• No results found

Chapter 4. Making data: study design and tools

4.7. Method of analysis

Stage five of the Kvale guidance on conducting an interview investigation is that of data analysis. Given that the data was made out of both descriptive quantitative information as well as exploratory qualitative information, the analysis had to both crunch the numbers as well as work with words (O‘Leary 2005).

4.7.1. Quantitative analysis: crunching the numbers

The purpose of the descriptive analysis was to collate the individual accounts of the narrative to capture and describe programme processes. Nominal and ordinal data was collected within the closed questions and response scales and therefore descriptive

analysis was used. Although non-parametric inferential statistics would allow for correlational or comparative analysis, relationships between constructs were not sought as this was not the purpose of the research.

The EXCEL spreadsheet containing the nominal and ordinal data was manipulated to produce descriptive statistics. Frequency tables were constructed as a way of presenting the summarised data to show the patterns of responses to the closed ended items and response scales (Appendix E).

4.7.2. Qualitative analysis: working with words

The purpose of the qualitative analysis was to explore individuals‘ assumptions about mentoring and what being involved meant to them. The interpretive motive behind this research was to make meaning from participants‘ thoughts, knowledge, attitudes and feelings, allowing research findings to emerge through the development of frequent or dominant themes within the interview data.

There is a wide range of literature that documents the underlying assumptions and procedures associated with analysing qualitative data, although as Creswell (1998) highlights, there is no consensus over methods of analysis. There are, however, common features. Morse and Richards (2002) identifies the key features of good qualitative analysis as synthesising, comprehending, theorising and re-contextualising: these strategies were used in this research to make sense and meaning from a complex, sometimes contradictory collection of individual experiences.

Seidman (1991) considers two distinct paths for analysing interviews - developing profiles and developing themes. Returning to the intent, motivation and expectations of this research the theme approach to analysis was determined most appropriate; it was envisaged that the themes emergent from the whole sample would provide generalisable learning for the CSLA and others with a policy or scholarly interest in school leadership and/or employer-led mentoring. Although profile analysis was not undertaken here, the individual transcripts and the mentor dyad information were available if additional context was needed to inform the thematic analysis.

The initial thematic frame was learning around McClellan et al.‘s (2008) first dimension of mentoring – to understand the process of mentoring in the CSLA. This theme involved both process and understanding of the interpretation and implementation of the mentoring policy. The initial topic coding of text was related to the process questions

asked and considered: place and time; what was important in the first meeting; the expectations of both mentors and mentees on purpose and process; the level of direction involved in the mentoring approach used; the documentation used to support the process and how the relationship evolved over time.

Data analysis to explore McClellan et al.‘s (2008) second dimension of mentoring required a number of stages as the interview schedule had deliberately not asked directly about the claims made over the anticipated outcome of mentoring but asked for commentary about experiences in broad areas. In order to move from data to abstraction, the transcripts were interrogated firstly by using first broad topic coding, then shaped through analytic coding, then conceptualised through a thematic frame. I was aware of the debate around data-driven or concept driven coding (Gibbs 2007) and at times became frustrated with the contested ground around open coding, axial coding and selective coding (Morse and Richards 2002). Although I was wary of ‗forcing data‘ - the criticism levelled by Glaser (1992) over the selective coding methods of Strauss and Corbins, I determined that analysis most coherent with the intent, motivation and expectation of the research would be concept driven, using a thematic frame as advocated by Ritchie et al. (2003).

A thematic frame was constructed as advocated Ritchie et al. 2003. Lists of thematic ideas, based around the research question, were taken from the literature and the knowledge, understandings and beliefs the researcher. The thematic frame altered as the results were interpreted as participants told stories which touched on a number of aspects, or reiterated themes throughout the interview which would move back and forth around one issue of importance to them.

The synthesising stage of this analysis required working with the words in the transcripts, categorising the content of the text, clustering natural units of meaning (Miles and Huberman 1994) and linking sections of text with thematic ideas (Gibbs 2007). Each transcript was fragmented in this way through a process of manual topic coding, putting the passages from each interview which exemplified the same idea, explanation, activity or phenomenon, together. A useful description of topic coding was offered in Morse and Richards (2002). Topic coding was needed as the first stage towards abstraction, moving the data from a series of responses to a set of questions towards a focus for thinking about the concepts that could help understand them.

The second stage was beginning to conceptualise and make meaning from the data through analytic coding. This was not, as suggested above, a strictly linear process; as

the data was topic coded from each transcript as it was completed, the development of the analytic codes shaped the topic codes from later interviews. What began as a series of collated responses to the questions within the interview schedule, based around the research question, was then refined into categories through a process of analytic coding, shaped into patterns and themes, as concepts, new meanings and some surprises, emerged.

4.8. Methodological quality; issues around validity and