Regime change and religious discrimination after the Arab uprisings

This article investigates how and when regime transitions intensify minority discrimination through an analysis of two types of religious persecution following the Arab uprisings. We argue that weakened institutions and the prevalence of religious outbidding during political transitions make societal-based religious discrimination (SRD) more likely to increase than government-based religious discrimination (GRD). This is because social divisions are often exacerbated and social unrest difficult to contain, while at the same time, policy change can be difficult to enact and enforce. We test these claims through a mixed-methods research design. Employing a synthetic control method, the cross-national, quantitative analysis from 1990 to 2014 confirms that GRD has not changed since the Arab uprisings, while SRD has substantially increased in those countries (i.e. Egypt, Libya, and Tunisia) that also experienced regime change. A case study of Egypt provides more direct evidence of the institutional and outbidding mechanisms. The qualitative analysis draws on ethnographic research conducted in Cairo during 2014, which includes in-depth interviews with Coptic Orthodox Christians. Our findings underscore the twin challenge of protecting and accommodating minority religions during periods of political transition.

How and when do regime transitions intensify minority, especially religious, discrimination? A growing body of scholarship acknowledges the challenges faced by minority communities in the interval between regimes -warning that political transitions, particularly the early stages of democratization, are prone to heightened religious and ethnic tensions, intercommunal conflict, and state repression (Mansfield & Snyder, 1995;Snyder, 2000;Hegre et al., 2001;Christensen, Nguyen & Sexton, 2019;Costalli & Moro, 2019). Additional research stresses that minority religious and ethnic groups are often the targets of the harassment, violence, and exclusionary state policies that emerge when a regime is replaced (Anderson, 2003;Lawoti, 2008;Zabad, 2017).
What we still know less about, however, are the channels through which minority discrimination occurs during political transitions. Are government restrictions and regulations the primary means of persecution? To what extent do majority group members who are not representatives of the state also persecute minorities? And what conditions either bolster or constrain these different kinds of minority discrimination?
This article advances our understanding of the propensity for and types of minority discrimination during regime transitions through an analysis of religious persecution following the Arab uprisings. In line with Geddes, Wright & Frantz (2014), we define a regime transition as a shift in the set of basic formal and informal rules that determine who influences the choice of leaders and, consequently, policies. Our concentration on religious discrimination allows us to leverage the growth in quantitative data on global religious repression in order to develop and test hypotheses about a key source of heterogeneity in minority discrimination: governmentbased religious discrimination (GRD) and societal-based religious discrimination (SRD) (see Fox, 2020). These two types of discrimination refer to restrictions on the institutions and practices of, along with harassment and violence towards, minority religions by the state or members of society who are not representatives of the government, respectively (Fox, 2017). 1 As such, this article makes theoretical and empirical contributions to the broader scholarship on political transitions and minority discrimination. Theoretically, we distinguish two key mechanisms that account for an increase in SRD, but not necessarily GRD, during periods of regime change. A rich literature explores these two types of discrimination, but there is still no consensus as to the causal pathways that lead to one or another (Fox, Finke & Eisenstein, 2019;Henne & Klocek, 2019;Fox, 2020;Sumaktoyo, 2020). We argue that weak or weakened institutions and the prevalence of religious outbidding make SRD more likely than GRD during regime transitions because social divisions become exacerbated and social unrest difficult to contain, while at the same time, policy change is challenging to enact and enforce.
Our empirical contributions are two-fold. First, we leverage quantitative data from round 3 of the Religion and State (RAS3) project to test the general relationship between regime transitions triggered by the Arab uprisings and the two types of religious discrimination. 2 Several studies have noted a rise in sectarian violence (Tadros, 2013;Abdo, 2017) and increased persecution of minority religions (Sarkissian, Fox & Akbaba, 2011;Fox, 2013) in the Middle East and North Africa (MENA), but scholars have not systematically examined what impact, if any, the Arab uprisings have had on these trends -save for one exception (Akbaba & Fox, 2019). What we find is that SRD, but not GRD, increased substantially in the four countries (i.e. Egypt, Libya, Tunisia, and Yemen) that experienced regime transitions as a consequence of the Arab uprisings compared to the other countries in the RAS3 dataset. 3 Additionally, we provide a plausibility probe of the institutional and outbidding mechanisms through a case study of Egypt. This analysis draws on ethnographic research by the second author, which includes in-depth interviews with Coptic Orthodox Christians in Cairo between January and June 2014. Even though this community experienced relatively more peaceful interfaith relations before and after the Arab uprisings than other parts of the country, we still find evidence that SRD has escalated there since 2011 and, at least in part, for the reasons we propose.

