Literature review

The question of ‘what causes terrorism’ has been extensively addressed in the field by many authors. The use of terrorist tactics in general has been explained through the fact that these can be successful and reach the goal (Pape, Reference Pape1997: 343–361; Kydd and Walter, Reference Kydd and Walter2006: 49–80). Another explanation looking at the causes of considers social and economic factors related to discrimination against minorities, explaining how these grievances can lead to an increase likelihood of terrorism (Piazza, Reference Piazza2011: 339–353). The same author (Piazza, Reference Piazza2012: 521–546) highlights in a large-N study that religious discrimination is not relevant in explaining terrorism.

Kydd and Walter (Reference Kydd and Walter2006) explore how terrorist groups actually use several strategies according to different contexts, for instance weak groups can increase popular support for them by trying to provoke forceful state response (ivi: 154). In other words, not all terrorist groups follow the same logic, an element that complicates the study of the phenomenon.

In fact, in terms of the general understanding regarding why people choose to become terrorists there are several possible explanations in the literature. Kavanagh (Reference Kavanagh2011: 106–132) suggests that poverty is a relevant factor, but at the same time highly educated people of poor areas are more likely to become terrorists, a result that reflects the theory of Gurr (Reference Gurr1970) regarding relative deprivation. Fattah and Fierke (Reference Fattah and Fierke2009: 67–93) underline the importance of humiliation as a driving factor of radicalization in the Middle East, arguing that this feeling of humiliation (after military interventions, for instance) is a crucial factor. Other authors, like Cottee and Hayward, focused instead of the individual psychological aspects of what causes people to join terrorist groups, arguing that desire for adventure and unconventional life is important (Reference Cottee and Hayward2011: 963–986). As it can be seen by this short review there is no overarching explanation of why people become terrorists, and not in small part because this element can be heavily context-dependent.

Regarding this last aspect, Bruce Hoffman argues that the role of the terrorist organizations in the Middle East is crucial to understand Islamist-related terrorism (Reference Hoffman2017). Other authors, mainly Sageman (Reference Sageman2008) disagree with this view and think that instead Islamist-related radicalization can also happen outside the tactics and the strategies of the terrorist organization, such as Al-Qaeda and ISIS. In both of these cases, however, the domestic aspect of grievances likely plays a role in explaining the radicalization of individuals in the West, also because the new media help the propaganda of extremist networks, which is aimed at spreading the sense of outrage against the Western world after the developments linked to the wars in Iraq and Afghanistan.

These elements are obviously relevant and play a role in the radicalization of Muslims in the Western world, but this process can also happen because of the domestic grievances of the Muslim minority toward the host state, regardless of what the leaders of the terrorist organization say. Olivier Roy (Reference Roy2017) suggests how the cases of radicalization in France are linked to global cleavages but that resonate in the specific context of the French Muslim community. Although foreign policy is often mentioned as a reason for the attack by perpetrators, generally these terrorists are not directly linked to the organization, even if they pledge allegiance to it. In other cases, like the Charlie Hebdo attacks in France, the reason of the attack is entirely domestic: in the specific case some Muslims were outraged by the provocative cartoons of the satirical magazine portraying the Prophet Mohammed in a bad light.

Following the argument of Olivier Roy, the reasons for the radicalization cannot be disentangled from the social and economic domestic conditions of minorities, which in turn increase the level of frustration of marginalized communities (regardless of whether the terrorists are marginalized themselves). This situation leads to political extremism, but in the words of the author the result of this process is not ‘a radicalization of Islamism, but an Islamization of radicalism’ (Roy, Reference Roy2017: Ch. 3). This would suggest that what occurs first is the radicalization process in terms of extremism, that it is later channeled into the religious aspect. The argument made here is that the transnational element of the strategy of the terrorist organizations cannot be disentangled from the domestic grievances of the Muslim minorities in Western Europe, also because the messages of jihadist propaganda would not resonate in the same way in the absence of a fertile ground caused by discrimination toward these communities.

