Making Sense of Human Rights Diplomacy: Symbolism or Concrete Impact? October 18, 2018

Transcription

1 Making Sense of Human Rights Diplomacy: Symbolism or Concrete Impact? Rachel Myrick Stanford University Jeremy Weinstein Stanford University October 18, 2018 This paper was prepared for the 2018 Journey in Politics Workshop at University of Iowa. Please do not circulate or cite without permission of the authors.

2 Abstract Existing studies of human rights reform in international relations consist mainly of cross-national analyses that assess the extent to which international legal institutions or human rights organizations change human rights practices. This literature largely sets aside the role of human rights diplomacy (HRD) efforts by government officials to engage publicly and privately on human rights with their foreign counterparts in changing human rights practices because diplomatic actions are often not publicly observable. In this paper, we exploit a unique opportunity to assess the relative effectiveness of HRD in practice: a coordinated effort by the U.S. government in 2015 to free twenty female political prisoners (the #Freethe20 campaign). We use two strategies to analyze the outcomes of the #Freethe20 campaign. First, we obtain a list of forty women initially proposed by the U.S. State Department for inclusion in the campaign and compare the outcomes of those who were on the initial list with those who were ultimately featured. Second, we construct a dataset of all female political prisoners imprisoned simultaneously in the countries targeted by the campaign and collect data on their release outcomes. Both approaches demonstrate that women featured in #Freethe20 were released from prison at a significantly faster rate than their peers. Next, we examine two strategies of human rights diplomacy used during the campaign: strategies to publicly shame target states ( naming and shaming ) or to induce reforms with explicit promises or threats ( carrots and sticks ). Based on a quantitative analysis of online searches and media coverage, we find minimal evidence for naming and shaming or sustained public attention to these cases after September Instead, using a series of in-depth interviews with U.S. officials involved in the #Freethe20 campaign, we demonstrate that the campaign reflected coordinated efforts by the U.S. government to engage in private diplomacy around this set of women. We draw on evidence from political prisoner cases in Azerbaijan and Uzbekistan to show that states are more effective at human rights diplomacy when they promise or threaten specific carrots and sticks in private. In order to do so effectively, however, states issuing promises or threats must both have leverage over the target state and be willing to prioritize human rights issues over competing security and economic interests.

3 I. Introduction A large literature in international relations explores the role of international institutions, human rights treaties, and transnational advocacy in generating human rights reform. However, human rights diplomacy the public and private efforts of government officials to directly engage with their foreign counterparts on human rights issues has received considerably less attention. In large part, scholarship on human rights diplomacy remains underdeveloped because actions taken by state officials are often invisible to the general public. While we observe public pronouncements by state leaders and government-issued reports intended to name and shame human rights abusers, the bulk of diplomatic action in the human rights space is conducted privately. Diplomats promote policy and institutional reforms, raise concerns about human rights abuses, and advocate on behalf of imprisoned activists as a regular part of bilateral relations. In a contemporary setting, the contents of these private negotiations and correspondence between government officials are rarely revealed. In addition, because diplomatic actions are undertaken strategically, identifying the causal effect of any single diplomatic action is challenging. This paper asks: Can human rights diplomacy conducted by states have concrete impacts on human rights practices? If so, why and under what conditions? We identify a unique opportunity to evaluate how human rights diplomacy works in practice: a coordinated effort by the U.S. government to free 20 female political prisoners in 13 different states as part of the #Freethe20 campaign. The #Freethe20 campaign, launched in September 2015 in advance of 70 th Session of the United Nations General Assembly, was initially conceived of as a media campaign to name and shame target governments. Simultaneously, members of the U.S. Mission to the United Nations, the U.S. State Department, and Congress took systematic efforts to privately engage with foreign officials. A cursory look at the release rate suggests the

4 #Freethe20 campaign was relatively successful. Within three years of its launch, 16 of 19 women 1 had been released. However, without a comparable set of similar cases, it is challenging, if not impossible, to evaluate the relative effectiveness of the campaign. In order to make sense of both the conduct of human rights diplomacy and its anticipated outcomes, we use two approaches. First, we assess the overall effectiveness of this campaign by collecting demographic data on #Freethe20 political prisoners and comparing release rates among this group with two pseudo- control groups. For the first control group, we obtain a list of political prisoners also considered by the U.S. State Department for inclusion in the campaign. For the second control group, we collect data on all female political prisoners highlighted in Amnesty International s Urgent Action reports who were imprisoned in the #Freethe20 countries between We systematically collect arrest and release information for these women and compare their release rates to women featured in #Freethe20. Using both strategies, we find clear and consistent evidence that women featured in the #Freethe20 campaign were released at a faster rate relative to other female political prisoners. Second, we identify and assess the relative importance of two major strategies of human rights diplomacy: the use of public pressure to shame target governments ( naming and shaming ) and the use of coercive diplomacy to coerce or induce human rights reform ( carrots and sticks ). Using a combination of data collected from news media and social media, as well as detailed interviews conducted with government officials directly involved in the campaign, we analyze the causal pathways through which human rights diplomacy influences state behavior. 1 The twentieth prisoner featured in the #Freethe20 campaign was symbolic: an unnamed North Korean political prisoner. 2 A recent exception is Katagiri and Min s (forthcoming) research that uses declassified documents from the Berlin

