Page 213 -

8.5 Interview strategies 201

CONTEXTUAL INQUIRY
Many HCI researchers and practitioners have found that simply asking people
about their practices is not sufficient for developing a complete understanding
of user requirements. If you ask someone who regularly makes scrapbooks
how they go about doing it, they may share certain interesting details that
demonstrate their explicit understanding—those parts of the process that they
can think of and easily describe to you. If you watch that same person complete
the task, you might find many implicit practices that are crucial for success,
even if they aren't stated directly.
A popular exercise used in HCI and other computing classes provides
a nice demonstration of the notion of implicit knowledge. The challenge
involves sandwich construction. Students are asked to describe how to make
a peanut butter and jelly sandwich, assuming one is given a loaf of bread and
new jars of peanut butter and jelly. Participants invariably find that seemingly
simple tasks—such as getting a knife full of peanut butter to be spread on
the sandwich—are complicated by challenges that may not be remembered
explicitly—in this case, removing the foil seal that might be found under the
lid of the unopened jar of peanut butter (Davis and Rebelsky, 2007). If you
limit your investigations to direct interviews, you might never come across
interviewees who remember this crucial step. If you instead choose to observe
someone in action, your first participant's attempts to remove the foil point to
the need to include it in your process.
Contextual inquiry techniques for conducting interviews (Beyer and
Holtzblatt, 1998) are specifically designed to uncover implicit knowledge about
work processes. Contextual inquiry starts from observation at workplaces, with
a focus on specific details rather than generalizations.
The simplest form of contextual inquiry is the contextual interview
that consists primarily of a few hours of observation as the user completes
his or her work. The goal is to form a partnership in search of a shared
understanding of work. The preferred approach to this is to have the researcher
and the interviewee work together in a manner similar to a master-apprentice
relationship, with the participant describing what she is doing and why as
she progresses through the various steps involved in completing her work.
Researchers conducting contextual interviews are generally much more
talkative than traditional apprentices, leading to a conversational partnership.
This collaboration extends into interpreting the data: the researcher begins to
build a model of how the interviewee is working and asks if it reflects the user's
understanding. If the interpretation is incorrect, the interviewee is likely to
clarify: “No, that's not quite right.” This discussion takes place in the context of
a focus on the project as a whole, as opposed to any smaller components, such
as the software that you might eventually design (Beyer and Holtzblatt, 1998).
(Continued)