The title of this chapter highlights succinctly the dilemmas faced when attempting to capture the dimensions of qualitative data analysis. For, as Atkinson and Delamont (2005) noted, the field is extensive, and the full range of methodological debates and practical approaches cannot be adequately covered even in an entire book. Inevitably, therefore, a contribution such as this must be partial and incomplete; indeed there may be some who find it superficial, or even inaccurate. The situation is not helped by ‘internal conflicts’ amongst qualitative researchers themselves (Morse 1994), which have resulted in qualitative data analysis becoming a ‘contested site of multiple practices’ (Schwandt 1997), lacking a consensus as to the best approach to adopt (Creswell 1998).

Little wonder then that qualitative analysis has been described as a:

It would therefore appear that neophyte qualitative researchers, hoping to achieve some order from their mass of seemingly ‘chaotic’ data, are faced with potentially even greater chaos if they consult the methodological literature. Clearly, then, this chapter cannot provide any definitive answers, nor do I intend to give a simplified ‘how to’ guide to data analysis, although I will provide some pointers as to where such approaches can be found.

Rather, following the advice of Seale et al (2004), I will adopt a more pragmatic stance and highlight some principles that I have found useful in my own research practice and, paraphrasing their words, this chapter will therefore ‘recount and reflect on my own research experience as well as that of others from whom I have learned’.

In setting the scene, the chapter begins with a very brief overview of what is meant by qualitative research, before providing a potential definition of qualitative analysis. Subsequently, and despite the variability noted above, I will identify a number of commonalities underlying qualitative analysis prior to highlighting some of the existing ‘frameworks’ for analysis.

Denzin and Lincoln (2005b, p. 2) suggested that ‘a complex interconnected family of concepts and assumptions surround the terms qualitative research’, and whilst most authors argue that the primary purpose of qualitative research is to generate theory (Lathlean 2006, Liamputtong 2005 and Morse 1994), agreement is by no means universal. In tracing the evolution of qualitative research over the last several decades, it is suggested that the overall aim has become increasingly more emancipatory and political, so that qualitative research is now seen as ‘a democratic project committed to social justice in an age of uncertainty’, helping people to move from ‘ideas to inquiry, from inquiry to interpretation, and from interpretation to praxis, to action in the world’ (Denzin & Lincoln 2005a).

Even if the focus is limited to inquiry and interpretation, the range of potential analytic strategies remains extensive, and in such circumstances it is essential that the ‘rationale’ for the choice of method adopted is clearly explained (Seale et al 2004). Some degree of informed choice is therefore always necessary and this is, in part, determined by the interests of the researchers involved. As Denzin and Lincoln (2005a, p. xi) noted, ‘the issues and concerns of qualitative researchers in nursing and health care … are decidedly different from those of researchers in cultural anthropology’. Moreover, of the several different methods of qualitative analysis available, ‘some are appropriate for exploring data, others for making comparisons and then for building and testing models, nothing does it all’ (Ryan & Bernard 2000).

With regard to this chapter my choice has been informed by the writing of Ryan and Bernard (2000). They identify two broad ‘traditions’ in qualitative analysis that they term the ‘linguistic’ tradition and the ‘sociological’ tradition. They see qualitative data as comprising mainly text, which itself can take multiple forms:

In ‘making sense’ of such ‘good stuff’ they consider that the ‘linguistic’ tradition sees the text itself as the object of analysis using techniques such as narrative, conversation and discourse analysis. In contrast, the ‘sociological’ tradition views text as an account of, or proxy for, experience. It is this latter approach that will be considered in this chapter.

However, qualitative analysis is not a ‘magical’ process (Morse 1994), and if we are to demystify what it is then some form of definition is required. After considering several definitions, the following by Schwandt (1997) appears to me to be amongst the most helpful:

But analysis is necessary and, even operating within a broad ‘sociological’ tradition, decisions as to the approach to use are still required. Miller and Crabtree (1992) likened this process to taking a photograph in which it is necessary to decide which camera to use, which scene to focus on, and which filter to select. Using this analogy the camera would relate to the broad methodological approach, the scene to the object of inquiry or study, and the filter to the theoretical orientation that is applied.