5 To supplement our quantitative analyses, we provide illustrations of successes and failures of the #Freethe20 campaign in target countries: Azerbaijan and Uzbekistan. Overall, we find that while the #Freethe20 campaign was broadly successful in achieving its objectives, it did not necessarily work as intended. While the campaign was designed to implement a naming and shaming strategy in order to pressure target states to release prisoners, there is little evidence that the media campaign reached a broad international audience. Instead, the effectiveness of the campaign came from private engagement by government officials. In essence, the #Freethe20 campaign served to coordinate U.S. government actions around a specific set of political prisoner cases and, in some cases, reflected coordination that preceded the start of the campaign itself. This focus on a small number of women and the highlevel involvement of U.S. officials, including Cabinet members and the President demonstrated the willingness of the U.S. government to take costly action on this particular issue. Our research suggests that government officials wanting to generate meaningful reforms in target states must be both willing to prioritize human rights diplomacy over competing interests and be able to identify specific sources of leverage to induce changes. When these conditions are met, diplomatic actions taken to pursue specific and concrete human rights objectives can be effective. These findings shed light on the importance of studying human rights diplomacy generally and private engagement between government officials in particular. II. The Study of Human Rights Diplomacy Grounded in a normative project about obligations of the international community to respond to and prevent human rights abuses, international relations scholars have pursed an empirical effort to understand when and why international action results in human rights reforms. This

6 academic literature focuses on one of three forms of international action taken to change state behavior: (1) international human rights law, (2) transnational advocacy, and (3) human rights diplomacy. In this section, we briefly outline these three forms of action. We argue that while transnational advocacy and international law have received considerable attention in IR scholarship, the literature on human rights diplomacy is underdeveloped for both theoretical and practical reasons. The first form of international action taken to curb human rights abuses consists of multilateral action by inter-governmental organizations most notably the United Nations in the form of international human rights law. The international human rights regime includes an array of human rights agreements and treaties, as well as a network of inter-governmental institutions. These institutions consist of non-judicial bodies (like the United Nations Human Rights Committee) that monitor the implementation of human rights agreements and judicial bodies (like the International Criminal Court) that investigate and adjudicate human rights abuses conducted by individual leaders. Existing research explores the independent effects these institutions exert on changing state behavior and deterring future atrocities (Lupu 2013; Jo and Simmons 2016; Kim and Sikkink 2010; Powell and Staton 2009; Prorok 2017; Simmons and Danner 2010; Sikkink and Walling 2007; Sikkink 2011). In addition to inter-governmental institutions, international human rights law encompasses human rights agreements, which can be expansive in nature like the International Covenant on Civil and Political Rights (ICCPR) and the International Covenant on Economic, Social, and Cultural Rights (ICESCR), or issue-specific like the Convention on the Elimination of All Forms of Racial Discrimination (CERD) and the United Nations Convention Against Torture (CAT). An extensive literature in political science and international law analyzes when and why states

7 ratify human rights agreements (Adeyeva 2007; Cole 2005; Goodman and Jinks 2004; Hathaway 2002, 2007; Hawkins 2004; Hollyer and Rosendorff 2011; Landman 2005; Nielsen and Simmons 2015; Risse-Kappen et al. 1999; Vreeland 2008; Von Stein 2015; Wotipka and Tsutsui 2008; Zhou 2014), and the conditions under which states comply with human rights agreements they ratify (Cardenas 2007; Cole 2012; Goldsmith and Posner 2005; Goodliffe and Hawkins 2006; Goodman and Jinks 2004; Hathaway 2002; Hafner-Burton and Tsutsui 2005, 2007; Keith 1999; Neumayer 2005; Simmons 2009, 2010; Von Stein 2015). In international relations, these approaches primarily consist of large-n, cross-national empirical analyses supplemented by qualitative evidence from specific human rights agreements. For example, studies about compliance with human rights law often take treaty ratification as the independent variable and observed human rights practices as the dependent variable, controlling for other state-level characteristics. A second form of international action taken to generate human rights reforms is conducted by international non-governmental organizations (INGOs) such as Amnesty International and Human Rights Watch, and their associated networks in civil society. These networks of activists engage in transnational advocacy spanning state borders in order to influence both public opinion and state behavior (Keck and Sikkink 1998). The academic literature on transnational advocacy analyzes the methods and strategies that INGOs use and their impact on political outcomes. To achieve their strategic objectives, INGOs raise public awareness of human rights abuses by providing information to the public in the form of detailed reporting (Ron et al. 2005) or performance metrics (Kelley and Simmons 2015; Kelley 2017). This information is often used to name and shame or shame and blame target states. IR scholars have explored the impacts of naming and shaming on domestic public opinion (Ausderdan 2014) and nonviolent protests

8 (Murdie and Bhasin 2011), human rights reforms (Keck and Sikkink 1998; Hafner-Burton 2008; Murdie and Davis 2012), and responses from the international community (Barry et al. 2012; Dietrich and Murdie 2017; Lebovic and Voeten 2009; Murdie and Pesken 2014). The third form of international action which we refer to as human rights diplomacy (HRD) consists of public and private forms of strategic engagement between two or more governments with the objective of improving human rights practices in a target country. This includes publicly observable activities undertaken by states to shame foreign governments in a multilateral forum (Lebovic and Voeten 2006; Terman and Voeten 2017) or issue public coercive threats. However, most of the activities that constitute human rights diplomacy are not publicly observable. Diplomats and government officials working in the human rights space communicate through private channels; the content of these conversations is often undocumented or classified. As a result, we lack clear and observable data on diplomatic actions. Another logistical obstacle to the study of HRD is the difficulty in evaluating its effectiveness. In part, this stems from our inability to observe diplomatic failures. While HRD successes result in observable political reforms, diplomatic failures typically maintain the status quo. This makes defining and evaluating a set of cases in which diplomatic efforts are more or less successful challenging. In addition, private diplomacy is often is used in conjunction with public pressure, which makes it difficult to disentangle any meaningful causal relationship. In many cases, the use of public pressure is a direct result of failed efforts to secure progress through private engagement. In addition to these logistical obstacles, scholars and policymakers have reason to be skeptical about the efficacy of HRD. This skepticism likely dissuades researchers from systematically examining the topic. From a theoretical perspective, international relations

9 scholarship has traditionally given HRD short shrift by dismissing private communications between states as cheap talk. A very large literature on audience cost theory suggests coercive threats and other signals from democratic leaders are more credible when issued publicly rather than privately (Fearon 1994, Smith 1998, Schultz 2001). The logic of audience costs implies that democratic publics will punish leaders who issue a threat or a promise and subsequently renege. To the extent that IR scholars focus on private diplomacy instead of public coercive diplomacy, this research is largely driven by formal modeling rather than empirical analysis. 2 Formal models of diplomatic signaling, for example, unpack the conditions under which private diplomacy may be a credible form of state-to-state communication (Guisinger and Smith 2002; Ramsay 2004; Kurizaki 2007). Policymakers also have reasons to be skeptical about the efficacy of HRD, albeit for different reasons. First, when there are competing interests at stake, policymakers may be reluctant to issue private threats in order to advance human rights reforms because they risk undermining other aspects of the bilateral relationship. For example, while the United States has had serious concerns about human rights practices in Egypt under both the Mubarak and Al-Sisi regimes, human rights diplomacy is often constrained by concerns about the essential cooperation the United States maintains with Egypt on counterterrorism and the Israeli-Palestinian conflict (Belfakir 2018). Second, policymakers often believe that threats issued with respect to human rights are less credible because these issues pose higher stakes for the target state. In other words, a dictator has stronger incentives to imprison outspoken political dissidents than the United States government does to work to free them. Thus, a coercive threat issued privately may have little influence over the dictator s behavior. Finally, an overarching skepticism about HRD stems 2 A recent exception is Katagiri and Min s (forthcoming) research that uses declassified documents from the Berlin Crisis to assess the relative efficacy of public and private threats.

10 in part from the fact that diplomacy is likely to be ineffective in places where human rights abuses are especially severe. This is because many of the worst human rights offenders states like Iran, North Korea, and Syria are relatively diplomatically isolated with weaker economic ties to states advocating for human rights reform. Despite these challenges in the study of HRD, this foreign policy tool merits much greater attention. In many democratic states, a substantial percentage of the bureaucracy dedicated to foreign affairs routinely engages in various forms of HRD. In the U.S. context, for example, the State Department formed a separate bureau in 1977 called the Bureau of Human Rights and Humanitarian Affairs (later renamed the Bureau of Democracy, Human Rights, and Labor) to better integrate into human rights concerns into the foreign policymaking process (National Archives Catalog). The next section details the different strategies of HRD and the mechanisms by which they generate human rights reforms. III. Strategies of Human Rights Diplomacy Human rights practices within a particular state reflect a calculation by the regime about how to most effectively manage domestic political opposition. In wanting to maintain their own political survival, leaders may choose to perpetrate human rights abuses with the goal of repressing political dissent. However, leaders must weigh the costs and benefits of these abuses. For example, while imprisoning large swaths of civil society may prevent coordination among the political opposition, it may also backfire by mobilizing other domestic actors or drawing unwanted attention from the international community. Extrapolating from this basic framework, strategies of human rights diplomacy attempt to either raise the costs of perpetrating human rights abuses or increase the benefits to human rights reforms in the target state. These costs and benefits could be tangible (for example, changes in

