De Facto Judicial Independence and Physical Integrity Rights

Economists, political scientists, and legal scholars have argued that independent judiciaries have an important role to play in promoting economic development and protecting property rights. We argue that judicial independence can also have a positive impact on the protection of human rights. To assess the human rights impact of a de facto independent judiciary, we also argue that scholars must account for the potential of endogeneity between judicial independence and protection of human rights. We examine whether greater de facto independence improves government respect for citizens’ physical integrity rights, using a comprehensive dataset of 193 countries from 1981 to 2010. Employing an instrumental variables approach to control for endogeneity, we find strong support for the argument that greater levels of de facto judicial independence improve government respect for physical integrity rights. These findings are robust to changes in measurement, estimation techniques, and model specification. Failing to account for endogeneity will tend to overemphasize the ability of completely independent courts to improve government respect for physical integrity rights.

Scholars of economics, international relations, and the courts argue that an independent judiciary can be important for economic development, the protection of property rights, and even the promotion of good governance (North and Weingast 1989;Barro 1991;Henisz 2000;Feld and Voigt 2003). But is judicial independence also important for the protection of human rights? Most of the existing research studies the effect of de jure, or formal, judicial independence upon the degree of human rights protection; in contrast, very little research has examined how de facto judicial independence affects government respect for physical integrity rights. 1 Our manuscript contributes to this important but understudied field. As it is generally understood, de facto independence connotes a willingness of courts to rule against the government as well as a reasonable degree of effectiveness in having those decisions respected (Ríos-Figueroa and Staton forthcoming). This concept looks beyond what is written in constitutions and takes into consideration the extent to which judges actually decide in ways consistent with their own "personal attitudes, values, and conceptions of the judicial role" (Becker 1970: 144). But does de facto independence actually translate into better human rights outcomes, or are proponents of courts as inherent protectors of rights deluded, as Shapiro (2008) suggests?
In investigating these questions, this article makes both theoretical and empirical contributions. First, we present a new explanatory framework that details how de facto independence affects government respect for physical integrity rights, drawing upon insights from both the human rights and judicial politics literatures. We argue along with others (McCann 1994;Keck 2009) that courts can play a role in evolving social struggles, through both direct means (e.g., ruling against the government) as well as indirect means (e.g., deterring governments from violating human rights). While courts often act strategically in their relations with other branches of government, we argue that courts free from political interference are more likely to be effective in protecting the core bodily integrity rights of citizens from abuse by their own national governments in comparison to courts lacking de facto independence. Second, as part of our argument, we contend that researchers must explicitly account for the possibility that some endogenous process affects both the presence of de facto independence as well as the level of government protection of physical integrity rights. We depart from previous work that does not address this concern and develop an instrumental-variables approach to test our argument about the impact of de facto judicial independence on government respect for physical integrity rights.
Third, in testing our expectations, we employ a new measure of de facto independence (Keith 2011) and utilize a more extensive dataset than any previous study of de facto independence (Howard and Carey 2004;Powell and Staton 2009;Keith 2011;Volcansek and Lockhart 2012): 193 countries from 1981 to 2010. While there are certainly trade-offs to studying this process in a large-N fashion (such as the lack of case-level data for large, cross-sectional time-series studies), we join other judicial scholars in arguing that taking a comparative, global approach to the study of courts offers us valuable insights about both institutions and the protection of individual rights (Epp 1998;Keith 2011;Uribarri et al. 2011;Volcansek and Lockhart 2012). With the "judicialization of politics" emerging as a significant trend across the world in the late twentieth and early twenty-first centuries (Tate and Vallinder 1995: 5), an increasing number of political questions are being transformed into legal questions, making it crucial that scholars consider the importance of judicial institutions (Gibson et al. 1998: 343). At the same time, more comparative scholars have begun to view courts as political institutions worthy of study (Shapiro 2008: 329).
Finally, our results demonstrate strong support for our hypothesis that greater levels of de facto judicial independence will improve government respect for physical integrity rights. These findings are robust to changes in measurement, estimation techniques, and model specification. We think that our research suggests two major implications for scholars of human rights and courts. First, our results contradict more pessimistic accounts that completely dismiss courts' ability to have a positive impact on repressive governments. There are fewer governmental abuses of human rights in countries with judiciaries that are independent in practice. However, our results also caution against an overly optimistic assessment of courts' effects, as we find that failing to account for endogeneity will tend to overemphasize the ability of completely independent courts to improve government respect for physical integrity rights.
The article proceeds as follows. We begin with a brief discussion of the existing research on state characteristics and government respect for human rights, and then we narrow in on work that has examined the consequences of both de jure and de facto independence for human rights protections. Building on this literature, we lay out our theoretical argument explaining the causal linkages between changing levels of independence and government respect for physical integrity rights. Next, we present the results of analyses that demonstrate the importance of modeling the relationship between de facto independence and human rights protections through an instrumental variables approach. Finally, we conclude by discussing the implications of our results for scholars of courts and of human rights.

