Template analysis is a flexible technique used to thematically organise and analyse qualitative data [King, 2004]. Having been developed by King [1998, 2004] in the late 1990’s, template analysis is a modern development within qualitative analysis and as such there is a relatively little literature in this area. Nevertheless, template analysis shares many similarities with other approaches such as ‘code-book analysis’ and ‘thematic coding’
and therefore it is possible to use some additional sources for guidance [Crabtree and
197This does not include the six purchasing clerk interviews which average around thirty minutes per interview.
174 Miller, 1999; Miles and Huberman, 1994]. Indeed, King himself states that template style approaches are widely employed across various types of qualitative research [King, 1998, p18].
To some extent template analysis may also be seen to share some similarities with the approach of grounded theory. Both approaches are inductive, iterative, and, use codes to organize and analyse large amounts of qualitative data. Nevertheless, there are some key differences between the approaches, which have led to template analysis being selected for this study. In reference to grounded theory King notes that ‘template analysis may be preferred by those who are not inimical to the assumptions of grounded theory, but find it too prescriptive in that it specifies procedures for data gathering that must198 be followed’
[King, 2004, p 257]. In contrast to grounded theory King finds that, ‘template analysis is, on the whole a more flexible technique with fewer specified procedures, permitting the researcher to tailor it to match their own requirements’ [King, 2004, p 257].
The level of flexibility afforded within the template approach allows the analytical process to evolve and adjust throughout the course of the study. Whilst grounded theory encourages the researcher to identify a ‘core’ variable at a relatively early stage, within template analysis there is the freedom to re-classify the importance and presentation of themes overtime. This is considered a major advantage in light of the relatively explorative nature of the partial model study. Equally, template analysis allows the researcher to develop a style of presentation best suited to answering the research questions within the study. Whether this involves focusing on one core theme or several key themes, once again, there is the freedom to make such decisions over the course of research.
However, whilst template analysis’ strengths are found in its pragmatism and adaptability, at the current time it is relatively underdeveloped in terms of guidelines for application.
King [2004] himself finds that it is the researchers own prerogative to tailor the loose framework of template analysis towards the needs of their research project. Therefore,
198 King [2004] places the word ‘must’ in italics based on Straus and Corbin [1990], where it is stated that,
‘Procedures must be followed in doing research...it is only by practicing procedures through continued research that one gains sufficient understanding of how they work, and the skill and experience that enables one to continue using the techniques with success’ [Straus and Corbin, 1990, p26].
175 despite the above critique of grounded theory, there may be certain areas where template analysis can learn from this approach. Indeed, King [2004] observes that template analysis and grounded theory are not inimical within their approach, thus opening the door to the possibility of one approach learning from the other. In light of this, insights from grounded theory have been employed within the process of refining codes/themes, the theoretical sampling carried out prior to round two research, and the process of memoing.
6.6.1 - Coding
Template analysis, like grounded theory, uses codes to organise and interpret qualitative data. King [2004] states that ‘put simply, a code is a label attached to a section of text to index it as relating to a thematic issue in the data which the researcher has identified as important to his or her interpretation’ [p257]. As such, each code is directly related to a theme in the template. Depending on the researchers style the code may be a number representing a theme, or it may simply be an abbreviation of a theme. Codes enable the researcher to transform a seemingly unmanageable amount of rich qualitative data into clearly identifiable themes related to the research question. Closely related to this Crabtree and Miller [1999] describe the development of a ‘code manual’, or ‘template’, as a means of organizing text within the larger interpretative process.
King [2004] notes that codes can be simple and descriptive, or deep and interpretative.
Typically each template will contain a variety of both, depending on both the phenomenon being examined and the researcher’s style of presentation. Within each template, codes are organized hierarchically, where a grouping of lower order codes are clustered together under a higher order code. Lower order codes are essentially used to more specifically define higher order codes, deepening the analysis of the text. Whilst the researcher may choose as many levels of coding as desired King [2004] recommends between two and four levels.
