Preferences or blocs? Voting in the United Nations Human Rights Council

Abstract

After four years in operation the United Nations Human Rights Council (UNHRC) is subject to criticism, and various scholars and practitioners alike present and discuss reform proposals. In the present paper we study systematically the controversial decisions in the UNHRC. We find that controversial proposals are introduced by countries with a blemished human rights record, and that in the votes on these proposals the council members belonging to the European Union (EU) vote very distinctly from the remaining members and have preferences quite different from those member states that violate human rights. Extending an empirical approach frequently used in parliamentary research we can also show that in votes in the UNHRC preferences of member states dominate over their membership to particular blocs. As controversial votes also heavily polarize the UNHRC we argue that the problems faced by the UNHRC’s predecessor, namely the Commission on Human Rights, have reappeared.

1 Introduction

From its very inception in 2006 the United Nations Human Rights Council (UNHRC) attracted its fair share of criticism. The assessment by former United States Ambassador to the United Nations (UN) John Bolton was clearly the most colorful: “We want a butterfly. We’re not going to put lipstick on a caterpillar and call it a success”.¹ After operating for four years as the successor of the even more vigorously decried Commission on Human Rights (UNCHR) the United Nation considers reforming this new body of the UN.² So far, however, no systematic studies exist to our knowledge elucidating how the UNHRC operates during its regular sessions (three per year).³

In the present paper we wish to contribute to filling this gap by offering a systematic study of the controversial votes that have taken place in the first thirteen sessions between 2006 and 2010 in the UNHRC. While a large majority of all resolutions are adopted without opposition or only abstentions, 70 votes between 2006 and 2010, however, revealed clear divergences of opinion with some countries opposing the proposal. In analyzing these votes we draw on the one hand on the literature on voting in international assemblies and on the other on recent work on parliaments in general. While the former literature has strongly emphasized bloc-voting, i.e., countries belonging to particular groups voting together, the latter suggests that when analyzing such groups (e.g., parties) one cannot neglect the role preferences play. Drawing on an empirical approach having become prevalent in parliamentary research we can show that the two blocs commonly assumed to be important in the UNHRC influence less strongly voting in this assembly than variables assumed to relate with the underlying preferences of states. Extending this analysis we can also show that countries with blemished human rights records are most frequently the authors of controversial proposals. The latter often pass against the council members of the European Union (EU), who in these votes adopt very distinct voting patterns. Analyzing more in detail how the identity of the proposer affects voting behavior we find clear patterns that these divisive votes play a considerable role in polarizing the UNHRC.

The remainder of the paper is organized as follows. In the next section we briefly present the development of human rights in the context of the United Nations as it led to the creation of the UNHRC in 2006. Section 3 discusses the insights scholars have gained by analyzing the voting behavior in other UN bodies, most notably the General Assembly (UNGA), which provides part of the theoretical backdrop for our study.⁴ Based on this literature we present three broad hypotheses on the way in which preferences and bloc membership, in conjunction with the proposer’s identity, should affect voting behavior. In Section 4 we present our empirical strategy which relies on hierarchical item-response theory models that allow for a direct estimation of the relevant parameters of interest to us. In Section 5 we present our empirical results before concluding in Section 6.

2 Human rights and the human rights council

Following World War II human rights have become a mainstay in international relations. The devastation by the War with the numerous atrocities and the continuous violations of international rules and norms called for a new governing body after the failure of the League of Nations. The creation of the UN in 1945 was the first step toward the increased importance of human rights. Soon thereafter, in 1946, the UN created a sub-body which would deal with the promotion of human rights, namely the Commission on Human Rights (UNCHR) (for an overview, see Tolley 1987). In 1948 the Universal Declaration of Human Rights (UDHR) was adopted with 48 yes and 8 abstentions in the UN General Assembly (UNGA) (Simmons 2009, 41). The UDHR applies to all people, but on its own is not a binding legal document in international law (Simmons 2009, 42). To ensure that the provisions set forth in the UDHR would be applied, the UNGA worked out the two additional International Covenants on Human Rights (1966) and the Optional Protocol, which are legally binding documents, and as a consequence can be enforced in courts (Simmons 2009, 46ff). Nonetheless, it is the UDHR that is widely cited when it comes to human rights, and it has become one of the foundations of our understanding of modern human rights. On the basis of these legal documents the international community set out countless conventions in various fields of human rights, sometimes in the framework of international conferences and initiatives, sometimes within the UNCHR (e.g., Carey et al. 2010).