State Characteristics and Government Respect for Human Rights
Scholars have developed a number of empirical models explaining why governments differ in their approach to, and enforcement of, human rights. There is a growing consensus concerning some relationships (see Poe 2004). Three main views seek to explain variations in government protections of their citizens' human rights. One approach emphasizes the role of threats to the regime as stimuli of repression (Gurr 1986;Poe and Tate 1994;Davenport 1995Davenport , 1996Gartner and Regan 1996;Poe et al. 1999;Bueno De Mesquita et al. 2003;Poe 2004). Another perspective explains repression as the result of state characteristics-mainly the level of democracy in a society (e.g., Mitchell and McCormick 1988;Poe and Tate 1994;Fein 1995;Davenport 1996;Poe et al. 1999;Regan and Henderson 2002;Davenport and Armstrong 2004) or its degree of economic development (Mitchell and McCormick 1988;Poe and Tate 1994;Poe et al. 1999;Zanger 2000;Hathaway 2002). Yet another viewpoint stresses the growing importance of transnational noneconomic and economic forces such as foreign aid, trade (Richards et al. 2001;Hafner-Burton 2005), arms sales (Blanton 2000(Blanton , 2005, foreign direct investment (Richards et al. 2001), and the activities of nongovernmental organizations (Welch 1995). There also is increased attention to the effectiveness of transnational actors. The theory of transnational human rights advocacy networks predicts that international human rights regimes can improve actual performance where such networks are strong (Keck and Sikkink 1998;Risse, Ropp, and Sikkink 1999;Ron et al. 2005). It also predicts that international governmental regimes, such as those that regulate international trade, the availability of capital for development, labor standards, may also significantly impact human rights practices (e.g., Hathaway 2002;Abouharb and Cingranelli 2007;Aaronson 2008). With all these factors in mind, we next turn to the literature examining the impact of independent courts on governmental respect for human rights.

Judicial Independence and Government Respect for Human Rights
Research has emphasized the important role an independent judiciary has upon the implementation of human rights legislation (Gibney 1991), a functioning legal system (Rosenthal 1990), maintaining the rule of law (Préfontaine and Lee 1998), the protection of political rights and freedoms (Cross 1999), and the enforcement of property rights and contracts (Sen 1999;Hayo and Voigt 2007). However, it is important to note that, like the conceptual and operational difficulties involved in defining "the rule of law" (Haggard et al. 2008), judicial independence is also the subject of much debate among the social scientists and legal scholars who wish to measure it (see Burbank and Friedman 2002). Thus, it is important to state clearly what we mean by the term before moving on to our explanation of how it can affect government respect for physical integrity rights.
The concept of judicial independence is typically broken down into two subcategories: de jure independence and de facto independence. De jure refers only to the formal protections for insulating the judiciary that are typically written into constitutional documents; in contrast, de facto independence is concerned with both the actions of courts and other political actors in practice. For the purposes of our investigation, we join others in defining de facto judicial independence as a characteristic of courts that allows judges to make decisions based upon their own values and reasons, not others' inclinations, without fear of retribution and with some reasonable expectation of implementation (Becker 1970;Herron and Randazzo 2003). De facto independence, under this definition, also means that courts can be "effective" in the sense that they may rule against the government and have a reasonable expectation to see those decisions enforced. 2 A substantial amount of work has focused on de jure judicial independence, examining how constitutions and existing legal frameworks insulate the judiciary from inappropriate external pressure (Keith 2002a(Keith , 2004Apodaca 2004;Ríos-Figueroa 2006). For example, Apodaca (2004) examines the rule of law explicitly stated in constitutions and its effects on political, civil, and economic human rights as a whole. Her study finds that de jure judicial independence significantly improves human rights. Other research also finds significant linkages between de jure independence and improvements in particular human rights practices (Cross 1999), although these are not as encouraging as human rights activists would hope (Keith 2002a). 3 Reviewing the findings of the de jure literature, Keith (2002b: 200) concludes, "We need to move beyond assessing formal independence to empirically determining the level of actual judicial independence experienced by a state." What has received less systematic attention is how de facto judicial independence affects government respect for human rights. Howard and Carey (2004) create a trichotomous measure of de facto judicial independence using US Department of State Reports for a seven-year period in the 1990s (1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999). They find a positive relationship between de facto judicial independence and the Freedom House index of political rights. A smaller study of predominantly EU countries (31 countries over one year) found mixed support for their expectation that judicial independence should positively impact human rights (Volcansek and Lockhart 2012). In this study, two of the four indicators of judicial independence, the Freedom House Rule of Law index and the World Bank Governance Index (both based on surveys of experts), were associated with improvements in political rights, civil liberties, and socioeconomic rights. However, their remaining indicators (public opinion about judicial trustworthiness and business leaders' perceptions about judicial corruption) were unrelated to human rights outcomes.
With respect to physical integrity rights in particular, only two studies that we are aware of explicitly examine the connection between de facto independence and government respect for these rights. Powell and Staton (2009) find that "effective" judiciaries reduce the frequency of torture during the period following ratification of the Convention Against Torture (1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000). More recently, Keith (2011) explores the connection between de facto independence and physical integrity rights over the 1980-2005 period. 4 She finds a significant relationship between the two and concludes that de facto judicial independence deserves its reputation as an "essential guardian of the rule of law" because of its ability to curb the tendencies of repressive regimes (Keith 2011: 189). However, none of these studies account for the possibility of endogeneity between de facto independence and the protection of physical integrity rights. We return to the implications of this problem later.
The notion that a court might be autonomous in practice and an effective check on the government seems to be closely associated with liberal democracies and therefore not likely to be present in autocratic societies. In order to rule out the possibility that independent courts are simply a function of democracy, we must observe variation in judicial independence across regime types. We turn our attention next to this argument and explore the connection between judicial independence and democracy.