Previous research on regime transitions and minority discrimination
The consequences of regime change for minority populations has emerged as an increasingly important area of study (see Bertrand & Haklai, 2013). While many academics and policymakers in the early 1990s shared the conviction that liberalization and democracy promotion could engender stable state-society relations, more recent scholarship underscores that the interval between political regimes can also be a harrowing time for religious and other minority groups as nationalist politics, scapegoating, and intolerant social attitudes often exacerbate pre-existing societal tensions (Mansfield & Snyder, 1995;Snyder, 2000;McCarthy & Menager, 2017). Two of the factors most often cited in the literature for explaining whether regime change leads to stable or 1 We acknowledge that GRD and SRD are likely more intertwined in practice than in theory. It is challenging, for example, to always know whether societal actors who vandalize a minority house of worship do so on their own or in coordination with government actors. We nonetheless aim to disentangle the two types of discrimination by focusing on the direct perpetrators as the key distinction (Fox, 2020;Grim & Finke, 2011). Operationalized in this way, GRD refers to instances where the government directly discriminates against minorities, whereas SRD refers to instances where discrimination is committed by nongovernmental societal groups.
2 These data cover 183 countries and independent territories from 1990 to 2014. 3 We also compare Arab uprising countries that did and did not experience regime change as a robustness check. In line with our broader findings, societal discrimination of minority religions increased substantially in the former, but not the latter. See the Online Appendix for these results. conflictual relations between the state, minority groups, and society at large are institutional frameworks and political outbidding (see Costalli & Moro, 2019). They point to a general expectation for when minority discrimination is likely to increase but say considerably less about which type(s).

Institutional frameworks
State institutions are one of the most commonly studied factors to explain the impact of political transitions on relations between the state, majority groups, and minority communities. In general, well-established procedures that guide the behavior of political actors buttress against minority discrimination by constraining the range of acceptable actions and alleviating uncertainty (Mansfield & Snyder, 1995Snyder & Mahoney, 1999). These formal and informal rules also provide opportunities for previously marginalized groups to voice their grievances through routine political activities (Kohli, 1997;Snyder, 2000).
Weak domestic institutions during regime transitions, in contrast, may undermine minority relations (Rotberg, 2004). They can deepen political instability, as well as limit a government's ability to check social violence and enforce the rule of law (Gates et al., 2006;Goldstone et al., 2010). This can further lead to surges in criminal and political violence as armed groups exploit the 'security gap' to advance their own agenda (Eizenstat, Porter & Weinstein, 2005). Frail political institutions undermine public trust since this impairs the government's ability to provide public goods, and also may lead to increased corruption among civil service and security forces (see Woodward, 2017). And the resultant mistrust can reduce the likelihood that ordinary citizens will comply with the law (see Levi & Stoker, 2000).

Political outbidding
A second mechanism often emphasized for understanding the impact of regime change on minority discrimination is political outbidding -or attempts by elites to advance and/or consolidate their authority by enhancing their nationalist, ethnic, or religious credentials with key domestic political audiences (Snyder, 2000;Toft, 2007). The process of outbidding has chiefly been applied to ethnic politics and conflict (Kaufman, 1996;Chandra, 2005;Vogt, Gleditsch & Cederman, 2021). More and more scholars of religion and politics, though, have also begun to investigate this strategy -noting that regime transitions can be especially ripe moments for leaders to mobilize along religious lines (Toft, 2013;Buckley & Wilcox, 2017). Toft (2007), for instance, details a succession of leaders who tendered competing religious bids in Sudan following the 1969 military coup when confronted with internal opposition and rebels in the south. Others demonstrate analogous processes in the Chechen Republic (Whitmeyer, 2015), Iran (Tabaar, 2018), and Indonesia (Sumaktoyo, 2020).
Religious and other forms of outbidding provide a number of benefits for elites (e.g. voter support and resources), but at a cost to minority groups. The process, for instance, necessarily favors one group at the expense of another, which can intensify exclusionary attitudes and reify ideational boundaries between groups (Saideman et al., 2002). It can also increase violence as such acts provide costly signals to demonstrate a group's credibility and capabilities (Farrell, 2020). Furthermore, once elites become beholden to a particular segment of the population, they have a strong incentive to continue to appeal to that group's interests (Horowitz, 1992;Goddard, 2006). This limits the state's ability to function dispassionately, such as in Sri Lanka where competition among Sinhalese political parties has led to the tacit acceptance of Buddhist nationalist organizations by some government officials (DeVotta, 2005).

Explaining types of religious discrimination during regime transitions
Past scholarship suggests minority, including religious, discrimination will increase during regime transitions when institutions are weakened and elites can gain an advantage in the new political environment by emphasizing social cleavages. This same body of work is relatively silent, though, on which type(s) of minority discrimination we should expect to intensify under these conditions. We argue that regime transitions characterized by weak institutions and religious outbidding are more likely to fuel SRD than GRD for three main reasons.
First, social divisions often become exacerbated and lead to violence under these conditions. One reason for this is that weak states are more easily captured by religious or ethnic elites and their constituencies and, subsequently, political, economic, and social resources are unlikely to be evenly distributed (Cederman, Weidmann & Gleditsch, 2011;Rudolfsen, 2017). This can produce an environment in which social identities of excluded groups become politicized and antagonistic intergroup relations intensify (Buhaug et al., 2011). Judicial institutions, even if not wholly replaced, are also unlikely to challenge the government when political power is concentrated in the hands of a small group of elites (Iaryczower, Spiller & Tommasi, 2002;Vondoepp, 2006). Under these circumstances, members of the majority may feel empowered to harass minority religions without the fear of being held accountable for their actions (Magnusson, 2001;Buhaug et al., 2011;Tajima, 2013).
Another explanation for heightened intergroup tensions and violence is the increasingly competitive environment of transitional societies, which makes religious outbidding an attractive strategy. Elites may leverage religious credentials in order to help capture the state, in the first place, and/or to consolidate and maintain their legitimacy. This is particularly likely when elites face an immediate threat from internal or external opposition (Toft, 2007;De Juan & Hasenclever, 2015). In other contexts where the political system is more open, rising counter-elites may find themselves almost forced to rely on religious outbidding to build a winning coalition (Snyder & Ballentine, 1996). For example, new elites in Serbia demonstrated little interest in religious nationalism until rising pressure from mass political participation created an incentive to do so (Silber & Little, 1997). Thus, even if the majority population or members of the majority do not share hostilities towards minority religions at the start of a transition, religious outbidding can increase the saliency of group boundaries and threat perceptions towards minority groups as the regime change unfolds and, ultimately, intensify SRD.
Second, and related, regime transitions characterized by weak institutions and religious outbidding are often unable to contain social unrest when it breaks out. One reason for this is the enfeebled state security forces and an ineffective rule of law characteristic of many weak states, as discussed above (Gates et al., 2006;Goldstone et al., 2010). Religious outbidding can also descend into cycles of increasingly more hardline positions that fuel intergroup mistrust, suspicion, and eventually incite violence (Gurr, 1970;Basedau, Pfeiffer & Vüllers, 2016;Basedau et al., 2017;Rudolfsen, 2017). When this is the case, political elites will have little incentive to moderate their positions or advocate for reconciliation since it would cost them the support needed to maintain and/or compete for power.
Third, government policy is likely to change less quickly than public behavior during transitions (Grim & Finke, 2011;Finke & Martin, 2014;Akbaba & Fox, 2019). Government reform is especially difficult to enact, let alone enforce, when political, judicial, and security institutions are weakened. Repressive state policies towards minorities, in particular, are challenging because they require both personnel and funds not always available to transitional regimes (Moore, 2000;Shellman, 2006;Davenport, 2007). Moreover, new regimes may be reluctant to enact legal restrictions against minority religions because these regulations can damage diplomatic ties and lead to potential sanctions, which they can ill afford to incur before they have consolidated their authority (Hathaway, 2002;Hafner-Burton, 2005;Sarkissian, 2015). Even if the elites that take power are hostile to religious minorities, therefore, they may have difficulty using their new authority to repress these communities. Moreover, the regime may not remain in power long enough to enact the longerterm policy changes they desire given the uncertainty and instability of many transitions, such as with the Muslim Brotherhood in Egypt.

