Page 233 -

8.10 Analyzing interview data 221

important made during the session. Even if you have to transcribe your handwritten
notes into an electronic format, the amount of transcription required will be substan-
tially less than that needed for a transcript of an audio recording of the same discussion.
These practical considerations play an important role in determining how you an-
alyze the data. If time and money are particularly tight, you might be best served by
an analysis of written records. Detailed examination of recordings is most appropri-
ate for situations where you are interested in digging into the details as deep as pos-
sible and you are willing to commit the resources (time and money) to do this work.

8.10.2 HOW TO ANALYZE
After you decide whether to work from a recording (either directly or via a transcript)
or interview notes, the next step is to decide how to approach the analysis. Interview
analyses usually rely heavily on qualitative methods for coding data, either through
emergent or a priori codes (Chapter 11). These methods attempt to find common
structures and themes from qualitative data. In the case of interviews, your goal is to
identify the important ideas that repeatedly arise during an interview.
One technique that is commonly used for analyzing interview data involves ex-
amination of the text of the interview for patterns of usage, including frequency of
terms, cooccurrences, and other structural markers that may provide indications of
the importance of various concepts and the relationships between them. This ap-
proach—known as content analysis—builds on the assumption that the structure of
an interviewee's comments provides meaningful hints as to what he finds important
and why (Robson, 2002). Discourse analysis goes beyond looking at discussions of
words and contents to examine the structure of the conversation, in search of cues
that might provide additional understanding (Preece et al., 2015). For example, do
users say “we log out of the system when we are done” or do they say “the proper
procedure is to log out when we are done”? The answer to this question might help
you understand differences between what users actually do (the first option) and
what IT managers might want them to do (the second option).
As interview and focus group research generally involves multiple partici-
pants, grouping of comments and resulting codes by participants can often prove
useful as well, particularly if you can identify differences between participants
that might be meaningful to the question at hand. You might find trends in re-
sponses that are associated with the age, educational level, and/or professional
background of the participants. Even if response content does not correlate with
demographic or other obvious variables, trends might indicate multiple “clusters”
of users with similar perspectives. Counts of the frequency of mention of vari-
ous terms, topics, or concerns (50% of participants over 65 years old expressed
interest in the proposed design, while only 25% of those under 65 wanted to learn
more…) might be one means of adding a quantitative perspective to otherwise
qualitative interview data.
If these techniques sound too abstract and theoretical for your taste, you might
want to try something simpler—an introductory approach is given in the Interview
Analysis for Novices sidebar.