Information Technology Reference
In-Depth Information
of information per case regarding 30 participants at year two of analysis. In order to
access the data considering the moment they were produced and the case they referred
to, we made use of NVivo's case nodes to represent our units of analysis and connect
each participant to its sources of information. The use of collections/sets allowed us to
group all the sources (i.e., SISDATS) according to the moment they referred to (e.g.,
1
st
interview). In this way we could see what happened in the data considering only
the information produced in a specific moment, for instance, what participants said
specifically after the first or the second interview.
NVivo also supported our process of building a coding scheme and coding
(probably the most popular technique of data analysis [8]). The
auto code
tool
enabled coding most of the data in our first level categories in a few minutes, after
preparing the data prior to importing it into the software. NVivo is able to
accommodate having
a priori
codes (categories) and creating others
in vivo
, building
directly on the data, two strategies adopted by us (as explained below). This process
usually requires readjusting the coding scheme, merging nodes or splitting them. We
were able to do this while guaranteeing that the original data remained intact. When
merging nodes the original indexes on the data were also merged and we could also
select information already coded at a specific node and then decode it. This meant we
didn't have to repeat operations already done and we could access the data at a
specific node (thematic or case) when considered relevant [9].
Integrating a deductive and inductive approach.
NVivo supported us in integrating
the use of a deductive and inductive approach to coding and analyzing the data. In the
initial exploratory approach we used thematic content analysis [10], adopting
SISDAT's dimensions as our first level categories (e.g., Satisfactions). At this stage
we followed a deductive approach, since our instrument informed our categories,
using NVivo's nodes as labels for themes (codes). During this process we also
developed a set of codes that reflected the different ideas in the text and that were
integrated in the first level categories, hence we also used interpretative codes [7]
(e.g., participants consider that the activity enabled development). At this point the
coding process was much more grounded in the data
2
and gave rise to an inductive
hierarchical coding scheme. This use of deductive and inductive coding enabled
organizing our data, reducing complexity [8].
At the level of analysis, we combined the meaning of a given theme
(category/code) with its quantitative expression, hence integrating inductive and
deductive approaches once again. For instance, in this phase of the project (six
participants) we looked at the percentage of references or words in each theme at a
given moment (see Table 1) which informed us about which theme the participants
explored the most at each moment. The results showed that participants tended to
stress satisfaction in all the moments of the process with the exception of moment
four (2
nd
interview), when dissatisfaction was more stressed. This made us look closer
at the data coded at dissatisfaction at m4 and try to understand what participants
expressed, focusing on their ideas without considering its quantitative dimension. We
2
Not to be confused with Grounded Theory.
Search WWH ::
Custom Search