Transnational voting rights and policies in violent democracies: a global comparison

In recent decades more than one hundred countries have enfranchised their diasporas, allowing emigrants to vote from abroad. However, this widespread formal recognition of extraterritorial voting rights does not always lead to increased participation of emigrants in home country politics. Migrant-sending countries have complex relation-ships with their diasporas, and this relationship is particularly fraught for countries with endemic violence. This article leverages a new dataset documenting the adoption and implementation of extraterritorial voting rights and restrictions for 195 countries from 1950 to 2020 to demonstrate how transnational voting rights and policies in violent democracies differ from other regimes. While violent democracies extend transnational voting rights to their emigrants at rates comparable to other regime types, they are less likely to implement those rights, and when they do implement them, they are more likely to restrict them to insulate domestic politics from external influence. voting most Electoral voting

1968). However, both the opportunities and challenges inherent to external voting may be more complicated in violent democracies. Violence can undermine state capacity, making it more challenging to organize voting abroad. Domestic political actors that rely on violence may be particularly resistant to adopting extraterritorial voting rights, as emigrant voters are more difficult to target with violence (and indeed, may have become emigrants in order to flee violence). On the other hand, violent democracies often are among the most economically dependent on emigrants (Pérez-Armendáriz, 2019), and, given global diffusion of extraterritorial voting, resistance to enfranchising external citizens has become more difficult.
This article presents a quantitative analysis of extraterritorial voting rights and policies comparing violent democracies with other regimes around the world. We first explore the relationship between violence and democracy, situating that tension in the broader global trend toward the extension of extraterritorial voting rights. We suggest that competing pressures over diaspora enfranchisement are often resolved in violent democracies in a distinctive pattern: while, as in other regimes, adoption of extraterritorial voting rights is likely to be widespread, implementation is likely to be delayed or postponed, and when it does occur, often involves greater restrictions on voting rights (such as through restrictive eligibility requirements or limiting voter access), suggesting a desire to insulate local political dynamics from 'excessive' external influence.

Violent democracies and emigrants as voters
Democracy is-at least in the narrowest definition of the term-a system in which rulers are determined by popular election. Although this minimalist definition of democracy does not mention violence, those who defend this definition typically justify minimalist democracy as the only political system that provides for peaceful change in government in response to the will of the people. Perhaps most famously, Popper (1945Popper ( (2020 defends democracy as the only system in which citizens can get rid of governments without bloodshed. Przeworski, the most prominent contemporary theorist espousing a minimalist conception, also defines democracy as a system in which political contestation, however imperfect, is waged through the ballot box rather than through violence. "In the end, the miracle of democracy is that conflicting political forces obey the results of voting. People who have guns obey those without them. " (Przeworski, 2018: 118).
Yet in many electoral democracies political actors regularly use violence, or the threat of violence, as a means to gain influence and power. As defined by Pérez-Armendáriz (2019: 1), "violent democracies" are states where "manifold political actors regularly use violence to compete for power and make demands within established democratic institutional frameworks" (see also Arias & Goldstein, 2010). Scholars increasingly focus on these violent democracies as important cases to better understand both the theoretical and empirical links between violence and political contestation, and thus better interrogate our understanding of democracy. While scholars have worried about the proliferation of "democracy with adjectives" (Collier & Levitsky, 1997), unpacking the political context of violent democracies challenges simplistic justifications of democracy by calling into question key assumptions scholars make about political contestation.
There are two intertwined aspects of violence in "violent democracies" that we think are of particular importance to understanding the political implications of this distinct regime type. The first is understanding variation of the ability and incentives amongst political actors to use violence as a tool to seek power, or seek to influence those in power. In this light, violence is a political resource akin to money or popular supporta tool that may be unequally distributed, but one actors can use (or credibly threaten to use) to help achieve their political aims. Second, violence, in addition to a political resource, inevitably becomes a political issue in its own right. Political actors develop distinct platforms that often include competing approaches to political violence. In some instances, political actors may implicitly or explicitly advocate violence as part of their platform, and in others political actors may advocate strategies aimed at reducing violence as part of their appeal.
Democracy is impossible without defining the scope of the demos, the people, and increasingly this includes debate over the inclusion of non-resident citizens (Bauböck, 2015). The general trajectory of democracy around the world over the past two and a half centuries has been to an increasingly inclusive conception of the people and increasingly inclusive electorates (see Kuo, 2020 for a review on democratization and franchise extension). Voting rights-a requirement for full membership in a democratic political community-have always been fiercely contested. Most early democracies started with narrow criteria for full citizenship that drew sharp boundaries limiting political participation. In most European countries and their colonies, voting rights prior to the twentieth century were limited to people who were male, propertied and white. But over time and through extensive (and often violent) contestation, more "universal" conceptions of suffrage became the global democratic norm.
