3.3 Data collection and analysis
3.3.4 Data analysis
In qualitative research, data collection and analysis is a simultaneous process (Miles
134 suggests that „without ongoing analysis, data can be unfocused, repetitious and
overwhelming in the sheer volume of material that needs to be processed‟ (p.171).
Ongoing analysis helps the researcher to test and refine the hunches and working
hypotheses emerging during data collection. However, data analysis does not end
with the completion of data collection. It becomes a more intensive process after all
the data has been collected (Merriam, 2009).
Qualitative research involves both synthesis and analysis. It is a process of both
taking things apart and putting things together. Stake (2010) explains,
As qualitative researchers, we try to be especially sensitive to what are wholes, things that resist being taken apart, but still we analyse them. And sometimes we put the facts together into new wholes, into new interpretations, into a new patch (Stake, 2010, p.134).
In this section, my intention is to elucidate how these processes of analysis and
synthesis are used in this research.
Bogdan and Biklen (1992, 2007) make ten important suggestions to help researchers
to make analysis an ongoing part of data collection. These include; „force yourself to
make decisions that narrow the study; to make decisions concerning the type of study
you want to accomplish, develop analytic questions, plan data collection sessions
according to what you find in previous observations, write many observer‟s
comments as you go, write memos to yourself about what you are learning, try out
ideas and themes on participants, begin exploring the literature while you are in the
field, play with metaphors, analogies and concepts, use visual devices‟ (Bogdan and
135 I have followed most of the above suggestions in this research. I entered the research
setting armed with a set of research questions which was later refined according to the
emergent needs of the setting and the relevant literature. It has helped me to narrow
the data collection and to focus on relevant data. I also carefully recorded the
hunches, feelings, reactions, initial interpretations, speculations and working
hypotheses, thoughts and questions about the setting, people and activities. These
notes helped me to plan the subsequent data collections and for further analysis at the
end of the data collection.
3.3.4.1 Analysis of the activity system of student learning
To address the research question „How do the students regulate their learning in
relation to the contextual demands in the undergraduate course and their own valued
outcomes?‟ I needed to analyse both contextual demands and individual actions.
According to the activity theory, individual actions are subordinated to the collective
activity and therefore the analysts have to construct and analyse the activity system
first and then the individual actions. To construct the activity system, I had to take
into consideration the students‟ and their lecturers‟ views and data collected from
observations and documents together. Therefore, in the data collection I made it a
point to test the emerging view of the activity system and its characteristics from the
perspective of the students and the lecturers. For example, when it emerged in the
ongoing analysis that there seemed to be a contradiction in the object or the collective
136 teachers as to how they viewed their purposes at different points in time of the course
and why, in the subsequent interviews.
The process that I adopted in my data analysis consisted of the following steps
derived from the suggestions made by Miles and Huberman (1994), Bogdan and
Biklen (1992) and Merriam (1998, 2009):
Transcription Organising data Category construction Sorting categories and data
Identifying patterns, themes and theorising.
Step 1: Transcription
All my taped data from students‟ interviews were in the Sinhala medium because the
students preferred to have their interviews in their mother tongue. I transcribed them
on my own and they were anonymised by allocating each student a pseudonym. The
field notes and the interviews with the lecturers were in the English medium. I also
transcribed the interviews with the lecturers on my own and they were also
anonymised in the process. All transcribed data, students‟ and their lecturers‟
reflective accounts and the field notes were word processed and stored in a computer
with backup files stored independently. The whole process of transcription and word
137 Step 2: Organising data
For easy retrieval, I gave an identity code indicating who and what was involved,
when and where it was collected/done to each interview transcript and other
documents. The narrative data was given page numbers to make it easy to trace back
units of text to their original context. A separate file was manually maintained to
store the memos and notes which recorded insights and hunches about what was
going on in the research setting that came to my mind throughout the research
process.
Step 3: Category construction
According to Merriam (2009), the practical purpose of data analysis is to answer the
research questions and the process is a highly inductive and comparative activity. The
method of analysis I have employed in this research is drawn from the constant
comparative method (Glaser and Strauss, 1967) used for developing grounded theory
by Glaser and Strauss. As pointed out by Merriam (2009), qualitative researchers
widely use the constant comparative method in their analyses without building
grounded theory. I started my analysis with interview transcripts of students and by
identifying „chunks‟ of data that were relevant to my research questions. I selected these „units of data‟ while paying attention to the two criteria suggested by Lincoln
and Guba (1985). The criteria were;
i. The unit should include information relevant to the study and stimulate the
reader to think beyond the particular bit of information;
ii. It should be the smallest piece of information about something that can stand
138 information other than a broad understanding of the context in which the
inquiry is carried out (p.345).
The process of category construction began with reading the first interview transcript.
