarly, Walls et al (1992) frame their design theories in
research (DR) in IS’ and question whether it is ‘research terms of meta-requirements and meta-design, pointing out
with design as a topic of investigation or research with that meta-requirements do not address a single problem
design as a method of investigation’ (p. 8). They also but a class of problems, and meta-design describes a class of
recommend that ‘design research’ should be allowed to systems rather than a single system. They also see the
have the breadth it has achieved in other design fields, implementation of a system (prototype) necessary only in
rather than constraining it to the constructive view, as the testing of the design theory.
advocated by Hevner et al (2004). Markus et al (2002) and Sein et al (2011), on the
The confusion is exaggerated by the terminology. Labels contrary, have applied a DSR strategy in which one
such as ‘design science research’ and ‘design research’ first builds an IT artefact in a specific client context and
are used more or less interchangeably in the IS litera- based on that experience attempts to distil lessons into
ture without any respect to the terminology used in the a generalized DSR contribution. Yet, they do not explicitly
50-year history of design studies (Cross 1993, 2001; introduce their views as an alternative DSR strategy.
Bayazit, 2004). Iivari (2012) contrasts the two, interpreting The rest of this section contrasts the two DSR strategies
that DSR is research with design as a method of investi- using a simple context-process-outcome framework from
gation and DR is broader, with design as a method of Denison et al (1996), which is supplemented to include
investigation and a topic of investigation. resources as a fourth aspect. Context describes differences
The present paper is strictly limited to DSR in the above in the research settings of the DSR projects, covering
meaning, which can be summarized in three interrelated the client relationship, problems to be addressed and the
points: associated major uncertainties. The outcomes aspect
focuses on the major artefacts developed in the case of
1. DSR produces new innovative meta-artefacts as its
the strategies and the nature of those artefacts. The process
constitutive and distinctive research outputs.
aspect addresses the two strategies in terms of drivers of
2. Constructive research on building new innovative
DSR process, research methods used and the process of
meta-artefacts is the core research activity of DSR.
generalizing the results. Finally, resources refer to some
3. New innovative meta-artefacts follow the epistemo-
practical differences between the strategies.
logy of utility rather than the epistemology of truth
At a more detailed level, the contrasting takes place
(likeness).
along 16 dimensions, which are significant to understand-
These points imply that the view of DSR adopted in the ing the essential differences between the strategies and/or
present paper is in line with the constructive view advo- are practically relevant to researchers attempting to make
cated by, for example, Hevner et al (2004). a choice between them. As a consequence, many of the
The early views of DSR seem to assume that one con- dimensions are specific to the two strategies, reflecting
structs the DSR contribution first and possibly instantiates it. their essential differences.
In the case of March & Smith (1995) and Hevner et al (2004), The purpose of the following analysis is to provide a
the very word ‘instantiation’ simply implies this – there is detailed point-by-point comparison of the two strategies.
something (constructs, models, methods) to be instantiated. Since there are not many published examples of DSR
Nunamaker et al (1990–1991) also assume this DSR strategy following Strategy 2, some points in the analysis are tenta-
when they advocate systems development as a research tive generalizations. The two examples of DSR Strategy 2
method for IS research. In their view the developed system (i.e., Markus et al, 2002 and Sein et al, 2011) are analysed in
serves as a proof of concept and provides an artefact detail in Iivari (2012) along the 16 dimensions included in
that becomes the focus of expanded, continuing research. Tables 1–4.

Table 1 Contrasting the two strategies of DSR – context


Dimension Strategy 1 Strategy 2

1. Researcher–client relationship A client may be involved, but not necessarily Client involvement is inevitable

2. Major problems to be addressed 1. A general problem (a class of problem), more or 1. A specific problem encountered by a client (or a
less informed by specific problems in practice set of clients)
2. The general problem (the DSR problem) to be
figured out during the DSR project

3. Typical uncertainty of a DSR 1. Uncertainty about the new, innovative general 1. Uncertainty about the specific solution to the
project solution concept to the class of problem specific problem encountered by the client (or a
2. Uncertainty about the total complexity of specific set of clients)
problems and their solutions in practice 2. Uncertainty about the possible DSR contribution

