Do Violent Leaders Harm their Own Cause? How the Use of Electoral Violence Advances Democracy

June 15, 2020

Do Violent Leaders Harm their Own Cause?

Master’s Thesis Word count: 9.845

Submitted by Megan van Melle

MSc Political Science

Specialization: International Politics

Leiden University, Faculty of Social and Behavioural Sciences s1789481

Submitted to Dr. S.P.A. Chauchard

Assistant Professor of Political Science

Leiden University, Faculty of Social and Behavioural Sciences

Second Reader: Dr. V. Tsagkroni

Lecturer in Political Science

Abstract

Authoritarian rulers frequently make use of violence or intimidation to increase their electoral chances. Upon re-election, these leaders however often witness a subsequent increase in the country's democracy score during their tenure. To explore possible mechanisms linking the use of electoral violence to the subsequent increase in democracy, this paper examines the 1986 presidential election in the Philippines and the legislative elections of Malawi in 1994, Benin in 1991, and Croatia in 2000. It considers three possible interpretations of the

correlation between the two variables. Firstly, the incumbent’s use of violence could lead to a backlash in the form of protest or negative feedback from the international community that generates a democratisation event. Secondly, the implementation of elections causes an increase in the country's democracy score, but could simultaneously motivate the incumbent to make use of violence. Lastly, the correlation might be weaker than expected or even non-existent as the result of a confounding variable or measurement error. The findings in this paper suggest that highly visible instances of electoral violence can aid in mobilizing a large group of people to protest, often leading to some form of democratisation. On the other hand, less visible forms of electoral violence, such as voter intimidation, are unlikely to lead to collective action. Moreover, this research shows that one of the variables in the dataset on which this paper builds is based on a measurement error. This suggests that the post-election increase in the democracy score is not related to the occurrence of pre-electoral violence.

Introduction

In 2001, the Zambian Movement for Multi-Party Democracy (MMD) cadres violently interrupted multiple campaign events (Burnell 2002). In the 2005 presidential election in Kyrgystan, the government harassed the opposition before the election (OSCE 2005). In Kenya’s 1997 elections, the president’s security apparatus violently clashed with opposition supporters before the election (Ajulu 1998). Although violence goes against the basic democratic principles, the incumbents all witnessed a significant increase in their country’s democracy score upon re-election. Hafner-Burton et al. 2016 even find that “violent leaders have a 25% chance of an increase in their regime’s democracy score during their tenure, compared to 8% of non-violent leaders” (p. 482). This thesis aims to explain this quantitative finding by answering the following question: To what extent is the subsequent increase in the democracy score related to the occurrence of pre-electoral violence?

To answer this question, this research conducted a small-n comparison of elections in the Philippines (1986), Malawi (1994), Benin (1992), and Croatia (2002). The process-tracing method was used to identify mechanisms that could explain the connection between the two variables in the individual cases.

Electoral violence can be defined as “acts or threats of coercion, intimidation, or physical harm perpetrated to affect an electoral process” (EC and UNDP 2011: 14). It is a tactic that is increasingly used, as a growing number of states are holding competitive elections (Birch and Muchlinksi 2017). Daxecker, Amicarelli, and Jung (2019) report that the Electoral Contention and Violence (ECAV) data set showed that in 50% of the elections, more than three violent incidents took place. Additionally, in 30% of the elections, the violence was deadly. These numbers demonstrate the importance of researching electoral violence (Birch, Daxecker, and Höglund 2020).

Moreover, research on the consequences of electoral violence has mostly focused on the effect on behaviour and attitudes on the individual level and the impact on existing conflict (Höglund 2009; Burchard 2015; Bekoe and Burchard 2017; Birch, Daxecker and Höglund 2020). Less has been written on how it affects a country on a systematic level, particularly its democratisation process. Moreover, the literature that does discuss this is often contradicting and mostly quantitative, providing little evidence for the actual mechanisms at work (Little and Tucker 2012; Hafner-Burton et al. 2013; Hafner-Burton et al. 2016; Taylor and Strauss 2017; Daxecker and Jung 2018; Harvey and Mukherjee 2018). By analysing the relationship between the use of electoral violence as an election manipulation technique and democratisation, this thesis provides a better understanding of the consequences of electoral violence and the conditions under which electoral violence can have a positive effect on the democratisation process, contributing to the literature on electoral violence, election manipulation, democratisation through elections, and the relationship between violence and democratisation.

The findings in this paper suggest that highly visible instances of electoral violence can aid in mobilizing a large group of people to protest, which could lead to a democratisation event. On the other hand, less visible forms of electoral violence, such as voter intimidation, are unlikely to lead to collective action. Moreover, the research shows that one of the variables in the dataset on which this paper builds is based on a measurement error. This suggests that the post-election increase in the democracy score is not related to the occurrence of pre-electoral violence. This finding implies that conclusions based on this particular variable might not be valid. One example of such research is that of Hafner-Burton et al. (2016), in which they find that incumbents who make use of electoral violence have a higher chance to win the elections. This correlation might be less significant when using a different dataset.

Theoretical Framework

Research on electoral violence has taught us that most electoral violence takes place before the election (Daxecker and Jung 2018; Bekoe et al. 2020) and that incumbents are more inclined to use violence in comparison to their challengers (Bekoe et al. 2020). Additionally, Hafner-Burton et al. (2013) find that incumbents are more likely to use violence as a tool to influence the election outcome when they believe there is a considerable chance they will lose and when there are little institutional constraints present. Hence, pre-electoral violence is more often used in countries where democratic institutions are new and, therefore, often weaker (Anderson and Mendes 2005). But is it useful for the incumbent to sponsor pre-election violence if there is a large change the country will become more democratic post-election?

2.1 The Effect of Pre-electoral Violence on Democratisation

Hafner-Burton et al. (2016) find that the use of violence as an electoral manipulation strategy increases the probability that the incumbent will be elected, but that this victory might be short-lived, as pre-electoral violence heightens the chance of post-election protests. These post-election protests make it five times more likely that the incumbent has to make concessions during their tenure (Hafner-Burton et al. 2016). The authors made use of Tucker’s (2007) explanation of how electoral fraud can lead to collective action to explain their findings. Tucker argues that the collective action problem explains when and why citizens protest. Each individual has a choice to participate or to not participate, which is dependent on the perceived costs, benefits, and the likelihood of success. One individual protesting would most likely lead to high costs in the form of punishment and a low chance of success as individuals are not likely to put much pressure on the government. Therefore, individuals often choose not to protest. However, when it becomes evident that the regime

committed electoral fraud, the perceived costs can seem much lower, as more people are aware of the problem making it easier to mobilize a large group. Additionally, when the outcome of an election is being questioned, the likelihood of success increases significantly, as the winners can be removed more legitimately. Hence, when it is clear that election manipulation has taken place, the costs are perceived to be lower than the benefits, and the likelihood of success is perceived to be higher making it is easier to mobilize a large group of people to protest.

That election fraud often leads to post-election protest is supported by the findings of Daxecker (2012). She finds that there is a higher chance of post-election violence when a fraudulent election is monitored by international election observers. She argues that these monitoring organizations are perceived as credible and that information published about election manipulation can motivate different actors to use violence post-election. Little, Tucker, and Lagatta (2015) also find that reports of fraud lead to an increase in the occurrence of protest. Hafner-Burton et al. (2016) state that this argument applies not only to election fraud but also to election violence.

