4.15 Planning and conducting phenomenographic interviews
4.15.3 Analysis strategies for Data Set 4: Graduates
The data comprised verbatim transcripts of six tape-recorded semi-structured interviews with graduates, each taking approximately 90 minutes, which generated 147 pages of text.
Detailed reading and re-reading of each transcript was undertaken until differences and similarities related to several themes or key topics emerged from the data. Critical variations in meaning across transcripts were the key aspects that I sought out in the reading of phenomenographically-derived data. Quotations were then selected from each transcript and put into a “data pool”. The decontextualised “pool of meanings’ was then sorted and re-sorted, searching for categories of description that seemed to capture the collective experience of the graduates. Within the categories the utterances displayed a commonality, but the categories were distinguishable from each other in their variation. Clear criteria for each category were gradually clarified and refined. The categories were neutral, with respect to empirical subjects (individuals) and to the context or life-world from which they emanate; by disconnecting the conceptions from the “temporal and situational context” it was possible to compare conceptions to other expressed, decontextualised conceptions in the “data pool” (Uljens, 1996). Phenomenographers claim only to analyse the participants’
“ways of functioning” or “forms of thought” and not their lived experience, so the words used are of critical importance (1981, 1984). Uljens (1996) stated that the object of phenomenographic analysis always consists of “expressed experience”, so the expressions used by each participant were critical when inferring meanings. The actual words, or the way in which the experience of curriculum and its various constituent parts was articulated by each participant, were also important for the identification of the referential aspect (what does it look like?) and the structural aspect (how is it seen?) of each conception.
170
Although every effort was made during the process of analysis to bracket knowledge about approaches to learning, and my personal understanding of graduates’ lived backgrounds, it was difficult to set them aside completely. In the result, the categories might have reflected some of my previous understandings about learning law and the particular curriculum that was the phenomenon under scrutiny. As a deep “insider-researcher”, I have great familiarity with the sub-cultures that operate within the Law faculty and so I was sensitive to the nuances of meaning in the data (cf. B. Edwards, 1999). I was able to treat the graduates in an empathetic manner, knowing the terrain which they described, but I had to exercise extreme caution in not reading my own interpretations or imposing my own understandings on their expressed experiences. The precautions taken by me, in separating my preconceptions from the realities within the data, are more thoroughly canvassed in Chapter 4 (above). It is notable, however, that in phenomenographic analysis, there is “an unavoidable relationship between the researcher and the phenomenon that is investigated”. The researcher of necessity must have a “thorough knowledge and understanding” of all aspects of the phenomenon that she is analysing. This is essential to enable the researcher to “discuss and query the learner about the various facets of the phenomenon” (Stamouli & Huggard, 2007).
Morris (2006) following Bowden and Walsh (2000) highlights two possible approaches to data analysis in phenomenography: either the neutral discovery of categories from the data, or an imposed “construction” of categories of description. Researchers such as Laurillard (1993), Crawford et al (1994) and Prosser and Millar (1989) discussed above (page 160) have specifically used predetermined categories to extend, replicate or compare empirical findings across different settings or in different disciplines The imposed “construction”
approach to producing an outcome space is associated with a structure of hierarchical relationships between the categories of description, such as Saljo’s (1979) conceptions of deep and surface approaches to learning, followed and extended by Marton et al. (1993).
“Discovery” of categories appears to mean a faithful capturing of conceptualisations that emerge from the data, similar in process to grounded theorising. It is unclear whether all the content of the data must be included or only selected data that “fits” the emerging categories. Bowden and Walsh recommend an inclusive approach, which incorporates all
171
data “even if the result is untidy, so that for example, the results do not have a logical structure” (2000).
It is notable, however, that cultural beliefs and practices have been shown to exert a role in learning and thus conceptions of learning in studies of “non-Western” students, e.g. Chinese and Nigerian, paradoxically do not align closely with the constructed categories of such typologies (Boulton-Lewis, Marton, Lewis, & Wilss, 2004; Cliff, 1998). In the context of student diversity that is represented in South African institutions it was thus preferable to follow a “discovery” approach to developing categories of description from the data, and to remain open to the possibility that a hierarchical relationship between these categories may emerge from the data. Both methodological approaches could be accommodated within the ambit of “variation” that is typically a feature of phenomenography, but the implications of each approach relate to the credibility and reliability of the research as discussed below.
During the process of analysing data, it is crucial to attempt to maintain an open mind, minimising any predetermined views, and by maintaining a focus on the transcripts and the emerging categories of description as a set (Åkerlind, 2005). The researcher will be better able to maintain focus on the collective experience rather than on individual transcripts and categories. There is a need to “bracket” the researcher’s presuppositions throughout the research process, to ensure that the focus remains on the participants’ (students’ and employers’) experienced world, acknowledging however that the researcher cannot suspend her commitment to certain guiding notions concerning the starting point or shared topic of the researcher and the participant (Ashworth & Lucas, 2000).