European Journal of Information Systems


Distinguishing and contrasting two strategies for DSR Juhani Iivari 109

Context Researchers following Strategy 2, on the contrary, start


As Table 1 notes, DSR following Strategy 1 does not with a specific problem encountered by a client. As
necessarily have any identifiable client(s). A researcher just a consequence they face the total complexity of the
has some understanding of the potential would-be clients specific problem right from the beginning. There is, of
and their problems, or a more general demand for a new course, uncertainty about the specific solution to this
solution. In the case of Strategy 2, a relationship with an specific problem, that is, how to address the client’s
identifiable client (or set of clients) is inevitable, since the specific problem effectively and efficiently.
whole DSR project commences with a practical problem As noted above, a researcher following Strategy 2 faces
encountered by a client and an attempt to find a specific considerable uncertainty about the potential DSR contri-
solution. bution: what is the general problem to be addressed and
DSR following Strategy 1 starts with a general problem what might be a new innovative solution to it? This
(a class of problem) to be addressed (e.g., Walls et al, 1992). generalization problem will be discussed later in the con-
On the contrary, DSR following Strategy 2 starts with text of Dimension 12. At this stage, a researcher can only
a specific problem encountered in practice. The challenge hope that the process of addressing the client’s specific
in the case of Strategy 2 is to identify the general problem problem will help in that search.
(the DSR problem). Therefore, the DSR problem in both
cases is a general problem (a class of problem).
Applying the distinction between an ‘empty head’ and Outcomes
an ‘open mind’ (Dey, 1993), Strategy 2 does not imply that Referring to Table 2, artefacts built in a DSR project
a researcher enters a DSR project ‘empty headed’ without following Strategy 1 comprise two types of artefacts:
an initial idea of what the DSR contribution might or will (1) conceptual IT meta-artefacts (e.g., constructs, models
be, although (s)he should maintain an open mind and be and methods), and (2) their optional instantiation as ‘real
sensitive to what the client’s real problem is, how to system implementations’. The adjective ‘conceptual’
address it and what is found when addressing it. Otherwise emphasizes that meta-artefacts are conceptual rather
(s)he can easily follow the aphorism, ‘If all you have is a than physical like the hardware components used in
hammer, everything looks like a nail’. technical IS implementation. The instantiation may serve
The two examples of DSR Strategy 2 (Markus et al, 2002; as a proof of concept (Nunamaker et al, 1990–1991) and
Sein et al, 2011) illustrate the need for an open mind, and may also be useful in the field evaluation of the conceptual
the usefulness of letting the experiences of addressing the artefact. The implementation implied by the instantiation
client’s specific problem guide the process. In both cases may also inspire learning. Yet, instantiations are only
the DSR contribution evolved considerably during the secondary DSR outcomes. The essential point is what the
process. This need is at least as important as in any design knowledge instantiations entail, that is, in the
empirically grounded qualitative research that aims at underlying conceptual meta-artefact.
theory building (Díaz Andrade, 2009), and perhaps more DSR Strategy 2 may comprise building three types
important since the hammer is not only used to make of artefacts: (1) real system implementations to address
sense of the world, but to change it. the specific problems of clients, (2) conceptual IT meta-
These two strategies comprise different uncertain- artefacts as DSR contributions and (3) their instantiation.
ties related DSR projects. Even though DSR following The design and implementation of the real system as
Strategy 1 often faces the issue of what is the problem, a specific solution to the client’s specific problem mainly
the key uncertainty concerns the general solution to serves as a source of inspiration for the DSR contribution,
the class of problem selected for the DSR project but does not necessarily instantiate the latter. Therefore, a
in question. This general solution concept should be separate instantiation may be needed as a proof of concept
a new innovative meta-artefact in order to qualify as a or for evaluation reasons. Furthermore, taking an analogy
DSR contribution. from theory building – theory testing, where it is not a
Although the general problem selected by a researcher good practice to use the same dataset for theory building
following Strategy 1 may be informed by specific problems and for theory testing, one can question if under Strategy 2
in practice, often (s)he faces uncertainty about the real it is justified to test a DSR contribution in the same client
nature of the specific problems in practice, and about the context where it was developed (constructed).
specific solutions to them. (S)he does not necessarily If we accept that DSR provides IT meta-artefacts that
understand the total complexity of specific problems – support building concrete IT artefacts (applications) in
complexity due to the specific problem encountered in a practice, it is obviously important to understand the
specific context (e.g., organization), possibly with unique nature of these concrete IT artefacts. Table 2 distinguishes
characteristics – implied by the class, especially if (s)he a priori designable artefacts and emergent artefacts (Iivari,
does not have a direct relationship with any specific client. 2003), suggesting that Strategy 1 assumes the former and
As a consequence, many times researchers following Strat- Strategy 2 the latter.
egy 1 are not so much concerned with these specific Both examples of DSR Strategy 2 emphasize emergence
problems, but leave it to practitioners to interpret, adopt in their studies – emergent knowledge processes (EKP) in
and adapt the constructed meta-artefact. the case of Markus et al (2002) and the emergent nature of

European Journal of Information Systems


110 Distinguishing and contrasting two strategies for DSR Juhani Iivari

Table 2 Contrasting the two strategies of DSR – outcomes


Dimension Strategy 1 Strategy 2

4. Artefacts built 1. Conceptual IT meta-artefact as a DSR 1. A real system implementation as a specific solution
contribution to a problem encountered in practice
2. Possibly a real system implementation 2. Conceptual IT meta-artefact as a DSR contribution
(instantiation) of the conceptual IT 3. Possibly a real system implementation
meta-artefact (instantiation) of the conceptual IT meta-artefact

