New Evidence for a Positive Relationship Between De Facto Judicial Independence and State Respect for Empowerment Rights

Does increased judicial independence lead to increased state respect for empowerment rights? Initial research on this topic has suggested an affirmative answer. Advances in measurement, however, call into question our understanding of the effects of judicial independence. In this paper, we re-examine the effect of de facto judicial independence on state respect for empowerment rights, making use of new measures and different modeling methods. In our empirical analysis, we find a positive association between the two concepts. This result is robust to a range of measures and modelling strategies. Increased judicial independence appears to substantially limit state violations of empowerment rights. Forthcoming in International Studies Quarterly ∗We would like to thank Christopher J. Fariss and Luke Keele for many helpful comments and suggestions. We would also like to thank M. Rodwan Abouharb, K. Chad Clay, and Linda Camp Keith for generously sharing their data with us. This research was supported by The McCourtney Institute for Democracy Innovation Grant and the College of Liberal Arts at Pennsylvania State University. All data files necessary to replicate the analysis presented in the article will be publicly available upon publication at dataverse repositories maintained by the authors. Introduction A major function of courts is to limit governmental power. Courts can constrain governmental exercise of power by nullifying or limiting governmental actions that violate constitutional or international laws. For example, the power of judicial review enables a constitutional court to nullify a duly approved law that oversteps the bounds of power allocated to the legislature by the constitution. The importance of courts extends beyond the ability of a constitutional court to invalidate legislation. Courts can also constrain government power in other arenas, ensuring that governments respect the rights of their citizens. Of course, not all courts are equally powerful; a court’s ability to be efficacious depends on its independence. Lacking the ability to implement their own decisions, courts are dependent upon the willingness of other political actors to follow their decisions in order for their decisions on paper to become binding law in practice. With this in mind, scholars have typically differentiated between courts that are independent by institutional design (de jure independence) and those that are independent in practice (de facto independence) (Linzer and Staton, 2015). Studies suggest that higher levels of both types of judicial independence, de facto and de jure, are associated with increased state respect for physical integrity rights (Abouharb, Moyer and Schmidt, 2013; Crabtree and Fariss, 2015; Lupu, 2013). Physical integrity rights encompass the rights to be protected from extrajudicial murder, forced disappearance, torture, and political imprisonment (Fariss, 2014a). Beyond physical integrity rights, empowerment rights represent another set of obligations which governments must respect (Richards, Gelleny and Sacko, 2001).1 While physical integrity rights are undoubtedly important, empowerment rights are also vital. They affect citizens’ fundamental relationship with their government: the ability of citizens to criticize the government, the ability to live their lives according to their own belief systems, and their ability to seek refuge from repressive gov1A common conceptualization of empowerment rights includes the right to electoral self-determination, the right to domestic movement, the right to foreign movement, the right to religious freedom, the right to freedom of speech, and the right to assembly and association (Cingranelli, Richards and Clay, 2015).


