Members of civil society groups protest the killing of Christopher Msando, electoral commission information technology manager, at a demonstration in Nairobi on Tuesday. Msando, an official crucial to running Kenya’s presidential election, was found tortured and killed, the electoral commission chairman said Monday, as concerns grew that the East African nation’s vote could again face dangerous unrest. (Ben Curtis/AP)

On Aug. 8, Kenyans will cast their ballots for about 1,880 positions, including president and vice president. Over the past year, the electoral process has been marred by violence perpetrated by politicians, party agents, protesters and security forces. Will this affect voters’ decisions to turn up at the polls?

Surprisingly, violence before an election doesn’t drive down the vote overall

As one of us, Stephanie Burchard, found, election violence occurs in roughly 50 to 60 percent of sub-Saharan Africa’s elections. Most incidents tend to be violent harassment and intimidation of voters and candidates, although about 20 to 30 percent of elections experience politically motivated assassinations. Some violence is strategic — deliberately organized to affect the outcome of the election — whereas other violence is incidental, breaking out as result of protests that devolve into riots. Incumbent parties and politicians usually instigate strategic election violence, but opposition parties also mobilize election violence.

When observers talk about election violence in sub-Saharan Africa, they’re almost always discussing violence that occurs before elections. In fact, pre-election violence makes up roughly 90 to 95 percent of violence associated with elections. Fortunately, post-election violence is rare; however, when it does occur, as in Kenya in 2007, it tends to be much more severe, resulting in hundreds or thousands of fatalities.

But election violence doesn’t affect overall voter turnout in Africa. That’s our conclusion in a recent article: Pre-electoral violence does not necessarily deter citizens from turning out to vote. That contradicts the thinking of both leading development institutions such as USAID and United Nations Development Program, and of political scientists such as Paul Collier, Pedro C. Vicente and Kristine Höglund, all of whom have presumed that such violence generally suppresses voting.

At first glance, our finding doesn’t make sense. Coercion and intimidation should scare voters away from elections that they fear will turn violent. After analyzing almost 300 elections held in sub-Saharan Africa from 1990 to 2014, however, we find that there’s really no systemic difference in turnout between elections where campaigning was peaceful and elections where violence took place.

Politicians typically use violence before an election for one of these three reasons

Here’s what we concluded. There are three primary reasons that politicians use violence during a campaign: suppression, mobilization and displacement. Politicians might be using violence to dissuade citizens from voting for a specific party or candidate, to try to compel them to vote for a specific party or candidate, or to displace them so they do not vote at all. Furthermore, if used at the same time, these violent strategies can effectively cancel one another out in the aggregate — and so the overall effect on voter turnout becomes difficult or impossible to determine unless it is extremely well coordinated.

We are not arguing that election violence is necessarily an ineffective strategy — it may have the intended results in targeted communities — but it is inefficient and a very difficult way to produce results that matter at the national level. By understanding that politicians have multiple motivations for violence, observers are better positioned to develop programs and interventions to mitigate and prevent these attacks.

Since Kenya became a democracy, pre-election violence served all three of these purposes, as we highlight in the elections of 1992, 1997, and 2007. In 1992, the Daniel arap Moi regime orchestrated clashes between Kalenjin and non-Kalenjin ethnic groups; more than 1,500 people died and about 300,000 were displaced. As a result, far more Kalenjin – who generally supported arap Moi’s government – went to the polls, while tens of thousands of opposition voters could not vote.

In 1997, local politicians and armed supporters of the ruling Kenya African National Union (KANU) in Coast Province attacked and drove out ethnic Kikuyu, Kamba, Luo and Luhya, seen as anti-KANU. Human Rights Watch documented that the armed groups mobilized support for KANU by turning local grievances into anger against these perceived “outsider” ethnic groups and promising land, property and jobs to those who helped chase them out.

The Waki Commission report after the 2007 election showed that supporters of opposition leader Raila Odinga tried to force Rift Valley citizens to vote for him; others – very often Kikuyu – were forced to flee if they refused. In other regions, such as Central Province, pro-Kikuyu armed groups attacked opposition supporters, intending to prevent them from casting their ballots.

So what’s been happening in the lead-up to Kenya’s election this weekend?

As the August 2017 polls approach, a familiar pattern has emerged. During the May and June primaries, rival candidates and their supporters harassed, intimidated and assaulted each other. In 19 percent of the 224 polling stations monitored by the Kenya National Commission on Human Rights (KNCHR), candidates and their supporters used violence to eliminate political opponents, prevent other candidates from running, disrupt voting and intimidate those considered outsiders. In other reports, politicians in drought-stricken Laikipia in the Rift Valley encouraged herders to invade farms in exchange for money and the promise that they could keep the land in exchange for supporting the sponsoring politician.

According to our research, those responsible for administering the upcoming Kenyan elections can prevent more violence and intimidation prior to the election (and in future elections) in several ways.

1. Secure the anonymity of the voting process. No constituent should worry that his or her vote can be known, and therefore suffer punishment as a result. A case in point: the announcement by Odinga’s National Super Alliance (NASA) that it will have five party representatives at each polling center, ostensibly to observe the electoral process, could make voters uneasy — especially supporters of opponents. Thus, clear rules for the number of party agents at polling stations, for example, can reduce voter anxiety.

2. Make it easier for displaced citizens to vote at the nearest polling center. Displaced voters may find it impossible to cast their ballots at the polling place where they are registered. Currently, voters are allowed to change their polling station several months before the vote. Allowing more flexibility for displaced voters might reduce violence and harassment close to election day.

3. Punish perpetrators. Ultimately, perpetrators of electoral violence must be punished. Following the party primaries, more than 60 people were arrested for various electoral offenses — including a few politicians. However, the KNCHR report seems to suggest many more political aspirants were involved in violence. To date, very few, if any, perpetrators of violence from previous elections have been held to account. Kenya’s poor record of prosecuting and sentencing electoral offenders makes violence a viable campaign tactic.

Dorina A. Bekoe, PhD, is an associate professor at the Africa Center for Strategic Studies in Washington. She is editor of Voting in Fear: Electoral Violence in Sub-Saharan Africa (USIP, 2012).

Stephanie M. Burchard, PhD, is a research staff member at the Institute for Defense Analyses in Alexandria, Va. She is the author of Electoral Violence in Sub-Saharan Africa: Causes and Consequences (Lynne Rienner, 2015).