W = 27,380 + 1.81x … or not?

Knowledge brokering13 Apr 2010Ko Colijn

The controversy surfaces during every armed conflict. Following the invasion of Iraq, estimates of casualties were painfully different. In 2006, the World Health Organization (WHO) reported 151,000 deaths in Iraq, while an article in The Lancet quoted a figure of 601,000. Both investigations were heavily criticized, which is hardly surprising given the considerable difference and the related political implications.

The same applies to the wars in Congo. The International Red Cross estimated that no fewer than 5.4 million people perished there between 1998 and 2007. After criticism by researchers involved in the Human Security Report (HSR), the total was adjusted to 2.8 million. This was also rejected by critics in 2008, and the HSR published a figure less than a third of the original 5.4 million claimed by the International Red Cross. Should we then conclude that, relatively speaking of course, things were not that bad?

In 2008 the British Medical Journal published an article analyzing estimates of war deaths based on health surveys in 13 countries over the period 1955–1994. The authors examined the method used by the Norwegian Peace Research Institute (PRIO), which had concluded that the number of war victims has decreased significantly over the past 15 years. They claimed that PRIO’s calculations were too low by a factor of three. In response, an article in the December 2009 issue of the Journal of Conflict Resolution (JCR) concluded that the Norwegians were right and that the illustrious BMJ was simply wide of the mark. The dispute continues.

It is a dispute about methods. One, the incidents-based method, involves collecting as many reliable reports and testimonies as possible for each war, and then counting the victims. The other, survey-based method analyzes population statistics, including health surveys, and sample surveys by WHO which involve asking people in countries with deficient population statistics how many family members they have lost due to war.

Survey-based research involves comparing expected and deviant trends, and attributing the difference to war. This method systematically generates death figures that are sometimes five times higher than those obtained using the incidents-based method. Et voilà, this largely explains the considerable differences between these examples. The BMJ even produced a mathematical formula that causes the actual number of war victims (W) to be systematically higher than the ‘wrong’ number (x) based on the Norwegian dataset: W = 27,380 + 1.81x.

The authors of the BMJ article applied the survey-based method to the same 13 countries examined using the incidents-based method included in the Norwegian dataset. They disputed the Norwegian outcomes because they believed the dataset was contaminated with unreliable media reports and did not consider genuine, unreported casualties.

The JCR retaliated, claiming it had drawn on much deeper sources than media reports. According to the JCR, the BMJ can not count either, since it quoted 5.4 million war deaths for this group between 1955 and 1994, while the Norwegians referred to 2.8 million. The difference is therefore a factor of two, not three. Applying the appropriate margin of error eliminates even that difference.

The definition of a war death also varies. The WHO talks of deaths that ‘respondents attribute to war’. The Norwegian definition is stricter, referring to ‘battleground deaths in conflicts whereby one of the two armed parties is a state’. This leaves a certain leeway, since there are numerous conflicts between non-state actors, and numerous one-sided conflicts, where an armed party attacks an unarmed one.

The WHO regards the victims of these types of conflict as war deaths, but they are not included in the Norwegian dataset. Genocide or politicide were occurring in eight of the 13 countries, and the definition of war deaths made a difference. Provisional Swedish research showed that the death statistics could be 55% higher as a result. The JCR article calls the BMJ’s methodology poor, condemns it for ‘mischaracterization’ of the Norwegian approach, and accuses the BMJ of being biased in its selection of countries.

Then there is the Norwegian claim that the number of war deaths has declined since the 1950s. According to the JCR, the BMJ’s criticism of this is ‘simply wrong’. The JCR claims that the WHO started collecting statistics just after a period of high victim levels as a result of the Korean War (which ended in 1953) and stopped before a period of lower victim levels between 1994 and 2002, which is when the Norwegians detected a drop in the number of worldwide victims by a factor of five. If WHO had extended the period slightly, the Norwegian optimism would have turned out to be credible.

The dispute between the two schools is welcome in so much as it will hopefully contribute to a convergence of definitions. The fact that such a dispute can actually take place is laudable because it means that, for the first time, multi-country studies can be systematically compared over lengthy periods. However, it is still a dispute, which is being keenly fought on the basis of widely varying viewpoints, and has considerable political implications for decision makers and public opinion.


Coghlan, B., Brennan, R., Ngoy, P., Dofara, D., Otto, B., Clements, M. et al. (2006) Mortality in the Democratic Republic of Congo: A nationwide survey. Lancet 367: 44-51.

Lacina, B., Gleditsch, N.P. and Russett, B. (2006) The declining risk of death in battle. International Studies Quarterly 50: 673-80

Mack, A. et al. (2005) Human Security Report: War and Peace in the 21st Century. New York: Oxford University Press/Human Security Reports Project. Original and subsequent reports can be found at

Obermeyer, Z., Murray, C.L.J. and. Gakidou, E. (2008) Fifty years of violent war deaths from Vietnam to Bosnia: Analysis of data from the world health survey programme. British Medical Journal 336: 1482–1486.

Spagat, M., Mack, A., Cooper, T. and Kreutz, J. (2009) Estimating war deaths: An arena of contestation. Journal of Conflict Resolution 53: 934-949.