Introduction
A major function of courts is to limit governmental power.Courts can constrain governmental exercise of power by nullifying or limiting governmental actions that violate constitutional or international laws.For example, the power of judicial review enables a constitutional court to nullify a duly approved law that oversteps the bounds of power allocated to the legislature by the constitution.The importance of courts extends beyond the ability of a constitutional court to invalidate legislation.Courts can also constrain government power in other arenas, ensuring that governments respect the rights of their citizens.
Of course, not all courts are equally powerful; a court's ability to be efficacious depends on its independence.Lacking the ability to implement their own decisions, courts are dependent upon the willingness of other political actors to follow their decisions in order for their decisions on paper to become binding law in practice.With this in mind, scholars have typically differentiated between courts that are independent by institutional design (de jure independence) and those that are independent in practice (de facto independence) (Linzer and Staton, 2015).
Studies suggest that higher levels of both types of judicial independence, de facto and de jure, are associated with increased state respect for physical integrity rights (Abouharb, Moyer and Schmidt, 2013;Crabtree and Fariss, 2015;Lupu, 2013).Physical integrity rights encompass the rights to be protected from extrajudicial murder, forced disappearance, torture, and political imprisonment (Fariss, 2014a).Beyond physical integrity rights, empowerment rights represent another set of obligations which governments must respect (Richards, Gelleny and Sacko, 2001). 1 While physical integrity rights are undoubtedly important, empowerment rights are also vital.They affect citizens' fundamental relationship with their government: the ability of citizens to criticize the government, the ability to live their lives according to their own belief systems, and their ability to seek refuge from repressive gov-ernmental actions.Recent research suggests that independent courts also affect citizens' empowerment rights, with countries that have independent courts more likely to respect citizens' empowerment rights (Keith, 2012).
Recently, however, advances in measurement have called into question our understanding of the effects of judicial independence.Traditional measures of de facto judicial independence, drawn (directly and indirectly) from reports issued by the U.S. Department of State, have indicated that courts worldwide have generally become less independent over the past three decades.In contrast, most measures suggest that de jure independence continues to increase.This has caused some to claim that there is a "growing gap between practice and promise" (Keith, 2012, 155).A new measure of judicial independence, one that better captures the underlying latent construct (Linzer and Staton, 2015), suggests that this perceived gap might not exist.Indeed, the trend over the past three decades has been one toward more de facto judicial independence.
The discovery that courts worldwide have become more, rather than less, independent over time calls into question the received wisdom about the relationship between judicial independence and human rights.Indeed, if the more valid measure of judicial independence were used, then perhaps the relationship between judicial independence and respect for empowerment rights may dissipate or more worryingly lead to the opposite conclusion: that more independent courts do a worse job of protecting citizens' rights.If we hope to understand the consequences of independent courts, we need to make sure that we have accurately estimated the actual relationship between judicial independence and rights protections.In a time when citizens' abilities to exercise their empowerment rights to publicize government abuses, seek refuge from abusive governments, and to exercise their ability to practice their religion without interference from government intrusion makes news almost daily, ascertaining the correct relationship between judicial independence and empowerment rights has both important scientific and policy implications.
In this paper, we re-examine the effect of de facto judicial independence on state respect for empowerment rights, making use of new measures of both de facto judicial independence and state respect for empowerment rights (Schnakenberg and Fariss, 2013;Linzer and Staton, 2015).In our empirical analysis, we find that independent courts exert a strong positive effect on state respect for empowerment rights.We obtain this result both when we use an aggregate index of empowerment rights and when we use disaggregated measures of individual rights.This result is also robust to a range of measures, modelling approaches, and specifications.Increased judicial independence is strongly correlated with increased state respect of empowerment rights.This important finding underscores the importance of recent efforts to empower judiciaries in countries with histories of human rights abuse.

Independent Courts and Rights Protections
Do independent courts safeguard human rights?The literature has been primarily concerned with the ability of independent courts to safeguard physical integrity rights.Increased independence enables a court to take a stand against a repressive regime because the court can do so with only minimal risk of effective reprisal by the regime.As (Keith, 2012) writes, "the legal institutions associated with democratic systems... can potentially provide the public and other political actors with the tools and venues by which they can hold the regime accountable should it fail to keep its formal commitments" (169).Of these legal institutions, independent courts are better able to hold regimes to their commitments than others because they have a separate base of support through which they are able to withstand reprisal by the regime.Put differently, courts are unlikely to take these actions if doing so will result in existential consequences for the court or jeopardize the continued tenure or welfare of the judges who sit on that court.As such, increased judicial independence the circumstances under which the regime is limited in its ability to punish a court for an anti-regime ruling should be associated with stronger respect for physical integrity rights.
Aside from the direct actions of courts, the litigation process by which courts operate provides a mechanism to publicize violations of human rights, thereby providing an indirect mechanism through which independent courts may lead to more respect for human rights.
According to this theory, the litigation surrounding these decisions may have powerful negative reputational and resource costs for the regime, providing an indirect mechanism through which independent courts are associated with more respect for rights; because regimes that would otherwise repress their citizens are fearful of these costs, they respect rights in order to limit their exposure to harmful litigation (Powell and Staton, 2009a;Keith, 2012).As such, the presence of an independent court serves as a reminder that the threat of harmful litigation is always present so long as that independent court is available to hear that litigation.
Prior studies have primarily analyzed the extent to which de jure judicial independence is associated with violations of physical integrity rights.While the empirical evidence presented in older studies is mixed, recent research suggests a strong positive correlation between de jure judicial independence and state respect for physical integrity rights (Cross, 1999;Keith, Tate and Poe, 2009;Keith, 2012;Powell and Staton, 2009b;Lupu, 2013).2 More recently, scholars have begun to examine the effect of de facto judicial independence on human rights abuses.Because de jure judicial independence measures only formal promises of independence, rather than the extent to which courts are independent in practice, one concern is that de jure judicial independence might overestimate the extent to which a court, in practice, is independent.In her impressive study on the subject, Keith (2012) presents persuasive evidence that independent courts constrain political repression.Using newer measures of both state respect for human rights and de facto judicial independence, Crabtree and Fariss (2015) provide additional evidence that independent courts can protect individuals from violent human rights abuses.
But does increased judicial independence also protect other human rights, such as empowerment rights?In contrast to physical integrity rights, which provide the individual with protection against physical harm, empowerment rights "provide the individual with control over the course of his or her own life and, in particular, control over the state."(Richards, 2003, 29).It seems reasonable that if increased de facto judicial independence is correlated with increased state respect for physical integrity rights, it might also be correlated with increased state respect for empowerment rights.
There are several reasons why increased judicial independence may be associated with increased protection for empowerment rights.First, violations of empowerment rights are typically easier to observe than violations of physical integrity rights, which frequently occur beyond public view.For instance, state laws that constrict domestic or international freedom of movement are typically more observable than state acts of torture.Since it is easier for courts to monitor and sanction abuses it can see than abuses it cannot, we should expect that independent courts have a greater capacity to limit state abuses of empowerment rights.
Second, the increased visibility of violations of empowerment rights makes the threat of harmful litigation as salient for violations of empowerment rights as it is for violations of physical integrity rights.Regimes know that violations of these empowerment rights may lead to litigation that could result in negative publicity and harmful reputational costs.Thus, just as it does for physical integrity rights, this threat of litigation provides one mechanism through which independent courts may be associated with greater respect for empowerment rights.
Third, many empowerment rights, in practice, provide citizens with protections to dissent against the government.Courts that are not independent are, by definition, closely linked to the ruling regime; this dependence should make these courts less likely to take a stand that empowers citizens to publicly exercise their dissent against the government.Conversely, an independent court has a base of power and support separate from that of the ruling regime, thereby enabling it to allow citizen expression of dissent against the government without fear of reprisal from the ruling regime against the court.This logic also applies to repressive policies, like censorship, that the regime may wish to impose.If a court is not independent, then it is unlikely to check the regime's attempt to impose the repressive policy.On the other hand, if a government imposes a repressive policy in a country with an independent court, the court can halt the policy, thereby respecting empowerment rights.
Together, these three explanations lead us to the following hypothesis: De Facto Judicial Independence Hypothesis: Increased judicial independence is positively correlated with increased state respect for human rights.
We are not the first to posit a connection between empowerment rights and judicial independence.Recent research, most notably Keith (2012), provides empirical support for this expectation.A potential limitation with this line of research, however, is that it relies on measures of de facto judicial independence that are indirectly or directly based on State Department reports.This is problematic because those reports are probably biased in favor of U.S. trade partners and military allies, among others (Crabtree and Fariss, 2015;Fariss, 2014b;Keith, 2012).As a result, the measure might not accurately capture changes in judicial independence over time and across countries.If this is true, measurement bias might be influencing empirical findings and, as a consequence, the inferences scholars make.
The release of a new latent measure of de facto judicial independence (Linzer and Staton, 2015) allows us to see if this is true.This measure draws on multiple sources in addition to the State Department reports and paints a different picture of judicial independence over time.While measures that use information from the State Department, such as Cingranelli, Richards and Clay (2015) or Keith (2012) indicators, show that judicial independence is decreasing, the Linzer and Staton (2015) measure shows that judicial independence is increasing.Figure 1

