The Infrastructure Advantage

Early literature on political Islam was highly attentive to how Islamists leveraged mosques to generate support. Munson (Reference Munson2001, 502), in describing the rise of Egypt’s Muslim Brotherhood, writes that ‘the mosque was the primary venue in which explicit recruitment to the organization took place’. In Iran, Ayatollah Khomeini himself claimed that ‘it was the mosques that created this Revolution, the mosques that brought this movement into being’.Footnote ³ In Algeria, ahead of the victory of the Islamic Salvation Front, Willis (Reference Willis1997, 122) writes that: ‘Full use was made of the network of some 9,000 mosques which Islamists had come to control, to coordinate activity and spread the party’s message at the weekly Friday prayers and sermon’. In Palestine, Abu-Amr (Reference Abu-Amr1993, 8) notes that ‘the Muslim Brotherhood’s most effective tool in spreading its influence was the mosques’. And in Tunisia, Waltz (Reference Waltz1986, 656) observes that: ‘it was actually in the mosques, in informally gathered seminars, that the MTI [Islamic Tendency Movement]… gained their first momentum’.Footnote ⁴

There are at least three reasons why mosques facilitate Islamist mobilization. The first is that even under dictatorship, mosques remained relatively free spaces for Islamists to mobilize. ‘Mosques were the only forum in which the government would permit large congregations of people during much of this period. Mosques were also relatively safe from police raids … even the state had to play by the rules in the mosques’ (Munson Reference Munson2001, 502). While governments did regulate who could preach and often locked up mosques in between prayers to prevent Islamists from leading informal study groups (Wiktorowicz Reference Wiktorowicz1998; Cammett and Luong Reference Cammett and Jones Luong2014, 196; Masoud Reference Masoud2014, 23), Islamists could often find independent mosques outside of the state’s control (Wickham Reference Wickham2002). And once dictators were toppled, Islamists enjoyed considerable freedom to mobilize in all of the country’s mosques.

Second, the mosque offers far more opportunities to interact than any other venue. Neighbours regularly and repeatedly come together at their local mosque to perform prayers. Islam prescribes five daily prayers and encourages believers to perform them at a mosque. Even for Muslims who only attend the mosque for the large Friday prayers (Salāt al-Jumu‘ah), the mosque still offers an opportunity for Islamists to meet them weekly. For Patel (Reference Patel2022, 1), it is this ‘unparalleled ability to communicate’ through mosques that explains why both Sunni and Shia Iraqis came to turn to Islamist preachers and parties post-US invasion.

The final reason, and the one that explains why Islamists benefit moreso than secularists from mosques, is that mosques bring together the relatively more religious subset of society. In Egypt, Munson (Reference Munson2001, 502) writes that mosques ‘served as a self-selection mechanism for potential recruits; those in attendance were already predisposed in some way to [the Islamists’] religious message … Only with a specifically Islamic message was the organization able to gain such effective advantages from mosques.’

While, of course, not all pious Muslims are Islamists, religiosity does, on average, correlate with support for Islamism (Tessler Reference Tessler2010; Gidengil and Karakoc Reference Gidengil and Karakoc2016). Survey data from Wave 2 of the Arab Barometer help to illustrate these dynamics. Surveying 12,000 Muslims across 10 countries, the Arab Barometer finds that Muslims who regularly attend mosque for religious lessons (Figure 2) or for Jumu‘ah prayers (Appendix, Figure A.2) are far more receptive to the Islamist project and appeal. Regular mosque attendees are significantly more likely to support the implementation of sharia in penal law (81 v. 69 per cent). They are almost twice as likely (61 v. 37 per cent) to think that the ‘country is better off if religious people hold public positions in the state’. They likewise highly value piety in political candidates when deciding who to vote for. The majority of mosque attendees, 55 per cent, ranked piety as among the top three factors they use in evaluating candidates, and they were almost twice as likely to rank it as the most important factor (28 v. 15 per cent). Mosques thus assemble a generally more sympathetic audience for Islamists than secularists.

Figure 2.

Mosque attendees are more likely to want Islamist candidates (Arab Barometer).

In some contexts, Islamist mobilization through mosques is explicit and visible. Politicized imams might exhort their congregants to vote for Islamists, Islamist candidates might present their platform in mosques, or Islamist parties and charities might hand out goods and services nearby (Wickham Reference Wickham2002; Clark Reference Clark2004; Brooke Reference Brooke2019; Medani Reference Medani2021, 58). Masoud (Reference Masoud2014, 27–28), for instance, describes Mohamed Morsi of the Egyptian Muslim Brotherhood ‘making the rounds of the district’s mosques’, delivering campaign speeches in between prayers ahead of the 2005 parliamentary elections.

