Is electoral violence in sub-Saharan Africa overreported? This new book looks at the data.

August 12, 2016

By Kim Yi Dionne and Stephanie Burchard* [caption id="attachment_31950" align="alignleft" width="300"] Supporters of President Edgar Lungu gather in Zambia’s capital, Lusaka, a few days before the Aug. 11 election. (Gianluigi Guercia/AFP/Getty Images)[/caption] In the run-up to Zambia’s election Thursday, there were violent clashes between supporters of the ruling Patriotic Front party and the main opposition party, the United Party for National Development (UPND). In one incident in July, police opened fire on opposition protesters, killing a UPND supporter.Zambia’s electoral commission temporarily suspended campaigning in the capital after that incident — and reports of other violence — claiming that doing so would quell electoral violence. How typical is this contemporary Zambian example of elections in Africa? What do we know about electoral violence in Africa? In this week’s African Politics Summer Reading Spectacular post, we feature a Q&A with Stephanie Burchard, author of “Electoral Violence in Sub-Saharan Africa: Causes and Consequences.” In her book, Burchard offers case studies of electoral violence in Kenya, Liberia and Senegal, and also analyzes data measuring election violence across sub-Saharan Africa from 1990 to 2014. Kim Yi Dionne: Early in your book, you distinguish between two types of violence: incidental and strategic. You say that the key distinction between the two is that incidental violence is not planned in advance (p. 28). Given the social norms against violence, isn’t it hard to know if violence was premeditated by political actors? As a researcher, how do you get around that obstacle of not knowing what’s in politicians’ heads? Stephanie Burchard: As I wrote this book and dug deeper into specific cases of election violence, it became clear that not all election-related violence was the same. In terms of both intensity and underlying motivation, there is a lot of variation across sub-Saharan Africa. The violence that took place before Senegal’s 2012 election, for example, looked very different from the violence that usually (but not always) accompanies elections in Nigeria and Kenya. One of the key differences was the planning and purpose of the violence. In Senegal, protesters were angry with the incumbent Abdoulaye Wade regime for manipulating the election in an attempt to remain in power beyond the two-term constitutional limit. These protesters engaged with state security forces for several weeks. Unfortunately, some protests turned violent. The government may have given security forces tacit support to respond with a heavy hand, but I found no evidence that violence was planned by either protesters or the state. This is the type of violence I categorize as “incidental.” Anecdotal evidence, however, also supports the notion that unscrupulous politicians have deliberately organized violence to affect the outcome of elections in several countries in Africa. Post-election analyses in Kenya, Nigeria and Zimbabwe, for example, have established that violence was planned months in advance as part of an overall strategy to ensure victory at the polls. Parties and/or politicians have hired youth groups for the explicit purpose of intimidating potential voters. In the case of Zimbabwe, for instance, youth groups were trained to bully voters to back Robert Mugabe in the March 2002 elections. This is the type of violence I term “strategic.” Measuring intent is obviously very difficult, especially when actors have a vested interest in disguising their motives. This is a fundamental shortcoming of any research into criminal behavior. Thus, the difference I make between incidental and strategic violence is more of a theoretical distinction than a strict typology. This is not to say it is unimportant — I think it is an important distinction to make in understanding the persistence of electoral violence and its effects — but the categories are admittedly fuzzy. Further complicating matters, both types of violence may occur in the same election, as the categories are not mutually exclusive. This is why I chose in my book to employ a combination of quantitative and qualitative analysis as a sort of quality check to ensure my assumptions and measurements accurately reflected reality. KYD: To be honest, I was surprised to learn from your book that so many African elections were violent. According to your data, 57 percent of the 289 elections held in sub-Saharan Africa between 1990 and 2014 experienced pre-electoral violence — meaning violence that occurred prior to election day. When I think of election violence, I’m imaging what you characterize as the most severe, “Category 3” cases, like in Côte d’Ivoire in 2010-2011, when 3,000 people were killed. Or I think of the election that opens your investigation: Kenya in 2007-2008.

However, the way you define the different types of pre-electoral violence suggests it’s broadly inclusive (pp. 35-36). For example, the most frequent form of violence is “violent harassment,” which was present in 37 percent of the elections you studied and includes instances of candidate disqualification and confiscation of opposition newspapers. The “violent harassment” category of electoral violence also includes party supporters fighting — and under that criteria, the 2016 American election campaign would be coded as violent. (And maybe rightly so.)