5. Primary role of the real system Instantiation as a proof of concept and The real system as a specific solution to a problem
implementation possibly used in the evaluation encountered in practice implemented primarily as a
source of inspiration
Instantiation as a proof of concept and possibly used in
the evaluation

6. Nature of the target IT artefacts A priori designable system Emergent system

7. Typical nature of the IT meta-artefact A new, innovative concept for a software- New, innovative design principles
hardware system or a new innovative
systems development approach, method,
technique

8. Innovativeness Innovativeness of the IT meta-artefacts Mixed tendencies


as the DSR contribution varies greatly + if an interdisciplinary research team, it may foster
creativity
+ practical problems may challenge existing solutions,
knowledge and wisdoms
– easily focused on client’s current problems
– clients may be reluctant to experiment with cutting-
edge technology

9. Practical relevance Varies greatly A priori better equipped to address immediate practical
problems

IT artefacts in Sein et al (2011). According to Markus et al Of course, the two examples of DSR Strategy 2 represent
(2002), ‘emergent processes are characterized by highly a very limited sample to draw far-reaching conclusions.
unpredictable user types and work contexts. The unpre- Nevertheless, there are three possible explanations for the
dictability extends to when and why the process is per- difference in the complexity of the meta-artefacts con-
formed and whether support tools will be used’ (p. 183). structed. First, researchers following Strategy 2 put consider-
Sein et al (2011) propose that their Action Design Resea- able effort into solving the client’s specific problem (see
rch (ADR) method is particularly suitable in the case Dimension 2 in Table 1), usually by building a real system
of ensemble artefacts (Orlikowski & Iacono, 2001) that implementation as a specific solution to that problem (see
emerge from design, use and on-going refinement in Dimension 4 in Table 2). Therefore, inventing and con-
context. Although Orlikowski & Iacono (2001) introduce structing the IT meta-artefact as the DSR contribution is
the ensemble view as one view of IT artefacts rather than only part of their research effort, even though it is consti-
as a class of IT artefacts, the positioning of their ADR tutive. Second, the DSR contribution inspired by the above
method in Sein et al (2011) seems intuitively reasonable. If real system implementation may be figured out and con-
the idea of DSR is to provide IT meta-artefacts for emergent structed as an afterthought. This may affect the outcome.
IT artefacts, it seems difficult to understand the latter Third, the IT meta-artefact produced under Strategy 2, even
without studying them as an emergent phenomenon. if light, is ‘grounded’ in practice and therefore its practical
Referring to Dimension 7 in Table 2, typically new inno- relevance may a priori be considered higher or more likely
vative IT meta-artefacts produced by DSR following (see Dimension 9 in Table 2). It may be that our research
Strategy 1 are concepts for software-hardware systems or community is ready to accept these lighter DSR contribu-
systems development approaches, methods and techni- tions of Strategy 2 research more easily than in the case of
ques. They tend to be relatively complex when compared Strategy 1, where they are not equally ‘grounded’ and their
with the IT meta-artefacts suggested by Markus et al (2002) practical relevance is not equally obvious.
and Sein et al (2011), which are fairly light design princi- The innovativeness of the IT meta-artefact as a DSR
ples for emergent IT artefacts (applications). contribution may vary greatly in the case of Strategy 1. More

European Journal of Information Systems


Distinguishing and contrasting two strategies for DSR Juhani Iivari 111

interestingly, Table 2 also suggests that Strategy 2 includes limit until it collapses. However, the client’s interest is
aspects that both support innovativeness and limit it. As usually that the technology employed is proven and robust
will be pointed out below, a research team using Strategy 2 and does not disturb the performance of their operation
may be multidisciplinary or interdisciplinary. The different and daily work. Having a technology that fails would be
worldviews and perspectives implied by such teams support unlikely to be appropriate or agreeable to the client.
creativity (Hage, 1980). The practical problems encountered Weedman (1998) describes such a conflict of interest in
may also challenge existing solutions, knowledge and wis- the Sequoia 2000 project. It was essentially a combined
dom (Markus et al, 2002; Sein et al, 2011). DSR and Action Research (AR) project, although she does
On the other hand, DSR following Strategy 2 easily not refer to AR in her paper. The project was comprised of
focuses on a client’s current problems rather than their computer scientists who acted as researchers and earth
future problems, the relevance of which may be difficult to scientists who acted as users. Weedman describes the con-
figure out. For example, major innovations in computing, flict of interest when the technology was repeatedly pushed
such as the relational model (Codd, 1970) and the object- to the limit until it broke, with consequent costs to the users
oriented features of the Simula 67 language (Dahl et al, when the system crashed and the users’ work came to a halt.
1968), would have been quite difficult to develop follow- Related to Dimension 9, Table 2 assumes that the
ing DSR Strategy 2, since there were very few, if any, clients practical relevance of DSR following Strategy 1 varies
who understood the practical significance of those inno- greatly. The relational data model (Codd, 1970) is a clear
vative ideas when they were developed. example of extremely high relevance, although it was
Existing technology may also limit the researcher–client proposed about 10 years before we had the technology
interaction. For instance, in the early 1970s, computers did to instantiate the concept in practice. A priori DSR adop-
not have the capacity to demonstrate the relational model ting Strategy 2 is better equipped to address immediate
in practical settings. Furthermore, since DSR by definition problems in practice, but as argued above, it has limi-
attempts to construct new and innovative meta-artefacts, tations in promoting more radical and disruptive inno-
it often operates at the edge of existing technology. vativeness (see Dimension 8 in Table 2).
Such cutting-edge technology is not necessarily robust.
Petroski (1982) claims that failure is central to understand- Process
ing engineering, and that lessons learned from failures can The two DSR strategies differ with regard to the major
advance engineering knowledge more than successful process drivers (see Table 3). As argued above, it is not
cases. Similarly, DSR may push a new technology to its necessary or sometimes even possible to test and evaluate