11 foreign aid flows) or intangible (for example, changes in a state s reputation). This section discusses two primary strategies of HRD. First, states can publicly shame target states by highlighting human rights abuses. Second, states can promise or threaten to enact policies designed to affect a leader s cost-benefit calculus. Then, we describe how these strategies can be employed in public settings or in private settings. We argue that the visibility of naming and shaming makes it a central focus of international relations scholarship. However, in the human rights space, private engagement constitutes a key feature of diplomacy between states. Naming and Shaming A common strategy states use to induce human rights reforms in a target country is a process of naming and shaming. When states name and shame (Hafner-Burton 2008) or shame and blame (Murdie and Davis 2012), they publicly highlight human rights abuses in another state. The intent of this strategy is to increase the discomfort of leaders in the target state, thereby making the particular human rights violation appear less attractive. In this way, we can conceive of naming and shaming as an attempt to impose direct but intangible costs on a target state. These intangible costs are intended to affect a regime s status or reputation through a variety of social behaviors, such as shaming, shunning, exclusion, and demeaning, or dissonance derived from actions inconsistent with role and identity (Johnston 2001, p. 499). Researchers have suggested that naming and shaming conducted by states, international organizations, and non-governmental organizations can also indirectly impose tangible costs on a state via two separate channels. First, international shaming can mobilize human rights organizations and civil society domestically in a target country, increasing public pressure on a leader. Using observational and experimental public opinion research, Ausderdan

12 (2014) shows that when states are shamed by the United Nations Commission on Human Rights, their citizens are more critical of the human rights conditions in their country. Murdie and Davis (2012) show that naming and shaming by human rights organizations (HROs) is most likely to be successful when there is an active civil society in the target state. The authors argue that without human rights organizations in a target state, repressive governments are unlikely to respond to shame. Second, in addition to increasing domestic pressure on a leader, shaming can mobilize other actors in the international community to impose direct, tangible costs on the target state. For instance, Barry et al. (2012) show that naming and shaming by INGOs leads to reductions in foreign direct investment in the target country. Researchers demonstrate that countries shamed for their human rights record are also less likely to receive multilateral foreign aid (Lebovic and Voeten 2009) or official development assistance (Dietrich and Murdie 2017); they are also more likely to be the target of economic sanctions (Peksen et al. 2014) and foreign military intervention (Murdie and Pesken 2014). While the preponderance of academic literature on naming and shaming focuses on actions taken by international organizations (both governmental and non-governmental), states often engage in similar practices. For example, states use multilateral forums as an opportunity to scrutinize one another s human rights practices and strategically invoke shame. Terman and Voeten (2018) show that states are significantly more likely to submit comments to the United Nations Universal Periodic Review about human rights practices for states with which they share geopolitical affinities or other special relationships; however, these reviews are also much less critical. In addition to acting through multilateral forums, states also name and shame by directly issuing press releases, official statements, or reports about human rights violations. For

13 example, the European Union issues annual reports on human rights and democracy ( EU Annual Report on Human Rights and Democracy in the World 2017). Similarly, the U.S. State Department releases detailed country reports on human rights practices that are submitted annually to U.S. Congress ( Human Rights Reports 2018). Carrots and Sticks An alternative strategy to naming and shaming is to coerce or induce human rights reforms abroad, a strategy also known as coercive diplomacy. As George (1991) summarizes: [T]he strategy of coercive diplomacy can use positive inducements and assurances as well as punitive threats to influence an adversary; when it does so it is often referred to as a strategy of carrots and sticks (10). With respect to human rights, states issue implicit or explicit threats ( sticks ) or promises ( carrots ) designed to alter the cost-benefit calculus of target states engaging in human rights abuses. These carrots and sticks are frequently economic in nature, such as increases or reductions in foreign aid, foreign direct investment, international trade, or security sector assistance, or the imposition of broad or targeted economic sanctions. Hathaway (2004), for example, argues that states ratify human rights treaties, like the UN Convention on Torture, to receive carrots such as more foreign investment, aid donations, international trade, and other tangible benefits (Hathaway 2004, 207). Nielsen (2013) provides evidence that states use sticks to punish human rights violators by strategically withholding foreign aid. Beyond explicitly economic tools, states may promise or threaten to add or remove a target state from an important international institution, such as a regional organization or preferential trade agreement. Membership in both formal and informal international institutions can confer tangible economic rewards and intangible rewards, like international legitimacy.

