Extremist and pro-violence attitudes of Spanish adolescents in secondary schools

Abstract This article distinguishes between attitudes towards extremism and attitudes towards violence and illegal acts. It operationalizes both groups of attitudes based on the proposed scales by Ozer and Bertelsen (2018), which are validated for the Spanish case. The combination of the two sets of attitudes favours the use of violence to pursue radical ideological, religious, or political views. The literature on the subject is reviewed, and the results of an exploratory study based on a questionnaire administered to adolescents attending 4 schools in Almería and 1 in Madrid (n = 1170), carried out in May 2022, are presented. The results of two multiple regressions on the indices of extremism and pro-violent and illegal acts, allow us to conclude that the boys, the youngest, of foreign parents, of mothers with a low educational level, right-wing, those who most frequently attend religious centres, show more extremism, and pro-violence and illegal acts attitudes. Those who believe that there is discrimination against LGTBI+ people are more extremist but not more pro-violence. Finally, those who belong to poor families, use many social networks a day, and have suffered physical violence show more pro-violence attitudes and illegal acts, but not more extremism. These results suggest that it is necessary to prioritize transforming violent extremism, especially among the most vulnerable groups.

The authors are PhDs in Sociology, Education and Psychology, and are professors at the University of Almería (and Málaga) where they teach Gender and Migration, Sociology, Educational Theory and Social Work. They have published numerous high-impact articles. This study is part of a larger investigation. Two books have been published in Spanish. The first is aimed at secondary school teachers and is entitled Transforming violent extremism (Octaedro Publishing House). The second book collects the results of the diagnosis and the intervention and is entitled "Intervention program in radicalization and violence in adolescents. Diagnosis and implementation in educational centers" (Delta Publishing House)

PUBLIC INTEREST STATEMENT
Our multidisciplinary research group (Investigación Internacional Comparada) is interested in exploring current social problems and in elaborating proposals of intervention to tackle the identified issues. This article presents the results of our survey on extremist and pro-violence attitudes of Spanish adolescents in secondary schools. The main objective of our research is to better understand the development of dangerous attitudes in teenagers in order to develop an intervention programme for preventing these attitudes. Our research allowed us to identify certain characteristics of young people which increase the possibility of developing extremist and pro-violence attitudes, such as male gender, foreign parents, mothers with a low educational level, right-wing, those who most frequently attend religious centres, among others. Interventions aimed at preventing extremist attitudes should focus on adolescents with these features.

Introduction
Violent extremist views and attitudes can manifest around a range of issues, such as politics, religion, cultural diversity, sexual orientation, or gender relations, producing political and social alienation. Such views could be catalysts for radicalization, especially among the most disadvantaged groups (Trujillo et al., 2006). No society, religious community or worldview is immune to such views, but it is of particular concern among adolescents, as adolescence is a clear risk factor for involvement in radicalization activities (UNESCO, 2016). As the 2020 UN report Violence against Children points out, children, especially adolescents, are more likely to engage in risky behaviour and attitudes than adults. Advances in understanding brain maturation and developments in neuroscience have highlighted that adolescents take risks because their increased appetite for novelty and sensation seeking precedes the growth of their self-regulatory competence and cognitive control system (UN, 2020, p. 9). In other words, adolescents have not yet fully developed their cognitive control system, and this makes it difficult for them to control their impulses or develop resistance to peer pressure. They are experiencing a phase of turbulence and reorganization, so that we can say that "adolescence per se is a risk factor of radicalization" (Campelo et al., 2018, p. 8). Rebellion and the desire to differentiate from parents as a way of asserting one's identity can be exploited through a mechanism of dehumanization by those who promise a community with strict fundamentalist rules, which asserts the existence of The Truth and promises a New Social Model (Doosje et al., 2013). But the increase of extremist and pro-violent attitudes (Rozado & Kaufmann, 2022) is not only related to the fact that their bearers are experiencing adolescence. There are other factors that help to understand why some adolescents are more extremist and pro-violent than others.
In this article we present the results of a research on violent extremist attitudes of adolescents in the provinces of Almería and Madrid (Spain). Strong migratory movements characterize Spain, and particularly Almeria. As a result, it is worth noting the high percentage of foreigners among the residents. The origin of the immigrants has changed over time, going from a provincial or interprovincial migration that came to integrate the small agricultural business, to a central or North African immigration with marked features of marginality and very difficult to quantify. In the province of Almería, the immigrant population represents more than 40% of the total in some localities, where three of the secondary state schools where we have carried out this study are located, with the Moroccan group being the most representative since it constitutes more than 90% of the immigrants in these agricultural areas. This area has been characterized by high birth rates that today have tended to homogenize with those of the surrounding area, although they are still slightly higher. As a result, it has a very young age structure, with ageing rates among the lowest in the country, with a high number of adolescents above the average. One of the schools we have included in this study is located in Almeria city and is public. In this case, the percentage of adolescents who are children of immigrants is lower. The secondary school in Madrid that we have included in the study is located in a poor neighbourhood of the capital and is a state-funded private school. In all of them, there is a high percentage of children of immigrants, and the families of the adolescents belong to the middle or lower-middle class. Both the province of Almeria and the suburb of Madrid are characterised by a high concentration of vulnerable populations, characterised by low income, a very high percentage of children of immigrants in secondary schools, and a lack of social integration (Capote & Nieto, 2017) which has led the authors to call these environments vulnerable environments (Lobato et al., 2021, p. 52), in which people may embrace violent groups (Gomez et al., 2021), fundamentalist and far-right alike.
Further to migratory waves in Almería and Madrid, it is worth highlighting the important economic downturns which Spain (and Europe) has suffered from 2007/2008 until the current economic crisis derived from the COVID-19 pandemic and the impact on young people and their families. The economic impact has been particularly severe on family savings, and families´ standards of living. Hence the importance of taking into account the economic factor on the process of youth radicalization.
We explore some socio-demographic characteristics of young people (gender, age, parents' nationality, family's ability to save money, mother's level of education, attendance to religious worship centres, political ideology, number of social networks used on a daily basis), together with their perception of discrimination against certain groups (LGTBI+, poor families and immigrants) and exposure to physical violence (polyvictimization). Attendance at religious services and political ideology were introduced with the intention of detecting whether there was a significant radicalisation phenomenon among adolescents. We will try to elucidate to what extent these variables help to understand extremist and pro-violence and pro-illegal acts and attitudes.

