Where Help Is Needed Most? Explaining Reporting Strategies of the International Trade Union Confederation

How do international non-governmental organizations select countries for naming and shaming? I argue that three focal actors influence non-governmental organizations’ shaming decisions: inter-governmental organizations, governments, and non-governmental organization members. Moreover, drawing on existing research, non-governmental organizations might respond differently to focal actors’ preferences, by either targeting states that have been criticized by focal actors or, alternatively, targeting those who have escaped their scrutiny. To test these propositions, the article conducts multiple interviews and gathers original data on shaming within the International Trade Union Confederation during period 1991–2011. The main findings are threefold. First, focal actors, except governments, have a significant influence on International Trade Union Confederation shaming. The members’ preferences regarding which states should be singled out have the strongest impact. Second, the International Trade Union Confederation is likely to adopt a bandwagoning strategy by shaming states that have been targeted by focal actors, rather than focusing on states that have escaped their criticism. Third, as a result of the bandwagoning approach, the International Trade Union Confederation mainly targets states with poor labor rights conditions.


Introduction
Over the last 40 years, international non-governmental organizations (INGOs) have played a crucial role in international politics. INGOs operate at various levels, providing information to inter-governmental organizations (IGOs), monitoring compliance with international norms, and influencing the domestic and international policies of states. To do this, they utilize a wide range of advocacy strategies (Murdie, 2014;Stroup and Wong, 2017: 65-88). One of the most critical assets of INGOs is their ability to generate and disseminate information that can mobilize domestic and international organizations (Keck and Sikkink, 1998;Murdie and Davis, 2012). To increase their relevance on the international arena, INGOs now operate through multiple channels. As their formal access to intergovernmental organizations (IGOs) has increased, INGOs spend considerable resources on monitoring and reporting on violators of international agreements (Johnson, 2014;Tallberg et al., 2013). For instance, within the United Nations (UN) system, there are currently thousands of registered INGOs that allocate substantial resources to "naming and shaming" states that violate international norms within IGOs like the Economic and Social Council (ECOSOC) and the Human Rights Council (UNHCR). 1 By reporting on states' violations, INGOs strive to hold perpetrators accountable for their wrongdoings and facilitate political change in these states. But how do INGOs decide which states to single out for "naming and shaming"? 2 While we have considerable knowledge of how and when governments within international organizations (IOs) target violators (Koliev, 2019;Terman and Voeten, 2018), we know little about what factors influence INGOs' decisions to target states for naming and shaming in IOs.
Consider the following examples. Between 2000 and 2008, the International Trade Union Confederation (ITUC), the leading labor rights INGO (Reinalda, 2019: 237), regularly criticized the government of Myanmar for its use of forced labor in the military. The ITUC compiled new evidence of forced labor and filed a report against Myanmar within the International Labour Organization (ILO), condemning the government and putting pressure on the ILO and its member states to punish Myanmar. In their efforts to eliminate the practice, the ITUC also compiled a blacklist of multinational companies working in Myanmar (ILO, 2001). In parallel with the Myanmar case, the ITUC conducted multiple interviews with workers in Mauritania and invested considerable resources in investigating the practice of slavery in the country. In 2001, the ITUC publicly alleged Mauritania for violating the ILO's conventions on forced labor. In their report to the ILO, the ITUC stated, for instance, that "a young man and a 13-year-old girl were forced to work for their masters as a shepherd and camel herder respectively before escaping and being caught with the help of the police," and demanded immediate measures against these practices (ILO, 2001). Yet, during the same period, the Nigerian government's anti-union measures and use of forced labor did not receive the ITUC's attention. This is despite the fact that ITUC members-Nigerian unions-accused the government of grave violations of major international labor conventions, such as the dissolution of central labor unions.
These examples generate an empirical puzzle regarding how INGOs select states for public criticism. This question is important in the light of previous research that links INGO and IGO naming and shaming with human rights improvements (Koliev et al., in press;DeMeritt, 2012;Murdie and Davis, 2012). It is also imperative to the study of NGOs and international politics to understand the politics behind INGOs' targeting behavior and the extent to which other factors and actors influence their decisions.
Drawing on previous research, this article argues that three focal actors shape INGOs' shaming decisions: IGOs, governments, and members or donors 3 (see, for example, Carpenter, 2014;Keck and Sikkink, 1998;Stroup and Wong, 2017). The preferences of these focal actors as regards which states to single out play a decisive role for INGOs' targeting decisions due to reputational and strategic benefits. But reputational and strategic considerations might be a doubled-edge sword (Gent et al., 2015), and, given the limited resources of INGOs, the focal actors' actions might incite INGOs to adopt different approaches to the selection of targets. Building on the theoretical work within the international political economy and constructivist literatures, the article offers two competing hypotheses, one that assumes that the ITUC adopts a strategic substitution approach and one that proposes the ITUC's use of a strategic bandwagoning 4 approach when selecting target states (Murdie and Urpelainen, 2015). The substitution logic builds on the idea that INGOs are likely to focus on countries that have escaped the scrutiny of the focal actors. In contrast, the bandwagoning strategy asserts that INGOs will follow the leads of focal actors as they seek to add more fuel to already articulated domestic and international criticism. While the existing literature stresses these actors' influence on INGO behavior, no previous studies have assessed their relative influence or effects on INGO shaming strategies.
