By Anthony Quinn, David Denney, Nick Hardwick, Rahul Jalil, & Rosie Meek
This research note arose from a collaboration between researchers at HM Inspectorate of Prisons and Royal Holloway University of London. Within the last two decades, HM Inspectorate of Prisons [HMIP] has collected a vast array of survey data from detainees in England and Wales. As part of HMIP inspections, the detainee voice is captured via a survey and it is triangulated with other evidence. The survey data inform the inspection report of each establishment as well as annual reports and thematic studies.
These survey data are important because they provide detainees with a rare opportunity to voice their experiences of incarceration. There are questions about a range of aspects of prison life. For instance, participants are asked about the amount of time that they spend outside of their cells, the quality of the prison food and how often they are able to receive visits from their family and friends. Currently there are 159 question items in total.
It goes without saying that in twenty years a dedicated research team has amassed a huge volume of data from detainees within the 117 prisons in England and Wales. HMIP have retained these data for inspectorate purposes and digital records of the paper surveys have been stored in an electronic archive. So, to analyse these historical survey data is it just a case of logging in to the archive and inputting the data into a statistical computing program?
Well, no… far from it. There are a number of complexities that must be addressed to make just one or a few larger datasets with these data. With these prisoner survey data, a major sticking point is the number of iterations of the questionnaires that there have been since the year 2000.i.e. there have been several different versions of the questionnaire and so questions and their response options have varied. This establishes the need to create metadata such as inventories and timelines so that available data can be easily identified.
A wealth of literature explains the challenges of opening up data to secondary users. Primarily, curating and maintaining datasets is costly and time-consuming. This is not an undertaking that should be taken lightly or, where it can be avoided, alone. Secondly, survey data can contain personal information and so it needs to be ensured that data are sufficiently anonymised. Thirdly, data can easily be misused or misinterpreted so it is vital to document and explain the data and their limitations for secondary users.
Any research involving places of detention and detainees raises significant ethical considerations. In this case, detainees had not explicitly agreed data from the surveys they returned to the inspectorate could be made more widely accessible. So, we conducted some focus groups with current long-serving prisoners (we would have conducted more if the epidemic had not halted our efforts) to ask what they thought – they were emphatic that the data should be shared if it would help improve prison conditions. Indeed, some said they would not have taken part in the survey if they had thought the data were NOT going to be used in this way. Further qualitative research with data subjects in order to ascertain their perspectives is certainly an endeavour to be pursued.
Within our Research Note we have put a spotlight on the intricacies involved in identifying administrative data, aggregating them and fully understanding the context within which they were collected. To achieve the latter aim it is vital that, where possible, those who have collected data play a prominent role in the collation of administrative data. This is not a task that should simply be outsourced. Rather, to do justice to such a potentially valuable resource, the expertise of a diverse collaboration of professionals is vital.
Oh, and there’s COVID-19 as well… this has prevented researchers from gaining access to prisons to talk directly to detainees. It has also highlighted the importance of making better use of existing operational and administrative data sources.
Read the full IJSRM article here.