Violent extremism
Violent extremism refers to the beliefs and actions of people who support or use violence to achieve radical ideological, religious, or political views. It brings together two aspects: the endorsement of extremism and the acceptance of violent means. Radicalization can be understood generically as the process of acceptance-and/or practice-of extremist ideology and action, as what defines violent extremism is the emphasis on the use of violence as a means, rather than the extreme views themselves (Neumann, 2013). However, cognitive radicalization does not have to lead to behavioural radicalization. Moreover, when we focus on cognitive radicalization, extremist attitudes need not necessarily lead to violent attitudes. In addition, suffering from other physical violence (in the peer group, family, etc.) may also influence the pro-violence attitudes of adolescents, as we will see later.
In fact, the concepts of radicalization and extremism are exposed to transformations in political, economic, social and security dynamics (Winter et al., 2020). The concept of radical extremism or violent radicalization has been widely used in recent years in connection with final behaviours of a terrorist nature (Abay Gaspar et al., 2020;Berger, 2016), which has to do with an intention to use the definition as a contribution to greater knowledge and its practical use in security and defence. But it has also shown the need to provide a convincing definition of the difference between radicalization as an attitude and its embodiment in actual violent behaviour (Pugliese, 2018). For not all harmful or violent actions stem from violent radicalization, not all terrorist actions embrace radical ideologies, and therefore violent radicalization should not be confused with terrorist behaviour (Francis, 2014;Horgan, 2009;Striegher, 2015).
There is no single profile or pathway for an individual to experience a process of behavioural radicalisation, nor can we even anticipate the speed at which it occurs. Nor does the educational level appear to be a reliable predictor of vulnerability to radicalisation. Campelo et al. (2018) point out that there are macroenvironmental factors (related to geopolitics or religiosity), micro-environmental factors (related to family dysfunction or being friends with radicalised individuals), and individual factors (psychological vulnerability, experiences of abandonment, psychiatric conditions, personal uncertainty and perceived social injustice) that, from a triggering event, lead to violent behavioural extremism. The point in the life cycle-adolescence -is a clear risk factor for involvement in radicalisation activities.
Moreover, cognitive radicalization is a necessary but not sufficient step for the activation of the individual. In itself, it does not mean that all cognitive radicalization ends in violent actions, in fact, in behavioural radicalization. Sageman (2017) maintains that cognitive radicalization is an ideological aspect that maintains and develops subsequent attitudes and subsequent actions in the individual, even if massively radicalized thinking does not result in violent actions. On the other hand, in the processes of behavioural radicalization, such as those related to terrorist groups, there is an attempt to promote attitudes of understanding and justification of the violent acts carried out, which increases extremist and pro-violence attitudes (Rodriguez et al., 2022).
Cognitive radicalization is related to behavioural radicalization also in the sense that proviolence attitudes are related to the experience that individuals have with online violence. At this point we would like to mention the role played by the public dissemination of violent attitudes, especially that which occurs in social networks through hate speech. In this line, Kruglanski et al. (2013) highlights that violence is evident in the content of hate speeches as an intermediate step between cognitive and behavioural radicalization.
Several authors have developed the idea that violence in social networks constitutes an effective violence that goes beyond the cognitive but does not reach the behavioural, remaining at an indeterminate point, apparently somewhere in between. Following these authors, Finkelhor et al. (2000) defined violence in social networks as threats and offensive acts that are sent or posted against a young person for other young people to see with aggressive intent. In the same vein, Besley (2009) highlights the concept of moral violence to underline the use of information and communication technologies, by individuals or groups, to commit deliberate, repeated and clearly hostile acts to morally harm others. The same is true for the use of insult and threat as a serious mechanism to damage the image of a person or collective (Hinduja & Patchin, 2008, 2010Juvoven & Gross, 2008;Li, 2008). Kowalski et al. (2012), elaborate on this step of cognitive violence developed on the internet as different from traditional violent bullying for four reasons: 1) it can be done anonymously; 2) it can be committed and suffered at any time; 3) it can create the fear of losing internet access for victims (who therefore refuse to report their attackers); and 4) it very often gives way to inaction by witnesses. In the same vein, Boyd (2014) shows that the everyday use of social networks by teenagers takes place because there is easy access to online content (especially through the use of keywords), which allows for a shift from ease of creating violent content to actual harm caused towards victims, even if the content is not perceived as destructive, on many occasions, by the young people who develop it. In any case, the authors conclude that the Internet can reinforce extremist ideologies through messages and narratives, creating avenues for adolescents to join networks where extremist ideas are spread, thus contributing to the normalization of violent behaviour (Amit et al., 2021;Hassan et al., 2020).