The article examines the shaming decisions of the ITUC from 1990 to 2011. The ITUC is by far the largest and most central INGO within the labor rights area. It represents 207 million workers in 164 countries. Its primary mission is to promote and defend workers' rights and interests "through international cooperation between trade unions, global campaigning and advocacy within major global institutions" (https://www.ituc-csi.org/aboutus?lang=en). Akin to many other INGOs, the ITUC relies on membership funding to carry out its work. As one of the leading labor rights INGOs, it performs the important roles of monitoring and improving labor standards through naming and shaming activities. By collecting and disseminating information on labor rights violations, the ITUC puts pressure on countries to comply with international norms. But despite its size and significance for international labor rights, the ITUC has gained surprisingly little, if any, attention among international relations scholars. 5 To test the two competing hypotheses, a novel times-series cross-sectional data set on the ITUC's shaming decisions-pertaining to four labor rights issues, including forced labor, child labor, discrimination, and unions' rights to organize and bargain collectively-were compiled from annual documents of the ITUC and the ILO. To assess the influence of IGOs and governments, the article focuses on the ILO's and its memberstates' public criticism of norm-violating countries. The ILO is a focal actor for the ITUC in the same way as the UN's Human Rights Council is strategically important to human rights INGOs like Amnesty International and Human Rights Watch (Thakur, 1994). To gauge the impact of members, the article gathers data on the member-unions' criticism. Although INGOs usually do not include labor unions as members, the ITUC is similar to other INGOs in the sense that they depend on their funding and strive to keep their beneficiaries loyal (Reinalda, 2019). In addition to the quantitative data, seven semi-structured interviews were conducted with senior officials at the ITUC, the ILO, and national and regional unions in order to provide supplementary information on what factors influence the ITUC's shaming decisions. 6 The findings of this article suggest that focal actors' preferences have a significant influence on the ITUC's selection of state targets. In short, the empirical analysis provides strong evidence that the ITUC is more likely to adopt the bandwagoning approach than the substitution approach in response to focal actors' preferences. The influence of governments on ITUC decisions is, however, not significant in the analysis. Accounting for alternative factors, such as respect for labor rights, economic development, regime type, and population, the preferences of member-unions have the strongest impact on the ITUC's shaming decisions, Taken together, the study makes several novel contributions to the literature on INGOs and labor rights. The article is the first to empirically assess the relative influence of IGOs, governments, and members on INGOs' shaming decisions. It theorizes how and when IGOs, governments, and members might influence the targeting decisions of INGOs, which has been overlooked in existing research. Moreover, this study moves beyond physical integrity rights and focuses on four different labor rights issues, including child labor, forced labor, and unions' rights to organize and bargain collectively.
The findings of this study have several important implications for INGO research and the literature on human and labor rights in IR. First, in order to understand INGOs' effectiveness, there is a need to first analyze how and why INGO strategies are shaped. This article suggests that future studies assessing the impact of INGO naming and shaming, or similar strategies, should account for the influence of focal actors. Most studies assume that INGO shaming is not influenced by focal actors, which introduces bias into their estimations. For example, reports by the UN's Human Rights Council arguably influence the targeting decisions of Amnesty International, just like George Soros' 100 million dollar contribution might influence the decisions of Human Rights Watch. When failing to understand how INGO shaming is shaped and correcting for this hurdle, examining the impact of shaming is difficult. Second, if the bandwagoning approach is favored because it is perceived as most efficient, as suggested by this article, then future studies should investigate the extent to which the bandwagoning strategy is effective in relation to the substitution approach. Third, for labor rights scholars, this study contributes with insights into how monitoring of international labor standards works and when norm-violators are exposed to the international community. More specifically, it nuances the role of the ITUC and unpacks its strategic considerations in relation to the ILO's and its members' preferences regarding the selection of violators for naming and shaming. The article suggests that members have a stronger influence on the selection of targets than the ILO and governments. In addition, one implication of adopting a bandwagoning approach might be that INGOs target "low hanging fruit" cases, as proposed by some scholars (Stroup and Wong, 2017: 66). The findings of this study indicate otherwise, suggesting that INGOs generally target norm-violating states, even when they adopt the bandwagoning strategy. Fourth, and more broadly, the article contributes to the research program on the role of INGOs in IR, offering novel insights into how focal actors and INGOs interact. The evidence suggests a more nuanced and unified theoretical model where INGOs are not solely treated as "shapers" , echoing other researchers' calls for future research on INGOs (Heiss and Johnson, 2016: 538). Indeed, the article paves way for research extensions, examining questions of whether, when, and why IOs may permeate INGOs' advocacy strategies.
The close interaction between the ILO and the ITUC might provide reasons to assume that it is a "most likely" case for the bandwagoning approach. In this context, it is important to note that this relationship is not exclusive to the ITUC as an INGO, and neither is it a necessary condition for bandwagoning. Whereas the decision-making in the ILO is consensus-based when it comes to adopting conventions and other instruments, the public identification of violators by the ILO is confrontational and contested (Koliev, 2019;Koliev and Lebovic, 2018). Although caution always should be raised when generalizing, the theoretical reasoning of this study should be applicable to other INGOs with similar organizational characteristics that work within the human rights area.
The remainder of this article is organized as follows. First, the article provides a brief literature review on the impact and determinants of INGO shaming. Second, the theoretical expectations are elaborated and a set of hypotheses is formulated. Third, the article proceeds to discuss the research design and the data, which is followed by the empirical analysis. Finally, the article offers conclusions.