Model and Results
To reexamine the relationship between judicial independence and respect for empowerment rights, we draw upon a "standard" model specification in the human rights literature (Poe and Tate, 1994;Keith, 2002;Keith, Tate and Poe, 2009;Keith, 2012).5This model specification includes a lagged dependent variable along with independent variables that capture betweenstate and over time differences in regime type, socioeconomic conditions, human rights treaty ratification, and domestic and international threats (Keith, 2012, 68). 6The virtue of this model specification is its widespread usage and the large amount of work that justifies the theoretical concepts included in the model, allowing us both to draw clear comparisons with prior findings and to be confident in the quality of the model specification.Table 1 presents descriptions and descriptive statistics for the measures in this model. 7everaging the availability of new measures for key theoretical concepts, we make one key addition to this model and two slight modifications.To test our hypothesis that independent courts are associated with improved state respect for human rights, we add a lagged latent variable measure of de facto judicial independence (Linzer and Staton, 2015). 8The variable is bound between 0 − 1 and ranges from 0.012 − 0.995 in our data.The measure varies within countries over time.For example, the de facto judicial independence score for Brazil ranges over time from 0.359 − 0.638.The measure also captures cross-country changes in de facto judicial independence over time.The mean de facto judicial independence score is 0.427 in 1982 but increases to 0.532 by 2008.9 This variable improves upon previous measures in several ways.First, it addresses the fact that de facto judicial independence is an unobservable construct that can only be measured with some uncertainty.This is important because coders cannot be certain of the exact level of de facto judicial independence for one country-year relative to another.Second, previous measures of de facto judicial independence are typically based on only one data source (Cingranelli, Richards and Clay, 2015;Keith, 2012).Linzer and Staton (2015), in contrast, use a measurement model that incorporates data from twelve separate observable indicators (i.e.manifest variables) that are theoretically related to de facto judicial independence.This ensures that the estimates for the latent variable are not strongly biased by any one data source (Linzer and Staton, 2015).Third, the Linzer and Staton ( 2015) is continuous, while other measures are ordinal and typically bound between 0−2 (Cingranelli, Richards and Clay, 2015;Keith, 2012).The advantage of a continuous measure is that it allows us to estimate the effect of small changes, rather than large shifts, in de facto judicial independence.This is important if we believe that judicial independence changes slowly from year to year and are interested in how these changes influence state respect for empowerment rights.
In addition to including a measure of de facto judicial independence in the model, we also make two modifications to the standard specification.The first modification we make is to replace the dependent variable with a latent variable measure of empowerment rights (Schnakenberg and Fariss, 2013).Prior studies typically use the CIRI Empowerment Index (Richards, Gelleny and Sacko, 2001;Keith, 2012).This is an additive index that captures the extent to which states respect seven different but related rights: freedom of foreign movement, freedom of domestic movement, freedom of speech, freedom of assembly and association, workers' rights, freedom of religion, and electoral self-determination (Richards, Gelleny and Sacko, 2001).The degree to which states violate these seven rights is measured on a 0 − 2 scale, with lower values associated with higher levels of violation.The index then ranges from 0, which indicates that a state does not respect any of these rights, to 14, which indicates that a state respects all of these rights.One possible concern with this measure is that the CIRI guidelines provide coders with some degree of discretion, which could result in the miscategorization of some state estimates for some country-years.By using the latent variable, we can relax the assumption that state respect for empowerment rights has been measured precisely (Schnakenberg and Fariss, 2013).Indeed, just as with the Linzer and Staton (2015), the Schnakenberg and Fariss (2013) latent variable provides us with a means to directly account for uncertainty in measurement. 10  The second change we make to the model is to replace the measure of democracy most frequently used in models of state respect for human rights, Polity IV (Marshall, Jaggers and Gurr, 2010), with a latent measure of democracy, the Unified Democracy Scores (UDS) (Melton, Meserve and Pemstein, 2011).We do this for two reasons.One, there continues to be debate over which manifest variable of democracy best captures the underlying construct (Melton, Meserve and Pemstein, 2011).As a result, we do not have strong reasons to prefer one manifest indicator over another.In this circumstance, we think it better to use a latent measure, such as UDS, that draws upon multiple measures and averages over the potential biases of any one indicator.Second, many published empirical findings are not robust to the inclusion of alternative measures of democracy (Elkins, 2000;Casper and Tufis, 2003).
One reason that this might be the case is that some indicators possess countervailing biases.
This further suggests that we should prefer a measure that incorporates data from multiple indicators.While we present results with the latent measure of democracy, our results are robust to using other alternative indicators of regime type, including the Democracy-Dictatorship measure (Cheibub, Gandhi and Vreeland, 2010), the Polity measure (Keith, 2012;Keith, Tate and Poe, 2009), and the Autocratic Regimes measure (Geddes, Wright   and Frantz, 2014a). 11  Given that our dependent variable is continuous, we test our hypothesis using a regression model.The full model specification is shown in Eq. (1).