Table 3 Contrasting the two strategies of DSR – process


Dimension Strategy 1 Strategy 2

10. Major process driver The constructed meta-artefact as a general solution Experiences from the process of addressing the specific
concept, if to be tested and evaluated on the field solution to a problem encountered in practice

11. Research methods Constructive (in building the meta-artefact) Action Research or Action Design Research (in the
Empirical (in the evaluation) intervention)
– laboratory experiment Constructive (when constructing the general solution
-–field experiment concept or IT meta-artefact)
– field study Other empirical (if separate evaluation)
– case study – field experiment
– action research -–field study
-–case study
– action research

12. Generalization Included in the problem statement (1) Identify various problems encountered while
constructing a real system implementation as a specific
solution to a problem encountered by a client
(2) Generalize these specific problems into a general class
of problems
(3) Identify lessons from (a) the real system implementation
as a specific solution to the client’s problem and/or (b)
the process of developing that specific solution
(4) Generalize these lessons into an IT meta-artefact as a
general solution concept (e.g., design principles)
(5) Associate the general solution concept with the class of
problems identified

European Journal of Information Systems


112 Distinguishing and contrasting two strategies for DSR Juhani Iivari

the constructed IT meta-artefact on the field when follow- terms of three processes: (1) generalization of the specific
ing DSR Strategy 1. However, if it is done, the process is problem instance into a class of problems, (2) gene-
typically driven by the meta-artefact in question. This ralization of the specific solution instance into a general
process of testing and evaluation may be iterative, consist- class of solutions and (3) derivation of design principles
ing of several cycles, with the meta-artefact being revised from the DSR outcomes. This process seems to assume
between cycles. Yet, the question is about Strategy 1, in that the authors have generalizations in mind regarding
which the evolving meta-artefact drives the process. the specific problem of the client and the solution deve-
DSR following Strategy 2, on the contrary, is driven by loped for it. Table 3 broadens this view by pointing out
the experiences of the process addressing the specific that a researcher may identify various problems and distil
problem in practice. However, once the DSR contribu- various solution ideas when developing a specific solution
tion is figured out, it starts more and more to drive the to the client’s problem. These need not have a direct
process. connection with the client’s problem and the specific
This difference between process drivers has some affinity solution to it, but may be any problems and solution
with the forcing and emergence debate in Grounded ideas (e.g., design principles) inspired by the developed
Theory (Glaser, 1992; Boychuk Duchscher & Morgan, system and its development process. Once generalized,
2004). If Strategy 1 includes field testing and evaluation, the general problems look for general solution concepts
it represents ‘forcing’, where a researcher has the IT meta- and the generalized solution concepts look for problems
artefact to be tested and evaluated. Actually, this forcing (cf. March & Olsen, 1976). They should be connected in
is stronger in the DSR context than in Grounded Theory, some way.
since the meta-artefact to be tested and evaluated is To illustrate the difference, Sein et al (2011) suggest three
used in an intervention to change the world, whereas in design principles as the DSR contributions of their Volvo
Grounded Theory forcing means forcing a preconceived case. These design principles clearly correspond to the
conceptual model on data. Strategy 2, on the other hand, design principles of the specific Volvo Information Portal
represents emergence rather than forcing, even though it (VIP) prototype they ended up creating in the process.
is in a weaker form. Since the purpose of DSR is not to Therefore, VIP can be considered as an instance imple-
build a theory that is faithful to reality and the data about mentation of the proposed design principles. On the
it, it is not necessary that the DSR contribution is closely other hand, Markus et al (2002) report a slightly different
connected to the experience of addressing the client’s story. Their work was based on experiences designing and
specific problem. The latter can just serve as an inspiration deploying a specific Technology, Organization, and People
for the DSR contribution. (TOP) system in practice. These experiences led them to
Referring to research methods, the distinction between formulate characteristics of EKP that can be considered a
the two strategies helps to understand their role in DSR. ‘theory for analysing’ (Gregor, 2006), meta-requirements
When following Strategy 1, constructing the IT meta- for IT support of EKPs, and meta-design for EKP support
artefact as a general solution concept is a central activity. systems and development principles. The development
Therefore, constructive research methods, such as those of principles clearly include aspects that are solutions to
Nunamaker et al (1990–1991), dominate DSR following problems concerning developing EKP support systems
Strategy 1. Empirical methods are mainly used in the and not generalizations of the features of the TOP system.
evaluation of the constructed IT meta-artefacts. Actually, Markus et al (2002) illustrate that a DSR project
DSR adopting Strategy 2 relies more on AR (Markus et al, as an empirical phenomenon may inspire the develop-
2002) or ADR (Sein et al, 2011) in the intervention. This AR ment of ‘theories for analysing’, ‘theories for predicting’
or ADR intervention comprises constructing the real sys- and ‘theories for explaining and predicting’. Yet, according
tem implementation as a specific solution to the client’s to the interpretation of DSR in the second section, these
problem encountered in practice. Constructive research in theories are not constitutive in the sense that they alone,
the context of Strategy 2 is confined to the process of without an innovative meta-artefact, would make a project
building the IT meta-artefact as a general solution concept. a DSR one.
If the constructed meta-artefacts are fairly light, construc-
tive research may look like a minor activity, especially
resource-wise. However, it is the core activity in the case Resource requirements
of Strategy 2. The innovativeness of the DSR in question Finally, Table 4 contrasts the resource requirements of the
lies in this activity of identifying and constructing new, two strategies. There is a great deal of variety in the case of
innovative IT meta-artefacts, however simple and light Strategy 1. A critical question is if one considers it neces-
they may be. sary to implement (instantiate) a system (see Dimension 5
As far as generalization is concerned, in the case of in Table 2), since it essentially affects the expertise needed
Strategy 1 the IT meta-artefacts as general solution con- in the project, the size of the research team, and the time
cepts are generally targeted to address a class of problem needed for and cost of the research effort.
right from the beginning (see Dimension 2 in Table 1). DSR following Strategy 2 means intensive collabora-
Strategy 2, on the other hand, is a more complicated tion with the client. Therefore, its resource requirements
situation. Sein et al (2011) discuss the generalization in can be expected to be very similar to intensive research

European Journal of Information Systems


Distinguishing and contrasting two strategies for DSR Juhani Iivari 113

Table 4 Contrasting the two strategies of DSR – resource requirements


Dimension Strategy 1 Strategy 2

13. Access to a client Not necessary Necessary, but may be challenging

14. Expertise needed Often disciplinary. A possible system Often multidisciplinary or interdisciplinary
implementation (instantiation) may require a
number of sub-specialists

15. Research team Size may vary from a single individual to fairly large The core research team is usually 3–10 members.
teams. A possible system implementation The real system implementation to address the
(instantiation) usually requires a larger research client’s specific problem usually requires additional
team, but all members do not necessarily have a members, but all these members do not necessarily
research interest in the project have a research interest in the project

16. Time and cost Varies greatly depending on the ambition and Usually requires intensive involvement in the project
complexity of the IT meta-artefact and more under a over a longer period of time. Overall, time-
researcher’s control. A possible system consuming and expensive
implementation (instantiation) is often time
consuming and expensive

