The analysis was conducted according to grounded theory.
Each interview was transcribed verbatim soon after it was conducted, and the text was read several
times and labeled with descriptive codes. This first level
(open coding) used line-by-line or segment-by-segment
coding. After open coding, selective coding was used to
sort the open codes into clusters thatwere re-coded with
a more specific focus. Questions such as ‘‘What is hap-
pening in the data?’’ and ‘‘What is expressed?’’ guided
the analysis. The open and selective coding processes
were facilitated by Open Code software (version 4.2).
36
The selective codes were grouped together into categories whose properties and dimensions were further identified. Thereafter, a theoretic coding was performed to find
links between categories and to identify a core category.
Throughout the process, from the interviews through the
analyses, notes andmemos were written and figures were
drawn. Thosememos complemented the transcribed text
and the figures captured and illuminated thoughts and
ideas that emerged during the analysis to clarify the interrelationship of the categories. Constant comparisons