CHAPTER FOUR METHODOLOGY
4.7. Data analysis
The phase of data fragmentation followed a systematic approach involving quantitative content analysis and qualitative narrative analysis. Once the complete set of 371
photographs from the 29 participants' data had been collected, systematically coded and stored in photograph albums, the complete data set was coded according to principles of quantitative content analysis as an accepted method of textual investigation (Krippendorf,
1980; Silverman, 2001). These principles are first, that sets of precise categories are established for the data and second, instances that fall into each category are counted (Silverman, 2001). The precision of the category definitions is considered a key feature of reliability and the precision of the counting process necessary for reliability
(Silverman, 2001). These issues were addressed by setting up specific criteria for each category and using a co-researcher to verify the counting process. Each photograph was coded according to (a) whether the participant was in the photograph or not; (b) thematic categories; (c) categories of settings and; (d) number of persons in each photograph according to race and gender. Where the themes were tentative, the interview data was engaged in establishing what the photograph was about. For example a photograph of a basketball court could be coded as 'sports participation' with knowledge from the interview that the participant had been playing basketball at the time. This process was less cumbersome than it might sound and resulted in a useful tabulation of the complete visual data set on a Microsoft Excel spreadsheet. Basic counting techniques and
descriptive statistics were then be applied to the data and presented graphically. The thematic categories were then compared across four groups defined as School A Black, School B Black, School A White and School A Indian. As there was much evidence in the verbal data that there were multiple subject positions in relation to racial identities, it is noted that comparison between these artificial racial 'groups' represented a limited view of the more complex cultural processes identified in the qualitative analysis.
(a) Participant as
Subject
• In photo
• Outside photo
(b)
Emerging thematic categories
• Explicit in photo
• Informed by interview
(c) Categorising
settings
• Explicit in photo
• Informed by interview
(d) Subject count
• Race & gender
• Background subjects coded 'B' and ignored
Figure Content analysis categories
As noted by Silverman (2001), although categorisation allows one to make sense of people and events it can serve to maintain a static view of identities. The second part of the fragmentation phase related to the narrative analysis of the verbal data. The qualitative narrative analysis of the verbal data as a process that was integrated and triangulated with the quantitative content analysis of the visual data. The content analysis helped to pose the questions that could be asked of the interview texts by identifying themes, microcultural settings and other participant concerns. The interview texts were used in clarifying categories or counted instances in the content analysis.
The language of narrative analysis may appear very technical at first, and the reader is guided to the glossary for detailed definitions of terminology. It is argued that once the process and its terminology are understood, the narrative method used provides a unique perspective on interview data that is complementary with the theory basis of this
research. The qualitative analysis focused on embedded narratives, defined as the diffuse stories which occur within the interview dialogue (Toolan, 2001; Mishler, 1986). An embedded conversation may be a collaboration of a principal story teller with multi-party conversational glossing which serves to amplify or clarify aspects of the story, as might occur in a focus group, or it may be occur in the interviewer-interviewee dialogue (Toolan, 2001). The narrative analysis was based partly on a study by McQueen and Henwood (2002) which used Labov's model of narrative categories to identify and organise embedded narratives. Labov's model is based on the hypothesis that there is a fundamental narrative structure that may be found in the ordinary narratives of personal experience (Mislhler, 1986). Labov suggested that stories are organised in a general pattern involving a 'headline' or orientation, an abstract which provides a sense of the what the story is about; a complicating action in which pivotal actions and experiences occur, an resolution or ending of the action and an evaluation in which the meanings for the action are given (McQueen & Henwood, 2002; Toolan, 2001). Embedded narratives may not always fit neatly Labov's categories of narrative action, however the organising structure provides a useful starting point from which to analyse the text (Toolan, 2001;
McQueen & Henwood, 2002).
With core narratives extracted from the interview dialogue and assembled according to Labov's categories, a process of data integration could occur (Ulin et al., 2002). The texts were read for tropes - the repeated figures of speech or expressions used by
speakers. The identification of tropes then allowed for recurring core narrative themes to be identified with careful attention to naming themes in a way which matched the words used by the participants (McQueen and Henwood; 2002; Silverman, 2001; Ulin et al., 2002). For example, one of the participants used the tropes 'show' or 'show off in his core narratives about boys behaviour with girls at nightclubs. From this trope was identified a main narrative theme of'displayed' masculinity.
Following Toolan (2001), narratives were then analysed for the position of the narrative voice or diegesis. As the narratives were framed as 'stories' told by the interview participant to the interviewer, they tended to have the quality of diegesis rather than mimesis, that is of an externally reported story rather than an immediate description. The terms intradiegesis, extradiegesis, heterodiegesis and homodiegesis were used to identify the narrative voice along with whether the pronoun used was in the first, second or third person or in the plural or singular form (Toolan, 2001). Intradiegesis and extradiegesis refer to whether the narrative voice was positioned within or outside the events of the story respectively. Homodiegesis and heterodiegesis refer to whether the narrative voice remained the same throughout the embedded narrative or whether there were diegetic shifts in the narrative voice. To return to the example above, the same participant related his story about boys at nightclubs heterodiegetically, that is he shifted from the position of an intradiegetic observer at one of the nights out in the abstract and action of the story to that of a removed, extradiegetic commentator on the boys' behaviours in the resolution and evaluation parts of the story (see Appendix N9 for the full transcript and its analysis).
Following Toolan (2001) and McQueen and Henwood (2002), attention was given to the orientational words of each narrative or its deixis and the spatiotemporality of each unit or unit cluster was considered with particular attention to chronologically retrospective or prospective accounts. The term 'analepsis' was used to describe events that occurred chronologically before the story and 'prolepsis' to in order to establish perspectival
'pastness' or 'futurity' of the narrative. Full definitions of these terms may be found in the glossary. Findings were organised with illustrative data and schematic
representations to depict the links between core narratives, tropes and the extracted main narrative themes as in the study by McQueen & Henwood (2002).
Immersion Fragmentation Integration
INTERVIEWER PROCESS
SUPERVISION DEBRIEFING
INTERVIEW TRANSCRIPTS
CONTENT ANALYSIS
TROPES MAIN THEMES
EMBEDDED NARRATIVES
CORE NARRATIVES
DIEGESIS DEIXSIS
SCHEMATIC SUMMARY
Reading Re-reading Verifying
Figure 2 Schematic representation of data analysis method
The schematic representation of the data analysis method (Figure 2) depicts the iterative, overlapping and non-linear processes of extracting and analyzing narratives from the verbal data. The reading, re-reading and verification of the data correspond to processes of data immersion, fragmentation and integration - the process suggested by Ulin, Robinson, Tolley, and McNeill (2002) for qualitative data analysis. At the reading stage, there were three interactive aspects: (a) the self-reflexive interview process; (b) peer debriefing and interviewer supervision and (c) the initial transcription of narratives. The reading stage produced immersion in the data from which were derived questions for purposive sampling and further analysis. During the re-reading phase, the 'conversation' among three processes produced a more focused reading of the 'fragmented' data. These processes were: (a) the content analysis of the complete visual data set; (b) identification of recurring words (tropes) and main themes; (c) embedded narratives extracted from the
transcripts and audio recordings. Integration of the data occurred through the verification of findings through a systematic analysis and consolidation of the embedded narratives by means of: (a) identifying core narratives; (b) examination of deixical and diegetical aspects of the text (spatiotemporal context and narrative voice) and (c) representing data in a schematic summary. Note that Appendix 1 contains examples of this data integration phase.