Yet, in Tunisia, such explicit uses of mosques have been less common. Under dictatorship, Zine El Abidine Ben Ali tightly controlled the country’s mosques. ‘Plainclothes security officials attended the sermons to make sure that the imams did not deviate from their pre-written speeches, and to report any “suspicious” activities, such as unauthorised meetings’ (Wolf Reference Wolf2017, 111). Post-revolution, as many as 1,000 of Tunisia’s 5,000 mosques were taken over by hardline Salafists (what would become Ansar al-Sharia), but they rejected elections and criticized Ennahda as heretics for participating in 2011 (Zelin Reference Zelin2020, 100). These were mostly retaken by the state before the 2014 elections (Wolf Reference Wolf2014). Moreover, with the 2014 constitution empowering the state to ensure ‘the neutrality of mosques and places of worship away from partisan instrumentalization’ (article 6), the government cracked down on politicized imams, monitoring and firing imams ahead of the 2018 municipal and 2019 parliamentary elections (Gall Reference Gall2015; Asharq al-Awsat 2018; Donker Reference Donker2019; Saidani Reference Saidani2019).

For its part, Ennahda, in May 2016, decided to formally specialize into a political party, excising any preaching or proselytizing roles from the movement (Marks Reference Marks2016). It voted to expel any leader from the party if they also served as an imam or preached in a mosque, as a signal that it has separated religion from politics (Merone Reference Merone2019; Meddeb Reference Meddeb2019). They likewise excised the few local-level da‘wa and charitable organizations that individuals had set up post-2011 (Sigillò, Reference Sigillò2022).

As a result of these developments, the 2019 elections, in particular, saw little explicit Islamist mobilization in mosques. In a nationally representative, face-to-face survey of 1,200 Tunisians conducted in January 2020, only 2.3 per cent said their ‘local imam encouraged them to vote for a certain candidate’. About 16 per cent said ‘political candidates were campaigning inside or around their local mosque’, but almost half of these accusations actually came from folks who say they never attended the mosque. Secular Prime Minister Youssef Chahed (2016–20) confirmed that his government nearly eliminated such explicit electioneering in mosques.Footnote ⁵

But even when Islamists are only praying in mosques and not preaching or campaigning, I argue that there might still be other, more subtle, mechanisms for how mosques help Islamists generate support. First, the mosque still provides a forum for Islamists to regularly and repeatedly meet other attendees. What begins as casual interactions in the mosque can, over time, develop into much stronger social bonds. Dina Zakaria, a spokeswoman for Egypt’s Muslim Brotherhood, told me that such interactions were her introduction to the movement:

Wickham (Reference Wickham2002, 134) similarly describes how ‘Islamic ideological outreach was typically a personal, even intimate process, rooted in face-to-face human relationships’. Notably, this individual outreach is replicating in mosques across the country, producing personal, local bonds on a national scale. Such regular meetings strengthen the ‘bonding’ social capital between Islamists and their supporters (Putnam Reference Putnam2000). Mosques thus allow Islamists to develop deeper personal ties with their voters than secularists can.

Moreover, Islamists do not have to personally meet everyone in the mosque. Their mere presence there provides their fellow attendees with two important pieces of information. First, their regular attendance signals their piety to other mosque-goers. Given that piety was highly valued in political candidates, the Arab Barometer surveys then asked how you can tell if someone is pious (Figure 2). Not surprisingly, the majority of mosque attendees said it is by whether they attend mosque (64 per cent), such as for Friday prayers (59 per cent).

Mosque attendance was a particularly costly signal of piety under a dictatorship when Islamists in Tunisia were often harassed and interrogated when they attended prayers (McCarthy Reference McCarthy2018, 99). As former Ennahda Vice-President Mohamed Akrout noted:

For Ennahda then, regularly attending the mosque sent a particularly strong signal of their piety. People ‘identified us as experts in social and religious issues’, noted one Ennahda activist.Footnote ⁸ That, in turn, produced votes. ‘I’ve chosen to vote for Ennahda because they are closest to Islam’, noted one Tunisian in 2011.Footnote ⁹ ‘Ennahda is the party of Islam’, said another.Footnote ¹⁰ In Egypt, Munson (Reference Munson2001, 502) similarly observes that mosque attendance ‘gave the Society a borrowed religious virtuousness’.

In addition, when Islamists attend the mosque, they also send another signal to their fellow attendees: that they are engaged in and embedded in the community.Footnote ¹¹ They are not just elite strangers who come to town around election time to appeal for votes. They are instead integral parts of the (religious) community, regularly attending prayers year-round. As one voter remarked about Ennahda candidate Abdelfattah Mourou in 2019, ‘He is one of us, and he knows the life we lead’.Footnote ¹² Indeed, for decades, Mourou had made a conscious effort ‘to go around in different mosques’.Footnote ¹³ ‘We are the sons of the people’, he could thus credibly claim in his campaign.Footnote ¹⁴

Each of these factors, in turn, breeds trust in the Islamists or a belief that they will keep looking out for one’s interests even once in power when they may have incentives to act otherwise (Hardin Reference Hardin2002).Footnote ¹⁵ Having repeatedly met members of the party and having seen their piety and embeddedness firsthand, mosque-goers are more likely to trust that the Islamists will remain faithful. In short, these interactions in the mosque helped produce the Islamists’ reputation. Their image as pious, authentic, and trustworthy, which Cammett and Luong (Reference Cammett and Jones Luong2014) highlight as the key to their success, is created in part through the mosque.Footnote ¹⁶

Notably, in free, democratic contexts, this trust is being built in mosques across the country. During Tunisia’s democratic era, Ennahda boasted between 80–100,000 members nationwide (Lübben Reference Lübben2016; Cimini Reference Cimini2020, 967). As a result, there was likely an Ennahda member in every one of Tunisia’s 5,000+ mosques. That gave Ennahda considerable reach. Ennahda leader Rached Ghannouchi (Reference Ghannouchi2018, 8) himself wrote that ‘each of the 5,300 mosques in the country receives on average 1,500 persons for Friday prayer services … making this an optimal platform for promoting to a broad and religiously observant audience a compelling and tolerant religious vision’. If we assume that each of Ennahda’s 100,000 members earned the trust of even five other mosque-goers, it would amount to 500,000 voters, or almost 20 per cent of the voting population in 2019 (Ennahda would win 561,132).

Trust, of course, does not automatically entail voting for a party. You could trust multiple parties or prioritize policy over personal familiarity. Yet, decades of research into voter turnout suggest that personal appeals from people you trust have an outsized influence (Rolfe Reference Rolfe2012; Green and Gerber, Reference Green and Gerber2019). Mosques, much like churches in the USA (Liu, Austin and Orey Reference Liu, Wright Austin and D’Andra Orey2009; Gerber, Gruber and Hungerman Reference Gerber, Gruber and Hungerman2015), can help to generate this trust and, in turn, voter turnout (Jamal Reference Jamal2005; Moutselos Reference Moutselos2020). As a result, trust in Islamists tends to correlate with votes for them (Grewal et al. Reference Grewal, Jamal, Masoud and Nugent2019; Livny Reference Livny2022).

This greater trust in the Islamists, in turn, helps them also win re-election. Throughout their time in office, party members continue to attend the mosque and have the opportunity to regularly explain to their supporters what their party is doing and why. ‘The mosque is a place of discussion…and exchange’, observed Ennahda MP Monia Brahim.Footnote ¹⁷ Egyptian MP AbdulMawgoud Dardery, from the Muslim Brotherhood’s Freedom and Justice Party, noted that conversations about politics were normal in mosques before and after prayers:

Ennahda members thus had the opportunity to regularly explain their performance and to do so through intimate, face-to-face, one-on-one conversations. Moreover, the trust that Ennahda had garnered among its supporters helped it to sell even the most difficult compromises. When Ennahda decided to form a coalition with remnants of the old regime (Nidaa Tounes), for instance, one Ennahda supporter remarked: ‘It is not right, and we know it, but it is smart … so we will be smart’.Footnote ¹⁹ The trust she had in Ennahda allowed her to give them the benefit of the doubt. Ennahda VP Akrout explained that:

In short, the trust that Ennahda had built with its supporters through repeated interactions in mosques was the glue that helped them retain their voters across elections. While other parties that lacked that trust crumbled, Ennahda could retain its core base even in the face of poor performance and political compromise.

Secular politicians and their supporters, of course, also attend mosque, but on average, less often than Islamists do. In a survey of candidates, Blackman, Clark and Sasmaz (Reference Blackman, Clark and Sasmaz2018) find that 62 per cent of Ennahda candidates regularly engage in religious activity like attending mosque, compared to only 39 per cent of Nidaa Tounes candidates. And Ennahda candidates were more than twice as likely to report doing so every day. As for voters, in my survey (discussed below), 56 per cent of Ennahda voters said they attend mosque daily or weekly, compared to only 30 per cent for secular voters. Mosques thus bring together Islamists more than secularists. Moreover, secularists lack a comparable infrastructure for their natural constituents. There is no pre-existing, nationwide network of institutions that regularly brings together the less religious segment of society in the way that mosques do for the more religious. Mosques thus provide a unique infrastructure advantage for Islamists. The exception that proves the rule is Tunisia, where the Tunisian General Labor Union (UGTT) has amassed a powerful, nationwide network of generally secular-leaning labour organizations (Yousfi Reference Yousfi2018), which has, in turn, facilitated mobilization for protests (Ketchley and Barrie Reference Ketchley and Barrie2020). However, the UGTT has thus far chosen not to endorse any parties for elections (Bishara and Grewal Reference Bishara and Grewal2022), and thus, Islamists, even in Tunisia, continue to enjoy an infrastructure advantage.

While intuitive, few studies have attempted to test Islamists’ infrastructure advantage. Although we have important qualitative accounts of Islamists mobilizing through mosques, we do not yet know if this is only occurring in a handful of mosques under Islamist control or instead is a more systemic, nationwide advantage even where Islamists are not preaching. Likewise, we do not yet know whether mosques in this story are simply epiphenomenal, just reflecting underlying religiosity and Islamism, or whether they instead have an independent effect of their own as theorized above.

Masoud (Reference Masoud2014, 169–174) takes a first stab at these questions, correlating the number of mosques per capita at the qism level in Cairo and Alexandria with Islamist vote share, but finds only mixed evidence. He also finds no survey evidence that mosque attendance correlates with Islamist votes (p. 177). He does find evidence in Tunisia of a correlation between mosques and Ennahda’s 2011 victory but underscores that he cannot control for underlying levels of religiosity (pp. 196–199). Meanwhile, Brooke and Masoud (Reference Brooke and Masoud2016) found no link between mosques and Islamist vote share in Egypt’s 2012 election. Outside of the Middle East, however, Bazzi, Koehler-Derrick and Marx (Reference Bazzi, Koehler-Derrick and Marx2020) show that religious institutions encouraged the rise of Islamists in Indonesia, and outside of Islam, religious institutions correlate with support for Christian democratic parties in Europe (Duncan Reference Duncan2015) and ultra-Orthodox Jewish parties in Israel (Freedman Reference Freedman2020). However, whether and to what extent mosques explain the Islamist advantage in the Middle East remains an open question.

In this paper, I examine the infrastructure advantage in Tunisia, leveraging original data on both mosques and mosque attendance. Tunisia is a particularly useful case for isolating the mechanism behind the theory. Because explicit campaigning in mosques did not occur in Tunisia, particularly in the 2019 elections, any effect of mosques on Islamist votes is likely driven instead by the more subtle mechanisms outlined above. Moreover, for that reason, Tunisia is also a hard case for uncovering any infrastructure advantage. If mosque effects are found in Tunisia when only subtle mechanisms were in operation, they are likely to be even stronger in cases where Islamists can leverage politicized imams and explicit campaigning.

If mosques truly provide Islamists an advantage, we should observe the following patterns:

• Hypothesis 1: At the individual level, Muslims who attend mosque more often should be more likely to vote for Islamists, and this should be due to greater trust in them.

• Hypothesis 2: At the district level, areas with more mosques per capita should see more votes for Islamists.

Test 1: Survey Data

The first test of the infrastructure hypothesis is through an original, nationally representative, face-to-face survey of 1,200 Tunisians conducted in January 2020, three months after Ennahda’s victory in the 2019 elections.Footnote ²¹ At the individual level, I show that Tunisians who attended mosque more often were more likely to vote for Ennahda and that this was due to their greater trust in Ennahda.

In the survey, 44 per cent said that they voted in the 2019 parliamentary elections (compared to 42 per cent in reality), and of those who voted, 19.9 per cent said they voted for Ennahda (compared to 19.6 per cent in reality). The survey thus captured Tunisian voting behaviour quite well. To measure mosque attendance, the survey asked respondents: ‘In general, how often do you visit your local mosque for reasons related to religious worship or prayer?’ The results reflect Tunisia’s polarized society; almost half of the sample (45 per cent) answered never, while another 33 per cent were very frequent attendees, attending mosque every week or every day.

Figure 3 (left) shows that the frequency of mosque attendance correlated strongly with voting for Ennahda in the 2019 parliamentary elections. While only 5 per cent of those who never attended mosque voted for Ennahda, about 17 per cent of those who attended daily did so, more than tripling support for Ennahda.Footnote ²² Mosque attendance likewise correlated with voting for Ennahda’s candidate in the presidential elections, Abdelfattah Mourou (see Appendix, Figure A.5), and at a similar magnitude. By either measure, mosque attendance strongly correlates with support for Ennahda. Meanwhile, mosque attendance does not correlate with voting for any of the secular parties (see Appendix, Table A.3).

Figure 3.

Mosque Attendance and Voting for Ennahda in the 2019 Elections.

Note: Figures created from Table A.2, models 1–2 (Appendix).

