Susie and I both began working on the project as part-time research fellows at the end of May 2015. Now as we go into our tenth month we would like to share with you some of the key pieces of work completed so far.
Our first few months were spent conducting the usual organisational activities required for a project of this length: developing a timeline of our planned activities, submitting our ethics applications, examining key readings, writing our impact pathway and importantly, getting access to our data via the Timescapes project archive.
In total, we are including six of the Timescapes projects in our project (Siblings and Friends; The Dynamics of Motherhood; Masculinities, Identities and Risk; Work and Family Lives; Intergenerational Exchange and; The Oldest Generation). Unfortunately, Youth Lives and Times is currently being updated and wasn’t available. Together this amounts to approximately 1,000 files, 165 cases (we plan to blog soon about what constitutes a ‘case’) and 1,873,553 words. Visual data in the form of activity sheets and photos adds further to the volume of data, and its complexity. Even from the beginning we could see the challenges of scaling up at this level.
Of course, we wanted to get straight into the analysis but we also knew (from experience!) that being able to find and categorise files was critical. We spend a considerable time re-labelling our data so that it had a common filing system across all six projects. All the data was uploaded to an Nvivo server project (a decision which will be another future post) and attributes attached to the data. This would enable us, we hoped, to query the data by characteristics such as age, gender, interview ‘wave’, household type or martial status at a later stage.
Although we were accessing the Timescapes data as secondary researchers, we should say that as a team we have a personal relationship to part of the data. Ros and Susie managed the Siblings and Friends project, while Lynn was responsible for Work and Family Lives. Not only has this connection to the data helped us understand better its origins, but it has also facilitated the many queries made to the original researchers (who have been very kind in answering). This has been invaluable in helping us resolve issues around missing and mis-labelled data – however, we also recognise that not all secondary qualitative researchers would benefit from this contact.
Once the data was organised, we moved on to scope out different approaches that we might take to analysing the data – what we have called a ‘test pit’ approach. We will discuss some of these in more detail in later posts, but our main interest has been in thinking about the ways in which care and intimacy might be identified and recognised within large volumes of qualitative data. We recognised that our usual approaches to analysis – in-depth narrative or biographical analysis – would not be possible given the sheer volume of the textual data. We therefore looked to keyword analysis as a means of getting, what Clive Seale (2013) calls, an “ariel view” of the data. Seale suggests that once complete, keyword analysis can then be used as a means for more detailed analytical work.
To explore this we have run several ‘tests’ including:
- An analysis of word frequencies across all six projects, including a detailed comparison of the most popular words across the data sets.
- Searches for the collocation of words and the presence of specific phrases.
- A comparison of the entire corpus of data with a pre-defined list of intimacy and care key words.
- Conducting in-depth readings of a selection of cases. This more conventional approach to qualitative data analysis was then compared to the results from computer-assisted keyword analysis to determine the extent more implicit discussions of instances of intimacy and care are identified through keywords.
There is not space in this blog to discuss the findings in-depth. However, our initial foray into keyword analysis has drawn several conclusions. First, and not surprisingly, much of the “ariel view” generated is shaped by the focus of the project and resultant direction of the questions (i.e. friendship is obviously central to Siblings and Friends, while work is central to Work and Family Lives). Second, while keywords may refer to component parts of intimacy they are not necessarily sufficient conditions for intimacy. Our detailed case file review revealed that intimacy is often discussed far more implicitly (or are unsaid), and also in a way unique to the individual participant. And finally, how can keyword analysis effectively help us capture time and the cumulative nature of intimacy? So far, temporality seemed difficult to hold onto in the analysis process.
As we now move onto the next phase of the research, we are considering ways in which we can focus our analysis onto one particular aspect of care and intimacy as a means of gaining greater theoretical purchase on the datasets. We will continue to update you as the study progresses on this blog. Please do follow our work and provide your comments and ideas.