Why the Arab uprisings?
We examine our argument for why SRD is more likely to increase than GRD during regime transitions in the context of the Arab uprisings. 4 These are a particularly useful set of cases for four primary reasons. The first is because they represent a broad universe of cases of interest to scholars of transitional societies. Earlier studies of regime change primarily focused on shifts from authoritarian to democratic rule, but more recent scholarship acknowledges that stalled democratic transitions that involve transitions to a new autocratic regime have been the predominant form of regime change over the past 75 years (Geddes, Wright & Frantz, 2014). Understanding how regime transitions following the Arab uprisings influenced types of minority discrimination can, therefore, have implications for a large number of transitions since the end of World War II.
A second reason for testing our argument in the context of the Arab uprisings is that the regime transitions triggered by those events are largely characterized by the institutional and outbidding mechanisms we identify as important. These cases, thus, provide the opportunity to directly observe whether or not the causal logic we outline accounts for the purported linkage between regime transitions and increased minority discrimination, especially through within-case analysis (see Goertz, 2017). For instance, Egypt and Tunisia possessed relatively stable administrative institutions prior to the uprisings. However, the rule of law, freedom of the press, and other state institutions weakened considerably after each regime collapsed (Mansfield & Snyder, 2012;Landolt & Kubicek, 2014;Yardımcı-Geyikçi & Tür, 2018). Public trust has also declined precipitously, with as many as 91% of young people in Tunisia reporting they do not trust parliament (Yahya, 2016). The picture is not as stark in Egypt, but confidence in political institutions has also fallen there in recent years (see Spierings, 2017).
Political actors in the regime change countries have also relied on religious outbidding to varying degrees following the Arab uprisings. In Egypt, a proliferation of new Salafist parties emerged, including the al-Nour (light), the al-Asālah (authenticity), and the al-binā'wa al-tanmīyya (building and development) parties. Each adopted increasingly more hardline religious positions in order to differentiate themselves from both secular parties and the more established Muslim Brotherhood's Freedom and Justice Party (FJP). A similar process unfolded in Tunisia, with new Islamist parties including the Ennahda Party, Jibhat al-Islah (the Reform Front), Hizb al-Asala (the Authenticity Party), and Hizb al-Rahma (the Mercy Party). Ennhada has maintained a relatively moderate position, but still made religious appeals to conservative voters (Churchill, 2011). Other parties, comprised primarily of Salafi youth and ultraconservative Islamists, went further, adopting increasingly more hardline views since the removal of Zine El Abidine Ben Ali (Lynch, 2011;Matesan, 2012;Marks, 2013). In a similar fashion, Islamist parties in Libya multiplied after Qaddafi's fall, chief among them the Justice and Construction Party (JCP). The JCP and others have also gradually leveraged religious politicsincluding gender segregation and compulsory wearing of the hijab in public universities -in order to gain an advantage over tribal leaders and other factions vying for power (Hamid, 2014;Sawani, 2018).
A third reason we focus our empirical analysis on the Arab uprisings and their aftermath is because they represent a hard test of our argument. While all of our proposed mechanisms are present in the majority of regime transitions countries, the MENA region also already ranks among the highest in the world for both GRD and SRD (Fox, 2020). This suggests that a further increase is less likely in the MENA region than other parts of the world where the baseline is lower.
A fourth and final reason we look at the Arab uprisings is because these cases offer variation in the size and types of religious minority groups. While Sunni Muslims make up the majority in all four regime transition countries, Egypt has a rather sizeable religious minority compared to the others -at 10% of the total population. These minority religions are mostly made up of Christians of various denominations, but also include Shia Muslims, Bahais, Jews, and Nubians. Minority religions in Libya account for about 3.4% of the population, and they primarily consist of Christians, Buddhists, Hindus, and Jews. In Tunisia, minority religions make up a mere 1% of the population and the Muslim population is overwhelmingly Sunni. The size of non-Muslim minority religions is equally small in Yemen. However, the Yemeni Muslim population has a substantial number of Shias (about 35% of the Muslim population, with the other 65% being Sunnis). This variation in religious landscapes enable us to guard against the possibility that our results are driven by a particular religious composition. The fact that we document below an increase in religious discrimination across these various religious landscapes suggests that it is our proposed mechanisms, as opposed to the size or type of the religious minorities, that are responsible for our findings.

Research design
We adopt a mixed-method research design to test the proposed link between regime transitions and minority religion discrimination. We begin with a statistical analysis of the RAS3 data, which covers 183 countries and independent territories around the world from 1990 to 2014. A case study of Egypt follows in order to provide more direct evidence of the institutional and religious outbidding mechanisms.
Quantitative analysis of regime change and religious discrimination Scholars have used the RAS3 data to examine different types of religious restrictions and their relationships with various political outcomes (e.g. Menchik, 2018;Arikan & Bloom, 2019;Henne, 2019). Akbaba & Fox (2019) also leverage the RAS3 dataset to explore the impact of the Arab uprisings on religious discrimination in the MENA region. Our findings are consistent with and extend those of that study. 5 We focus on two variables: GRD and SRD. GRD is concerned with restrictions imposed by the government on the institutions and practices of religious minorities. These can include constraints on the public or private observance of religious services, regulation of what religious symbols or clothing may or may not be worn in public, and restrictions on conversions to minority religions. It is measured using 36 indicators (available in the Online appendix) with possible scores ranging from 0 to 108. SRD differs from GRD as it is concerned with 'discrimination, harassment, prejudice, or violence against members of minority religions by members of society who are not representatives of the government' (Fox, 2017). Examples of the indicators used to assess the variable are whether or not there are anti-religious minority propaganda in the private media or attacks against properties owned by minority religious groups. Twenty-seven indicators are used to calculate the variable, with potential scores ranging from 0 to 81. For both GRD and SRD, higher scores indicate higher levels of discrimination.
To examine whether the levels of GRD and SRD indeed changed in countries that experienced a regime change following the Arab uprisings, we employed a synthetic control method (SCM). This method aims to create a synthetic control unit that matches the pre-treatment characteristics of the treated unit. Following Abadie, Diamond & Hainmueller (2015), for each dependent variable (i.e. GRD and SRD), we combined the scores of the countries that experienced a regime change (i.e. Egypt, Libya, Tunisia, and Yemen) to create a single composite score. This represents our treatment unit. We constructed our synthetic control unit, on the other hand, from the donor pool consisting of all other countries in the dataset that did not experience regime change or major protests during or shortly after the Arab uprisings. 6 SCM constructs the synthetic control by assigning weights to the potential control units in the donor pool according to their contributions. Greater weights are given to potential controls with greater contributions. Potential controls that do not contribute (are not included) in the creation of the synthetic control are given a weight of zero.