These domestic grievances can resonate more with the Muslim minorities if within the community there is a high number of religious and cultural networks and meeting places, which could connect grievances and increase the likelihood of radicalization of small portions of Muslims, therefore increasing also the risk of terrorist attacks in the host country. This link has been analyzed mainly in the media rhetoric and by populist parties rather than in the academia, where however some articles look at the role of mosques in society in general (Saint-Blancat and Schmidt di Friedberg, Reference Saint-Blancat and di Friedberg2006: 1083–1104), finding mixed results (Shaffer, Reference Shaffer2016: 383–394). Rabasa and Benard (Reference Rabasa and Benard2015: 110–117) highlight how some of the terrorists became radicalized in prison: the most common pattern is the one of second or third generation Muslim committing petty crimes because of social marginalization. These same people then come in contact, while in prison, with Islamist extremists, and when once they go out they become radicalized. That is because ‘for petty criminals without a future, radical Islam offers a way to rise above their condition in life’ (ivi: 110). It can be seen also in this case how the root cause of radicalization is still linked to grievances and lack of integration in the host society: if this was not the case, then there would have been no need to commit crimes in the first place.

In terms of the specific case studies presented in this paper, the literature generally focuses on explanations related to alternative hypotheses rather than to the variables presented here. In this comparison of countries, Italy is the only state which (so far) did not suffer any fatal Islamist-related terrorist attacks. It should be kept in mind that however Italy is a target of ISIS, also because the city of Rome represents the center of Christendom and therefore it has a strong symbolic value, as the ISIS propaganda repeatedly show explicitly threatening the city of Rome. In some occasions during the past few years, the Italian police and the intelligence services were able to prevent terrorist attacks to the Capital and other parts of Italy and arrest and expel the wannabe terrorists (Tgcom, 2017). Groppi (Reference Groppi2017: 20–27) writes that this ‘Italian exceptionalism’ is rapidly diminishing, and he attributes the lack of terrorist attacks to the fact the Islamic community is still relatively small. According to his view, whereas terrorists constantly threatened Rome, the rationale for the threat is to appeal to the sympathizers more than to actually plan a terrorist attack, as the concept of ‘Rome’ is used most likely as a metaphor to indicate Christianity. Yet, terrorist groups did try a number of times to attack Rome, even if they were prevented by the Italian law enforcement and Intelligence agencies. Moreover, there is a large Muslim population in Italy too, even if the proportion from the percentage of Muslim in other countries is different. Vidino (Reference Vidino2011: 389–418) realized an unusual specific case study, the terrorist cell of Buccinasco, a small Italian village in the North in the outskirts of Milan, explaining the dynamics and the reasons of this group of terrorists who once arrested decided to cooperate with the Italian police.

Regarding other cases, the literature addresses mainly the problem of the integration of the second generation of immigrants in the country (Algan et al., Reference Algan, Dustmann, Glitz and Manning2009; Wojtowicz, Reference Wojtowicz2012; Stevenson et al., Reference Stevenson, Demack, Stiell, Abdi, Clarkson, Ghaffar and Hassan2017), finding that the presence of a second generation of immigrants who are not integrated has a strong positive effect on increasing the likelihood of terrorist attacks. Finally, a relevant part of the literature focuses on the role of domestic intelligences in Western Europe in relation to their ability in the prevention of terrorist attacks. This is the case of Groppi (Reference Groppi2017), Samaan and Jacobs (Reference Samaan and Jacobs2018), and Shaffer (Reference Shaffer2016).

The relevant independent variables in this study are accounted in many other articles in the field of security. However, they are almost always taken separately and independently, whereas I argue that they interact with each other. In particular, to my knowledge no authors took in consideration the relevance and importance of the mosques, other networks, and their interaction with discriminatory dynamics toward minorities. That is the gap that this paper aims to close.

The research design

The paper uses a qualitative comparative case-study approach, focusing on three countries in Western Europe: Italy, France, and the United Kingdom. These cases are comparable because these countries share close relationships with the European Union, a common cultural background and host large Muslim communities (but with relevant differences in terms of numbers). The main difference between the countries in the study is in the number of fatal Islamist-related terrorist attacks they suffered. They also differ regarding how they regulate religion, and in the number of mosques and other networks on their territory.

The main hypothesis is that in a comparison of countries, those with higher level of social and religious discriminations toward Muslims are more likely to suffer Islamist-related terrorist attacks in the presence of socio-economic grievances in the Muslim community. If these grievances are brought to public debate through religious and cultural networks, this effect will be stronger. There is an interaction between discrimination toward Muslims and the number of networks where Muslims can gather. This interaction makes terrorist attacks more likely. In general, however, this interaction can enhance the mechanism, but the sufficient condition for the mechanism to take place lies in the discrimination, not in the number of mosques or religious or networks available.