Compared to quantitative approaches, where data are structured and strategies for analysis are often predetermined, qualitative analysis is more flexible (Donovan & Sander 2005) and is neither linear nor predictable (Liamputtong & Ezzy 2005). Furthermore, whereas in quantitative research data collection precedes analysis, the relationship is different in qualitative research. Data collection and analysis proceed in an iterative fashion (Ritchie 2003, Donovan 2005 and Lathlean 2006), each informing the other. Therefore, data analysis begins early and guides subsequent data collection (Lathlean 2006).

For example, in my own early research I was interested in the experiences of carers who supported a family member, and who used respite care services (Nolan 1990, Nolan 1992a and Nolan 1992b. Virtually all of the literature at the time described caring as burdensome or difficult and so the focus of my early questioning was about the extent to which respite care provided relief from such burden. However, initial analysis of the data suggested that, whilst caring was often difficult, it was by no means always a problem, and many carers also gained considerable satisfaction from their role. As a result of this early analysis, the focus of data collection, and indeed the study as a whole, broadened to consider the potential rewards and satisfactions of carers and the impact of respite care on these. Further analysis of the data identified differing sources of satisfaction, and also highlighted the fact that carers’ acceptance of respite care hinged critically on the extent to which the cared-for-person had a positive respite experience. As a result, we proposed a ‘mid-range’ theory that provided practical ways in which respite care could be made more acceptable to both the carer and the cared-for-person (see Nolan & Grant 1992b and later in this chapter).

Qualitative data analysis is therefore intimately linked to data collection and involves both cognitive processes and the application of varying structured techniques. However, several authors identify a number of shared goals, although the terms they use differ. Some of the more commonly cited approaches to analysis are summarised in Table 33.1.

Ritchie et al (2003) argued that two common processes underlie all qualitative analysis:

Descriptive accounts are closer to the data and would be readily recognised by participants in the research. Later these descriptive accounts may be ‘classified’ into more abstract ideas that retain the original meaning of the data but use more complex language. More sophisticated explanatory accounts seek to find links and connections between two or more phenomena, in order to generate theory.

In taking stock of the various arguments regarding the sequencing and purpose of qualitative analysis, it seems to me that these might be summarised using an ‘alliteration’ of C’s. These are briefly outlined in Table 33.2.

This format captures, hopefully in an easily remembered fashion, the way in which data analysis moves from the ‘raw data’ and generates ever more abstract ideas that are however still connected with, and are traceable back to, the data. Depending on their purpose, studies may stop at any point along this continuum. So, for example, using Ritchie et al’s (2003) suggestions, studies that are primarily descriptive and exploratory might go no further than the creation of concepts. At a slightly higher level of abstraction, studies that wished to ‘classify’ concepts would aim for the creation of categories. Researchers who wanted to provide more sophisticated explanatory accounts would need to find connections and draw conclusions from their work. Of course, in qualitative work, ‘conclusion’ does not signal some ultimate or final result but rather indicates that this is the furthest point it is possible to teach with the data available.

Once again an example from the respite care study mentioned earlier might help to illustrate how this process works. One major component of this study was to understand the respite care experience from the older person’s perspective, as initial data analysis had clearly indicated that carers got most benefit from respite care when the older person had a positive experience. The service in question worked on a rota basis in which the older person spent two weeks in hospital, followed by six weeks at home, with this cycle being repeated over a number of months, or even years. It therefore seemed very important to find out why some older people seemed to enjoy their stay, whereas others found it extremely distressing. In order to appreciate this, interviews were conducted with the older people themselves, and extensive periods of observation were undertaken in the hospitals concerned. Staff and carers were also interviewed about their views on respite care. This generated masses of data, detailed analysis of which identified a ‘typology’ of differing types of respite user in which three main groups, one with a number of subgroups, were identified (Nolan & Grant 1992b).

The three main groups were the: