Chapter 3 METHODOLOGY
3.7 DATA ANALYSIS
Processing the array of evidence gathered as outlined in the previous section on research procedures began with careful data management. Transcripts were completed for each taped interview; interview forms with additional handwritten notes were filed for each participant; field notes were filed separately. An Excel database was set up to record responses on all survey questions.
The analytic process in my study, described by Robert Yin (2003) as maintaining “a chain of evidence” (p. 105) began with studying the first transcripts for their contribution to the key constructs of social cohesion and internet use, examining the Excel data in table form, and as work proceeded from Time 1 to Time 2, beginning to compare and contrast individuals, groups, cases and all of these again over the second data collection period, with continual
reflection on the two propositions underlying the research goal. In this section I explain the procedures used to analyse the quantitative data and then the qualitative data.
3.7.1 QUANTITATIVE ANALYSIS
• DATA ENTRY
An Excel spreadsheet was needed for quantitative data management, as survey data were collected in face to face interviews at Time 1 and Time 2 in this study. As each interview was completed, data were entered into three separate worksheets relating to Section A (sense of belonging), B (internet use) and C (community-oriented activities and involvement - civic engagement) of the interview. Participants were each assigned a code number, and Excel columns labelled with each survey item. A ‘master’ spreadsheet was retained, so that should an analytic procedure be relevant, columns of data could be copied to new sheets.
For the Internet Connectedness Index (refer section 3.5.1) the columns of values for each of the eight ICI elements (evaluation of the internet; how much would one miss the internet when absent; time spent online; history of home computer use; time spent on online activities; scope of goals in internet use; scope of online activities; scope of places of internet use) in the master spreadsheet were copied to a new spreadsheet. Calculation formulae were set up so Excel standardised all items to a factor of 12, and then averaged the total for each individual, generating a decimalised index between 0 and 12. Results are shown in Table 4.1 (page 142). The procedure was repeated after Time 2 with the remaining nine participants. Results are shown in Table 4.2 (page 152).
3.7.2 THEMATIC ANALYSIS
Grounded theory, developed by Anselm Strauss and Barney Glaser in the 1960s as a branch of qualitative research aiming to generate theory in the sociology of health care, has some applicability to the process used to analyse the material in the present study. Patton (2002) explains that grounded theory uses a “constant comparative method, comparing research sites, doing theoretical sampling, and testing emergent concepts with additional fieldwork” (ibid., p. 125), an apt description of the analytic procedure followed in this study. Data were continually placed in the context of a gathering stream of findings - from the other case study, from other methods, other family contexts, and earlier periods - and in this way illuminated the propositions underlying the research goal, and hypotheses gradually began to form.
Grounded theory makes use of a close interaction between the researcher and the real world “so that the results and findings are grounded in the empirical world” (Strauss & Corbin, 1998, p. 101), and at the same time it emphasises objectivity and rigour in procedures. It requires researchers to be “systematic and creative simultaneously” (Patton, 2002, p. 127) with a balance between objectivity and sensitivity required during analysis:
Objectivity enables the researcher to have confidence that his or her findings are a reasonable, impartial representation of a problem under investigation, whereas sensitivity enables creativity and the discovery of new theory from data. (ibid., p. 128)
This approach requires categorisation of data once description has taken place, so that the researcher identifies properties or characteristics among the data that will be noted, compared and carefully considered for relationships and patterns. For example, in the current study processing of interview transcripts involved multiple re-readings and reflection on similarities and differences between participants, and on detail and threads of information that gradually began to cohere as themes relevant to the research goal. One such theme is the role of opinion leaders at Case A, which formed on the basis of my growing awareness over repeated cycles of transcript analysis that certain interviewees were similar in some respects: in
sociability, outspokenness, esteem from others, willingness to volunteer and take leadership roles, and public individuation. This theme took on greater importance in the context of other social cohesion data on social solidarity at Case A (see section 4.2.2, pages 195 - 196), and began to point towards one of the key outcomes of the research.
Another example of the iterative analytic process described in paragraph 1 of this section above is the data collected on the length of time people had lived in the neighbourhood that showed participants across both cases tended to be recent arrivals in their area, yet there was a sense of pride in belonging to that locality and a strong sense of neighbourliness (J.
Williams, Sligo, & Wallace, 2004a). More detail on this perhaps surprising evidence of social cohesion at Time 1, despite domestic transience, is found in section 4.2.2 (page 185 - 186). Despite their unexpectedness these findings are considered in relation to other results on dimensions of social cohesion and internet use, again leading to one of the key outcomes of the study.
3.7.3 CODING
“Open coding” is an example of ways the inductive analysis above may be achieved through thorough, repeated analysis of the data using a classification or coding scheme. This is essentially a content analysis of all the interview transcripts and associated notes, together with raw and processed field notes. As soon as data collection commenced, it became necessary to design a means of organising and coding the data because although the study involved just two cases and around 30 participants, the volume of data was copious.
• CODING OF INTERVIEW TRANSCRIPTS AND FIELD NOTES
The objective in processing the interview transcripts and field notes was to achieve a system in which the material would not overwhelm but could be read and systematically studied in an early phase of analysis. Open coding, “the analytic process through which concepts are identified and their properties and dimensions ... discovered in data” (Strauss & Corbin, 1998, p. 101) involved the processing of transcripts and notes to identify material relating to the propositions underlying the research goal. As shown in chapter 2, Figure 2-3 (page 66), and especially in Table 3-4 in this chapter (pages 115 – 117), social cohesion has been defined and operationalised for this study as comprising eight dimensions such as place attachment and identity. Initial reading involved identifying particular phrases and whole statements that related to these dimensions. Multiple re-readings helped in summarising perspectives, noting contrasts and consolidating themes. It became possible from an early stage in transcribing and studying interview transcripts to identify, compare, contrast and distil particular
observations that, over time, developed into themes in the process of “interplay between researcher and data” (Strauss & Corbin, 1998, p. 13). This definition highlights the inductive aspect of moving from the data to reflection, to further investigation, identification and movement towards conclusions.