This study uses several different databases: to measure the restriction of religion and other discriminatory factors, I use the Association of Religion Data Archive (ARDA) looking at the variable concerning societal discrimination of religion and regulation of religion. To examine discriminatory laws, I also consider news archive and official government sources to see if the countries discriminate against religious clothing, for example banning the wearing of the hijab. This prohibition makes the respect of religious principles impossible in some cases, as it forces (unwillingly) to violate religious prescriptions. This is possibly the strongest aspect of discrimination, because it goes beyond the fact that Muslims cannot publicly practice their religion. In addition to this element, bans on religious clothing make impossible to respect the laws of the host country and follow religious prescriptions at the same time in the aspects of daily lives, which is not the case for other forms of discrimination. The ARDA dataset is also used in this paper to identify the number of Muslims in the different countries (in addition to secondary sources if a regional breakdown is needed) and other quantitative data (percent of Muslim population, total population, and adherents to other religions).

To assess the increase in the likelihood of terrorist attacks, I use the Global Terrorism Database (GTD). In the analysis, I include all cases of fatal terrorist attacks related to Islamic terrorism after 9/11 and until 2019 included. I consider only cases entailing fatalities as these are more representative of the level of organization, scale and intent of the attack. I also consider cases where the perpetrator is not from the host country based on two main motivations: The first one is that there are only a few situations where this was not the case, especially if we consider the prolonged presence in Europe of foreign perpetrators. The second one is that even if the perpetrators were foreigners, the pre-requirement to carry out a successful terrorist attack entails the existence of a local support network. All the main terrorist attacks involved citizens of the attacked countries or of other European states. This paper focuses on the domestic aspect of the issue of radicalization, not on the strategy of terrorist groups. There is no event where the attack was entirely planned in another country outside of Europe and carried out only by foreigners and with no connections to local networks, which goes also against the theory of a direct impact of foreign immigration on terrorism.

The data regarding the number of mosques and other networks are more complicated to measure. The contentious issue here is what constitutes a mosque. If we consider the so-called ‘prayer rooms’ as mosques, the numbers are obviously larger than if we consider only the main official mosques. To assess the overall number of mosques in Europe I rely on the dataset built by Stefano Allievi (Reference Allievi2010, Reference Allievi2014), which however does not breakdown religious places of worships according to typology, which constitutes a major limitation for the dataset and the analysis. For this reason, I also rely on secondary media and government sources to identify, where possible, the number of official mosques. The distinction between the number of official mosques and prayer rooms might be relevant, because large official mosques have a larger audience, and therefore more potential to connect to existing grievances leading to radicalization, in situations where there is a high number of mosques of this kind. However, empirical evidence shows that radicalization can also happen in smaller and unofficial mosques, although the audience of these is smaller.

Is the nature of Islamist-related terrorist attacks transnational?

So far, I argued that the domestic grievances of the Muslim communities are likely to play a relevant role in the explanation of terrorist attacks. Yet, it might be argued, because of the strategic nature of the role of terrorist organizations (Kydd and Walter, Reference Kydd and Walter2006), the nature of these attacks might be merely international, and it could be mainly caused by decisions related to foreign policies of specific countries.

In that sense, it is worth to note two main things: The first one is that virtually all the Islamist-related terrorist attacks on European soil involved either local or other European citizens, even when the main organizers of the attacks were citizens of countries in the MENA region. This category of people is identified by Nesser (Reference Nesser2018) as the category of the ‘entrepreneurs’. However, entrepreneurs have to resort to ‘misfits’ of ‘drifters’ (ibidem) to carry on their plans. These categories are generally composed of marginalized people, often citizens, or at least long-time resident, of the country that the terrorist organizations want to target. Even when that is not the case, the grievance aspect toward the host society is still a driving factor of recruitment.

This paper focuses on the latter categories (‘misfits’ and ‘drifters’), as they are the one linked to the successful realization of terrorist attacks. It is clear that if there are fewer people to recruit in a specific place, or if there are other elements which are capable of making the creations of network more difficult, then the central organization of the terrorist group, generally based in the MENA region, would have more challenges in successfully carrying out attacks in those countries.