As the researcher reads through the text codes are applied where appropriate to sections that relate to a particular theme. Within template analysis parallel coding is allowed, where the researcher is able to classify the same segment of text with two or more codes if
176 required [King, 2004]. Crabtree and Miller [1999, p167] find that through coding, ‘the template process reduces the amount of data being considered at any one time and brings together related pieces of text earlier in the process, which can facilitate making connections’. By providing structure template analysis also helps to reduce the fabricating, discounting and misinterpreting of evidence, which can unconsciously take place in the interpretation process [Crabtree and Miller, 1999]. Miles and Huberman [1994], also support the template approach, as a means of displaying an objective interpretation of complex social phenomena.
6.6.2 - Data Interpretation: Developing the templates
Based on the approach of critical realism the interpretation of data from interviews begins with the assumption that participant responses give us direct access to their lived experiences. This does not mean that responses directly reflect reality, but rather it is necessary to explore the underlying trends within participant experiences in order to understand the generative mechanisms at work within the phenomenon being studied.
Data interpretation has therefore proceeded with the initial objective of appreciating respondent’s experiences at the individual level, later advancing to the stage of exploring the underlying trends and linkages between respondent’s experiences.
The approach of template analysis itself does not specify any particular process for the interpretation of data. Observing this problem, King finds that, ‘a strategy must be developed which fits the aims and content of the particular study’ [King, 2004, p266]. In light of this, the interpretative strategy for this study was designed with two main concerns: firstly, the exploratory nature of the early stages of data collection, and secondly, the study’s aim of presenting a focused response to research questions. As such, interpretation began by using a mechanism to organise and interpret a broad array of exploratory, themes, data and conceptual ideas, and overtime it channelled this broad amount of data and ideas into a clearly defined understanding of each theme within the final template. This process occurred in two distinct phases: firstly, the selection and categorization of codes into templates, which was accompanied by the process of memoing, and secondly, the chunking of related sections of text and data immersion.
177 6.6.2.1 - Selection and Categorization199
Beginning with the initial stages of coding, the process of selecting and hierarchically organising concepts into themes was a key aspect of the analysis within this study. Not only does the presentation of themes within the template demonstrate what themes are present within the data, but the hierarchical structure helps to show both the varying importance and relationship between themes in the data. As such, the template acts as the cornerstone upon which the various elements of data analysis are based, and each stage of template development represents a key stage in the journey towards the final research findings.
There were three main factors affecting both the selection of key themes, and the organisation of lower order themes within the template for each of the key themes.
Firstly, themes were ordered based on how they relate with the research questions for the study. Research questions provided a solid bench mark against which all themes, both a priori and emergent, could be measured. This has resulted in a style of presentation for the research findings that is both clear and relevant in its connection with the research questions for the study. The central importance of the research questions within the methodology of this study is conceptually demonstrated in figure 6.3 on the next page.
Secondly, themes were organised based the richness of the data related to each theme. The richness of data was considered in equal measure to the link each theme had with the research questions. Data richness was determined by a number of factors including; the level of explanation given by each respondent, the quality of examples related to the explanation, whether the issue emerged without prompting within the respondent’s narrative, the level of emphasis placed on the issue by the respondent, and the respondents level of expertise in the area on which they are commenting.
199 For a full demonstration of the process of coding and template re-organisation please refer to appendices number four and fourteen.
178 Thirdly, and least importantly themes were selected based on the regularity with which they occurred in the data. This does not signal a content analysis type approach, but rather this technique is used to simply help draw attention to key generative mechanisms, as supported in the earlier section on critical realism200.
Figure 6.3 – Research Questions and the Analytical Process
Source: Author
King [2004] notes that the re-organisation of templates will occur throughout the template analysis process. Template analysis enables the flexibility for themes to be inserted, deleted and re-classified at any stage leading up to the presentation of the final template. Indeed,
200 Within the section on critical realism the identification of ‘demi-regs’ was observed as a process within the critical realists methodology.