INGO Naming and Shaming
The extensive literature on the impact of INGO naming and shaming has demonstrated that the international exposure of state violations can lead to compliance with international agreements. For instance, empirical studies have suggested that INGO shaming can induce compliance by mobilizing domestic reform-minded groups and by triggering economic and political considerations at the elite level (Ausderan, 2014;DeMeritt, 2012;Franklin, 2008;Hendrix and Wong, 2013;Krain, 2012;Murdie and Davis, 2012). Researchers have also shown that INGOs' shaming activities or dissemination of negative information can lead to economic sanctions, a loss of foreign direct investments (FDIs), and humanitarian interventions (Lebovic and Voeten, 2009;Murdie and Peksen, 2013;Woo and Murdie, 2017).
While few scholars disagree with the notion that INGO naming and shaming can influence state behavior, studied have come to debate the politics behind targeting strategies in order to uncover potential biases introduced by selection practices and to explain the strategies and logics of INGOs (Bob, 2012;Cooley and Ron, 2002;Hadden and Jasny, 2019;Ron et al., 2005). Several studies on the determinants of INGO shaming have demonstrated that INGOs target states strategically, particularly states that constitute frequent human rights violators (Asal et al., 2016;Hendrix and Wong, 2014;Ron et al., 2005). Other studies have argued that INGOs are likely to target "low hanging fruit" in order to be perceived as efficient and worthy of financial support (Stroup and Wong, 2017: 66). Moreover, the existing literature has suggested other, equally important factors that may explain INGOs' selection of targets. For instance, Ron et al. (2005) find that INGOs' decisions are shaped by the size of media coverage, population, and foreign aid. Hendrix and Wong's (2014) study shows that INGOs strategically choose their targets based on security links with the United States, highlighting the influence of governments on Amnesty International's shaming decisions. Murdie and Urpelainen (2015) find that environmental INGOs target states where domestic actors lack opportunities to criticize the government. Other studies suggest that INGOs primarily base their decisions on their donors' and/or members' preferences and the domestic political environment that they act within Meernik et al., 2012;Murdie and Urpelainen, 2015).
While these studies enhance our understanding of how INGOs select their targets, we still lack sufficient answers to the questions of, first, how and when INGOs target states within IOs and, second, whether and how focal actors' preferences shape INGOs' naming and shaming decisions. No previous studies (known to the authors) have empirically assessed the relative influence and behavioral effects of IGOs, governments, and members on INGO shaming even though each of these actors' influence on INGOs has been highlighted in separate studies (Cooley and Ron, 2002;Hendrix and Wong, 2014;Meernik et al., 2012;Tallberg et al., 2013). Moreover, with a few notable exceptions, most studies of the determinants of INGO shaming have focused on physical integrity rights and Amnesty International, which raises questions about the generalizability of these findings across different types of INGOs and issue areas.
The article makes several contributions to the literature on INGOs and labor rights. First, it theorizes how and when IGOs, members, and governments may influence the targeting decisions of INGOs. Second, while previous studies have focused mainly on Amnesty International and the human rights area, this article focuses on the ITUC and moves beyond physical integrity rights by focusing on several different issue areas, including child labor, forced labor, and unions' rights to organize and bargain collectively. Third, unlike previous scholarship, the article focuses on how INGOs select target states within IOs through their own reports and press releases.

Theory
In this article, INGOs are viewed as self-interested and morally driven actors that strategically select targets and respond to the preferences of their members or donors (Asal et al., 2016;Bob, 2005;Dellmuth and Bloodgood, 2019;Mitchell and Schmitz, 2014;Prakash and Gugerty, 2010). To achieve their goals, INGOs calculate which strategies are most likely to be effective (Bloodgood, 2011;Bob, 2005). Yet, their calculations regarding shaming decisions are not conducted in a vacuum. While the focal actors can vary depending on the issue area and the specific INGO, this article contends that INGOs' shaming decisions are significantly influenced by the preferences and actions of IGOs, governments, and members. While the members of INGOs are crucial for financial and organizational ("survival") reasons (Cooley and Ron, 2002), IGOs and governments constitute the cornerstones of the international system, making them focal actors for INGOs' abilities to reach their multifaceted goals (Keck and Sikkink, 1998). INGOs must be loyal to their members and donors in order to secure the resources necessary to carry on their work. They also seek to be viewed as politically impartial and as credible sources of information, thus maintaining a good reputation and legitimacy vis-à-vis these focal actors (Asal et al., 2016: 241;Gent et al., 2015;Hendrix and Wong, 2014: 35;Tallberg et al., 2013). As observed by other scholars, INGOs not only shape but are also shaped by the focal actors that they interact with (Aldrich, 2008;Hendrix and Wong, 2014;Stroup and Wong, 2017).
Much like the way in which INGOs observe and respond to decisions of media outlets to report on certain countries, they are also influenced by the targeting decisions of members, IGOs, and states. Given the prominent role of the ILO within the labor rights area, for instance, its prior criticism of countries might influence the ITUC's selection of targets . The same logic applies to the role of the UN's human rights bodies and their influence on, for instance, Amnesty International's targeting (Thakur, 1994: 147). For example, in 2014, Sweden received harsh criticism from the UN Human Rights Council for violating the UN Convention against Torture. According to the UN, Swedish authorities had repeatedly detained children in isolation, thus breaching the convention. This criticism arguably influenced Amnesty International, which one year later singled out Sweden for this specific violation in their campaigns and reports. By the same token, governments' preferences and actions toward violating states might also influence INGOs' naming and shaming politics since governments play an important role in pressuring the violator states to comply with international agreements (Keck and Sikkink, 1998).