Nesser disagrees with this interpretation and downplays the importance of grievances in explaining the radicalization of ‘marginalized individuals’ (ivi: 269). In this paper, I argue instead that the concept of discrimination should be broadly intended and that religious discrimination, socio-economic discrimination, outrage at the foreign policy of the host country are all factors that are likely to cause grievances in the Muslim minorities across Western Europe and should therefore be considered.

The aspect of religious discrimination is also particularly relevant. Scandinavian countries know this well, as the only reason why they became a target for Islamist extremism was the publication of sarcastic cartoons portraying the prophet Mohammed in the Danish newspaper Jyllands-Posten in 2005 (ivi: 284). This is the same logic that inspired the Kouachi brothers in the Charlie Hebdo attack (ivi: 103–122; see also Vargo, Reference Vargo2021: part I). In this case the discrimination aspect of this dynamic lies in the fact that the Muslim community in Europe felt outraged by the publishing of these cartoons.

Explanations presented in the literature

The role of networks

So far, the explanations presented by the literature are not convincing. The main element of novelty in this paper is the hypothesis that in the presence of relevant grievances, places of aggregation can help radicalization of individuals because they can be a ‘multiplier’ of said grievances. The mechanisms of radicalization happen mostly through networking of like-minded individuals. This mechanism can take place in mosques, but also in other places with a high concentration of members of the minority holding grievances against the host state, such as prisons. Virtual places can also have the same effect, although there are not many cases where perpetrators radicalized online, for instance through social media (Thompson, Reference Thompson2011). It is also worth to highlight how the Imams of the mosques do not generally have a strong effect on radicalization. The main element enhancing grievances is not the way of preaching in the mosques, but simply the fact that several persons holding grudges against the host state can connect in one place

Cultural and Muslim student associations might have the same effect too, even though they are not related to religion in a direct way (Rabasa and Benard, Reference Rabasa and Benard2015: 103–122) and they are also less important as focal point of the community. Regardless, the main driver of the ‘echo effect’ is not necessarily the fact that the meeting place is a religious one, but it is rather the gathering of people who feel discriminated on religious and social ground by the host society in a specific place and moment in time. These networks can develop also in a more random way: For instance, prisons are known to have the same effect and recruitment through the jail network is fairly common, especially in France (ivi: 110). That is because the category of people that Nesser refers to as ‘drifters’ can come in contact with ‘entrepreneurs’, who can recruit them in the terrorist network while in prison.

Yet, the role of religious networks, as explained in the section regarding the attacks in the United Kingdom, often remains at the center of the logistical base for the plotting of terrorist attacks. Several perpetrators of the July 2005 attacks in London met in mosques, which is a dynamic that takes place also in other kind of Islamist-related terrorist attacks. The role of social networks is a bit murkier: Social media can be used as a recruitment and propaganda tool (Thompson, Reference Thompson2011; Gates and Podder, Reference Gates and Podder2015). Yet, social media and Internet tools are mostly used to recruit jihadis in the Middle East area (‘foreign fighters’) and very rarely the perpetrated attacks were realized by individuals radicalized on the Internet (the only case where such an argument can plausibly be made is the beheading of French teacher Samuel Paty in October 2020, BBC 2020).

Defining discrimination

Defining what constitutes discrimination is not easy. However, some basic assumptions can be made in terms of what the Muslim minorities in the three aforementioned countries care about: First, the presence of the ban on religious clothing is a clear indicator of religious discrimination as it prevents the possibility of following religious prescriptions. The French ban on religious clothing of 2004 is a clear-cut example of a discriminatory law. Second, socio-economic discrimination refers to the social and economic indicators of the minority group. These are not as straightforward to measure, but government reports and other official sources can give a general idea in terms of unemployment rates and other key measures. The further these measures are from the same indicators for the rest of the population, the higher the discrimination, as in the cases presented in the UK example. It is important to note that although the goal of the government is not to create a situation of disadvantage for minorities, the very fact that this situation exists is likely to cause a perception of discrimination. Finally, the levels of perceived discrimination are also addressed looking at survey data from the EU-Midis II survey, the results are analyzed in the sections on the countries.