In contrast, Harvey and Mukherjee (2018) find that voter intimidation does not increase the probability of post-election protest. They also build their argument on that of Tucker (2007) but add to it by arguing that the kind of manipulation tactic used determines the likelihood of post-election protest. Manipulation tactics that require a certain level of organizational and financial strength, such as vote-buying and voter intimidation, signal that the incumbent has power over these resources. This lowers the perceived chance of success. Administrative fraud, on the other hand, signals less information about the capabilities of the regime, which heightens the likelihood of protest. This suggests that electoral violence is not very likely to increase the likelihood of protest.

An explanation for these contrasting findings is that electoral violence can increase the likelihood of post-election protest when the violence is very visible, for example when the opposition leader is imprisoned. Bell and Chernykh (2019) look at the relationship between human rights violations and post-election protest and find that political imprisonment of the opposition increases the likelihood of post-election protest when the incumbent wins.

Additionally, Beaulieu (2014) argues that electoral protest takes place when negotiations between the incumbent and the opposition collapse. She explains that two conditions can lead to this collapse: “(1) misrepresentation of private information and (2) commitment problems” (p. 4). Incumbents can misrepresent the level of manipulation that they are willing to use, while the opposition has the information on how much manipulation they can tolerate. When the incumbent overestimates the opposition’s toleration of election manipulation, they might use too much manipulation in the eyes of the opposition, motivating the opposition to call for protest. Secondly, even when both sides have all the information, negotiations can still collapse due to credibility problems.

Furthermore, it might not be solely domestic pressure that makes the incumbent decide to reform. Many developing countries receive some kind of support from developed countries or Western or international institutions, such as financial aid (Levitsky and Way 2002). When these external actors know about the incumbent's use of electoral manipulation tactics, they might pressure the incumbent to democratize by, for example, threatening to stop this financial aid. Other options for external actors to pressure the incumbent government are other economic sanctions, shaming campaigns, exclusion from international forums, aid conditionality, and human rights prosecutions (Hyde 2011; Escribà-Folch and Wright 2015).

Moreover, with the increase in the use of election monitoring these external actors gain more information on whether the incumbent made use of violent manipulation tactics (Hyde 2011). The international benefits that accompany internationally certified elections

have motivated incumbent governments to invite election observers. However, many leaders keep making use of (strategic) manipulation to ensure their victory. These manipulation tactics are mostly covert but also include tactics that are overt but unlikely to be reported as fraud by election observers, such as intimidation of or using violence against voters (Schedler 2002; Beaulieu and Hyde 2009; Hyde 2011). However, as election observers are seen to give trustworthy reports, their inclusion of the incumbent’s use of election violence may increase the chance that external actors are motivated to pressure the incumbent government to reform or resign. The visibility and scale of electoral violence are again of influence in this kind of scenario.

Based on the abovementioned arguments, I expect that the incumbent’s use of electoral violence in the period before an election in the form of (overt) harassment of the opposition, or in the form of overt violence against voters, as opposed to covert voter intimidation, is likely to lead to a backlash that could to lead to an increase in the country’s level of democracy, especially when these acts are reported by election observers or are very visible in general. This could explain the subsequent increase in democracy after the occurrence of pre-electoral violence.

2.2 The Effect of Democratisation on Pre-electoral Violence

There is, however, an alternative explanation for the positive relationship between the use of pre-electoral violence and an increase in democracy. An increase in a country’s democracy level could also heighten the probability of the occurrence of government-sponsored pre-electoral violence.

Especially after the fall of the Berlin Wall (1989) and the dissolution of the Soviet Union (SU) (1991) the pressure from the European Union (EU) and the United States (US) to have freer and fairer elections intensified, as the SU could not provide an anti-democratic

counterbalance anymore (Beaulieu 2014). This pressure from external actors and the growing demand for legitimacy led many authoritarian rulers to start holding (multi-party) elections (Schedler 2002; Levitsky and Way 2002). This external pressure took on the same forms as the abovementioned ways in which external actors can respond to manipulated elections. When these actions make the political costs too high, the incumbent might give in by allowing for true democratisation to take place (Lindberg 2009; Escribà-Folch and Wright 2015).

However, most incumbents do not want to leave the continuation of their rule up to faith. Therefore, many autocrats have tried to either control or distort the electoral process to ensure their victory by using electoral manipulation strategies (Schedler 2002). This could explain the correlation between the use of pre-electoral violence by the incumbent and the increase in the level of democracy. The democracy level goes up as elections are held for the first time but simultaneously the implementation of elections motivates the incumbent to use violence to influence the outcome of the election.

2.2 Confounding Variables and Measurement Errors

Another option is that there is no significant interaction between the occurrence of pre-electoral violence and the subsequent rise in a country’s democracy score. An increase in democracy can, for example, be the result of a conflict settlement within the country.

There could also be a confounding variable that leads to a post-election backlash. For example, the incumbent might have used both electoral violence and fraud. A backlash could be the response on the announcement that the incumbent used fraud, leaving their use of electoral violence without attention.

Lastly, the correlation between the two variables might be weaker than expected or even non-existent due to measurement error. When a variable is measured in a way that does

not reflect what the researchers aim to measure, the results do not support the argument made. For example, electoral violence is very difficult to measure. When there are no datasets available that are specifically aimed at electoral violence, researchers often search for variables that can serve as a proxy. However, these variables that are then used might be too far removed from what the authors try to research. Taking a closer look at the individual cases might reveal such errors.

Methodology

3.1 Research Method

To find the mechanisms that could explain to what extent the occurrence of pre-electoral violence and the subsequent increase in democracy are connected, this research has conducted a small-N study of four cases. Within these cases, I made use of theory-building process-tracing.

A small-N comparison allows for the identification of multiple mechanisms connecting the dependent and independent variables. One unit of analysis would not be sufficient to estimate causal effects, nor to estimate probabilities (Lijphart 1971; Gerring 2004). Additionally, a finding is more robust when it holds for multiple cases with different conditions and/or in a different context. Nevertheless, adding more cases to the analysis was simply not possible within the limits of time, space, and resources (Lijphart 1971; Blatter and Haverland 2012).

Moreover, a large-N study does not give us the needed insight into the interaction between X and Y. Large-N studies provide a correlation with a possible explanation for why this correlation exists, but this is not sufficient to answer the question as to how X and Y interact. An in-depth case study design allows us to see the intermediate causes that lie between the dependent and independent variables (Gerring 2004).

Furthermore, a study that analyses a limited number of cases allows for data triangulation. The use of multiple sources to measure one variable will increase the validity of the data. In a large-N study, this would be too time-consuming (Blatter and Haverland 2012).

For the in-depth analysis of the four cases, the research made use of the method theory-building process-tracing, which “involves building a theory about a causal mechanism between X and Y that can be generalized to a population of a given phenomenon, starting from a situation where we are in the dark regarding the mechanism” (Beach and Pederson 2012: 11). Although we are not completely in the dark about possible mechanisms connecting X and Y, no theory or hypothesis aims to explain this correlation. Theory-building process-tracing is used when there is a correlation between X and Y but no available explanation about what links the two, nor do we know what the intermediate steps are in the process that is taking place between the two variables.