As I went through the transcript I wrote down notes, comments, observations and
queries in the margins. After going through the entire transcript I looked at them
again to check whether those comments or the codes that go together could be
grouped in some way to arrive at more encompassing categories. I made a running list
of these categories, which was fairly long, on a separate sheet. Then, I took the
second transcript and followed the same procedure while bearing in mind the first set
of categories that I had identified from the first transcript. I made another separate list
of developing categories using this second transcript. Finally, I compared the two lists
and merged them into one master list of categories. This tentative set of categories,
which was further refined as I went along with categorising the data set, was
subsequently used to identify recurring patterns of data in the whole data set.
Step 4: Sorting categories and data
The tentative set of categories derived at the end of the above process was used to sort
all the data. Separate folders labelled with different category names were used to store
each unit of data that belonged to specific categories. The units of data were cut and
put into the file manually, after attaching identity codes containing the respondent‟s
139 Step 5: Identifying patterns, themes and building theories
Sorting data into categories helped me to describe the data and to interpret them to
some extent. However, to become more theoretical I had to be engaged in a deeper
thinking process to make inferences, develop models or generate theory. In the words
of Miles and Huberman (1994) I had to move up :
...from the theoretical trenches to a more conceptual overview of the
landscape. We‟re no longer just dealing with observables, but also
with unobservables, and are connecting the two with successive layers
of inferential glue‟ (Miles and Huberman,1994, p. 261).
According to Merriam (2009) thinking about data helps us to develop theory that
explains „some aspect of practice and allows a researcher to draw inferences about future activity‟, (p.188). Thinking about and comparing concepts and data helped me
to identify patterns and themes as summarised in Table 8.
Table 8: Initial categories, patterns and themes that emerged in the analysis of the activity system
Initial categories Patterns Themes
Nature of the object or the collective purpose.
Views on the object
Views on the
curriculum, teaching and assessment.
Dual nature of the collective purpose
Changing views about the object/collective purpose Changing understanding about teaching, curriculum and assessment
Primary contradiction in the collective purpose
Understanding the collective purpose is a gradual process
Understanding teaching, curriculum and assessment is a gradual process
140 Maximising grades, Competition vs. collaboration Mimicking Mastery. Following additional vocational courses Utilising opportunities to enhance employability skills Improved self efficacy Self awareness Understanding others Interpersonal skills Employability skills Mismatches between expectations and realities of teaching, curriculum and assessment. Collective responses to assessment demands Collective responses to the need to enhance employability Reported outcomes Increased action possibilities Identity transformation Managing interpersonal relations
Tensions or contradictions between the collective purpose, teaching, curriculum and assessment in classrooms where teaching was based on transmission approaches
Tensions emerged between institutional assessment practices, teaching, curriculum and the collective purpose in classrooms where teaching was based on social constructivist approaches.
Collective responses to the contextual demands
Indications of expansive learning at the individual level at varying degrees
Finally, I have used the themes identified in the analysis of the activity system and the
findings of cross case analysis to build a model to explain student learning in the
undergraduate course. In the next section, I explain the process of analysis of case
141
3.3.4.2 Analysis of case studies
One of the key steps in case study analysis is within-case analysis (Eisenhardt, 2002).
Here, the researcher has to construct detailed case study reports for each case. These
reports are „simple, pure descriptions‟. However, they play a key role in generating
insights. They help the researcher to become familiar with each case as a single unit
and to identify unique patterns emerging from each case before searching for patterns
across cases. So, in this research I developed detailed descriptive write-ups of the
three cases, using the three sets of interviews I had with each of them at different
intervals of their course, my own observations and their reflective comments.
Within-case analysis informed by the concept of expansive learning (Edwards, 2005a;
Roth and Lee, 2007), helped me to identify five categories to compare the cases.
According to Eisenhardt (2002), cross case comparisons „force the researchers to go
beyond initial impressions by using structured and diverse lenses‟ (p. 19). They help to improve the likelihood of accurate and reliable theory that fits closely with the
data.
The next step was to „compare systematically the emergent frame with the evidence
from each case in order to assess how well or poorly it fits with the case data, (p.20).
In this thesis, I have compared the emerging five factors that affected expansive
learning at the individual level and the evidence from each of the case studies that
supported, or otherwise, the effects of those factors. Then, I have compared the
emerging findings with the related literature. As argued by Eisenhardt (2002), linking
emergent findings to existing literature enhances internal validity, generalisability and
142
The final stage of Eisenhardt‟s model for building theories from case studies is reaching closure which is marked by „theoretical saturation‟. Theoretical saturation is
the point where incremental learning is minimal because the researchers are
observing phenomena seen before (Glaser and Strauss cited in Eisenhardt, 2002,
p.26). In this research, theoretical saturation is achieved after comparing the three
cases which had been selected on the basis of the nature of the reported outcomes.
The three cases represented a range of outcomes which were categorised as most
elaborated, moderately elaborated and least elaborated.