In contrast with other universal human rights, however, citizenship and voting rights have been tied to geography-i.e. recognized membership in a particular nation-state. The tension between universal human rights and territorially-sovereign nation-states is a staple in the study of international law and human rights, but one that is particularly important when it comes to thinking about suffrage (Bauböck, 2007). And while this tension would be less consequential in a world in which migration is low, that is not the world we live in. International migration has increased 84% since 1990, with 281 million immigrants currently living in countries other than those where they were born (UN-DESA, 2019). Economic remittances from diaspora citizens have correspondingly skyrocketed, reaching 714 billion in 2019, surpassing foreign direct investment as the largest source of external income to low-and middle-income countries (World Bank, 2020). In addition to economic remittances, numerous scholars have also noted transmission of values, education, and skills back into home countries, so-called "social remittances" (e.g. Levitt, 1998).
With the remarkable increase of the "domestic abroad" (Varadarajan, 2010), both state and non-state actors alike have sought to strengthen ties with non-resident citizens to encourage economic and political involvement. In addition to formal ministries, liaison offices abroad, or tax incentives (Gamlen, 2019), the majority of governments around the world now formally extend voting rights to emigrants. Although external voting is not new-extended initially during the Roman empire and implemented by Wisconsin during the Civil War and New Zealand in 1890 (Ellis et al., 2007)-there has been a global wave of diaspora enfranchisement in recent decades, with 100 countries extending voting rights to citizens abroad for the first time since 1990. Currently, 141 states, or around 70% of all the countries in the world, have legally extended the right to vote to citizens living outside of the country (Wellman, Allen, and Nyblade, Forthcoming). The ability for emigrants to formally participate in the politics of their home countries as voters has never been greater.
Numerous scholars have linked the emergence of extraterritorial voting with concurrent global trends of democratization, but through different pathways. 1 First, scholars have linked the extension of voting rights to non-resident citizens during periods of genuine democratization, whether part of the transition to multipartyism in Africa in the early 1990s (Hartmann, 2015), following the so-called Color Revolutions in eastern Europe in the 2000s (Bunce & Wolchik, 2006), or as part of political reforms following the Arab Spring uprisings throughout the Middle East and North Africa (Brand, 2014). These periods of political liberalization provide a window of opportunity for formerly disenfranchised groups to make claims for their inclusion (Rhodes & Harutyunyan, 2010). Moreover, the phenomenon of including refugees in post-conflict elections is seen as a symbol of reconciliation: an invitation to participate as an invitation to return (Grace & Mooney, 2009). Scholars examining enfranchisement through political transitions have recognized that the extension of voting rights to emigrants occurs in authoritarian regimes, not just democracies (Brand, 2010), but enfranchisement may unfold differently depending on regime type and a wide range of political considerations, including the magnitude and anticipated impact of enfranchising the diaspora (Umpierrez de Reguero et al., 2021). They have also highlighted that the effective inclusiveness of the adoption of transnational voting rights varies greatly (Caramani & Grotz, 2015). However, they have generally not considered how the distinctive dynamics that occur in violent democracies affect transnational voting rights and policies.
Other scholars have looked at the wave of non-resident voting as a reflection of international policy diffusion associated with emerging norms of democratic inclusivity in the post-cold war era (Rhodes & Harutyunyan, 2010;Turcu & Urbatsch, 2015). The adoption of global norms may occur due to the internalization of such norms by domestic political elites, or as an instrumental political strategy: states may adopt policies that "signal" democratic bona fides in order to increase legitimacy or attract investment or assistance from the international community (Hyde, 2011). Thus, states may legally extend voting rights as a signal of democratization to the international community, akin to inviting election observation monitors (Hyde, 2011;Kelley, 2012) or establishing gender quotas in legislatures (Bush, 2011).
A third set of scholars has looked at the electoral inclusion of non-resident citizens as linked to transnational mobilization by political parties (e.g. Burgess, 2020;Østergaard-Nielsen & Ciornei, 2019). Although the diaspora has long been a source of financial support for numerous political parties in countries around the world, the rise of transnational voting entails parties campaigning abroad for not only campaign donations, but also votes, requiring new research on "political parties abroad" (Kernalegenn & van Haute, 2020;Rashkova, 2020;van Haute & Kernalegenn, 2021). Additionally, non-resident citizens may influence the vote choices of family and friends back home (Paarlberg, 2017).