Interviews with ILO and ITUC officials provide more insights. ILO officials emphasize local members' credibility and knowledge of the situation in their countries and their financial support to the ITUC as sources of their influence on ITUC decisions (Interview 1: Interviews with ILO official in Geneva, Switzerland, and via e-mail, 2018 and Interview 2: Interview with ILO official. Geneva, Switzerland, 1 June 2017). ITUC officials, on the other hand, underscore the ILO's competence within the labor rights area as the main reason for why the ITUC observes its actions (Interview 3: Interview with ITUC official. 24 August, via Skype, 2018 and Interview 4: Interview with ITUC official. Geneva, Switzerland, 30 May 2018). Moreover, ITUC officials acknowledge that they closely observe the decisions of IGOs, such as the ILO and the World Trade Organizations, and of their member states and to some extent act based on these.
H1. The preferences of focal actors have a significant influence on ITUC's shaming decisions.
For strategic reasons, however, the ITUC might respond differently to the decisions of IGOs, governments, and members. While INGOs strive for a good reputation and credibility among focal actors, this may lead INGOs to adopt different strategies, especially in an environment with multiple objectives. Indeed, INGOs are faced with several (potentially conflicting) objectives when singling out countries for criticism, including (1) exposing the violations of states, (2) inducing change in target states by mobilizing thirdparty actors, (3) serving the interests of members and donors, and (4)  Building on the previous literature and the accounts of political economy and constructivist theorists, this article argues that INGOs may adopt two competing reporting strategies: the strategic substitution approach or the bandwagoning approach (Murdie and Urpelainen, 2015: 354). According to the substitution approach, INGOs are likely to target states that would otherwise escape the scrutiny and criticism of focal actors, whereas the bandwagoning approach expects INGOs to single out states that have previously been criticized by IGOs, governments, and members. The two different logics generate different expectations.

The Arguments for the Substitution Approach
Given the influence of focal actors on INGOs' selection of state targets, there are theoretical and empirical reasons to assume that INGOs might adopt different strategies in response to these actors' actions. There are three main reasons for why INGOs may adopt the strategic substitution approach.
First, INGOs are likely to target uncriticized states since these states otherwise might avoid the public scrutiny of international and domestic actors. INGOs might, then, expect greater benefits-reputational and strategic-from targeting states that would otherwise have received no, or less, scrutiny and criticism from focal actors. Second, INGOs are likely to adopt a substitution approach because IGOs' demand for new or additional information is one of the main reasons for INGOs' access to IGOs (Dellmuth and Tallberg, 2017;Johnson, 2014;Tallberg et al., 2013;Vabulas, 2013). By "filling the gap," they are complementing the scrutiny gaps created by IGOs and governments. Third, the substitution approach also serves the purpose of demonstrating INGOs' important function to the less resourceful members or members from oppressive states-conceivably reminding them of how much these members' success and survival depends on their work. Anecdotal evidence illustrates these dynamics.
In 2008, the ITUC published several reports targeting Iran and its wrongdoings with regard to anti-discriminatory policies toward women. While the Iranian national unions had been asked by the ILO to report on the government's implementation of the convention on discrimination, none of the unions had filed a complaint against the governmentarguably for political reasons. Hence, the ITUC used its networks in the country to reveal to the ILO and its member states that Iran had been secretly applying quotas restricting women's access to university (ILO, 2008). By adopting the substitution strategy, the ITUC not only made Iran accountable for its violations of several fundamental conventions; it also demonstrated its efficiency and value vis-à-vis the ILO, governments, and its member-unions in Iran. Interviews conducted with the ITUC representatives, ILO officials, and national labor unions provide additional insights. According to interviews, the ITUC often adopts a substitution strategy when the ILO fails to pay sufficient attention to certain violations and when national unions lack the opportunities to assess and report on violations (Interview 3: Interview with ITUC official. Via Skype, 24 August 2018 and Interview 4: Interview with ITUC official. Geneva, Switzerland, 30 May 2018). The ITUC official stressed that the ITUC's independence and resources allow it to focus on countries that need its help the most, "while also helping members from developed states" (Interview 3: Interview with ITUC official. Via Skype, 24 August 2018). Moreover, by adopting a substitution approach, the ITUC strives to bring up new or less exposed countries on the global labor rights agenda. Taken together, the implication of the substitution logic is that the ITUC is expected to on average target countries that have escaped the scrutiny of focal actors.
H2A. The ITUC targets states that have not been criticized by the ILO.
H2B. The ITUC targets states that have not been criticized by governments.
H2C. The ITUC targets states that have not been criticized by its member-unions.

The Arguments for the Bandwagoning Approach
The bandwagoning logic offers a competing story. In order to maximize their visibility and amplify the existing pressure, the bandwagoning approach suggests that INGOs recognize the benefits of selecting states that have already been targeted by the focal actors rather than those that have escaped scrutiny. There are several reasons for this logic. First, the focal actors' criticism mobilizes INGOs to act as these decisions indicate sufficiently serious violations for INGO attention (Kelley and Simmons, 2015: 58-59). Second, in contrast to the substitution approach, this logic emphasizes that INGOs expect more leverage from joining in on previous or existing criticism since the combined pressure is expected to increase the chances of producing change (Keck and Sikkink, 1998). In this view, INGOs may also gain attention for their work without serving as a substitute for absent domestic/member activism (Cooley and Ron, 2002;Ron et al., 2005). Third, seeking to uphold a credible and impartial image, targeting the same countries as focal actors may portray INGOs as credible collaborators (Stroup and Wong, 2017). The anecdotal evidence also provides support for the bandwagoning hypothesis.
