Page 218 -

206 CHAPTER 8 Interviews and focus groups

a broad range of users—for example, systems aimed at meeting the needs of all
patrons in a large metropolitan library—broadly based focus groups representing
multiple viewpoints may be more helpful. Groups that are too diverse may pose a dif-
ferent set of problems, as a lack of any common ground or shared perspectives may
make conversation difficult (Krueger, 1994). In any case, participants in focus groups
should have an interest in the topic at hand and they should be willing to participate
constructively (Brown, 1999).
Focus groups may be inappropriate for addressing sensitive or controversial top-
ics. Many participants may be reluctant to discuss deeply personal issues in a group
setting. Controversial topics may lead to arguments and bitterness that could destroy
the group's effectiveness (Krueger, 1994). Although such concerns may seem unre-
lated to much HCI work, group discussions can take on a life of their own, possibly
bringing you unanticipated difficulties. If you have any concerns at hand about dif-
ficult issues, you may decide to use one-to-one interviews instead.
Although most focus groups are at least somewhat unstructured, structured focus
group techniques can be useful for building group consensus on topics of common
interest. The Nominal Group Technique (NGT) (Delbecq and Van de Ven, 1971) asks
users to answer a specific question. Participants start by writing individual responses
to the question, which are then provided to a moderator and discussed with the group.
Participants then prioritize their “top 5” responses, and a ranked tally is generated
to identify the most important consensus responses to the question at hand (Centers
for Disease Control, 2006). An NGT inquiry into the information needs of home-
care nurses and their unmet information needs in dealing with geriatric patients after
hospital discharge asked participants “In your experience, what information-related
problems have your elderly patients experienced that contributed to hospital readmis-
sions?” Respondents identified 28 different needs in six different categories, includ-
ing medication, disease/condition, nonmedication care, functional limitations, and
communication problems (Romagnoli et al., 2013).

8.7 TYPES OF QUESTIONS

As seemingly small differences in the phrasing and form of interview questions can
lead to big differences in responses, you should pay careful attention to what you ask
your interviewees and how you ask the questions. Although writing these questions
is more of an art than a science, there are some guidelines that should help you get
started in the right direction.
One of the first considerations in the construction of any interview question in-
volves the degree of structure. Structured, closed questions limit users to a small
number of predefined choices. Examples include yes-no questions, multiple choice,
true-false, and Likert-scale questions, asking for ratings on a scale of 3, 5, 7, or more
possibilities (Robson, 2002). These questions have the advantage of being easy to ana-
lyze, as responses can be tabulated across all participants, and statistical methods can
be used to describe the distribution of responses. However, giving your interviewees a