These elements can coexist and overlap. For instance, according to a recent study (Abdelgadir and Fouka, Reference Abdelgadir and Fouka2020) the 2004 hijab ban in France had the effect of making teenage girls drop out of the French public school system, causing in the next decade an increase in the unemployment rate (this discrimination is both religious and socio-economic). The existence of multiple and overlapping grievances linked to this kind of discrimination is likely to increase the risk of terrorist attacks as a retaliation against specific laws/acts, a point also made by Sayia and Manchanda (Reference Sayia and Manchanda2019) who look at the effect of burqa bans in Europe on terrorist attacks through a large-N regression. Some studies in the field hint at this potential mechanism, which is analyzed more in detail for the European scenario in this paper: For instance, Gathak (Reference Gathak2016) shows that discrimination against minorities is indeed a significant predictor (worldwide) of domestic terrorism. The novel contribution of this paper is the finding that this mechanism can be magnified by the presence of networks, which facilitate the logistics and dynamics of recruitment for terrorist organizations.

France

France suffered the greatest number of Islamic-related terrorist attacks, according to the GTD. It also faces a deep contradiction between two elements: It is the country of secularism and historically has always been very clear in the distinction between church and state (despite a majority of Christian citizens) and at the same time it hosts a high number of Muslims, coming from former colonial ties, second- and third-generation immigrants and also immigrants coming from Africa and the Middle East with the latest waves. According to ARDA, 8.6% of the French population is Muslim (representing roughly 5.7 million people, mainly concentrated in great urban areas). This situation can cause serious problems, as the laicity of the state and an almost assimilationist approach is in stark contrast to the prescription of the Islamic religion. These links are also studied by Ivekovic (Reference Ivekovic2004: 1117–1119) from a more feminist perspective, as these prohibitions and discriminations regard mainly women, especially regarding rules for religious clothing, although other feminist authors argue that the Islamic prescription for religious clothing are inherently discriminating toward women in the first place (Chesler, Reference Chesler2010: 33–45). This is indeed true when the choice is not voluntary, but that impeding people from wearing symbols associated with their religions if they choose to wear them freely is a strong form of religious discrimination. It also results in a direct form of socio-economic discrimination, as shown by the example of the previously mentioned study by Abdelgadir and Fouka (Reference Abdelgadir and Fouka2020).

The hijab ban constitutes therefore a clear-cut case of discrimination. In 2004, the French Parliament passed a law which banned religious clothing in public school, including crucifixes, kippahs, and Islamic veil (hijab) (Sciolino, 2004). The law was framed sometimes as anti-Islamic in that prohibited the wearing of veil, but the rationale for the law was the laicity of the state, as it is important to note that all religious symbols were banned. However, it arguably affected the Muslim community more deeply as there were more Muslim girls wearing hijab than Jewish men wearing kippahs or Christian wearing (at least ‘ostensibly’, as the law says) religious symbols. There is also a gender issue, as the law affects mainly Muslims women. In 2010, a new law prohibited the wearing of the ‘integral veil’ (niqab) in public spaces. These issues raised also the concerns of groups such as Human Rights Watch, which saw these laws are an attack against the right of religious freedom (Human Rights Watch, 2014). On the 2016 EU-Midis II survey, 50% of Muslims with a Sub-Saharan African background and 46% of Muslims with a non-African background felt discriminated in the last 5 years (p. 29).

These grievances enter a context where Muslim communities are already in a disadvantaged position in terms of the role in society. In France, Muslims generally face higher level of unemployment, lower wages, and higher level of crimes (for the economic variables, see Algan et al., Reference Algan, Dustmann, Glitz and Manning2009). This mix of element (religious and social discriminations) led also in 2005 to the so-called Banlieu Riots, or riots in the suburbs (BBC, 2005).

France also hosts a high number of mosques. Official government data are lacking, and there is not always a clear distinction between prayer rooms, as the improvised temporary ones in the basements of residential buildings, and official mosques. According to the Allievi 2010 dataset, there are more than 2000 mosques. Among them, there are many Salafi mosques which preach more radical messages. Often, mosques of this kind have been shut down by the French police (Serhan, Reference Serhan2016). These Salafi mosques preached a message of violence and encouraged the jihad. Their radical message found a fertile ground in the grievances of the Islamic community at large, and the radical imams were able to channel frustration and anger into the jihadist propaganda. As a result of these laws, French Muslims took a strong stand against these bans (Reuters, 2021). It clearly appears that the combination of discrimination toward Muslim, bad socio-economic conditions and availability of religious networks caused outraged reactions in response to these discriminations, and all these elements also constitute an increase in the likelihood of a fertile ground for terrorist attacks in France.