The first step is to investigate the empirical record, using a wide range of sources. When this has been done, the gathered evidence is used to infer an underlying causal mechanism. The evidence does not identify the causal mechanism by itself. Often, scholars look for inspiration in existing work and observations to assist in generating possible hypotheses that are generalizable under certain conditions (Beach and Pedersen 2012; Bennet and Checkel 2014).

In this paper, the case study analyses will first present the context and chain of events surrounding the election. It then focuses on the role electoral violence played in the chain of events, especially on the question of to what extent the post-election increase in democracy was influenced by the occurrence of pre-electoral violence. The second part of the case analysis refers back to the theoretical framework to identify possible mechanisms.

3.2 Case selection

This research uses the reproduction files of the paper of Hafner-Burton et al. (2016) for the case selection. Within their dataset, cases were selected that met specific conditions: the occurrence of pre-electoral violence is organised or pushed for by the incumbent; the incumbent who used this tactic must have won the election, and there was a subsequent increase in the democracy level during the incumbent’s term. The variables Hafner-Burton et al. (2016) used to code these conditions are pre-electoral violence, Incumbentwins, and

polity2. After creating a list in which the first two conditions were met, the original database

was put next to this list to see whether polity2 had increased by the next election, i.e. at the official end of the incumbent’s term. This was done on a case by case basis. Out of the list of all the cases that met these conditions, the four cases with the largest increase in polity score were selected. This allowed me to find a clear reason for the increase, as the reason behind the smaller increases might not be detectable. In the cases with a high increase in the democracy level, it is easier to trace back the processes that led to the increase. Furthermore, picking the four cases with the largest increase also ensures that there is no cherry-picking in the case selection.

Year Type of

Election Democracy Score Before Election Democracy Score at the next Election Increase in Democracy Score Philippines 1986 Presidential -6 8 +14 Malawi 1994 Legislative -9 6 +15 Benin 1991 Legislative -5 6 +11 Croatia 2000 Legislative -3 8 +11

Case Analyses

4.1 The 1986 Presidential Election in the Philippines

Context

After being the president for seven years, Ferdinand Marcos declared martial law in 1972, marginalizing the oligarchic elite by only providing patronage to a small number of regime insiders. This increased elite opposition to his rule, especially when the economic conditions deteriorated during the 1981 financial crisis (Putzel 1999; Shock 1999; Lee 2009; Rice 2010; Fukuoka 2015). Meanwhile, this elite, the Catholic Church, and the US became worried about Marcos’ inability to repress the mobilization of the Communist Party of the Philippines (CPP). The continuation of the Marcos regime started to pose a serious threat to their interests (Shock 1999; Lee 2009; Rice 2010; Fukuoka 2015).

Moreover, while the situation in the Philippines was worsening, Benigno Aquino Jr., one of the former opposition leaders, returned from the US in 1983 to continue his campaign against the regime. He was shot before his feet touched the ground. The assassination was perceived to have been ordered by Marcos, which made Aquino a national martyr and caused widespread protests. Additionally, the public's response to the assassination made the oligarchic elite fear that the continuing changes might lead to an uncontrollable revolution, which also would not be in their interest. This fear, together with the communist threat, led the US and the Catholic Church to promote non-communist alternatives to the Marcos regime, which led to the establishment of the National Citizens’ Movement for Free Elections (NAMFREL) in October 1983. This organization brought together different forces in civil society and would later monitor the 1986 election. Moreover, the assassination led the opposition to gather momentum around Aquino's widow, Corazon Aquino. She would become the main opposition candidate in the 1986 election. Lastly, it also led the public to shift from passive acceptance to actively opposing the regime. Before Aquino's assassination, there had been no anti-Marcos demonstrations, the funeral procession of Aquino can be seen as the first (Timberman 1987; Putzel 1999; Shock 1999; Lee 2009; Rice 2010; McGeown 2011; Fukuoka 2015).

US pressure and growing domestic tension led Marcos to organize snap elections in 1986 (Fukuoka 2015). To ensure he would win the election he made use of violence, vote-buying, and fraud (Lee 2009; Rice 2010; Fukuoka 2015). However, when Marcos announced his victory the election-monitoring body NAMFREL exposed his use of fraud and declared Aquino had won, resulting in a wide-spread campaign of civil disobedience (Rice 2010; McGeown 2011; Fukuoka 2015).

Simultaneously, a group of prominent military officers openly defected and staged a coup (McGeown 2011; Fukuoka 2015). When Marcos ordered his troops to use violence against them, the Catholic Church called on the people to support the rebels. This led to a massive non-violent protest. Marcos then ordered his troops to clear the streets with violence, but they refused to use force against unarmed citizens. Additionally, the US now also urged Marcos to step down. These developments led Marcos to leave the country and Aquino to become the new president of the Philippines (Timberman 1987; Putzel 1999; Lee 2009; McGeown 2011; Fukuoka 2015).

The Role of Electoral Violence

The assassination of Mr. Aquino played a significant role in mobilizing the people of the Philippines to protest the regime, eventually leading to a significant increase in the country's democracy score. The assassination in combination with the exposure of Marcos' use of fraud were the main causes of the post-election protest. On the other hand, Marcos' use of voter intimidation might have been part of the individuals' motivation to protest but it has not significantly increased the likelihood of protest.

The assassination of Aquino led to a "collective outpouring of grief and anger in the streets of Manila", with about two million people attending his funeral procession (Rice 2010: 272). This can be seen as the first public showing of support for the opposition. This, in

turn, showed the oligarchic elite, the US, and the Church that the situation could lead to a revolution that they could not control, resulting in their promotion of non-communist alternatives and the establishment of the NAMFREL (Kessler 1986; Rice 2010; Fukuoka 2015). Moreover, it made it possible for the opposition to rally around Aquino’s widow, creating a new, broadly supported candidate in the 1986 election (Rice 2010; Fukuoka 2015; BBC 2016). Lastly, it created a general idea of political illegitimacy (Putzel 1999; Schock 1999). These developments increased the likelihood of a backlash in the form of protest significantly, as broad public support for the opposition signals that the cost of collective action will most likely be low, and the perceived illegitimacy increases the perceived likelihood of success. Although these consequences increased the likelihood of protest, the assassination did not solely cause the People Power revolution. It was the announcement of election fraud and the call for action from the Church leaders to aid the military officers that were the direct causes of the collective action that led to the fall of the Marcos regime. Yet, the existence of NAMFREL and the support from the Church are mainly consequences of the assassination and the public's reaction to it.

This argument supports the expectation based on previous literature that the incumbent’s use of electoral violence before an election in the form of harassment of the opposition increases the likelihood that there will be a backlash that could lead to an increase in the country’s level of democracy. The assassination of Aquino was a very visible attack against the opposition, which supports the argument that electoral violence can increase the likelihood of post-election protest when the violence against the opposition is very visible. The reason for this is that the visibility signals that a high number of people are aware of it, increasing the likelihood that the protest will draw more people. This in turn lowers the perceived costs and heightens the perceived chance of success of protesting. This makes it more likely that the collective action problem will be overcome.