All of these pathways to enfranchisement are likely to unfold differently in violent democracies. Endemic violence pressures states and political actors, motivating extra efforts by domestic political elites to signal compliance with international democratic norms, and increases the economic importance of emigrants while simultaneously making diaspora engagement riskier for local power brokers. We explore our expectations for how the extension of transnational voting rights is likely to play out differently in violent democracies than in other contexts below.

Violent democracy and non-resident voting rights: concepts and hypotheses
As a general matter, we suggest that external voting may entail greater risks and greater rewards in violent democracies compared with other regimes. In violent democracies, multiple actors-from civil society organizations to organized cartels, everyday citizens to incumbent leaders-use violence as a means to further a range of political ends, whether to protest government action or inaction (Holland and Rios Contreras 2017), force negotiations with political elites (Assies, 2004), or secure favored electoral outcomes (Wilkinson, 2006;Giraudy 2009). Violence also becomes a subject of contestation, with some electoral actors competing based on promises to reduce violence, while others use violence as a method of signaling to voters their positions or credibility (Daly, 2019). In these contexts, the real and perceived distribution of political support of domestic actors and in diasporas may reflect substantial differences, and the means by which elites can compete for support amongst domestic and diaspora votes may be quite unequal.
Both a legacy of conflict and present-day violence may influence the relationship between diasporas and domestic political actors. The political orientation of the diaspora (real or perceived) shapes the extension and implementation of external voting in ways that distinguish violent democracies from other post-conflict contexts. For countries experiencing civil conflict and post-conflict transitions, refugee inclusion in transitional elections is often considered an essential component of moving forward (e.g., Grace, 2007;Van Houte, 2014). The imperative to enfranchise citizens displaced by violence as an invitation to return is underscored by the involvement of international organizations (IOs) in efforts to organize external voting for refugees as well as internally displaced people. IOs, including the United Nations High Commissioner for Refugees (UNHCR) and International Organization for Migration (IOM), have facilitated external voting for numerous transitional elections, including the 2013 Mali elections following the Tuareg uprising (Caux & Mahoney, 2013), the 2014 Libyan elections (Labovitz, 2014), and the referendum for the independence of South Sudan (Carter Center, 2011). Thus inclusion of a diaspora from a violent democracy becomes more crucial for a meaningful democratic transition both domestically and in the eyes of international actors.
While the pressure to include diasporas in elections may be greater in violent democracies given democratic norms and perceptions of the international community, the strategic risks to political elites of expanding the electorate to include citizens in violent democracies may be particularly complex. Conflict diasporas may have strongly localized ties and partisan biases that suggest political interests different from domestic constituencies. Furthermore, the means by which local elites mobilize both domestic voters and voters abroad may be substantially different. Incumbent political elites may believe they have relatively more influence over domestic voters than diaspora voters. Many of the tools that local elites employ in violent democracies-not only violence, but also patronage, public goods provision, and personal ties-may be seen as less useful in mobilizing potential non-resident voters given citizens abroad are much less likely to benefit directly.
Violence, in particular, is often a highly localized tool. In Mexico, India, Kenya and Zimbabwe, violence has been used to displace voters in particular electoral districts (e.g., Hammar, 2008;Robinson & Torvik, 2009; more generally see Dunning, 2011). This kind of geographically specific electoral violence is not as easily applied to citizens outside of the state, often dispersed across locations abroad. If violence is indeed a constitutive part of local politics, political actors who rely on violence to ensure their support may only want to include citizens who can be influenced through its strategic deployment.
Furthermore, the context of violence as a productive tool for, and constitutive of, democratic politics in these contexts may preclude the effective inclusion of emigrants as voters. Combatting electoral violence at an operations level is a challenge for electoral management bodies, requiring additional resources and support. In contexts of limited resources and capacity, electoral security domestically may take priority over building the global infrastructure for voting abroad (Pallister, 2020). Whereas we may observe high levels of de jure enfranchisement in violent democracies, particularly if they are also migrant-sending states, in-country political violence may preclude effective de facto enfranchisement during elections.