During the period 2006-2008, the Hungarian government, led by Viktor Orbán, interfered in the collective bargaining system and passed laws that weakened the position of labor unions on the labor market. The ILO criticized the government's policies for restricting the unions' fundamental rights to bargain collectively and thus weakening their status vis-à-vis employer organizations. In 2008, five Hungarian unions carried out demonstrations, criticizing the government for its anti-union agenda, and filed official complaints to the ILO. The next year, the ITUC decided to join the criticism and thus publicly criticized the Hungarian government for its lack of plans to improve the labor rights situation. Hungarian labor union representatives indicated that the ITUC's criticism was imperative for the national unions' ability to put international pressure on the government, stating that "we always want them [ITUC] to join our cause" (Interview 5: Interview with Hungarian trade union representative. Geneva, Switzerland, 30 May 2018). For the ITUC, the bandwagoning strategy also represented a way to intensify the criticism of serious labor rights violations and to show their loyalty with members and domestic groups (Interview 1: Interviews with ILO official in Geneva, Switzerland and via e-mail, 2018; Interview 6: Interview with Swedish trade union representative. Stockholm, Sweden, 26 September 2018; Interview 7: Interview with Guatemala trade union representative. Geneva, Switzerland, 2 June 2018). This logic does not necessarily imply, however, that states with oppressive regimes-without domestic and focal actor criticism-are completely ignored by INGOs. Rather, the argument is probabilistic and stresses INGOs' strategic approach in order to maximize the leverage of their shaming actions. This discussion leads us to the strategic bandwagoning hypotheses: H3A. The ITUC targets states that have been criticized by ILO.
H3B. The ITUC targets states that have been criticized by governments.
H3C. The ITUC targets states that have been criticized by its member-unions.

Research Design
The ITUC provides us with a suitable environment to test the hypotheses as it offers a unique opportunity to empirically assess if and how IGOs, governments, and members shape INGOs' targeting decisions, which has not been possible in previous studies. 7 In order to assess the influence of IGOs and governments, the article focuses on the ILO and its member states.
One should note, though, that the close relationship between the ILO and the ITUC might limit the generalizability of the findings of this study to INGOs that have a close relationship with IGOs, formal access, and a member base structure. It could also be viewed as a "most likely" case for IGO influence. However, a close relationship with an IGO is by no means exclusive to the ITUC, but also applies to other INGOs, such as Greenpeace and the UN's Environmental Programme (UNEP). Furthermore, it is important to acknowledge that the ITUC by design has a considerably narrower political agenda than the tripartite ILO. The ILO puts the interests of member organizations first, while the ITUC represents narrow interests of its members. While the ILO is unique for its tripartite structure and is perceived as more consensus-oriented with regard to decision-making related to the adoption of conventions and recommendations, the ILO's public identification and scrutiny of norm-violating states is confrontational and contested (Koliev, 2019;Koliev and Lebovic, 2018;Weisband, 2000). In this respect, which is of analytical interest to this study, the ILO resembles other IOs, such as the UN's Human Rights Council (Lebovic and Voeten, 2006). While acknowledging that caution should be paid when generalizing, the article contends that the theoretical logic here could be applied to INGOs like Amnesty International and Greenpeace. The ITUC's formal access to the ILO also allows us to partially address the critical question of whether INGOs are more likely to be influenced by their members' decisions than the decisions of powerful IGOs (Gent et al., 2015;Stroup and Wong, 2017).
In order to test the hypotheses, the article utilizes an original cross-sectional time-series data set on ITUC naming and shaming from 1990 to 2011. 8 The analysis centers on the ITUC naming and shaming with regard to the ILO's core conventions (Table 1). While the ILO monitors compliance with 180 conventions, I focus on the eight so-called fundamental conventions, which are considered the major international labor conventions and are explicitly prioritized by the ILO as well as the ITUC. They are also the most widely ratified ILO conventions, reaching close to universal ratification.

Dependent Variable
The main dependent variable captures whether a country has been targeted by the ITUC within the ILO system. Since the ITUC can target states multiple times in the same year with respect to fundamental conventions, it is a count variable with country-year as the unit of analysis. Note though that while the data does not capture the intensity of shaming, this approach still provides important insights regarding INGOs' selection strategies. In total, the ITUC performed 1060 naming and shaming acts, directed at 138 different countries, during the study period, with Myanmar (11), Sudan (10), Guatemala (9), Colombia (9), and Venezuela (9) being the five most targeted states. Figure 1 displays the average number of targeted states by the ITUC over three different time periods. During the period 1990-1997, the ITUC on average targeted two countries per year. Among the most targeted states were Saudi Arabia, Venezuela, Libya, Colombia, Iran, and Pakistan. During the period 1998-2005, the ITUC on average targeted 16 countries per year. Guatemala, Zambia, Sri Lanka, Indonesia, and the Dominican Republic were among the most targeted countries during this period. During the period 2006-2011, the ITUC on average targeted 68 countries per year. The data suggest that during the early 1990s, the ITUC mainly focused on the worst violators of labor rights but steadily increased its scope to other states, including developed countries like the United States, Canada and the United Kingdom. There are historical explanations for this pattern. Before the creation of the ITUC, the international trade unions were divided, less resourceful, and poorly coordinated. However, this changed when the two major unions, the International Confederation of Free Trade Unions (ICFTU) and the World Confederation of Labour (WCL), merged into one considerably more resourceful organization: the ITUC. 9 Since 2005, the ITUC has dedicated increasing resources and attention to its reporting activities. Another explanation for the increase in shaming decisions is the ITUC's increasingly "confrontational" approach to shaming, presumably influenced by the success of the shaming strategies of other INGOs, such as Amnesty International, over traditional methods of dialogue and other union mechanisms to put pressure on governments. In addition, as a robustness check, the article employs another dependent variable that captures ITUC shaming in the reports published on the ITUC's website. I expect the hypotheses to hold even here for our main independent variables. 10 These shaming acts are published as press releases, reports, and resolutions. This variable is also a count variable that measures the total number of shaming activities per country and year. The data are limited to the period 2006-2011 since the ITUC started issuing the outside reports in 2006 and the main explanatory variables lack data after 2011. Over the period 2006-2011, the ITUC targeted 56 different countries through press releases and resolutions, on average 23 countries per year. Guatemala (6), Columbia (6), Iran (5), Turkey (5), and Honduras (5) constitute the five most frequent targets of the ITUC through press releases, reports, and resolutions during this period.