The Connection Between Judicial Independence and Democracy
While democracy and judicial independence often co-vary, Helmke and Rosenbluth (2009) caution scholars not to assume that democracy inevitably leads to judicial independence. Instead, judicial independence can be thought of as one manifestation of certain institutional and normative characteristics of democracy (Dahl 1971;Przeworski 1991). Like competitive elections, an independent judiciary fits into a general notion of constraints or checks on unlimited power and is one mechanism for offering nonviolent means to resolve conflict and to promote broad acceptance of societal choices (Schmitter and Karl 1991).
However, not all democracies have independent judiciaries, and not all independent judiciaries exist in democratic states. In fact, examining the 193 countries in our study reveals only a moderate correlation between democracy and de facto judicial independence. 5 Among democratic regimes, three examples illustrate states with nonindependent judiciaries. In South America, the Peruvian judiciary "suffered from corruption and political influence" and was placed under control of the executive in 1992 (US Department of State 1992: 472), while in Venezuela political parties were successfully able to "influence judicial decisions in particular cases" during 1993 (US Department of State 1993: 570). In Italy, the judiciary is identified as partially independent for the entire period of 2001 to 2010. Several reports during this time mention the widespread intimidation of judges by those involved in organized crime, including six arrests related to an incident in which a judge received a letter containing bullets from a known crime family. Going further, about two thirds of the countries identified as autocratic regimes have, at some point during the 1981-2010 period, had courts that are somewhat or completely independent. For example, Bangladesh, an authoritarian regime, 6 exhibited a partially independent judiciary from 1981 to 1990 (with the exception of 1986). While the civilian courts in Bangladesh were often backlogged with cases, they were "generally considered fair" and not subject to undue influence (US Department of State 1987: 1117). Another authoritarian example is Egypt during the 1981-2004 period. 7 Over this time, the judiciary was classified as partially independent for 16 years and completely independent for an additional eight years. For instance, in 1983, the State Department observed "the regular court system is independent" and identified no instances of improper influence from government officials (US Department of State 1983: 1116-1117. 8 These examples beg the question of why a dictator would tolerate or even establish an independent judiciary when such an institution could act to limit their power. One perspective advanced by Helmke and Rosenbluth (2009: 355-356) is that structural norms, such as limited jurisdiction or appointment of judges with the same values as the ruling regime, may make judicial independence relatively costless for a dictator. Examples of the former include Spain under Franco (Toharia 1975) and the latter, Chile under Pinochet (Hilbink 2007). However, in these scenarios, judges' ability (and willingness) to be responsive to claims of human rights violations may be limited or nonexistent. 9 Dictators may also be willing to establish independent courts in their "final hours" as a check on an incoming democratic regime (Ginsburg 2003;Hirschl 2004).
Another set of explanations highlight the fact that judicial independence can yield benefits for dictators and democratic leaders alike. Increased levels of judicial independence and improved rule of law can have a variety of economic benefits, including attracting foreign direct investment (Li 2006). The "credible commitment" literature further argues that, because courts can protect property rights and can enforce terms of contracts, independent courts can promote investment and economic growth (North and Weingast 1989;Barro 1991;Cameron 2002: 142-143). To the extent that dictators and democratic leaders share goals of retaining political office (Bueno de Mesquita et al. 2003) and economic progress (either enriching themselves or their constituents), they may be inclined to respect the autonomy of their judicial institutions.
Given that judicial independence is not necessarily a function of a democratic regime, we turn next to our theoretical argument about how independent judiciaries affect the protection of physical integrity rights within a country.

How Independent Courts Can Protect Physical Integrity Rights
The last component of our argument has to do with the causal mechanism by which judicial independence can protect physical integrity rights. Judicial independence may contribute to the protection of physical integrity rights through both direct and indirect pathways. The direct mechanism refers primarily to instances in which a court actually rules against the regime in a way that protects physical integrity rights of the individual. The indirect mechanism refers to the deterrent effect that de facto independent courts may have on regime actions. Unfortunately, testing for a direct effect requires case-level data with information about physical integrity rights abuse. At this point, however, no large-scale, case-level dataset with this kind of detail exists. For this reason, our discussion places more emphasis in explicating the indirect mechanisms and returns to a more speculative discussion of direct effects at the end of this section.
As discussed above, independent judiciaries are not necessarily byproducts of democracy; even autocratic regimes have incentives for establishing or maintaining de facto independent courts. However, the existence of a de facto independent court brings along with it the possibility that such a court could sanction the government for abuses of physical integrity rights. Even the mere threat of being deposed can be a powerful motivator for autocratic rulers (Bueno de Mesquita et al. 2003), and, consequently, governments may curtail their violations in anticipation of monitoring by the courts. Support for this contention comes from Powell and Staton (2009). Governments contemplating torture consider their courts' effectiveness when deciding to join international treaties. This is because an effective domestic judiciary can enforce the treaties' legal requirements. Thus, an independent court can have a deterrent effect upon a government inclined to violate physical integrity rights.
Independent judiciaries may also indirectly influence the actions of the public in ways that affect the frequency of human rights abuses by the government. The public is likely to view decisions by courts that are free of inappropriate government intervention as fairer because the decisions appear to be based on the law rather than political expediency (Gibson et al. 1998). The public's belief in the integrity of the system will tend to mute the reactions of groups that lose out, reducing the likelihood of violent protests against the state and making the protests that do occur seem less threatening to the incumbent regime (Moore 1998). As a result, governments faced with fewer threats tend to repress less (Poe 2004).