Although we expect that the pressures for adopting external voting should be as high, or even higher, in violent democracies as in any other regime, the political risks to increasing diaspora inclusion and electoral power, and the practical challenges in implementing external voting, may also be greater. Ultimately, we suggest that this tension results in violent democracies exhibiting a distinctive pattern when we examine both the extension and implementation of transnational voting rights, one that may be easily missed if one considers only the nominal adoption of transnational voting rights. We expect that, although violent democracies are likely to adopt emigrant voting rights, just as we have seen globally across other regime types, they are more likely to delay the implementation of those rights and restrict those rights in order insulate local-level politics and political elites. We briefly summarize each of these expectations below. 1) Widespread adoption: As the democratic and international pressures for diaspora inclusion in violent democracies is as great (or greater) than in other political regimes, we expect violent democracies will follow larger global trends and extend non-resident voting rights in a similar fashion to other regimes. 2) Delayed implementation: Given the instability of violent democracies, we expect that there may be more delays in organizing extraterritorial voting compared with other contexts. This delay can occur in multiple ways. First, violent democracies may have adopted diaspora voting policies relatively later than other democracies. Second, given concerns of security and capacity, it may be more difficult for electoral commissions to organize voting abroad when they also must address violence within the domestic electoral context. Thus, we expect violent democracies are more likely to delay implementation, potentially waiting numerous electoral cycles before organizing voting abroad. In some cases, that may mean the state has never implemented. 3) Local-level insulation: Scholarship on violent democracies has identified that political actors that use violence to achieve their political goals often do not target elections of national-level offices but rather local and regional politicians, such as mayors (Calderón, 2018) or state governors (Trejo & Ley, 2017). Not only are these offices and officials the targets of violence, but they also can be the perpetrators of political violence as well (Giraudy 2010;Rosenzweig, 2017). Violence (or the threat of violence) as a tool of political coercion requires the target to be local. Thus, political actors who deploy, or would want the option to deploy, local violence for political ends should be less interested in including diaspora voters. As non-resident citizens are insulated from local violence, and may have different interests than those they are attempting to mobilize in-country, sub-national politicians in violent democracies may want to insulate their electoral contests from diaspora votes. Thus, we predict that external voting in violent democracies is more likely to be restricted to national elections, i.e. less likely to include diaspora voters in sub-national legislative or state elections.

Conceptualizing and measuring violent democracy
What constitutes a violent democracy? While any typology of regimes inevitably simplifies both complex concepts and rich variation in institutions and practices across political systems, building typologies and leveraging them is crucial to make sense of political systems that migrants (and people more generally) engage with. Attempting a global comparison of external voting in violent democracies requires measurement criteria for a universe of cases that is simple, straightforward, and systematic. Although we recognize that both violence in societies and democratic governance operate along a continuum and that there are challenges in operationalizing both concepts, to streamline our analysis we develop a binary typology along two axes: violent/non-violent and democracy/non-democracy. One limitation of this approach is that it fails to capture the full dynamism in both violence and democracy and how these characteristics of political system change over time. We focus on countries that can be reasonably classified as violent democracies in recent years, with levels of violence and democratic political institutions that are persistent. That said, there are important changes over time that are inevitably missed in a global analysis that relies on a static regime type classification. Even without capturing dynamic nuances in our measurement strategy, we believe that there are important patterns in the adoption and implementation of emigrant voting rights in violent democracies using the approach we adopt here.
To measure violence, we use the Rule of Law Index generated by the World Justice Project (WJP) (World Justice Project, 2018). The WJP index measures rule of law across eight separate variables using data from public and expert surveys of each country in the dataset. WJP is recognized as a central measure of the rule of law in political science literature (Hadfield & Weingast, 2014). Beyond political science, WJP is used in economics, law, and sociology, including work analyzing education accountability (Botero et al., 2013), enterprise spending on security (Besley & Mueller, 2018), and participation in cryptocurrency markets (Décary-Hétu et al., 2016).
For this article, we rely on one of the WJP variables, Order and Security, to identify violent countries. The Order and Security variable is itself an aggregate score of three sub-factors: "Crime is effectively controlled" (i.e. low burglary, kidnapping, etc.); "civil conflict is effectively limited" (i.e. citizens are protected from armed conflict and terrorism); and "people do not resort to violence to redress grievances" (i.e. citizens do not resort to violence to solve civil disputes). WJP then combines the scores of these three sub-factors to generate the Order and Security measure for a country-year. These scores range from 0.31 [Pakistan] to 0.93 [Singapore], with a mean of 0.72 and a standard deviation of 0.13. Order and Security measures are available for the years 2014 through 2017. Using the Order and Security variable allows us to parse out sub-state violence from broader rule of law issues that may not directly pertain to physical coercion, such as the transparency of government data or the enforcement of government regulation. While not all countries are included in every iteration of the Rule of Law Index, WJP generated scores for a total of 112 countries during this time period.
