Philosophical Perspective and Methodology
4.3 Template analysis
4.3.2 The main features of template analysis
A template analysis was found to be the most appropriate analysis technique for the investigation. This section of the chapter will provide the reader with an understanding of the main features of template analysis.
Rather than a single, specifically defined technique, template analysis can be characterised by its flexible nature, enabling a researcher to adapt it in accordance with his or her own requirements (King, 2004; Slade et al., 2009). A key feature is that codes, referring to interpreted themes within the data (Richardson et al., 2002), are identified
and hierarchically organised into a template of related higher and lower level categories (King, 2004; Waring & Wainwright, 2008). A final template usually consists of between two and four levels of codes, ranging from broad (high level) to narrow (low level) themes, which are supported by relevant exerts from interview transcripts (Slade et al., 2009). In some cases, specific segments of text can be used to support more than one code at the same level (King, 2004). By organising data in this way, the researcher has a detailed summary from which to anchor his/her analysis (Stratton et al., 2006), making it easier to identify patterns in the presented experiences of respondents (King et al., 2004).
The process starts with the construction of an initial template. This is developed by the researcher as he/she begins coding themes that emerge from a sub-set of the data (King, 2012). As stated earlier, the initial template can also include some a priori themes (Richardson et al., 2002). These preliminary codes are then clustered into hierarchically organised groups that structure the themes based on their broad or narrow relation to relevant aspects of a subject (King, 2012). Once an initial template has been developed, however, it is continually modified in order to allow it to adequately account for the further collection, reading, and interpretation of data (King, 2004; Stratton et al., 2006). In cases where none of the existing codes appropriately accommodate a section of text, the template is likely to require the insertion of a new code (King, 2004). Alternatively, a predefined code will be deleted if it fails to be supported by the data. This will also be the case on occasions when one code, although representative of a theme identified in the data, overlaps with another, more appropriately defined one (King, 2004). Another common revision is an adjustment of the hierarchal ordering of the codes with the aim of developing a more coherent structure (King, 2004). This modification process continues until all the data has been adequately accounted for (Slade et al., 2009).
Owing to the diversity of qualitative data, template codes do not necessarily have to accommodate data from every transcript (Slade et al., 2009). According to King (2004), it can be difficult deciding the stage at which the template can be considered sufficient. What is important, however, is that the final template has a coherent structure that appropriately accommodates all of the collected data that is relevant to the research
subject, irrespective of whether or not it supports a researcher’s developing theory (Waring & Wainwright, 2008).
As the final template informs (and is a part of) the way the data are analysed and structured in the findings chapters it is appropriate to discuss this at the end of the methods chapter. As such, it makes structural sense within this thesis to have the method chapter culminate in an analysis section that discusses how the data were thematically organised into a template of hierarchically organised codes. This will conclude with the presentation of the final template.
4.4
Conclusion
I began this chapter by demonstrating how the underlying ontological and epistemological assumptions of this investigation are consistent with critical realism. The dominant masculine norm, as discussed in chapter three, can be seen as a transitive structure that vicariously reveals itself through the common patterns of men’s behaviour, as identified in the literature (Anderson, 2001; Bennett, 2007; Phillips, 2007; Bishop et al., 2009). Within critical realism, transitive structures are socially pre-established representations of reality (Pilgrima, 2000; Dobson, 2002; McEvoy & Richards, 2003; Bergin et al., 2008) that are internalised by agents (Holmes & Smyth, 2005; Averill, 2006) and perpetuated in society via the influence they can exert over their behaviours (Brown et al., 2002; Madison, 2005; Stewart & Usher, 2007). Therefore, through their continued reconstruction, transitive structures can be relatively enduring (Archer, 1995, Dobson, 2001).
Critical realism does not refer to a cause-effect relationship between transitive structures and patterns of behaviour, and agents are always viewed to retain the potential to act in novel ways (Carspecken, 1996; Zembylas, 2006). Congruent with this assumption, I acknowledge that individual men demonstrate a multiplicity of different behaviours that are temporally and contextually influenced, as a number of authors have suggested (Speer, 2001; Wetherell & Edley, 2003; Allen, 2005; Paechter, 2006). My assumption is
that only certain patterns of their behaviour, particularly those demonstrated during relations between men (Anderson, 2001; Hall, 2002; Whitehead, 2005; Phillips, 2007), revolve around actual attempts to appear masculine (Maccoby, 2000; McHale et al., 2003; Vogel et al., 2003; Whitehead, 2005). In this sense, I am theorizing that the dominant masculine norm, as a transitive structure, only exerts an influence within certain social contexts. The aim of this investigation, then, is to gain a rich understanding of the dominant masculine norm, as constituting a transitive structure, and the ways and reasons for its activation.
As transitive structures are assumed to influence social behaviour (Wilson & McCormack, 2006; Bergin et al., 2008), verbal practices – though never complete nor precise accounts – are seen to constitute a means by which to acquire information about them (Dobson, 2002). It, therefore, emerged that this information is best obtained by qualitative interviews that have the potential to gather rich verbal data. During the chapter, I presented my case for choosing to use a semi-structured, rather than an unstructured, interview technique and to gather data during focus groups and one to one interviews. I also explored why radical relativist informed discursive approaches are incompatible with this investigation’s philosophical perspective.
Template analysis, IPA, and grounded theory approaches are ideal for analysing qualitative data in critical realist informed research. Through comparisons between them, this chapter found template analysis to be the most suitable means of thematically organising and analysing the research data for this particular investigation. Whereas a review of the literature influenced my tentative theories and understandings of the subject, when using a grounded theory or IPA approach, these must be induced from the data (McGhee et al., 2007; Smith, 2008). A template analysis, on the other hand, enables researchers to acknowledge and report their provisional expectations via the development of some a priori themes (King, 2012). Unlike template analysis, IPA adopts an idiographic approach to data analysis (Smith, 2008), making it less comfortable with data gathered from focus groups (Smith, 2004) and samples in excess of ten respondents (King, 2012).