Variables
We focus on five covariates of interest for the synthetic control in this study. The first two covariates are also the dependent variables: GRD and SRD. Matching on the pre-intervention values of the dependent variables helps to account for the effects of the non-observables (Abadie, Diamond & Hainmueller, 2015).
The third covariate is the Polity IV score, which is used as an indicator of a country's democracy level. More democratic countries generally have more clearly defined separation of religion and state, as well as less government involvement in religion (Fox, 2007;Brathwaite & Bramsen, 2011). In addition, living in a democratic country may also inoculate in the citizens a respect of diversity through the process of democratic learning (Peffley & Rohrschneider, 2003), hence lowering the likelihood of societal discrimination.
The fourth covariate is GDP per capita. We use this as an indicator of a society's level of socio-economic development, which the modernization theory suggests should shape levels of GRD and SRD (Inglehart & Welzel, 2005). Lastly, the fifth covariate is the level of religious diversity (Pew Research Center, 2014). Sarkissian (2015) shows that religious diversity, in combination with level of political competition, shapes the level of GRD in a society. Several studies have also linked social diversity to higher probabilities of conflict and discrimination (see Montalvo & Reynal-Querol, 2005), while others argue that diversity can actually reduce conflict (see Dowd, 2016).

Results
Our analysis indicates that GRD has not changed since the Arab uprisings, while SRD has substantially increased in those countries that experienced regime change. We discuss each finding, in turn. Figure 1 presents the levels of GRD from the composite treatment unit (solid line) and the synthetic control (dashed line). 7 We observe an increasing trend that started even before the uprisings began. What is most notable from the figure, however, is the parallel trend shown by both lines. We see that the hypothetical level of GRD shown by the synthetic control closely matches that of the composite treatment both before and after the uprisings. This suggests that the regime change experienced by Egypt, Libya, Tunisia, and Yemen as a result of the Arab uprisings did not significantly change the levels of GRD in these countries. Figure 2 presents a different pattern. The levels of SRD of the composite treatment and the synthetic 6 The countries that experienced sizeable or major protests during the Arab uprisings are Bahrain, Syria, Algeria, Iraq, Jordan, Kuwait, Morocco, Oman, and Sudan. 7 Further details about the estimation results are available in the Online appendix.
control are roughly similar from 1990 to 2010. However, the two lines diverge from each other starting in 2011. The composite treatment unit exhibited a spike in the level of SRD, whereas the synthetic control was more of a flat line. Because the two lines largely overlap in the pre-uprisings period, the divergent trends that are

Robustness checks
To examine how dependent our results are on model specifications, we ran four robustness tests (see Online appendix). The first two are recommended by Abadie, Diamond & Hainmueller (2015): an in-time and an in-space placebo test. The in-time placebo test hypothetically assigned the Arab uprisings to a year in the preintervention period (in this case, 2007). Since no actual uprising or regime change happened in that year, then no diverging trends should emerge between the composite treatment and the synthetic control units. The in-space placebo hypothetically assigned a regime change to each country in the donor pool. Since these countries did not actually experience a regime change, then we should not expect them to experience changes similar to the composite treatment unit. Results from these two placebo tests are in the expected direction and conform to our general findings.
The third robustness test limits the donor pool to only Muslim-majority countries. Such an exercise enables us to address the argument that Islam has a unique relationship with religious restrictions by ensuring that the treated units and the donor countries are in a similar cultural zone, namely the Muslim world (Inglehart & Welzel, 2005). The results from this exercise are practically identical to those we obtained from the main analysis.
Our fourth robustness test confirms the importance of regime change. A critic might ask whether it was these political transitions that drove the changes in SRD or the unrest and protests that preceded them. We ran an identical SCM model to the one used in the main analysis, except that our composite treated unit consisted of only Arab uprising countries that experienced major protests but not regime change (see Footnote 5). If it was regime change (as opposed to civil unrest more generally) that mattered, then we should not see a similar increase in SRD among countries that, despite having major protests during the Arab uprisings, did not experience a political transition. Our results support this expectation. When we consider major protests as the treatment, we find no significant increases in the levels of GRD or SRD among the treated countries. This suggests that regime change is a critical factor that drove an increase in SRD after the Arab uprisings. But, to what extent do the institutional and outbidding mechanisms account for this pattern? We explore these through a case study of Egypt.