Italy

Italy did not suffer any fatal Islamist-related terrorist attacks so far. Few authors and media journalists tried to answer why, since Italy is both an important country for Western culture in general and for Christianity (and Catholicism) more specifically. For Groppi (Reference Groppi2017), the answer lies merely in numbers. This ‘Italian exceptionalism’, as he refers to it, is therefore not meant to last long, and the rhetoric of ISIS in the ‘coming to Rome’ campaign had more a rhetoric value rather than representing an actual strategic planning of operations. In this sense, however, years after the publication of the article, this ‘Italian exceptionalism’ seems to be continuing.

Italy hosts roughly a million Muslims, representing 2.45% of the population according to the ARDA. The Italian Muslim community is not as large as the other in this study, but two things should be noted. The first one is the geographical concentrations of the Muslim community around great urban areas. More than a quarter of the Muslim Italian population lives in Lombardy, in the city of Milan and the surrounding outskirts. In Lombardy alone, there are 368 thousand Muslims, mainly coming from Albania and Kosovo, but increasingly also from Tunisia and Egypt (Polchi 2018).

Obviously, the absence of a second generation of Muslim immigrants is not clearly identifiable as cause of grievance for the lack of integration, as most of the community still consists of first-generation migrants, however in the next decades this indicator will probably become more relevant. The other overlooked aspect by Groppi is that there actually were plans for terrorist attacks in Italy, mainly in Rome and Milan, and they were successfully thwarted by Italian forces and intelligence services.

A distinctive feature of Italy, especially when compared to the French case, is the different interpretation of secularism. In fact, there is no mention of ‘secularism’ in the Italian Constitution and Articles 7 and 8 grant a special status to the Catholic Church, the only Church with agreements directly embedded into the Constitution (Italian Constitution, 1948).

This framework says that Italy is a somewhat secular country, but that Catholicism is clearly favored in terms of relation with the state. There is no law prohibiting the wearing of religious clothing in any case. At the same time, it is worth noting that Muslims can wear hijabs in school, possibly exactly because of a different interpretation of the role of ‘religious freedom’. That means that there is no direct discrimination against the prescriptions and rules of Islam at least regarding the private sphere of individual freedom. As of today, wearing religious symbols of any kind is not against the law in Italy (although it is forbidden to completely conceal the face for reasons of public order).

When it comes to building mosques, however, things change. In Italy, there are few large official mosques. Allievi counts more than 720 when including prayer rooms but is very hard to find large mosques with minarets except for a handful of cases. Author Maria Bombardieri counts only five mosques of this kind (Reference Bombardieri2011). Italy is also perhaps the country in Western Europe with more links to Catholic religion and the Vatican, and where religion still plays a relevant role in politics and social life in general. It is therefore no surprise that the planning for building mosques, sometimes made by the progressive left, are faced with strong bipartisan opposition (Talbot, Reference Talbot2006). In the Italian case, there is tolerance for the private sphere of the individuals and very low tolerance for public display of Muslim religion, meaning clearly visible signs of other religions in the public urban space (Allievi, Reference Allievi2010: 64). That does not necessarily apply to religious clothing, but it applies to places of worship, especially mosques following the traditional architecture and style of Arab countries but in a modern Western environment. The second element represents a permanent public sign of religion, an element which is not present in religious clothing, which although publicly displayed is temporary, and it also pertains to the individual sphere.

