and implications section, but I felt that a more holistic interpretation of the complexities of the multiple realities under investigation, which can be one of the strengths of case study (Stake, 1995), would have been welcome.
Research Framework
Research Questions
By the time I presented my thesis (Wyatt, 2008), my case study of five teachers posed the following overall research questions:
1 What changes do teachers report in their TSE beliefs?
2 To what extent do changes in their TSE beliefs reflect changes in their practical knowledge?
3 Which aspects of the BA TESOL program may have influenced changes in their practical knowledge and TSE beliefs, and how?
I now explain how I came to settle on five teachers.
Gaining Ethical Approval and Sampling
After gaining permission to use an informed consent form, guaranteeing confidentiality, anonymity, and the right to withdraw at any time, I asked for volunteers in September 2003; 16 teachers responded positively. While none withdrew, I reduced the number I was focusing on to 12 then 10 then 6 throughout 2004 (and then to 5 in 2006 during the writing-up stage), but without signaling this process to the research participants, to avoid the risk of hurting feelings and/or reducing learning opportunities, given that I realized participating may have brought some benefits through the more extended interaction it afforded.
When making decisions about sampling, Stake (1995, p. 4) argues that “the first criterion should be to maximize what we can learn.” Indeed, I quickly realized in my research that participating teachers who were relaxed and engaged throughout, had interesting and revealing stories to tell, listened carefully, and were able to expand fluently in clear voices (easily picked up by the tape recorder) made very good interviewees. These were reasons to retain them. In contrast, a teacher who appeared to be suffering from mild interview fatigue was quickly dropped.
I also employed “purposive” and “theoretical” sampling (Silverman, 2010). Of these, “purposive sampling” requires us to “think critically about the parameters of the population we are studying and choose our sample cases carefully on this basis” (Silverman, 2010, p. 141). Through considering factors such as gender, school type (first or second cycle, single sex or mixed, following the new or old curriculum at a time of curriculum renewal) and location (town, rural or mountain village), I could focus on achieving some balance and variety (Stake, 1995).
“Theoretical” sampling, sampling on the basis of hypotheses (Silverman, 2010), was also crucial. One of my hypotheses was that growth in practical knowledge and TSE beliefs would be more evident in dimensions of the teachers’ work they cared about most, where they were focused on self-directed goals (Henson, 2001). Just over halfway through their three-year course, the teachers were asked to articulate goals explicitly when choosing research topics for their dissertations. As I had started with a wide focus (Nisbet & Watt, 1984) in the first research interviews, these topics were well represented in the data. For example, one teacher identified his main research interests in September 2004 as: motivation, group work, adapting materials, and assessment (in no particular order). In the previous year, I had elicited data from him relating to three of these topics, and his choice of dissertation then combined two. It seemed ecologically valid (Cohen et al., 2011) to align my focus with his, to concentrate on themes he was most interested in exploring for himself.
A second and related hypothesis that fed into the theoretical sampling (Silverman, 2010) was that teachers who were planning to research their own practice through conducting action research (Burns, 1999) might, through engaging in processes such as reflection and experimentation, experience the kind of growth in practical knowledge and TSE beliefs that I was particularly interested in investigating. My final selection was therefore of teachers who were planning action research. Implicit in the above hypothesis, however, was a further hypothesis, that the change that would occur was positive, that the BA TESOL would support teachers to become more learner-centered, with a somehow deeper “moral, affective and aesthetic way of knowing” (Golombek, 1998, p. 449). It seemed necessary to test this hypothesis by choosing “a deviant case” (Silverman, 2010), one that did not seem to be following that pattern, to gain a richer, deeper understanding of the theory. Considering this, I was conscious from observations and interviews conducted during the second year of the program that one of the teachers I was investigating was possibly not developing in the expected direction. This teacher was amongst those five finally retained.
Collecting and Analyzing Data, and Writing up the Research
In collecting data, I employed the three key methods discussed by Stake (1995) that are characteristic of qualitative case study research: observations, in which my own role was as a “non-participant observer” (Cohen et al., 2011); semi-structured interviews (Kvale, 1996); and documentary analysis. I used these methods as follows.
While observing in schools, I kept an open narrative record, jotting down descriptive notes relating to actions and reactions, movement, words spoken by teacher and learners, and written on the whiteboard; this commentary was punctuated occasionally by exclamations and questions to remind me of key incidents later. Post-observation discussions were held immediately afterwards in the quietest room we could find and audio-taped on micro-cassette with the teacher’s permission. I elicited feelings about the lesson, memories, reactions, highlighting key incidents and inviting reflection on them, teasing out evidence, encouraging links to public theory and summarizing. While we talked, I added to observation notes, in different colored ink, as exclamations and questions were dealt with. Next, I had prepared questions relating specifically to my research to ask, these generally having arisen from data analysis, reading since the last interview, theorizing, and reflection. I approached this semi-structured phase flexibly, in terms of the order in which questions were asked and the way in which topic areas were explored through follow-up questions, aiming to follow “quality” procedures (Kvale, 1996).
I also analyzed documents, chiefly the teachers’ written assignments, many of which encouraged both engagement with theory and reflection on the planning, implementation, and evaluation of tasks, materials, and activities. These documents included action research dissertations and earlier proposals for these. Analyzing these documents allowed me to compare the participating teachers’ written words with their observed actions and their reports in interview, which would facilitate not only “methodological triangulation,” but also “data source triangulation,” since I would be able to explore changes over time (Stake, 1995).
The data analysis, which was “interactive” and “iterative” (Lincoln & Guba, 1985), informed successive rounds of data collection, and progressed through stages. Observation notes were written up within a few hours, as Stake (1995) recommends, and audio-recorded interviews were also quickly transcribed. This enabled me to use the data “to think with” (Hammersley & Atkinson, 1995). I coded in relation to my research questions, reading and rereading the corpus of data, revised research sub-questions in relation to each teacher to cover emerging patterns and themes, and created categories. I then adopted a “template approach” (Robson, 2002), with key codes, determined by research questions and sub-questions, serving as a template into which coded text segments were placed, which created a matrix. This allowed for the data to be read in a way that facilitated the move from coding to interpretation.
This analytical process supported the making of “thick description” (Geertz, 1973), which involved selecting, organizing, and presenting “interconnected data” (Holliday, 2002) in a way that aimed to support vicarious experience. The narratives I developed, drawing on the different sources of interconnected data, aimed to render the concrete particularities of experience immediate and employ verbal imagery that appealed to the senses (Clandinin & Connelly, 1990). Feedback on draft narratives from PhD supervisors and a critical friend familiar with the context, who together could supply a degree of “investigator triangulation” (Stake, 1995), suggested a need