SRD in Egypt after the Arab uprisings
Levels of religious discrimination in Egypt were already among the highest in the world prior to the Arab uprisings (Fox, 2020). Yet, even here, SRD has intensified. To evaluate the extent to which our proposed causal mechanisms influenced this additional increase, we draw on seven months of ethnographic research by the second author. This includes daily observations of a Coptic community in Cairo and in-depth interviews with 27 Coptic Orthodox Christians. 8 We supplement these data, where appropriate, with secondary materials, including policy reports and international and Egyptian newspapers. Our focus is on the Coptic community because Christians in Egypt comprise the largest religious minority group, and the majority of Christians belong to the Coptic Orthodox Church.
The ethnographic research was conducted from January to July in 2014 during the transition period between Muhammad Morsi's removal from office in July 2013 and Abdel Fattah al-Sisi's assumption of the presidency in June 2014. This represents a period during which many Christians experienced short-term relief, quickly followed by increased fear of being attacked and persecuted. This fear was largely driven by attacks on church buildings and minority religions even after Morsi's ouster. In this context, some Christians declined to be interviewed, worrying they could be targeted if identified. The researcher, therefore, conducted all interviews in spaces identified by the participants as safe (e.g. local churches) and ensured confidentiality. The final interview participants included 13 men and 14 women whose ages ranged from 21 to 55. Approximately one-hourlong interviews were conducted in a mixed-income area in Cairo, which had one of highest concentrations of Christians in the city.
Our interviewees represent some of the least-likely Copts to experience physical violence. These residents agreed that their neighborhood has a longstanding reputation for nonviolence and peaceful interfaith relations. Many were proud that Muslims and Christians worked together to protect their properties during the Arab uprisings. For instance, residents took shifts to keep thieves from entering apartment buildings, and they also actively identified trespassers (interview with Bishoy on 27 June 2014). While the interviewees claimed they never personally experienced direct physical attacks, they did report experiences of micro-aggressions, as well as concerns about increased overt discrimination and violence towards the broader Coptic community following the removal of Hosni Mubarak and, subsequently, Morsi.

Deteriorating interfaith relations
Relations between the majority Sunni Muslim and minority religious, especially Christian, communities have deteriorated considerably in Egypt since 2011. Numerous media and human rights reports, as well as academic research, document this spike in hostilitieswhich includes abductions, assaults, rape, and forced conversions (see Gabbay, 2018;. Some of the worst violence occurred in August 2013 when Muslim Brotherhood members burned approximately 80 churches within a single month in the city of Minya (Human Rights Watch, 2013).
During interviews in 2014, most Cairenes expressed concern about both the increased number and boldness of attacks on religious minorities. Demiana, for instance, shared how Christians were being increasingly ill-treated by Muslims. She spoke with a worried face about younger Muslim students proudly saying, 'because he's Christian, I beat him'. She also compared this incident to past interfaith relations from her childhood in the 1960s, recalling that 'when we were little, we didn't talk like that'. More generally, she emphasized how religion has become a more salient indicator that divides Egyptians particularly after the Arab uprisings, adding that Christian children also speak negatively about their Muslim counterparts (interview on 4 July 2014).
Deteriorating social relations were a common theme among the Copts interviewed, older and younger adults alike. In separate interviews, both Abanoub and Bishoy talked about how a fast food restaurant, known to be run by Islamists, refused to hire a Christian in their neighborhood. According to Abanoub, '[t]he restaurant were looking for someone to hire but said they're not hiring Christians' (interview on 9 May 2014). Bishoy added that the owner of the restaurant made it public that they do not intend to hire Christians by '[putting] the ad in front of the store' (interview on 27 June 2014).
Karim, a man in his mid-30s, talked about an even more personal experience. His six-year-old daughter had recently asked why a Muslim classmate told her that 'I am not going to play with you because you're Christian'. He was disturbed and added that this is why he wants to leave the country. Recalling his childhood, Karim compared the current period to the 1980s: Other respondents shared that they have adjusted their behavior in the wake of an increase in SRD towards Christians. Sara, a woman in her 30s, spoke about how she no longer visits the malls run by Islamists after a particularly embarrassing moment at al-Tawheed wa al-Nour (unification and light), a nationwide shopping mall run by the Muslim Brotherhood. A Muslim store owner ordered her to 'cover your hair as Muslim women do'. She refused, saying 'no, I am Christian' and 'walked away'. Yet, Sara believed the owner already knew this from her unveiled hair, cross necklace, and a Coptic cross tattoo on her right wrist (interview on 23 May 2014).
These conversations underscore what has been captured in our quantitative analysis. Despite already high levels, SRD in Egypt has further intensified since the Arab uprisings, and the Coptic community, in particular, has experienced increased levels of insecurity and fear (see also . Our interviewees further shed light on what explains this escalation.