Additionally, the argument confirms the findings of Daxecker (2012), Tucker and Lagatta (2015), and Harvey and Mukherjee (2018), who argue that electoral fraud heightens the chance of post-election protest. Moreover, the occurrence of voter intimidation during the pre-election period is not mentioned as a cause of collective action. Although I do not rule out that it was part of individual motivations to protest, voter intimidation did not lead to the collective action that took place. This supports Harvey and Mukherjee’s (2018) finding that voter intimidation does not significantly heighten the likelihood of protest.

Furthermore, one expectation that comes forth out of the discussed literature is that the incumbent’s use of electoral violence in the period before the election increases the probability that external actors will pressure the incumbent to democratize, leading to a higher level of democracy. This seems true for this case, as American pressure has played a large role in Marcos' final resignation (Kessler 1986; Lee 2009; Rice 2010). However, it was not Marcos’ use of violence that led them to pressure him to let go of the presidency but their fear of a communist take-over. It was generally known that Marcos made use of violence as a means of repression and many earlier elections have seen the occurrence of electoral violence, but the US did not condemn this use of violence (Lee 2009; Kessler 1986).

Hence, pre-electoral violence in the form of the overt assassination of an opposition leader has had a large influence on causing the post-election protest and the increase in democracy that followed, while voter intimidation did not play a significant role.

4.2 The 1994 Legislative Election in Malawi

Context

Two years after Malawi's independence in 1964, it officially became a one-party state under the rule of the Malawi Congress Party (MCP) led by Kamuzu Bandu. From 1989 onwards, multiple developments and circumstantial factors led to the country’s political liberalisation.

In 1989 purges and redeployment of the civil service, together with a deteriorating economy, created cross-regional resentment against the government (Ihonvbere 1997; McCracken 1998; Kaspin 1995; Venter 1995). In March 1992 eight bishops publicly criticised Banda, especially his disregard of human rights, in a pastoral letter that was widely distributed. This sparked riots, protests, and strikes throughout the country (Ihonvbere 1997; Kaspin 1995; Venter 1995). Additionally, the imprisonment of exiled pro-democracy activist Chakufwa Chihana on his return to Malawi in April 1992, led to demonstrations calling for multi-party politics and the end of the Banda regime (Ihonvbere 1997; Venter 1995). In response to these overt calls for multi-party politics, two main opposition parties were formed: The Alliance for Democracy (AFORD) and the United Democratic Front (UDF) (Van Donge 1995).

Simultaneously, the developments outside of Malawi pressured Banda to reform. During the Cold War, the West prioritized their struggle against communism, lending their support to rulers who were against left-wing movements. After the Cold War, the West started to shift their focus and started to set political and human rights conditions to foreign aid. In 1991 they made it clear that if Banda did not meet these conditions, foreign aid to Malawi would be frozen. This happened in 1993 (Ihonvbere 1997; McCracken 1998; Kaspin 1995; Venter 1995). At the same time, Malawi lost South Africa’s support as they started the process of ending apartheid, a wave of democracy was sweeping through the rest of the African continent, and pro-democracy movements were set up by exiled Malawians (Venter 1995).

In October 1992 Banda decided that repression of collective action was no longer feasible and announced that a referendum would be held to determine whether people wanted a multi-party democracy or retain the existing one-party system (Ihonvbere 1997; Van Donge 1995). This referendum was held in June 1993 and showed that, despite the use of widespread intimidation and harassment by the MCP, two-thirds of the voters had chosen for

a multi-party democracy. As a result, it was announced that the first multi-party elections would be held in May 1994 (Ihonvbere 1997; Venter 1995).

It was the UDF who won both the parliamentary and the presidential election. In the parliamentary elections, the UDF won 84 seats, the MCP 56, and AFORD 36. UDF's Bakili Muluzi won the presidential election with 46%. Banda had 33% of the votes and Chihana, the leader of AFORD, had 19,5% of the vote (ICJ 1994; Van Donge 1995; Ihonvbere 1997).

The main concern of human rights groups before the election period was that the MCP, in particular their youth group the Malawi Young Pioneers (MYP) and the Nyau group, would use violence and intimidation to increase their chance in the elections. Despite efforts to disarm the MYP, members of both groups have been reported to have used violence and intimidation during the election campaign by the United Nations’ Joint International Observer Group (JIOG) and the Independent Electoral Commission (Venter 1995). Both the UDF and AFORD were also involved in election violence and intimidation but the MCP and its followers were responsible for the majority of incidents (Venter 1995). However, according to both reports, these incidents did not have a significant effect on the election results. (ICJ 1994; Carver 1994; Van Donge 1995; Kaspin 1995; Venter 1995; Ihonvbere 1997).

The Role of Electoral Violence

The increase in democracy, in this case, can be attributed to the fact that it was the first multi-party election since 1964. Moreover, the ruling MCP used electoral violence, such as voter intimidation, to increase their chances in the upcoming election. Nevertheless, looking at this case in more detail proves that the incumbent did not win. It was the opposition party that gained most votes in the legislative elections even though the dataset coded incumbentwins as yes. The discussion on the context of the case does support the argument that the ruling

government will try to increase their chances in the first election by making use of electoral violence. Although a backlash is not likely in this case, as it was not the ruling party that won the elections. It also shows that external pressure played a large role in this democratisation process, as did domestic discontent with the country’s economic situation, and the open criticism on the regime by leaders of the Church.

We cannot, however, identify any underlying causal mechanisms because this case does not meet the conditions of the case selection. This also means that the correlation is weaker than it seems, as the correlation is not present in this case.

4.3 The 1991 Legislative Election in Benin

Context

After Benin gained independence from France in 1960, it began holding multiparty elections. However, the three ethno-regional based parties did not manage to come up with a stable division of power, which led to an intervention of the army. The period that followed was marked by multiple coups, creating military dictatorships. A coup in 1972 gave power to Major Mathieu Kérékou, who established a Marxist-Leninist regime in which all parties, other than his Parti de la Révolution Populaire du Bénin (PRPB), were banned (Gisselquist 2008; Morency-Laflamme 2015). Kérékou’s regime came under pressure in the 1980s, as a result of the region-wide economic recession. As the crisis worsened, the resources that members and supporters of the regime received began to dry up, lessening elite support (Gazibo 2005; Gisselquist 2008). In 1984, the regime could not pay its civil servants and was forced to lay off a number of them, again creating more discontent. Kérékou sought help from the International Monetary Fund (IMF) in the form of a structural adjustment loan. To receive this loan, the country first had to comply with the conditions of the loan, which included major economic reforms. These reforms were particularly ill-received by organized,

urban constituencies. After getting the first IMF loan in 1987, three state-owned banks collapsed in 1988, and in 1989 the government could again not pay the salaries of civil servants, including soldiers and teachers. These developments were perceived to be the result of government mismanagement. In response, strikes were held, which quickly turned into major protest (Morency-Laflamme 2015; Bierschenk 2009; Gisselquist 2008). When it became clear that the support of the elite and the military began to crumble and France began pressuring the regime to organize a national conference, Kérékou gave in (Houngnikpo 2001; Gisselquist 2008). A National Conference was held in February, during which a transitional government, led by Nicéphore Soglo, was appointed and a new constitution establishing multi-party elections was approved. Both legislative elections and presidential elections were planned to be held in 1991 (Kohnert and Preuss 1992; Houngnikpo 2001; Magnussen 2001; Gisselquist 2008; Morency-Laflamme 2015; BBC 2019).

