The Partnership for Conflict, Crime and Security Research (PaCCS): Security in an Africa of networked, multi-level governance

This project, led by Professor David Leonard, addressed fundamental problems in the rebuilding of effective states in conflict and post-conflict African countries. All states are subject to multiple levels of real authority, ranging from the local community to a variety of international actors. The effectiveness of governance depends not just on national government but also on the performance of all these authorities and the complex networks linking them.

Ethical issues

Our studies were conducted in African states that had experienced substantial violent conflict in their rural areas, and in most cases these conflicts were ongoing or might recur. Countries studied were the Central African Republic, Democratic Republic of Congo, Ivory Coast, Mali, Sierra Leone and the Somali polities. The units of analysis in our study presented special issues in the treatment of human participants.

We were interested in the experiences of rural African communities during periods of conflict and in the networks of relationships between international actors who tried to assist them. Thus our units of analysis were communities, their experiences, and international assistance networks. The individuals we interviewed were not our units of analysis, but informants about the experiences and actions of communities.

There were two major implications of our context and units of analysis. First, because the latter were social entities, not individuals, participants were providing information about those entities, not about themselves, and hence not subject to anonymisation measures granted for individual participants. In describing social units, their experiences and their networks, the identity of participants cannot be fully anonymised in the raw data sets – someone who is a central actor in a unit is a much stronger source of evidence than someone who is only a distant observer.

The biases associated with different roles in the unit have to be considered as well, and one cannot trace networks without knowledge of the names or positions (and preferably both) of those interviewed. As a consequence, raw interview data from participants cannot be anonymised in the way in which it is when those interviewed are participants and themselves the units of analysis.

Second, confidentiality of sources can be maintained by control and secure storage of the raw data set, but either usability or confidentiality will be lost if the data are put in a public data bank. If confidentiality is to be maintained, the interview transcripts that provide evidence for a social unit’s experience or behaviour cannot be made public.

An alternative is to deposit the coding of the units at an appropriate data service provider, for example the UK Data Service (UKDS), but without the evidence supporting that coding. This was feasible and approved by UKDS in the case of South and Central Somalia, where the nature of the community peace initiatives that had been attempted was already public, and in the case of Sierra Leone, where the experiences being characterised were then over a decade old and therefore no longer sources of conflict.

However, in places where the conflicts were recent or on-going (such as eastern Congo, the Ivory Coast or Mali), revealing the views of villagers whose names can be guessed could lead to those places or people being targeted and possibly killed. Similarly, details of the interviews with members of international assistance networks could easily reveal their identities and compromise their work, whereas analyses of a network as a whole would not.

Lessons

The context and nature of a research project may make it methodologically and ethically inappropriate to handle interview data in the ways that are standard for sample surveys, where individual participants themselves are the units of analysis and all individual identifiers can be removed from the interview records. Our experience is that ESRC and UKDS are quite responsive to requests for variances in their protocols when a well-reasoned and ethically sensitive case is presented.