There are several alternative explanations, each of which I control for in Figure 3 (right).Footnote ²³ First, mosque attendees may be more religious than non-attendees, which may, on its own, increase support for Ennahda. To control for piety, I include whether respondents pray five times a day,Footnote ²⁴ and whether they had any outward signs of piety (such as a hijāb or zabība). Neither correlated with voting for Ennahda in the full model, suggesting the important variation is instead in mosque attendance. Indeed, even among respondents who pray five times a day, those who do so in mosques were significantly more likely to vote for Ennahda (17 per cent) than those who do not (6 per cent), a difference of similar magnitude as in the full sample.

Second, mosque attendees may be more supportive of Islamist ideology or more likely to identify as Islamist, and that might be driving their support for Ennahda (Livny Reference Livny2022). I thus control for whether they self-describe their political views as Islamist. Mosque attendance continues to strongly correlate with voting for Ennahda, suggesting it has an effect independent of greater Islamism. Similarly, one might vote for Ennahda out of support for the revolution or for democracy, yet the correlation holds despite these controls.

Finally, perhaps attending mosque exposed them to the more explicit uses of a mosque, including politicized imams, political campaigns, or social service provision. I thus control for whether respondents said their ‘local imam encouraged them to vote for a certain candidate’, whether ‘any political candidates were campaigning inside or around their local mosque’, and whether any party offered them food, cash, or help with the bureaucracy in exchange for their vote (only 0.8 per cent said Ennahda did so).Footnote ²⁵ More generally, perhaps mosque attendees were more likely to receive government jobs or reparations from Ennahda – I thus control also for employment, income, and their perceptions of the economy and of corruption. Despite these controls, mosque attendance continues to show a robust correlation with voting for Ennahda, suggesting it is not operating through any of these mechanisms.

Instead, I argue that what drives the correlation between mosque attendance and voting for Ennahda is greater trust in Ennahda, built through heightened social interactions with them. The survey asked: ‘on a scale from 1-10, how much do you trust Ennahda?’Footnote ²⁶ Figure 4 (left) shows that mosque attendance indeed positively correlates with trust in Ennahda. Moreover, a mediation analysis (Figure 4, right) shows that trust in Ennahda fully mediates the relationship between mosque attendance and voting for Ennahda. The direct effect is not significant; in other words, there is no reason other than greater trust in Ennahda that mosque attendance correlates with voting for Ennahda. While the data are correlational, not causal, the analysis provides suggestive evidence that trust in Ennahda is the primary vehicle by which mosque attendance correlates with voting for Ennahda.

Figure 4.

Mosque Attendance and Trust in Ennahda.

Note: Figures created from Tables A.2, model 3 and Figure A.4 (Appendix).

These effects are not limited to men. Women are far less likely to attend mosque regularly (with only 13 per cent saying they attend daily or weekly, compared to 54 per cent for men), but those who do exhibit the same patterns. Among women only, mosque attendance still correlates with greater trust in and the likelihood of voting for Ennahda (Appendix, Table A.4).

Finally, mosques appear to have helped Ennahda retain its core base across elections. Figure 5 compares the respondents who voted for Ennahda in 2019 to the voters who had voted for Ennahda in the past but did not do so in 2019. The results are illuminating: the voters that Ennahda lost over the course of the transition were those who went to mosque less often, a finding that holds with the addition of all controls (Appendix, Table A.5). Interactions in mosques thus appear to be key to Ennahda’s resilience.

Figure 5.

Ennahda retained the more regular mosque-goers.

Note: Figure created from Tables A.5, model 1.

In short, the survey uncovers strong evidence of the individual-level micro-foundations of the theory, showing that mosque attendance breeds trust in Ennahda and, in turn, votes for them. To see if this relationship generalizes spatially, I now turn to a district-level test.

Test 2: Spatial Data

The second method of testing whether mosques give Ennahda an advantage is through an ecological analysis of mosque density and electoral returns. I show that Ennahda performed significantly better in districts with more mosques per capita in the 2011, 2014, and 2019 parliamentary elections.

The outcome of interest is Ennahda’s vote share in these elections. I collect this data at the delegation level. Administratively, Tunisia is divided first into 24 governorates (wilāyāt) and subsequently into 264 delegations (mu‘tamadiyāt). Delegations are thus a rough equivalent of US counties, with an average population of about 33,000. All electoral results are obtained from the electoral commission, the Instance Superiore Independante pour les Elections (ISIE).Footnote ²⁷ For illustration, Figure 6 shows Ennahda’s vote share in the 2011 and 2019 elections. There is considerable variation, ranging in 2019 from a low of 5.7 per cent in El Ayoun (Kasserine) to a high of 65 per cent in Ghannouch (Gabes).Footnote ²⁸

Figure 6.

