INTRODUCTION
Juvenile delinquency is a core issue in criminology and criminal justice, and various elements of this phenomenon have been subject to extensive research (Khoury-Kassabri, Mishna, and Massarwi Reference Khoury-Kassabri, Mishna and Massarwi2019; Loeber and Farrington Reference Loeber and Farrington2012; Thompson and Bynum Reference Thompson and Bynum2016). Specifically, numerous studies have examined the risk and protective factors of involvement in juvenile delinquency (Orlando and Farrington Reference Orlando and Farrington2021; Zhao, Ren, and Chen Reference Zhao, Ren and Chen2023). For instance, parental attachment (Lee, Moon, and Garcia Reference Lee, Moon and Garcia2020) and religious orientation (Mohammad and Banse Reference Mohammad and Banse2023) have been identified as associated with lower levels of juvenile delinquency, while high impulsivity (Geerlings et al. Reference Geerlings, Asscher, Stams and Assink2020) and low socio-economic status (Shong, Abu Bakar, and Islam Reference Shong, Bakar and Islam2019) predict more involvement in this type of behaviour. However, alongside this extensive research, one major gap remains – the antecedents of juvenile delinquency in the unique context of political conflict, i.e. a state of dispute over the established political order of society (Trinn and Wencker Reference Trinn and Wencker2018), are still unknown.
Several theories have dealt with the notion of resistance to social or political order as a factor leading individuals, and especially members of racial and ethnic minority groups, to take part in risky and delinquent behaviours (e.g. see Factor et al. Reference Factor, Mahalel, Rafaeli and Williams2013; Rios Reference Rios2012). Such resistance is predominantly present in social conflicts, where members of non-dominant minority groups feel they are being treated illegitimately and unfairly (Atkin-Plunk, Peck, and Armstrong Reference Atkin-Plunk, Peck and Armstrong2019; Johansson and Vinthagen Reference Johansson and Vinthagen2016). In the current study, we utilize this notion to argue that minority youth exposed to political conflict may turn to delinquency as a means of coping with this discrimination. Specifically, we propose that exposure to political conflict may lead to perceived injustice, finding expression in attitudes that oppose the social order (Canetti et al. Reference Canetti, Galea, Hall, Johnson, Palmieri and Hobfoll2010). Consequently, this could lead to juvenile delinquency as an active form of resistance against the prevailing social order.
In the present study, we seek to investigate juvenile delinquency’s risk and protective factors in the context of majority–minority political conflict, i.e. a conflict arising from a dispute between the dominant majority and marginalized minority groups. We hypothesize that, in this context, due to perceived unfair treatment and discrimination by authorities, involvement in delinquency by youths will bear a strong political core and be predicted by attitudes reflecting resistance to social order. To examine this hypothesis, we utilized data from a survey conducted in East Jerusalem, an area characterized by majority–minority political conflict, among a representative sample of 814 youths who study in schools from various local neighbourhoods.
Juvenile Delinquency: Risk and Protective Factors
Juvenile delinquency has received widespread attention in the literature over the years (Loeber and Farrington Reference Loeber and Farrington2012; Moffitt Reference Moffitt1993; Mohammad and Banse Reference Mohammad and Banse2023; Orlando and Farrington Reference Orlando and Farrington2021; Thompson and Bynum Reference Thompson and Bynum2016). This phenomenon bears severe psychological, physical and social harms, which could manifest themselves in the long term, such as developing a criminal career (Khoury-Kassabri et al. Reference Khoury-Kassabri, Sharvet, Braver and Livneh2010; Loeber and Farrington Reference Loeber and Farrington2012). Thus, it is not surprising that much effort has been spent on identifying the scope of the problem, its causes, and effective interventions (see Cohen Reference Cohen1955; Fernández-Molina and Bartolomé Gutiérrez Reference Fernández-Molina and Gutiérrez2020; Khoury-Kassabri et al. Reference Khoury-Kassabri, Mishna and Massarwi2019; Wilson, Brennan, and Olaghere Reference Wilson, Brennan and Olaghere2018).
The grave implications of juvenile delinquency have led to a growing body of research into theories that might explain it, and from which several risk and protective factors of juvenile delinquency can be drawn. One central and dominant theory is Hirschi’s (Reference Hirschi1969) social bond theory, which emphasizes the role of positive bonds in preventing delinquency. According to Hirschi, youth with positive bonds with central agents of their socialization process – such as parents or peers – and are involved in normative activities are at lower risk of engaging in crime. These positive bonds include elements of attachment to socialization agents and commitment to them, substantial involvement in positive activity, and a belief system that opposes criminal behaviour.
In this regard, several factors that relate to positive bonds with the environment are associated with lower levels of juvenile delinquency. For instance, parental involvement and attachment have been found to be associated with normative behaviour (Khoury-Kassabri et al. Reference Khoury-Kassabri, Mishna and Massarwi2019; Lee et al. Reference Lee, Moon and Garcia2020). In the same vein, residing in a disrupted family environment, such as one with divorced parents, which diminishes the effectiveness of parental informal social control over youth, could serve as a risk factor for juvenile delinquency (Zhao et al. Reference Zhao, Ren and Chen2023). Likewise, positive attitudes towards the school and neighbourhood have found to be negatively related to delinquent behaviour (Liu and Miller Reference Liu and Miller2020; Zhang et al. Reference Zhang, Zhao, Zhao and Ren2014). Finally, involvement in normative activities such as religious practices and work has predicted lower levels of juvenile delinquency (Chan Reference Chan2019; Mohammad and Banse Reference Mohammad and Banse2023).
Another theory that emphasizes the role of the environment is Moffitt’s (Reference Moffitt1993) developmental taxonomy, which was supported by numerous empirical studies (e.g. Reckdenwald, Ford, and Murray Reference Reckdenwald, Ford and Murray2016; Widdowson et al. Reference Widdowson, Ranson, Siennick, Rulison and Osgood2020). Moffitt argues that there are two types of juvenile offenders: those whose delinquency is limited to adolescence and those who show chronic criminal offending even after this period (life-course persistence). According to Moffitt, one distinctive element among those who show stable criminality over time is an environment typified by criminogenic characteristics. To illustrate, one factor shown to be associated with juvenile offending is delinquent family members, especially criminal parents (Antle, Gibson, and Krohn Reference Antle, Gibson and Krohn2020; Swisher and Shaw-Smith Reference Swisher and Shaw-Smith2015).