There are also arguments in the transitional justice literature about the justice cascade (Sikkink 2011: 5) that point to a shift in the legitimacy of the norm of individual criminal accountability especially for state officials' past human rights violations. Of particular interest for this article is the finding that prosecutions of human rights violations in transitioning countries improves government respect for human rights in neighboring countries (even in those not undertaking prosecutions), as state officials in these neighboring countries also change their behavior. Some of these officials may change their behavior because they are deterred through the fear of prosecution, while others may change their behavior because of shifting norms (Sikkink 2011: 183-188). This is especially well illustrated in both Latin America and Europe, where many of the foreign prosecutions of violations that took place within these regions began in other countries in the same region (Sikkink 2011: 138-139). Likewise, European courts have also been the source of prosecutions in Latin American countries when it was thought that judiciaries in these states were not willing or able to undertake such prosecutions. As we discuss in more detail below, there is some indication that part of what has facilitated these human rights prosecutions has been an independent judiciary. While the transnational consequences of greater judicial independence is beyond the scope of this article, it does lead to the intriguing possibility that improved levels of judicial independence in one country may actually benefit human rights protection in neighboring countries where the judiciary is able to rule freely. (For example, courts may pursue these cases using the concepts of universal jurisdiction that have now been embedded in a variety of international human rights conventions like the Convention Against Torture and the Inter American Convention on Forced Disappearance.) We now turn our attention to how a direct causal mechanism might operate. There are two relevant questions here: First, will a court actually rule against the government, and second, what effect would such an adverse ruling have? Even given strategic action by judges and courts, we speculate that examining outcomes in individual cases would reveal that, ceteris paribis, de facto independent judiciaries should be more likely to protect physical integrity rights than courts that are not autonomous. 10 Recent research on transitional justice has also documented the proliferation of criminal prosecutions of state officials for past human rights violations beginning in the mid-1970s (Olsen, Payne, and Reiter 2010;Sikkink 2011). While there is some debate about the exact configuration of institutions (i.e., trials, truth commissions, and amnesties) required to improve human rights outcomes in these transitioning states, the basic premise of this work is that accountability for past violations reduces the likelihood of future violations. One dominant perspective argues "holding individuals accountable through prosecutions deters future perpetrators from committing violence [and] establishes and legitimizes the rule of law" (Olsen, Payne, and Reiter 2010: 983). For example, Sikkink (2011: 82) notes that "Argentine judges had somewhat more autonomy, and this may have given them more room for independent judicial decisions in human rights prosecutions than colleagues elsewhere." 11 In sum, if judiciaries with greater independence are more likely to undertake criminal prosecutions of human rights violations, then the transitional justice literature provides an additional avenue linking judicial independence to improved governmental respect for physical integrity rights.
With respect to the second question (i.e., what effect would an adverse ruling have), we must confront the objection that courts are limited in their abilities to affect social change (Rosenberg 1991) and to protect the rights of political minorities from the ruling regime (Dahl 1957). While we acknowledge these critiques of courts' institutional capacity, we join others (McCann 1994;Keck 2009) who have taken a broader perspective in arguing how court action can play a role in evolving social struggles. McCann (1994: 10) observes that groups who pursue a legal mobilization strategy may reap a variety of benefits, even apart from "short-term remedial relief" for victims of government abuse. For instance, court action can lead to the development of favorable precedents, which may help victims in other cases. Favorable rulings can also help build a movement, can generate public support for new rights claims, and can provide leverage to supplement other political tactics (McCann 1994: 10). In addition, attributes of independent courts, such as institutional legitimacy (Gibson and Caldeira 2003) and their moral standing (Barnett and Finnemore 2004), may allow them to draw public (and even international) attention to violations by the government through their opinions and rulings. 12 These actions by the judiciary may affect that nation's reputation in the international community or make it a less appealing location for some foreign investors. Finally, losses in court can also benefit a cause, as Moustafa and Ginsburg describe in the context of courts in authoritarian regimes: "Even when activists do not win particular cases, courts can facilitate rightful resistance by providing publicity about government malfeasance, deterring future abuses and developing skill sets for activist leaders" (2008: 13). This discussion leads to the following expectation: HYPOTHESIS: Higher levels of de facto judicial independence will improve government respect for physical integrity rights.
Existing literature has found a number of important factors that need to be controlled for in order to assess the impact of judicial independence on government respect for physical integrity rights. More democratic countries as well as wealthier countries tend to have governments that provide more respect for the physical integrity rights of their citizens (e.g., Poe et al. 1999;Poe 2004). Countries with relatively sizeable populations and large population increases place greater stress on limited government resources (Henderson 1993). In turn, states may not be able to accommodate all these demands, increasing the likelihood of government repression to maintain domestic control of their citizens (Poe 2004). Likewise, high levels of domestic conflict and involvement in interstate war increase the level of threat faced by governments (Davenport 1995;Landman 2005). In response to this increased threat, governments attempt to maintain control by abusing citizens' physical integrity rights (e.g., Davenport 1995;Poe et al. 1999;Poe 2004; see also Landman 2005). These factors were included as control variables in the analysis.

