In analysing the data, audio-tapes o f interviews were first fully transcribed. Similarly to the idiographic process o f analysis outlined by Smith, Jarman & Osborn, 1999), which begins with individual cases and slowly working up to more general categorisation or theory, interviews were then analysed according to the following stages;
1. Looking for themes
Each interview was initially analysed individually. In analysing interviews, transcripts were read a number o f times, using one side o f the margin to note down key words or phrases (‘meaning units’) that captured the essential essence o f the text. An example o f this from an extract o f one participant’s pre-release interview follows:
I: So how are you feeling about going?
Doesn 7 know how he feels
P: Weil, I don’t know yeah.... I don’t know if I’m about leaving
supposed to be happy or anything... but really I can’t
really explain how I am feeling, because I’m not really Feeling nothing
feeling anything. Like to me it’s just I ’ve been through
a phase. I’m coming out o f there. That’s all it is really. Coming out o f a phase
it’s nothing big. It’s just the next step o f the ladder,
that’s all it is. That’s the way I see it. Release no big deal - just one step on the ladder
Comments, including attempts to summarise, draw connections and associations were also made, along with more preliminary interpretations. These were documented as illustrated below using an extract from another participant’s interview - in the opposite margin:
Re-interpretation o f I: And have you got a girlfriend?
girlfriend's failure to visit
to re-establish a sense o f P: Yeah. But she don’t come. I don’t want her to see
personal control me in jail. I swear, because I ’m in jail she could be doing things behind my back. I tell my friends to keep
Development o f paranoia? an eye on her.. I don’t care anyway.
Finally, all comments and identified meaning units were re-read to identify themes which appeared to run throughout the interview. In most instances themes were highlighted that appeared to encapsulate the essence o f a number o f similar meaning units - particularly if the meaning units recurred across different areas o f questioning. The potential clinical importance o f concepts expressed was also borne in mind in delineating themes however.
2. Producing a table of themes for each interview
In the first instance, a list of an interview’s emerging themes was compiled and examined for connections and clusters. Particular attempts were made to identify superordinate concepts or themes which helped to order the data.
On the basis o f this, a final table o f themes which seemed to encapsulate the participant’s major concerns was then drawn up for each individual interview. This table included superordinate and subordinate themes as appropriate.
3. Producing a master list of themes for the group
The above stages were repeated for each interview separately. A master list o f themes was then constructed including superordinate and subordinate components apparent across interviews. A list o f relevant extracts for each theme and sub-themes was also compiled during this stage o f the analysis. Initial passages o f text were referred back to repeatedly during this process, in order to both check interpretations and elucidate the nuances o f themes across accounts. This stage involved perhaps the greatest amount o f interpretation by the researcher, and required a certain amount o f selectivity. Factors such as the prevalence of themes within the data, the richness of associated passages and the potential relevance of themes (e.g. to the mental health and well-being of participants) were taken into account during this process.
In addition to this, in selecting themes some consideration was given to the target audience o f the research. Several authors have commented that qualitative research should be applicable and o f use to parties beyond those who participated (e.g. Stiles, 1993, Lincoln & Cuba, 1990). Thus in the master list themes were retained which would be o f interest and relevance to YOI and YOT staff working with YOs, even if these were somewhat unrelated to the original research questions. More generally, efforts were made to document themes in an accessible and jargon-fi'ee manner to facilitate the understanding of a non-psychologist readership.
Finally, reliability checks were made at all stages o f the process by an ‘analytical auditor’ (Elliott et al, 1999). Specifically, the supervisor o f the project made checks of the analyses o f 5 interviews (3 pre-release and 2 post-release interviews) to ensure that themes produced fi’om the data were credible and warranted (Lincoln & Guba, 1985). The list o f master themes and accompanying list o f extracts were also audited by this individual. The few minor differences o f opinion arising fi*om these checks were discussed and resolved. The final agreed list o f themes and extracts provided the basis for the written account.