**6. Findings**

*6.1. Building Variables Representing Two Dimensions of Religious Restrictions*

The factor scores are derived from the governmen<sup>t</sup> religious restriction index (NX) from the RAS3 dataset. The NX index is about governmen<sup>t</sup> restrictions on all religions and majority religions, and it comprehensively covers governmen<sup>t</sup> religious restrictions in various dimensions.

To discover the latent dimensions of religious restrictions, I run exploratory factor analysis on the NX index. I performed orthogonal rotation when I generated the factors, so the two factors have a minimal correlation. The correlation between two factors is 0.09 (*p* < 0.05). Based on an examination of both the eigenvalues and the questions, I decided to limit the number of factors to two. Based on two different kinds of rotations of the factor analysis, the patterns of the two factors seem quite stable. As shown in Table 2, the first factor has high loadings mostly on items focusing on whether any level of governmen<sup>t</sup> imposes limitations on religious organizations, such as churches, schools and associations. For example, items regarding governmen<sup>t</sup> restrictions on religious parties, governmen<sup>t</sup> restrictions on religious members and organizations not sponsored by the states, and government restrictions on religious clergy and/or organizations engaging in public political speech all have high loadings on factor 1. I labeled this factor as "institutional restrictions". Countries, such as Egypt, Iraq, Saudi Arabia and Syria have noticeably high scores on this factor, which indicates that Middle Eastern countries tend to impose higher religious restrictions on organizations than other countries. This could be explained by application of Islamic law in such countries, which puts restrictions on religious groups, but also how theocratic regimes control religions in general.

As shown in Table 2, the second factor has high loadings on items concerned with the governmen<sup>t</sup> restrictions on individuals' expression and practice of their religions, such as public observance of religious practices, public religious speech, and access to places of worship. For instance, items regarding restrictions on the public observance of religious practices, restrictions on access to places of worship, whether people are arrested for religious activities are all have high loadings on factor 2. Thus, the second factor is focused on the religious restrictions on individual practice of religion and is labeled as "individual restrictions". Countries, such as China, North Korea, Cuba, Turkmenistan, Uzbekistan and Vietnam scored high on this factor in comparison to the rest of the countries, which implies that these countries impose higher levels of restrictions on individual religious practice. This could be explained by the legacy of the totalitarian regimes from the post-communist states that controlled all aspects of social life.

As the indexes load upon two dimensions–institutional and individual religious restrictions, they can be categorized into these two groups. I followed Grim and Finke (2006)'s methods to generate the indexes for the two dependent variables. As the indexes generated from factor score and raw score additive approach are highly correlated, I will only present the latter results. The simple additive methods have the advantage of being simple and parsimonious. In order to keep the index to the same scale as the single index to facilitate the interpretation of the statistical models, I use the following equation to calculate the indexes for institutional religious restrictions and individual religious restrictions:

Institutional/Individual Religious Restrictions = sum of raw scores of indicators /number of indicators


**Table 2.** The Factor Analysis Loadings for Government Religious Restrictions (NX) Index.


**Table 2.** *Cont.*

#### *6.2. The Effect of HROs on the Aggregate Religious Restrictions*

Table 2 presents the results of my hypotheses tests using the aggregate index of religious restrictions. The first model only controls the variables in general human rights studies, and the second model additionally controls some religious variables including the percentage of Christians and Muslims, religious homogeneity and religious establishment.

HROs are helpful in the protection of aggregate religious restrictions, which includes all indicators of religious restrictions. As shown in Table 3, HROs are related to a better protection of religious rights for organizations as indicated by model 1 and 2, which is consistent with the Hypothesis 1. Substantively, one human rights organization in a country can decrease 0.04% of one standard deviation of religious restrictions; an increase on the number of human rights organizations from 1 to 139 can lead to the decrease of around 55.6% of one standard deviation of religious restrictions on average in model 1. The size of effect reduces slightly after the percentages of Christians and Muslims, religious homogeneity and establishment are held at constant in model 2. Around 139 HROs can decrease the restriction of religious organization by almost 46.3% of one standard deviation when these religious variables are held constant. Therefore, HROs are effective in pressuring states to change their policy and practice in religious rights.



 *p* < 0.01; *p* < 0.05; *p* < 0.1.

#### *6.3. The Effect of HROs on the Institutional and Individual Religious Restrictions*

Table 4 presents the results of my hypotheses tests using the institutional and individual religious restrictions variables. The first and second columns are the results for institutional religious restrictions and the third and fourth columns are the results for individual religious restrictions. Models 3 and 5 only control the variables in general human rights studies, while models 4 and 6 additionally control some religious variables including the percentages of Christians and Muslims, religious homogeneity and religious establishment.


**Table 4.** The Effect of Domestic Presence of HROs on Institutional and Individual Religious Restrictions.

> \*\*\* *p* < 0.01; \*\* *p* < 0.05; \* *p* < 0.1.

The domestic presence of HROs is effective in protecting both the institutional and individual religious restrictions, which is consistent with Hypotheses 2 and 3. For religious institutions, one human rights organization in a country can decrease 0.1% of one scale of institutional religious restriction index, which is 0.25% of one standard deviation; an increase on the number of human rights organization from 1 to 139 can lead to the decrease of 13.9% of one scale of institutional religious restriction, which is around 34.75% of one standard deviation of institutional religious restrictions on average. After holding constant the percentages of Christian and Muslim, the level of religious homogeneity and religious establishment status, the effect size remains the same.

HROs can also mitigate the oppression from the governmen<sup>t</sup> for religious individuals. According to models 5 and 6, one human rights organization in a country can reduce 0.1% of one scale of individual religious restriction index, which is 0.5% of one standard deviation; an increase on the number of human rights organizations from 1 to 139 can lead to the decrease of 13.9% of one scale of individual religious restriction, which is around 69.5% of one standard deviation of individual religious restrictions. The effect size also remains the same, after holding the percentages of Christian and Muslim, the level of religious homogeneity, and religious establishment status constant.

Thus, HROs can reduce both institutional and individual religious restrictions, even though the underlying mechanisms might be different, based on the distinct nature of these two dimensions of human rights. Religious institutions are visible, thus HROs provide support and resources to religious institutions and mediate between them and the governmen<sup>t</sup> to reduce the repercussions from political disruptions in a direct way. Religious individual restrictions are covert, HROs can also transmit information they gather on religious individual restrictions to overseas and pressure states from the third parties such as UN and some western democracies. HROs can also provide knowledge and resources to individuals and mobilize them to fight for their rights.

#### *6.4. Other Variables Explaining Variation in Religious Institutional and Individual Restrictions*

Even though the control variables are not the focus of this study, it is necessary to explore how they are related to the different levels of religious institutional and individual restrictions. Based on models 3–6, those control variables are associated with institutional and individual restrictions differently. Therefore, they support argumen<sup>t</sup> that institutional and individual religious restrictions are inherently different, and the distinctions between the two dimensions of human rights should be further examined in the future studies.

Some factors have a significant effect on the institutional religious restrictions but not on individual religious restrictions. The size of GDP in a country is negatively related to the institutional religious restrictions. It is consistent with literature that the bigger the size of the economy is, the lower the religious restrictions are (Mitchell and McCormick 1988; Keith et al. 2009; Grim and Finke 2010). The models also indicate that stable regimes have a lower level of institutional religious restrictions (Grim and Finke 2010). However, both variables are not related to the individual religious restrictions. The models also show that countries with an established religion or a homogenous religious scene have a higher level of restrictions on institutions, because governments are more likely to adopt favoritism towards the established or majority religion in a society in exchange for their support. Restrictions on religious institutions also tend to be higher when the countries have a bigger population. Literature in human rights indicates that there is a higher chance for the occurrence of human rights violations when the population is big (Poe and Tate 1994; Poe et al. 1999; Keith et al. 2009; Richards et al. 2001). It is also true that coders are more likely to find restrictions in a more populous country, where the number of religious people and institutions are also bigger. Interestingly, population size is not related to the individual religious restrictions.