Studies about extremist and proviolence attitudes in adolescents
Over the last few years some research on extremist and pro-violence attitudes of adolescents have been carried out internationally. In the context of our research project, the following studies, carried out in Belgium and Switzerland, as well as a meta-analysis of research on cognitive and behavioural radicalization seemed especially interesting, because, even though focusing on slightly different topics and using different methodologies, they offer important insights, which the design of our own survey took into account.
These studies directly use-or are inspired by - Moskalenko and McCauley's (2009) scales on Activism and Radicalism. Moskalenko and McCauley (2009) introduced the Activism and Radicalism Intention Scales to understand how individuals move from political attitudes and beliefs to political action, including political violence and terrorism. The Activism Intention Scale assesses readiness to participate in legal and no-violent political action, whereas the Radicalism Intention Scale assesses readiness to participate in illegal or violent political action. These are two correlated but distinguishable dimensions. They tested the scales in ad-hoc samples of U.S. and Ukrainian undergraduates (n = 140 and 146), and in an Internet panel survey representative of the U.S population (n = 429). They conclude that there are two views of the relation between activism and radicalism. In the first, there is a single dimension of political action that runs from passivity to activism to radicalism. In the second, radicalism is an appraisal that justifies or even requires political violence as the only possible path to political change. Only a small minority (from 11 % to 17%) of those with activist intentions have also radical intentions. The authors conclude that the perpetrators of political violence, including terrorist violence, are the apex of a pyramid of sympathizers and supporters. Frounfelker et al. (2019) explored the nexus between discrimination and radicalization among adolescents and young adults in Belgium. The authors assumed that there would be an association between discrimination based on political views and religion/faith and support for violent radicalization. Furthermore, they thought that a connection between experiencing discrimination during interactions with the criminal justice system and seeking employment and endorsement of radical action would exist. Therefore, the authors carried out a survey intended for identifying the mentioned context. 2,037 young adults between the ages of 16 and 30, mainly students in secondary schools in the French speaking Brussels Capital Region, the Dutch speaking Antwerp city region and elsewhere in Flanders participated in the study. The sociodemographic data of the participants were registered, as well as their answers concerning their perceived discrimination. Furthermore, their replies to questions belonging to the Radicalism Intention Scale (RIS) developed and validated by Moskalenko and McCauley (2009) were correlated to these data. More than a third of all participants (37.6%) reported experiencing discrimination; among them, more women than men (41.1% vs. 34.3%, respectively). A majority of Muslims (55.6 %) declared having perceived discrimination, but only 30.6% of those who reported not being part of any other listed religion. In what refers to sociodemographic characteristics and radicalization, Frounfelker et al. (2019) indicate that "men and respondents who did not speak Dutch or French as their first language had higher scores as compared to women and those who spoke one of the official languages'' (p. 30). The authors also found a correlation between discrimination and radicalization: "Experiencing any discrimination was associated with higher RIS scores. There was also an association between number of reasons for discrimination and higher RIS scores, with individuals reporting five or more reasons for discrimination having higher scores than those reporting four or less" (p. 30). Individuals who had perceived discrimination due to political views showed the highest RIS scores. In the discussion, the authors highlight that, whereas the identified relationship between having perceived discrimination and a tendency to supporting extremist behaviour does not seem surprising, the lack of finding a relationship between discrimination based on religion/faith and support for radical action in their survey seems surprising. This result is at least valid in the context of Belgium. Having equally found a positive relationship between having grown up with a first language different to one of Belgium's official languages and higher scores on the RIS scale, Frounfelker et al. (2019) suggest enhancing efforts to recognize linguistic diversity in Belgian schools and fostering language inclusion as antidotes for perceptions of discrimination that would consequently diminish intentions for radical behaviour.
A Swiss research group (Kamenowski et al., 2021) recently presented a study of religion as an influencing factor of right-wing, left-wing and Islamist extremism. The approach of the authors distinguishes between religious affiliation and religious attachment and tries to measure the influence of these two factors on extremist attitudes among young people. The survey was carried out in 10 of Switzerland's 26 cantons and above all aimed at young people aged between 17 and 18 years. A total number of 8,317 students took part in the survey; the gender ratio was balanced and there was a high proportion of young people with a migration background (51.2%). The analysis showed that religious affiliation correlates to extremist attitudes in different ways. A "Christian religious affiliation is associated with a slightly higher acceptance of right-wing extremist as well as Islamist-extremist attitudes" (p. 13). Furthermore, being a Muslim is accompanied by an approval of Islamist extremism. In what refers to Muslim youth, the finding of the survey is that the more religious they are, the more important extremist attitudes are to them. Claiming religious exclusiveness, as well as religious intolerance, is also accompanied by a higher agreement with extremist attitudes among Christians and Muslims. Taking into account gender, the study finds that "male adolescents are more open to extremist attitudes" (p. 14). On the other hand, the survey could not identify a clear correlation between the association with delinquent friends and extremism, which is often mentioned in criminological literature (for instance, Wolfowicz et al., 2021). This latter finding is not consistent with the results of another survey carried out in Switzerland (Nivette et al., 2017). The authors of this study tried to examine the influence of collective strain on support for violent extremism. The sample consisted of a group of ethnically diverse Swiss adolescents; the analysis focused on the data collected in two waves of surveys conducted with 1,675 children, of which a large proportion had a migration background. The theoretical background of the research was the general strain theory, which states that exposure to different types of strain, such as interactions perceived as unjust, the removal of positively valued stimuli or negative experiences with parents, peers, police, and employers, can produce negative emotions, which demand corrective action. According to this theory, crime is a type of corrective action that seeks to injure, damage, or seek revenge on the presumed sources of strain (p. 759). The analysis came to the conclusion that collective strain in the age group 15-17 years did not have a direct effect on violent extremist attitudes. As the other mentioned studies also pointed out, males are more likely to support violent extremism. In this context, the authors state that "adolescents who justify the use of violence more generally and who dismiss the 'bindingness' and legitimacy of the law are more likely to support the use of violence to achieve political, social, or other ideological goals" (p. 778), i.e., teenagers, who justify violence and rule breaking are more vulnerable to extremist violent pathways, especially when vicariously exposed to conditions of collective social and economic strife, conflict, and repression (p. 779). Bliesener et al. (2021) take a similar line of reasoning in a study of the development of ideological beliefs, acceptance of violence, and extremist behaviour. A survey conducted with 6,700 9 th grade students from Germany showed that different extremist ideologies (right-wing, left-wing, Islamism) covary on the level of attitudes, acceptance of violence and extremist behaviour. Further analysis confirmed that a model of radicalization patterns starting at the cognitive level, followed by the acceptance of deviant means to achieve ideological goals, and finally leading to extremist acts seems to be a reasonable scientific assumption.
A comprehensive meta-analysis by Wolfowicz et al. (2021) states that factors known to be predictive of criminal attitudes are related to radical attitudes. The criminogenic factors also point to the usefulness and relevance of criminological perspectives in understanding and explaining radicalization. For example, juvenile delinquency, a criminal history, contact with the police, low self-control, thrill-seeking and risk-taking, similar peers, deviant-radical peers, moral neutralizations (and dehumanization) are factors associated with self-control theory, social control theory, as well as social learning theory, each of which have been suggested as useful frameworks for understanding radicalization. In addition, the authors consider it likely that individuals who experience a loss of self-importance for various reasons will seek to regain this importance, considering violence as a possible means.
A relatively small amount of research studies the factors responsible for radical behaviour. According to the authors' meta-analysis, the same factors identified as causing radical attitudes seem to play a role; additionally, factors such as (male) gender, unemployment and recent job loss are mentioned. Similarly, current and past military experience may be a factor that carries a higher risk of producing radical behaviour. However, Wolfowicz et al. diagnose a lack of research that satisfactorily explains why an individual with radical attitudes and intentions becomes a person who commits radical acts.
Even though referring to different countries and focusing on the correlation of partially different factors with the development of attitudes in adolescents, which could be classified as extremist, pro-violence and pro-illegal, the reviewed research and meta-analysis hint at questions that the design or our own tool had to take into consideration. First, the relationship between holding extremist and pro-violence attitudes and gender should be examined, because all studies mention that male adolescents are more likely to develop such attitudes than their female peers. Furthermore, it seemed recommendable to examine the context between religious affiliation and attachment and the acceptance of extremist attitudes. Next to these factors, we should be able to examine the correlation between experiences of discrimination and victimization and the likelihood of developing such attitudes. Registering socio-demographic factors such as the educational and economical level of the parents, as well as their cultural background, also seemed paramount.

