Why the World Can Do More to Protect Civilians in Conflict

Noele Crossley, University College London, and Christina Qiu, Yale University, offer some insights on recent trends of mass violence.

Our ability to collectively prevent the worst forms of political violence has become better over the past three decades since the publication of the report of the International Commission on Intervention and Sovereignty, The Responsibility to Protect, and the endorsement of the protection principle in the Outcome Document of the UN’s World Summit in 2005. But more can still be done. In what follows below, we outline several reasons why. These reflections are based on consideration of data available in the Uppsala Conflict Data Program (UCDP) and new methods for conducting empirical research. Policymakers can harness this new knowledge, making use of advanced quantitative methods to shed more light on conflict dynamics and their effects on civilian populations.

There has been progress in reducing the number of conflict-induced deaths. The average number of total deaths, including both civilian and combatant deaths in situations in which armed violence occurs during conflict has dropped considerably in the past two decades, and is today about a fifth of what it was at the turn of the millennium, suggesting that crises have become less deadly overall. The table below shows the average number of total deaths in crises between 1989 and 2019. 1994, the year of the Rwandan genocide, constitutes an outlier year in terms of the average number of total deaths.

Generally, the average number of deaths has declined in this time period – and this although the number of crises has systemically increased. Emerging international norms of protection may account for this change. The genocide in Rwanda and the breakup of Yugoslavia were both decisive in providing the impetus for global action to prevent genocide and mass atrocities. The Responsibility to Protect has reframed the way the international public thinks about the protection of civilians and, arguably, has helped bring about more ‘political will’ to act in crises that put human security at risk.

However, the number of episodes of violent conflict has increased. At the same time, the table above also shows that the number of crises has increased three-fold over the same period of time. This suggests that while international actors – states, the UN, humanitarian organisations – have become better at preventing casualties during crises, more work still lies ahead in finding ways of preventing episodes of violent conflict altogether. The table below shows the distribution of crises across regions post-2000. Asia had the largest number of crises, three times more than the region with the second-highest frequency of crises, Africa. Genocide and mass atrocities – one-sided violence – are forms of identity-based violence targeting civilians. The graph below also shows that while one-sided violence constituted a third of all crises in Africa, it characterised about 1 in 10 crises in Asia.

We now understand the conditions under which political violence is more deadly. A closer look at the available data shows that a small proportion of violent episodes account for a majority of overall deaths. This suggests that if international actors continue to learn about the conditions under which violence against civilians is more likely, it will be possible to tailor measures to prevent violence that accounts for the greatest number of civilian deaths. The table below displays the distribution of death estimates across all crises (the x-axis is log-scaled to mitigate the extent of the skew). The table shows that the majority of the crises had a low death count, while a small number of violent episodes accounted for a large number of deaths.

We now understand the circumstances in which civilians are most likely to be targeted. The graph below maps the relationship between the number of total deaths and the proportion of civilian deaths in crises. The graph shows that the proportion of civilian deaths seems to be roughly bimodal at its extreme ends – in other words, that a large proportion of crises have either no or a complete proportion of civilian deaths. It also shows that the majority of one-sided crises have a complete proportion of civilian deaths, and that the majority of crises recorded are state-based conflicts – that is, conflicts in which the government of the state concerned is involved in the fighting. Early warning for situations in which there is a risk of mass atrocity crimes, as outlined in the Framework for Analysis of Atrocity Crimes, can help identify crises in which there is a higher risk of one-sided violence.

Small-scale conflicts are the least understood. As visualized in the figure for point (1) above, average deaths per crisis have decreased as the number of crises increase. The red line in the graph below is the linear fitted model that relates crisis scale (in terms of estimated number of deaths) and reporting discrepancy. The negative slope suggests that more consistent information is reported for larger-scale crises. The graph shows that the magnitude of reporting discrepancies lower as crises become more serious. Though crises with large death tolls often mark important turning points in conflict, this observation presents a challenge to the accurate mapping of crises, since an increasing majority of crises are small in scale. For this reason, renewed emphasis on tools for mapping the onset of armed violence, such as tools for predicting conflict before it begins, can help alert policymakers to risk situations and draw attention to emerging crises.

The loss of human life as a consequence of conflict is today more preventable than it has been in the past, owing to the growing influence of protection norms and their institutionalisation domestically and internationally. The protection of civilians as a practice of peacekeepers, norms of human protection enshrined in international law, and an increasingly complex network of state-based and civil society actors, the evidence suggests, have contributed to reducing overall fatalities during episodes of violent conflict. More is now known about the conditions under which conflict turns deadly, and a wider repertoire of measures – multidimensional peacekeeping, diplomacy and mediation, targeted sanctions, naming and shaming of norms violators – is now at the disposal of states and civil society in protecting civilians during conflict. Additionally, with new methods of analysis—such as synthetic control, a statistical method that constructs a “control” to examine the impacts of one-time events—sticky questions regarding cause-and-effect, intervener motives, and heterogeneity of intervention techniques can be examined more precisely. The world has become better at delivering assistance to those most in need. However, as these data also show, serious protection gaps still remain.

Noele Crossley is affiliated with University College London. She is the author of Evaluating the Responsibility to Protect: Mass Atrocity Prevention as a Consolidating Norm in International Society (Routledge 2016).

Christina Qiu is a PhD student in Economics at Yale University. She holds an M.Sc. in Global Governance and Diplomacy from Oxford University and a B.A. in Applied Mathematics from Harvard College.

Title image: "Blue Helmets" by riacale is licensed under CC BY-NC-ND 2.0.

If you are interested in submitting a blog post for the ECR2P’s Fresh Perspectives series, then please contact Richard Illingworth by Email (r.illingworth@leeds.ac.uk) or Twitter (@RJI95).