Performing Subjectivist Studies in the Qualitative Traditions Responsive to Users - Evaluation Methods in Biomedical Informatics - page 288

Qualitative analysis software is one method for supporting this analyti-
cal process, but some researchers find handwritten lists or spreadsheets
more workable at early stages of a study because they are more compati-
ble with the personal style of the investigator. (Even the most flexible soft-
ware imposes a process or structure that the investigator may not find
congenial.) The investigative team needs to come to agreement on patterns
and themes that emerge from the data; this is best done face to face. Track-
ing these discussions is critical because the records of them become a highly
distilled form of data. As larger themes emerge from the data, it is possible
to revisit the older data with new insight. The discussion of themes often
generates new research questions or new foci for future fieldwork. As soon
as some themes seem to be coalescing, it may be time to introduce member
checking, taking the results back to the informants and asking if they seem
right. As the study progresses, the themes might become codes for a code
manual or a template for further data collection. At this point, an “outsider”
who has not been involved in collecting the data can review both the data
and the ultimate themes as a further confirmatory step.
Another way of analyzing the data is to use narrative analysis. This kind
of analysis examines both the content of the text and its structure. For
example, the structure of informants' stories can be analyzed for their plot
development. Stories are narrative descriptions that generally follow a stan-
dard format: the stage is set; something happens; there is an ending. The
stories of different informants can also be compared with one another. For
example, Stavri and Ash 25 analyzed stories told by informatics experts using
this comparative approach. Analyzing the structure as well as the content
of the stories, they found that stories about successful implementations of
computerized physician order entry usually began with descriptions of prior
failures. This finding was important because it raised a new research ques-
tion (how might failure of CPOE systems breed future success?) and it pro-
vided insight into the thinking of the storytellers (their memories focus on
the failures first, perhaps because those are more vivid than memories of
success).
Computer Software
Software can be helpful during both the data analysis and result reporting
processes, especially when there is a large amount of data to be analyzed
by multiple researchers. Such software may be highly compatible with the
working style of persons aligned with the field of informatics, who may
prefer reading and coding using a laptop over marking up paper documents.
The software must be able to easily import text documents in a standard
word processing format since undoubtedly transcripts and field notes will
be in such a standard format. It is up to the investigator to determine the
granularity of the coding: by line, sentence, paragraph, or section. During
coding, the investigator reads each segment and assigns terms just as he