According to Gisselquist (2008), there were two incidents of election-related violence. First, supporters of Kérékou attacked Soglo supporters in one instance and resulted in two deaths. This was, however, between the first and second round of the presidential election in March, while the legislative elections were mid-February. The database created by Hafner-Burton et al. (2016) states that there was also pre-electoral violence related to the legislative elections but such incidents are not mentioned in the reports. Secondly, some think the mysterious disease Soglo had during the campaign can be attributed to poisoning. Additionally, the New York Times reported some scattered incidents of intimidation and violence, such as some citizens in the north, a Kérékou stronghold, throwing stones as Soglo while he was campaigning. The same happened to Kérékou while he was campaigning in Benin’s capital, which is Soglo’s political base. However, there is little evidence that these instances of intimidation and violence are government-sponsored (Noble 1991). Despite these

incidents, electoral observers reported that both the legislative and presidential election were free and fair (Department of State 1992; Wiseman 1992; Battle and Seely 2007).

The party with the highest number of votes in the legislative election was the Union for the Triumph and Democratic Renewal, which was a coalition of pro-Soglo parties. The presidential election had two rounds and was eventually between Kérékou and Soglo. The latter won with 67,5% of the votes (Gisselquist 2008).

The Role of Electoral Violence

The increase in the democracy score after the 1991 elections can be attributed to the reintroduction of multiparty elections after almost 30 years of military dictatorships. The turn to multi-party elections was again highly influenced by external pressure and the economic situation. This suggests that the relationship between pre-electoral violence and the subsequent increase in the country's democracy score can be explained by the argument made in the theoretical section: when the democracy score goes up due to the implementation of elections, there is a high probability that the incumbent has used electoral violence during his campaign as a strategy to increase his or her chance of staying in power. However, we cannot be sure about this relationship between the variables in this case, as we encounter the same problem as in the Malawi case: that of a measurement error. All the cases in this research are selected on multiple conditions, of which one of them is whether the incumbent that used or sponsored pre-election violence also won the election. In the legislative elections of 1991, the ruling party did not win, even though the database did code it as a win for the ruling party. Hence, the variable incumbentwins does not tell us whether the previously ruling party won the election, even though Hafner-Burton et al. frame it like it includes both legislative and presidential elections.

Hence, the case of the Benin 1991 legislative elections cannot be seen as a case in which the government used violence and won. This increases the likelihood that the correlation is much weaker than expected, or even non-existent. This suggests that pre-electoral violence has no significant effect on the subsequent increase in democracy.

4.4 The 2000 Legislative Election in Croatia

The Context

After the death of Josip Tito in 1980 and the end of the Soviet Union's control of Eastern Europe, the support for parties with nationalistic programs increased. When the first free multiparty elections were held in 1990 the communist party lost to Franjo Tudjman and his conservative, nationalist party, the Croatian Democratic Union (HDZ). Tudjman and his party asserted the independence of Croatia in 1991, against the wishes of the Serb minority living in Croatia. Slobodan Milosevic took advantage of the situation and started a conflict that is known as the Croatian War of Independence. The Serb militants, with support of the Yugoslav military, seized almost one-third of Croatian territory. The war ended when the US and NATO intervened in early 1994. In 1995, Croatia retook the occupied areas (CSCE 2000; Bellamy 2001; Bunce and Wolchik 2011; BBC 2018).

While this conflict took place, Tudjman managed to secure more and more power at home, developing authoritarian tendencies. Before the elections of 1992, 1995, and 1997, Tudjman and his party changed the electoral regulations, harassed the opposition, intimidated voters, and used other forms of electoral manipulations to secure their win. Meanwhile, independent media was restricted, making it difficult to bring these manipulations to light (CSCE 2000; Bellamy 2001; Castaldo and Pinna 2019).

However, by the end of the 1990s, there was a shift in Croatian public opinion. Citizens felt more certain about the continuity of their country’s independence and territorial

integrity. Nationalistic sentiment decreased and people started to criticize the government’s corruption and cronyism. Other developments, such as the revelation of Croatian atrocities in the Bosnian War and the end of the Mečiar regime in Slovakia, changed the public mood as well. Additionally, minimal economic growth, high unemployment, continuing isolation from the rest of the world, as well as international criticism of the Tudjman regime created more dissatisfaction among the Croatian people. The opposition made use of the situation by pressing for reform, which, together with pressure from the EU and the US, resulted in an agreement on basic principles for holding elections. To minimize competition in the next election, the opposition formed two coalitions. The two largest parties, the Social Democratic Party (SDP) and the Croatian Democratic Community (HSLS), formed a coalition. The remaining opposition parties formed a coalition known as the Group of Four (Done 1999; CSCE 2000; Hedl 2000; Bellamy 2001; Castaldo and Pinna 2019).

Despite these changes, the ruling party remained largely authoritarian, increasing their efforts to manipulate the upcoming elections in their favour. However, when Tudjman’s health quickly deteriorated resulting in his passing away, the HDZ became unable to actively manipulate the election. This was a result of the inability of the HDZ to capitalize on his death and the fact that there was no clear successor, leading to disunity within the party (Bellamy 2001; Bunce and Wolchik 2011; Ramet and Goldstein 2019).

The results of the legislative elections, taking place on January 2, 2000, showed that the opposition coalition of the SDP and HSLS won with 41% of the votes. Together with 16% of the votes going to the Group of Four, the opposition had a majority in parliament (CSCE 2000; IRI 2000; Jasic 2000; OSCE 2000; Bellamy 2001; Bunce and Wolchik 2011).

The reports of the International Republican Institute, the Commission on Security and Cooperation in Europe, and the Organisation for Security and Co-operation in Europe Office for Democratic Institutions and Human Rights all deemed the election free and fair, looking

both at the period of campaigning and the election day itself. Only the broadcast media continued to give the ruling party more screen time and used a more positive tone of reporting when talking about the HDZ. Additionally, there were some isolated cases of intimidation towards voters and domestic observers, but this did not affect the result of the elections. (CSCE 2000; IRI 2000; Jasic 2000; OSCE 2000; Nadler 2000).

The Role of Electoral Violence

In this case the ruling party did not win despite the information given in the dataset as well. Therefore, it is again not a case in which the conditions for the case selection are met. This supports the argument that the occurrence of pre-electoral violence has little influence on the subsequent increase in democracy, as the correlation is based on a flawed dataset.