Preventing the construction of large mosques is a form of strong direct discrimination too. In fact, according to the EU-Midis II (EU-Midis II, 2017: 23–58) survey based on Muslim European population, Italy is the country where the perceived discrimination is the highest. However, this is a form of discrimination that does not entail the impossibility of respecting religious rules at the individual level. It is also a form of discrimination that prevents the establishment of networks. In France, grievances against the hijab ban in Salafi mosques were quite relevant, whereas the presence of this kind of more indirect discrimination is not facilitating the connection of grievances. The main point is that the leap from outrage to planning of terrorist attacks is harder for wannabe terrorists in the absence of networks in which their plans can resonate, even in the presence of high levels of perceived discrimination. This is not the say that a country can discriminate as long as it does allow the building of mosques: many terrorist plots were thwarted in Italy, and discrimination alone increases this risk. However, the absence of networks makes thing harder than in other countries. A common argument in the political debate is that the few official mosques in Italy are based on moderate principles and strictly cooperate with the state, whereas improvised prayer rooms attract more radical extremists. As a consequence, progressive politicians often argue for the construction of more mosques to counter Islamic extremism, on the basis that those are actually easier to control. The same logic was also shared by moderate conservative politicians and government officials (Strnieriinitalia.it, 2015). This discourse overlooks two important elements: it is true that as of today the Imams of the main Italian mosques have moderate peaceful views and cooperate with the Italian state and security forces, but that is possible because there are few mosques. If there were thousands of mosques it would clearly be more difficult for the Italian state and police to have the same level of control and cooperation with the Islamic community. The second problem is that intervening with police and special operation forces in large mosques creates a different impact than shutting down improvised garage rooms where Muslim pray. For instance, we can easily imagine that a large-scale police operation in the mosque of Rome, the biggest European mosque, will have a tremendous negative repercussion in the relationships between the Muslim community and the state. The argument that large mosques are more moderate and easier to control still holds, but only when the mosques are few, as the case of France clearly shows.

United Kingdom

The United Kingdom is second only to France with regard to the number of fatal Islamist-related terrorist attacks and number of victims. However, the scale of the phenomenon is somewhat different in terms of fatalities. The most notorious terrorist attack was the London public transportation bombing by Al Qaeda on the 7th of July 2005. Great Britain's colonial history explains why the numbers of the Muslim community are large, and also why the main Muslim community is from Pakistan and Bangladesh (Wojtowicz, Reference Wojtowicz2012) and from South Asia more in general. According to data, the geographical concentration of Muslims, representing around 3.4% of the UK population (ARDA), is also extremely geographically concentrated and clustered around metropolis, as in the case of Italy. Of the entire UK's Muslim population, 40% lives in London, and is made of more than 600,000 people (ivi: 8).

Regarding rules for religious clothing, the UK does not have a national law enforcing this kind of ban, but local towns, villages, and schools can do so. It might be the case to highlight how, according to a recent European survey (EU-Midis II), Muslims wearing traditional religious clothing in general feel more discriminated than those who do not (EU-Midis II, 2017: 30), although only 22% of Muslims with a Sub-Saharan African background felt discriminated in the last 5 years, and only 17% for Muslims with a Non-African background (p. 29).

If religious discrimination is arguably not as relevant as in France, the ARDA dataset still indicates a high value of religious societal discrimination for Western standards, 3 out of 10. However, the main grievances of the Muslims in the UK derive from social and economic disadvantages, as there are higher levels of unemployment, poor neighborhoods, and slums. In general, British Muslims do not own houses and score low on every social indicator compared to non-Muslim British (Wojtowicz, Reference Wojtowicz2012: 12). This set of problems also targets people at the individual level, and it is cause of grievances that can be linked to religion in radicalization processes, especially in cases of overwhelmingly geographical concentration creating ghetto dynamics and hindering integration processes.

The UK hosts a high number of mosques in all places where there are large Muslim communities. It is unclear from the data how many of these are prayer rooms and how many of these are official large mosques with minaret. According to Wojtowicz, there are 352 mosques in London alone (Wojtowicz, Reference Wojtowicz2012: 16). In all the main cases of Islamist-terrorist attacks in the UK perpetrators were British citizens affiliated to terrorist networks and regularly attending services at mosques. In the case of the July 2005 London bombing, three of the four perpetrators regularly met at a mosque in Beeston (Leeds) (Burke et al., Reference Burke, Barnett, Bright, Townsend and Thompson2005), even though they were banned by moderate Muslims because they were showing signs of radicalization. This highlights in clear detail how religious networks facilitate encounters among people with similar views, enhancing the risk of radicalism despite the generally strong opposition of the religious leaders. The larger the audience, the lower the level of control of the leaders of the institution. In Great Britain the number of mosques is large: even if the total (Allievi, Reference Allievi2010) represents relatively low numbers, the geographical concentration of mosques coincides with the presence of Muslim communities in a way that represents per capita number of mosques actually higher than what Allievi calculates (because he calculates it at a national level). The attacker of Manchester Ariana Grande's concert, Salman Ramadan Abedi, for instance, was a regular at the mosque of Disdbury, near Manchester (Evans et al., Reference Evans, Ward, Mendick, Farmer, Dixon and Boyle2017). As it can be seen, most of the perpetrators in the UK shared a somewhat similar story: in particular, they were born and raised in the UK and they attended mosques.