Moffitt (Reference Moffitt1993) also argues that another element that distinguishes chronic criminal offenders from those who only offend in adolescence is the presence of neurological/genetic traits, such as impulsivity (Geerlings et al. Reference Geerlings, Asscher, Stams and Assink2020; Khoury-Kassabri et al. Reference Khoury-Kassabri, Mishna and Massarwi2019). In this regard, it is noteworthy that the role of impulsivity in predicting juvenile delinquency is a central component in another established theory – Gottfredson and Hirschi’s (Reference Gottfredson and Hirschi1990) General Theory of Crime. According to the latter, delinquent tendencies are more prevalent among juveniles characterized by a deficiency in self-control, manifested as impulsivity and a propensity for seeking immediate gratification.
Finally, a third theory that sheds light on the causes of juvenile delinquency is the General Strain Theory (Agnew Reference Agnew1992), whereby individuals who experience strain from different sources – a low socio-economic background, lack of academic success, or problematic relationships – are more prone to criminal behaviour (Li, Zhang, and Cheng Reference Li, Zhang and Cheng2022; Rebellon et al. Reference Rebellon, Piquero, Piquero and Thaxton2009). The reason is that strain leads to negative emotions such as fear, anger or frustration, which, in turn, leads to criminal activity as a possible response (Brezina Reference Brezina2017). Although not focused directly on youth, this framework has been useful in explaining juvenile delinquency (e.g. Barn and Tan Reference Barn and Tan2012; Snyder et al. Reference Snyder, Hartinger-Saunders, Brezina, Beck, Wright, Forge and Bride2016). Specifically, strain producers such as low socio-economic status have been found to predict youth offending (Shong et al. Reference Shong, Bakar and Islam2019).
In the context of the present study, it is noteworthy that, in recent years, the General Strain Theory has been extended to explain terrorism and radicalization (see Agnew Reference Agnew2010). According to this perspective, individuals facing significant strains, such as political oppression or economic marginalization, may be more prone to engage in acts of terrorism due to the negative emotions these strains generate, in conjunction with a lack of conventional coping mechanisms (Agnew Reference Agnew, LaFree and Freilich2016). This extension is also applicable at the group level, where collective experiences of strain can contribute to the emergence of violent political activities among marginalized or oppressed communities (Wolfowicz et al. Reference Wolfowicz, Litmanovitz, Weisburd and Hasisi2021).
Through these theories and others, we can explain various aspects of juvenile delinquency and derive various risk and protective factors. The role played by the latter has been tested in various places and contexts (e.g. see Green et al. Reference Green, Musci, Johnson, Matson, Reboussin and Ialongo2016; Orlando and Farrington Reference Orlando and Farrington2021; Zhang et al. Reference Zhang, Zhao, Zhao and Ren2014); however, their role in one unique context, namely that of political conflict, has remained neglected.
The Impact of Political Conflict on Children and Youth
A political conflict can be defined as “a dispute between two or more political actors (e.g., governments, challengers, third parties) over the pursuit, maintenance or distribution of power” (Zhukov, Davenport, and Kostyuk Reference Zhukov, Davenport and Kostyuk2019, 604). As Trinn and Wencker (Reference Trinn and Wencker2018) note, the term “political” encompasses more than just the narrow definition of the political system, as it broadly refers to state institutions and those arising from societal self-regulation. This phenomenon takes place in different places around the world (Caplan Reference Caplan2019; Keller Reference Keller2014; Tonge Reference Tonge2013), and exposure to such conflicts, i.e. living in an atmosphere characterized by political disputes, is correlated with various undesirable outcomes, such as poor health and lowered social trust (Canetti et al. Reference Canetti, Galea, Hall, Johnson, Palmieri and Hobfoll2010; De Juan and Pierskalla Reference De Juan and Pierskalla2016; Miller and Rasmussen Reference Miller and Rasmussen2010). Regarding children and youth in particular, studies have shown that living in an atmosphere of political conflict is associated with unfavourable mental health symptoms, such as anxiety, depression and post-traumatic stress disorder (Dvir Gvirsman et al. Reference Dvir Gvirsman, Huesmann, Dubow, Landau, Shikaki and Boxer2014; Merrilees et al. Reference Merrilees, Taylor, Goeke-Morey, Shirlow and Cummings2022; Siegel et al. Reference Siegel, Brickman, Goldberg, Pat-Horenczyk, Hoven, Amsel and Tyano2019). In addition, several studies have revealed a positive relationship between youth exposure to political conflict and delinquency outcomes. For example, Nuttman-Shwartz (Reference Nuttman-Shwartz2017) found that children and adolescents who face continuous security threats are more prone to show aggressive behaviour. Such findings recurred when different aspects of delinquent behaviour were examined among youth exposed to political conflict (Cairns Reference Cairns1996; Dubow et al. Reference Dubow, Huesmann, Boxer, Smith, Landau, Gvirsman and Shikaki2019; Huesmann et al. Reference Huesmann, Dubow, Boxer, Landau, Gvirsman and Shikaki2017; Muldoon and Trew Reference Muldoon and Trew2000).
Although this research area has focused primarily on the consequences of exposure to such conflicts, it is worth mentioning that some (see Baier Reference Baier2018; Frounfelker et al. Reference Frounfelker, Frissen, Vanorio, Rousseau and d’Haenens2019; Khoury-Kassabri, Khoury and Ali Reference Khoury-Kassabri, Khoury and Ali2015) have examined its effect on active participation in political violence, defined as any deliberate action directed towards achieving a political objective by employing violence or the threat thereof against others (Bueno de Mesquita et al. Reference Bueno de Mesquita, Fair, Jordan, Rais and Shapiro2015). For example, De Waele and Pauwels (Reference De Waele and Pauwels2014) conducted a study among Flemish youth and found that various risk and protective factors, such as impulsivity, peer delinquency and religious authoritarianism, explain involvement in politically motivated violence and property crimes. Previous studies have also examined supportive attitudes towards the use of radicalization and political violence, meaning perspectives that endorse the use of violence against state agents, whether they be security forces or civilians. For instance, Frounfelker et al. (Reference Frounfelker, Frissen, Vanorio, Rousseau and d’Haenens2019) found that perceived discrimination was associated with support of political violence. In the same vein, in a recent and thorough meta-analysis, Wolfowicz et al. (Reference Wolfowicz, Litmanovitz, Weisburd and Hasisi2021) have identified numerous risk and protective factors for radicalization. Interestingly, they have found juvenile delinquency to be a significant and consistent predictor of supporting attitudes towards political violence.
