Relative Risk of Mass Killing Onset in 2019: Results from a Comparison Survey

According to participants, who voted on a series of head-to-head match-ups, the top 30 countries at risk for mass killing in 2019 are: 

1. Yemen
2. Somalia*
3. Iraq*
4. Democratic Republic of Congo*
5. South Sudan*
6. Syria*
7. Mali
8. Central African Republic*
9. Afghanistan*
10. Sudan*
11. Burma/Myanmar*
12. Egypt
13. Chad
14. Nigeria*
15. Pakistan*
16. Republic of the Congo
17. Niger
18. Burundi
19. Libya
20. Ethiopia*
21. Turkey
22. Zimbabwe
23. Guatemala
24. Eritrea
25. North Korea*
26. Philippines*
27. Rwanda
28. Venezuela
29. Sri Lanka
30. Israel

Countries in bolded with an asterisk had ongoing episodes of mass killing at the end of 2017 (EWP staff are currently compiling data on 2018 onsets). Though it is possible these countries will experience a new onset of mass killing in 2019, respondents might have confused the risk of a new episode—in which either a different perpetrator group begins a new mass killing, or a the same perpetrator group targets a different group than the one(s) already being targeted—with the persistence of an ongoing one. 

Participants voted Yemen to be the highest risk for new mass killing in 2019. Violent conflict between a Saudi-backed coalition supporting the government in exile and Houthi rebels has produced more than 60,000 total conflict deaths—6,480 of which were civilian fatalities—since January 2016, according to estimates from ACLED (the conflict began in 2014). Many thousands more have died from disease and hunger caused by the economic crisis as well as a coalition-imposed blockade. EWP might already count Yemen as having an ongoing mass killing, but the project’s definition of “mass killing” excludes cases in which civilians are killed by an armed group from outside of a country’s borders. 

Mali is the second highest-risk country without an ongoing mass killing, according to the survey. At the end of December, armed men killed at least 37 Peul (also known as Fulani) civilians in central Mali, in what many believe was an ethnically motivated killing, the culmination of a year of escalating intercommunal violence. The Early Warning Project recently conducted an in-depth analysis of potential scenarios of mass atrocities against civilians in Mali, and the recent attack mirrors one of the scenarios identified in the report—that intercommunal violence could escallate in central Mali between the Peul, Dogon, and Bambara ethnic groups, with civilian Peul as the most vulnerable target group. 

Both Yemen and Mali have moved up in the rankings two years in a row, with Yemen rising from 15th (2017) to 4th (2018) to 1st (2019) and Mali from 20th (2017) to 14th (2018) to 6th (2019). 

The third highest ranked country on this year’s survey without an ongoing mass killing is Mozambique. Although rarely in the news, Mozambique is showing signs of fragility. The repressive government allegedly has been a perpetrator of violence in an ongoing civil conflict spanning central Mozambique, and a militant group in the country’s north has staged attacks on police and civilians to which the military allegedly responded with arbitrary detentions and summary executions. Mozambique has presidential elections scheduled for October 2019, which are recognized by many to be triggers for increased violence in high-risk countries. 

The annual comparison survey (formerly called “wiki survey”) is one of three quantitative methods used by the Early Warning Project to assess atrocity risk worldwide (see our statistical risk assessments and public opinion pool). Like the opinion pool, the comparison survey methodology draws on a “wisdom of the crowd” approach. Our survey was open to the public for the month of December 2018, and specifically promoted to experts, policymakers, NGOs, and scholars in international affairs. 

A pairwise comparison survey involves a single question—in this case, Which country is more likely to see a new episode of mass killing in 2019?—with many possible answers on which participants vote, one pair at a time. The response options presented to participants (all are countries with a population greater than 500,000) are randomly generated at the beginning of the survey period. As the survey progresses and more people record responses, the survey mechanism “learns” from previous responses and presents pairings that are close in risk estimation as the survey progresses. 

The results are mapped below. The darker the shade of red, the greater the risk that a new episode of mass killing will begin in 2019, according to survey respondents. See all results on All Our Ideas. 

Comparison Survey results: EWP data, Heatmapper

The results from our annual comparison survey do not contribute directly to our statistical risk assessments, but they do inform the selection of cases we track in real time with our public Opinion Pool. We are currently tracking 17 countries, 16 of which scored in the top 40 in the comparison survey. To participate in regular forecasting, and see results in real time, create an account on Good Judgment Open.

Methodology

We used All Our Ideas to conduct this pairwise comparison survey. For more on how a pairwise comparison survey works, see this working paper by the instrument’s creators. For details on what we mean by “mass killing” and “onset,” see the blog post that led respondents to the survey. We launched the survey with that blog post on December 1, 2018 and closed it on December 31, 2018. 

We do not know the identities of our survey respondents. Participation in All Our Ideas surveys is anonymous, and our blog post was shared publicly. However, we do know that participants cast 5,687 votes in 90 unique user sessions. Some individuals likely voted in more than one session, so the total count of respondents is some unknown figure less than or equal to 90. 

The map below shows that, as in previous years, most of the votes in our comparison survey came from the United States and Europe. We welcomed votes from Asia and East Africa as well, but we are eager to attract more participation from those and other regions in future surveys. Note that the locations of 38 of the voters could not be located.