Foreign born vs. domestic perpetrators

So far, the paper did not focus on the difference between foreign-born perpetrators and radicalized domestic citizens. Although the literature stresses the difference between the two, it is extremely difficult in the recent Islamist-related terrorist attacks in Western Europe (after 9/11) to find perpetrators who are not domestically radicalized perpetrators (‘home-grown terrorists’). When the terrorists are not born in the state where the attack is perpetrated, often their place of birth is another European state. In most cases, terrorists are born on European soil. In the few cases where they are not, they still spent a significant number of years in the country where the attack took place. It is important to note that the reference here is to perpetrators on the ground, not the planners of the attacks, which indeed are more likely to take place abroad. Still, without people ready to carry out the attack in the country they live in, these terrorist attacks would be a lot harder to execute. The argument made here is that it is easier to find people willing to take part in these attacks in situation of discrimination and grievances in the countries where the attack is supposed to be taking place.

As an example, the suicide bombings in London on 7th July 2005, perpetrated by an Al-Qaeda cell, were planned in the Middle East, but looking closely at the perpetrators on the ground, three out of four were born and raised in the UK. The remaining one, born in Jamaica, moved to the UK when he was 5 years old (The Independent, 2015).

In the November 2015 Paris attacks, most of the Brussels terrorist cell was made by Belgian and French citizens. The 2015 January attack to a kosher supermarket in Paris were perpetrated by Paris-born terrorists, the Kouachi brothers. The list could go on. What is clear, once again, is that without the logistic support of domestic radicalized terrorists these attacks could not have happened. In fact, there are zero instances in which a foreign-born terrorist reached a Western European country and immediately proceeded to organize an attack. The fact that in virtually all the main attacks there is a heavy involvement of homegrown terrorists strengthens the hypothesis that domestic discrimination is at the center of this dynamic. Without a large support network, it becomes significantly harder for terrorists to carry out violence.

According to archival research (see the replication data on the associated dataverse repository for sources of the nationality of every single terrorist attack taking place in France and the UK), this was the distribution of homegrown and foreign-born terrorists summarized in Table 1: for every fatal Islamist-related terrorist attack in the GTD for France and the UK since 2001, I looked up either reliable media sources or reliable outlets to address the places where the perpetrators were born. The dataverse repository has the list of the attacks, the list of the perpetrators, and every source I used in relation to that attack.

Table 1. Place of birth of the perpetrators of fatal Islamist-related terrorist attacks in France and UK

This distribution makes it clear that without the support of the homegrown jihadis, the majority of the attacks would not have occurred. Conversely, the fact that almost the totality of the attacks was perpetrated by either homegrown terrorists or immigrants living in the country since childhood also signals the relevance of domestic grievances in explaining radicalization processes. In the absence of grievances against the host state, the planners of these attacks could not have relied on their foreign network to carry out the attacks.

Comparisons and results

The result of the study is summarized in Table 2. Regarding the countries with the highest values of societal discrimination of religious minorities and regulation of religion, higher scores are found in countries with more terrorist attacks. The additional variable regarding discrimination is based on national and local bans on religious clothing, which prevents the respect of religious principles and therefore constitutes the strongest level of discrimination toward Muslim communities, as argued in the previous paragraphs. I coded this variable, assessing whether the country has laws regulating religious clothing, and the extent of these bans. France is the only country banning hijabs in public schools; the United Kingdom delegates the matter to local authorities. However, this ban is likely to affect a very low number of people as the majority of Muslim women does not wear integral burqas, and those who do are unlikely to work in one of the aforementioned places. In Italy there is no such law, even though the burqa is theoretically banned because the Italian laws prohibits to walk in public places in a way that completely conceals the identity of the person for reasons of public safety. The conditional effect of mosques also appears clear looking at the 2010 Allievi dataset: Italy has similar numbers of mosques per Muslim per capita compared to France and the UK but very few large mosques as explained earlier, and lower levels of discrimination in terms of clothing ban and religious discrimination and regulation according to the ARDA.

Table 2. Discrimination, mosques and fatal Islamist-related terrorist attacks in Italy, France and the UK

^a Variable coded by the author, explanation in the text.

^b Not including attacks where the perpetrator was the only victim.