Theory and Research Approach
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135
136 137 138 139 140 141 142 143 144 145 146
Research and practice reports on UML are diverse and scattered (Cho & Kim, 2001, 2002;
Sim & Wright, 2001, 2002; Johnson, 2002). We find documentation of negative as well
as positive effects of UML deployment. Also, UML — while studied and described in
many computer science papers — has generally not been studied in the evaluative sense:
what is its economic contribution to a firm? How does the precision of description it
provides trade off against the time it takes to develop and maintain? What is its cost
through the lifetime of an application and for use throughout an organizational department?
Most profoundly, in our view the established literature typically addresses only
a small subset of UML-related issues within a very narrow area of the project effort.
Reflecting upon how UML is used in the organizational context, it quickly became
apparent that decisions and implementation at multiple levels must be involved. Clearly
the organization must permit, if not encourage, its use. Particular projects may subscribe
or not subscribe to the UML approach — perhaps for engineering reasons such as a
project whose mission is not deemed consistent with this approach or for human resource
reasons such as it would take too long to acquire UML development skills to allow the
project to be completed on time. We came firmly to believe that use of UML requires the
resolution of organizational, project, group, and individual issues. With the exception
of textbooks, contributions addressing UML in this larger context could not be found.
The challenge of conducting research on UML in an organizational context is further
discussed in the subsections of search for a theory, theory-driven versus data-driven
research, and use of causal mapping.
Search for a Theory
The lack of an integrative perspective in UML research may not be typical in research
addressing similar research objectives. A review of published research within the topic
area of management information systems identified diffusion theory as a likely candidate
(Fichman, 2000). If we consider UML as a “technology” in the broad sense of a technique
and approach to achieving a set of results, the way that it is diffused and adopted among
organizations, projects, groups, and individuals would perhaps be a reasonable framework
for attempting to address UML usage and the impact it has on project outcomes.
Rogers’ (1995) formulation, sometimes called Rogers’ theory, has been fairly thoroughly
examined using highly precise and rigorous measures — generally gathered through
widely distributed survey questionnaires (Brancheau & Wetherbe, 1990; Moore &
Benbasat, 1991; Prescott & Conger, 1995; Fichman & Kemerer, 1999). Rogers’ theory has
also been extensively criticized (Larsen, 2001; Lyytinen & Damsgaard, 2001). For
example, organizational IS/IT innovations are more complex than Rogers’ theory specifies.
The IS/IT innovation processes unfolding in the “adopting” organization are richer
and more diverse than sigmoidal. The division of users into the categories of early
adopters, early majority, late majority, and laggards is at best unproven but more likely
an introduction of social complexity, yet simplicity, that implies semantic meaning
contrary to innovation process realities (Van de Ven, Polley, Garud & Venkataraman,
1999). Decisions are not purely individual but distributed among individual actors,
groups, project managers, and organizational units. We concluded that using diffusion
theory as the platform for studying UML in an organizational context would be inappropriate.
Our concern parallels the debate that has appeared in venues ranging from top journals
to informal discussion groups about the “rigor” versus “relevance” and “theory” versus
“description” characteristics of research studies (for example, Robey & Markus, 1998;
Benbasat & Zmud, 1999; Nurminen & Eriksson, 1999; Goles & Hirscheim, 2001; Weber,
2003). Given how little knowledge links UML to organizational outcomes, the study of
UML would suggest “relevance” and “description.” At least among the organizations
we ended up visiting, the use of UML remains either a new technology that could
potentially be adopted or one that is partially in use with the jury still out regarding its
precise cost-benefit results. Therefore, adding to the knowledge of UML in the organization
would potentially be of benefit to these practitioners. In order to integrate our
observations regarding the research characteristics of UML with theories of diffusion,
we decided to take a theory building approach.
Theory-Driven vs. Data-Driven
Research
We found ourselves in a predicament — staying theoretically pure or reporting on the
use of object-oriented analysis and design as it unfolds in organizations today. We came
to believe that there is a dialectic struggle between theory and practice (Van de Ven &
Poole, 1995). The employment of theory requires a precise definition of constructs,
variables, relationships, and units of analysis, or has these elements as the planned
outcome of the research effort. Conversely, increasing our understanding of a complex
phenomenon, such as the use of object-oriented design and analysis, requires a
multilevel research design where the outcome cannot be predicted based on available
research reports (Klein & Kozlowski, 2000).
Our research focus on increasing our understanding of UML in its broad organizational
setting suggests that many loose ends and disconnected bridges would emerge. The
indication is that chaos rules. Although this might be expected, we also believe that some
degree of order will exist. In organizational settings, actors allocate time and resources
to issues and phenomena that they believe have importance. A high demand for
resources in defining and executing action related to the issues and phenomena
addressed, as is the case with UML, will increase the likelihood that actors take it
seriously. It is reasonable to suggest that actors will concentrate on particular issues and
phenomena that they deem as being important. Directing attention to issues and
phenomena is a necessary but not sufficient prerequisite for understanding why actors
would pay attention. We also believe that actors are concerned about outcome. That is,
when time is spent on defining issues and phenomena, attention would also be given to
the impact that one of these may have on other issues and phenomena.