Independent Variables
The key independent variables capturing the targeting behavior of the focal actors are ILO criticism, Government criticism, and Labor union criticism. The ILO criticism variable captures the total number of so-called "observations" issued for each country-year by the ILO's main supervisory body, the Committee of Experts on the Application of Conventions and Recommendations (CEACR). The observations are considered as public criticism of target states and imply that the government has failed to comply with its obligations Weisband, 2000). The government criticism, which is expressed during the annual conference, captures the number of times that a violating state has been verbally condemned by governments at the annual sessions. The variable Labor union criticism captures the criticism of the ITUC's members, that is, labor unions. The member-unions submit their statements to the ILO, before the ILO and governments make their decisions. Figure 2 shows that average number of states targeted by the ILO, governments, and national unions. While the governments are most restrictive with public criticism, the ILO on average targets about 109 countries per year. 11   suggests that the ITUC must make strategic decisions since previous criticism by its members and by the ILO exceeds its average number of targeted states.
Using the CIRI's workers' right categorization, 35% of labor union criticism originates from countries with "severely restricted" labor rights conditions, 45% comes from countries with "somewhat" restricted labor rights, and 20% originates from countries with "fully protected" labor rights (Cingranelli et al., 2014). 12 Arguably, one potential reason for the lower incidence of union criticism from countries with "severely restricted" labor rights is their lack of resources and opportunities to criticize their governments. As for the ILO and the government criticism, about 60%-65% is directed toward countries with "severely restricted" or "somewhat restricted" labor rights conditions. In general, the descriptive statistics suggest that countries with poor labor rights conditions are more likely to be criticized by their labor unions as well as by the ILO and governments than countries with protected labor rights.

Control Variables
The previous scholarship and interview material suggest other sources of influence that merit several control variables in the regression models. First, worker rights violations and country size have been shown to be significant predictors of INGO shaming (Ron et al., 2005). Hence, the control variables capturing the level of worker rights protection and the population size are included.
Second, research shows that economic variables can explain the variation in shaming decisions by INGOs. In order to account for economic factors and conditions, three variables are included: gross domestic product (GDP) per capita, Trade, and FDIs. As suggested by some interviews, the ITUC might be more willing to target richer countries since member-unions from these countries are more resourceful and may have a greater impact on international labor standards. However, the ITUC might also target countries with a low level of economic development if the circumstances suggest that labor rights are violated. As for the trade and FDI variables, research suggests a mechanism where INGOs are likely to target trade-and FDI-dependent countries as these may be more sensitive to international criticism due to the fear of losing future investments (Franklin, 2008;Koliev et al., in press).
Third, regime type might be a significant determinant of INGO shaming. For example, Ron et al. (2005) find that Amnesty International targets democracies more frequently than autocracies, while Murdie and Urpelainen (2015) suggest that environmental INGOs target autocracies more than democracies (Murdie and Urpelainen, 2015). Fourth, existing research provides reason to suspect that the ITUC might target US allies more frequently than other states in hope of triggering the US government to threaten them with hard sanctions (Hendrix and Wong, 2014). This prompts the inclusion of the two variables US military aid and US alignment that capture countries' ties with the United States. Fifth, three context-sensitive variables are included. The ITUC might be biased in favor of left-leaning states that embrace labor rights, rhetorically or otherwise, justifying the inclusion of the variable Left. The number of ratified conventions may also influence the shaming decisions of the ITUC as well as the number of labor-oriented NGOs in a country.
To account for temporal dependence, all models include the time trend variable and a lagged dependent variable. Theoretically, previous ITUC shaming may also tell us whether the ITUC keeps its targets under pressure or lets them "off the hook." Since the ITUC cannot observe the current year's actions by the ILO and governments, the two variables are lagged one year, while the information on the union-criticism is accessible to the ITUC before it makes shaming decisions that year. All control variables in the models are lagged one year. For summary statistics and information on the measurement and sources of variables, see Table A in Online Appendix.

Model and Analysis
Given the nature of the dependent variables, there is a need for an appropriate model that can be used for count data. The mean value of the main dependent variable, ITUC shaming, is 0.48, while the variance of 1.04 is twice as big. For the ITUC shaming through press releases and resolutions, the mean value is 0.22 and the variance is thrice as big: 0.66. These values indicate that the negative binomial regression is better suited than the Poisson model since the data are over-dispersed. 13 The Vuong tests show insignificant z-statistics (0.95), suggesting standard negative binomial regression over zero-inflated models.
