Database Reference
In-Depth Information
surprisingly, they are seldom willing to accept an untrained observer as part of their
team. Since information visualization researchers are of necessity highly trained
themselves, it is rare that an information visualization researcher will have the neces-
sary additional training to become accepted as a participatory observer. However,
domain expertise is not always essential for successful participatory observation.
Expert study participants can train an observer on typical data analysis tasks - a proc-
ess which may take several hours, and then “put them to work” on data analysis using
their existing tools and techniques. The observer keeps a journal of the experience and
the outcomes of the analysis were reviewed with the domain experts for validity.
Even as a peripheral participant, valuable understandings of domain, tasks, and work
culture can be developed which help clarify values and assumptions about data, visu-
alizations, decision making and data insights important to the application domain.
These understandings and constructs can be important to the information visualization
community in the development of realistic tools.
Laboratory Observational Studies:
These studies use observational methodologies
in a laboratory setting. A disadvantage of in situ observations is that they often require
lengthy observations. For instance, if the observer is interested in how an analyst uses
visual data, they will have to wait patiently until the analyst does this task. Since an
analyst may have many other tasks - meetings, conference calls, reports, etc. - this
may take hours or even days. One alternative to the lengthy in situation wait is to
design an observational experiment in which, similarly to a laboratory experiment, the
experimenter designs a setting, a procedure and perhaps even a set of tasks. Consider,
for example, developing information visualizations to support co-located collabora-
tion. Some design advice on co-located collaborative aspects is available in the com-
puter supported cooperative work literature [35]. However, while this advice is useful,
it does not inform us specifically about how teams engage in collaborative tasks when
using visual information. Details such as how and when visualizations will be shared
and what types of analysis processes need to be specifically supported in collaborative
information visualization systems were missing. Here, an observational approach is
appropriate because the purpose is to better understand the flow and nature of the
collaboration among participants, rather than answering quantifiable lower-level ques-
tions. In order to avoid temporal biases in existing software, pencil and paper based
visualizations were used. This allowed for the observation of free arrangement of
data, annotation practices, and collaborative processes unconstrained by any particular
visualization software [36].
Contextual Interviews:
As noted in Section 5.1, interviewing in itself is core to
qualitative research. Conducting an interview about a task, setting, or application of
interest within the context in which this work usually takes place is just one method
that can enrich the interview process. Here the realism of the setting helps provide the
context that can bring to mind the day-to-day realities during the interview process
(for further discussion see Holtzblatt and Beyer 1998). For example, to study how
best to support the challenging problem of medical diagnosis, observing and inter-
viewing physicians in their current work environment might help to provide insights
into their thought processes that would be difficult to capture with other methodolo-
gies. A major benefit of qualitative study can be seeing the big picture - the context in
which a new visualization support may be used. The participants' motives, misgiv-
Search WWH ::
Custom Search