Since we treat "violent democracy" as our main regime category of interest, we require a cut-off point to sort "violent" from "non-violent" contexts. Accordingly, we code a country as "violent" if they have an Order and Security score one standard deviation below the mean score at any time between 2014 and 2017. This simplification generates a static dichotomous variable that provides a simple, intuitive measure across a wide number of countries. Nevertheless, this measurement strategy has its limitations. While the WJP improves our ability to compare and analyze country contexts with a more nuanced measure for the concept of violence in a given state, these measures are also quite new, making it impractical to study change in violence over time. However, of the 31 countries with scores that identify them as "violent, " nearly half of them (15) qualify as "violent" in at least three of the four years of the WJP index we sampled. For the cases that fell a standard deviation below the mean just once, we also err on the side of inclusion.
To identify countries as democracies, we employ the widely-used Polity2 variable from the Polity5 dataset (Marshall, Gurr, and Jaggers 2020). Polity2 provides a -10 to 10 point measure of political regimes measuring executive constraints, executive recruitment, and political competition. Following the suggestion of the dataset authors (Center for Systemic Peace, 2020), as well as numerous studies in political science (e.g. Fearon & Laitin, 2003;Tsebelis, 2017), we treat the cut-off for "democracy" as a Polity2 score of 6 or higher. In order to match the time period of the Polity2 variable with the period covered by the WJP Order and Security score, we then code any country with a Polity value of 6 or higher in three out of the four years between 2014 and 2017 (the final year included in our dataset of extraterritorial voting rights) as a "democracy. " The more demanding inclusion criteria for the "democracy" measure (3 of 4 years), as compared with the "violent" measure (1 of 4 years), is applied because the time-series coverage in the Polity5 dataset is comprehensive across cases. 2 We then combine the democracy variable with the violent systems variable to divide regimes into four types. Out of 112 total countries included in both the 2017 WJP and Polity5 datasets, we identify 17 countries as violent democracies. 3 In contrast, 66 are non-violent democracies, about 59% of the sample. There are 29 non-democracies in the WJP sample, split between violent (14) and non-violent (15) subtypes.
There are notable gaps in our global coverage. Small countries (those with less than 500,000 people) are not covered by Polity5 and thus not included in our measure of violent democracy. Likewise, the WJP coverage focuses on larger countries, though without the clear population cut-point offered by Polity5. This has two important implications. First, as Collyer and Vathi (2007: 17) note, in small countries, diasporas can have a disproportionate influence on electoral outcomes. As such, small countries tend to correlate with low levels of electoral inclusion for emigrants. Our sample, then, is biased toward larger countries that are more likely to enfranchise non-resident citizens. Second, the gaps in the WJP coverage results in an undercount of violent democracies. For instance, countries that arguably fit the "violent" criteria, and are coded "democratic" according to our Polity criteria but are not included in WJP, include Timor-Leste and Kosovo. So although our sample of countries covers the vast majority of the world's population, we recognize it may miss an distinct dynamics seen in smaller violent democracies.
There are many other measures of both violence and democracy that could be reasonably used to create a violent democracy regime typology. While countries that may be at the margin of the threshold between democracy and non-democracy or violent and nonviolent could be reclassified using alternative measures, we believe that the countries identified by the approach we have adopted here identifies a reasonable set of countries that are violent democracies. In a supplementary appendix to this article (Additional file 1), we use two different data sources, the Armed Conflict Location and Event Dataset (ACLED) (Raleigh et al., 2010) and Varieties of Democracy Dataset (V-DEM) (Coppedge et al., 2011) to generate an alternative violent democracy typology. We show that our results are robust to minor differences in the regime type classifications that result from different data sources.

Measuring voting abroad
To assess the dynamics of emigrant enfranchisement in violent democracies, and how they compare with other country contexts, we employ the Extraterritorial Voting Rights and Restrictions (EVRR) dataset (Wellman, Allen and Nyblade, Forthcoming). EVRR is the first global time-series dataset of non-resident voting policy and procedures that captures adoption, implementation, and a variety of restrictions, covering 195 countries between 1950 and 2020. It is the most comprehensive dataset to date on external voting in terms of geographic scope, temporal coverage, and policy details, with over 20 variables covering how non-resident citizens are included as voters. The unit of observation for this analysis is at the country-year level (n = 13,845). Nyblade et al. Comparative Migration Studies (2022) 10:27 The dataset distinguishes between legal enfranchisement (i.e. adoption) and effective enfranchisement (i.e. implementation). Legal enfranchisement is coded as a yes if the following three criteria are met: first, enfranchisement extends to citizens whose primary residence is abroad (vs. being out of the country on a temporary basis); second, non-resident citizens are able to cast their vote from abroad (vs. having to physically return to the home country to vote, such as in Nigeria and Ireland); and finally, the nonresident voter eligibility is relatively similar to resident voter eligibility (vs. restricting external voting to particular professional categories, e.g. state employees).