The results of negative binomial regression, with standardized coefficients, are presented in Figure 3(a) and (b). 14 Figure 3(a) presents the results of the determinants of ITUC shaming within the ILO, while Figure 3(b) shows the results for ITUC shaming via press releases and resolutions. Model 1 in both figures, indicated by a circle, includes only the three main independent variables. Model 2, indicated by a square, includes the workers' rights and the economic variables. Model 3, denoted by a triangle, is a full model.
The analysis provides strong support for the bandwagoning hypothesis in both figures. First, the findings suggest that focal actors have a significant influence on the ITUC's shaming decisions within the ILO as well as outside of it. But hypothesis 1 is partly supported since governments' previous actions do not have a significant influence on ITUC shaming (only significant in Model 1). The impact of previous ILO criticism is significant and positive across all models, suggesting that the ITUC selects targets that previously have been criticized by the ILO (H3A). A one standard deviation increase in ILO criticism leads, on average, to a 0.21 increase in the log odds of being targeted by the ITUC.  Out of the three focal actors in the model, the union criticism has the strongest impact on ITUC shaming, lending support to hypothesis 3C. A country that receives criticism from its labor unions is likely to experience a 0.61 increase in the log odds of being targeted by the ITUC (Model 3 in Figure 3). Taken together, the analysis of the main independent variables implies that the ITUC adopts a bandwagoning approach in response to the ILO's and member-unions' criticism and that the members' preferences in relative terms have the strongest impact on the ITUC decisions.
The analysis also reveals other sources of influence on ITUC shaming. The ITUC is more likely to shame states with poor labor rights conditions than states with protected labor rights. This result is in line with previous findings that INGOs in general target rights-violating states. A one unit change in FDIs leads to 0.21 log odds of being shamed, suggesting that the ITUC selects states that are likely to be sensitive to international criticism. However, trading countries are less likely to be targeted by the ITUC. Although the effects are small, I speculate that the ITUC might prioritize FDI-dependent countries more because they often come with commitments of labor rights improvements, whereas trade is not coupled with such commitments. Furthermore, the results indicate that having a left-leaning government decreases the log odds of being shamed by 0.10. Each additional ratified ILO convention increases the log odds of ITUC shaming by 0.32. The analysis, however, provides no support for the influence of factors such as population size, economic development, or the level of democracy.
While the above models assess the influence of focal actors on ITUC shaming across different labor rights issues, there are reasons to examine how the ITUC reacts to focal actor preferences within each of the labor rights issues. For example, this might reveal whether the ITUC adopts different approaches depending on the issue area. To do this, a binary dependent variable is constructed, indicating whether a country has been targeted by the ITUC in relation to four different labor rights issues: forced labor, the right to organize and collective bargaining, discrimination, and child labor. Due to the binary nature of the dependent variables, I perform logistic regression analysis.
The results presented in Table 2 suggest that the ITUC adopts a bandwagoning approach across all four labor rights issues. It also reveals that the bandwagoning approach is more substantial within the area of rights to organize and collective bargaining, compared to the other issue areas. Note also that the workers' rights variable is only significant within the same issue area. One possible explanation for this is that the ILO has focused considerably more on countries that violate the rights to organize and collective bargaining, prompting the ITUC to act in line with the ILO's priorities . Moreover, according to interviews with labor union representatives, the rights to organize and bargain collectively are considered more central for labor unions' existence than any other core labor rights issue. Previous government criticism is still not significant in most models except for Model 3 on discrimination, where there is some evidence that the ITUC adopts a substitution approach. The insignificant results might reflect the highly politicized government shaming, where shaming is used as a political tool to punish adversaries and reward friends, or, in addition, might reflect the low number of government criticism. In contrast to the negative binomial models in Figure 3, the Left variable is not significant, the Population size and Trade is only significant in the forced labor model (Model 1, Table 2). The presence of labor rights NGOs is only significant and negative in one of my models (Model 1, Table 2), which suggests that the ITUC adopts the substitution approach as the number of labor NGOs increases. In general, however, the findings are similar to Figure 3.  One implication of adopting the bandwagoning approach might be that INGOs, steered by their willingness to enhance their reputation and credibility in the eyes of IGOs and donors, are likely to target "low hanging fruit" cases. INGOs might, then, target "easy" cases in order to show their members that they can alter the behavior of governments and, hence, to be viewed as worthy of investments (Stroup and Wong, 2017: 66). For instance, the ITUC might adopt the bandwagoning strategy toward countries that respect labor rights and are more willing to comply with international labor conventions than countries with persistently poor records of labor rights. I test this preposition by interacting the ILO's and member-unions' criticism with countries' labor rights conditions. For this purpose, I dichotomize the CIRI workers' rights variable-where 0 represents countries with restricted labor rights protection and 1 represents countries with fully protected labor rights-and interact it with the ILO and union criticism variables. The average marginal effects of interaction terms are plotted in Figure 4. The findings provide strong support that the ITUC is likely to target countries with poor labor rights conditions in response to ILO and union criticism.