Still, does this greater frequency of lower-level harassment lead people to overestimate the proportion of “violent” elections in Africa? SB: First, I do think that certain violent events surrounding the 2016 U.S. election are cause for concern. And I think a better understanding of African experiences with election violence would help us understand what is taking place in our own elections, what steps could be taken to mitigate any problems, and what the long-term consequences for democracy might be if no action is taken. But I am not an expert on American politics, so I will leave that discussion and analysis to my more qualified colleagues. Does my categorization overstate violence in African elections? No, I would argue that my analysis is a more nuanced treatment of electoral violence than we have seen to date. Disaggregating election violence into multiple categories provides us with more analytic leverage than if we applied a simple binary assessment of “violent or not violent” to an election and set a minimum threshold of protests, injuries or fatalities. By looking at changes in the type and intensity of electoral violence over time, we can closely track not only any decline in electoral quality but also improvements — and we can assess what factors might account for changes in either direction. I’d like to emphasize that 43 percent of elections held in Africa are peaceful, which is commendable. And, as you mention, the majority of elections that I identify as violent are of the less intense variety. These elections are characterized by harassment and intimidation from state security forces, hired thugs and/or party functionaries. Opposition candidates or supporters may be detained, but only briefly. Fights within and between partisan supporters break out but are contained relatively quickly. Anti-government newspapers are shuttered but generally resume publication shortly thereafter. These activities might not escalate to the level where fatalities occur, but they still are coercive. I think that any violence or coercion has the potential to be disruptive to democratic development, especially if it is recurring and can directly be connected to the electoral process. High-profile election violence, such as has occurred in Kenya and Côte d’Ivoire, typically garners attention and international intervention. But less intense forms of violence, like the events that have taken place during the recent elections in Zambia, are just as important to monitor and understand. Left unaddressed, lower levels of violence can be a precursor to major violent episodes in future elections. The 2007 post-election violence was not Kenya’s first experience with turbulent elections. Violence had accompanied all multiparty elections in Kenya since 1992; only the intensity and perpetrators have varied over time. Electoral harassment or intimidation, my Category 1, might be common, but that doesn’t mean something can’t or shouldn’t be done about it. KYD: In your case study on Liberia focusing on the 2011 elections, you mention the work of Ushahidi Liberia. Ushahidi is a nonprofit technology organization developed to map reports of violence in Kenya after the 2007 election. It has since worked in or developed platforms used in multiple crisis situations, including other African elections and the aftermath of the Haiti earthquake. You shared some of Ushahidi Liberia’s crowdsourced data on 2011 electoral violence with caveats. These caveats included potential for citizens to over- or underreport violence, the low Internet penetration in Liberia, and the lack of cellphone infrastructure outside the capital city of Monrovia. As cellphone coverage and Internet accessibility continues to grow on the continent, what potential do you see in using future Ushahidi data for scholars and students studying election irregularities and electoral violence? SB: Honestly, the potential is limitless, and the innovative work that groups such as Ushahidi are doing is invaluable. But, unfortunately, less-than-democratic governments have also discovered this dissent channel and are working to frustrate efforts at crowdsourced reporting on elections. Just within the past six months, Uganda, Ethiopia, Congo-Brazzaville and Chad have all imposed social media blackouts during elections, under the guise of national security. Even Ghana, arguably one of the more mature democracies in Africa, recently floated the idea of shutting down social media during its upcoming December 2016 elections to prevent the promulgation of hate speech and dampen the potential for violent mobilization. As long as governments have and are willing to exercise control over information and communications technology in this manner, researchers will have to continue to take extra care to ensure that any data they analyze is representative, reliable and externally valid. KYD: Finally, I think exploring the data you analyze might be of interest to some of our readers. Whether they accept the arguments you made or have ideas for other potential factors shaping election violence, they can look and see for themselves what the data say. Is there a website where people could download the data or analyze it online? Or related sites that you think readers would find interesting? SB: I don’t have a website, but am preparing my data set to be housed online at the Harvard Dataverse. In the interim, I am more than happy to make my data available to anyone who would find it useful — people are welcome to contact me on Twitter. I am a big fan of transparency and replication. I’d love to see others use my data to continue furthering our understanding of the causes and consequences of electoral violence in Africa. Or anywhere for that matter (cough *United States* cough). *Washington Post.Stephanie Burchard is a research staff member in the Africa Program at the Institute for Defense Analyses and occasionally teaches courses at Georgetown University and American University. You can follow her on Twitter at @smburchard.