Making Democracy Safe: Explaining the Causes, Rise, and Decline of Coercive Campaigning and Election Violence in Old and New Democracies
69 Pages Posted: 13 Aug 2009 Last revised: 23 May 2013
Date Written: 2009
Ethnic grievances, socio-economic cleavages, past conflicts, and other macro-level factors associated frequently with political violence cannot explain subnational and cross-national variation in geographic and temporal patterns of coercive campaigning and post-election violence. Why do so many democracies — old and new, diverse and homogenous — experience election violence, often long after founding elections? Why does it occur in some electoral districts and not others? I develop a common set of explanations for this unique form of political violence, proposing why, where, how, and when parties and candidates risk reprisals, punishment, and reputational costs to influence elections through “undue influence.” Considering the array of available non-coercive strategies, such as negative campaigning, vote buying, and boycotting available to politicians, to name a few, the choice to use violence or to allow supporters to do so is a rare and, often, conscious choice. I aim to expand understanding of this phenomenon. First, drawing on historical case studies of particularly acute eruptions of election violence, I describe the enigmatic historical and contemporary patterns of election violence that contemporary explanations, which focus primarily on recent episodes, tend to overlook. It is study of these cases, as well as preliminary, theory-building research trips to observe both rounds of Indonesia’s 2004 presidential elections, on which I base my theory, while the collection of data and tests of this theory will be carried out independently and separately from this presentation of theory, hypotheses, and empirical expectations to minimize bias and report transparently and honestly when the empirical results are inconsistent with my initial propositions. While the theory shaped the research design and data collection, data has not been analyzed before full articulation of the theory in its current form.
I begin by developing a typology of election violence — an undertaking that responds to one of the first studies in political science to call attention to the topic (Rapoport & Weinberg, 2001b, p. 35). I follow by presenting a theory to explain the probability, spatial diffusion, typological variation, lethality, and long-term rise and fall of election violence in the electoral histories of many polities.
In the first of four propositions, I contend that parties and candidates are most likely to initiate or tolerate election violence when both uncertainty and incentives to cultivate a personal vote are high. Electoral uncertainty arises from exogenous sources of competitiveness. Institutional uncertainty arises following pro-democratic electoral reforms or tougher monitoring and enforcement of electoral corruption laws, increasing the costs of nonviolent fraud relative to violence. Personal vote incentives (a) minimize party desire and ability to control candidate campaign behavior and (b) maximize the number of actors willing to provide violence when faced with the imminent gain or loss of the private benefits that particular candidates target to loyal supporters.
Second, I argue that the timing, targets, perpetrators, number of people involved, and other components of the typology vary across electoral system families and regime type. At one extreme, in Closed List Proportional Representation Systems (CLPR), violence occurs primarily during the intra-party, pre-campaign and/or coalition-formation stages of competition and involves candidate sponsorship of violence against one another. At the other extreme, violence in First-Past-the-Post (FPTP) electoral systems occurs primarily during the inter-party campaign and election day phases, targeting voters and supporters. Following the posting of results, post-election violence tends to occur when the national distribution of competitive constituencies is such that at least one party and its supporters estimated apriori equal probabilities of winning or losing the ability to govern alone at the national level, a situation that can occur in both CLPR and FPTP systems. This and other types of violence tend to occur at adolescent stages of democratization, rather than primarily during founding elections.
Third, I suggest that ethnic grievances, socio-economic cleavages, past conflicts, and other correlates of deaths common in the broader political violence literature cannot predict when and where election violence occurs. I hypothesize instead that these predisposing factors determine the severity and lethality to which coercive campaigning and election violence escalate in particular countries and constituencies.
Fourth, I propose that election violence is endogenous to democratization. Cleaning up elections can, in the short term, increase incentives for competitors to engage in coercive campaigning and election violence. In turn, eruptions of election coercion can disrupt path-dependent, institutionalized electoral bias and fraud more by generating mass awareness of and demand for pro-democratic electoral reforms than can nonviolent election fraud alone.
The theory’s subnational and cross-national empirical expectations will be evaluated using three original datasets, informed by qualitative fieldwork in Algeria, Newark, and Pakistan. The Election Violence Incidents Database (EVID) includes narratives and micro-level coding of the features of election-related incidents reported in major national newspapers four months before and one month after elections in Algeria, Egypt, Ghana, Newark, Pakistan, and Sri Lanka. Variables include dates, geographic location (latitude-longitude), perpetrator and victim affiliations, and positions, the tactics used by each actor in each event (from vandalism to bombings), deaths, injuries, property damage, and electoral consequences, as well as an index of report reliability indicators. EVID encompasses elections since the 1960s, but initial analysis will include only two elections for each case.
The cross-national analysis employs the Global Violent Elections Database (GVED), which indicates whether each national election worldwide between 1890 and 2005 included violence, fraud, or both, along with number of injuries and deaths (for 1945-2005). I combine this data with that compiled by other scholars for uncertainty, ICPV, electoral systems, degree of democracy, electoral reform, and control variables such as ethno-linguistic diversity and poverty. Due to the lack of literature and measurement of causes of electoral reform, it is difficult to implement a dynamic model that addresses the mutually causal relationship between election violence and electoral reform. For a tentative test of the endogeneity hypothesis, I will use data on the timing of major electoral reforms and constitutional changes in neighboring countries during the previous five years as an instrument for electoral reform in a structural equation model.
I created the Election Laws on Election Crimes (ELECD) database, which codes national election crimes laws current as of 2005 (if specified in the constitution or electoral law and amendments) for nearly all countries in the world. I model the content, complexity, and levels of fines, penalties, electoral remedies for violent and nonviolent electoral crimes as a function of a country’s past experience with election violence.
In describing and explaining election violence and its institutional consequences, I contribute to research on micro-level variations in violence, the role of violence as a mechanism of institutional change, and the incremental process by which even flawed elections further democratization. My findings should interest practitioners involved in election observation, reform, administration, and security, as well as those involved in designing institutions in new democracies.
Keywords: election violence, endogenous democratization, democratization by elections, election fraud, electoral reform
Suggested Citation: Suggested Citation