Robustness Checks
A series of robustness tests are performed. First, in Table C1 and C2 in the Online Appendix, I show the results from zero-inflated, standard Poisson and rare logistic regression models. These tests show that the results for the main independent variables are robust across different regression models. The impact of the workers' rights variable is robust in Poisson and rare logistic models but not in zero-inflated models. One viable explanation for this is that ITUC is mostly influenced by focal actors' reactions rather than formal worker rights protection in the country. However, when looking at each labor issue separately, it is clear that labor protection rights are only salient within the rights to organize and bargain, which as mentioned above, constitutes a vital issue for labor movement and prioritized by the ILO. Second, the results of negative binomial models remain robust with year fixed effects, as well as when I restrict the analysis to the period 1998-2011, omitting the years with a low frequency of ITUC shaming (Table E in Online  Appendix). 15 Third, I estimate the results in Table 2 with rare logistic regression models and obtain similar results (Table G in Online Appendix). Fourth, there are reasons to suspect that ILO criticism might be triggered by previous ITUC shaming. To ensure that it is not previous ITUC shaming that triggers ILO criticism, all cases where ITUC shaming occurred a year before the ILO criticism were omitted in the model (about 90 cases were identified). The results for all issue areas except for child labor-where the ILO criticism variable is not significant-showed no substantial differences (see Table H in the Online Appendix; see also Table I in Online Appendix with restricted time period models). Finally, I utilize Anderson-Hsiao and Arellano-Bond estimations to control whether inclusion of an LDV introduces bias. The estimations using these two approaches indicate that the results are robust (see Table F in Online Appendix).

Conclusion
This article asked if and how focal actors influence the shaming decisions of INGOs. Drawing on previous studies, the article argued that IGOs, governments, and members are likely to influence INGOs' targeting decisions. Moreover, it suggested that INGOs can adopt either a bandwagoning or a substitution approach in response to these actors' preferences.
The findings suggest that the ILO's and members'-but not governments'-targeting decisions influence ITUC shaming. Moreover, the analysis provides strong evidence that the ITUC adopts a bandwagoning approach in response to the ILO's and member-unions' actions. The member-unions' criticism has the strongest impact on ITUC shaming, suggesting that INGOs value their members'-or donors'-preferences more than those of other focal actors when selecting countries for shaming. The bandwagoning approach does not lead to the selection of "easy" cases. Rather, through the bandwagoning approach, the ITUC generally targets countries with worse labor rights conditions.
The results have substantial implications for our understanding of INGOs and shaming strategies. More broadly, the article contributes to the research program on the role of INGOs in international relations. It invites researchers to reconsider INGOs' status as "shapers" and to adopt a view where INGOs' actions are also shaped by their focal actors (Heiss and Johnson, 2016: 533). I see a value in asking questions about how INGOs' access to and interaction with focal actors like IOs also might shape their campaigns, strategic considerations, and selection of targets. Indeed, the article provides reasons for research extensions, involving broadly formulated questions of whether, when, and why IOs may permeate INGOs' advocacy strategies (Johnson, 2014;Vabulas, 2013). No less important, to enhance our understanding of INGOs and their strategies, I also see a value in examining not only if but also which focal actors succeed in influencing INGOs and which fail to do so and why.
The previous ideas of what influences INGO shaming-such as rights violations, population size, and ties with powerful countries-are partly confirmed but failed to provide us with comprehensive answers regarding how INGOs select their targets. Using new data, this study suggests that INGO shaming is significantly influenced by the preferences of members and IGOs, which has been unacknowledged in previous empirical studies. If this is the case for other INGOs, which this study suggests, researchers should account for these sources of influence when assessing the impact of INGO shaming on states' compliance with international agreements. Arguably, resolutions of the UN's Human Rights Council might influence the shaming decisions of INGOs like Amnesty International. In addition-though the ITUC's shaming decisions within the ILO resembled its decisions outside of it-a case can be constructed where INGOs adopt different approaches through their resolutions and reports. Future studies could show when and why this is more likely, for example, by paying close attention to the characteristics of the audience and to the internal as well as interactive factors of INGOs' environments.
If the bandwagoning approach is favored because it is perceived as most efficient, as this article suggests, then future research should investigate to what extent bandwagoning is effective as a strategy in relation to the substitution approach. Caution should be raised, however, that if some IGOs' shaming decisions are politically motivated , there might be reasons to suspect that the bandwagoning approach introduces political bias into INGOs' decisions. For the same reasons, donor preferences might also lead INGOs to target states that might not be the worst or most frequent violators. However, as this study shows, the influence of members or donors does not always imply that the shaming decisions of INGOs are biased, that is, motivated by factors other than actual violations. Acknowledging the influence of these focal actors is the first step toward a research agenda that can unpack the shaming decisions of INGOs in this broader, more interactive political context.
To be clear, while I assert that the theoretical logic in this article could be applied to other INGOs and issue areas, there are also reasons to expect empirical variation across different INGOs. Previous studies point to organizational goals, membership base, funding media reliance, domestic institutions, cultural affinities, and networks as important factors in explaining NGOs' strategies (Bob, 2005;Dellmuth and Tallberg, 2017;Heiss and Johnson, 2016). Our results, at the very least, suggest that the findings of this article are generalizable to INGOs that share the ITUC's characteristics. I cannot entirely rule out, however, that specific institutional designs of focal IGOs like the ILO, as well as members of INGOs, might generate different strategies to target violators. Future research should examine more thoroughly how these factors influence INGOs' targeting strategies. As regards issue area, the article shows that the bandwagoning approach is consistent over four different issues, including broader human rights issues, such as forced labor, child labor, and discrimination-and narrower labor rights, such as unions' right to organize and bargain collectively.

Supplementary information
Additional supplementary information may be found with the online version of this article. Content Table A1. Labels, Measurements and Sources. Table A2. Summary Statistics. Table B