In summary, there is a solid empirical basis for the relationship between exposure to political conflict and juvenile delinquency (as well as for several risks and protective factors that were identified as spurring youth involvement in political violence). However, alongside the findings mentioned above, one substantial gap remains – it is still unclear which mechanisms play a role in leading youth to delinquency in the unique context of political conflict. In other words, although children and youth exposed to political conflict are at risk of exhibiting delinquent behaviour, less effort has been paid to identifying possible risk and protective factors of delinquency in this context.
As we will argue, there is reason to believe that juvenile delinquency in political conflict will have some unique antecedents. Youth may witness and experience ongoing political tension in states of political conflict. Among those belonging to non-dominant minority groups, such exposure could influence perceptions regarding how fairly they are being treated by the dominant group, as well as their experiences of discrimination. This may lead them to embrace behaviours expressing dissatisfaction with the social order. More specifically, as detailed below, we will propose that, in this special context, juvenile delinquency is derived, among other things, from social resistance.
Social Resistance and Delinquency
Findings of studies from all over the world show that minorities of racial and ethnic groups are overrepresented in the criminal justice system (Anderson, Wooldredge, and Cochran Reference Anderson, Wooldredge and Cochran2022; Boon, van Dorp, and de Boer Reference Boon, van Dorp and de Boer2019; Kutateladze et al. Reference Kutateladze, Andiloro, Johnson and Spohn2014; Mears et al. Reference Mears, Craig, Stewart and Warren2017). The leading explanation for this overrepresentation is that racial and ethnic minorities suffer discrimination at different stages of the criminal process (MacFarlane and Stratton Reference MacFarlane and Stratton2016; Spohn Reference Spohn, Luna and Phoenix2017), which manifests, for example, in higher prosecution rates and harsher punishments (MacDonald and Donnelly Reference MacDonald and Donnelly2019; Wu Reference Wu2016). However, contrary to this view, which sees minorities as passive agents, some theoretical perspectives offer explanations that include an active component. They do so by focusing on the precursors for risky and delinquent behaviours among minorities and, more specifically, by looking at delinquency as an act of resistance to the social order. The central concept here is that minorities resort to risky and delinquent behaviours as a way of expressing their discontent with the established social order. This extends beyond a narrow interpretation confined to the “political system”, encompassing a broader spectrum.
The origins of the idea that delinquency among minorities is driven to some extent by resistance lies in the principles of procedural justice theory (Tyler Reference Tyler2006). This theory holds that behaviour is affected, among other things, by whether people perceive the treatment they receive as fair or unfair (Donner and Olson Reference Donner, Olson, Ivković, Maskály, Donner, Mraović and Das2022). Simply put, a perception of being treated fairly has a positive impact on behaviour, whereas unfair treatment may lead to unfavourable outcomes, including non-compliance with the law (Rattner and Yagil Reference Rattner and Yagil2004). In the case of racial and ethnic minorities, it is often argued that perceived unfair treatment by formal authorities in the criminal justice system reduces the legitimacy given to society’s focal values and institutions, especially the law and legal system, thereby increasing delinquency (see Atkin-Plunk et al. Reference Atkin-Plunk, Peck and Armstrong2019; Nuño Reference Nuño2018; for a critique, see Nagin and Telep Reference Nagin and Telep2020).
However, perceptions of fairness are also affected by the treatment of informal agents, as drawn from the General Strain Theory (Agnew Reference Agnew1992) mentioned above. The latter holds that when individuals believe that they are subject to unjust treatment by those with whom they interact on a personal level (e.g. parents or peers), this perceived unfairness serves as a source of strain that can ultimately contribute to the development of delinquency. Moreover, it can even lead to political radicalization when the source of strain is “more powerful” outgroup members (Agnew Reference Agnew, LaFree and Freilich2016). This idea was empirically validated by Rebellon et al. (Reference Rebellon, Manasse, Van Gundy and Cohn2012), who found that perceptions of unfairness constitute a significant form of strain, eventually promoting delinquency.
The above-mentioned active component is manifested in the notion that, aside from passive non-compliance with the law, non-dominant minorities actively engage in acts that show their resistance to power relations in society (Johansson and Vinthagen Reference Johansson and Vinthagen2016). These ideas are at the core of some theoretical perspectives, which focus on the role of resistance to the social order; the social resistance framework (Factor, Kawachi, and Williams Reference Factor, Kawachi and Williams2011; Factor et al. Reference Factor, Mahalel, Rafaeli and Williams2013) is one example. This framework holds that everyday discrimination experienced by non-dominant minority groups leads them to resist society’s central values and institutions. In order to express their dissatisfaction with the social order, individuals from these groups actively engage in “everyday resistance” (Scott Reference Scott1986) – practices that contradict the dominant group’s values. These acts include, among other things, risky and delinquent behaviours (Factor et al. Reference Factor, Mahalel, Rafaeli and Williams2013). This framework is supported by empirical evidence regarding various racial and ethnic groups (e.g. Langley et al. Reference Langley, Ariel, Tankebe, Sutherland, Beale, Factor and Weinborn2021; Letki and Kukołowicz Reference Letki and Kukołowicz2020; Waterworth et al. Reference Waterworth, Dimmock, Pescud, Braham and Rosenberg2016).
Another example of a theory that emphasizes resistance to social order as a root cause of delinquent behaviour is Rios’s (Reference Rios2012) “crimes of resistance”. Rios argues that marginalized youth, specifically youth of colour, are often labelled as delinquent and deviant. Thus, they often commit crimes to resist this label imposed on them by society. Rios claims that it may lead even law-abiding and non-delinquent youth, who are usually committed to positive goals, to participate in minor acts of delinquency by way of exhibiting their frustration with their position within the social order.