14 When human rights reforms are a prerequisite for accession to an institution, the benefits of membership can be an attractive enough carrot to induce changes in human rights practices. For example, Kelley (2004) demonstrates that membership conditionality in the European Union and the Council of Europe resulted in substantive reforms in Baltic and Central European states. Increasingly, regional and international economic institutions such as preferential trade agreements also incorporate clauses related to democracy and human rights (Milewicz et al. 2016; Hofmann, Osnago, and Ruta 2018). Some scholars argue that states are more likely to comply with human rights standards integrated into economic agreements than with human rights agreements more generally, which lack specific mechanisms for ensuring compliance (Hafner-Burton 2005). Public and Private Diplomacy Having reviewed the two primary strategies of HRD, we now draw a distinction between whether these strategies are employed publicly or privately. In the human rights space, we frequently observe states naming and shaming because it is a publicly visible action. While government officials may invoke shame when privately engaging with their foreign counterparts, naming and shaming is fundamentally a mechanism of social influence and therefore more effective when an audience is present. In contrast, strategies of coercive diplomacy can be pursued both publicly and privately. While international relations scholarship would generally suggest that threats issued publicly by democratic states are more effective in crisis bargaining, it is uncommon to see human rights practices closely tied to public, explicit threats, which are typically reserved for major national security issues. In general, states are unwilling to take coercive action on human rights issues when: (a) those actions are unlikely to result in any meaningful reform, or (b) there are

15 competing security interests at stake that override human rights concerns. In the first case, states that take additional coercive action against target countries with abysmal human rights records, such as North Korea, would simply be adding to extensive, pre-existing sanctions regimes. In the second and arguably trickier case, government officials manage a delicate balance of competing interests in an existing bilateral relationship with a target state. Often, these conflicting interests are apparent within the foreign policy bureaucracy, creating a sort of deadlock that prevents progress on human rights diplomacy. Consequently, we would expect that diplomatic engagement on human rights issues is either raised privately with foreign counterparts or the topic is generally avoided altogether. IV. Evidence from Human Rights Diplomacy around Political Prisoners In order to assess the relative effectiveness of different strategies of human rights diplomacy, we evaluate the outcomes of a specific human rights campaign conducted by the U.S. government to free female political prisoners. According to official guidelines outlined by the Parliamentary Assembly of the Council of Europe (PACE), a political prisoner is an individual that meets one of five criteria: (1) The detention violates basic guarantees in the European Convention on Human Rights, particularly freedom of thought, conscience, and religion; freedom of expression and information; and freedom of assembly and association. (2) The detention is imposed for purely political reasons. (3) The length or conditions of detention are out of proportion to the offense. (4) He or she is detained in a discriminatory manner as compared to other persons. (5) The detention is the result of judicial proceedings that are clearly unfair and connected with the political motives of authorities (Resolution 1900). We focus on human rights diplomacy around political prisoners for a variety of reasons. First, imprisoning individuals for their beliefs is a clear, stark example of human rights abuse

16 that infringes on both civil and political liberties. While there are no reliable estimates of the population of political prisoners worldwide, many of the world s most repressive countries unjustly detain hundreds or thousands of individuals for their political beliefs. For example, human rights groups estimate over 9,000 political prisoners are held in China, 60,000 in Egypt, and well over 80,000 in North Korea (Lai 2018, Hammer 2017, Hincks 2017). The arrest and release of political prisoners not only has tangible consequences for these individuals, their families, and their communities, but also can serve a bigger symbolic function to indicate major changes in human rights practices, draw international attention to a specific cause, or rally domestic political opposition. Some of the world s most well-known political prisoners Aung San Suu Kyi in Myanmar, Nelson Mandela in South Africa, and Vaclav Havel in the Czech Republic became major political figures or even heads of state. From a methodological standpoint, tracking data on political prisoners is a more concrete way to measure a set of human rights abuses than an aggregated country-level index. The international relations literature on naming and shaming, for instance, generally relies on country-level indices from Freedom House, the Polity Project, Cingranelli-Richards (CIRI) Human Rights Data Project, and similar source to trace trends in human rights practices crossnationally. While these approaches are useful for understanding macro-level trends, they are less helpful for our purposes in disaggregating when and how diplomatic actions undertaken by government officials produce concrete results. Finally, we focus on political prisoners because the #Freethe20 campaign, a human rights campaign launched by the U.S. government in September 2015 to free 20 specific female political prisoners, presents a unique opportunity to study the efficacy of HRD around this particular human rights issue. The campaign involves multiple forms of HRD, including efforts