Weakened political institutions
One reason is the weakening of political institutions following the removal of Mubarak in 2011. In particular, the Egyptian government's inability or unwillingness to curb sectarian attacks against religious minorities, especially Copts, stands out in both our interviews and independent reports. For example, international human rights organizations have documented a variety of ways the Morsi regime 'routinely failed to condemn incendiary speech, including anti-Semitic and anti-Christian speech in mosque sermons and during broadcasts by Islamic "televangelists"' (US Department of State, 2014).
Several of our interviewees mirrored these concerns. They especially worried that the law was not applied evenhandedly to Muslims and Christians. Rather, in cases where both engaged in violence, the latter were punished, while Muslims were acquitted. An interview with Ibrahim captured this common sentiment among the Coptic community in Cairo: Our interviewees' sentiments about unfairness and insecurity are also in line with reports about Copts living in different regions. In the southern city of Qena in 2011, a group of Salafis allegedly accused a Christian man of renting an apartment to Muslim prostitutes. In reaction, the Salafis burned his apartment, beat him, and cut off his right ear. They were, however, never arrested or prosecuted, and the Christian man complained, 'the police refused to help' (Trofimov, 2011). Other Upper Egyptians have similarly asked, but never received, help from the police and army. A 2013 Human Rights Watch report, for example, captures the frustration of Coptic clergy with the lack of police response to the aforementioned church bombings in Minya. Following one attack by some 200 Muslims, a priest complained, '[t]hey took everything, all the equipment, furniture, everything. I called the police and army on their hotlines [ . . . ] no one came, the church is gone' (Human Rights Watch, 2013).
Another incident in northern Cairo underscores Egyptian security forces' indifference towards the Coptic community. In April 2013, an armed group attacked Christians walking out of the largest cathedral in the capital and killed six people (Kingsley, 2013a (Kingsley, 2013b). Similarly, in the case of the above-mentioned murder of a Coptic woman in Ain Shams, police did not intervene during the incident, nor was the woman's death well covered by the Egyptian media.
These episodes illustrate how regime change opened space for increased sectarian violence, not just in Southern Egypt, a relatively less developed and more conservative area of the country, but also in the capital. In particular, our interviewees and other Christian voices powerfully capture how weakened political institutions exacerbated tensions between majority and minority religious communities, opened space for discrimination to occur with impunity, and undermined public trust in the government.

Religious outbidding
An increase in Islamist propaganda and religious politics also followed the regime change in Egypt. Morsi, for instance, immediately started to enhance his religious credentials after his 2012 election as president by increasing the number of Muslim Brotherhood members in the government and strengthening the independence of al-Azhar scholars (Cook, 2013). There was also a surge in prosecutions and convictions over statements considered blasphemous or denigrating to Islam under the Morsi administration (Chick, 2013).
The Salafi parties that emerged during the transition similarly tried to gain a competitive edge in the new political environment by engaging in religious politics. Party members and Salafi supporters have, for instance, organized mass protests to mobilize their supporters. At those and other times, they have displayed the Salafi party symbol on banners and flyers in several cities, including Cairo, Alexandria, and Fayum. Also, their events often include the chanting of the slogan 'Islamic, Islamic, we don't want secular' (Elyan & Youssef, 2011).
The al-Nour Party has been particularly successful in leveraging their religious credentials and economic development programs to gain popularity (Ghalwash & Phillips, 2017). Since its inception in 2011, the party has stressed that its platform is in line with the Salafi Da'wa group, which aims to establish a nation based on sharia with the authentic principles of the Prophet Muhammad and early Muslims (Elyan & Youssef, 2011). 9 They have also strategically emphasized their religious ideology, which helped them appear to be more ethical and less corrupt than the past Mubarak regime and the FJP. This has led to support from a wide range of Muslim voters (Lacroix, 2012).
The al-Nour party has also been critical of efforts to incorporate Coptic Christians into government. The Party's spokesman, Mohammad Nour, condemned President Morsi's intention to appoint a Copt as vice president (Brundtland, 2012). Similarly, al-Nour officials and other Salafis condemned the appointment of a Copt as the new Qena governor in 2011, raising concerns about Christians ruling over Muslims (Trofimov, 2011). The government initially delayed their decision to appoint a Coptic governor for three months, but they eventually appointed a Muslim governor due to Salafis' violent protests (Naiem, 2018).
Salafi politicians have also made derogatory remarks about minority religious communities in their public rhetoric. It is not uncommon, for instance, to hear Copts referred to as 'kafir' (infidels) or calls to 'purify' the national identity (Lacroix, 2012: 10). The Asala Party, in particular, has called for the replacement of the current legal system with sharia (Islamic Law) and called those who oppose their perspective 'adulterers, thieves, and immoral people' (see Dabash, 2011).
Furthermore, Salafi politicians' political viewpoints have largely overlapped with and been reinforced by Salafi clergy and preaching against Christians, secularists, liberals, and Muslims who do not pray (see Arafat, 2017). During the 2011 parliamentary election, one Salafi preacher went so far as to tell a local newspaper it is unlawful according to Islam to vote for Christians, secularists, liberals, and Muslims who do not pray (Al-Masry al-Youm, 2011).
To summarize, the case of Egypt provides more direct evidence of the causal pathways we argue lead from regime change to increased SRD. Political institutions in Egypt weakened considerably after Mubarak's ouster, while religious politics has become a mainstay since 2011. Our interviewees provide a number of examples of how this subsequently increased tensions between the majority Muslim and minority Christian populations but did not necessarily lead to new government regulations.