Regardless of the UNCHR’s success with introducing the Universal Declaration of Human Rights of 1948, the UNCHR became over the decades a forum that was too heavily exposed to political influence (e.g., Edwards et al. 2008). It was referred to as the “shame of the UN”⁵ by outsiders as well as by insiders such as the former Secretary General Kofi Annan. The latter stated in his special report entitled “In Larger Freedom” that the Commission was “undermined by the politicization of its sessions and the selectivity of its work”⁶ and that “the Commission’s capacity to perform its tasks has been increasingly undermined by its declining credibility and professionalism.”⁷ The report also emphasizes the possible reputational damage of the Commission to the UN as a whole since “the credibility of the Human Rights Commission has eroded to the point that it has become a blot on the reputation of the larger institution.”⁸ This report would later guide the reforms of the Commission. Others, such as the US ambassador to the Commission at the time, called the election of countries with a poor human rights record just “absurd.”⁹ Secretary General Kofi Annan even stated that, “[s]tates have sought membership of the Commission not to strengthen human rights but to protect themselves against criticism or to criticize others.”¹⁰

Edwards et al. (2008, 394) find in their study that members could have had several reasons for joining the Commission, namely to “influence the agenda to blame other states..., in an act of self-defense, enabling them to insulate themselves from investigation... and to deflect attention from themselves.” In addition, developing countries tried to shift the focus from civil and political toward economic rights. As history has shown, this is not uncommon within the UN framework as the interpretation and focus on human rights can be very diverse. Edwards et al. (2008) even find that countries with a poor human rights record had higher chances of getting elected to the Commission than those with high human rights standards.¹¹

All this criticism was not in vein and following the publication of the UN Special Report the General Assembly decided¹² to abolish the Commission and replace it by the UNHRC in March 2006. The newly created Council reports directly to the General Assembly and therefore has a higher status than the Commission. It defines itself as “an intergovernmental body within the UN system made up of 47 States responsible for strengthening the promotion and protection of human rights around the globe.”¹³ These 47 Member States are elected for a three year term by the UNGA in accordance with the members’ “contribution to the promotion and protection of human rights and their voluntary pledges and commitments made thereto” and “shall uphold the highest standards in the promotion and protection of human rights.”¹⁴ According to Cox (2010, 89ff), however, especially when considering the much more far-reaching reform proposal by the United States, the changes introduced regarding the membership are rather negligible. While members of the UNCHR were elected by the ECOSOC committee of the UNGA, the regional groups still play a central role in the election process.¹⁵ Nevertheless, Cox (2010, 108) notes that the share of countries that are considered as not free by Freedom House slightly dropped among the members in the first sessions of the UNHRC compared to the last sessions of the UNCHR.¹⁶ More substantial changes concerned the Council’s working mechanisms, which include the following:¹⁷

• 1. The Universal Periodic Review, “which serves to assess the human rights situations in all United Nations Member States.”

• 2. An Advisory Committee “which serves as the Council’s ‘think tank’ providing it with expertise and advice on thematic human rights issues and the Complaint Procedure which allows individuals and organizations to bring human rights violations to the attention of the Council.”

• 3. “The Human Rights Council also works with the UN Special Procedures established by the former Commission on Human Rights and now assumed by the Council.”

3 Voting in international organizations and human rights

While voting in the UNHRC has to our knowledge not been studied systematically (see for some discussion and partial exceptions Besant & Malo 2009; Cox 2010; Seligman 2011; McMahon 2012), some scholars have looked under various perspectives at voting in its predecessor, i.e., the Human Rights Commission (e.g., Smith 2006; Lebovic and Voeten 2006 see below). Much more detailed studies on voting in international bodies focused on the UNGA. Early studies relied heavily on the notion of “blocs,” which were assumed to emerge in the voting record. Studying possible bloc building appeared to be important as it can suggest the direction which an assembly may take and suggest ideas about voluntary or involuntary cooperation and coordination among countries on specific topics. The notion of blocs proved, however, difficult to define. Hovet (1960) offers a comprehensive discussion of this notion, while Lijphart (1963) criticizes the various attempts to rely on geographical groups as defined in the Charter itself, geographical proximity, closeness in other respects etc. in earlier work (e.g., Ball 1951; Furey 1954; Riggs 1958; Ogley 1961; Nicholas 1962).