Scales about extremism and pro-violence attitudes
A series of studies published by Moskalenko and McCauley (2009) on mechanisms of political radicalisation and mobilisation has received considerable attention in the academic world. In our context it is interesting that the authors deliberately distinguish between political activism and radicalisation. The scales on activism and radicalism which they applied to adolescents in the US and Ukraine allowed them to diagnose both attitudes as separated dimensions. Whereas political activism refers to the readiness to participate in legal and no-violent action, political radicalism implies a willingness to participate in illegal or violent action. Even though slightly interconnected, both attitudes are not the same; Moskalenko and McCauley's research only found a small minority of adolescents who hold at the same time activist and radical intentions. According to the authors' pyramid model, only a very small group of individuals, supported by larger groups of supporters and sympathizers, recurs to political violence. Further recent research developed these fundamental ideas of the authors and served as a model for the design of our own survey.
The Frounfelker et al. (2019) scale contained four items related to radical activism and assessed an individual's willingness to engage in illegal and violent behaviour on behalf of their group or organization. Kamenowski et al. (2021) tried to measure extremist attitudes (right-wing extremism, left-wing extremism, and Islamist extremism) with a different scale. Nivette et al. (2017) measured support for violent extremism defined as attitudes that "encourage, endorse, condone, justify, or support the commission of a violent criminal act to achieve political, ideological, religious, social, or economic goals" (Nivette et al., 2017, p. 757). In our study we were interested in measuring extremist attitudes, but also pro-violence attitudes, so we opted for the scales proposed by Ozer and Bertelsen (2018). Ozer and Bertelsen (2018) highlight the importance of developing and validating two generic scales for empirical research on violent radicalization, pertaining to (1) endorsement of extremism and (2) acceptance of violent and/or illegal means. The main reason why they propose to analyse separately extremism and pro-violence attitudes is that, according to their opinion, phenomena such as "terrorism", "radicalization" and "extremism" have become increasingly the object of research in the social sciences, but the corresponding terms have not yet been exactly defined. In this, their diagnosis is shared by Austin and Giessmann (2018, p. ii) who state that "no unified or even very explicit understanding exists of how to define the phenomena of violent extremism, armed radicalization and terrorist activity." Ozer and Bertelsen (2018) insist on distinguishing between ideology and violent action because the development of radical attitudes does not necessarily lead to actual acts of terrorism; insofar, "radicalization can generically be understood as diverse processes of gradually accepting extremist ideology and action (p. 1)." Their concept of extremist attitudes considers them related to two interdependent dimensions: (1) endorsement of comprehensive personal and societal change and (2) intolerance toward other groups of citizens objecting to these societal changes. The construction of the two scales is based on their assumption that extremism in itself is not accompanied by violent and criminal acts, so that most extremist attitudes are completely legal; therefore, the acceptance of employing violent/criminal acts should be operationalized independently from extremism. This procedure would allow, according to the authors, to distinguish between extremism within and beyond the law, avoiding the confusion of legal extremism with extremism employing illegal means.
Ozer and Bertelsen (2018) denominated their scales (a) Extremism Scale and (b) Pro-violence and Illegal Acts in Relation to Extremism Scale. We reproduce the two scales in Tables 1 and 2 below. The first scale was developed to measure the attitude toward comprehensive sociocultural change and intolerance toward others through group dynamics, following the mentioned definition of extremism. The first component included lifestyle and culture, socioeconomic founding, and governmental system, whereas the second component referred to us-them distinction, devaluation of others, breakdown in deliberation, 2. If one can´t live with the majority´s lifestyle and culture, it is necessary to create a totally different lifestyle and culture for oneself and ones like-minded.
3. It is necessary to totally change the economic system that is the basis of society.
4. Those who think like me have to thoroughly change the foundation of our own life (economy, job, consumption, well-being). The rest of the society can do what they want.
5. It is necessary to do away with the democratic form of government if we want to have a decent society.
6. Just let the rest of the society choose democracy-I, and those who think like me, work to establish up a different system in our own milieu. 7. I, and those who think like me, in fact share nothing with the rest of the society.
8. There is only one way to live the good and correct life.
9. If one doesn´t live in agreement with the good and correct life, then one has chosen to withdraw from the community.
10. Those groups in the society that don´t support the good and correct life should be deprived of their rights.
11. It is a waste of time to try to find common solutions with those whose thoughts about life are completely different than ours.
12. It is wrong to make compromises with what oneself stands for.
13. It is wrong and immoral to live peacefully side by side with people who don´t live the good and correct life.
14. In the end, there must be a confrontation-one can´t forever live peacefully, side by side with people who live a completely different life than they are obligated to live.