Moreover, the occurrence of pre-electoral violence is not very clear in this case. The first multi-party elections that took place in Croatia were in 1990, which were followed by elections in 1992, 1995, and 1997 (Jasic 2000; Castaldo and Pinna 2019). In all these elections, the ruling party made use of several manipulation techniques, among which harassment of the opposition and voter intimidation, to stay in power. However, the 2000 legislative election was different. In the words of one of the founders of the largest domestic NGOs that did the vote monitoring during this election, the election “represented a welcome departure from previous elections that had been widely regarded as “free, but not fair” (Jasic 2000: 160). Furthermore, the reports from independent bodies tell us that there were only a small number of incidents in which voters or domestic observers were the victims of intimidation. These incidents did not alter the outcome of the election and do not seem to have played a role in the subsequent increase in democracy. This also highlights a limitation of the database. The variable for pre-electoral violence is built upon nelda15 and nelda33, which are focused on government harassment of the opposition, including the media, and on

whether there were civilian deaths resulting from significant violence around the election time. In this case, the reason for coding this case as one in which pre-electoral violence occurred is most likely a result of the HDZ's control of the media, which is seen as harassment of the opposition. Yet, controlling the media is not a form of pre-electoral violence. Hence, this gives a distorted picture of how often pre-election violence in the form of harassment of the opposition and voter intimidation/violence against voters happens.

The main direct cause of the significant increase in democracy after the election, in this case, is the death of Tudjman. As the HDZ could not agree upon who would become Tudjman's successor, the space for opposition activity increased and made it almost impossible for the HDZ to manipulate the elections without impunity (CSCE 2000; Bellamy 2001; Bunce and Wolchik 2011). Therefore, they made no significant use of electoral violence, nor had the limited number of occurrences of pre-electoral violence any influence on the increase in democracy post-election.

Findings and Discussion

In a large number of cases, the incumbent government uses violence and intimidation to stay in power, which goes against the pillars of real democracy. However, in many of these cases, the democracy score of the country went up during the incumbent’s tenure. In some cases, this score even goes up significantly. The aim of analysing the four cases in more depth was to gain a better understanding of what leads to this increase in democracy and to what extent this was a result of the incumbent’s use of pre-electoral violence.

The analysis of the four cases teaches us multiple things. First of all, the Philippine case suggests that a highly visible attack on the opposition can mobilize people to protest. It also suggests that the use of violence against the opposition might not be enough to cause major protest by itself. The use of fraud during the election was also a significant cause of the post-election protest. This supports previous research, which identifies fraud as one of the

direct causes of post-election protest, especially when the election is monitored by election observed (Daxecker 2012; Tucker and Lagatta 2015; Harvey and Mukherjee 2018).

Additionally, this case suggests that Harvey and Mukherjee are correct in arguing that voter intimidation does not heighten the likelihood of protest. However, they argue that the government’s use of voter intimidation signals a high level of organization and financial strength. I would argue that visibility is mainly of significance here. In all four cases included in this research, the government was relatively weak, struggling financially, while still making use of voter intimidation. This counters Harvey and Mukherjee’s argument. The visibility argument would be in line with Bell and Chernykh’s (2019) finding that imprisonment of the opposition leader does increase the likelihood of protest. Moreover, the Philippine case shows that the highly visible assassination of a public opposition figure did lead to collective action.

The other three cases were, as was the Philippine case, selected on three specific conditions. As this research is based on the calculation of Hafner-Burton et al. (2016), stating that violent leaders are more likely to experience an increase of democracy during their tenure than non-violent leaders, this research has used their replication files together with the NELDA dataset on which their dataset is built, to conduct a non-biased case selection. However, the analyses of these cases prove that the variable 'incumbentwins’ does not actually refer to whether the ruling party won any legislative elections. The variable is built out of two variables of the NELDA dataset: What was the name of the leader who was in office before the election? And what was the name of the leader who was in office after the election? I assume that this refers to the president, who has not been replaced after the legislative election because in the cases included in this research the legislative elections took place before the presidential ones. This means that many of the legislative elections for which the variable ‘incumbentwins’ is coded as yes, the ruling party did not win. This might paint a

distorted picture of the correlation between the incumbent’s use of pre-electoral violence and a subsequent increase in democracy.

Moreover, the pre-electoral violence variable is based on two factors: harassment of the opposition and civilian deaths immediately before, during, or after the election. Included in harassment of the opposition is harassment of the media, by for example shutting down opposition newspapers. Whether this falls into the category of pre-electoral violence seems questionable. Additionally, Hafner-Burton et al. see voter intimidation as falling into the category of pre-electoral violence but only look at civilians in terms of the number of deaths. This exclusion of voter intimidation and the inclusion of harassment of the media in their

pre-electoral violence variable significantly affects how their findings should be interpreted.

Generally, the interpretation of Hafner-Burton et al. (2016) of what the measurements of the variables are telling us versus what they are actually measuring are quite far removed from each other. This highlights the limitations of statistical research based on variables build from data far removed from the topic under research. As a result, the answer to the question to what extent is the subsequent increase in democracy a consequence of the pre-electoral violence is that the increase in democracy is not significantly affected by the occurrence of pre-electoral violence. If Hafner-Burton et al. consider the variable ‘incumbentwins’ as giving us a clear answer to whether the incumbent party on the legislative election, the correlation is most likely based on erroneous information and does not exist.

A different case-selection strategy, resulting in a selection of cases in which all three conditions are present, might have altered the outcome of the research. However, this is very unlikely. To control for this, I have made three random selections of cases and checked whether the incumbent was re-elected during the election included in the database. In all three selections, there was at least one case in which the ‘incumbentwins’ variable was coded as yes but the incumbent party lost the election.

To find more reliable answers to these research questions, further research could perform similar tests making use of more reliable data. A newer dataset, the dataset for Electoral Contention and Violence (ECAV), includes more detailed information on electoral violence and could give us a better understanding of the relationship between electoral violence and democratisation (Daxecker, Amicarelli, and Jung 2019). Additionally,

Conclusion

By considering the impact of pre-election violence on the process of democracy, this thesis gives us a better understanding of the consequences of pre-electoral violence. It clarifies when and how pre-electoral violence can lead to a backlash by arguing that visibility and scale have a significant impact on whether collective action is feasible. This research also hints at other conditions, such as the economic situation and the presence of domestic and international pressure, under which the regime's use of electoral violence can have a positive effect on a country’s political liberalization. These findings can be of significance for governments and organizations that promote democracy. It points out the invisibility of voter intimidation, emphasizing the need for the presence of election observers during the pre-election period, who pay special attention to signs of these kinds of manipulation tactics. As intimidation of or violence against voters is often not seen as making an election fraudulent by election observers, democracy-promoting institutions should think about alternative ways in which incumbent governments can be demotivated to make use of such tactics.

Moreover, this research emphasizes the limitations of statistical research based on variables build from data far removed from the topic under research. Therefore, as argued for in the discussion, further research should perform similar tests with data of a higher quality. Additionally, comparative case studies could explore the conditions under which electoral violence can aid the democratisation process by for example comparing cases in which there

is an increase in the democracy score versus cases in which there is no increase or a decrease in the democracy score.

Bibliography

Ajulu, Rok. "Kenya's Democracy Experiment: The 1997 Elections." Review of African

Political Economy 25, no. 76 (1998): 275-85.

Anderson, Christopher J., and Silvia M. Mendes. "Learning to Lose: Election Outcomes, Democratic Experience, and Political Protest Potential." British Journal of Political

Science 36, no. 1 (2006): 91-111.

Battle, Martin and Jennifer C. Seely. 2007. “It’s All Relative: Competing Models of Vote Choice in Benin”. Working Paper No. 78. Afrobarometer. https://acrl.ala.org/IS/wp-content/uploads/2016/01/IS-Publications-Chicago-Manual-of-Style-Quick-Guide.pdf. Accessed 06/05/2020.