The interpretive role of the researcher must be acknowledged, in that the final outcome space represents a partial understanding of the phenomenon, reflecting the analysis of the data as experienced by the researcher (Svensson & Theman, 1983). It is however essential that the results of the research, the outcomes, are firmly based upon the empirical data (Åkerlind, 2002). This attitude was uppermost in my mind throughout the analysis. Although my personal interest and knowledge of the topic informed my interpretation of the data, by using many direct quotations from the transcripts to support my analysis, I was able to add credibility to the interpretation.
172
The stages in the analysis of the data began with several careful readings of the transcripts as a “sensitisation” to the data, before beginning the search for meaning. The initial data exploration seeks critical variations in meaning, looking for expressions that reveal how the phenomenon appears to the participants across interview transcripts. It is imperative that a focus is maintained on the transcripts as a set, rather than as individual transcripts (K Trigwell et al., 1999, p. 70). The search for structure should be delayed in the early stages of the process, until the aspects of meaning are fully appreciated (Åkerlind, 2002).
The analysis of each data set was treated as a separate process and was carried out in sequence, i.e., the graduate data (Data Set 4) was analysed completely into an outcome space before the employer data (Data Set 5) was analysed at all. Bowden (1995) recommends at least six readings, to focus on a different perspective each time, so as to clarify what the interviewee meant in relation to key elements of the phenomenon. His approach is to treat the entire transcript as a set of interrelated meanings, best understood in relation to each other. In practice, transcripts are allocated to the emerging categories, which can result in an overemphasis on an individual focus in analysing the data, rather than a collective one (Åkerlind, 2002). I preferred to adopt Marton’s approach (1986) which is to first undertake a “selection process”, during which quotations from transcripts are identified and marked as being of interest to the questions under investigation. These quotes are added to a decontextualised “pool of meanings” from which the researcher begins to develop categories, based on similarities between the utterances. The categories must be distinguishable from each other based on the differences between them (Marton, 1986).
This approach has the advantage of ensuring greater clarity on key aspects of the phenomenon and facilitating the management of the data, yet it may lead to the researcher not giving sufficient weight to the context within which the selected quotes are made.
The time-consuming and complex nature of the process of phenomenographic analysis, juggling concepts, parts and whole in one’s head, is effectively captured by Trigwell, who comments:
I find the analysis very demanding, and I tend to put it off as long as I can…but the only time I can do the analysis effectively is when I have absolutely no interruptions.…Having to keep a lot of ideas active in your head at the same time (is demanding)…you try to hold two, three, and so on…getting started is very
173
hard…the parts and the whole define each other dialectically…There are times when the focus is, and has to be, on the parts. But in order to see whether these parts are, in fact, parts of the same category, the focus has to be on all the transcripts (Trigwell, 2000, p. 70).
This resonates exactly with the process I experienced in constantly shuffling between the transcripts and relevant excerpts that expressed aspects of the phenomenon under review.
The procedure entailed selecting quotes from the transcripts that were relevant to particular key aspects of the phenomenon of experiencing the law curriculum and how it prepared a student for professional practice, focusing on finding the “dimensions of variation” for each aspect of the phenomenon (Marton & Booth, 1997). The quotes were interpreted in two contexts: the interview from which they had been selected, and in the context of the “pool of meanings”. Following Marton’s (1986) approach, the quotes were gradually allocated to categories on the basis of their similarities, and the categories were distinguished from each other. This allowed the criterion attributes for each category to become clear.
Thereafter, an iterative process of moving between the emerging themes and the original data, to confirm, contradict or refine the categories, as well as searching for structural relationships between meanings, continually sorting and re-sorting the data took place (Åkerlind, 2005, 2008). The key qualitative differences between the categories were refined, which entailed grouping and regrouping selected quotes from transcripts, to develop hierarchies in relation to given criteria, until there was “a decreasing rate of change and eventually the whole system of meanings is stabilised” (Marton, 1986).
Bowden and Walsh caution that there is a risk that researchers may explicitly reflect their own professional judgement in imposing categories, and thereby potentially ignore aspects of the data in their interpretation, so as to create a “tidy construction that is useful for some further explanatory or educational purpose” (2000). This cautionary note was heeded in the analysis, to avoid developing too “neat” an outcome space.
It is imperative that the writing up of the research be explicit about the nature of the process engaged in, as issues such as the researcher imposing her own presuppositions, or insufficiently “bracketing” her conceptions of the forms and concepts within her discipline
174
may intrude on the accurate reflection of the participants’ conceptions (Ashworth & Lucas, 1998). This caveat was adhered to in the data analysis to enhance the validity of the study.