Empowerment Rights Index = β 0 + β 1 Empowerment Rights Index (lagged) We estimate this model using panel data for 177 countries from 1982-2008.Our data are clustered with multiple observations nested in each country.This means that we need to account for the fact that our observations are not necessarily independent of each other (Wooldridge, 2010).To do this, we estimate a multilevel regression model that includes country-level random effects (Gelman and Hill, 2007;Raudenbush and Bryk, 2002).We use a multilevel model because this approach has a number of advantages over other methods that are often used to analyze panel data, such as OLS with fixed effects or panel-corrected standard errors.These advantages include increased efficiency and more accurate standard errors (Shor et al., 2007).Our results, however, are robust to more traditional means of analyzing panel data, such as using ordinary least squares with classic and robust standard errors and including year-level random effects. 12e also need to account for uncertainty in the point estimates of the latent variables included in our model.Latent variables provide both a point estimate for each observation, which is the mean value of the posterior distribution, and a measure of uncertainty for these estimates, which is the standard deviation of the posterior distribution.We include this information in our model by following the recommendations of Schnakenberg and Fariss (2014) and Crabtree and Fariss (2015).Specifically, we duplicate our dataset 1,000 times and then assign a random draw from the posterior distribution of the latent variable to each country-year observation.We then use this new value as the measure.We preform this procedure for the dependent variable, the lagged dependent variable, the Linzer and Staton (2015) measure, and the Melton, Meserve and Pemstein (2011) measure.After that, we estimate a set of 1,000 random-effects models, saving and combining the results across the the multiple sets of data to create one set of coefficient and standard error estimates.This procedure is substantively important because it allows us to to relax the assumption that theoretically important variables are measured perfectly and without error (Mislevy, 1991;Schnakenberg and Fariss, 2014).Rubin (1987) developed the equation used to combine the estimates from each of the 1,000 models Rubin (1987).Crabtree and Fariss (2015), Mislevy (1991), and Schnakenberg and Fariss (2014) show how this approach should be used in relation to latent variable models.
According to the De Facto Judicial Independence Hypothesis, increased court independence should correlate with increased state respect for empowerment rights.Before we take our model to the data, we first look to see if the these concepts are positively correlated.
Figure 2 presents a bivariate plot of de facto judicial independence and state respect for empowerment rights.The dashed grey 45-degree line represents where we would expect the points to fall if there was a perfect linear relationship between the two variables.The black line represents the estimated slope from a bivariate regression. 13The line suggests a positive relationship between de facto judicial independence and state respect for empowerment rights.
Of course, both judicial independence and respect for empowerment rights are plausibly related to many other confounding factors.For example, factors such as whether or not a country faces an interstate threat or whether it is ruled by an authoritarian regime are likely related to changes in both state respect for empowerment rights and de facto judicial independence.In order to address possible confounders and to place this relationship in context, we estimate the model shown in Eq. ( 1).The results of this model are presented in Figure 3 and Table 2.As predicted, the model shows that state respect for human rights is significantly higher in states with independent courts.This is indicated by the positive and statistically significant coefficient on De Facto Judicial Independence.The effect of increased judicial independence is also substantively large.A change in de facto judicial independence from the 25th percentile to the 75th percentile value is associated with an approximately 0.258 increase in state respect for empowerment rights.This effect size is substantively significant.Increasing de facto judicial independence from the 25-75th percentile has a 35% larger effect that a change in democracy from the 25-75th percentile. 14These results suggest that increased judicial independence can have a meaningful effect on individuals' empowerment rights. 15  Our argument is that increased de facto judicial independence is positively correlated with increased state respect for empowerment rights.One way to examine this hypothesis is by seeing if de facto judicial independence positively correlates with the Empowerment Index.Another way to do this would be to see if de facto judicial independence positively correlates with the several measures that comprise the Empowerment Index.We estimate seven additional models, replacing the latent Empowerment Rights variable we use in Model 1 with one of the Cingranelli, Richards and Clay (2015) indicators that comprise the Empowerment Index.This allows us to see if the relationship is driven by a particularly strong relationship between de facto judicial independence and one or more of the individual rights in the index measure.As a reminder, the individual indicators, described in Table 3, are coded from 0−2, with higher values associated with increased state respect for empowerment rights.
The individual measures are ordered, so we estimate ordered logit models with country-14 ( 0.2577304 0.1910949 ) • 100% = 134.8704% 15We find similar results when we use the Cingranelli, Richards and Clay (2015) Empowerment Index.
Appendix A contains those results.Our results also hold when we account for the extent to which states protect empowerment rights in their constitutions (Keith, 2012).Table 7 presents the results of our model with Keith's (2012) 'four freedoms' variable.This measure accounts for the extent to which states protect the freedoms of speech, association, assembly, and religion in their constitutions (Keith, 2012).
The idea here is that since constitutional protections of these rights could increase both de facto judicial independence and state respect for empowerment rights, we ought to control for that in our empirical analysis.Appendix B contains these results.level random effects.Since these models include independent variables that are latent measures, we estimate a 1,000 of these models, and then save and combine the results of the models as described above. 16To ease model estimation, we set the thresholds parameters to be equidistant.Figure 4 and Table 4 display the results of these models.Both present odds ratios for the variables in our model and 95% confidence intervals.If de facto judicial independence is positively correlated with state respect for empowerment rights and statistically significant, we would expect the odds ratio for the measure to be greater than 1 and the bounds for the 95% confidence intervals to only include values greater than 1. Figure 4 shows that this is the case across all models.Indeed, the size of the estimated effect is substantial, indicating that a full one-unit change in de facto judicial independence is associated with a dramatic increase in the odds that a state respects empowerment rights.Table 4 highlights the relative importance of this relationship.In each model, the estimated odds ratio for de facto judicial independence indicates that an independent judiciary is one of the most important correlates of improved state respect for individual empowerment rights.