While differing, these various early definitions have nonetheless some blocs in common, like the Soviet Union with its satellite states (also called the Soviet-bloc), the African bloc (sometimes mixed with the Arabs), the Arabs, the Latin American bloc (Lijphart 1963), and later also the Western countries. Newer studies covering longer time periods (e.g., Holloway 1990), argue that the developing countries or also the “Muslims”¹⁸ appeared as new blocs, but also the “imperialists” or the “neutrals” (Newcombe et al. 1970), depending on the factors analyzed. Holloway (1990) shows, however, that depending on the time period studied the blocs are not located constantly in the same place spatially speaking. Certain countries moved around in this space, so the “shapes” of the blocs have also been changing. Apart from studies focusing on how the different blocs are taking form, others explore the internal life of the blocs themselves. For example some authors study whether there are blocs within the blocs (for the African “bloc inside the bloc” see Meyers 1966).¹⁹

With the end of the Cold War, the long established Warsaw Pact group (Soviet-bloc), often considered as the most cohesive bloc, disappeared at least in theory. In the 1990s the European Union (EU) appeared as an important new bloc (analyzed by, among others, Young and Rees 2005; Smith 2006; Luif and Radeva 2007).

The way in which these blocs are identified, however, is largely inductive. More specifically, authors most often employ some sort of a scaling approach to represent countries in a low-dimensional policy space based on their voting records. Clusters of countries are then identified on the basis of notions of blocs favored by the authors. Proceeding this way, however, allows not to distinguish two possible ways in which such clusters might emerge.²⁰ On the one hand it might be that a group of countries, for instance the member-states of the European Union (see below) wish to coordinate their stance on human rights issues. This should lead to a more cohesive and distinct voting pattern of these member states. On the other hand these same member states might vote more often together simply because they have similar preferences when it comes to resolutions on human rights. In what follows, drawing on an empirical strategy employed in parliamentary research, we propose a way to separate out these two elements, while also considering the role the proposers of controversial resolutions play in explaining voting patterns.

3.1 Voting in the UNHRC: Expectations

In the most comprehensive study of the UNGA, Voeten (2000, 191) argues that voting heavily clusters member states and thus classifies various groups of countries assumed to appear in UNGA voting. Similarly, two recent studies focusing more closely on human rights issues in the UN implicitly or explicitly rely on the idea of “blocs.” Smith (2006) assesses directly how one specific group of countries, namely the member states of the European Union (EU), behaves in votes in the third committee of the UNGA (dedicated to human rights issues) and the UNCHR. As Smith (2006) argues the EU puts considerable emphasis on appearing united in human rights issues on the international scene (see also Wouters et al. 2008). Despite this emphasis, she finds considerable dissent in the votes she analyzes. Lebovic and Voeten (2006) offer a more comprehensive study of voting in the UNCHR, finding that reputation effects appear to have played an increasing role over the lifetime of this commission. These authors subsequently show that condemnation of particular countries by the UNCHR affect multilateral (though not bilateral) aid flows (Lebovic and Voeten 2009).

Boockmann and Dreher (2011) study voting on human rights issues in the UNGA and argue that the average level of human rights violations in a country’s “peer group” dominates the individual human rights record of a country in affecting the latter’s voting decisions. While this seems an interesting empirical result, given that the “peer groups” are defined by the World Bank classification it is hard to make sense out of this finding in terms of a causal mechanism.

The two only systematic studies focusing on voting in the UNHRC that we are aware of focus on specific aspects. Cox (2010, 111ff) suggests that resolutions targeting particular countries (most notably Israel) are as prevalent in the new body as they were in its predecessor. This assessment is confirmed by Seligman (2011) who also proposes to evaluate three broad hypotheses by comparing the voting behavior of groups identified by their degree of political freedoms as measured by Freedom House in resolutions voted upon in the first sessions of the UNHRC. He finds very distinct voting patterns especially for resolutions targeting Israel.²¹

These studies suggest, as does the broader literature on voting in international assemblies, that at least two elements should influence voting in a body like the UNHRC. On the one hand, studies like those by Voeten (2000), Smith (2006), Wouters et al. (2008) and Boockmann and Dreher (2011) suggest that belonging to particular groups of countries or blocs (defined in whatever way) should influence the way in which UNHRC member states vote. On the other hand, authors like Lebovic and Voeten (2009) and Seligman (2011) suggest that factors more related to the preferences of states should affect the voting behavior. Consequently, we propose two very broad hypotheses that rely on the arguments discussed above:

• H1:  Belonging to a group of countries taking clear positions on human rights affects the voting behavior of its member-states (in a more human rights respecting way, respectively in the opposite direction).

