Data analysis
Qualitative interview analysis
Following data collection, the data was transcribed from its digitally recorded format and transferred into Microsoft Word documents. This resulted in the production of several hundred pages of interview transcripts. All of the preliminary data gathered was organised into spreadsheet format so as to allow for comparability at a glance and also to allow for it to be accessed with relative ease during the quantitative analysis. In truth, the data analysis began long before the interviews were transcribed. Throughout the data collection phase of the study, notes and reflections were made after each day of data collection. These were recorded in order to help with data analysis later, but the observations also helped in refining the questions that were asked and the points that were elaborated on with each prisoner day to day. Any of the analytical memos made daily during the data collection phase were also written up, specifying whether they were in relation to a particular participant or whether they related to the study in general. The interview data was analysed in line with Miles and Huberman’s (1994) three-step approach to qualitative data analysis: data reduction, data display and conclusion drawing and verification. These steps were utilised in the analysis of the interview data.
The transcribed interviews were loaded to NVivo, a popular software tool often used by qualitative research to manage and organise data. The first task was to reduce the data by conducting a preliminary sweep of the interviews. During this phase of the process, each interview was read carefully. Where any issue or theme was observed, a node was created to reflect it. This procedure was followed throughout the entirety of the interviews, and even where a theme appeared to be unrelated to a research question a node was created to reflect it nonetheless. It was possible at that early stage to create a hierarchical structure where relationships appeared to be materialising even during the data collection phase. Once this preliminary sweep of the data had been completed, the researcher set about taking the themes and nodes which had been created and trying to display them in an organised structure so as to be able to observe relationships and categorise themes more effectively. By recoding back and forth several times, it was possible to put a workable structure on the data. This is fulfilling all three of the flows mentioned by Miles and Huber-man (1994). Not only does it refine data, reducing its volume and retaining the most relevant parts, the hierarchical structure also ensures that data is displayed in an accessible manner which lends itself to conclusion and verification. This verification and conclusion drawing is ongoing as the data is being reduced and displayed.
The various findings from the interview data will be outlined in the following chapters, particularly Chapter 4 (prison education), Chapter 5 (desistance — narrative theory and cognitive change) and Chapter 6 (desistance — structural theories).
Quantitative survey analysis
Once the surveys were completed, they were scored in accordance with the weighting system and all of the relevant information was entered into a Microsoft Excel spreadsheet. Demographical data was added to the Excel spreadsheet such as ages, offences, relationships, and so on. These variables were given codes which were developed and recording in an accompanying “codebook”. The scores for each of the five themes, along with the total score for both social capital and the questions relating to the recognition of prison-based capital, were all entered into the spreadsheet. When the spreadsheet was completed, it was imported into SPSS, a statistical analysis tool in order to test the survey data.
Two particular tests formed the basis of the quantitative data analysis. T-tests are appropriate for use when examining the difference between two groups in relation to one single variable. In the present research, these t-tests were used when examining the difference between the social capital and prison-based social capital scores of those who were and were not participating in prison education. The second type of test relied upon for quantitative data analysis was Pearson and Spearman correlations, which are used to measure the strength of a relationship between two continuous variables. This in turn indicates both the direction of the relationship and the strength of it. These tests were carried out when examining the relationship between social capital and prison-based social capital, for example. The gathering of a significant amount of preliminary data also allowed for a considerable amount of testing on the data, such as in relation to previous occupational history, previous levels of education and so on. The results of the relevant tests will be outlined and discussed in Chapter 7.