It is important to emphasize that, while sharing theoretical resemblances, resistance theories fundamentally diverge from other recognized theories that attribute delinquency to the experience of unfairness and discrimination, which, in turn, may lead to strain (Agnew Reference Agnew1992; Tyler Reference Tyler2006). Indeed, resistance theories do view perceived unfairness as a source of strain. However, they assert that delinquent behaviour functions not as a mere coping mechanism for negative emotions that stem from this strain but as an active demonstration of individuals expressing their discontent with the existing social order (Factor et al. Reference Factor, Kawachi and Williams2011, Reference Factor, Mahalel, Rafaeli and Williams2013). In other words, this active expression aims to challenge society’s core values. In this regard, resistance should not be seen solely as a stand-in for perceived unfairness but rather as a proactive stance challenging the fundamental values of society. As we propose, given the prevalence of perceived unfairness among youth experiencing political conflict (Canetti et al. Reference Canetti, Galea, Hall, Johnson, Palmieri and Hobfoll2010), a central precursor of juvenile delinquency within this context is the manifestation of social resistance.
The Current Study
The literature suggests various risk and protective factors of juvenile delinquency that stem from diverse theoretical perspectives (Antle et al. Reference Antle, Gibson and Krohn2020; Orlando and Farrington Reference Orlando and Farrington2021; Zhao et al. Reference Zhao, Ren and Chen2023). However, the underlying mechanism for this behaviour in the exceptional context of political conflict remains unknown. Our research question, therefore, is: What are the risk and protective factors of juvenile delinquency among youth exposed to political conflict?
Indeed, there are reasons to assume that the antecedents for delinquent behaviour of youths living in an atmosphere of political conflict would differ from those in “regular” contexts. First, as shown above, exposure to political conflict has a dramatic negative effect on the well-being of children and youth (Dubow et al. Reference Dubow, Huesmann, Boxer, Smith, Landau, Gvirsman and Shikaki2019; Nuttman-Shwartz Reference Nuttman-Shwartz2017; Siegel et al. Reference Siegel, Brickman, Goldberg, Pat-Horenczyk, Hoven, Amsel and Tyano2019), and these effects might interplay with other aspects in their lives, including those associated with delinquency. For example, the protective effect of factors such as parental attachment, commitment to school, or religiosity on delinquency could vary among adolescents struggling with anxiety or depression.
Second, and more importantly, in political conflict, children and youth are exposed to everyday tension between social groups. For members of non-dominant minority groups, this could affect how fair they perceive the dominant group’s treatment of them to be, as well as their levels of experienced discrimination (Sargeant, Davoren, and Murphy Reference Sargeant, Davoren and Murphy2021). As we have seen above, such perceptions might lead them to actively engage in delinquent behaviours by exhibiting their resistance to the social order. Consequently, we might also expect the risk and protective factors of delinquency among youth exposed to political conflict to be, at least in part, closely linked with resistance to the social order. Thus, drawing upon the principles of social resistance theories, we hypothesize that while juvenile delinquency in the context of political conflict will be explained by the well-established risk and protective factors from existing literature, it will also be characterized by a strong political core.
As we will show below, our research question was explored in the context of majority–minority political conflict – the Israeli–Palestinian conflict – among adolescents from East Jerusalem. The Israeli–Palestinian conflict’s origins can be traced back to 1948 when the Arab countries bordering the newly established Jewish state – Egypt, Syria, and Jordan – had instigated a war against Israel after refusing to accept the United Nations partition plan (Karsh Reference Karsh2014). Israel won this war, and its victory resulted in roughly 700,000 Palestinians becoming displaced and seeking refuge in neighbouring Arab countries (Plascov Reference Plascov2017). More than two decades later, in 1967, several Arab armies launched another war against Israel, known as the Six Day War. During this campaign, Israel had gained control over territories in the West Bank, including East Jerusalem (Odeh Reference Odeh1992). Since then, this area has served as a microcosm encapsulating the complexities and tensions of the Israeli–Palestinian conflict, characterized by frequent escalations and outbreaks of violent acts between the Arab and Jewish populations (Hasisi, Itskovich, and Khoury-Kassabri Reference Hasisi, Itskovich and Khoury-Kassabri2023).
The area of East Jerusalem shares similar features with other regions characterized by high levels of political conflict worldwide (see Corkalo Biruski Reference Corkalo Biruski, Alon and Bar-Tal2016; Knox Reference Knox2002). First, the inhabitants of East Jerusalem, who are not considered fully fledged Israeli citizens but rather permanent residents, do not identify with the majority group in Israeli society but with their own in-group, i.e. Palestinian society (Yair and Alayan Reference Yair and Alayan2009). Second, the primary enforcement authority in East Jerusalem is not the Israel Police but the paramilitary border police (“Magav”), leading to frequent political clashes between the Arab residents and Israeli security forces (Volinz Reference Volinz2018). Third, residents of East Jerusalem are highly involved in resistance activities, manifested, in some cases, in military and terrorist attacks (Dumper Reference Dumper2013). Finally, East Jerusalem residents suffer from low physical conditions, such as poor sanitation and infrastructure, and economic deprivation manifested in poor educational attainment and high underemployment rates (Shlomo Reference Shlomo2017; Shtern Reference Shtern2019; Yair and Alayan Reference Yair and Alayan2009). These factors make the residents highly exposed to and involved in political violence (Khoury-Kassabri et al. Reference Khoury-Kassabri, Khoury and Ali2015), rendering this area appropriate for examining our research question.
METHOD
Data and Sample
The data used in the present study were obtained from a survey conducted by the present authors in 2018 among Arab male students from East Jerusalem aged 12 to 18 years (grades 7 to 12). We chose to include only male participants since they are significantly more likely to participate in serious physical violence compared to females (Khoury-Kassabri Reference Khoury-Kassabri2019). The survey was conducted in middle- and high schools in East Jerusalem using two-stage cluster sampling: first, we randomly selected 11 out of 26 of these schools. Then, we randomly selected two classes from each grade and offered all the students from these classes the opportunity to participate. Following this method, a total sample of 814 male students (mean = 14.48, standard deviation = 1.48 years) with a response rate of 86% was obtained. We used the response rate number 2 (RR2) formula to calculate this rate, which returns the ratio of complete and partial interviews to the overall number of eligible respondents (American Association for Public Opinion Research 2016).
All students who agreed to participate in the study filled out an anonymous self-report questionnaire under the supervision of a trained research assistant. The students’ parents received a consent form describing the study goals and were allowed to refuse to have their child participate (about 2% of the parents indeed decided to do so – and their children were, therefore, not included in the response rate calculation). Next, the students selected to participate were also offered the opportunity to withdraw their participation (the refusal rate was approximately 5%). The institutional review board and Ministry of Education approved all the study materials, including the informed consent forms and questionnaires.