methods such as AR. First, access to a client interested case of enumerative induction is if it can reasonably be
in a collaborative research effort and committed to it may based on one particular instance. One can expect that this
be challenging, even though the contribution to the client is a typical situation in the case of DSR Strategy 2.
may be expressed in terms of a fairly concrete system Putting this conceptual problem aside, it is my con-
implementation. An attempt to address the multi-faceted tention that one cannot conclude that Strategy 1 relies on
complexity of the client’s total problem often justifies a deductive reasoning only and Strategy 2 on inductive
research team that is multidisciplinary or interdisciplinary. reasoning only. It is obvious that sources of ideas –
On the basis of published examples (Majchrzak & Gasser, practical problems and opportunities, existing meta-arte-
2000; Markus et al, 2002; Lindgren et al, 2004; Sein et al, facts, analogies and metaphors, and existing theories
2011), the core research team itself, when following Stra- (Iivari, 2007) – essentially affect the forms of reasoning
tegy 2, is not necessarily very large, maybe 3–10 members. in the DSR process. Both DSR strategies may be inspired
In addition, there may be more peripherally involved by all these sources. For example, when following DSR
members in the implementation of the real system to Strategy 2, a researcher may find that an existing theory –
address the client’s problem, but they do not necessarily the ‘theory for explaining and predicting’ or the ‘theory
have a research interest in the project. As an intensive for design and action’ (Gregor, 2006) – informs about
research approach, DSR following Strategy 2 is fairly time a potential solution to the client’s problem. Therefore, the
consuming, with a realistic project requiring an interven- reasoning in that case may be highly deductive from
tion of at least a few months and possibly lasting 1–2 years. theory to the particular solution. Yet, it may be an inter-
As a consequence it is also fairly expensive. esting research project to investigate the forms of reason-
ing at a more detailed level in the case of the two
strategies.
Conclusions The two DSR strategies were contrasted along 16 dimen-
The main contribution of this essay is the explication sions that represent the context, process, outcomes and
of the two DSR strategies. To my knowledge nobody has resource requirements of DSR. It is not my claim that this
introduced this earlier, although the review team raised set of dimensions is comprehensive. For example, one
the question if the two strategies are somehow related to can question if following Strategy 2 is more conducive to
the forms of reasoning in DSR (see Fischer & Gregor, 2011). theory building than Strategy 1 – a theory in the sense of a
It would take too much space to discuss this in detail. One ‘theory for predicting’ or a ‘theory for explaining and
problem is the concept of inductive reasoning. According predicting’ (Gregor, 2006) – since DSR adopting Strategy 2
to Vickers (2009), the traditional interpretation of induc- comprises real systems development in practice. One
tive reasoning as a ‘process of inferring a general law or should note, however, that DSR following Strategy 1 often
principle from the observation of particular instances’ is includes real systems development, too, although usually
outdated. It corresponds to a form of enumerative induc- in a university or research laboratory context. It is not clear
tion in the modern view. Fischer & Gregor (2011) seem to that this context is less important, less informative and less
have this enumerative form of induction in mind when conducive to theory building than the client context in
speaking about inductive reasoning. One question in the the case of Strategy 2.

European Journal of Information Systems


114 Distinguishing and contrasting two strategies for DSR Juhani Iivari

Implications to researchers interested in DSR in IS It is not my intention with the distinction between
The distinction between the two DSR strategies has a the two DSR strategies to suggest implications as deep as
number of benefits to researchers interested in DSR and/ Gibbons et al (1994). Assuming that Mode 2 research gets
or DSR in IS. First of all, it provides a more balanced view of more footing, it will likely increase the popularity of
DSR in IS. The seminal work of Hevner et al (2004) has DSR Strategy 2 in IS. Yet, practicing Strategy 2 does not
been criticized for adopting a biased and narrow view of IS mean that one adopts all canons of Mode 2 research. For
and IS design (e.g., McKay et al, 2008). Although this may example, DSR following Strategy 2 may be discipline-based
