Performing Subjectivist Studies in the Qualitative Traditions Responsive to Users - Evaluation Methods in Biomedical Informatics - page 278

outline a time frame for data collection, and decide what resources and
budget are needed.
Site and Informant Selection
In many evaluations, the sites for field investigation are predetermined, but
this is not always the case. When there is flexibility in site selection, the
choices are invariably driven by the study goals or research questions. For
example, if the goal is to understand how house staff (interns and residents)
view a new EMR, teaching hospitals should be the focus. On the other hand,
to compare views of clinicians at teaching and nonteaching hospitals, it
would be necessary to do fieldwork at at least one site of each type. Even
if the evaluation site is predetermined, the investigators still need to decide,
for example, which unit(s) of the organization might be the focal points for
fieldwork.
Whether selecting a study site or selecting the individuals to interview or
observe, the logic of the selection procedure is similar. Qualitative methods
usually entail a “purposive” selection, meaning that deliberate selection is
based on the purpose of the study, and the selection strategy can evolve as
the progress of the study reveals initially unanticipated needs for subjects
who bring potentially novel viewpoints. Investigators use their judgment to
select appropriate sites and subjects. They may begin selecting individual
study subjects, customarily known as “informants,” by finding the most
knowledgeable people, those who know something about the focus of the
evaluation (the EMR, for example) and who also know other people in the
organization. These important contacts can then identify others in the orga-
nization who may be productively included in the study, and these subjects
can in turn recommend others, creating a “snowball” effect.
This approach contrasts with convenience sampling, which is less delib-
erate, and based purely on availability and happenstance. For example, an
investigator studying the utilization of a new bioinformatics consulting
service based in a health sciences library might engage as an informant
anyone who happens to use the service on the days the investigator is on
site. Because measurement and estimation, as discussed in Chapters 4 to 8,
are typically not performed in qualitative studies, random sampling does
not enjoy in qualitative work the preeminent role it plays in quantita-
tive/objectivist studies.
Careful purposive selection of informants is almost always the preferred
method for qualitative investigation. This is especially the case when choos-
ing subjects for interviews that are very expensive to conduct when the costs
of transcription and analysis are added to the very real costs of the time of
the interviewer and interviewees. Informants should be selected based on
the information or expertise that they can share. 13 As a general rule, both
expert and nonexpert users of an information resource should be included.
It is useful to seek out the outlier and the skeptic in addition to those who