KO H8

 8 Moving towards analysis The scope of qualitative analysis OVERVIEW Introducing qualitative a...

 8 Moving towards analysis The scope of qualitative analysis OVERVIEW Introducing qualitative analytic methods suitable for beginners A flexible foundational method: thematic analysis An experiential and interpretative approach: interpretative phenomenological analysis An inductive yet theorised approach: grounded theory An approach for looking at what language does: discourse analysis By the time you have finished transcribing your data, you’ll be very familiar with it and may have started to note down some initial analytic ideas – this is one reason transcrip- tion is often described as part of the analytic process, part of the process of familiarising yourself with your data. But once you’re done transcribing, the analysis really begins. Before you jump in and get started with the exciting task of data analysis, you need to decide what method you’re going to use to analyse them (though you should already have a sense of that when designing a project, before collecting data). This chapter introduces seven key methods of analysis, four of which (the pattern-based methods – thematic analysis [TA], interpretative phenomenological analysis [IPA], grounded theory [GT] and pattern-based discourse analysis [DA]) we discuss further in this book. THE SCOPE OF QUALITATIVE ANALYSIS Qualitative analysis covers a spectrum from descriptive (and exploratory) through to more interrogative, theorised, interpretative analysis; this spectrum closely aligns with the experiential/critical orientation outlined in Chapter 2. Descriptive work aims to 08-Braun & Clarke_Ch-08.indd 173 27/02/2013 3:55:55 PM  174 Successfully analysing qualitative data ‘give voice’ to a topic or a group of people, particularly those we know little about. For instance, US psychologists Marianna Litovich and Regina Langhout (2004) explored experiences of heterosexism in an interview study with five lesbian families. They reported that the parents often claimed that their families had not been victims of hetero- sexism, but nonetheless discussed multiple incidents of heterosexism and prepared chil- dren for heterosexism in a number of ways. The authors make sense of this discrepancy in their data by suggesting that lesbian parents deal with heterosexist incidents in ways that minimise the negative repercussions for their children. Qualitative analysis which is interpretative aims to go further than descriptive analysis, unpicking the accounts that are given, and asking questions like ‘What’s going on here?’ and ‘How can we make sense of these accounts’? It tries to gain a deeper understanding of the data that have been gathered, and often looks ‘beneath the surface’ of the data, as it were, to try to understand how and why the particular accounts were generated and to provide a conceptual account of the data, and/or some sort of theorising around this. For example, Victoria took this approach when analysing data from interviews with lesbian and gay (LG) parents about their children’s experiences of homophobic bullying (Clarke et al., 2004). Just as in the Litovich and Langhout (2004) study, parents either reported that their children hadn’t been bullied or they minimised the extent and impact of any bullying their children had experi- enced. Instead of treating these as factual accounts of children’s realities, she asked what effects these types of accounts have, in talk about LG parenting. Taking a language practice approach to qualitative data, Victoria made sense of this talk within a socio-political context relatively hostile to LG parents, where the possibility of homophobic bullying is used to support arguments against LG parenting. The analysis showed that parents’ accounts were designed to protect LG parents from outside critique. Such interpretative accounts go well beyond what is obvious in the data, to explore meaning at a much deeper lever. INTRODUCING QUALITATIVE ANALYTIC METHODS SUITABLE FOR BEGINNERS There are many different methods of qualitative data analysis, but some are more com- mon in qualitative psychology than others, and some are easier to learn and do than others. We focus on describing and demonstrating those that are both common within psychology and relatively accessible to those new to qualitative research – TA, IPA, GT and pattern-based DA. Table 8.1 provides a quick overview for comparison of these pattern-based methods. Three other methods that require more advanced skills – discursive psychology (DP), conversation analysis (CA), narrative analysis – are briefly described by expert practitioners of these methods, to give you a sense of the wider scope and diversity of qualitative analysis (see Boxes 8.1–8.3). A FLEXIBLE FOUNDATIONAL METHOD: THEMATIC ANALYSIS Some sort of ‘thematic’ coding is common across many qualitative methods within the social sciences. TA as a named approach was first developed by Gerald Holton, a 08-Braun & Clarke_Ch-08.indd 174 27/02/2013 3:55:55 PM  Table 8.1 (Continued) Basic method What is it? Varieties? What is it? 08-Braun & Clarke_Ch-08.indd 