be an extreme interpretation, the present essay is ready in the sense that one attempts to contribute to the body of
to accept that Hevner et al (2004) are biased towards knowledge of IS.
DSR Strategy 1. Therefore, it is significant to understand
that DSR in IS also covers Strategy 2, which may be less
susceptible to similar criticism. Implications to researchers conducting DSR
Second, the distinction implies two widely different The intent of this essay is to make design science research-
research strategies, the relative pros and cons of which are ers aware of the two widely different DSR strategies and
still unclear because we do not have much practical expe- their differences, and consequently make them more
rience with DSR Strategy 2. Therefore, it is desirable to reflective of the DSR strategy of their work. More concre-
build a cumulative research tradition of Strategy 2 research tely, the two strategies and their differences in Tables 1–4
so that its viability can be better evaluated. This requires might help design science researchers choose an appro-
that researchers conducting DSR are aware of the research priate DSR strategy. At the risk of dissuading researchers
strategies. from Strategy 2, Tables 1–4 suggest that there are consi-
Third, the distinction also helps to better understand derable risks involved in that choice. First, if a researcher
some controversies around DSR in IS. For example, it has access to a client that is interested in research colla-
helps to make sense of the relationship between AR and boration in a problem area of interest to a researcher,
DSR (e.g., Figueiredo & Cunha, 2007; Järvinen, 2007; Iivari a researcher may jump into the dark to solve the client’s
& Venable, 2009; Sein et al, 2011). The above analysis problem, hoping that it leads to a DSR problem. As a
suggests that AR is indeed a significant research method in consequence, (s)he will face considerable uncertainty
DSR adopting Strategy 2. But the role of AR is much more about the possible DSR contribution. DSR Strategy 2 also
distant in the case of Strategy 1. Yet, in both cases AR is just means intensive involvement in the collaborative pro-
a research method, while DSR is more a research ject over a longer period of time. A researcher should
orientation. consider if (s)he can commit to such an involvement.
At a more general level, the two DSR strategies have Overall, Strategy 2 research is time-consuming and expen-
some affinity with Modes 1 and 2 research as introduced sive, implying that finding funding for the research effort
by Gibbons et al (1994) and Nowotny et al (2001), and as is a big concern.
discussed by Figueiredo & Cunha (2007) in the IS context. On the other hand, Strategy 2 research may be really
Mode 1 research is largely governed by academic interests; rewarding. It implies access to the problems faced in
it happens predominantly in university environments, it is practice, and it possibly leads to a DSR contribution that
discipline-based and quality control is maintained mainly is ‘rooted’ in practice and likely is relevant for practice.
through peer reviewing (Figueiredo & Cunha, 2007). In Working in a multidisciplinary or interdisciplinary project
Mode 2, knowledge production is solution-focused. Know- may be interesting and stimulating, even though there is
ledge is generated within a context of application, which always the risk that the project focuses too much on the
refers to the total environment in which scientific pro- client’s immediate problems and safe, routine solutions.
blems arise, methodologies are developed, outcomes are It is naturally up to each researcher to make up his or her
disseminated and uses are defined (Nowotny et al, 2003). mind based on considerations like these. Yet, since we do
Although there is some similarity, one should note that not have much evidence about the relative pros and cons
the distinction between Modes 1 and 2 is much broader of the two strategies, especially in the case of Strategy 2, it
and ambitious. Mode 2 is suggested as an alternative would be beneficial to have more carefully conducted
model of knowledge production that will supersede Mode Strategy 2 research projects in the future. This requires
1 research, characterized by the hegemony of theoretical that researchers pay attention to the research strategy and
science, by an internally driven taxonomy of disciplines methodology adopted, and carefully follow and report
and by the autonomy of scientists and their host institu- their experiences of the project. That is the way to build
tions (Nowotny et al, 2003). The ambition of the present a cumulative tradition of DSR following Strategy 2.
essay is more modest – to clarify DSR in the IS context.