In 1980, only 21 countries around the world (13%) legally enfranchised citizens abroad; by 2020, that number increased to 141 countries, 72% of the countries in the dataset. Effective enfranchisement is coded as a yes if the following two criteria are met: first, there is evidence (i.e. from published election results, news media coverage, and/or electoral monitoring reports) that the government organized external voting for citizens abroad in the last national election; second, that implementation fulfilled criteria consistent with legal adoption. In 1980, only 13 countries around the world (8%) had organized polling abroad; by 2020, that number increased to 127 countries organizing voting abroad for at least one election (65%).
In addition to these baseline measures of external voting, EVRR also codes practical restrictions that significantly shape the degree of emigrant inclusion in electoral competition. These restrictions include election type (executive, legislative, diaspora legislators, and referenda), voter eligibility restrictions (registration policies, identification requirements, citizenship status), and external voting method (in-person, proxy, online, or postal balloting). While scholars have long recognized how restrictions shape the nature and scope of emigrant inclusion through the extension of voting rights (e.g., Lafleur, 2015), no other global dataset systematically captures these important characteristics of transnational voting.

Legal adoption of non-resident voting rights
Most violent democracies have extended legal rights to non-resident citizens. As can be seen in Table 1, in our sample of 17 violent democracies, 14 (82%) had undertaken some form of legal change recognizing the rights of non-residents to vote from abroad by 2017. This is slightly higher than the total percentage of non-violent democracies in our sample (72%). Though this modest difference is not statistically significant, it does indicate that violent democracies have kept up with the global pattern recognizing voting rights on non-resident citizens.
However, this static finding obscures differences in the timing of legal extensions. EVRR also allows us to identify the year of legal enfranchisement. The United Kingdom, for instance, extended non-resident citizens the right to vote from abroad in 1985. 4 Thailand first legally recognized voting abroad in 1997. 5 Figure 1 shows the adoption of extraterritorial voting rights by year of adoption for violent democracies, as well as for non-violent democracies, violent non-democracies, and non-violent non-democracies. Overall, violent democracies are later enfranchisers than non-violent democracies. On average, violent democracies adopted extraterritorial voting rights in 1999, whereas in non-violent democracies the average year of adoption was in 1990, nearly a decade earlier.
This difference of nine years reflects distinct adoption trajectories for the two groups of countries. Of the contemporary non-violent democracies, 18% (11) had already extended non-resident citizens the right to vote from abroad by 1980. The same is true Table 1 Adoption and implementation of extraterritorial voting in violent democracies Data from Wellman, Allen and Nyblade (Forthcoming) and reflect adoption and implementation as of 2020 * External voting was originally extended in South Africa in 1993 for the 1994 transitional election, rescinded in 1998, and subsequently re-instated by a constitutional court ruling in 2009 (Wellman, 2021)

Country Adoption Implementation
Argentina 1991  of only one violent democracy (Colombia), which in fact was one of the first countries to broadly extend voting to citizens abroad beginning in 1961 (Lafleur, 2015). However, Colombia's experience is exceptional. In terms of legal extension of voting rights to nonresident citizens, violent democracies appear to follow rather than innovate. With a few exceptions, the earliest cases of emigrant enfranchisement tended to occur in relatively peaceful, democratic countries. Non-resident voting picked up significantly in the 1990s and early 2000s, reflecting broader democratization trends including transitions to multiparty systems, movement away from military to elected leadership, and new constitutions. It was after the practice of non-resident voting was widely established elsewhere that most of the violent democracies in our sample enfranchised emigrants. In these contexts, legal extensions of voting rights to non-resident citizens often occurred as part of broader reforms, including comprehensive electoral overhauls, e.g. Guatemala 2016 (Pallister, 2020) or new constitutions, e.g. Kenya 2010 (Wellman & Whitaker, 2021).