Conclusion
This article offers theory and evidence to demonstrate that regime transitions are more likely to intensify SRD than GRD. Our empirical focus was on religious persecution following the Arab uprisings. And while the majority of the countries in our treatment unit experienced an observed increase in SRD, it is worth noting that one did not. That case, Yemen, is also the only one of the four transitional societies that did not hold open elections. Instead, regime change occurred through a power-transfer agreement brokered by the Gulf Cooperation Council.
This variation points to a puzzle for future research. We have argued that weakened institutions and religious outbidding contribute to an increase in SRD following regime transitions. But we do not make a strong case for whether both mechanisms must be present or if one may be a more influential factor. In Yemen, state institutions were weakened but religious outbidding did not occur because a new leader was appointed rather than elected. How much that process explains the stability in levels of SRD remains an open question.
Our argument and findings also have implications for how we think about political transitions, minority discrimination, and sectarian violence more broadly. Most notably, this study raises difficult questions about the promise of democratization for minority communities. A strong consensus persists among scholars and policymakers alike that religious discrimination in the MENA region ranks among the highest in the world because of a preponderance of authoritarian regimes (see Koesel, 2014;Sarkissian, 2015). But, our findings suggest that democratization, at least in the short term, does not necessarily hold more promise (Fox, 2020).
There are also reasons to remain wary of the longerterm benefits of democratization for religious minorities in the MENA region and beyond. SRD is frequently a precursor to more direct pressure exerted on political leaders to enact legislation that reduces the freedoms of minority groups (Grim & Finke, 2011;Finke & Martin, 2014). The increase in SRD that we observe, therefore, offers a cautionary note about other forms of religious discrimination, especially GRD, that might follow.
Furthermore, religious discrimination is often linked to larger-scale political violence, including civil war. Members of minority communities are not passive recipients of harassment and intolerance. They form grievances and mobilize in response to discrimination, especially when coupled with state inaction or complicity (Gurr, 1970;Basedau et al., 2011Basedau et al., , 2017. Also, this mobilization frequently leads to domestic terrorism and armed opposition movements (Akbaba & Taydas, 2011;Basedau, Pfeiffer & Vüllers, 2016;Ghatak, 2016;Saiya, 2018).
Future studies would do well, therefore, to further tease out the relationship between GRD and SRD, as well as their connection to larger-scale political violence. Because the systematic collection of information on religious discrimination remains a relatively recent endeavor, we are limited in how far we can extend our analysis. But as more data become available, other quantitative studies can analyze the longer-term effects of the increase in SRD that we observe. Historical case studies could also explore this relationship in now democratized countries, as well as uncover other mechanisms that might drive the relationship between regime change and other forms of religious discrimination and violence. These remain open questions, but of critical importance for hundreds of minority faith communities around the world.
This article also points to important policy implications for states with sizeable minority populations who are now experiencing political transitions, such as Myanmar and Sudan. The mechanisms we identify offer a useful way to understand how and when tensions will escalate between majority and minority religious communities. They also underscore how open elections can be double-edged swords that allow previously sidelined groups to participate in politics while at the same time allowing for new forms or increased levels of marginalization (Mansfield & Snyder, 1995;Bertrand & Haklai, 2013). Our findings, therefore, emphasize that elections in democratizing countries need to be managed in such a way that is sensitive to the possibility of exacerbating religious tension. This is, of course, no easy task. It will require emerging democracies to find a balance between free speech and the protection of minority groups.