17 to name and shame and use coercive diplomacy, coupled with attempts to publicly and privately engage foreign counterparts. The outcomes from the campaign (the release of individual political prisoners) are observable and measurable. By conducting interviews with government officials and human rights advocates involved in the campaign, we are able to detail the diplomatic strategies that were used, including those employed privately that would typically be unobservable. In the remainder of this section, we provide a brief overview of the campaign and assess its overall effectiveness. Overview of #Freethe20 campaign The #Freethe20 campaign was led by the United States Mission to the United Nations (USUN) and crafted as a response to the Beijing+20 conference held concurrently that marked the twentieth anniversary of the Fourth World Conference on Women held in Beijing, China. The initial intent of the campaign was to draw attention to the hypocrisy of the Chinese government. At the time, despite hosting a celebration of women s rights, an estimated 23 percent of over 9,000 political prisoners in China were women (Lai 2018). In addition, earlier that summer, on July 9, 2015, the Chinese Communist Party accelerated a crackdown on more than 300 human rights lawyers, over 250 of which were temporarily detained (Palmer 2017). The U.S. Mission to the United Nations (USUN), under the leadership of Ambassador Samantha Power, developed a campaign to feature 20 political prisoners including 3 women from China from 13 target countries: Azerbaijan, Burma, China, Egypt, Ethiopia, Eritrea, Iran, North Korea, Russia, Syria, Uzbekistan, Venezuela, Vietnam. Ambassador Power featured #Freethe20 women on 20 days in September at the State Department Press Briefing alongside a larger media campaign deployed via online videos, social media, and government press releases to name and shame the 13 target countries. Simultaneously, members of the U.S. Mission to

18 the United Nations, the U.S. State Department, and Congress undertook systematic efforts to exert private pressure on foreign officials. Assessing the effectiveness of a human rights campaign requires the construction of a plausible counterfactual. In this case, the relevant counterfactual question is: Would 16 of 19 women have been released in the absence of the #Freethe20 campaign and subsequent efforts? Would they have been released as quickly? Because this counterfactual is impossible to observe, we use two different strategies to construct plausible control groups with characteristics similar to #Freethe20 women. Then, we compare the release outcomes of these prisoners to the treated group, the set of 19 prisoners featured in the campaign. We supplement these analyses with qualitative research based on in-depth interviews from government officials directly involved in the #Freethe20 campaign who formerly worked at U.S. Mission to the United Nations, the U.S. State Department, or select Congressional offices. 3 After providing an overall evaluation of the campaign, we turn to this qualitative evidence to understand the different strategies officials used to conduct human rights diplomacy around political prisoners. Control Group 1: Free the 20 Long List The first control group we consider is a set of female political prisoners who were proposed for inclusion in the #Freethe20 campaign but ultimately not featured. In developing the campaign, the State Department undertook an extensive process of internal consultation in an effort to identify political prisoners that should be featured. The objective was to feature a diverse set of women from different regions and backgrounds for whom raising their case would not be harmful to their own safety or the safety of their families. This internal vetting process also enabled the Department s regional bureaus those primarily responsible for bilateral 3 This was a set of seven semi-structured interviews obtained through snowball sampling and conducted on background (under conditions of anonymity).

19 relationships to exercise some influence over which countries would be targeted. Overall, 40 women were proposed in a long list, 19 of whom were ultimately highlighted. We obtained a list of the other 21 women from the U.S. Department of State, researched their cases, and compared the outcomes of the women that made the short list to those on the long list. 4 In order to compare outcomes of these groups of women, we use survival analysis, a method for analyzing data when the dependent variable of interest is the occurrence of a particular event (in this case, release from prison). We use survival analysis because many of the outcomes we observe are right censored, meaning that we do not observe the event occurring within the specific period of time (in this case, within three years of September 2015). In Figure 1, we generate a Kaplan-Meier plot, which plots the survival curves of the women in each group (the short list in blue versus the long list in pink). A survival curve estimates the probability that an event has not yet occurred at a given time. Here, the relevant event is release from prison, and the time unit is t months after September , the start of the #Freethe20 campaign. Substantively, Figure 1 demonstrates over time, long list prisoners are less likely to be released from prison than women featured in #Freethe20. A log-rank test allows us to assess whether the survival curves of the two samples are statistically different from one another. The test is significant at the 99 percent confidence level (p < ), which suggests that the survival curve of the women on the short list significantly differs from the survival curve of the women on the long list. 4 Three of these 21 cases, all journalists from Ethiopia (Reyot Alemu, Mehlet Fantahum, and Edom Kassaye) were released in July 2015, so we do not include them in this analysis because it would not have been feasible to ultimately incorporate them in the campaign. Including these cases does not substantively change our results, but does decrease the statistical significance between Short List and Long List outcomes (p=0.058). 5 Results are the same if t is measured as time since arrest rather than time since #Freethe20 launch.