Robustness Checks
In order to ensure that these results are robust, we re-examine the relationship between de facto judicial independence and state respect for human rights using different modeling strategies.First, we investigate whether our findings are determined by the cases we include in our data.We employ k -fold cross validation to guard against such overfitting (Efron and Gong, 1983;Hill Jr and Jones, 2014;Ward, Greenhill and Bakke, 2010).Specifically, we conduct 1,000 simulations, within each we randomly partition our data into a training set and nine test sets (k = 10) and then estimate a series of linear models with the empowerment index as the dependent variable.The baseline model includes only the lagged dependent variable on the right-hand side, while the other models contain either one or more independent variables.Figure 5  Second, we check whether our findings are dependent on parametric assumptions, such as that the relationship between de facto judicial independence and state respect for human rights is linear or even smooth.In order to account for non-linearities, interactions, and other functional form possibilities, we specify a series of random forest models (Hill Jr. and Jones, 2013; Jones and Linder, 2015).In each model, we use the independent variables from our regression model to predict state respect for one of the empowerment rights measures used in Models 1-8.Since these models contain latent variable measures, we estimate a 1,000 random forests, saving and combining the goodness-of-fit statistics provided by these models as described above.Table 5 presents the results.Each cell in the table contains the mean permutation importance value for an independent variable as calculated from the results of the 1,000 random forest models.This value captures the mean decrease in classification accuracy caused by permutating the values of an independent variable.The intuition is that if an independent variable is not an important predictor, then randomly changing the values of that variable will not decrease prediction accuracy.Variables that have higher importance values then are stronger predictors.Across all models, the importance values for de facto judicial independence are relatively high.In fact, the random forest results suggest that the lagged de facto judicial independence measure is consistently an important predictor of state respect for empowerment rights.