Variables
Our dependent variable was “juvenile delinquency”. This variable was measured using eight items (α = 0.90) from the Arabic version of the self-report delinquency scale (Elliott and Ageton Reference Elliott and Ageton1980) developed by Khoury-Kassabri et al. (Reference Khoury-Kassabri, Khoury and Ali2015). These items include statements such as “You carried a weapon such as a knife or a gun” and “You were involved in gang fights”. Participants indicated how often they were involved in such acts on a scale of 1 (“never”) to 5 (“more than 10 times”; see Table 1 for the factor loading and item wording), indicating that they belong to the same construct. The variable was constructed using the mean of the eight items.
Table 1. Exploratory Factor Analysis of the Dependent and Independent Variables (Principal Factor with Varimax Rotation) a
a Only loadings greater than 0.4 are presented. Kaiser–Meyer–Olkin measure of sampling adequacy = 0.85.
Our model includes two sets of independent variables. The first set comprises variables about attitudes and behaviours, all reflecting risk and protective factors of juvenile delinquency under the theoretical perspective of the theories mentioned above (Agnew Reference Agnew1992; Hirschi Reference Hirschi1969; Moffitt Reference Moffitt1993). Importantly, this set includes two attitude variables: “attitudes towards general violence”; and “attitudes towards political violence”. The latter assessed the role of political orientation in predicting juvenile delinquency, as it reflects resistance to social order. The former served to determine the role of general violence attitudes, as well as a benchmark for the impact of said political violence attitudes, since it indicates which type of attitudes has a stronger effect. The variable “attitudes towards general violence” was measured using three items (α = 0.53) from the translated version of the Attitudes Towards Violence Scale (Funk et al. Reference Funk, Elliott, Urman, Flores and Mock1999) introduced by Khoury-Kassabri et al. (Reference Khoury-Kassabri, Sharvet, Braver and Livneh2010), such as “If someone hits you, you should hit them back”. Note that this scale’s relatively low α value is probably due to the small number of items comprising this scale (see Swailes and McIntyre-Bhatty Reference Swailes and McIntyre-Bhatty2002), and, therefore, we also calculated the mean interitem correlation – 0.28 – which fell within the optimal range of 0.2–0.4 (Briggs and Cheek Reference Briggs and Cheek1986). The variable “attitudes towards political violence” was measured using five items (α = 0.81) composed for the present study based on questions from the Attitudes Towards Violence Scale. This variable included questions such as “It is okay to use violence against Jews even if there is no reason for it”, and participants indicated their agreement on a scale of 1 (“completely disagree”) to 4 (“completely agree”). The two attitude variables (and the other variables in this set) were constructed using the mean of their survey items.
This first set of variables also includes three variables representing bonds with positive socialization agents. The first variable is “parental attachment”, which was measured using five items (α = 0.77), such as “To what extent do you feel that your parents understand you?” with responses being on a scale from 1 (“not at all”) to 5 (“to a large extent”). This scale was constructed by Hirschi (Reference Hirschi1969) and adapted to Israeli youth by Shechory and Laufer (Reference Shechory and Laufer2008). The second variable is “neighbourhood attachment”, which was measured using five items (α = 0.71), such as “My neighbourhood is a good place in which to live and grow up”, with responses being on a scale from 1 (“completely disagree”) to 4 (“completely agree”).
The third bond variable pertains to religiosity and is divided into two variables (see Khoury-Kassabri et al. Reference Khoury-Kassabri, Khoury and Ali2015) – “relationship with God”, measured using five items (α = 0.91), such as “Reading the Koran helps me strengthen my relationship with God”, on a scale of 1 (“never”) to 5 (“to a large extent”), and “request and retribution”, measured using four items (α = 0.72), such as “If I behave badly, God will make my life difficult”, with participants indicating their agreement on a scale of 1 (“completely disagree”) to 4 (“completely agree”). These two variables were constructed in 2011 by Pickering, Buzzetta, and Aten in 2011 and used by Eseed and Khoury-Kassabri (Reference Eseed and Khoury-Kassabri2018) among Arab students in Israel. Finally, another variable that was included in this model is “impulsivity”. This variable was measured using three items (α = 0.62; e.g. “I do things without thinking”) from the Teen Conflict Survey (Bosworth and Espelage Reference Bosworth and Espelage1995), with responses ranging on a scale of 1 (“never”) to 5 (“always”).
We performed an exploratory factor analysis to ascertain whether this set’s dependent and independent variables represent distinct constructs. The results of this analysis have confirmed their distinctiveness, and each of the eight items exhibited a large eigenvalue (all other extracted factors had eigenvalues much lower than 1; see Table 1 for the factor loading and item wording). In addition, confirmatory factor analysis has indicated that all items are significant (p < 0.001) and a good fit to the data (comparative fit index = 0.96; root mean square error of approximation = 0.05; standardized root mean square residual = 0.06 [Kline Reference Kline2015]).
The second set of independent variables consists of sociodemographic measures, which were also found to be correlated with juvenile delinquency (Antle et al. Reference Antle, Gibson and Krohn2020; Chan Reference Chan2019; Zhao et al. Reference Zhao, Ren and Chen2023). These include “age”, “employment” (0 = no, 1 = yes), “parental marital status” (0 = not married, 1 = married), and “family member arrest by the police” (0 = no, 1 = yes). In addition, we included a measure of “family socio-economic status”. Following Davidov and Khoury-Kassabri (Reference Davidov and Khoury-Kassabri2013), this scale was constructed using the mean of three standardized variables: mother’s and father’s education level, ranging from 1 (elementary) to 5 (academic), and family income level, ranging from 0 (very low) to 5 (high). Since we could not obtain such information, this scale was not standardized by household size. Table 2 presents descriptive statistics on the research variables.
Table 2. Descriptive Statistics of the Study Variables
Data Analysis
We began with a bivariate analysis of the research variables. Then, to test our research hypothesis, we conducted a multilevel analysis (Kreft and de Leeuw Reference Kreft and Leeuw1998). This method is appropriate when observations are clustered at a higher level, which poses the risk of violating the independence of errors assumption (Snijders and Bosker Reference Snijders and Bosker1999). In the present study, students were clustered within 11 schools; thus, multilevel analysis was suitable. Note that, although the number of clusters (schools) is relatively small, it still meets the guidelines of at least 10 clusters (Clarke and Wheaton Reference Clarke and Wheaton2007), and, in any event, a small number of clusters should have a small impact on the fixed effects, which was the main point of interest in the present study (Łaszkiewicz Reference Łaszkiewicz2013).