Ennahda Vote Share, 2011 (left) and 2019 (right).

The primary independent variable is the number of mosques per capita at the delegation level. Since mosques are continually being built, I collect this variable at two snapshots in time: first in 2011 (from Masoud Reference Masoud2014) and second in January 2019 (directly from the Ministry of Religious Affairs, Wizārat al-Shu‘ūn al-Dīniyya).

Tunisia had 4,716 total mosques in 2011 and 5,934 in 2019. In other words, since Ennahda took office in December 2011, over 1,000 new mosques have been built or registered. It is thus possible that Ennahda may have influenced where new mosques were built.Footnote ²⁹ If Ennahda reallocated mosque construction to its strongholds, it may artificially create a correlation between mosques and Islamist vote share by the 2014 and 2019 elections. Empirically, the opposite is true: new mosques were disproportionately built where Ennahda was weak in 2011 (see appendix, Figure A.11). However, the possibility remains, and thus I privilege as the primary independent variable the mosque data from 2011, prior to Ennahda’s time in office. However, I show in the appendix that results are robust to using the 2019 data instead (Table A.11), and I also explore the effect of the new mosques below.

For both years, I standardised the number of mosques per delegation by that delegation’s population, as recorded in the 2004 census for 2011 and the 2014 census for 2019. I thus calculate mosque density as the number of mosques per 10,000 people. Figure 7 presents this data in each of Tunisia’s delegations in 2011 and 2019. There is likewise a large variation, ranging in 2019 from 1.16 mosques per 10,000 people in Soukra (Ariana), a posh suburb of Tunis, to 31.5 mosques per 10,000 people in Matmata (Gabes).

Figure 7.

Mosque Density in 2011 (left) and 2019 (right).

Figure 8 shows that there is a significant bivariate correlation between mosque density and Ennahda’s vote share in all three elections, lending credence to the infrastructure advantage (see corresponding regression table A.6 in Appendix). Ennahda performed better in areas with high mosque density, such as in the southern and interior regions of the country, while performing worse in areas with low mosque density, such as in the northwest.

Figure 8.

Bivariate Relationship, Mosques and Ennahda Vote Share.

Note: Figures created from Table A.6, models 1–3 (Appendix).

I then test the robustness of these correlations to a variety of covariates.Footnote ³⁰ First, one potential confounder is religiosity: perhaps areas that are more religious build more mosques and vote for Islamists, creating a correlation between them. Likewise, mosques themselves may increase religiosity. I control for religiosity in two ways. First, I create a survey-based measure of piety by combining six nationally representative, face-to-face surveys: the four waves of the Arab Barometer in Tunisia (2011, 2013, 2016, and 2018) and two surveys of my own conducted with the same survey firm, One to One for Research and Polling, in 2019 and 2020 (the latter being the one used in Test 1). The only measure of piety in all six surveys is how often respondents pray, a metric that remains quite stable every year: between 60–70 per cent of respondents in each survey said they pray at least once daily (see appendix, Figure A.6). Given its stability over time, I combine all six surveys to produce a total of 8,203 respondents, or about 31 per delegation, allowing us to make tentative claims about religiosity at the delegation level.

Still, these surveys were representative at the national level and not necessarily at the delegation level. As an alternative measure of religiosity, I follow Ciftci, Robbins and Zaytseva (Reference Ciftci, Robbins and Zaytseva2021) and Livny (Reference Livny2021) and collect satellite imagery of night-time lights during the month of Ramadan. Since many rituals during Ramadan occur after sunset (including the meals of iftār and suhūr, and the tarawīh prayers), we should observe an increase in night-time lights in areas with greater religious observance compared to the months before and after Ramadan. Using publicly available satellite data,Footnote ³¹ I thus calculate each delegation’s per cent increase in night-time lights during Ramadan compared to the average of the months before and after as a proxy for sub-national devotion (for more details, see appendix).

Beyond religiosity, I control for a number of socioeconomic variables thought to shape support for Islamists (Pellicer and Wegner Reference Pellicer and Wegner2014). Across the region, Islamists tend to perform better in urban, densely populated areas, especially among unemployed and lesser-educated youth (Willis Reference Willis1997). From the census, I thus control for the delegation’s population, per cent youth (aged 20–29), per cent of the workforce involved in agriculture (as a proxy for urban/rural), the per cent unemployed, and the per cent with higher education. For the 2019 elections, I am also able to control for a new delegation-level measure of poverty released in 2020 by Tunisia’s National Institute for Statistics.Footnote ³²

