3.7.1. Reliability
Reliability relates to the extent to which data is consistent, stable and can be repeated (Brink, 1991). One way in which reliability was maintained was through the control of some interviewer effects; same interview schedule, same interviewer and primary data analyst. However, it is important to recognise that these data
91 were collected in context and thus it may be difficult to assess their reliability. Several methods of reliability were considered, but these were not found to be appropriate for the current study that is rooted within a contextualist epistemology. For example, multiple-researcher coding and the use of code-books (methods illustrated by Boyatzis, 1998; Crabtree & Miller, 1999) were considered, however, Fielding & Fielding (1986) argue that the use of such methods within the contextualist constructionist framework would only add value to the current understanding of the experience, as opposed to functioning to converge the findings. Furthermore, Seidel & Kelle (1995) argue that measures designed to improve reliability in qualitative applications are meaningless; consistent coding using a coding template for example, would suggest that the raters have interpreted the data in a similar way, rather than this being an example of reliable coding. Finally, the use of a coding template may have restricted the elaboration of context-sensitive data (Manning & Cullum-Swan, 1994), potentially limiting the overall quality of data that were developed. As such, formal measures of reliability were considered but not deemed appropriate for use in the current study.
3.7.2 Validity
Yardley (2009) suggests a number of criteria for enhancing the validity of qualitative research such as triangulation, participant feedback and evidencing a paper-trail to verify themes. Secondary researcher coding for purposes of triangulation was not used, however it was felt important to include other researchers in the analysis process to increase the credibility of the findings; reducing the potential bias caused by just one researcher conducting the analysis. Coding discussions were held between the researcher and their colleagues and these were ongoing as codes and themes were developed, in an attempt to generate a more comprehensive and shared understanding of the findings. This would include discussions about whether the themes were valid and where they were located within a broader clinical psychology context. This method was useful
92 in that it added value to substantiate the findings, whilst embracing the subjectivity of the themes, in line with the contextualist-constructionist epistemology.
Participant feedback methodology was also not used here as Sandelowski (2002) argues that such member-checking processes can sometimes become the form of the themes that the researcher has developed, thus reducing the utility of this method as a means for assessing validity.
Instead, the method of disconfirming case analysis was used to demonstrate validity for themes identified within the study. Disconfirming case analysis involves searching for data that serves to test the hypothesis or disconfirm the theme or pattern that has been recognised within the data. Disconfirmation case analysis therefore provides an indication as to the limits of the theme and the extent to which it can be generalised (Booth, Carroll, Ilott, Low, Cooper 2013; Pope and Mays, 1995). Each theme developed as part of the thematic analysis was scrutinised for disconfirming cases and evidence of this is discussed within each chapter where appropriate.
Yardley (2009) suggests that a paper trail can help to improve the subjective status of the findings. The paper trail presented in appendix 3-I additionally illustrates how certain decisions were made6, to increase transparency with reference to how themes were developed and to demonstrate credibility of the findings overall (Koch, 1994). In addition, the frequencies of each sub-theme are noted in parentheses next to each sub-theme heading in the following results chapters, to illustrate the relative prevalence of the themes across the data. Data which illustrate the variability in contributions from participants and theme frequencies are also reported within appendix 3-I.
6
This is available electronically as part of the NVivo software that was used to collect the data. Some of the later stages of the analysis were transferred to paper for ease of developing themes. Mind-maps are also available in appendix 3-I.
93 Finally, to demonstrate the quality of the qualitative analysis the COREQ checklist (Tong et al., 2007) was used to assess the quality of the research detailed within this thesis. The COREQ is a 32-item checklist that aims to encourage the reporting of qualitative research conduct. For example, some items enquire about
overall theme representativeness and clarity. The full checklist as applied to this research study can be found in appendix 3-J.
94
CHAPTER FOUR
THEME 1: PART OF WHO I AMINTRODUCTION
The Part of Who I Am theme captures the origins of violent thoughts and fantasies (VTF) and how patients make sense of these experiences in terms of their identity as an offender or mental health patient. The theme name is so termed to boldly convey ownership over the experience of violent thinking, whilst simultaneously highlighting the ascribed importance that violent thinking has to
M VTF
as a consequence of how they were brought up; the experiences they encountered as children and adolescents such as being brought up to use violence and/or being a victim of violence during childhood. As such, the experience of violent thinking was usually portrayed as an abnormal experience but one which patients claimed was normal or regular for them, because of who they are - their identity as a mentally disordered offender.
This theme illustrates the origins of VTF and describes how these A , it provides a useful introduction to the results section and for the subsequent themes that will be
I I
detail, alongside the two sub-
A at the end of the chapter,
within the context of existing literature and psychological theory.