• H2:  A country’s preferences in the area of human rights records affect its voting behavior.

To empirically evaluate these two hypotheses we need on the one hand to identify the groups of countries taking stances on human rights issues, and on the other identify variables that should be related to the countries’ preferences in the area of human rights. Regarding the first issue, we argue that the EU and the Organization of the Islamic Conference (OIC, founded in 1969)²² are the blocs that take most often the clearest stances. Regarding the EU, the studies by Smith (2006) and Wouters et al. (2008) amongst many others underline the importance the EU attributes to human rights, especially as they are discussed in international bodies. Thus, Macaj and Koops (2012, e.g., 72) discuss in detail how the EU’s Common Foreign and Security Policy has led to an important coordination effort in the area of human rights. Regarding the OIC, Besant and Malo (2009, 145) note the importance of this group (see also Lukács 2010, 2011),²³ while Steinberg (2012, 374) goes as far as stating that OIC plays a dominating role in the UNHRC (for a similar point, see Smith 2010, 225). As Macaj and Koops (2012, 81) argue, this is related to the growing importance of the notion of “defamation of religion.” In this debate the OIC plays clearly a dominant role. Thus, if any blocs should constrain the voting behavior of UNHRC member states the EU and the OIC should certainly be among them.

In order to separate out the effect of preferences from the one exerted by bloc-membership we also need to identify variables likely to be related to the underlying preferences of states in the area of human rights. Based on the literature, we argue that three variables can serve as proxies for these preferences. First, as several studies have argued, the respect for human rights can be assumed to be akin to revealed preferences by states. Consequently we will rely on Cingranelli and Richards’ (2010) data and for robustness tests on the “Political Terror Scale” (Wood and Gibney 2010). Second, as numerous studies have shown (e.g., Park 1987; Henderson 1991; Poe & Tate 1994; Hathaway 2002; Neumayer 2005; Vreeland 2008) democracies have on average a better respect of human rights. Consequently, we assume that being a democracy will also relate to preferences important in votes on human rights issues.²⁴ Finally, we also consider that ratifying treaties in the area of human rights should be related to preferences in this area. Several scholars have demonstrated that signing and ratifying treaties and conventions in this area is related to respect for human rights but not necessarily in a straightforward way (see Hathaway 2007; Vreeland 2008; Simmons 2009; Hollyer & Rosendorff 2011).

While the primary focus of this paper is to assess what affects the voting patterns (and/or revealed ideal-points), we also wish to take into account how the authorship of a proposal affects these patterns. We do this for two reasons. First of all, several studies have highlighted that agenda-setting is closely related to voting behavior (see for instance Carrubba et al. 2008). Second, and relatedly, not all decisions in the UNHRC are reached in votes. A series of resolutions and other decisions are adopted without a vote. Consequently, the observed votes and the members’ voting behavior in the UNHRC can be considered as roll call votes, as at least one member wishes to make the voting result public. As Hug (2010) shows, such votes often differ regarding several characteristics from secret votes. Carrubba et al. (2008) derive from a theoretical model, however, the insight that the content of the proposal plays in this context a central role.

Consequently, in our analyses we also control for the authorship of the proposal to account at least in part for the possible selection biases identified by Hug (2010).²⁵ As will become apparent below, the authorship of controversial proposals is heavily concentrated among a small set of countries. We expect that resolutions proposed by countries with rather extreme views in the area of human rights will polarize more strongly the UNHRC member states. Hence we will also evaluate whether countries frequently submitting controversial resolutions lead to a polarization of voting in the UNHCR:

• H3: Resolutions submitted by countries with less respect for human rights polarize the voting behavior in the UNHCR.

As we show below, four sets of countries have submitted the largest share of controversial votes, namely Cuba, Egypt, Pakistan and member states of the EU. Consequently, we will test this third hypothesis by assessing how polarizing resolutions proposed by these four groups compare to all the other resolutions.