Discussion
A growing empirical literature suggests that independent courts can increase state respect for empowerment rights.Unfortunately, measurement issues prevent scholars from assessing the validity of past results.In this paper, we have used new measures to re-examine this relationship.We find strong evidence that de facto judicial independence is positively correlated with empowerment rights.This empirical finding is robust to a wide-range of model specifications and estimators.
Our results have important policy implications.Many of the specific rights included under the broader umbrella of empowerment rights have been the subject of intense scholarly interest in recent years, with scholars investigating the extent to which regimes engage in censorship (repression of freedom of speech) (Cain, 2013;Charles Crabtree and Kern, 2015; Gehlbach and Sonin, 2014;Howard, 2010;Kalathil and Boas, 2010;King et al., 2013;King, Pan and Roberts, 2014;Lorentzen, 2014;Shadmehr and Bernhardt, 2015), violations of voting rights (Gandhi, 2008;Gandhi and Lust-Okar, 2009;Levitsky and Way, 2002;Przeworski, Stokes and Manin, 1999;Stokes et al., 2013;Alvarez, Hall and Hyde, 2009;Beaulieu and Hyde, 2009;Beber and Scacco, 2012;Simpser, 2008), limitations on the ability to worship freely (Toft, Philpott and Shah, 2011;Vala and O'Brien, 2007), and the freedom to assemble for political or non-political purposes (Davenport, 2014;Earl, Soule and McCarthy, 2003;King et al., 2013;Shadmehr and Bernhardt, 2011).Our results suggest that one of the strongest mechanisms to limit political repression comes through the judiciary.By empowering independent courts, states can limit the extent to which their citizens' rights are curtailed.
Our results also suggest that more democratic states tend to respect empowerment rights.
Indeed, the regression results provide strong evidence that increases in de facto judicial independence and democracy are strongly correlated with increased empowerment rights.
The results from our predictive models provide additional evidence that these two domestic factors are strongly related to increased empowerment rights.
Our results then shed some light on the relative force of domestic and international factors to influence the ability of citizens to exercise their rights.In particular, our results suggest that domestic, rather than international, factors play an important role in moderating state respect for empowerment rights.Thus, those wishing to improve the state of empowerment rights in a country might be better served seeking to strengthen domestic institutions within a country than through other means.This should be good news.While it is difficult to influence the development of domestic institutions and behavior, it is often far easier than trying to shape international forces.
illustrates this relationship.In Panel (a) it plots the trimmed country-year means for the(Keith, 2012) measure across the years 1980-2008. 3These values approximately represent the extent of judicial independence across the model for a given year.Panel (b) plots the trimmed country-year means from the Linzer and Staton (2015) measure.Higher values for both measures are associated with higher levels of de facto judicial independence.By comparing the patterns of these values across panels, we see that the two measures capture different dynamics over time. 4This underscores the potential limits of State Department-based measures.It also suggests that results from models that use a measure primarily based on State Department reports might not hold when that measure is replaced with the improved Linzer and Staton (2015) measure.

