4 Chapter Four Methods
4.7 Analysis
4.7.1
Introduction
There are a variety of methods of analysis that are applied to qualitative data, which should be applied with rigour, scrutiny and integrity, factors that are important in making the basis of solid analysis (Green and Thorogood, 2014). Choosing which type of qualitative analysis is informed by factors such as; the type of data collated (Spencer et al., 2014), the aims of the study, the research questions, research paradigm and its intended outcomes, such as whether it is used as a function to describe or whether explanations can be deduced from the description or whether a theory can be derived (Green and Thorogood, 2014) (Spencer et al., 2014, Gale et al., 2013). A thorough qualitative analysis should result in the
illustration of the depth of the subject being researched (Green and Thorogood, 2014). Analysis is by nature an iterative process that often requires theories and thoughts to be re-interrogated, interpreted and continuously developed (Ormston et al., 2014) and thus is a process that should start at the beginning of qualitative data collection. These
considerations and those that follow are applied to the decision making used to identify the most appropriate form of analysis for this study.
There are two overarching approaches to qualitative analysis; substantive and structural (or constructionist) (Spencer et al., 2014). The substantive approach concentrates on what the text is saying and is therefore relevant to analysis such as thematic analysis, grounded theory, content analysis and framework analysis. On the other hand, the structural approach centres on what the text is doing as opposed to what it is saying, i.e. how the language is structured, which is thus suited to narrative and discourse analysis (Spencer et al., 2014, Gale et al., 2013) . At this point it can be said that a substantive approach will be taken in this study.
The epistemological approach one takes also has an influence on the method of qualitative analysis identified, i.e. whether it is inductive and thus looking for the patterns that emerge from observations made or deductive whereby theory and ideas result from the process of research and analysis (Ormston et al., 2014) . Qualitative research is generally regarded as being inductive (Strauss and Corbin, 1998) however both induction and deduction often
102 occur at different stages of the research process (Ormston et al., 2014). On the basis of this understanding for this study, a predominately inductive approach shall be used.
4.7.2
Data to be analysed in this study
The focus groups are to be analysed to ascertain the views and understanding of the participants in regards to evidence based practice and what it means to them. In
addressing these areas, further insight into the three overarching elements of the PARIHS framework; evidence, context and facilitation will also be explored. In the semi structured interviews the data will be analysed to establish the applicability of the PARIHS framework cross culturally. Therefore, to enable the identification of an appropriate analysis for this study, consideration will be given to two substantive approaches; thematic analysis, and framework analysis as the analysis of the data collected in this study is concentrating on the meaning of the data. As a result this section will review these two approaches and provide an in depth overview of the chosen method of analysis.
4.7.3
Thematic analysis
Thematic analysis is a method of qualitative analysis that looks at the content of the data in order to identify the themes and patterns in it (Spencer et al., 2014) and to compare these themes from one set of data to another and identify patterns (Green and Thorogood, 2014). Whilst themes and patterns are identified from the data, themes are not always those that emerge from the data, as some may be pre assumed, identified from existing literature or based on professional experience and used as part of the process of data collection (Ayres, 2008). In keeping with other qualitative methods of data analysis, the process of thematic analysis is not sequential and individual ‘steps’ of analysis; coding, data management, identification of themes and patterns, may be revisited (Ayres, 2008). As a method of analysis it has wide accessibility as it is not connected to a particular theoretical construct (Spencer et al., 2014). Thematic analysis can be used as a stand-alone method of analysis yet it is also a component of other analyses such as grounded theory (Spencer et al., 2014) and framework analysis. It is commonly used in health research as it is applicable to research that intends to provide answers to specific issues or find common responses (Green and Thorogood, 2014). However, it is limited if one wishes to transcend the themes on the surface or aims to identify theories that may evolve from the data, which is the intention of this study. Therefore, thematic analysis will not be employed in this study,
103 however it is important to acknowledge that it is the basis of framework analysis, which allows for the deeper analysis of data.
4.7.4
Framework analysis
Framework analysis was developed by the National Centre for Social Research (NCSR) in the 1980’s, since when it has continued to evolve (Spencer et al., 2014). The NCSR have a remit for the development of social and public policy and therefore framework analysis was originally developed with a focus on the production of robust policy (Green and Thorogood, 2014).
As a means of qualitative analysis it is clearly structured with defined steps that act as a guide through the process. These steps, like other analytic procedures run either as a continuum or they can be operated in parallel and can be repeated during the analytical process (Ritchie and Spencer, 1994). The orientation of the approach is reflected in the name ‘framework’, as the essence of the process is the organisation of themes from the data into a written framework, i.e. a matrix (or matrices) populated with participant data (Ritchie and Spencer, 1994) and themes from the literature. The matrices work to maintain data integrity of individual respondents as the data is continuously linked to its original source (Green and Thorogood, 2014) (Gale et al., 2013) By structuring the data in this fashion, comparisons within and across data are more easily facilitated (Gale et al., 2013) (Ritchie and Spencer, 1994).
Framework analysis is independent of any epistemological or theoretical links and can be used with either inductive or deductive coding (Ritchie and Spencer, 1994) (Gale et al., 2013), both of which are factors that increase its flexibility and application (Gale et al., 2013). Framework analysis has been identified as being suitable for case studies (Ritchie and Spencer, 2000) and semi structured interviews (Gale et al., 2013), is clearly structured, and allows for comparison across themes (Gale et al., 2013). Themes may be those which emerge directly from the data (inductive) or those that are applied to the data from
existing literature or theories (deductive). Furthermore, framework analysis is a clear, step by step process of data analysis which has been applied to a variety of topics in an
extensive international arena.
