light of later analysis and the growing perceptions of the
researcher and each quotation was confirmed on multiple
occasions, adding to the strength of the researcher’s interpretation.
(3) Lincoln and Guba’s method of peer debriefing
was used to confirm interpretations and coding decisions
including the development of categories. (4) Automatic logs
of coding changes were maintained by Atlas-ti and supported
by an archive of project file backups for future
reference. These allowed an audit trail to be maintained,
tracking the development of analysis with annotations for
major decisions and researcher input. (5) Emergent code
categories were tested with a specific view to dependability
and confirmability. In generating themes, tests for co-occurrence
of concepts using text retrieval tools within Atlas-ti
allowed a large range of tests to be applied to the coding.
The results of co-occurrence were checked by reading them
in context and in the form of a list of quotations. Each acted
to verify the analysis and highlight any inconsistencies or
errors. (6) Diagrams illustrating code relationships were
used to visually identify and compare patterns and inconsistency.
(7) A final test of the analysis, and the most
important for this study as a naturalistic inquiry, was the use
of member checking, which confirmed the results as a true
representation of the perceptions and experience of the
participants.