Delayed implementation
Legal recognition of diaspora voting rights does not guarantee the ability to vote abroad. Where violent democracies have granted non-residents the right to vote from abroad, they are often slow to implement these rights. This pattern manifests in several ways. As observed in Table 1, both delays and failure to implement are particularly noticeable. Despite violent democracies adopting emigrant voting at slightly higher rates than other countries, they have implemented external voting at relatively lower rates. In 2017, 63% of countries from our sample implemented non-resident voting rights in their most recent relevant election. For violent democracies, only 59% had implemented (10 of 17). This was slightly higher than violent non-democracies (57%) but noticeably lower than both non-violent democracies (65%) and non-violent non-democracies (67%). Although these differences are not large, the fact that they reverse the pattern seen with legal adoption -where violent democracies appear eager to formally extend rights -indicate an important gap between law and practice. Second, similar to legal adoption, violent democracies that implement external voting are also relatively later than other countries. In violent democracies that have organized external voting, the average year of first implementation was 2002, whereas the average year of first implementation for nonviolent democracies was 1991. Violent democracies are more similar to non-democracies in this respect (violent non-democracies average year of adoption: 2004, non-violent non-democracies: 2003). The overall patterns of implementation by regime type can be seen in Fig. 2.
Violent democracies, in short, have been slower to implement voting rights for nonresident citizens than non-violent democracies. An additional metric that allows us to demonstrate this pattern is an analysis of the implementation gap, or the number of years between legal adoption and implementation. We can measure implementation gaps for countries where we have both a year of adoption and year of initial implementation between 1950 and 2017. In total, this includes 120 cases in the EVRR dataset; out of those 120, 74 of them are included in the Polity5 and WJP datasets.
In our sample, the average implementation gap is 3.8 years. For the non-violent democracy subtype, however, the average is 2.6 years. This is in notable contrast to violent democracies, which have an average implementation gap of 7.9 years. In between these extremes, both forms of non-democracy have an implementation gap of 4.3 years. Figure 3 visualizes this major temporal distinction between violent and non-violent contexts.

Restrictions
The EVRR dataset measures three types of restrictions on non-resident voting: structural incorporation (e.g. for which elections non-residents can vote), voter eligibility (e.g. who can vote), and voting method (e.g. how non-residents can vote). We summarize external voting restrictions in violent democracies that have implemented extraterritorial voting in Table 2, focusing on three tangible metrics of institutional incorporation,  . 3 Extraterritorial voting adoption-implementation gaps by regime type. Notes: Data from Wellman, Allen and Nyblade (Forthcoming). Implementation gaps are only reported for countries that both adopted and implemented extraterritorial voting between 1950 and 2020 eligibility, and voting method: right to participate in legislative elections, identification requirements, and remote voting options (i.e. not in-person). The results indicate a tendency to insulate sub-national politics from emigrant voters.
The first sign of insulation is apparent in the tendency to exclude emigrants from participating in legislative elections. Even when a country has legally enfranchised non-residents and implemented some form of voting abroad, they can still exclude non-residents from important electoral competitions (e.g. voting in presidential elections but not referendums). Focusing on countries that allow non-residents to participate in legislative elections provides an important metric of diaspora inclusion. National legislative elections, more so than executive elections, involve high numbers of locally-rooted political actors pursuing political office. As such, legislative elections may be particularly important to political actors who engage in violence, as well as emigrants who remain particularly embedded in regional economies back home (Bada, 2014). A disjuncture between incorporation in executive-level elections and exclusion from legislative elections could indicate an interest in insulating local political dynamics from non-resident influence.
We find tentative evidence in support of the insulation hypothesis. In our sample, 81% of countries that enfranchised non-residents allow these citizens to participate in legislative elections. For nonviolent democracies, 87% extend the vote in legislative elections to residents abroad. This is true of only 57% of violent democracies, which is well under the percentage for the other country types (violent non-democracies: 80%; non-violent nondemocracies: 90%). 6 It is useful here to disaggregate parliamentary and presidential systems, as the former have the opportunity to exclude emigrants from legislative elections while still incorporating them into national competitions for political office. Notably, of the violent democracies that have parliamentary systems (India, Pakistan, South Africa), all three have legally enfranchised non-residents. However, legal enfranchisement of emigrants in these cases has been accompanied with non-implementation and retrenchment of rights. Pakistan has never organized diaspora voting, India has yet to organize a mechanism for non-resident Indians to vote from outside of the country, and South Africa abolished foreign voting in 1998 following the 1994 transitional election. When South Africa re-introduced non-resident voting in 2009, it only enabled voting at the national list level; it no longer included the ability for citizens abroad to vote in provincial elections (Wellman, 2021).