The Importance of Accounting for Endogeneity
The controls mentioned above account for alternative explanations that could mediate the relationship between judicial independence and respect for physical integrity rights. However, they do not account for the possibility that there are endogenous processes that jointly increase the likelihood of some countries developing effective, independent judiciaries and improving respect for physical integrity rights. As we were unable to identify any published work that predicts the emergence of de facto judicial independence, we draw upon the widely cited work of La Porta et al. (1999), who argue that a variety of demographic, religious, historical, and locational factors affect the effectiveness of institutions (including the judiciary). For example, the different geographic locations of countries that were colonized have been linked to the development of (mal)functioning institutions. This is because inhospitable conditions encouraged what has been described as "extractive" states that were designed to maximize the transfer of resources to the colonizers with minimum investment, little provision for the protection of property rights, and scant regard for any checks and balances on the behavior of government officials in these colonized states. In contrast, conditions that were more conducive to colonial settlement led to the development of "Neo-Europe" states with a greater emphasis on rules respecting private property and checks against government power (Acemoglu et al. 2001). We include these measures in the first stage of our model predicting judicial independence.
The empirical consequences of not accounting for endogeneity can be severe. If there are endogenous processes taking place that jointly determine higher levels of judicial independence and improved government respect for physical integrity rights then this means from an econometric perspective that our key measure of interest is not randomly assigned since there is an underlying process that makes judiciaries more likely to function better in some countries than others and makes human rights outcomes also better in some countries than others. This is a potentially a very serious issue for researchers interested in causality, because failing to account for endogeneity undermines researchers' ability to give a causal interpretation to the model results (Cameron and Trivedi 2009: 171) and can lead to biased coefficients. In order to deal with this problem, we adopted an instrumental variables approach that first predicts de facto independence and then predicts its effect on the protection of physical integrity rights, which we explain in more detail below.

Research Design
We employ a cross-national, annual time-series dataset comprised of 193 independent countries with a population of at least 500,000 in 1981 using the Correlates of War (2008) framework. Our unit of analysis is the country year. The data span the time period from 1981 to 2010, going five years beyond the most extensive analysis of this question (Keith 2011). All the independent and control variables are lagged one year to minimize the possibility of reverse causality, and robust standard errors are also included to limit the effects of heteroskedasticity and within panel autocorrelation. 13 Our analysis examines whether de facto judicial independence is associated with improved government respect for physical integrity rights. As discussed above, we account for the possibility of an endogenous relationship between countries that have both independent judiciaries and are protective of their citizens' physical integrity rights by using an instrumental variables approach. Specifically we selected an instrumental-variables regression Limited Information Model (LIML) estimator because LIML "has been found in recent research to outperform both [Two Stage Least Squares] 2SLS and [General Methods of Moments] GMM [models] in finite samples" (Cameron and Trivedi 2009: 175), regardless of the strength of the instruments (Cameron and Trivedi 2009: 199). Additionally, the LIML approach is more accurate than 2SLS in situations where overidentification (i.e., more instruments than regressors) is a problem. Finally, LIML offers a number of diagnostics to evaluate the strength of the instruments, and these diagnostics are unavailable for ordered models that account for endogeneity. However, because the dependent variables in both stages of our model (de facto judicial independence and government respect for physical integrity rights, respectively) are ordinal measures, we recognize that an ordered model would be more appropriate for our data. Therefore, to ensure the robustness of our findings, we also estimated our model using a bivariate-ordered probit model, which better reflects the distribution of our dependent variables at both stages (Sajaia 2011). The findings from this alternate model mirror the findings presented in the main analysis below (see Appendix D).
In our analysis, we treat our measure of de facto judicial independence as our endogenous regressor. In the model specification, it is important to have one or more additional variables (excluded exogenous variables) that are correlated with our measure of de facto judicial independence. These variables do not directly affect our physical integrity rights measure (the dependent variable in the second equation). Due to the paucity of research that has tried to systematically predict the emergence of judicial independence, we draw upon previous research that has predicted the quality of institutions (La Porta et al. 1999) as a proxy for estimating the factors that make judicial institutions function properly. These excluded exogenous variables (our instruments) are used to predict the degree of de facto judicial independence: ethnolinguistic heterogeneity, proportions of populations by religious affiliation (Catholic, Protestant, Muslim), legal origin (English common law, French, German, Scandinavian, Socialist/Communist), and latitude. 14 The included exogenous variables form the basis of our second-stage regression equation and are displayed in our results. These variables directly affect our physical integrity rights outcomes and are automatically included as instruments (StataCorp 2009: 745). Below, we elaborate on the operationalization procedures used for the key independent and dependent variables to ensure replicability.

Independent Variables De Facto Judicial Independence
While a number of de facto measures exist (Henisz 2000;Feld and Voigt 2003;Howard and Carey 2004;Cingranelli and Richards 2008;Keith 2011), we focused on three criteria as the basis for our selection of variables: construct validity, broad geographic coverage, and nonsystematic patterns of missing data. 15 It was important that our measure would tap into purely de facto aspects of judicial independence, not de jure, and would provide a good approximation of the extent to which judges are able to be the authors of their own opinions and to have their decisions reasonably respected. Secondly, to ensure generalizability, it was important that the measure included observations for a wide range of countries. Finally, we wanted to avoid problems related to nonrandom patterns of missing data that have been identified with de facto measures (Ríos-Figueroa and Staton forthcoming). Previous research has observed that de facto measures based on the United States State Department Human Rights Reports do not exhibit statistically significant patterns of missing observations that vary by level of economic development (Ríos-Figueroa and Staton forthcoming: 19).