We started our regression analysis with a null model, which includes no independent variables, only the dependent variable with the variances of the individual (youth) and cluster (school) levels (Snijders and Bosker Reference Snijders and Bosker1999). We did so to calculate the intraclass correlation (ICC) and determine the degree of variance in juvenile delinquency deriving from the clustering of youth in schools. If this variance is significant, the use of multilevel modelling is justified. Next, we ran four models: the first only included attitudes towards general violence as an independent variable to examine its effect on juvenile delinquency; in the second, we added the attitudes towards political violence variable to examine its effect and whether its inclusion changed the effect of attitudes towards general violence; the third model included all above-mentioned risk and protective factors of juvenile delinquency; and, finally, in the fourth model, to further examine the role of participants’ political stand – a main point of interest in our study – we added an interaction term between the variables “attitudes towards political violence” and “family member arrest by the police”. The rationale here is that if support for political violence, as reflecting resistance to social order, is indeed a predictor of juvenile delinquency, then the experience of the arrest of a family member by the authorities (the state of Israel, in the present case) should strengthen the youth’s political resistance, and, in turn, the effect of attitudes towards political violence on juvenile delinquency.
Our models are represented in the following equation:
$$\matrix{{JuvenileDelinquenc{y_{ij}} = {\gamma _{00}} + {B_{1j}}PoliticalViolenceAttiude{s_{1ij}}} \hfill \cr
{\quad \quad + \;{B_{2j}}GeneralViolenceAttitude{s_{1ij}}\; + {B_{3j}}ParentalAttachmen{t_{1ij}}} \hfill \cr
{\quad \quad + \;{B_{4j}}NeighbourhoodAttachmen{t_{1ij}} + {B_{5j}}RelationshipwithGo{d_{1ij}}} \hfill \cr
{\quad \quad + \;{B_{6j}}RetributionandReques{t_{1ij}} + {B_{7j}}Impulsivit{y_{1ij}} + {B_{8j}}Ag{e_{1ij}} + {B_{9j}}Employe{d_{1ij}}} \hfill \cr
{\quad \quad + \;{B_{1j}}ParentalMaritalStatu{s_{1ij}} + {B_{1j}}FamilyMemberArres{t_{1ij}}} \hfill \cr
{\quad \quad + \;{\gamma _{10}}\left( {PoliticalViolenceAttiude{s_{ij}}FailyMemberArres{t_{ij}}} \right) + {U_{0j}} + {r_i},} \hfill \cr } $$
where i represents an individual in school j,
${\gamma _{00}}$
is the grand-mean intercept, and
${U_{0j}}$
and
${r_{ij}}$
represent the variances of level-2 intercepts and students’ residuals around each school’s slopes, respectively. Note that the variance inflation factor calculated for the models did not suggest multicollinearity and that Little’s test had indicated that the data are missing completely at random (Li Reference Li2013).
RESULTS
Table 3 presents bivariate correlations of the study variables. As shown, except for socio-economic status (r = 0.01, NS), all variables significantly correlate with juvenile delinquency. Specifically, parental attachment (r = –0.21, p < 0.001), the two religiosity measures – relationship with God (r = –0.22, p < 0.001) and request and retribution (r = –0.14, p < 0.001) – and neighbourhood attachment (r = –0.21, p < 0.001) are all associated with lower levels of juvenile delinquency. In contrast, impulsivity (r = 0.27, p < 0.001) and supporting attitudes regarding general (r = 0.13, p < 0.001) and political (r = 0.35, p < 0.001) violence are associated with higher levels of juvenile delinquency. As for the sociodemographic variables, older age (r = 0.13, p < 0.001), employment (r = 0.19, p < 0.001) and family member arrest (r = 0.29, p < 0.001) are positively associated with more juvenile delinquency, whereas the parental marital status of marriage (r = –0.22, p < 0.001) is associated with less juvenile delinquency.
Table 3. Bivariate Analysis of the Study Variables
* p < 0.05; ** p < 0.01; *** p < 0.001.
The results of the null model (Table 4) show that the ICC equals 0.07, meaning that 7% of the variance in juvenile delinquency between individuals is derived from the variance between schools (
$\sigma _{u0}^2$
= 0.04, p < 0.05). In addition, a likelihood ratio test supports the use of multilevel analysis over a simple linear regression, as it reveals that the variance between the schools is significantly different from 0 (χ
2
(1, n = 774) = 29.53, p < 0.001).
Table 4. Multilevel Analysis of Juvenile Delinquency at Two Levels (Null Model) (n = 774)
* p < 0.05; *** p < 0.001.
The results of our multilevel analysis appear in Table 5. From model 1, we discover that supporting attitudes towards general violence are positively correlated with juvenile delinquency (b = 0.14, p < 0.001). However, when including attitudes towards political violence in model 2, the effect of general attitudes on juvenile delinquency disappears (b = 0.06, NS), while political violence has a significant and positive effect (b = 0.29, p < 0.001). This holds true in model 3 (which includes all independent variables), indicating that supporting attitudes towards political violence predict more involvement in juvenile delinquency (b = 0.19, p < 0.001), contrary to supporting attitudes towards general violence, which is insignificant (b = 0.01, NS). Out of the bond with socialization agent variables, parental attachment (b = –0.08, p < 0.05), higher levels of relationship with God (b = –0.07, p < 0.05) and neighbourhood attachment (b = –0.09, p < 0.05) are significantly associated with lower levels of juvenile delinquency. Impulsivity (b = 0.09, p < 0.01), employment (b = 0.21, p < 0.01), family member arrest (b = 0.25, p < 0.001) and higher socio-economic status (b = 0.12, p < 0.001) are associated with higher levels of juvenile delinquency, while parental marital status is associated with lower levels of juvenile delinquency (b = –0.55, p < 0.001).
Table 5. Multilevel Model (Random Intercept) of Juvenile Delinquency (n = 633)
SE, standard error; CI, 95% confidence interval.