Table 2. Items of pro-violence and illegal acts
1. Using physical violence is the only thing that really works when it is a matter of . . . creating proper conditions for those with whom one feels a solidarity.
2. Using physical violence is the only thing that really works when it is a matter of . . . creating a new and better society.
3. Using physical violence is the only thing that really works when it is a matter of . . . creating proper conditions for those one is closely connected to. 4. Using physical violence is the only thing that really works when it is a matter of . . . creating respect for one's own rights and security. 5. Using physical violence is the only thing that really works when it is a matter of . . . preventing repression and assault of my people.
6. Using physical violence is the only thing that really works when it is a matter of . . . advancing a higher cause (ideological, religious). 7. Breaking the law is the only thing that really works when it is a matter of . . . creating proper conditions for those with whom one feels a solidarity.
8. Breaking the law is the only thing that really works when it is a matter of . . . creating a new and better society.
9. Breaking the law is the only thing that really works when it is a matter of . . . creating proper conditions for those one is closely connected to.
10. Breaking the law is the only thing that really works when it is a matter of . . . creating respect for one's own rights and security.
11. Breaking the law is the only thing that really works when it is a matter of . . . preventing repression and assault of my people.
12. Breaking the law is the only thing that really works when it is a matter of . . . protecting a higher cause (ideological, religious). and the impossibility of coexistence. The second scale was designed to measure attitudes of proviolence and acceptance of using illegal means in regard to extremism centred on various domains, such as society, a higher cause, family and friends, and group. Both scales were tested with high school students in Denmark and the United States and produced coherent results; besides background information about age, gender, and country of birth, the measurements contained five self-report measurement scales: The Extremism Scale consisted of 14 items scored on a 7-point scales, and the Pro-violence and Illegal Acts in Relation to Extremism Scale included 6 items pertaining to accepting the use of violence and 6 relating to the acceptance of using illegal means. Both scales range from (1) Strongly disagree to (7) Strongly agree. According to the authors, both scales confirmed the initially supposed distinction between activism and radicalism; furthermore, they considered the Extremism Scale being able to identify extremism by the criteria of a deeply felt wish to change life conditions and setting aside concerns about human coexistence. In the next epigraphs, we present our own survey and replicate Ozer & Bertelsen´s two scales (ES and PIARES) in order to regress them on sociodemographic variables and violence and discrimination scales and variables.

Objectives and hypotheses
This research has as one of its main objectives to apply Ozer and Bertelsen (2018) scales (ES and PIARES) to Spanish adolescents. In the case that the scales would be validated, a question arises as to which are the sociodemographic characteristics of adolescents who express extremist and pro-violent attitudes.
Hypothesis 1 (H1). If Ozer and Bertelsen (2018) scales would be validated for adolescents in Denmark and the US, then these scales should also be valid for Spanish adolescents (in secondary schools of Almeria and Madrid).
The research reviewed highlights that factors such as male gender, having suffered discrimination, belonging to the extreme right, religious attachment, exposure to different types of stress, as well as unemployment may constitute risk factors for violent radicalization. It is also to be expected that those who consume more social networks will develop more extremist and pro-violence attitudes.
Hypothesis 2 (H2). Regarding the sex variable, boys will score significantly higher than girls for ES and PIARES. Adolescents with a foreign parent, those who are younger, those from low saving families, those of mothers with lower level of education, those who more frequently attend religious services, those who are more on the right of the political ideological scale, and those who use a higher number of social networks on a daily basis will score significantly higher in ES and PIARES than adolescents with a Spanish parent, those who are older, those from high level of saving families, those of mothers with higher levels of education, those who less frequently attend religious services, those who are more on the left of the political ideological scale, and those who use a lower number of social networks on a daily basis.
In our research design we did not include questions on delinquency, but we did include questions on physical violence suffered by adolescents. As a third objective, this research proposes to explore the relations of extremist and pro-violent attitudes with the adolescents´ experiences of polyvictimization. For this purpose, we have designed a battery of questions about the persons from whom they have received violence (friends, father, mother, classmates, intimate partner, and unknown people) and construct an index about polyvictimization.
Hypothesis 3 (H3). Adolescents who have suffered more violence (poly-victimization) will show significantly more extremist and pro-violent attitudes.
Research has highlighted the relationship between violent extremism and perceptions of social vulnerability. The fourth objective aims to explain whether the perception of discrimination against the LGTBI+ group, low-income peers and immigrants are related to higher extremism and proviolent attitudes.
Hypothesis 4 (H4). The adolescents with a higher perception of discrimination will show significantly more extremist and pro-violent attitudes.
As commented above, Ozer and Bertelsen (2018) suggest measuring extremism and pro-violence attitudes separately, proposing two scales. Therefore, it is likely that some of the variables associated with one of the scales will not be statistically significant for the other.