176 Grounded Theory Developed by the US sociologists GT-lite Aims to generate a taxonomy of categories (clusters (GT) Glaser and Strauss in the 1960s (Glaser of related codes; similar to themes in GT and IPA) & Strauss, 1967) and has evolved from data, with some indication of the relationships considerably since then, with many between concepts and the relative importance of different varieties of GT on offer; focuses concepts to the research question (see Pidgeon on building theory from data and, & Henwood, 1997); leaving aside questions of because of its sociological origins, there epistemology, the outcome of GT-lite is very similar is an emphasis on understanding social to that of thematic analysis and IPA (a set of themes processes; analysis is organised around or categories that fit together in various ways) key categories (similar to themes) (see (Full) GT There is some debate as to what constitutes a ‘full’ Pidgeon & Henwood, 1997; Charmaz, GT, but generally it aims to build a theory from the 2006) data (Glaser & Strauss, 1967); theoretical sampling is used, saturation is achieved and similar concepts are grouped together into categories and used to generate a theory Positivist GT Aims to represent reality (see Glaser, 1978, 1992) Contextualist Acknowledges the role of the researcher in shaping (constructivist) GT the analysis, views meaning as contextual, and argues that it is not possible to generate one ‘true’ reading of data (see Charmaz, 2006) (Radical) Similar to DA-lite and constructionist TA; pays constructionist GT closer attention to language use; emphasis on the discourses that shape accounts the discursive devices through which participants construct and give meaning to their lives; and acknowledges ambiguity and inconsistency (see Madill et al., 2000) 27/02/2013 3:55:55 PM  184 Successfully analysing qualitative data AN INDUCTIVE YET THEORISED APPROACH: GROUNDED THEORY GT is a very popular qualitative method, particularly in the US, and has the longest history of the methods we discuss. Developed in the 1960s by the US sociologists Barney G. Glaser and Anselm L. Strauss in a ground-breaking study on dying in hos- pital (Glaser & Strauss, 1965), the principles of grounded theory were first outlined in their classic book The Discovery of Grounded Theory (Glaser & Strauss, 1967). Glaser and Strauss came from different theoretical backgrounds: Glaser was interested in the standard hypothetical-deductive method, Strauss in the work of symbolic interac- tionists, who were concerned with the role of interpretation and meaning in the social world – they argued that meaning derives from social interaction (Blumer, 1969). What united them was a critique of sociologists’ preoccupation with testing ‘grand’ theories, rather than generating more contextually situated theories that were relevant to the lives of the people being studied. Glaser and Strauss chose the term grounded theory to capture the idea of a theory that is grounded in a close inspection of qual- itative data gathered from concrete, local settings (Pidgeon & Henwood, 1997a). In essence, GT is an approach to qualitative research (not just an analysis method), concerned with constructing theory from data (Charmaz & Henwood, 2008). GT has been used to research a wide range of topics, such as chronic illness (see Illustrative Research Example 8.3), the phenomenon of ‘pro-anorexia’ (Williams & Reid, 2007) and the impact of wind turbines on people living in their locality (Pedersen, Hallberg, & Waye, 2007). GT has become hugely popular across the social sciences. Many different versions have been developed, with no one predominating. The original proponents publi- cally split and developed their own versions of GT: Glaser’s version is more positivist (Glaser, 1992, 1978); Strauss’s, particularly that developed with Juliet Corbin (e.g. in the hugely influential text Basics of Qualitative Research, Corbin & Strauss, 1990), is more contextualist (constructivist) in theoretical orientation. There are now ver- sions of grounded theory situated across the epistemological spectrum: positivist (Glaser, 1992; Glaser, 1978); contextualist (Charmaz & Henwood, 2008; Pidgeon & Henwood, 1997a); and constructionist (Madill et al., 2000). We primarily draw on the version developed by British social psychologists Karen Henwood and Nick Pidgeon (Henwood & Pidgeon, 1994, 2003, 2006; Pidgeon, 1996; Pidgeon & Henwood, 1996, 1997a, 2004), whose work is heavily influenced by US sociologist Kathy Charmaz (2006), former PhD student of Strauss, and one of the world’s foremost proponents of GT, and Birks and Mills’s (2011) very practical and accessible book – which syn- thesises the work of the original proponents as well as ‘second generation’ grounded theorists like Charmaz. Henwood and Pidgeon situate their approach within a, broadly speaking, contextualist framework; Charmaz (2006) locates hers within the symbolic interactionist tradition in sociology, and writes about ‘constructing grounded theory’, rather than discovering theory in data (as per Glaser and Strauss’s, 1967, original model). She argues that