BBC. 2019. "Benin Profile". BBC News. https://www.bbc.com/news/world-africa-13040372. Accessed on 02/05/2020.

BBC. "Croatia Profile". 2018. BBC News. https://www.bbc.com/news/world-europe-17217954. Accessed 08/05/2020.

BBC. "How Filipino People Power Toppled Marcos". 2016. BBC News.

https://www.bbc.com/news/av/magazine-35526200/how-filipino-people-power-toppled-dictator-marcos. Accessed on 21/04/2020.

Beach, Derek., and Rasmus Brun Pedersen. Process-tracing Methods Foundations and

Guidelines. Ann Arbor: University of Michigan Press, 2013.

Beaulieu, Emily, and Susan D. Hyde. "In the Shadow of Democracy Promotion: Strategic Manipulation, International Observers, and Election Boycotts." Comparative Political

Studies 42, no. 3 (2009): 392-415.

Beaulieu, Emily. Electoral Protest and Democracy in the Developing World. West Nyack: Cambridge University Press, 2014.

Bekoe, Dorina A., Stephanie M. Burchard, Gabrielle Lynch, and Peter VonDoepp. "The Use of Electoral Violence." In Routledge Handbook of Democratisation in Africa, 258-72. 1st ed. Vol. 1. Routledge, 2020.

Bellamy, Alex J. "Croatia after Tudjman: The 2000 Parliamentary and Presidential Elections." Problems of Post-Communism 48, no. 5 (2001): 18-31.

Bell, Sam R., and Svitlana Chernykh. "Human Rights Violations and Post-Election Protest."

Political Research Quarterly 72, no. 2 (2019): 460-72.

Bennet, Andrew, and Jeffrey T. Checkel. Process-Tracing: From Metaphor to Analytical

Tool. West Nyack: Cambridge University Press, 2014.

Bierschenk, Thomas. "Democratisation without Development: Benin 1989–

2009." International Journal of Politics, Culture, and Society IJPS 22, no. 3 (2009): 337-57.

Birch, Sarah, and David Muchlinski. 2017. "Electoral Violence: Patterns and Trends". In

Electoral Integrity and Political Regimes: Actors, Strategies and Consequences,

100-112. New York: Routledge.

Birch, Sarah, Ursula E. Daxecker, and Kristine Höglund. "Electoral Violence: An Introduction." Journal of Peace Research 57, no. 1 (2020): 3-14.

Blatter, Joachim, and M. Haverland. Designing Case Studies: Explanatory Approaches in

Small-n Research. Research Methods Series. New York: Palgrave Macmillan, 2012.

Bunce, Valerie J, and Sharon L Wolchik. "The 1998 Elections in Slovakia and the 2000 Elections in Croatia: The Model Solidifies and Is Transferred." In Defeating

Authoritarian Leaders in Postcommunist Countries, 53-84. Cambridge: Cambridge

University Press, 2011.

Burchard, Stephanie M. Electoral Violence in Sub-Saharan Africa: Causes and

Consequences. Colorado: First Forum Press, 2015.

Burnell, Peter. "Zambia's 2001 Elections: The Tyranny of Small Decisions, 'non-decisions' and 'not Decisions'." Third World Quarterly 23, no. 6 (2002): 1103-120.

Carver, Richard. 1994. "Malawi: Between The Referendum and The Elections". Refworld -

UNHCR. https://www.refworld.org/docid/3ae6a6be8.html. Accessed 28/04/2020. Castaldo, Antonino, and Alessandra Pinna. "After Competitive Authoritarianism Hybrid

Regime Legacies and the Quality of Democracy in Croatia." Journal of Contemporary

European Studies 27, no. 4 (2019): 486-502.

Center for Systemic Peace. "The Polity Project". 2018. Systemicpeace.org. https://www.systemicpeace.org/polityproject.html.

Collier, David. “The Comparative Method.” In Political Science: The State of the Discipline

II, edited by Ada W. Finifter, 105-119. Washington, DC: American Political Science

Association, 1993.

Commission on Security and Cooperation in Europe. 2000. "CROATIA S

PARLIAMENTARY ELECTIONS". Washington: Commission on Security and Cooperation in Europe. https://www.csce.gov/international-impact/publications/report-croatias-parliamentary-elections. Accessed on 07/05/2020.

Daxecker, Ursula, and Alexander Jung. "Mixing Votes with Violence: Election Violence around the World." The SAIS Review of International Affairs 38, no. 1 (2018): 53-64. Daxecker, Ursula, Elio Amicarelli, and Alexander Jung. "Electoral Contention and Violence

(ECAV): A New Dataset." Journal of Peace Research 56, no. 5 (2019): 714-23. Daxecker, Ursula E. "The Cost of Exposing Cheating: International Election Monitoring,

Fraud, and Post-Election Violence in Africa." Journal of Peace Research 49, no. 4 (2012): 503-16.

Department of State. Country Reports on Human Rights Practices for 1991. U.S. Government Printing Office: Washington, 1992.

Done, Kevin. "Tudjman to Call Croatia Election Soon. (WORLD NEWS: EUROPE)." The

Financial Times, 1999.

Escribà-Folch, Abel., and Joseph G. Wright. Foreign Pressure and the Politics of Autocratic

Survival. Oxford: Oxford University Press, 2015.

European Commission, UNDP. 2011. "Thematic Workshop: Elections, Violence, And Conflict Prevention". Barcelona: Joint Taskforce on Electoral Assistance. Freedom House. "Benin". 2020. Freedom House.

https://freedomhouse.org/country/benin/freedom-world/2019. Accessed 02/05/2020. Fukuoka, Yuki. "Who Brought down the Dictator? A Critical Reassessment of So-called

'People Power' Revolutions in the Philippines and Indonesia." The Pacific Review 28, no. 3 (2015): 411-33.

Gazibo, Mamoudou. "Foreign Aid and Democratisation: Benin and Niger Compared." African Studies Review 48, no. 3 (2005): 67-87.

Gerring, John. "What Is a Case Study and What Is It Good For?" American Political Science

Review 98, no. 2 (2004): 341-54.

Gisselquist, Rachel M. "Democratic Transition and Democratic Survival in Benin." Democratisation: Deviant Democracies 15, no. 4 (2008): 789-814.

Hafner-Burton, Emilie M., Susan D. Hyde, and Ryan S. Jablonski. "Surviving Elections: Election Violence, Incumbent Victory and Post-Election Repercussions." 48, no. 2 (2018): 459-88.

Hafner-Burton, Emilie M., Hyde, Susan D., and Jablonski, Ryan S. "When Do Governments Resort to Election Violence?" British Journal of Political Science 44, no. 1 (2014): 149-79.

Harvey, Cole J., and Paula Mukherjee. "Methods of Election Manipulation and the Likelihood of Post-Election Protest." Government and Opposition, 2018, 1-23. Hedl, Drago. "Living in the Past: Frango Tudjman's Croatia." Current History 99, no. 635

(2000): 104.

Höglund, Kristine. "Electoral Violence in Conflict-Ridden Societies: Concepts, Causes, and Consequences." Terrorism and Political Violence 21, no. 3 (2009): 412-27.