20 Figure 1 Release Outcomes for Free the 20 Short vs. Long List Strata + f20=0 + f20= Probability Unreleased p < Time in Months

21 While these results suggest that #Freethe20 was relatively effective, there are reasons to be skeptical of these findings. In particular, a primary concern is that officials working on the campaign selected cases among the set of 40 that they expected they would be most likely to influence. To understand the case selection process, we conducted in-depth interviews with seven government officials and civil society leaders involved in the #Freethe20 campaign. These officials generally expressed that while the aim of the campaign was clearly to make progress on individual cases, women were not excluded from consideration for the campaign because they had particularly difficult or intractable cases. Rather than featuring the 20 most famous female political prisoners or 20 cases that had previously received no media attention, the campaign sought to highlight a mix of different individuals. As one official described: I don t think we just went after low hanging fruit in terms of who was already being profiled Some of the cases, like Nadia [Savchenko] and [Liu Xia] were high profile prisoners that were getting significant attention from others. At the same time, we had cases of individuals who were not on anyone s radar, like the women from Ethiopia [B. Mesfin, M. Alemayehu, N. Wondifraw] and Syria [Rasha Sharbaji] I think we had a mix of cases. Officials we interviewed also noted the campaign wanted to feature a diverse set of women, both in terms of occupation and region, so if two women shared a similar background and country, typically only one made the short list. Importantly, both lists included cases in countries with whom the United States has a very deep and complicated bilateral relationship (ex: Egypt, Russia, Syria). However, it was also clear that in some cases that a few women did not make the short list because the State Department was balancing human rights concerns with a set of competing security or economic interests. For example, political prisoners in Bahrain were likely not included in the short list due to the strategic importance of Bahrain in U.S. military presence in the Gulf Region ( U.S. Security Cooperation with Bahrain ).

22 Control Group 2: Amnesty International Urgent Actions Given the small sample size and our concerns about how cases on the short and long list may have differed, we used a second strategy to assess the relative effectiveness of the campaign. We constructed an alternative control group from the set of all female political prisoners in recent years imprisoned in the #Freethe20 target countries: Azerbaijan, Burma, China, Egypt, Ethiopia, Eritrea, Iran, North Korea, Russia, Syria, Uzbekistan, Venezuela, Vietnam. As previously discussed, there are considerable challenges to collecting reliable information about arrests and releases of political prisoners, particularly in repressive contexts with neither independent media nor active civil society. Since no comprehensive database of political prisoners exists, we identified this set of cases from Amnesty International Urgent Action reports. Amnesty International is international non-governmental organization that focuses on human rights advocacy. A core component of their work is the creation of an Urgent Action Network that encourages citizens around the globe to write letters, s, faxes, and social media posts to exert public pressure on a target government with respect to individual cases of human rights abuses ( Urgent Action Network ). Amnesty International issues Urgent Actions (UAs) to it global network, many of which feature political prisoners or other individuals unjustly targeted by their government. One virtue of working with data on UAs is that this subset of political prisoners is already benefiting from international attention and pressure. As a result, a key difference between the #Freethe20 women and the larger set of Amnesty cases was the diplomatic pressure exerted by the United States. To construct a dataset of female political prisoners, we started with an existing database of Urgent Actions based on work by Judith Kelley and Dan Nielson (Kelley and Nielson 2015). We updated this data through September 2015, the start of the #Freethe20 campaign (over 1100

23 unique UAs). After identifying all Urgent Actions from the target countries between January 2000 and September 2015, we looked at the underlying report from each of the UAs and coded the name and sex of all individuals listed in the report (over 2800 unique names). Figure 2 shows the total number of UAs and the distribution of males and females mentioned in each, broken down by target country. To construct a comparable set of cases for a control group, we focused only on named females featured in Urgent Actions (n=455). Since there are often multiple Urgent Actions for each case, we began by aggregating data at the individual level. Prior to collecting additional information, we looked at how the demographic characteristics of women featured in #Freethe20 differed from the rest of the women in the UA dataset. Figure 3 plots the standardized mean difference between these two subsamples across different characteristics: the number of times they appear in UA reports, the concerns detailed in their case, their country of origin, and their entry date into the database.

24 Figure 2 Females in Amnesty International Urgent Actions ( ) 1500 Count 1000 Male Female Azerbaijan China Egypt Eritrea Ethiopia Iran Myanmar Country Russia Syria Uzbekistan Venezuela Vietnam

25 Figure 3 Standardized Bias Plot (Freethe20 vs. All Female UAs) Count of UAs Death Penalty Concerns Medical Concerns Prisoner of Conscience Torture Concerns Trial/Legal Concerns Year of Entry Azerbaijan China Egypt Eritrea Ethiopia Russia Iran Myanmar Syria Uzbekistan Venezuela Vietnam Standardized Bias