4 Data and model

In this section we briefly describe the data we employ before discussing in more detail the empirical model used to test our hypotheses.

4.1 Data

The data employed for this study consists of the recorded votes of the UNHRC’s first thirteen ordinary sessions as of April 2010.²⁶ Overall the UNHRC reached 271 decisions on resolutions during the first thirteen sessions.²⁷ The large part of resolutions were, however, adopted without votes. As for the analyses we will carry out below we will rely on votes and, in an even more constraining fashion on non-unanimous votes, it is important to understand what characterizes the resolutions that led to a vote. Table 1 reports for those 198 resolutions that were drafted and submitted by a single actor whether or not a vote occurred, its authorship and in how many sessions the proposer was member of the UNHRC.²⁸

Table 1 Drafters of resolutions adopted with and without votes

Table 1 first shows that only four countries drafted and submitted ten resolutions or more, namely Cuba, Egypt, France and Pakistan. Three of these countries had, according to Cingranelli and Richards’ (2010) data, a rather blemished human rights record in 2008.²⁹ Second, as about one fourth of all resolutions are adopted in votes and mostly with controversy, it is striking that the resolutions introduced by three of these countries (the exception is France) are more frequently submitted to a vote than the average. This is also the case for some other countries, but given the small numbers involved, these might be outliers. Obviously, the number of sessions in which Cuba (13), Egypt (13), France (11) and Pakistan (13) were members might also explain the high numbers of resolutions introduced by these countries.³⁰ Several other countries were, however, member of the UNHRC for equal number of sessions and did not introduce as many resolutions, and certainly not as many controversial ones.

Looking more closely at Table 1 one notes that following the three top-ranked countries (Pakistan with 23 resolutions, Cuba with 20 and Egypt with seven) thirteen different countries proposed the remaining ones. Among these, seven stem from EU-member countries.³¹ As we will analyze how EU member countries vote in controversial votes, it is useful to assess how the EU fares in these 70 votes as a function of who proposed the resolution. As Table 2 nicely shows, whenever Cuba and to a lesser degree Pakistan or Egypt propose a resolution, the chances that the EU-member countries will be on the winning side are small or even minute.³²

Table 2 Proposer’s identity and the EU’s success

Thus, it is obvious that simply focusing on the decisions reached in votes in the UNHRC may introduce a bias, as they do not reflect the full business of the UNHRC (see for a more detailed analysis along these lines on the UNGA Hug 2012). Information on the individual voting records is, however, only available for these votes on resolutions³³ plus some votes on amendments and motions (eight overall). This potential bias should be indirectly addressed when evaluating hypothesis 3, namely when considering how the authorship of the proposal voted upon affects the polarization of UNHRC members.

For our explanatory variables we rely on standard sources. To measure the respect for human rights we rely on Cingranelli and Richards’ (2010) torture scale in 2008 as a general measure. As measure for democracy we rely on Cheibub et al. (2010) dichotomous coding for 2008.³⁴ Finally, to have an assessment of the ratification behavior of UNHRC member states of human rights conventions we rely on Hathaway’s (2007) data. The latter author provides for 2003 (and the years before) the ratification status of the major human rights convention.³⁵ We used the number of these conventions ratified in 2003 (to allow for a sufficient lag) as measure as it takes some time for treaty ratifications to affect behavior.

In addition to these variables allowing us to evaluate our three hypotheses we also consider two control variables which might influence voting in the UNHRC.³⁶ On the one hand we control for GDP per capita measured in 1995 PPP by relying on the Penn World Tables V7.0 (Heston et al. 2011). On the other we also use as control variable the Net Development Assistance and Aid (in constant USD) from the World Development Indicators (http://data.worldbank.org/data-catalog).