Data
Data in this research were collected from students in public and state-funded private secondary schools located in Almería (4) and Madrid (1) during May 2022 in the context of a research project about violent radicalization of adolescents. The total number of participants in this study was 1,170, which fulfilled the rule of thumb and conventions for sample size in multiple regression analysis with eight independent variables (Maxwell, 2000). The database, together with the questionnaire and a methodological note, is available in the repository of the University of Almeria (http://repositorio.ual.es/handle/10835/14093).
The questionnaire was answered by all the adolescents present in the classrooms with the prior consent of the School Councils of the Centres. Professors and members of the research team were present. Researchers explained the general guidelines of the questionnaire, ensuring the adolescents that their answers would be treated anonymously. In order to answer the questionnaire, the research team provided the adolescents with a QR code which they used to answer the questionnaire in their mobile phones for approximately 30 minutes. In case they did not have mobile phones, the questionnaire was provided on paper.
The questionnaire was carried out in Spanish, even though we were aware that some of the adolescents spoke two languages, one at school and the other at home. In Almeria there is a population from more than 100 countries, which made it difficult to translate the questionnaire for each of the foreign languages. During the administration of the questionnaire, there were some terms that the adolescents did not understand. The teachers present in the classroom helped to explain the meaning of these terms.
Although the previous intention of the research team was to reach all groups in the secondary schools from the second year of secondary school and above, some groups had to be excluded because, in the two designated days for the fieldwork, these groups had scheduled curricular activities that prevented them from answering the questionnaire (tests, etc.). Moreover, in some cases it was impossible to carry out the survey with some adolescents, as they had just arrived in Spain and had a total lack of knowledge of the language. In one of the schools in Almeria, we carried out the survey in two groups of first year secondary school students, as we wanted to check if they would understand the questionnaire well. The results of the data analysis for these two groups suggest that there are no major differences between them and the rest of the courses.
The questionnaire included sociodemographic questions (sex, age, parent´s nationality, family's level of saving, mother´s highest educational level, frequency of attendance to religious services, self-positioning in political ideological scale, number of social networks used on a daily basis) and questions about the adolescent´s integration in the secondary school, their perception of discrimination and violence, their attitudes towards violence and the use of violence. All Ozer and Bertelsen (2018) questions were included in order to reproduce the scales: ES and PIARES (see Tables 1 and 2). The translation into Spanish was carried out following the indications of Beaton et al. (2000), with a literal translation from English to Spanish by a translator who is part of the research group (knows the object of the translation), and then a reverse translation (from Spanish to English) by a translator who does not know the object of the work (naïve translator). Both translators compared their work based on semantic and idiomatic equivalence criteria.
There are some aspects in the reviewed studies which we did not consider in our questionnaire, as the association with delinquent friends and extremism is a factor which is frequently mentioned in studies carried out from the point of view of security held by the police and other armed forces but does not seem relevant for research that tries above all to diagnose the development of radical and violent attitudes in teenagers. On the other hand, we also have identified a factor which has not been considered in the quoted studies: The frequency of the use of social networks.

Results
Before analysing the socio-demographic characteristics of secondary school students, we carried out a multilevel regression analysis, taking schools as the subject of analysis. Although we had an important variety of schools (state/private-funded, in the West of the province of Almeria and in the city of Almeria and in a suburb of Madrid) we could not refute the null hypothesis, so we decided to continue without taking into account the distribution of the sample by school. Table 3 shows frequencies and percentages of all independent variables used in this analysis. Our questionnaire was answered by a similar percentage of boys and girls. But, concerning sex, we offered a third option, "other sex", for which 3% of people opted. It is certainly still very much in the minority, so we had to recode the variable to introduce it into the multivariate linear regression model, treating "other sex" as a missing value, since very few people chose this option.

Characteristics of the sample
Adolescence is a risk factor in the process of radicalization, so we collected the age of the adolescents. The interviewees were between 12 and 19 years old, so it can be observed whether extremism and pro-violence attitudes decrease or increase with increasing age.
As already pointed out, Almeria and Madrid have received important migratory waves. Therefore, it is worth analysing whether the variable Parents´ Nationality is a factor which impacts on extremism and/or pro-violent attitudes. Among respondents, 63,2% declared that their parents (father and mother) were Spanish compared to 36,8% who declared that their father and/or mother were foreigners. Therefore, our sample contains a high percentage of adolescents whose father and/or mother are foreigners.
To the question of Family Savings Level, 19,3% of respondents answered that their families had spent savings or borrowed money last year compared to 80,7% of respondents who said that it was enough, or it was possible to save. As noted above, in Almeria the migrant population works in low-skilled jobs related to intensive agriculture or the service sector, which leads many adolescents to perceive that their families do not have many economic difficulties. Despite this, almost 20% of the families of these adolescents have to borrow money. In Madrid, the income levels of the school families we have included in the sample are lower than in Almería. The context of economic downturns in Spain and the economic impact of COVID-19 pandemic have also played an important role in family savings level, particularly on the most vulnerable ones.
Concerning the Highest Level of Education Achieved by the Mother of the respondent, almost 36% of respondents answered that their mothers have not attended school, or only primary school or less, while 38,2% (447) answered that she had attended secondary school, and 26,5% (310) that she possessed a university degree. The lowest educational levels correspond to mothers of foreign origin.
In the case of the variable Frequency of Attendance to Religious Services, 25% of respondents attended religious services only on specific holidays, and 43,2% never or practically never. The adolescents who most frequently attend religious services are Catholics or Muslims, but we also find Orthodox and Evangelicals.
Regarding self-positioning on the political ideology scale, it can be observed how most of the adolescents interviewed place themselves in the centre, since we did not offer them the possibility  Rodriguez Martinez et al., Cogent Social Sciences (2023) of not answering the question. The rest of the adolescents are almost equally divided between left and right, with those on the left predominating.
Regarding the variable Number of Social Networks Used Daily, the majority of teenagers (62% of respondents) use 3 or 4 social networks. Interestingly, on the extremes, 23% of respondents replied that they do not use social networks, or they use just 1 or 2, and 14% of interviewees use 5, 6 or 7 social networks on a daily basis. The most used social networks are WhatsApp, Instagram, Tik Tok, YouTube, Snapchat, Twitter and Facebook.
Exposure to violence is an important risk factor on radicalization, extremism, and pro-violent attitudes. To check this, we have constructed an Index of Diversity of Experienced Physical Violence from Sources such as schoolmates, friends, teachers, father, mother, partner, Unknown Person (what we have called Poly-victimization) after performing a Cronbach's analysis, the result of which allowed us to construct an index. According to this Index, 64% of respondents did not receive physical violence, 21% received violence from 1 source, while the rest of the respondents (15%) received violence from 2, 3, 4, 5, and 6 sources. The most frequent source of physical violence is schoolmates, and the least frequent source is the intimate partner.
We have also used three variables which try to measure the perception of respondents regarding the discrimination against three minority groups, LGTBI+, Children of Poor Families, and Children of Immigrants. With regard to the variable Groups that Suffer Discrimination in School (LGTBI+), 51,1% of respondents answered Yes. With regard to the variable Groups that Suffer Discrimination in the School (Children of Poor Families), 12,6% of respondents answered Yes. Finally, regarding the variable Groups that Suffer Discrimination in School (Children of Immigrants), 25,6% of respondents answered Yes. Table 4 presents the means of the extremism and pro-violence scales for each of the independent variables that we will later introduce in the multivariate linear regression model, with the scores for each category of each of the variables.