researchers are part of the things we explore and the data we generate: ‘we construct our grounded 08-Braun & Clarke_Ch-08.indd 184 27/02/2013 3:55:56 PM  Moving towards analysis 185 ILLUSTRATIVE RESEARCH EXAMPLE 8.3 Grounded theory: Chronic illness US sociologist Kathy Charmaz (1983) examined the loss of self experienced by chronically ill adults using a symbolic interactionist-informed version of GT, and drawing on 73 in- depth interviews with 57 chronically ill people living in Northern California who had various diagnoses including cancer, diabetes and multiple sclerosis. Symbolic interactionism views the self as fundamentally social – the self is developed and maintained through social relationships – so that changes in the self-concept occur across the lifespan. Applied to the study of chronic illness, this perspective highlights how the illness experience creates situations in which people learn new understandings of self and relinquish old ones. The participants ranged in age from 20 to 86 years (most were aged 40–60); two-thirds were women. The data analysed in the paper were limited to participants who were housebound or severely debilitated by their illnesses. Charmaz identified four types of suffering experienced by these participants: leading restricted lives, experiencing social isolation, being discredited and burdening others, each of which led to a loss of self. Each type of suffering was discussed in turn; analytic claims were illustrated with interview extracts. For example, living a restricted life demonstrated that unlike ‘healthy’ adults, these chronically ill people lived narrow, restricted lives, which focused on their illness and offered few possibilities for creating valued selves. Charmaz argued that ‘values of independence and individualism’ (p. 172) intensified the experience of restriction: chronically ill people could not do the things (to the same extent) they enjoyed and valued before they became ill. If participants felt they had the choice and some freedom to pursue valued activities (even if they could not always exercise that freedom), they felt less restricted and maintained a valued self-concept. For others, the restrictions imposed by treatment regimes became ‘daily reminders of the lessened freedom, and often, diminished self, that these patients experience[d]’ (p. 173). Participants experienced a loss of control over themselves and their lives, which led to losses of self, with their past lives and selves seeming increasingly distant. Charmaz argued that the practices of health professionals sometimes reinforced restricted lives when patients were not supported in reducing their suffering (and consequent loss of self), and that a world designed for the healthy and able often led to unnecessary restriction (partly through the failure of the chronically ill to question social norms and through using the lives of the ‘healthy’ as the yardstick by which they judge themselves). The unpredictability of some forms of chronic illness also led some participants to voluntarily restrict their lives more than was perhaps necessary. For the chronically ill who were increasingly socially isolated, their restricted life ‘foster[ed] an all-consuming retreat into illness’ (p. 175). However, for participants whose health had improved, their past illness became a path to self-discovery and self-development – the freedom from ordinary existence led them to reflect on who they were and who they wanted to be. Charmaz concluded by noting that the language of suffering was a language of loss – the loss of self and a meaningful life (see Chapter 11 for a comparison with an IPA study on a similar topic). 08-Braun & Clarke_Ch-08.indd 185 27/02/2013 3:55:56 PM  186 Successfully analysing qualitative data theories through our past and present involvements and interactions with people, perspec- tives, and research practices’ (Charmaz, 2006: 10, emphasis in original). In this respect, Charmaz’s version of GT is rather similar to the emphasis on the researcher’s interpre- tative activity in IPA (which also draws on symbolic interactionism; see Smith, 1999b). Although GT is used to address a number of different types of research questions, because of its concerns with social and social psychological processes within particular social settings (Charmaz, 2006), it is probably best suited to questions about influencing factors and the (social) processes that underpin a particular phenomenon. The interview is a key method of data collection (as it is for IPA). Other possible methods include partici- pant-generated texts and secondary sources (Charmaz, 2006) (see Table 3.3 in Chapter 3). The production of a ‘full’ GT is a demanding process, and only possible in larger research projects (not constrained by time and resource pressures); there is also some debate among grounded theorists as to what constitutes a ‘grounded theory’ (Pidgeon & Henwood, 1997a). In practice, many researchers only complete the earlier stages of GT (initial coding and con- cept development), which we refer to as ‘GT-lite’. This ‘lite’ version of GT is ideally suited to smaller