4.2 Model

Analyses of voting decisions in assemblies carried out to understand what underlying conflict lines can be detected have made considerable advances, also when it comes to studies on international bodies. Early work, for instance by Alker (1964) (see also Alker & Russett 1965) largely employed factor analytic models to determine the underlying conflict lines. As scholars mostly dealing with the US congress noted, such factor analytic models lack, however, a solid theoretical underpinning to allow for generating information on the relevant conflict lines. Based on this critique Poole and Rosenthal (1985) developed a theoretically informed estimator based on a spatial model of decision making.³⁷ This estimation procedure, called NOMINATE, was used among others by Voeten (2000) in his study on voting in the UNGA. More recently scholars proposed estimating the underlying conflict lines by relying on the so-called item-response theory (IRT) model stemming from the school testing literature (e.g., Fox 2010). Clinton et al. (2004) propose this approach to estimate the ideal-points of legislators (see also Jackman 2004; Gelman & Hill 2006).³⁸ All these models rely on parametric models with quite constraining assumptions underlying the estimator. Poole (2000) thus proposes a nonparametric technique to estimate ideal-points of legislators.³⁹ In some recent critical articles Spirling and McLean (2006, 2007) alert the reader that under certain circumstances these estimators may fail to yield meaningful estimates, namely if the assumed spatial model of voting is not appropriate, for instance if government and opposition vote against each other and the latter are dispersed on opposite sides on the ideological scale.

All these methods have been applied to voting in international bodies in order to deduce the preferences of member states based on their spatial location (for instance after a factor analysis). If states belonging to particular blocs appeared to be located close to each other “bloc voting” was induced. This inductive approach is unsatisfactory as “blocs” are inferred from similar voting patterns, even though they might also simply be due to similar preferences. In the approach we propose preferences are estimated based on voting patters but also linked to explanatory factors of these preferences. In addition, we also estimate how bloc-membership constrains the voting patterns of their members beyond the variables that we hypothesize affect preferences.

These external sources of information can easily be taken into account when estimating models, especially those based on the item-response theory as estimated in a Bayesian framework. For instance, Høyland and Hansen (2010) employ additional information on preferences to assess whether in the Council of Ministers of the European Union a push toward more consensus exists. Similarly, Malecki (2008) proposes, based on work by Fox and Glas (2001) (see also Fox 2007, 2010, 141–192), a hierarchical item-response theory model, where the ideal-points are “explained” in part by exogenous variables.⁴⁰ Gabel et al. (2008) use this approach to assess what elements affect voting behavior on the European parliament (EP), while Høyland (2010), drawing on a suggestion by Clinton et al. (2004), studies how party group pressure is exerted in legislative matters in the EP, compared to votes on non-legislative matters.⁴¹

In our study of the UNHRC we also rely on the classic “two-parameter” item-response theory (IRT) model. In this model the probability (π _ij ) of a yes-vote (y _ij ) by actor i on issue j is following Jackman (2009, 455)

$$\begin{array}{@{}rcl@{}} \pi_{ij}&=& Pr(y_{ij}|\theta_{i},\beta_{j},\alpha_j)\\ &=&F(\theta_{i}\beta_j-\alpha_j) \end{array} $$

(1)

where θ _i is in our context the revealed ideal-point, β _j the item discrimination of issue j and α _j the item difficulty of issue j. F being a cumulative density function either of a normal or logistic type. For identification purposes, θ _i is usually assumed to be normally distributed with μ _θ = 0 and $\sigma ^{2}_{\theta }=1$.

Fox and Glas (2001) (see also Fox 2007, 2010, 141–192) present the basic ideas how an IRT model may accommodate a hierarchical data structure. Based on this Malecki (2008) models θ as being dependent on some person specific characteristics. When addressing the issue what member countries of the UNHRC vote together we adopt such a hierarchical IRT model allowing the θs to vary systematically with characteristics of the member states (x _i ). More precisely we assume that

$$\begin{array}{@{}rcl@{}} \theta_{i} &=&x_{i}\beta_{\theta} + \epsilon_{\theta} \end{array} $$

(2)

In a second step we assume that the item discrimination parameter β _j varies with characteristics of the resolution voted upon, more specifically its authorship (x _j ).⁴² For this we assume that

$$\begin{array}{@{}rcl@{}} \beta_{j} &=&x_{j}\beta_{\beta} + \epsilon_{\beta} \end{array} $$

(3)

IRT models are most frequently estimated in a Bayesian framework, given the numerous parameters to estimate and the distributional assumptions required for identification. We proceed similarly in this framework and implemented these models in JAGS (Plummer 2010).⁴³