A “conception” of a phenomenon is developed through the relationship, or the understanding that a participant has of the experience, through simultaneously discerning and focussing on critical aspects of the phenomenon (Bond & Le Brun, 1996). In this study, the critical aspects of curriculum that were discerned by each participant were identified and used to develop an understanding of the collective experience of the graduates, and then of their employers.
Each conception is composed of two intertwined aspects: the referential aspect, which denotes the particular global meaning of the individual object conceptualised, ”so as to discern it from its context” (Marton & Fai, 1999); as well as a structural aspect, which is the combination of features discerned and focussed upon by the participant. Awareness of a phenomenon is constituted by the “what” aspect, which corresponds to the object itself, and the “how” aspect, which relates to the act of discerning it, in a dynamic relationship (Marton & Booth, 1997). This distinction between the “what” and “how” aspects is, however, merely an analytical tool that is only used by the researcher to assist in the analysis (Stamouli & Huggard, 2007, p. 183). Students’ experiences of curriculum had a referential aspect, which is the meaning aspect (what does it look like?), which relates to the direct object under scrutiny, the curriculum, as well as a structural aspect that concerned the constituent parts of the whole experience and their relationship to each other, including contextual factors (how is it seen?) (Morgan & Beaty, 1997). The graduates’ different experiences of curriculum were described in terms of the dynamic relationship between discerning the “what” (referential) aspect, which refers to their understandings of the global meaning of curriculum, and discerning the “how” (structural) aspect, which is the act of discerning curriculum through focussing on the constituent parts that comprise it. The
“how” aspect can be further broken down into how they went about experiencing the curriculum: the act of experiencing it, as well as the indirect object of the curriculum experience, such as their goals and motives for experiencing curriculum. Figure 10 indicates the analytical structure by which awareness of a phenomenon can be deconstructed in phenomenographic analysis. In analysing the data it was helpful to use the tools of this
175
theory of awareness to distinguish the various component parts of the experience and conceptions of curriculum.
The experience of
curriculum
How aspect
What aspect
Act of curriculum Indirect object of curriculum
Direct object of curriculum
Figure 10 The experience of curriculum (adapted from Stamouli and Huggard, 2007, p.183)
The phenomenographic analytical tools which take account of how awareness is developed accommodated the nuances of curriculum such as the hidden curriculum, as well as more obvious features, such as the contents within its ambit of analysis, the participants’ motives, and their goals. In the six graduate interviews clear differences between their structural understandings and their referential understandings of aspects of the curriculum were identified in the course of developing the categories of description in the hierarchically- ordered outcome space. Similarly, the corresponding duality inherent in the employers’
conceptions became evident in the outcome space where their different perceptions were described.
Figure 11 indicates how the phenomenon cannot be seen in isolation. The focus of the study is the way the phenomenon was understood and experienced by the graduates. The aim of the research was to discover this object that was experienced, by analysing the relationship between the graduates and the curriculum, the phenomenon which is being studied (Stamouli & Huggard, 2007).
176
Figure 11 Object of the study in a phenomenographic study (Bowden, 2005)
In order to access the personal reality of the participant and to ensure that the researcher pays attention to the “subtleties of the actual life world”, “bracketing” or epoche is recommended in phenomenographic analysis (Ashworth & Lucas, 1998, p. 421). However, the contrary approach of using existing categories and applying them in different contexts, in different countries, or across different disciplines, has emerged as a strong trend in phenomenography (Crawford et al., 1994; Laurillard, 1993; Marton & Booth, 1997; Prosser
& Millar, 1989; Saljo, 1979).
It was virtually impossible to apply bracketing in an absolutely rigorous way as the completion of a prior literature review, which provided the necessary background for the study, by its very nature, planted the seeds of ideas that are likely to have influenced my approach to the analysis, albeit to a limited extent. I was conscious throughout of the need to develop categories that reflected the empirical-derived descriptions of the experiences and conceptions of the participants. Thus, I have used quotations from the data to support the characterisations I have developed for each “category of description” in the analysis.
177
I chose to first “discover” categories of description from the data, before comparing my outcome space with an Australian study that also related to law students and learning at university (Reid et al., 2006) in Chapter 7. The purpose of contrasting my study with the Australian study was to establish whether there were any points of correlation or dissonance between the two analyses and interpretations, after I had developed my own analysis of my data.
Within each category of description, which together comprises, in effect, “empirically- derived characterisations” of the key aspects of the participants’ experiences of the phenomenon, I have expanded on the range of ways participants within each category experienced the component parts of the phenomenon, so as to develop a sense of the whole and the parts of the phenomenon (Åkerlind, 2002). The purpose of this was to develop a deep, textured interpretation of the various categories which might provide insights for improving curriculum development in legal education. The map of the collective mind will thus have contours and rich topographic detail to add to the understanding of the legal education landscape.