qualitative projects, and so this is the version we focus on in this book. Table 8.4 provides an evaluation of GT’s strengths and weaknesses. GT is famous for distinctive procedures such as ‘line-by-line coding’, ‘constant comparative analysis’, ‘memo writing’, ‘theoretical sampling’, ‘saturation’, and for not engaging with the relevant literature prior to beginning the analysis (to avoid it being shaped by preconceptions from existing research, rather than being truly grounded in the data). Some of these, such as theoretical sampling (see Box 3.3 in Chapter 3), are specific to full (rather than lite) GT; those relevant to GT-lite are discussed through Table 8.4 Evaluating grounded theory Strengths Weaknesses Different versions of GT to suit different There are so many versions of GT, and so many theoretical and epistemological different sets of guidance for doing GT, not to frameworks mention different terminology, that it can be difficult A useful method for researchers to know where to start (Birks & Mills, 2011) interested in social and social Some versions of GT procedures are bafflingly psychological processes (rather than complex individual experience) GT was developed in sociology and so emphasises Some clear and comprehensive accounts sociological concerns, such as social structures and of analytic procedures (some are less processes, rather than psychological ones clear – see weaknesses) Completing a full GT is highly demanding and time Many GT procedures such as line-by-line consuming; in practice, many grounded theorists use coding and memo writing are useful in GT-lite almost any kind of qualitative analysis Difficult or impossible to complete a ‘full’ GT in a small project 08-Braun & Clarke_Ch-08.indd 186 27/02/2013 3:55:56 PM  Moving towards analysis 187 Chapters 9–11. Some of these procedures have assumed a status akin to Christianity’s ten commandments: ‘you must generate a code for every line of data’; ‘you mustn’t read any literature until after you have completed your analysis’. Such prescriptive and purist ideas are problematic, and set you up for a sense of ‘failure’. In practice, it’s virtually impossible not to have engaged with some of the relevant literature prior to beginning your research – research proposals, funding and ethics applications all require that we situate our proposed research in relation to the relevant literature; and you certainly can’t do research without at least one of these (ethics). Moreover, as scholars we continually accumulate a wide variety of disciplinary knowledge, which is impossible to ‘un-know’. And to do research that isn’t a waste of time, we need to have some sense of whether our question hasn’t already been answered. A good maxim to apply is that ‘there is a difference between an open mind and an empty head’ (Dey, 1999: 251); we can still strive to approach our (GT, or indeed any) research with an open-mind, even though we have some prior knowledge about our topic. AN APPROACH FOR LOOKING AT WHAT LANGUAGE DOES: DISCOURSE ANALYSIS DA is another immensely popular method, but ‘discourse analysis’ is also one of the most confusing terms in qualitative research. It’s used in so many different ways, just within psychology, that one ‘DA’ can bear little correspondence at all to another ‘DA’. If you look outside the discipline, such as into English, linguistics or media studies, it can get even less familiar. British psychologist Nigel Edley described DA as ‘an ever- broadening church, an umbrella term for a wide variety of different analytic principles and practices’ (Edley, 2001a: 189). We’re going to try to narrow this down a bit, and explain some core concepts, but our main focus is on versions of DA that look at pat- terns (of meaning or language practice) across linguistic datasets (see Parker & the Bolton Discourse Network, 1999, on non-linguistic data). DA is not a method (like TA); it is not even an approach to qualitative research (like IPA or GT). Rather, it is a whole approach to psychology and knowledge. It emerged in (British) social psychology in the 1980s, which was in a state of epistemological and ontological upheaval, following the ‘crisis in social psychology’. This ‘crisis’ saw diverse paradigmatic, political and conceptual challenges to what had become the dominant framework and form of social psychology: a positivist cognitivist science using labora- tory-based experimentation as its primary method of inquiry, married to an impoverished and diminished vision of ‘the social’ (see Parker, 1989). Now-classic texts, Changing the Subject: Psychology, Social Regulation and Subjectivity (Henriques, Hollway, Urwin, Venn, & Walkerdine, 1984), Discourse and Social Psychology: Beyond Attitudes and Behaviour (Potter & Wetherell, 1987) and Arguing and Thinking: A Rhetorical Approach to Social Psychology (Billig, 1987), took a knife to the foundations of (social) psychology, offering up a whole different way of seeing and doing psychology: different theoretical frameworks, methodologies and analytic concepts. Rather than locating psychology as 08-Braun & Clarke_Ch-08.indd 187 27/02/2013 3:55:56 PM