5 Empirical results

As only 76 votes occurred in the first thirteen sessions of the United Nations Human Rights Council, we rely on a rather small empirical basis. In addition, as noted above, we drop six unanimous votes from our analysis, as they are uninformative for the estimates we wish to provide.⁴⁴ This leaves us with 70 votes and 64 members who voted in part in these votes. Each of the 64 member states voted on at least 9 of these topics. We use this empirical base to evaluate our hypotheses, as discussed above, with a hierarchical IRT-model. This model estimates at the same time the revealed “ideal-points” (or preferences) for the member states, but also how the latter are influenced by various independent variables (among them bloc-membership and variables related to preferences). While voting in the UNHRC is ternary, namely comprising yes- and no-votes as well as abstentions, we focus our analysis only on the first two alternatives. Proceeding like this is justified, as the rules of procedure (http://ap.ohchr.org/documents/E/HRC/resolutions/A_HRC_RES_5_1.doc) clearly specify that with very few exceptions, abstentions are not relevant for the voting outcome, as decisions are reached by simple majority by the member states present and voting. Consequently, abstentions are likely to follow a different pattern and can hardly be conceived as some intermediary category (though see Voeten 2004; Boockmann & Dreher 2011).

As discussed above, our hypotheses relate to the effect of membership in two groups, on the one hand, and democracy, human rights records and ratification of human rights treaties on the other. Table 3 reports the second-level estimates of two IRT models in which we first only estimate the parameters in Eqs. 1 and 2 and then also integrate Eq. 3 in the second model.⁴⁵ The estimates (with their respective credible intervals) suggest that all our hypotheses point in the right direction. Considering hypothesis 1 related to the effect of belonging to a particular bloc, we find that the EU and the OIC pull their member countries in opposite directions from the remaining UNHRC member-states. The credible intervals for both of these effects fail to comprise the value of zero. Consequently, we have quite considerable evidence that belonging to one of these two blocs actually influences systematically the voting behavior of UNHRC members.

Table 3 Hierarchical IRT model: second-level estimates

Regarding our second hypothesis both models reported upon in Table 3 offer considerable support that preferences as measured by our three variables influence the voting in the UNHRC. Countries engaging often or frequently in torture vote systematically differently from those not engaging in torture. Democracies and countries having ratified more human rights treaties appear to vote in the opposite direction, but the credible intervals for these two variables include the value of zero. Regarding our two control variables we also fail to find any systematic effect. Receiving development aid and/or being a rich or poor country fails to affect the preferences that find expression in UNHRC voting. Thus it seems that the preferences emerging in this assembly are related mostly to human rights practices, lending support to our second hypothesis.

Our first model reported upon in Table 3, however, is clearly too simple, especially given our results showing that the proposers of resolutions affect the results from the EU’s perspective. To assess this and evaluate our third hypothesis we estimate a second hierarchical IRT-model, this time with the parameter for the item discrimination (β) varying as a function of the proposer’s identity, thus estimating the parameters in Eqs. 1, 2, and 3. For simplicity’s sake we categorize the proposers into four groups, namely the ones used in Table 2. As base category we employ the residual category of “remaining” proposers. Consequently the estimated parameters reported for the second model in Table 3 indicate whether resolutions proposed for instance by EU member countries, discriminate more or less strongly among council members than resolutions by the remaining countries.⁴⁶ Similarly the sign indicates whether the yes and no votes are on the same sides as in votes on resolutions proposed by the remaining countries or not.

The results reported for model 2 in Table 3 underline again the crucial role played by Cuba in controversial votes. Resolutions proposed by this country discriminate most strongly among council members and compared with those proposed by other countries and the credible intervals for this effect excludes the value of 0. The direction of the effect is similar for resolutions proposed by Egypt and Pakistan, though the credible intervals for these coefficients are large and comprise 0. The coefficient for the EU is positive, suggesting that resolutions introduced by EU member states are on average less discriminating than those of the remaining UNHRC members. The credible interval for this coefficient is large, however, and comprises 0, implying that we cannot distinguish this effect from a null-effect.

When considering the estimated coefficients for blocs and our preference variables we find largely the same results as above. Figure 1 illustrates the effects estimated for our second model graphically with their respective credible intervals based on the second model in Table 3.⁴⁷ The figure clearly provides support for our first hypothesis. The EU member states vote quite distinctly from their fellow UNHRC members, as do the members of the OIC though in another direction. Consequently, our results lendsupport to studies on the EU, like Young and Rees (2005), Smith (2006) and Wouters et al. (2008), who document an important coordination effort among the member states. The same also appears for the OIC, for which Besant and Malo (2009, 145) also highlight the considerable coordination work. The effects are even stronger for our main preference variable, namely whether countries engage in torture or not.