These criteria led us to employ Keith's (2011) measure of de facto judicial independence. Based on the annual State Department Human Rights reports, the measure is a trichotomous variable covering 193 countries for the years 1981 to 2010. 16 As such, it provides the most extensive coverage of any of the existing de facto measures. Keith (2011: 154) codes the measure as follows: A "2" is labeled as a "fully independent judiciary." The judiciary is reported as "'generally independent' or is independent in practice with no mention of corruption or outside influence." A "1" is reported to "be somewhat independent in practice with reports of (some) pressure from the executive 'at times' or with occasional reports of corruption." A "0" is labelled as a "non-independent judiciary." The judiciary is reported as "not being independent in practice; is reported to have significant or high levels of executive influence or interference; or is reported to have high levels of corruption."

Government Respect for Physical Integrity Rights
The level of government respect for physical integrity rights is the dependent variable in the second stage of our model. The index comes from the CIRI human rights dataset (Cingranelli and Richards 2008). It comprises the level of government respect for four different physical integrity rights-freedom from extrajudicial killings, forced disappearances, political imprisonment, and torture. The sources of information used to develop this dataset were the annual U.S. State Department Country Reports on Human Rights Practices and Amnesty International annual reports. Each of the four physical integrity variables is coded on a three-part scale where 0 = frequent violations of the right (50 or more), 1 = some violations (1-49), and 2 = no violations. The index is constructed from the summed value of these four components and can range from a score of "0" indicating a government's frequent violations of all physical integrity rights to a score of "8" indicating a government's complete respect for all physical integrity rights. Table 1 provides a summary of the operationalization of all variables used in the analysis to predict physical integrity rights violations. These controls include economic factors: gross domestic product (GDP) per capita, change in GDP per capita, and trade as a percentage of GDP. They also include political factors, such as the level of democracy, the log of population, the density of population, and the ratification of the International Covenant on Civil and Political Rights (ICCPR) covenant (United Nations 1966). To account for the effect of conflict on physical integrity violations, we control for both the level of interstate conflict and civil conflict. Finally, we include a year variable to control for any temporal trends that have changed government respect for physical integrity rights. 17 One of our key control variables, democracy, deserves a fuller explanation. We constructed the democracy measure from the Polity IV dataset (Marshall, Gurr, and Jaggers 2009), one of the most popular data sets used in both international relations and comparative politics literature because of its broad geographic and long historical scope, as well as the careful and transparent manner in which the data are generated. The existing Polity IV dataset's measure of democracy (Marshall et al. 2009: 14-15) is derived from four components, including the "competitiveness of political participation," the "openness of executive recruitment," the "competitiveness of executive recruitment," and the "constraints on the chief executive," resulting in a 0-to 10-point scale. Part of the coding on the "constraints on the chief executive" includes those imposed by the judicial branch. We adjusted this measure by removing the "constraints on the chief executive" component, resulting in a new measure that ranges from 0-6. This step was undertaken to ensure that our independent variables were structurally independent of one other. Pairwise correlations amongst all our control variables indicate no problems of multicollinearity (see Appendix B). Furthermore, Variance Inflation Factor (VIF) tests show that no variable obtains a value greater than 2.49, providing additional evidence that multicollinearity is not a problem in any of our models. We now move to a discussion of our results.

Temporal Factors Year
Year Constructed

The Impact of Judicial Independence on Government Respect for Physical Integrity Rights
Figure 1 provides some simple descriptive statistics about government respect for physical integrity rights across different levels of de facto independence. The far left bar represents countries with a "0" (not independent) on the Keith (2011) index, showing a relatively low score on the CIRI index (3.5 out of 8). In comparison, the next two bars ("somewhat independent" and "completely independent") show an increasing improvement in respect for physical integrity rights (to a high of about 6.5 out of 8, for completely independent courts). While Figure 1 suggests preliminary support for our hypothesis, it of course fails to control for alternative explanations or endogenous processes. We next turn to a more rigorous examination of our expectations. In Table 2, the second column displays the results of an analysis that does not account for endogeneity, the typical approach taken in existing research, compared to an instrumental variables approach that does control for the possibility of endogeneity (presented in Column 3). 18 As hypothesized, de facto independence is positive and significantly related to improved human rights outcomes in both models. (Recall that higher values on the index correspond to fewer violations of physical integrity rights.) While this finding is robust to the choice of estimation techniques, what does change across the models is the substantive nature of the relationship between de facto judicial independence and government respect for physical integrity rights. The single-stage model overemphasizes the positive impact of de facto independence, predicting a 0.53 increase on the CIRI index for a one-unit change in level of independence. In comparison, the instrumental variables model predicts a more modest 0.43-unit increase on the CIRI index, after accounting for endogeneity. Holding all other factors to their mean or modal value, the predicted level of government respect for physical integrity rights moves from about 4.45 (where courts are not independent) to about 5.33 (where courts are independent), almost a whole unit change on the CIRI index. This would be indicative of moving from a human rights situation found in countries like the Dominican Republic, Kazakhstan, and Tunisia for most of the 2000s to one found in The importance of accounting for these endogenous processes becomes even more apparent when we graph the predicted effects of the two models side by side (see Figure 2). In countries where the judiciary is not independent, the single-stage model underestimates government respect for physical integrity rights, compared to the LIML model. However, for completely independent courts, the reverse is true: The single-stage model overestimates the level of government respect for physical integrity rights. For countries with judiciaries that are "somewhat independent," the predictions of the two models are fairly close together, though the single-stage model still offers an overly optimistic assessment of the influence of independent courts.