* p < 0.05; ** p < 0.01; *** p < 0.001.
Next, to further understand the role of attitudes towards political violence, we produced a model for the interaction between this variable and the family member arrest variable (model 4). To test whether the inclusion of this interaction term significantly improves the model, we performed a χ 2 deviance test, which indicated a better fit of this model compared to model 3 (χ 2 (1, n = 633) = 14.63, p < 0.001). The results show that the interaction term is significant (b = 0.24, p < 0.001), meaning that the effect of attitudes towards political violence on juvenile delinquency is conditioned upon the arrest of a family member. Marginal effect displays of juvenile delinquency based on the interaction term (Figure 1) revealed that, for those whose family members had been arrested, attitudes towards political violence have a positive and significant effect on juvenile delinquency (z = 6.70, p < 0.001). In contrast, this effect is weaker among those whose family members have not been arrested (z = 2.23, p < 0.05). More specifically, among students whose family members had never been arrested, there is a difference of 0.27 between those with low (1) and high (4) levels of attitudes towards political violence. In contrast, the difference is more pronounced among students whose family members had been arrested, reaching 1.00.
Figure 1. Predicted juvenile delinquency for attitudes towards political violence and family member arrest by the police using multilevel regression.
DISCUSSION
Juvenile delinquency has severe consequences, from immediate physical and mental harm to the potential development of a lifelong criminal career and even prison sentencing (Khoury-Kassabri et al. Reference Khoury-Kassabri, Sharvet, Braver and Livneh2010; Loeber and Farrington Reference Loeber and Farrington2012). In an attempt to shed light on this phenomenon, scholars have offered and explored various risk and protective factors of the involvement of youth in criminal behaviour (Antle et al. Reference Antle, Gibson and Krohn2020; Geerlings et al. Reference Geerlings, Asscher, Stams and Assink2020; Liu and Miller Reference Liu and Miller2020; Orlando and Farrington Reference Orlando and Farrington2021). However, to date, no research has referred to these risk and protective factors in the special context of political conflicts. We sought to fill this gap by examining the risk and protective factors of juvenile delinquency in the context of a majority–minority conflict.
In the present study, we utilized the perspectives of resistance theories (Factor et al. Reference Factor, Mahalel, Rafaeli and Williams2013; Rios Reference Rios2012). These maintain, as do other well-established theories (e.g. Agnew Reference Agnew1992), that delinquency among disadvantaged groups stems from experienced unfairness and discrimination, ultimately serving as a source of strain. However, they also differ from them because they view delinquency as an active and dynamic display of individuals articulating their dissatisfaction with the prevailing social structure (Factor et al. Reference Factor, Kawachi and Williams2011).
Our hypothesis was examined among youth from East Jerusalem. Within this region, one finds a condensed representation of the intricacies and strains characteristic of the Israeli–Palestinian conflict, marked by recurrent escalations and instances of violence between Arab and Jewish communities. The population in this area, composed of residents rather than fully fledged Israeli citizens, is also extremely disadvantaged. East Jerusalem Arabs face challenging living conditions alongside economic struggles reflected in low educational achievement and high rates of underemployment. These elements notably incline them towards resistance against the Israeli state and its representatives, making this area a suitable focus for our research inquiry.
We hypothesized that juvenile delinquency in East Jerusalem would stem not only from the well-documented risk and protective factors of juvenile delinquency but from a mechanism with a strong political core, too. The results support our hypothesis. Our multilevel model showed that resistance to social order, as reflected in supportive attitudes towards political violence, is a strong and significant predictor of juvenile delinquency among youth from East Jerusalem, especially those whose family members have been arrested. Notably, the results show that the variable “attitudes towards general violence” is not a significant predictor of juvenile delinquency after controlling for “attitudes towards political violence”.
These results suggest that an “active ingredient” of juvenile delinquency in East Jerusalem is a political view that resists social order. Among the youth of East Jerusalem, who are constantly exposed to the ongoing conflict between Israel and the Palestinians, it is not general support for violence that leads to delinquency but a view that supports violence directed at agents of the Israeli state – whether security forces or civilians. Moreover, this effect intensifies in cases in which a family member had previously been detained by the (Israeli) police, serving as evidence that juvenile delinquency in the context of political conflict is driven by political orientation in which youth see themselves as opposing the outgroup. More generally, it supports the notion of resistance theories whereby members of minority groups actively engage in risky and delinquent behaviours to express their dissatisfaction with power relations in society (Factor et al. Reference Factor, Mahalel, Rafaeli and Williams2013; Johansson and Vinthagen Reference Johansson and Vinthagen2016; Rios Reference Rios2012). Note that these findings are in line with Massarwi and Khoury-Kassabri (Reference Massarwi and Khoury-Kassabri2017), who found that one predictor of serious physical violence among Arab youth is perceived ethnic discrimination.
The current findings also have implications for the literature pertaining to radicalization. As noted above, in their recent meta-analysis, Wolfowicz et al. (Reference Wolfowicz, Litmanovitz, Weisburd and Hasisi2021) found that juvenile delinquency is a strong predictor of supportive attitudes towards political violence. The present study’s findings suggest that this relationship’s direction may be contrary to prior assumptions, as we have found that a supportive attitude towards political violence (social resistance) predicts general delinquency. More generally, this issue relates to the notion that, just as attitudes harbour the potential to shape behaviours, behaviours can influence attitudes (McCauley and Moskalenko Reference McCauley and Moskalenko2017).
As for the other risk and protective factors included in our model, our results revealed that in accordance with previous research (Chan Reference Chan2019; Khoury-Kassabri et al. Reference Khoury-Kassabri, Mishna and Massarwi2019; Mohammad and Banse Reference Mohammad and Banse2023; Swisher and Shaw-Smith Reference Swisher and Shaw-Smith2015; Zhang et al. Reference Zhang, Zhao, Zhao and Ren2014; Zhao et al. Reference Zhao, Ren and Chen2023), high impulsivity and arrest of a family member are associated with higher levels of juvenile delinquency. Likewise, religiosity (relationship with God), parental attachment, attachment to the neighbourhood, employment and married parents all predict lower levels of juvenile delinquency.