Testing 4 hypotheses
In the following lines, we test the four hypotheses proposed above in the hope to contribute to the understanding of the risk factors and general patterns of the extremist and pro-violent attitudes of adolescents in secondary schools of Almería and Madrid, Spain. To do so, we entered all the independent variables described above into two multivariate linear regression models, one for the extremism scale and the other for the pro-violence and illegal acts scale. We intend to observe which variables are significant and which lose significance when controlling for the rest. In this way we could test our hypotheses. Before introducing them, we carried out an analysis to validate the scales and to know to what extent they are correlated.
Regarding the first hypothesis (H1), our analysis validates both scales. In the case of ES, our data show a Cronbach's alpha of 0.84. In the case of PIARES, it is 0.92. Confirmatory Factor Analysis (CFA) was performed for both scales, with results of good fit in the case of ES and need for changes in the case of PIARES, so in this case Exploratory Factor Analysis (EFA) was carried out, eliminating four items (PIARES: 8 items) and performing CFA again, verifying the good fit of both scales to the sample used.
Subsequently, multi-group (gender) CFA was performed to test the measurement invariance of both scales. Comparisons of the proposed models were made, verifying configurational invariance for men and women, metric (restriction of factor loadings), scalar (restriction of intercepts) and strict (restriction of residuals), verifying that there are no significant variations in the CFI (ΔCFI <0.01) and RMSEA (ΔRMSEA <0.015) indices, in order to be able to assess whether the scales could adequately explain the differences between men and women. Baseline indices were calculated for both scales (extremism and pro-violence) to facilitate intra-and intersubject comparison.  Correlation analysis between ES and PIARES is 0.628 and statistically significant at the level of 0.01 (bilateral). This means that both indexes have much in common but are not exactly the same. Therefore, we can conclude that it is possible that the variables which explain both indexes (ES and PIARES) could not be the same. Table 4 shows the means of ES and PIARES for all the variables introduced in the analysis. Sex, Age, Spanish or Foreign Parents, Highest Level of Education achieved by the Mother, Frequency of Attendance to Religious Services, Political Ideology Selfpositioning, Number of Social Networks Used Daily, Index of Diversity of Physical Violence Received (Poly-victimization Index), and Perceived Discrimination in the School against the following social groups: LGTBI+, Children from Low Income families, and Children of Immigrants.
We have developed two multivariate linear regression models with all the variables, one Model for ES and the other Model for PIARES. The F-ratio in the ANOVA table for extremism and proviolence tests reveal that the two regression models show a good fit for the data. The independent variables predict with statistical significance the dependent variable, F(12, 1157) = 14,833, p  < .0005 (i.e., the regression model is a good fit of the data). In the case of the pro-violence index, the independent variables statistically significantly predict the dependent variable, F (12, 1157) = 13,418, p < .0005 (i.e., the regression model is a good fit of the data). Table 5 provides the results for the ES regression analysis and Table 6 for the PIARES regression.
It was hypothesized (H2) that boys would score higher than girls in ES and PIARES. Also, that adolescents with a Foreign parent, those who were younger, those from low saving families, those of mothers with lower level of education, those who more frequently attend religious services, those who are more on the right of the political ideology scale, and those who use a higher number of social networks on a daily basis would significantly score higher in ES and PIARES than adolescents with a Spanish parent, those who were older, those from high level of saving families, those of mothers with higher level of education, those who less frequently attend religious services, those who are more on the left of the political ideology scale, and those who use a lower number of social networks on a daily basis. This hypothesis was largely confirmed, but not in every case. In the case of ES, the level of saving of the family was not statistically significant, nor the number of social networks used on a daily basis.
According to H3, we should expect that the fact of having suffered violence (poly-victimization) could imply higher scores of the adolescents in ES and PIARES. Our analysis concludes that polyvictimization is statistically significant to explain PIARES. However, poly-victimization is not statistically significant to explain ES.
H4 proposed that the fact of perceiving discriminations (against LGTBI+ adolescents, against children of low-income families, and against children of immigrants) in secondary school would mean that the adolescent would develop more extremist and pro-violence attitudes. In this case, our analysis suggested that the perception of discrimination against children of immigrants was not statistically significant for ES and PIARES. Furthermore, in the case of PIARES, the perception of discrimination against children of low-income families, and against LGTBI+ adolescents was not statistically significant. On the contrary, the perception of discrimination against the LGTBI+ group and against children of low-income families, increased the probability of expressing extremist attitudes.