Among the violent democracies with presidential elections, 10 of 14 have legally enfranchised non-resident citizens. Of these, only six (Argentina, Colombia, Dominican Republic, Mexico, Philippines, Sierra Leone) allow non-residents to vote in legislative elections. Among these countries, several have taken additional steps to insulate legislative competitions from non-resident voters. Philippines, for instance, has a mixedmember electoral system for the House of Representatives. Voters within the country can cast ballots both in single-member districts and for a party list in a nationwide proportional competition. Non-resident citizens, however, only cast ballots for the national party lists. Colombia and the Dominican Republic, meanwhile, insulate the legislative competition by other means. In the Dominican Republic and Colombia's Chamber of Representatives, rather than integrating legislative votes into territorial constituencies, they set aside extraterritorial seats for non-resident citizens. Although this provides non-residents with direct representation within the legislature, it also ensures that votes from abroad cannot tip the balance in any domestic district (Palop-García, 2018).
Moving on to eligibility restrictions, we see some modest signs that violent democracies restrict access via identification rules. 7 For non-residents, keeping up-to-date identification documents can be particularly onerous as they often lack easy access to administrative offices. In the most permissive systems, countries do not require any specific type of identification and will accept a range of documents, sometimes even if expired. 20% of the countries in our sample apply permissive identification rules; notably, not one of them is a violent democracy. By contrast, 29% of non-violent democracies use permissive identification rules. At the other end of the continuum, 17% of countries in our sample use strict criteria requiring two or more specific types of identification (often a valid passport and a country-of-residence visa). In violent democracies, 33% use the stringent criteria, more than any other subtype (24% of non-violent democracies use strict criteria). The higher eligibility requirements may indicate an elevated concern with the issue of ballot security, or an additional barrier in order to reduce the relative influence of the diaspora vote.
In terms of voting methods, violent democracies tend toward in-person rather than remote voting options, which are often more costly for emigrants. Like most countries, violent democracies follow the norm of organizing in-person voting for non-residents at embassies, consulates, and/or other locations in the country of residence. In our sample, 90% of countries with voting abroad utilized in-person voting methods. A similar percentage (91%) of non-violent democracies use in-person voting. Among violent democracies, 82% offer in-person voting.
In addition to in-person voting, there are three additional voting methods used for external voting: postal, proxy, and electronic. These remote modes of voting do not require in-person participation at a polling booth on the part of the voter: in other words, voting at a diplomatic post using electronic means is not considered electronic voting. By 2017, 40% of our sample countries with non-resident voting rights used some form of remote voting, often in addition to in-person voting. Over half (56%) of nonviolent democracies allow some form of remote voting. In contrast, not a single violent non-democracy offers any form of remote voting, while violent democracies fall between these two extremes, with 25% offering a form of remote voting. This is similar to the percentage of non-violent non-democracies (20%) that provide a remote voting option.
Not only do violent democracies rely more heavily on in-person voting than non-violent democracies, they are more likely to restrict access based on country of residence. Countries that offer in-person voting can choose to organize polling stations within a limited selection of countries. In our sample, approximately one-third (34%) of countries employing an in-person voting method also restricted the geographic location of the stations. For non-violent democracies, only 27% restricted polling stations to specific countries, while the non-democratic sub-types restricted in close to half of all cases (violent non-democracy: 44%; non-violent non-democracy: 50%). Violent democracies land in between, with 38% of cases with in-person voting imposing some form of geographic restriction. Bolivia, for instance, limited voting abroad to Argentina, Brazil, Spain, and the United States in the year of its first implementation (Lafleur & Sánchez-Domínguez, 2015). Kenya only organized external voting at their high commissions in neighboring countries in 2013 (Burundi, Rwanda, Tanzania, and Uganda), although the Kenyan diaspora is predominantly located in the United States and United Kingdom (Wellman & Whitaker, 2021). While these geographic manipulations may be more common in violent democracies than in non-violent democracies, it is important to recognize that this is not the norm amongst either type; most democracies attempt to enact in-person voting broadly.
Overall, the comparative descriptives we present are consistent with the hypotheses suggested earlier. We observe widespread adoption of extraterritorial voting rights in violent democracies, though implementation is less consistent and more delayed, and there are greater restrictions on voter eligibility and the types of elections in which emigrants from violent democracies are allowed to participate.

Conclusion
Emigrants are key actors in violent democracies around the world, and the many roles they play within the political, economic and social dynamics of their home countries are extensively documented throughout the cluster of articles devoted to this topic.