However, in contrast with previous studies (e.g. Shong et al. Reference Shong, Bakar and Islam2019), which showed that high socio-economic status is associated with less youth crime, the opposite was true in our model. This result could be explained by previous findings whereby the economically advantaged are more prone to political participation (Castillo et al. Reference Castillo, Miranda, Bonhomme, Cox and Bascopé2014; Silalahi Reference Silalahi2022). In the context of East Jerusalem, as we have argued, juvenile delinquency is an act with a political core and, thus, more common among youth from high socio-economic levels. Our results also showed that age is not a significant predictor of juvenile delinquency, contrary to the findings of Steffensmeier, Lu, and Na (Reference Steffensmeier, Lu and Na2020). This finding aligns with previous studies on school pupils that did not find age to significantly affect juvenile delinquency (e.g. Khoury-Kassabri et al. Reference Khoury-Kassabri, Mishna and Massarwi2019).
Overall, our model suggests that the risk and protective factors identified in the literature as associated with juvenile delinquency are also helpful in predicting it in the context of a majority–minority political conflict. However, the novelty of the present study lies in the fact that it has unveiled one distinct element of juvenile delinquency in this special context – resistance to political order. We found that youth from East Jerusalem who have supportive views regarding political violence are more likely to be engaged in juvenile delinquency, while supportive attitudes towards general violence did not have a significant effect on such behaviour. Thus, we concluded that although this type of behaviour shares similar aspects in both regular and political contexts, it bears a unique political nature in the latter.
The present study also has practical implications for the prevention of juvenile delinquency. Given our finding that a main factor of juvenile delinquency in East Jerusalem is resistance to social order, the negative consequences of the Israeli–Palestinian conflict on youth behaviour could be mitigated by lowering levels of social resistance among them. One way of achieving this could be through a focused treatment that addresses aspects that affect this resistance, such as strengthening their attachment to broader Israeli society (Factor et al. Reference Factor, Mahalel, Rafaeli and Williams2013). Alternatively, intervention programmes could help them channel their frustration with the social order in more positive directions, such as participating in social change movements.
The present study is, of course, not without limitations. The first limitation regards our operationalization of resistance. As stated, we relied on the extent to which youth showed supportive attitudes towards political violence, reflecting resistance to the social order. However, this measure is merely a proxy and is not necessarily exhaustive of the full essence of youth’s resistance. Operationalizing political and social resistance with direct and comprehensive measures (e.g. see Factor et al. Reference Factor, Mahalel, Rafaeli and Williams2013) will be an important contribution to future studies.
A second limitation pertains to our juvenile delinquency and attitudes variables. As noted by Nunes, Pedneault, and Hermann (Reference Nunes, Pedneault and Hermann2021), the relationship between attitudes and actual behaviour raises the concern that attitude measures do not fully capture the nuanced attitudes towards specific types of delinquent acts. In the current study, it is notable that the items comprising the attitude variables may not exhibit the requisite granularity towards the specific types of behaviours encompassed within our delinquency measure. For instance, the attitude variable exclusively encompasses items related to violence, whereas the juvenile delinquency variable encompasses a broader range of juvenile behaviours, including non-violent acts. While we do not perceive this limitation as undermining the significance of our findings, we suggest that future studies explore these issues with more nuanced measurement techniques.
More broadly, another limitation is that our results are based on self-reports, which are generally open to bias, especially when asking participants about “sensitive” topics. In the present study, youth from East Jerusalem were asked to provide information regarding, for example, delinquent behaviours, attitudes towards violence, and socio-economic status, which could raise some concerns about the reliability of their responses. Future studies could overcome this limitation by validating the data with other sources of data (e.g. parents or police). A related limitation is the absence of data on school-level variables that could potentially influence the relationships explored in this study. This also presents an avenue into which future research could delve.
More limitations pertain to our sample and study context. First, our data were obtained from surveys conducted among male adolescents. However, the mechanisms we discovered could differ among female adolescents or adults. Second, we examined our research question in the specific context of a political conflict stemming from a majority–minority dispute. Thus, the patterns observed here may not necessarily apply to other forms of political conflicts. In the same vein, this study is limited to the context of East Jerusalem. Although this area shares similar features with other politically conflict-characterized areas (Corkalo Biruski Reference Corkalo Biruski, Alon and Bar-Tal2016; Knox Reference Knox2002), its uniqueness lies in local Arab inhabitants not being considered fully fledged citizens but permanent residents. This unique attribute of the study sample may affect the generalizability of this study. In the future, it would be important to examine whether the current study’s findings are also valid among other populations.
CONCLUSION
In the present study, we sought to shed light on juvenile delinquency mechanisms in the context of political conflict. The results suggested that several well-known risk and protective factors of juvenile delinquency are also helpful in explaining it under political conflict; however, this phenomenon is unique because one important predictor of it is resistance to political order. Scholars, as well as practitioners, should take this risk factor into account when interacting with delinquent youth exposed to political conflict.
Acknowledgements
This article is based on a study supported by a Ministry of Justice, Israel research grant. The authors would also like to thank the many young students who generously provided their time and support to make this study possible.
Eran Itskovich is a PhD candidate at the Institute of Criminology of the Hebrew University of Jerusalem. His research interests include the sociology of crime, wrongful convictions and political violence. He was awarded the Rothschild Academic Excellence Award for PhD Students in honour of Professor David Weisburd, the Robert Wistrich Prize for Outstanding Advanced Students, and the Olivier Vodoz Prize for the Study of Racism and Antisemitism.
Mona Khoury-Kassabri is a full professor, Vice President for Strategy and Diversity and previous Dean of the Paul Baerwald School of Social Work and Social Welfare at the Hebrew University of Jerusalem. Professor Khoury-Kassabri’s research revolves around issues related to child and youth welfare. It focuses on child and adolescent deviant and delinquent behaviours in three particular areas: school violence, cyberbullying, and juvenile delinquency and political violence. Her research examines how socio-political context influences child adolescent development and adjustment cross-culturally within Israel and in other societies.
Badi Hasisi is a full professor and Chair of the Institute of Criminology, Faculty of Law, The Hebrew University. His work focuses on the interaction between minority communities and criminal justice agencies. He also specializes in homeland security and the crime–terrorism nexus. He was awarded the 2018 prize for best article by the Israeli Organization of Law and History and the Fattal Prize for Excellence in Legal Research and Criminology. Professor Hasisi has also served as the Executive Editor of the Journal of Quantitative Criminology and acts as the current chair of the Israeli Society of Criminology.
Open access
