Page 183 -

170 CHAPTER 7 Case studies

Integration of the multiple units of analysis is an important aspect of embedded
case study design. In Sara's case, insights from the various tasks were combined in
a classification of challenges that she faced, including control, efficiency, portabil-
ity, and interoperability. Just as these categories provide additional understanding
of the individual tasks, individual units of analysis in an embedded design might be
grouped or viewed from common perspectives.
The inclusion of multiple participants in a case does not necessarily imply an
embedded case study. If participants are not discussed individually, with analyses
identifying similarities and differences between them, they are not distinct units of
analysis. In this case, the group is the unit of analysis in a holistic study. A study of
virtual collaboration in a school in Finland provides an illustration (Lakkala et al.,
2007). Although the class involved 14 students and seven teachers, the case study
does not discuss students and teachers in any detail. Specific comments from both
teachers and students are cited in the paper, but there is no attempt to discuss any of
the participants as individuals, making this a single-case, holistic study. The sidebar
on the International Children's Library presents an example of a case study involving
a theoretical replication across four comparable groups, each of which is a single
unit of analysis.
A paper discussing strategies for sustaining a “community computing infrastruc-
ture” provides an interesting example of an embedded case study (Farooq et al.,
2007b). This single-case study examined an online community aimed at support-
ing professional development for teachers. Four “design interventions”—contact and
bug forms, “needed features” group, task list, and help desk—were chosen as the
units of analysis, due to their differences in terms of goals, primary mode of commu-
nication, participants, and implications for use. Separate discussions of each of these
interventions complemented a general examination of how they worked together to
support the continuing success of the community.
Although the distinction between holistic and embedded analysis might ideally
be made before the study is conducted, the need for multiple units of analysis may
not be clear until after data collection has started. A study of the use of a groupware
tool in a corporate setting might start out as a holistic study of the tool's use in a given
group, only to evolve with time to include embedded analyses of the differing tasks
for which the tool would be used, the roles of the various members in the group, or
the types of project for which it might be used.

7.6 RESEARCH QUESTIONS AND HYPOTHESES
As with almost any other form of research, a good case study is built on the founda-
tions of a theoretical model. Although these theories might not be cleanly testable
hypotheses that can be easily disproved, they can be used to describe what you are
looking for, what you think you might find, and how you will use your data to sup-
port your theories.