Houngnikpo, Mathurin C. "Democratisation in Africa: Double Standards in Benin and Togo." The Fletcher Forum of World Affairs 25, no. 2 (2001): 65.

Houngnikpo, Mathurin C. "The Military and Democratisation in Africa: A Comparative Study of Benin and Togo." JPMS: Journal of Political and Military Sociology 28, no. 2 (2000): 210-229.

Hyde, Susan D., and Nikolay Marinov. "Which Elections Can Be Lost?" Political

Analysis 20, no. 2 (2012): 191-210.

Hyde, Susan D. "Catch Us If You Can: Election Monitoring and International Norm Diffusion." American Journal of Political Science 55, no. 2 (2011): 356-69. Ihonvbere, Julius O. "From Despotism to Democracy: The Rise of Multiparty Politics in

Malawi." Third World Quarterly 18, no. 2 (1997): 225-48.

International Commission of Jurists. 1994. "Parliamentary and Presidential Elections In Malawi". Geneva: International Commission of Jurists. https://www.icj.org/wp-content/uploads/2013/10/Malawi-elections-mission-report-1994-eng.pdf. Accessed 28/04/2020.

International Observer Delegation. 1986. "A Patch to Democratic Renewal". Washington, DC: International Observer Delegation.

https://pdf.usaid.gov/pdf_docs/PNABK494.pdf. Accessed on 21/04/2020.

International Republican Institute. 2000. "Republic of Croatia Parliamentary Election". Election Observation Mission Report and Recommendations. Washington.

https://www.iri.org/sites/default/files/fields/field_files_attached/resource/croatias_2000 _parliamentary_elections.pdf. Accessed 06/05/2020.

Inter-Parliamentary Union. 1994. "MALAWI: Parliamentary Elections National Assembly, 1994". Archive.Ipu.Org. http://archive.ipu.org/parline-e/reports/arc/2195_94.htm. Accessed 28/04/2020.

Jasic, Suzanna. "Monitoring the Vote in Croatia. (Election Monitoring)." Journal of

Democracy 11, no. 4 (2000): 159-168.

Kaspin, Deborah. "The Politics of Ethnicity in Malawi's Democratic Transition." The Journal

of Modern African Studies 33, no. 4 (1995): 595-620.

Kessler, Richard. "Marcos and the Americans." Foreign Policy, no. 63 (1986): 40. Kohnert, Dirk, and Hans-Joachim Preuss. "Vom Camarade Zum Monsieur:

Strukturanpassung Und Demokratisierung in Benin [From 'Camarade' to 'Monsieur': Structural Adjustment and Democratisation in Benin]." IDEAS Working Paper Series

from RePEc, 1992, IDEAS Working Paper Series from RePEc, 1992.

Lee, Terence. "The Armed Forces and Transitions from Authoritarian Rule: Explaining the Role of the Military in 1986 Philippines and 1998 Indonesia." Comparative Political

Studies 42, no. 5 (2009): 640-69.

Levitsky, Steven, and Lucan Way. "Elections without Democracy. The Rise of Competitive Authoritarianism." Estudios Políticos no. 24 (2004): 159-78.

Lijphart, Arend. "Comparative Politics and the Comparative Method." The American

Political Science Review 65, no. 3 (1971): 682-693.

Lindberg, Staffan I. "Democratisation by Elections? A Mixed Record." Journal of

Democracy 20, no. 3 (2009): 86-92.

Little, Andrew T., Joshua A. Tucker, and Tom Lagatta. "Elections, Protest, and Alternation of Power." The Journal of Politics 77, no. 4 (2015): 1142-156.

Magnusson, Bruce A. "Democratisation and Domestic Insecurity: Navigating the Transition in Benin." Comparative Politics 33, no. 2 (2001): 211-30.

McCracken, John. "Democracy and Nationalism in Historical Perspective: The Case of Malawi." African Affairs 97, no. 387 (1998): 231-49.

McGeown, Kate. 2011. "Long Road to Philippine Democracy". BBC News.

https://www.bbc.com/news/world-asia-pacific-12567320. Accessed on 20/04/2020. Morency-Laflamme, Julien. "A Missing Link? Elite Factionalism and Democratisation in

Africa." Canadian Journal of African Studies / Revue Canadienne Des études

Africaines 49, no. 3 (2015): 459-77.

Mydans, Seth. 1986. "Observers of Vote Cite Wide Fraud by Marcos Party". Nytimes.Com. https://www.nytimes.com/1986/02/10/world/observers-of-vote-cite-wide-fraud-by-marcos-party.html. Accessed on 21/04/2020.

Nadler, John. "Hope Springs in Croatia with New President: Election Seen as Clear End to Dark Days of Isolation and Violence." Edmonton Journal (Edmonton, Alta.), 2000. Noble, Kenneth B. "FREE VOTE IN BENIN ITS FIRST IN YEARS. (Foreign Desk)." The

New York Times, 1991.

OSCE. 2000. "REPUBLIC OF CROATIA PARLIAMENTARY ELECTIONS (HOUSE OF REPRESENTATIVES)". Warsaw: OSCE/ODIHR.

https://www.osce.org/odihr/elections/croatia/15668. Accessed 07/05/2020. OSCE. 2005. "INTERNATIONAL ELECTION OBSERVATION MISSION

PRESIDENTIAL ELECTION, THE KYRGYZ REPUBLIC, 10 JULY 2005". Bishek. Putzel, James. "Survival of an Imperfect Democracy in the Philippines." Democratisation:

The Resilience of Democracy Persistent Practice, Durable Idea 6, no. 1 (1999):

Ramet, Sabrina P, and Goldstein, Ivo. "Politics in Croatia since 1990." In Central and

Southeast European Politics since 1989, 273-308. Cambridge: Cambridge University

Press, 2019.

Rice, James A. "From Marcos to Arroyo – What Went Wrong for Democracy in the Philippines?" In Public Governance in Asia and the Limits of Electoral Democracy, edited by Brian Bridges and Lok Sang Ho, 253-283. Cheltenham, UK: Edward Elgar Publishing, 2010.

Schedler, Andreas. "Elections Without Democracy: The Menu of Manipulation." Journal of

Democracy 13, no. 2 (2002): 36-50.

Schock, Kurt. "People Power and Political Opportunities: Social Movement Mobilization and Outcomes in the Philippines and Burma." Social Problems 46, no. 3 (1999): 355-375. Timberman, David. "Unfinished Revolution: The Philippines in 1986." Southeast Asian

Affairs 14, no. 0 (1987): 239.

Tucker, Joshua A. "Enough! Electoral Fraud, Collective Action Problems, and Post-Communist Colored Revolutions." Perspectives on Politics 5, no. 3 (2007): 535-51. Van Donge, Jan-Kees. "Kamuzu's Legacy: The Democratisation of Malawi: Or Searching for

the Rules of the Game in African Politics." African Affairs 94, no. 375 (1995): 227-57. Venter, Denis. “Malawi: The Transition to Multi-Party Politics” in Democracy and Political

Change in Sub-Saharan Africa, edited by John A. Wiseman, 164-204. London; New

York: Routledge, 1995.

Wiseman, John A. "Early Post-re-democratisation Elections in Africa." Electoral Studies 11, no. 4 (1992): 279-91.