Page 171 -

158 CHAPTER 7 Case studies

These three approaches to technology biography provide opportunities for gather-
ing insights that might be difficult to acquire using only one method. By asking Sara
to talk about both past experiences and future aspirations, the research design al-
lowed for the possibility of understanding changes in her relationship to technology.
Sara's demonstrations of the tools provided an example of current use of artifacts.
The examination of tools (such as the tactile watch and screen-reading software) can
be an important source of data for case studies, particularly when you are interested
in understanding how users complete tasks.
Multiple data sources can also provide corroborating evidence to increase your
confidence in observations. In a case study of workplace information management,
you might start your data collection with interviews of employees. These discussions
provide useful data but they are limited: participants may have different understand-
ings of practices and habits, they may be unwilling to comment on the details of
their work, or they may simply forget important details (Chapter 8). Investigation of
the artifacts of their work—computer files, paper records, archives, and e-mail mes-
sages—can provide concrete understanding of actual practices, free from the limita-
tions of interviews. This analysis may confirm statements made in interviews, thus
increasing your confidence in their validity. The use of multiple sources to provide
corroborating evidence is known as data triangulation—a reference to the practice
of taking measurements relative to multiple known reference points in order to pre-
cisely measure location.
Multiple data sources can also help deal with any concerns about the quality of
the data provided by any single source. Due to the relatively small number of cases
involved, the use of any single data collection technique with a particular case may
not give you the data that you really need. For example, if Sara had some residual vi-
sion that allowed her to make use of some visual display components on a computer
screen she might not be an appropriate participant in the case study. We have more to
say about selecting cases in Section 7.7, but for now, we mention that simply asking
Sara about her use of technology might not have revealed her use of visual displays.
A combination of interviews along with direct observation of her work might provide
more appropriate measurements; logs of computer activity—taken when she wasn't
being directly observed—might be even more realistic.
Of course, the use of multiple data sources does not guarantee nice, clean cor-
roboration of results—if only it were that easy. Two scenarios may arise that make
life more interesting. Your data sources might diverge, with each source of data cov-
ering different observations. This is not necessarily a problem, as all of the observa-
tions may have some validity. When this happens, your use of multiple data sources
has not increased the validity of your analyses—you simply have many observations
that fail to support each other. You may need to be cautious about your interpretation,
refraining from strong claims until you can find some corroboration.
The possibility of contradiction is a more troubling concern. Suppose one source
says that something is true, while another says that it is not? You may need to look
carefully at the specific details of the claims and the specific sources, in order to
determine which is plausible. Contradictions may also motivate you to dig deeper,