The Changing Landscapes study develops new ways of bringing together existing qualitative longitudinal research together. We have followed a very specific plan of action, which has involved methodological work, such as synthesis and secondary analysis of data collected from across our network; and archive and repository work: developing the Timescapes Archive, which now contains two themed collections of qualitative longitudinal research data (families and relationships over the lifecourse; the third sector). New ways of data collection and analysis have therefore been core to this study.
Our motivation in this study is not only about the usefulness of these ways of working for enhancing our knowledge and understanding about the third sector, nor about the methodological challenges and academic gains from the research. We are also responding to very good practical reasons for thinking about how we collect and store data and how they may be re-used, as the major UK funding councils have devolved responsibility for research data onto the organisations (such as universities) that generated the research. Not only must research data be properly stored, but also research-active organisations are responsible for properly curating those data and making them available for re-use.
This question of curation and re-use leads us directly into considering whether and how any data set is only of value if it can be found and understood sufficiently well to allow researchers to reuse the data. This might include researchers from the original study as, in qualitative longitudinal research, the very nature of the research involves the need to go back to data again and again, and re-analyse in the light of new and emerging data. However, re-use more broadly refers to how new researchers may want to ask different questions of data generated by an existing study, or work across a range of thematically linked data sets. Therefore, this need to understand how to get to data, and what those data are, becomes even more acute.
What are data?
‘Data’, however, do not only mean the transcripts of interviews, but any documents, recordings and images that form the basis for research theorization and outputs. More than that, they can include diaries, field notes and even bus tickets. ‘Data’ include any contextual materials that were produced or collected in the process of producing the study. If we think about historical archives dating back, for example, fifty or even one hundred years it is clear that so much of the taken-for-granted and experiential knowledge of that social period has been lost with time. If modern day researchers want to find out about how people went about getting hold of their participants, say, because they want to know who was actually spoken to and how that shaped the findings of the research, a collection of bus tickets, or photos of participants’ localities, would be invaluable. Fieldnotes would be even better. Any contextual information that would provide insights into how the research was shaped and therefore how the findings were arrived at would be of immense value.
These same questions about context apply equally to contemporary research that wants to understand who the participants were and, therefore, how and why what they say is of relevance to the research questions being asked.
These contextual data are frequently called ‘metadata’, and this is where the fun starts because the word ‘metadata’ means different things to different people. In some places, researchers are increasingly talking about ‘paradata’, which refers to much of the marginalia collected in research. [Abstracts of a good collection of papers discussing this can be found on: http://www2.le.ac.uk/conference/previous-conferences/paradata/abstracts]. Elsewhere, this discussion about what we mean by ‘data’ has exploded in recent years with what has been described as the Big Data phenomenon where the digital capture of much of human activity has given us unprecedented access to the minutiae of social life. In Changing Landscapes, however, we used the word ‘metadata’, and this involved a lot of translation between the people working on the data repository where the Timescapes Archive is held, and the social scientists on the study.
Metadata and the repository: Graham’s view
Qualitative, and increasingly quantitative, research produces documents and transcripts and recordings and images that form the basis for research theorization and outputs. These supporting materials may be at different levels of refinement and may fit together in complex ways. Managing and understanding all of this material or data benefits from organisation, and contextual and management information.
For the individual or project team this “metadata” may be provided by file naming conventions and directory structures or may be within documents that list and describe the files produced by the research. Those from outside the project team will need help finding the files.
For those working in a closely related area this might be achieved through the use of a DOI (digital object identifier) within the publication pointing to the location of the data. [Graham, could you expand on this a little, please]
For these and other interested people it will be helpful if the publication is not the only route to the data. They will want to be able to search for and discover the material based on some set of criteria. What is the research about – keywords, discipline area? Who produced the data, when, where? How was the data generated and in what form – interviews, measurements, transcripts, audio, video?
For those managing the repository it will be important to understand other aspects of the data set. Are there any conditions on its distribution based on time or location or intended reuse? What are the file formats used and hence what might need to be done to ensure the files can be used perhaps years into the future? How will the ownership and contact details change over time?
Once we begin to ask these questions it becomes clear that researchers have to have an eye to the future; to have a sense of heritage about their data and to build in this sort of information about their research as it develops. Research teams are clearly best placed to characterize their data in ways that facilitate re-use and the need to capture identifiers of their research data is becoming increasingly obvious and unavoidable in order that they can be made available through whatever repository is used to store the data.
Metadata: Kahryn’s view
Metadata for social scientists do not only point us to those data which constitute the research (photos, transcripts, diaries), but are research data in and of themselves; data which can be analysed in the context of other data generated within the research in order to refine and test theory.
A good example of this is in the Intergenerational Exchange study under Timescapes (Hughes and Emmel, 2005-2012), where we conducted research on mid-life grandparents’ experiences of poverty in low income communities. Our analysis of participants’ interviews suggested that acute deprivation, both personally and within a locality, meant that people had to work hard to make ends meet on a day-to-day basis. They often struggled to knit together their personal timescapes (daily routines, when they could and couldn’t be available) with the timescapes of formal services and service providers who were often needed to provide crucial resources for the grandparents’ lives. This uncertainty in the short-term management of their time led to difficulties for families planning anything longer term and their attitudes towards their futures, and the futures of their grandchildren, was often very tentative and fragile.
We gained this level of insight not only through what people told us but also through an analysis of how we were or weren’t able to conduct our research with the participants; when they were and weren’t available; how far ahead they were able to plan (usually no longer than a week, and commonly only two or three days). Bringing together different sorts of data in the research meant we were able to observe how vulnerable our participants were to ‘tipping points’, small emotional or financial shocks, that precipitated them into chaos. And through these analyses over time, we were able to observe how meager their capacity to control the circumstances of their lives (http://www.timescapes.leeds.ac.uk/assets/files/methods-guides/timescapes-emmel-analysing-time.pdf) .
In this way, ‘metadata’ refer to all the data collected as part of the research and not only provide a context within which the core data are situated, helping us to make sense of the characteristics of our data, but are able to help in developing and refining theory on the basis of the research.