Third, I control for two political grievances thought to undergird Ennahda’s support in Tunisia. The abuses of former dictator Zine El Abidine Ben Ali, including police repression and torture, are thought to have fueled support for the opposition, in particular for Ennahda, which bore the brunt of this repression. Ben Ali’s corruption likewise fueled support for the opposition, including Ennahda. To measure these political grievances, I use the number of cases submitted to the Truth and Dignity Commission (Instance Vérité et Dignité, IVD) in each governorate for a) human rights violations and b) corruption.Footnote ³³

Fourth, for the 2019 elections, I am also able to control for three alternative mechanisms: mosques might provide a location for Islamists to preach, provide social services, or explicitly campaign. As we saw in Test 1, the January 2020 survey included questions to capture each of these. Those questions were also included in a second face-to-face survey of 1,008 Tunisians conducted by the same survey firm in September 2019, right before the elections, as well as in an online survey of 11,329 Tunisians that ran in between.Footnote ³⁴ Combining these three surveys produces a total of 13,542 respondents or about 51 per delegation. If mosque density correlates with Ennahda’s vote share even after controlling for these reports of Ennahda using mosques to preach, offer services, or campaign, it will help to rule out these three alternative mechanisms.

Figure 9 presents the results. Despite these controls, mosque density continues to exhibit a significant, positive correlation with Ennahda’s vote share in all three elections. Substantively, Ennahda performed about 10–20 percentage points better in areas with the highest mosque density. In 2014 and 2019, that represented roughly a doubling of Ennahda’s vote share in areas with the highest v. lowest mosque density.

Figure 9.

Mosque Density and Ennahda Vote Share in 2011, 2014, and 2019.

Note: Figures created from Table A.9, models 1 and 3, and Table A.10, model 1 (Appendix).

In addition to mosque density, piety is also a consistent predictor of voting for Ennahda, validating our survey-based measure of religiosity.Footnote ³⁵ Moreover, the fact that mosque density remains significant after controlling for piety suggests that the link between mosques and Islamist vote share is not simply reflecting underlying levels of religiosity.

Likewise, the results from 2019 suggest that the alternative mechanisms are not driving the mosque effect. Mosque density remains a significant predictor of Ennahda’s vote share in 2019, even when controlling for the (few) reports of Ennahda vote-buying, campaigning near mosques, or politicized imams. None of these variables, meanwhile, ever positively correlate with Islamist vote share.

Instead, the link between mosque density and Ennahda’s vote share appears to be driven by more subtle interactions in the mosque breeding trust in Ennahda. The same three 2019–2020 surveys also asked respondents how much they trust Ennahda on a 1–10 scale. Figure 10 (left) shows that delegations with more mosques per capita showed significantly higher trust in Ennahda. That heightened trust, in turn, produced Ennahda’s higher vote share. A causal mediation analysis (Figure 10 right) shows that 70 per cent of the total effect of mosque density on Ennahda’s vote share was operating through heightened trust in Ennahda.

Figure 10.

Mediation: Mosque Density, Trust in Ennahda, and Ennahda Vote Share.

Note: Figures created from Table A.15, model 1, and Figure A.10 (Appendix).

The infrastructure advantage does not appear to be limited to Ennahda. Figure A.8 (Appendix) shows that mosque density likewise correlates with the vote share of the Dignity Coalition (‘itil ${\bar {a}}$ f al-kar ${\bar {a}}$ ma), a new hardline Islamist party that emerged in 2019, as well as the vote share of all Islamist parties combined (Ennahda, Karama, and Errahma, which ran in a few districts). Mosques seem to benefit Islamist parties across the board.

Two extensions lend further credence to the theory. First, mosques matter most when frequently used. Interacting mosque density with a survey-based measure of mosque attendanceFootnote ³⁶ reveals a powerful and robust interaction effect. Figure 11 shows that in all three elections, effects are concentrated in delegations that have both high mosque density and high mosque attendance. Results are thus not driven by some selection effect through which mosques came to be there, but rather whether those mosques are actually used and thus producing the interactions theorized above.

Figure 11.

Interacting Mosque Density with Mosque Attendance (2011–2019).

Note: Figures created from Table A.12, models 1–3 (Appendix).

Finally, mosques appear to have helped Ennahda retain its voters across elections (Figure 12). While Ennahda lost about 60 per cent of its votes between 2011 and 2019 in delegations with the lowest mosque density, it lost only 40 per cent in delegations with the highest density. In addition, Ennahda also retained significantly more of its voters in areas where new mosques were built between 2011 and 2019. Even though those new mosques were disproportionately constructed in areas where Ennahda had been weak, they seem to have helped Ennahda retain its voters there.

Figure 12.

Mosque Density and Ennahda Vote Retention (2011–2019).

Note: Figures created from Table A.13, model 3 (Appendix).