Next, we turn to a discussion of our control variables. Countries with higher levels of GDP per capita show significantly improved levels of government respect for physical integrity rights, as expected. Consistent with the literature (e.g., Davenport and Armstrong 2004), more democratic countries were significantly associated with improved levels of government respect. Our results indicate that more populous countries (indicated by the log of population) are associated with worsened human rights outcomes. Although there is some evidence that countries with greater population density should be associated with worsened human rights outcomes, the results from our analysis show that population density is associated with improved human rights outcomes. As others have found (Vreeland 2008), ratification of the ICCPR was associated with worsened human rights outcomes, although our finding is only at the .10 level of confidence. And while interstate conflict was not statistically significant, higher levels of internal conflict were associated with worsened human rights outcomes.
Finally, we subjected our findings to a series of robustness tests (see the Appendices). Across all of these alternative specifications, we see strong support for our hypothesis: Higher levels of de facto judicial independence remain associated with improved government respect for physical integrity rights, at the .01 level of confidence or higher. 20 Below, we describe each specification in more detail.
First, we estimated each model from Table 2 using a bivariate-ordered probit (see Appendix D). This method more closely approximates our ordered dependent variables, with the limitation that it does not allow us to assess the strength of our instruments. Again, our measure of de facto judicial independence is associated with improved government respect for physical integrity rights, significant at the .001 level of confidence. 21 Next, we explored the possibility that structural differences in country's constitutions, and not de facto independence, are driving the effect on physical integrity rights. To account for this alternative explanation, we included a control for de jure judicial independence (Keith 2011) in the second stage of our model (see Appendix E). 22 The coefficient for de jure independence is negative and statistically significant here, but substantively negligible (-0.01). More importantly, including this control does not affect the direction or significance of de facto independence. Controlling for de jure independence does reduce the size of the effect of de facto independence somewhat (from .437 to .396); however, comparing the two types of independence shows clearly that, to the extent judicial independence is associated with improved human rights outcomes, it is because of de facto independent courts.
Another set of models considered the possibility that our findings are an artifact of the CIRI index of physical integrity rights violations. To ensure the robustness of our results, we estimated the instrumental variables LIML model using another commonly used indicator, the Political Terror Scale (PTS). Appendix F shows the results from models using three different variants of the PTS: the average of Amnesty International and the State Department, the Amnesty International Index, and the State Department Index. (The indices were inverted for ease of comparison.) Across all three models, de facto judicial independence remains associated with improvements on the PTS scale, significant at the .001 level.
Finally, we evaluate an alternative explanation from Davenport and Armstrong's work (2004): namely, that improved human rights outcomes resemble a step function and tend to come at higher levels of democracy. To rule out the possibility that the misspecification of democracy is affecting our results, we ran two models with different, binary specifications of our democracy variable (see Appendix G). The first dichotomizes our 0-6 measure, coding values of 4 or greater as a "1" to indicate a democracy; the second specification uses a higher threshold of 5 and above to indicate a democracy (coded a "1"). Both of our binary measures are statistically significant in the expected direction, and the larger coefficient on the higher threshold variant (the second specification) offers support for their arguments. As before, our measures of de facto judicial independence are associated with improved government respect for physical integrity rights in support of our hypothesis.

Discussion
In the most comprehensive analysis yet on this topic, we find strong support for the argument that de facto judicial independence will lead to fewer instances in which governments violate their citizens' physical integrity rights. This finding has important policy, theoretical, and empirical implications. On the policy front, we counsel cautious optimism about the degree to which human rights advocates can count on courts, by themselves, to stop human rights abuses. At a minimum, our results suggest that state and nonstate actors interested in promoting human rights respect should campaign for and demand improved levels of de facto judicial independence, as one part of a multipronged strategy to improve respect for physical integrity rights.
From a theoretical perspective, our work offers a measure of support for both proponents of independent courts as protectors of human rights, as well as for more cynical observers. Judicial institutions can indeed be effective in ensuring greater protection of citizens' physical integrity rights. However, while independent courts are able to curtail some governmental violations of physical integrity rights, the evidence shows that they do not stop all abuses completely, even under the best-case scenario of complete independence. Put simply, judicial independence is not a panacea, but neither is it an empty promise.
Empirically, our research underscores the importance of accounting for any endogenous process that makes governments more likely to have both independent courts and greater human rights protection. Research that does not account for this endogeneity will tend to overstate the positive impact of courts on human rights protection at high levels of independence and underestimate the influence of courts that are not independent. Future empirical research should build on this insight to delve further into the factors that contribute to the rise (and decline) of de facto independence.
The question of judicial independence and human rights protection is a topic worthy of continued study. Given the literature on courts in autocratic regimes (Ginsburg 2008), it would be instructive to compare the ability of judiciaries to constrain physical integrity rights abuse under conditions of both democracy and dictatorship. Future research could also explore whether the transnational consequences of prosecutions on human rights found in Sikkink's (2011) work apply to other aspects of the judicial process. For instance, does increased de facto judicial independence in one country promote better respect for human rights in neighboring countries, due to fear of prosecution by the state with the independent judiciary? In addition, research using case-level data across multiple countries and years could help unpack the precise causal mechanism that produces the aggregate effect we identify here. Such analyses might control for the political environment, the ideological preferences of the judges, and the capacity of the legal system to investigate and prosecute human rights claims. Along these lines, research should also take into account the procedural and jurisdictional rules that dictate the extent to which courts are able to hear (and investigate) individuals' claims against the government. Data collection efforts like those of Haynie et al. (2007) will be vital in the next wave of research on the impact of judiciaries, worldwide, to protect human rights. (2011) is one notable exception. 2. We concur with Burbank and Friedman (2002: 11-12) that no court enjoys "complete decisional independence" from any conceivable constraint (Seidman 2001). Given the wealth of scholarship showing the influence of extralegal factors in judicial decision making at the individual level (Segal and Spaeth 1993;Hall and Brace 1996;Helmke 2002), such a definition sets the bar too high for judicial independence and is probably undesirable in any event. Our definition of de facto independence also accepts that judges may be (properly) constrained by legal and institutional factors in their decisions and that political entities may properly use the regular procedures of the court to advocate for their position without necessarily compromising the independence of the court (Kornhauser 2002: 48).