About the Author

Juhani Iivari is an AIS fellow and professor emeritus at the scientific head of INFWEST/INFORTE programmes for
Department of Information Processing Science, University 10 years. These programmes are joint efforts of a number
of Oulu, Finland. Before his retirement, he also served as a of Finnish universities to support doctoral studies and

European Journal of Information Systems


Distinguishing and contrasting two strategies for DSR Juhani Iivari 115

advanced professional education in IT. Iivari has served in acceptance of information systems, and design science
various editorial positions in leading IS journals such as research in IS. He has published in journals such as
European Journal of Information Systems, Information Sys- Communications of the ACM, Data Base, Information &
tems, Journal of the AIS, and MIS Quarterly. His research has Management, Information & Software Technology, Information
broadly focused on theoretical foundations of information Systems, Information Systems Journal, Information Systems
systems, information systems development methods and Research, Journal of Management Information Systems, Journal
approaches, organizational analysis, implementation and of AIS, MIS Quarterly, and Omega.

References
BASKERVILLE R (2008) What design science is not. European Journal of JÄRVINEN P (2007) Action research is similar to design science. Quality &
Information Systems 17(5), 441–443. Quantity 41(1), 37–54.
BAYAZIT N (2004) Investigating design: a review of forty years of design KUECHLER W and VAISHNAVI V (2008) The emergence of design research in
research. Design Issues 20(1), 16–29. information systems in North America. Journal of Design Research 7(1),
BOYCHUK DUCHSCHER JE and MORGAN D (2004) Grounded theory: reflec- 1–16.
tions on the emergence vs. forcing debate. Journal of Advanced Nursing LINDGREN R, HENFRIDSSON O and SCHULTZE U (2004) Design principles for
48(6), 605–612. competence management systems: a synthesis of an action research
CODD EF (1970) Relational model of data for large shared data banks. study. MIS Quarterly 28(3), 435–472.
Communications of the ACM 13(6), 377–387. MAJCHRZAK A and GASSER L (2000) Top modeler: supporting complex
CROSS N (1993) Science and design methodology. A review. Research in strategic and operational decision making, information, knowledge.
Engineering Design 5(2), 63–69. Systems Management 2(1), 95–110.
CROSS N (2001) Designerly ways of knowing: design discipline versus MARCH JG and OLSEN JP (1976) Ambiguity and Choice in Organizations.
design science. Design Issues 17(3), 49–55. Universitesforlaget, Bergen, Norway.
DAHL O-J, MYHRHAUG B and NYGAARD K (1968) Some features of the Simula MARCH ST and SMITH GF (1995) Design and natural science research on
67 language, Proceedings of the Second Conference on Applications of information technology. Decision Support Systems 15(4), 251–266.
Simulations, pp 29-31. MARKUS ML, MAJCHRZAK A and GASSER L (2002) A design theory for
DENISON DR, HART SL and KAHN JL (1996) From chimneys to cross- systems that support emergent knowledge processes. MIS Quarterly
functional teams: developing and validating a diagnostic model. 26(3), 179–212.
Academy of Management Journal 39(4), 1005–1023. MCKAY J, MARSHALL P and HEATH G (2008) An exploration of the concept of
DEY I (1993) Qualitative Data Analysis: A User-friendly Guide. Routledge, design in Information Systems. In Information Systems Foundations:
London. Answering the unanswered questions about design research, The 4th ANU
DÍAZ ANDRADE A (2009) Interpretive research aiming at theory building: Workshop on Information Systems Foundations, Australian National
adopting and adapting the case study design. The Qualitative Report University, Canberra, Australia.
14(1), 42–60. NOWOTNY H, SCOTT P and GIBBONS M (2001) Re-thinking Science: Know-
