Understanding delayed access to antenatal care: a qualitative study Rosalind Haddrill

3.2.3 Data analysis

There are many different qualitative data analysis methods, influenced by theoretical and methodological perspectives and the aims of the research, though fundamentally all qualitative data analysis is the search for general statements about relationships and underlying themes (Liamputtong and Ezzy 2005, Marshall and Rossman 2011). It is a process of “discovering patterns, themes and categories in one’s data” (Patton 2002: 453), bringing together “components or fragments of ideas or experiences, which often are meaningless when viewed alone” (Leininger 1985: 60). It seeks to generate and develop analytic categories and theoretical explanations, to describe and explain social phenomena, often from extensive data. It does not aim to quantify data or to present it as if it is statistically representative; as such, methods using counts or frequencies (such as content analysis) are inappropriate (Pope, Ziebland and Mays 2006).
As discussed earlier in the chapter, the inductive approach of grounded theory has been highly influential in this study. The process of coding and data analysis in grounded theory shares many of the characteristics of other qualitative methods, in terms of its search for themes and patterns across the entire data set. Green and Thorogood (2014) argue that there are many elements of the grounded theory approach that are useful for any analysis. However, the ultimate goal of a grounded theory analysis is to generate and/or develop a substantive theory (Polit and Beck 2012). This process of conceptualisation and abstraction of participants’ perceptions and actions moves beyond description, aiming to discover the ‘core category’ which links all categories or concepts identified in the data, arguably an objectivist approach (Bryman 2008, Holloway and Wheeler 2010). As previously mentioned, this may be achieved to a greater or lesser extent, and there is significant debate about the value of ‘modified grounded theory’ or what some have called grounded theory ‘lite’ (Braun and Clarke 2006). The aim of the study was not explicitly to generate theory, but rather to examine the full range of women’s views; presenting an inductive exploration of themes in relation to late booking. This reflects the pragmatic roots of the study, which weren’t aligned to any single pre-existing theoretical framework, and the likely diversity of perspectives, based on initial literature searches. Consequently, a decision was made to choose a method of analysis appropriate to the broad aspirations of the study.
Thematic analysis is a method which some have argued lacks a clear identity, but which is widely used in qualitative research. It is a process for encoding qualitative information, identifying common threads and facilitating the evolution of ‘simple’ interview data into more sophisticated codes, and subsequently into concepts and explanations that link the interviews together (Morse and Field 1995, Mays and Pope 2000). Some authors have argued that thematic analysis is not a distinct method of analysis, as the search for themes is an activity common to many if not all approaches to qualitative data analysis (Bryman 2008). Its use crosses different qualitative methods without being specifically described, and it has been referred to by many different names, such as ‘inductive analysis’ (Boyatzis 1998, Patton 2002). Other authors propose that thematic analysis should be considered as a method in its own right, as it has many advantages. For example, Braun and Clarke (2006) argue that it is an accessible and theoretically flexible approach to analysing qualitative data from different sources, in essence a foundational method for many forms of qualitative analysis, which allows for different interpretations of the data. Thematic analysis can summarise the key features of a large body of data and create results which are accessible and useful for informing policy development. It can be used to develop taxonomies or classifications, or to develop models or diagrams that express the connections between themes. The choice of thematic analysis reflects the study’s aim to articulate and understand these connections.
Thematic analysis has been presented as the simplest form of qualitative analysis used in healthcare research and the most inductive, although the reality is a mixture of both inductive and deductive methods, the latter particularly in the final stages of the analysis (Patton 2002, Liamputtong and Ezzy 2005, Pope, Ziebland and Mays 2006). It is a method which is highly appropriate for both the constructivist and interpretivist approaches which underpin this study. Thematic analysis mirrors many of the techniques of grounded theory: both seek to allow themes and explanations to develop from the data itself and to avoid the imposition of an explanatory framework by the researcher. Crawford, Brown and Majomi (2008) see this inductive approach being a key strength, in that the analysis is firmly anchored in the data and able to challenge existing theoretical presuppositions. Reflecting this, the analytical process is cyclical and iterative rather than sequential, and forms part of the process of constant comparison. This method results in the interweaving of theoretical sampling, data collection and the testing of emerging ideas as the study progresses, enabling the development of themes from the data itself (Murphy et al 1998, Liamputtong and Ezzy 2005, Pope et al 2006).
Thematic analysis takes a more inductive approach than other methods, such as the matrix-based framework analysis (Ritchie and Spencer 1994). This is described as ‘a content analysis method which involves summarising and classifying data within a thematic framework’ (Green and Thorogood 2014: 218) and is widely used in health research. Framework analysis is a systematic and transparent process of organising qualitative data, with many strengths. Although partially ‘grounded’ in the original accounts of the people studied, it usually starts deductively from the aims and objectives of the study itself and a priori reasoning, reflecting, in part, its large-scale policy research heritage (Pope et al 2000, 2006, Gale et al 2013). Though Ritchie and Spencer argue that the framework approach is not a mechanical process and that dynamism and interpretive thinking are key features, critics have argued that this deductive focus has the potential to remove several key aspects of traditional qualitative analysis. This includes the induction of themes, flexibility in design and sampling, and an iterative approach, with the potential to constrain what is studied and suppress complexity (Donovan and Sanders 2005, Lacey and Luff 2007). Whilst the aims of the study and the prior knowledge of the research team have been influential on the study design, the process of data analysis has been conducted independently of this, and as such thematic analysis was considered more appropriate for this inductive approach.
Whilst a flexible method, the process of thematic analysis needs clear and concise guidelines (not simply ‘anything goes’), a clear statement of purpose and a systematic approach, to guide the whole process and maximise credibility and rigour (Wolcott 2009). Braun and Clarke (2006) argue that those undertaking thematic analysis need to make their assumptions explicit: what they are doing, why and how; theoretical and methodological assumptions need to be explicit too. The process aims to avoid a superficial analysis where data is simply organised rather than analysed, and/or criticisms of anecdotalism (Pope et al 2006, Bryman 2008). Validity is ensured through the provision of a detailed and descriptive narrative which offers insight into ‘what’s going on’ in terms of late booking, attempting to identify connections or relationships within the data and integrate these into existing theories around the subject (Ziebland and Wright 1997, Green and Thorogood 2014). Such analysis aims to represent data holistically rather than simply reduce it down to its component parts, demonstrating a consideration of different explanations for the findings, and presenting findings which reflect the breadth of participant views rather than one sided accounts (Murphy et al 1998).
To support this detailed analysis requires the ‘thick description’ discussed previously. However successful analysis requires even more than this: it is a complex and creative process which is difficult to prescribe and describe completely (Green and Thorogood 2014). As Corbin and Strauss (2008) comment, ‘something occurs when doing analysis that is beyond the ability of a person to articulate or explain’ (p9). Wolcott (2009) examines the difference between early ‘analysis’ – the description and examination of data using systematic and standardised measures and procedures – and ‘interpretation’ – a more intuitive and reflective activity which takes place later the process. He discusses how a well-balanced study should show evidence of both. Braun and Clarke (2006) propose a step-by-step guide to doing thematic analysis which attempts to articulate this balance. This echoes Leininger (1985) who proposed similar sequential conceptual steps for thematic and pattern analysis in relation to nursing studies; both reflect influential phenomenological approaches such as the method proposed by Giorgi (1985). This process is elucidated using (loosely) the six stages they propose in the following chapter, though the reality was less linear, and illustrates some of the tensions and complexities. Reflexive thinking formed part of this process of analysis, to try to avoid making premature assumptions about the data and the emerging analysis, as did independent data verification by the research team.