Keith
4. While Keith's (2011) study does not control for the endogenous process that affects both the level of de facto judicial independence and level of government protection for physical integrity rights, she does model the likelihood of de facto judicial independence as a function of de jure independence. In a robustness test, we include a measure of de jure judicial independence as an additional control. Our core findings remain consistent. 5. Across the range of democracy and judicial independence, the correlation between the two variables is .34. When the democracy variable is limited to observations with a relatively high democracy score (4 or higher on our 0-to 6-revised democracy score), the correlation rises only to .61. 6. Bangladesh received scores of "0" on both our revised democracy score as well as the original Polity democracy measure for the same 1981-1990 period. 7. Egypt also received the lowest scores of "0" on our revised democracy score as well as the original Polity democracy measure for the period 1981-2004. 8. For an in-depth analysis of why authoritarian Egypt established a constitutional court with the power of judicial review, see Moustafa (2007). 9. In another example, Tate and Haynie (1994) describe how Filipino judges appointed under the Marcos regime were selected primarily because of their loyalty and personal connections to Marcos. Because of this congruence, no pressure from the administration was necessary for favorable treatment by the court. 10. Judicial decisions that check the government are not made in a political vacuum. Courts do evaluate the political conditions before making an adverse ruling against the government (Iaryczower, Spiller, and Tommasi 2002). Judges also consider whether there is public awareness and understanding of a case, since the presence of these two factors increases the cost of noncompliance by the executive (Vanberg 2001;Carrubba and Zorn 2010). 11. Sikkink (2011: 82-83) also notes that some civil law systems (e.g., Argentina) permit private citizens to pursue human rights prosecutions when state prosecutors are unwilling to do so. 12. This is similar to the concept known as "naming and shaming." Naming and shaming is "a popular strategy to enforce international human rights norms and laws [whereby] nongovernmental organizations, news media, and international organizations publicize countries' violations and urge reform" (Hafner-Burton 2008: 689). We think that this concept also has applicability to national-level judiciaries. 13. One way to evaluate our hypothesis would be to examine judicial outcomes across a large number of countries over time and to see how often courts ruled against the regime in physical integrity rights cases, controlling for other factors. Unfortunately, comprehensive case-level and judge-level data of this type are not available for the large number of countries and years of our study. (The High Courts Judicial Database currently includes case-level information for only 11 countries and a variety of different time periods.) Despite these limitations, we believe that our approach is valid and can shed some light on independent courts' ability to influence government protection of physical integrity rights. 14. La Porta et al. (1999: 239, 244) note that "temperate zones have productive agriculture and healthier climates" that has in turn helped them to develop their economies and institutions.
In addition, because their discussion of functioning institutions posits specific expectations for Protestant countries compared to Muslim and Catholic countries, the variables we include reflect this emphasis and exclude "all other religions" (a residual category for which we have no specific expectations). However, when this "other" category is included in the analysis, all results that are significant at the .05 or better remain robust to the new specification. 15. For a more thorough comparison of the strengths and weaknesses of different measures of judicial independence, see Ríos-Figueroa and Staton (forthcoming). 16. There is no evidence of systematic bias in the reports for the period we examine (Poe et al. 2001). 17. Descriptive statistics for all variables are available in Appendix A.
18. We generated diagnostics to evaluate the strength of our first-stage instruments from our instrumental variables LIML model. The results give a significant (p < .001) F-statistic for the models that is well above 10, the value considered to be the demarcation of strong instruments (Cameron and Trivedi 2009: 193). This gives us confidence that our results have properly instrumented the endogenous process that makes some states more likely to have independent judiciaries and better human rights practices. 19. The results from the first-stage model indicate that countries where more than 80 percent of the population is Catholic or Muslim have lower levels of judicial independence, in comparison to countries where more than 80 percent of the population is Protestant. The results also show consistently negative associations of the legal origins variables (compared to English common law systems that is the excluded category) with judicial independence. Additionally, higher latitudes, countries further away from the equator, are associated with greater levels of judicial independence. A summary of these results appears in Appendix C. 20. Unless stated otherwise, each alternate model uses the instrumental variables LIML Model. 21. We also ran models using the Cingranelli and Richards (2008) measure of judicial independence but opted not to present them because of the measure's hybrid nature (Ríos-Figueroa and Staton forthcoming: 9-10). It combines both structural elements of judicial independence, such as the extent of constitutional provisions for an independent judiciary, together with dynamic elements, typically found in de facto measures such as if the judiciary acted independently of the government during that year. This measure also indicated higher levels of judicial independence were associated with improved respect for physical integrity rights. 22. Correlation between de facto and de jure independence variables is .04.