Page 219 -

8.7 Types of questions 207

small set of predefined responses might discourage elaboration and further comments.
If you ask someone “Did you like the design of the home page?,” they might just say
“yes” or “no.” However, if you ask “what do you think about this home page?,” inter-
viewees may be more inclined to elaborate, describing their reactions in more detail.
This second example—asking “what do you think about…?”—is an open-ended
question. These questions ask for responses, opinions, or other feedback, without im-
posing any external constraints on the responses. This freedom invites the respondent to
answer in depth, exploring any aspect of the issue that may be of interest. Such answers
can often stimulate conversation and generate insights that closed questions might not
reveal. Increased difficulty in analysis is the price that you pay for this insight. Instead
of simply counting answers in different categories, you'll have to analyze the content of
responses to open-ended questions, using techniques described in Section 8.10.
Knowing how you will analyze answers may help you determine which kind of
question to ask. If you want to divide participants into groups, a closed question ask-
ing them which group they belong to is ideal. If you want to understand the relation-
ship between education level and reactions to a proposed community information
system, you might ask people to state their highest level of education completed.
This would clearly establish your categories of interest. In other circumstances, you
might find it useful to divide interviewees into those that have unfavorable, favor-
able, or neutral reactions to an existing system. In this case, a closed question with
three choices would be more helpful than an open-ended question that might lead to
a more ambiguous response.
If you're not quite sure how you're going to use the data, you might be better off
starting with the least restrictive approach. If you're not sure how you intend to use
interviewee age, you might prefer to ask for exact ages rather than ranges (such as
20–29, 30–39, etc.). This will preserve the option of reporting age statistics and ag-
gregating them into ranges for a histogram. If you start by asking for the ranges, you
can't switch to numeric values later.
Other forms of interview “questions” are tasks or exercises that ask participants
to provide useful information, without presenting a question as such. You might
ask users to complete a sentence: “The task that I would most like to be able to
complete with my word processor is….” (Krueger, 1994). This may not be all that
different from asking a direct question, but it does add some variety to the interview
process. Another possibility involves conceptual mapping: asking participants to
draw pictures or graphical layouts that describe their understanding of a situation.
(Krueger, 1994) For a study of perceptions of websites, you might provide a list of
20 sites, asking interviewees to organize the list into groups of similar sites. In one
study of user perceptions of web security, interviewees were asked to draw diagrams
depicting their understanding of how secure web connections work. These pictures
provided concise and informative illustrations of how users understood—and mis-
understood—web security (Friedman et al., 2002) (see Figure 8.2).
Interview questions should be as simple as possible, without any technical terms
or jargon. You don't want your questions to be puzzles that confuse your interviewees.
Compound questions with multiple parts may cause problems for some participants