In sum, the data from Tunisia suggest that mosques may lend Islamists an advantage in elections, with mosque density strongly correlating with the vote share of multiple Islamist parties across multiple elections. Combined with the individual-level survey data in Test 1, there appears to be strong evidence of the infrastructure advantage in Tunisia.

Test 3: Arab Barometer

To explore how far the theory travels, I now turn to region-wide survey data from the Arab Barometer, which has conducted six waves of nationally representative face-to-face surveys in the Arab world. Three of these waves include all the survey questions needed: Wave 2 (2011), Wave 3 (2013), and Wave 5 (2018–19).Footnote ³⁷ In all three waves, I show that respondents who attend mosques more often were more trusting of Islamists.

While the Arab Barometer surveys did not ask about voting for Islamists, they do include our core mechanism: respondents’ level of trust in the main Islamist party in their country.Footnote ³⁸ Wave 2, conducted in 2011, asked this question in Egypt and Tunisia (N=2415). Wave 3 asked it in all countries surveyed, namely Algeria, Egypt, Iraq, Jordan, Kuwait, Lebanon, Libya, Morocco, Palestine, Sudan, Tunisia, and Yemen (N=14,809). Finally, Wave 5 asked it in Egypt, Iraq, Jordan, Libya, Morocco, Palestine, Sudan, Tunisia, and Yemen (N=10,319). In all countries, trust in Islamists is asked on a four-point scale from absolutely no trust (1) to a great extent of trust (4). I treat this as a continuous variable and run standard OLS regressions, though the results are robust to ordered logistic regression.

The Arab Barometer likewise does not include a question on general mosque attendance, but they do ask how often respondents attend Friday prayers (Sal ${\bar {a}}$ t al-Jumu‘ah). Friday is the most holy day in Islam, and the Jumu‘ah prayers are the most important of all prayers. Notably, unlike other prayers, the Friday prayer must be performed in the mosque; indeed, the more literal translation of Sal ${\bar {a}}$ t al-Jumu‘ah is congregational or communal prayer. Friday prayers are thus the most commonly attended prayer at the mosque and, accordingly, a useful measure of mosque attendance (Brooke, Chouhoud and Hoffman Reference Brooke, Chouhoud and Hoffman2023).

Of course, attendance at Friday prayers not only reflects exposure to the mosque (the concept of interest) but also religiosity, which could also increase trust in Islamists. To distinguish between these, I include in every model a control for the frequency of prayer in general, which would capture both prayers at the mosque as well as prayers performed individually. Indeed, the correlation between how often one attends Friday prayer and how often one prays in general was just 0.33 in 2011, 0.36 in 2013, and 0.17 in 2019.Footnote ³⁹ Including a control for frequency of prayer in general thus more credibly suggests that any effect of Friday prayers is operating through mosque attendance, not religiosity.

I also control for a number of variables that capture the existing explanations for support for Islamists (for full question wording, see appendix). First, to capture Islamist ideology, I include respondents’ support for Sharia, antipathy to the USA and Israel, and opposition to the rights of non-Muslims and women. To proxy for a desire for Islamist services and redistribution, I examine respondents’ evaluations of whether their government is tackling inequality and corruption. To measure Islamist reputations, I measure how highly respondents rank ‘integrity’ and ‘piety’ as qualifications for political leadership (wave 2 and 3) and whether they believe religious leaders are less corrupt than non-religious ones (wave 5). For a reputation for opposing dictatorship, I include respondents’ support for democracy, which would indicate they might value such a reputation. Finally, I include standard demographic variables, including age, gender, marriage, education, income, unemployment, and country fixed effects.

Figure 13 presents the results for Wave 2 (top), Wave 3 (middle), and Wave 5 (bottom) of the Arab Barometer. Across all three waves, attending the mosque for Friday prayers is a major predictor of trust in Islamists. In 2011, for instance, an individual who ‘always’ attends Friday prayers was about half a point more trusting of Islamists (on a 1–4 point scale) than someone who ‘never’ attends Friday prayer. Notably, this effect is significant (p ${\lt}$ 0.001) despite controlling for frequency of prayer, suggesting that mosque attendance has an additional, independent effect from religiosity.

Figure 13.

Friday Prayers at the Mosque and Trust in Islamists (Arab Barometer).

Note: Figures created from Table A.17, models 1–3 (Appendix).

Moreover, mosque attendance is one of the few variables that is consistently significant across all three waves of the Arab Barometer. The others are support for sharia and support for democracy. The importance of mosque attendance does not appear to be limited to men: an interaction by gender is only (weakly) significant in one of the three waves (see appendix, Table A.18). In short, mosque attendance, as proxied by Friday prayers, appears to be one of the most important predictors of trust in Islamists.