Conclusion and discussion
Our analysis has validated Ozer and Bertelsen (2018) two scales (ES and PIARES) for adolescents from 12 to 19 years old in five secondary schools in Spain, 4 in the city of Almería and 1 in Madrid (H1). This demonstrates the consistency and reliability of the scales, so we can say that they are a good instrument to measure extremism and pro-violence attitudes and illegal acts. Both scales have proved to be correlated in our analysis but not completely. That meant that it is likely that some variables were not significant in the two models. Also, the high correlation between both scales suggests the need to transform violent attitudes, especially among those who express extremist attitudes.
This research therefore supports the need to measure extremist attitudes and pro-violence and illegal acts separately. Although these two groups of attitudes are related, they should not be confused. In effect, the main problem of violent extremism is not only the fact that people have extremist attitudes (which should not be a problem in democratic societies), but that in order to defend extremist ideas, the use of violence or the carrying out of illegal acts is defended and/or justified, as some authors point out (Neumann, 2013;Sageman, 2017). Often one thing leads to the other.
In the case of ES, the risk factors for violent extremism focus on adolescents with a foreign parent, those who were younger, those of mothers with lower level of education, those who more frequently attend religious services, and those who are more on the right of the political ideology scale. In the case of PIARES, the risk factors focus on adolescents with a foreign parent, those who were younger, those from low saving families, those of mothers with lower level of education, those who more frequently attend religious services, those who are more on the right of the political ideology scale, and those who use a higher number of social networks on a daily basis.
Political ideology is significant in our regression model, with more violent attitudes being observed in right-wing positions. This could be explained by contextual causes characteristic of the territory, which activate more aggressively the positions of individuals who have scored to the right. The importance of the context is intuited in comparison with other conflicts where the activated context coincided with left-wing positions, as in the case of the violent left-wing movements in the Basque Country and Navarre that resulted in a behavioural radicalization, not only with ETA terrorism, but with various groups and actions inspired by a model of violence rooted in the left-wing political context and its highly coinciding discourse of the cognitive and behavioural, as shown by authors such as Muñoz Alonso (1982Alonso ( , 1985Alonso ( , 1986Alonso ( , 1988 or (Llera et al., 2017(Llera et al., , 2022.
Some variables are not significant in the two models. Family economic status is significant in the case of pro-violence attitudes and illegal acts, which corroborates studies indicating that proviolence attitudes are found more in vulnerable populations. However, the economic level of the family is not relevant in the regression model on extremism, so we can say that extremist attitudes are manifested in all socio-economic groups of adolescents.
The increased use of social media has no implications for extremist attitudes. However, it is significant for the development of pro-violence attitudes and illegal acts. These results possibly indicate that social network abuse leads adolescents to develop intolerant attitudes, which could have consequences for the use of offline violence.
Concerning poly-victimization, it is statistically significant to explain PIARES but not ES. Thus, the experience of having suffered physical violence helps to generate pro-violence attitudes and illegal acts but does not significantly influence extremist attitudes. In this sense, it can be said that violence generates violent attitudes.
Finally, our analysis shows that the perception of discrimination against children of immigrants is not statistically significant for ES and PIARES. In the case of PIARES, the perception of discrimination against children of low-income families, and against LGTBI+ adolescents is not statistically significant, but increases significantly the probability of expressing extremist attitudes (ES). The perception of discrimination is not related to the development of violent attitudes, although it leads adolescents to develop extremist attitudes, particularly in the case of discrimination against LGTBI+ people and children from poor families.
Transforming attitudes related to violent extremism among adolescents and young people is an investment in the future. This study provides results that could be used to develop programs and activities in secondary schools. As Austin and Giessmann (2018) point out, there are several approaches to addressing violent extremism: countering violent extremism, preventing violent extremism, and transforming violent extremism. Countering Violent Extremism (CVE) was initially an improvement on increasingly criticized counterterrorism approaches that rely on intelligence gathering and a repressive repertoire (Bhui et al., 2012; United Nations Development Programme, United Nations Office on Drugs and Crime, World Health Organization, 2014). CVE has evolved into a security-focused approach to tackling extremist violence, which uses a plethora of tools and entry points, but remains rooted in a "hard" approach to power. Second, Preventing Violent Extremism (PVE) was first introduced in the 2015 UN Plan of Action presented by then UN Secretary-General Ban Ki-moon. The emphasis is on identifying and addressing push and pull factors, addressing structural root causes and strengthening individual and community resilience. Thirdly, it highlights the recent introduction of Transforming Violent Extremism (TVE), which stresses the need and possibility of changing actors, and their means of violence, rather than only strengthening security or resilience to protect and prevent Thus, it is considered that the reasons and motivations that lead an adolescent to be attracted to violent extremist movements can be transformed into a different kind of agency or engagement. It is about altering the dynamics that lead to extremist violence. For the field of Online Disruptive Technologies there are also some interesting proposals (Amit et al., 2022).
By and large we can say that the findings of our research do not contradict the results of the international studies which we have reviewed but add specific knowledge with regard to the correlation between factors that have not been taken into account or have been examined with different accentuations in those investigations. In the same vein as Ozer and Bertelsen (2018), the results of our study suggest that it is interesting to differentiate between extremist and proviolence attitudes. In addition, we point out the groups which need our special attention: boys, young people, children of foreign parents, mothers with very low levels of education, those who attend religious services frequently, those on the extreme right, those who use many social networks, and those who have suffered physical violence.

Limitations
Extremist and pro-violence attitudes are beginning to be studied both in Spain and internationally. There are some limitations though. First, concerning the design of the research project, our questionnaire was designed for a particular age group, and the data collection was geographically limited. Therefore, our findings cannot be generalized. We need more studies that contain representative samples of the entire adolescent population, and even of the adult population.
Another limitation has to do with the design of the questionnaire. Since we wanted to develop a questionnaire that adolescents could answer in 30 minutes at most, and the batteries of questions to measure extremist and pro-violence attitudes were time-consuming, we could not introduce more questions that would have given us more information about the experience of physical violence or the use of social networks. Furthermore, we did not look into the correlation between the association with delinquent friends and extremism, even though it is a factor which is frequently mentioned in studies carried out from the point of view of security. We also had difficulties in introducing the question of the economic status of the adolescents' families.
Teenagers often ignore their parents' income range, so that it only was possible to establish the approximate economic status of the family.
In sum, we consider that the results of our study should be considered as exploratory but contribute to advancing knowledge about extremist and pro-violence attitudes and illegal acts. It is important to develop future empirical and theoretical research on these phenomena to better understand, prevent, and transform them.

Research ethics statement
The authors declare that they have respected the laws, rules and regulations of their community (the present work has the certificate of the ethics committee of the University of Almeria, which is consistent with the ethical standards articulated in the 1964 Declaration of Helsinki and its subsequent amendments and Section 12 ("Informed Consent") of the ASA's Code of Ethics). Also, of school (through the permission to participate in the research of adolescents by parents and school), work and country. During the research process, the researchers have behaved with integrity, fidelity, and honesty. Therefore, we openly assume responsibility for our actions and will only enter into agreements that we intend to keep.