On the basis of these factors, it’s working application and its suitability to case study, framework analysis has been identified as the method of analysis best suited to this study.
104 The following outlines the process of framework analysis indicating how it is applied to analysing data.
4.7.4.1 Framework analysis in practice
There are a number of studies in adjacent areas to this thesis which have successfully used framework analysis. In a study looking at the influences on and the process of decision making of guideline development groups, (Gardner et al., 2009) used a mixture of data analysis methods including framework analysis. Framework analysis was applied to the analysis of data from verbatim meeting transcripts and semi-structured interview transcripts (Gardner et al., 2009). In another study looking at experiences and views of general practitioners (Tonkin-Crine et al., 2011) looked at antibiotic prescribing practices in five European countries. They used framework analysis to analyse semi-structured
interview which were held to explore the experiences and views of general practitioners usage of antibiotics. In South Africa, (Marais and Petersen, 2015) looked at factors within the healthcare system regarded as blocking or facilitating the implementation of a model of integrated mental healthcare through a series of semi structured interviews. Framework analysis was identified as suitable because the study outcomes were directed at improving policy and legislation, which is what framework analysis is not only often used for but was developed to do. Additionally, it highlighted the barriers and facilitator to the
implementation of the mental health policy. As a final example, in Iran Imani-Nasab et al (2014) used framework analysis to evaluate a series of semi-structured interviews held with those involved in the development of evidence based health policy documentation (Imani- Nasab et al., 2014). In this example through the use of framework analysis, 3 main areas; behavioural, normative, control beliefs were identified as barriers and facilitators to development.
The reasons outlined above underpin why framework analysis will be used to analyse the case study data; focus groups, semi-structured interviews and observation data. As framework analysis will be used a further explanation of its processes is given below.
4.7.4.2 Five steps of framework analysis
Whilst like other types of data analysis, framework analysis has evolved since it was initially developed, it still maintains its form as a five step approach to data analyses;
105 interpretation (Ritchie and Spencer, 1994). Each stage is important and will be briefly defined to provide an understanding of the process that not only underpins the theory of framework analysis but to clearly define how this analysis will be used in relation to this study.
Familiarisation is the first stage of the process and is regarded as forming the basis of the whole analysis. The emphasis of this stage is the importance of getting to know the data well, resulting in an in-depth understanding and overview of all the data (Ritchie and Spencer, 1994). To achieve this, the researcher must re-visit and re-read notes, re-listen to recordings and transcripts until the point at which they are sufficiently immersed in the data (Gale et al., 2013). In cases in which there is a particularly large amount of data, then careful selection of what one needs to be familiar with must become the focus (Spencer et al., 2014 -a). During this process of familiarisation the aim is to identify and record relevant ideas, observations, interesting and recurrent ideas in the data as well as recorded any observations about the interview situation as a whole (Spencer et al., 2014 -a). These notations form the basis of the thematic framework (Ritchie and Spencer, 2000). In this study, familiarisation will focus on the transcripts from the focus groups, transcripts from semi-structured interviews and observations made and recorded.
The notes created during data familiarisation, often result in an all-encompassing list of relevant items, however for the thematic framework to take shape, this is reduced (Spencer et al., 2014 -a). During this process, it is imperative that the researcher continually cross checks this emergent list to the original aims of the research and
previously identified factors (Ritchie and Spencer, 2000). From this list, an initial thematic framework is developed with the data grouped into themes and subthemes and an
explanation of the subthemes recorded to ensuring there is clarity about how they are used and their meaning (Spencer et al., 2014 -a). As qualitative analysis is a dynamic process the thematic framework may often change over the course of the analysis as it is revised and new themes developed (Gale et al., 2013). By recording the data under themes, the researcher then has an ability to sort and compare the data (Spencer et al., 2014 -a). The process is very reliant on analytical thinking and application of the original question. Once the framework has been constructed the data is ready to be annotated according to the themes and subthemes, this is referred to as indexing (Spencer et al., 2014 -a). It is important to note that a section of data can have multiple themes and subthemes applied to it (Spencer et al., 2014 -a). During this process the researcher continuously develops
106 views about the data (Ritchie and Spencer, 2000). The process itself calls for strict
documentation particularly as references are indexed, linking them back to the main text, thus increasing its accessibility to others who may wish to ascertain how, or on what, conclusions were based or decided upon (Gale et al., 2013). When all the data has been indexed, it is sorted so that similar data is grouped together. The overall role of sorting the data is that it enables the researcher to give attention to the details and differences of each topic, although data may be sorted into more than one location (Spencer et al., 2014 -a). Following this, charting takes place, whereby charts are created by pulling out the data under themes. The charts act as data summaries and by collating under themes it allows for comparisons to be made both within and between individual interviews (Green and Thorogood, 2014). During this process the text is continuously linked to the source data, again increasing the robustness of the data trail (Ritchie and Spencer, 1994). The final step of the analysis journey is referred to as mapping and interpretation. By this point, all data has been arranged under themes and subthemes and the process of looking for patterns, explanations and relationships between themes and interpretation of the full data set begins (Green and Thorogood, 2014, Gale et al., 2013). The process of framework analysis as defined here will be applied to this study.