Objectives. This research supports the ultimate objective of the Humanitarian Data and Trust Initiative (HDTI) Wilton Park Dialogue: to allow government donors to request and humanitarian organizations to responsibly and safely share data on crisis-affected people with donors whilst doing no harm. It assesses whether and how the risks related to the sharing of data that were discussed at the Dialogue materialize in practice. Separate research conducted by the University of Manchester is looking at how and why donor governments request data, and how they use it. Those findings, combined with the recommendations in this report, will inform the development of principles for safe and responsible data sharing by the HDTI Wilton Park Dialogue.
Methods. The analysis of the risks in this report is based on the observations of humanitarian and donor government staff working in or on the three countries examined as case studies: Bangladesh (response to the Rohingya refugee crisis), Nigeria, and Syria. The research only looked at the past five years (2016-today), to account for the fact that data sharing risks and mitigation measures are changing rapidly. It draws on 35 confidential interviews with humanitarian staff, and ten with donor government agency representatives.4 They included both field staff and staff at headquarters or regional offices. Depending on the structure of their organization and their role in it, the interviewees have different levels of exposure to data sharing with donors. A few primarily focus on data or information management, but most interviewees work in External Relations, Protection, Monitoring and Evaluation, and Programme Management. The organizations and donors selected for interviews were participants in the 2020 HDTI Wilton Park dialogue, or other international humanitarian organizations receiving direct government donor funding for their work in the three case study countries. Individual interviewees were identified thanks to recommendations by the Research Board, personal contacts of the researchers, and suggestions by other interviewees. (See illustrations 1 and 2).
Limitations. The research process faced three important limitations:
Interview participation was limited. The research team defined the participation of six different humanitarian organizations per context as the absolute minimum. The team managed to secure five per context. This limited diversity of humanitarian organizations did not make it possible to draw decisive conclusions about how risks materialize. However, the rich information from participating humanitarian organizations and donors provides a sufficient basis for identifying risks that should be considered when sharing data related to crisisaffected people.
An important part of the proposed method was to conduct a data stress test to evaluate the re-identification risk in available data and donor reports. Unfortunately, the research team did not obtain enough suitable datasets to conduct the planned stress test. This report therefore only summarizes the qualitative interview data. We have indicated in the text below where we expect important insights from such a stress test.
The research into donor practices by the University of Manchester follows a different timeline from this research. It was therefore not possible to take into account the findings and recommendations where the two efforts look at complementary aspects of the same issue.