FIGUEIREDO AD and CUNHA PR (2007) Action research and design in ledge and the Public in an Age of Uncertainty. Polity Press, Cambridge.
information systems: two faces of a single coin. In Information Systems NOWOTNY H, SCOTT P and GIBBONS M (2003) Introduction: mode 2
Action Research: An Applied View of Emerging Concepts and Methods revisited: the new production of knowledge. Minerva 41(3), 179–194.
(KOCK N, Ed), pp 61–96, Springer, New York, NY. NUNAMAKER JF, CHEN M and PURDIN TDM (1990-1991) System develop-
FISCHER C and GREGOR S (2011) Forms of reasoning in the design science ment in information systems research. Journal of Management Informa-
research process. In Service-Oriented Perspectives in Design Science tion Systems 7(3), 99–106.
Research: 6th international Conference, DESRIST 2011, LNCS 6629 ORLIKOWSKI WJ and IACONO CS (2001) Research commentary: desperately
( JAIN H, SINHA AP and VITHARANA P, Eds), pp 17–31, Springer, Heidelberg. seeking the IT in IT research – a call theorizing the IT artifact. Information
GIBBONS M, LIMOGES C, NOWOTNY H, SCHWARTZMAN S, SCOTT P and TROW M Systems Research 12(2), 121–134.
(1994) The New Production of Knowledge: The Dynamics of Science and PEFFERS K, TUUNANEN T, ROTHENBERGER MA and CHATTERJEE S (2007-2008)
Research in Contemporary Societies. Sage, London. A design science research methodology for information systems
GLASER BG (1992) Emergence vs Forcing: Basics of Grounded Theory Analysis. research. Journal of Management Information Systems 24(3), 45–77.
Sociology Press, Mill Valley, CA. PETROSKI H (1982) To Engineer Is Human: The Role of Failure in Successful
GREGOR S (2006) The nature of theory in information systems. MIS Design. Vintage Books, New York.
Quarterly 30(3), 611–642. SEIN M, HENFRIDSSON O, PURAO S, ROSSI M and LINDGREN R (2011) Action
HAGE J (1980) Theories of Organizations: Form, Process, and Transformation. design research. MIS Quarterly 35(1), 37–56.
John Wiley & Sons, New York. VAISHNAVI VK and KUECHLER Jr. W (2008) Design Science Research Methods
HEVNER A and CHARTTERJEE S (2010) Design Research in Information Systems. and Patterns. Auerbach Publications, Boca Raton, FL.
Springer, New York. VAN AKEN JE (2004) Management research based on the paradigm of
HEVNER AR, MARCH ST, PARK J and RAM S Design science in information design sciences: the quest for field-tested and grounded technological
systems research (2004) MIS Quarterly 28(1), 75–105. rules. Journal of Management Studies 41(2), 219–246.
IIVARI J (2003) The IS core – VII: towards information systems as a science of VICKERS J (2009) The problem of induction, Stanford Encyclopedia of Philo-
meta-artifacts. Communications of the Association for Information Systems sophy, Spring 2009 edn (Zalta N, Ed) (http://plato.stanford.edu/archives/
12: 568–581. spr2009/entries/induction-problem/) (accessed 25 October 2012).
IIVARI J (2007) Paradigmatic analysis of information systems as a design WALLS J, WIDMEYER GR and EL SAWY OA (1992) Building an information
science. Scandinavian Journal of Information Systems 19(2), 39–63. system design theory for vigilant EIS. Information Systems Research 3(1),
IIVARI J (2012) Two strategies for design science research. Working Paper, 36–59.
2012 (https://www.researchgate.net/profile/Juhani_Iivari/contributions/ WEEDMAN J (1998) The structure of incentive: design and client roles
?ev=prf_act) (accessed 14 October 2013). in application-oriented research. Science, Technology, & Human Values
IIVARI J and VENABLE J (2009) Action research and design science research – 23(3), 315–345.
Seemingly similar but decisively dissimilar. In Proceedings of 17th WINTER R (2008) Design science research in Europe. European Journal of
European Conference on Information Systems (ECIS). Verona, Italy. Information Systems 17(5), 470–475.

European Journal of Information Systems

You might also like