Voting, fraud, and violence : political accountability in African elections
- Author(s): Long, James Dunaway
- Long, James Dunaway
- et al.
My dissertation examines whether and how elections in Africa's emerging democracies support political accountability. Elections in developing countries often fail to meet acceptable standards of fairness, and this in turn can result in protest, violence, and fragility. Prior studies argue that ethnic fragmentation may produce these outcomes, but an "ethnic headcount" election in many African countries does not yield any one group's winner-- candidates must appeal to voters beyond their co-ethnics. I take a different approach and argue that Africans utilize a diverse set of information sources beyond ethnicity, including cues about performance and policy preferences, to demand accountable representation. In spite of these attempts, politicians frequently curtail legitimate electoral practices through corrupt manipulation of the vote, particularly when they perceive voters are likely to unseat them. Fraud creates incentives for violence between government security forces and opposition party supporters, activating a security dilemma between and within communities that encourages further fighting. Using the case of Kenya's 2007-08 election crisis, I test my argument focusing on three substantive areas : voting behavior, electoral fraud, and electoral violence. In Kenya, the fraudulent 2007 election sparked protests and wide-spread violence resulting in 1,500 deaths, 700,000 displaced people and the reversion of democratic progress. Contrary to standard accounts, my findings demonstrate that Kenyans supported well- performing candidates, regardless of their ethnicity. But commitment problems guaranteeing a free and fair race and a lack of an independent third party electoral commission to support a credible electoral process allowed for wide- spread rigging. Fraud instigated post-election protest and generated violence between party supporters, security forces, and communities. Ethnic fragmentation on its own did not cause this outcome. Novel data for my dissertation come from an exit poll, household surveys, election forensics, event count violence data, and ethnographic research from two years of fieldwork