26 A few results are worth noting. First, for the most part, #Freethe20 women look fairly similar in terms of their demographic characteristics when compared with other women from the UA dataset. The main differences are that, on average, #Freethe20 women entered the dataset more recently and their UAs were less likely to contain reports of torture. There are also some regional imbalances in the data. Specifically, women from Iran make up a high percentage of the UA data (roughly 30 percent) but only 1 Iranian woman (Bahareh Hedayat) was included on the #Freethe20 list. The most important takeaway from Figure 3 is that the number of times each of the #Freethe20 woman appeared in the dataset ( Count of UAs ) does not significantly differ from the number of times the average woman appeared in the data. In other words, we interpret this to mean that #Freethe20 women were not disproportionately profiled prior to the launch of the campaign relative to the rest of the sample. Among this set of female political prisoners, we then systematically collected data about their release. We first determined whether or not the individual case could be plausibly coded a prisoner for a campaign like #Freethe20. Here, individuals had to meet two criteria: (1) have been reported arrested, detained, or disappeared (some UAs, for example, report threats of violence or harassment without detention), and (2) have been detained or imprisoned for a sustained period of time, a minimum of two months. Third, we recorded the date the individual was arrested or detained as reported in the UA. Fourth, we conducted additional research to see whether or not the individual had been released and, if so, recorded the day and month of release. Finally, we coded indicators about our level of certainty about the outcome.

27 We use survival analysis in order to analyze the differences in release outcomes for #Freethe20 women versus the broader sample of female political prisoners. We use Cox proportional-hazard models in order to control for other covariates without imposing a specific functional form on the data. Cox models are semi-parametric ; a core assumption made in the model is that the effect of the covariates on survival time (also conceptualized as time to event, which, in this application, means time to release from prison in months) is constant over time. Table 1 displays the results of four different Cox models, which estimate the relationship between a set of covariates and the probability of an event occurring. The unit of analysis is the individual, and the main coefficient of interest is the coefficient on a #Freethe20 indicator variable (coded as a 1 if the woman was featured in #Freethe20 and a 0 otherwise). In different specifications, we control for other covariates, including indicators for characteristics of the political prisoner and regional dummies. 6 Intuitively, positive coefficients are associated with a higher likelihood that the event will occur (in this case, will be released from prison). The coefficient on Free the 20 is positive and statistically significant at the 90% confidence level across the different specifications in Table 1. For a more substantive interpretation, we can consider the exponentiated coefficient, which ranges in magnitude from 1.86 to 2.22 across different models specifications. In a Cox model, the exponentiated coefficient can be interpreted as a hazard ratio, which for a dichotomous variable would capture the ratio of the risk of an event occurring when X = 1 relative to the risk of an event occurring when X = 0. Substantively, this means that the rate at which women featured in #Freethe20 were released from prison is roughly two times that at which other female political prisoners were released. 6 The regional dummies coded for the #Freethe20 target countries are: Asia (China, Myanmar, Vietnam), Eurasia (Azerbaijan, Russia, Uzbekistan), Africa (Eritrea, Ethiopia), Middle East (Egypt, Iran, Syria), and Americas (Venezuela).

28 Table 1: Survival Analysis for Female Political Prisoners Cox Proportional Hazard Models Time to Release (1) (2) (3) (4) Free the (0.279) (0.321) (0.303) (0.349) Prisoner of Conscience (0.215) (0.219) Any Torture Concerns (0.252) (0.261) Death Penalty Concerns (0.240) (0.253) Any Legal Concerns (0.256) (0.259) Any Medical Concerns (0.283) (0.287) Asia (1.045) (1.074) Middle East (1.050) (1.080) Eurasia (1.091) (1.123) Africa (1.097) (1.123) Americas (1.417) (1.456) Observations Note: p<0.1; p<0.05; p<0.01

29 V. Evidence for Naming and Shaming Both strategies used in the previous section suggest that women featured in the #Freethe20 campaign were released from prison at a faster rate than they would have been otherwise. The natural next question, therefore, is why? The #Freethe20 campaign was designed primarily to bring public pressure, by virtue of naming and shaming, on target states. As one official told us, #Freethe20 was conceived of as an effort to draw attention and create some energy around [these cases] and break through some of the noise (Interview 1, August 8, 2016). In this section, we explore whether public attention to these women substantially increased as a result of the campaign. While we do find quantitative evidence that both media coverage and online searches of women featured in #Freethe20 increased during the campaign launch, our qualitative evidence suggests that public pressure itself did not change the cost-benefit calculus of leaders in the targeted states. To name and shame #Freethe20 countries, government officials at the U.S. Mission to the United Nations and partners in different divisions of the U.S. State Department used a variety of strategies. In total, there were five public-facing elements of the campaign. First, the U.S. State Department issued press releases and daily press briefings in which they featured stories of the individual women throughout the month of September Second, the campaign created and maintained an online presence, which included a website, online videos about each individual woman, and a social media presence via Facebook and Twitter. Third, stories featuring the #Freethe20 campaign were featured in major media outlines like The New York Times, The Wall Street Journal, USA Today, and internationally in outlets like The Guardian, Radio Free Europe, and Voice of America. Fourth, the U.S Mission to the United Nations created a public-facing display in their building on First Avenue visible to press and foreign