Figure 1 :
Figure 1: Mean Values of the State Respect for Empowerment Rights and De Facto Judicial Independence Over Time

Figure 2 :
Figure 2: Bivariate Plot of State Respect for Empowerment Rights and De Facto Judicial Independence Across Countries

Figure 3 :
Figure 3: State Respect for Empowerment Rights Across Countries

Figure 4 :
Figure 4: State Respect for Individual Empowerment Rights Across Countries presents the results of the 10-fold cross validation.It plots the average percent reduction in mean square error of various model specifications compared to the model with just the lagged dependent variable.This shows the additional predictive power of individual variables or combinations of variables.The figure illustrates that adding de facto judicial independence to the model dramatically improves its predictive ability.The amount of error reduced by including de facto judicial independence is roughly the same as the amount of error reduced by including the three indicators of regime type (i.e. the Democracy, Military Regime, and Monarchy variables).More importantly, the amount of error reduced by including de facto judicial independence in the model is greater than the error reduced by including variables related to conflict onset (i.e.Interstate Conflict Intensity and Civil War Intensity) or socioeconomic differences (GDP Per Capita (logged), GDP Growth (logged), Population (logged), Population Density (logged)) in the model.

Note:Figure 7
Figure 7 plots the mean country-year values of our dependent variable and primary independent variable over time.Panel (a) presents the mean country-years values of the Empowerment Rights Index measure over time.Panel (b) presents the mean country-year values of the de facto judicial independence measure over time.While judicial independence appears to be increasing over time for the countries in our sample, state respect for empowerment rights does not.

Figure 7 :
Figure 7: Mean Values of the State Respect for Empowerment Rights and De Facto Judicial Independence Measures Over Time 10 Appendix D contains a plot of the mean country-year values of the Empowerment Rights Index measure over time. 11We present the results of these models in Appendix C. + β 2 De Facto Judicial Independence (lagged) + β 3 Democracy + β 4 Military Regime + β 5 Monarchy + β 6 GDP Per Capita (logged) + β 7 GDP Growth (logged) + β 8 Population (logged) + β 9 Population Density (logged) + β 9 ICCPR Ratification + β 10 Interstate Conflict Intensity + β 11 Civil War Intensity + ,

Table 1 :
Variable Descriptions and Descriptive Statistics Descriptive statistics calculated based on the 3797 country-year observations in our sample.

Table 2 :
State Respect for Empowerment Rights Across Countries (1982-2008) Standard errors are shown in parentheses.Data come from 3797 country-year observations from 1982 to 2008.The dependent variable is Empowerment Rights Index.See text for more information about the model and data. Note:

Table 3 :
CIRI Variable Descriptions and Descriptive StatisticsNote: All definitions taken directly from the Cingranelli, Richards and Clay (2015) codebook.Descriptive statistics calculated based on the 3797 country-year observations in our sample.

Table 5 :
State Respect for Empowerment Rights Across Countries (1982-2008) -Random Forest Results Cells for each variable contain the mean permutation importance value from a series of 1,000 random forest models.The random forests were estimated on a data set that include 3797 country-year observations from 1982 to 2008.The dependent variable for each model is Empowerment Rights Index.Each random forest model corresponds to a multilevel model presented in Table 2 or Table 4. Model 9 corresponds to Model 1, Model 10 to Model 2, Model 11 to Model 3, Model 12 to Model 4, Model 14 to Model 5, Model 14 to Model 6, Model 15 to Model 7, and Model 16 to Model 8. See text for more information about these models and data.