Some other factors are associated with a higher level of religious restrictions for individuals but not for institutions. When there are internal wars, individuals in a society are more likely to be oppressed. Therefore, human rights, including religious freedom, are more likely to be violated. Even though the polity score measure of democracy is not associated with individual religious restrictions, judiciary independence is. Surprisingly, when the judiciary branch becomes more independent, the level of individual religious freedom is actually lower. In addition, economic growth rate is positively related to the level of individual religious restrictions. In other words, individual religious restrictions tend to be higher in countries where the economic growth rate is high. Poe and Tate (1994) explain that rapid economic growth also can increase instability which induce the states to resort to coercion. The only variable seems to be able to reduce the individual religious restrictions is the HROs. Therefore, the study on HROs is of grea<sup>t</sup> significance in the study of individual religious restrictions.

#### **7. Conclusions and Discussion**

This study finds that religious restrictions have two dimensions: institutional and individual restrictions with an exploratory factor analysis and provides evidence that these two dimensions of religious restrictions should be treated separately in future studies.

This study also examines the effects of HROs on the protection of religious freedom and its two dimensions over time. The mere presence of HROs within a country not only reduces overall religious restrictions; that presence also alleviates both institutional and individual religious restrictions in the 14 years covered. Based on the distinct features of

religious institutions versus individual adherents, restrictions on the former are more overt while more covert on the latter. Therefore, the approach HROs adopt might be more direct for institutions and more indirect for individuals. Domestic HROs play an important role in working with religious institutions to pressure states to change their policies and practices in religious freedom. HROs channel resources to religious institutions to engage in legal or social movement activism for religious freedom. Protection of individual religious freedom might entail a bottom-up process, in which domestic institutions transmit information to HROs on repression or persecution of individuals, which then can name and shame violating states on the international stage. In this way domestic institutions, working in concert with international NGOs and sympathetic states, can help empower and mobilize individuals to fight powerful states to protect their rights through litigation or social movement activism.

This study has grea<sup>t</sup> theoretical implications. It expands the current literature on HROs and human rights by focusing on religious rights and their two dimensions. The findings indicate that the domestic presence of HROs is effective in the religious rights protection, which increases with their number and density, a finding consistent with most current literature on other kinds of rights protection. A more textured finding is that HROs are critically important both in the protection of institutional and individual dimensions of religious freedom, but through different underlying mechanisms. HROs employ top-down and bottom-up processes, using direct and indirect approaches for religious institutions and individuals, respectively. This indicates the need to differentiate individual and institutional rights in international human rights studies, both to advance more novel theories and develop more practical policies in human rights protection.

This study also has a grea<sup>t</sup> implication for policies to advance international human rights protection. The domestic presence of HROs is useful in protecting the rights of religious institutions and individuals. Among all the variables tested in this study, HROs are the only variable effective addressing religious individual restrictions. Thus, we need to acknowledge that HROs are improving the protection of human rights on the grassroots level, despite of the lack of enforcement in international human rights regimes. As moral agents, transnational advocacy networks play an important role to pressure states to change how they treat their citizens with the vital information and tangible resources. HROs and other international actors need to strengthen existing strategies and develop new ones that hold states accountable for violations.

There are also a few limitations in this study which can be addressed by future studies. The first lies in the limitation of the data, which only covers 1990 to 2003 (given the available data on HROs). The results would have stronger internal validity if the years covered in this study are longer. Future studies can collect more data with longer time span and test the same theories in this study. Second, measurement on HROs can be improved. As the study done by (Murdie and Peksen 2015), the HROs with a specific focus on certain kind(s) of human rights can be more effective in combating the corresponding human rights violations. A way can improve this study is to collect the data on the number of religious HROs on religious freedom across years and countries. The relationships found in this study might be strengthened with these new measures and data.

**Funding:** This research received no external funding.

**Data Availability Statement:** The data presented in this study are openly available in Harvard Dataverse at https://doi.org/10.7910/DVN/1LHW5S, V1.

**Acknowledgments:** My thanks to Allen Hertzke, Roger Finke, Colin Barry, *Religions* editors, and anonymous reviewers for their helpful comments and suggestions on previous iterations of this article.

**Conflicts of Interest:** The author declares no conflict of interest.
