CHAPTER 4 METHOD: OVERVIEW, QUALITATIVE DATASETS AND ANALYSIS
4.5 Sample sizes and data availability
Table 4.6 provides a summary of the datasets available for analysis, with corresponding numbers of participants involved and number of coded instances derived from the present discourse analysis (an indication of how substantial each of the datasets are for the purposes of the present study). The number of groups involved in each research project51 is also given.
51 In the case of the 2003 interviews, this number (n=21 interviews) is divided by three for equivalence with longer focus group discussions.
102 Table 4.6: Datasets and extent of data available for analysis, 1997-2010
Dataset Number of participants (no. groups)
Number of codes generated (approx. codes per
participant)
Data stage (no. participants)
1997/8 14 (6) 405 (29)
‘Early’ datasets (n=33)
2000 19 (4) 592 (31)
2002 24 (3) 770 (32)
‘Interim’ datasets (n=45)
2003 21 (7) 730 (35)
2007 84 (9) 984 (12)
‘Recent’ datasets (n=131)
2010 47 (10) 2150 (46)
Total 209 (39) 5631 (27)
As can be seen from table 4.6, there is a general trend towards an increase in the amount of data available over time. Whilst there exists no clear clustering of data at particular time periods, for the sake of convenience the data stage in table 4.6 indicates approximately where, relative to the overall time period, these research projects are situated. Thus where research findings are reported subsequently in this thesis, this is often with reference to ‘early’ and ‘recent’ data – though individual research projects are also referred to in turn.
Overall, the quantity of data available is considerable, however is more limited towards the earlier period. Given the smaller participant pool from 1997/8, in particular, interpretation of longitudinal trends in discourses is often achieved through considering datasets together (e.g. of ‘earlier’
datasets against ‘recent’ data).
103 4.6 Evaluating the qualitative research
Taylor (2001) suggests that rigour in discourse analysis can be linked, firstly, to the richness of detail in both the body of data used and the analysis presented; secondly, this author suggests rigour is linked to the explication of the process of analysis.
I suggest that the richness of detail of the datasets available is substantial – indeed the present research is fortunate through being able to use secondary data, and so to have a larger corpus to draw on than do most primary studies. Taylor’s criterion for the richness of detail in the analysis as presented, is addressed through the provision of a comprehensive set of participant excerpts and detailed interpretation for each discourse (presented within chapters 5 to 8). Due to restrictions of length, summaries only are given in the main text of the thesis; however, the full set of arguments and evidence for each discourse identified, are placed in the appendices of the thesis to which the interested reader can refer.
Taylor’s (2001) suggestion that rigour in discourse analysis is also connected to the explication of the process of analysis is addressed in detail in the commentary given above. I have set out in some detail the procedure by which discourses are constructed, in terms of the identification of their component parts, overarching ontology, and consideration of function. In addition, there is detailed consideration of the means by which the analysis balances the application of prior constructs with a data-driven approach, clarification of how context is attended to, and a framework presented for the identification of variation over time in discourses.
Taylor (2001) also suggests that discourse analysis may be evaluated through triangulation. In the present thesis, this is approached within the qualitative analysis by the use of multiple datasets to derive discourses. Triangulation of methods is also enabled through the use of survey methodology, one objective of which is to measure levels of agreement with statements that reflect discursive framings. One means by which the validity of the discourses presented in the qualitative findings can be appraised, is therefore through an assessment of how ‘commonplace’ these are found to be in a representative survey sample.
Criteria for evaluating qualitative research more generally are provided by Yardley (2000) who suggests four main characteristics of good qualitative research. These are reproduced from Yardley (2000:219) in table 4.7; following this author, essential qualities are shown in bold, with examples shown in italics of the form these qualities can take.
104 Table 4.7: Characteristics of good qualitative research
Sensitivity to context
Theoretical; relevant literature; empirical data; sociocultural setting; participants’ perspectives;
ethical issues
Commitment and rigour
In-depth engagement with topic; methodological competence/skill; thorough data collection;
depth/breadth of analysis
Transparency and coherence
Clarity and power of description/argument; transparent methods and data presentation; fit between theory and method; reflexivity
Impact and importance
Theoretical (enriching understanding); sociocultural; practical (for community, policy makers, etc.)
Whilst each of these 17 separate possible criteria for evaluating research will not be considered separately in this section, I do suggest some of the ways in which the essential qualities are addressed.
Firstly, one important facet of sensitivity to context in the present study, is in terms of what Yardley terms the ‘sociocultural setting’. Particularly for the social and personal discourses identified (chapters 6 and 7) there is an attempt to present in some detail how people’s understanding of self and society informs perspectives on climate change. From a researcher perspective, sociocultural context further informs the interpretation of participant talk – for example, through attention to how individuals manage normative injunctions to act pro-environmentally. In terms of attention to sociocultural factors particular to the time period (e.g.
the changing political climate) this is considered separately and in some detail with respect to each discourse identified, in chapter 11.
With respect to commitment and rigour, I argue that Yardley’s criterion of ‘breadth/depth of analysis’ is met in the present study. Yardley (2001:221) suggests that rigour can be conceived of as the “completeness of the data collection and analysis”: the ability of the data to supply all the information needed for a comprehensive analysis, and attention to the variation and complexity observed in the data. In the present study, as noted above, multiple datasets representing a large corpus of data are used. The completeness of analysis, I suggest, is revealed through the detailed account of the diverse ways in which climate change is understood – as outlined in chapters 5 to 8.
105 Attention to complexity of the data is attended to particularly in terms of the diverse range of concepts and their use within each discourse; variation within discourses is also of particular relevance in the present study for interpreting change over time.
As well as reporting of findings within chapters 5 to 8, I have included within the Appendices extensive additional detail for each discourse identified. This approach to reporting is commensurate with the criterion that Yardley terms ‘transparency’: using this approach, numerous excerpts from transcripts are able to be presented as evidence for conclusions drawn. In addition, I have detailed in this chapter both the processes of data collection used, and analyses applied.
In terms of the impact and importance of the research, such is likely indeed only be revealed some time following the production of this thesis52. Nevertheless, another means by which Yardley suggests impact can be assessed, is in the presentation of a novel, challenging perspective on a topic, which can be seen as having theoretical worth. It is the aim of the present study that this should be an outcome of the research undertaken: this is returned to in the Discussion chapter, in light of the implications of the study findings for current understanding of public perspectives.