Fig. 1

Hierarchical IRT Model: mean estimates of second-level coefficients for preferences and credible intervals

To have more confidence in our results we reestimated our two models by replacing our two central variables linked to preferences by other measures.⁴⁸ Instead of using Cingranelli and Richards’ (2010) torture scale we relied on the “Political Terror Scale” (Wood and Gibney 2010). We also replaced Cheibub et al.’s (2010) dichotomous measure of democracy with the 21 point scale provided by the Polity indicator (Marshall et al. 2002).⁴⁹ The results depicted in Table 4 clearly demonstrate the robustness of our results regarding preferences (i.e., hypothesis 2). We still find a considerable effect for our main variable related to preferences, namely the “Political Terror Scale.” These effects are particularly clear when we control for the authorship of the resolution voted upon (model 2). Similarly, in this second model we also find that the credible interval for our democracy measure excludes the value of 0, suggesting that there is a systematic effect of this variable on the voting behavior of UNHRC members. The effect of this variable, however, is rather small (see also Fig. 2).

Table 4 Hierarchical IRT model: second-level estimates, robustness check

Fig. 2

Hierarchical IRT Model: mean estimates of second-level coefficients for preferences and credible intervals (robustness test)

While thus the robustness check strengthens our confidence in hypothesis 2, the results depicted in Table 4 suggest that hypothesis 1 is on much less solid ground. While in model 1 we still find a considerable effect for OIC bloc-membership, the credible interval for the EU fails to exclude the value of 0. In model 2 this is even the case for both blocs, suggesting that bloc-membership does not systematically affect voting behavior in the UNHRC. Regarding the control variables we still fail to find a systematic effect, both in models 1 and 2. Our results regarding the influence of Cuba as a proposer of resolutions survives this robustness check as well (model 2).

We depict the effects related to hypotheses 1 and 2 again graphically in Fig. 2. This figure shows that support for hypothesis 2, namely that variables related to preferences affect voting in the UNHRC, is quite robust and solid. On the other hand, it appears that the effect of bloc-membership is not as robust, as it depends on whether or not we control for the authorship of resolutions, and on what preference variables we employ. Support for hypothesis 1, taking into account Fig. 2, is thus weaker. Quite to the contrary, the two sets of analyses lend considerable support to hypothesis 3 related to the effect of authorship of resolutions those resolutions authored by Cuba appear to discriminate more strongly among UNHRC members.

6 Conclusion

Our systematic analysis of controversial votes in the UNHRC suggests that voting behavior is largely affected by what we might consider the preferences of member states. Despite controlling for the major human rights relevant “peer groups” we find that their effects pale compared to those of variables related to preferences. Consequently, both the human rights record of a country and to a lesser extent its level of democracy prove important factors in explaining the preferences related to voting in the UNHRC. Countries with poor human rights records vote systematically differently from those that do not engage in torture.

On the other hand we find only mixed evidence that blocs are relevant in this new UN institution. Whether we find evidence in support of bloc-voting depends on the measures we use for our preference variables and whether we control for the authorship of the resolution voted upon. Here we find that resolutions proposed by Cuba strongly polarize the member states of the UNHRC in their voting.

While our analyses are still based on a rather limited empirical basis, we can draw two broad conclusions. First of all, from a methodological perspective our analysis demonstrates that work on international assemblies should employ approaches that allow to distinguish between the effects of blocs and preferences when looking at voting behavior. When doing so, accounting for the authorship of the proposals also proves important. Second, from a substantive point of view our analyses suggest that the criticisms leveled against the UNHRC’s predecessor have not lost all their relevance. When assessing the effect of a resolution’s sponsor we find that especially one set of resolutions can be distinguished. Those proposed by Cuba heavily discriminate among members of the UNHRC and thus find strong reflection in the voting patterns. They differ from proposals submitted by other countries, especially compared to those authored by EU member countries. The latter countries also systematically lose in the controversial votes initiated by Cuba, and to a lesser extent by Egypt and Pakistan.

Hence our results suggest that not all defects of the UNCHR have been mended with the UNHRC.⁵⁰ Controversy in this assembly is largely introduced by countries with blemished human rights records. As a consequence, many of the conclusions presented on the UNHRC’s predecessor still seem to apply. Hence, despite a heavy dose of “lipstick” the animal has not lost all its “caterpillar”-features.