The Role of the Social Environment in the Relationship Between Group-Focused Enmity Towards Social Minorities and Politically Motivated Crime

Politically motivated crime against a person is characterised by the fact that the victims are interchangeable in the sense that individuals are seen as representatives of their social or ethnic group. The current study examines whether the existence of group-focused enmity (GFE) is linked to the perpetration of discriminatory behaviour, in the form of politically motivated crime, which is considered a classic example of the attitude–behaviour relationship. Although various survey studies show connections between group-focused enmity towards social minorities and discriminatory behavioural intentions, there is less knowledge available on whether attitudes can also be regarded as potential risk factors for actual behaviour. The role of the social environment in this relationship, as well as how the attitudes of the social environment are related to the perpetration of politically motivated crime, regardless of personal group-focused enmity, are further analysed. Using a representative student sample from the year 2018 (N = 2824), results show that anti-immigrant, anti-gay, ableist and anti-homeless attitudes are potential risk factors for the perpetration of politically motivated crime against these social minority groups, even after controlling for various other correlates of politically motivated crime. Group-focused enmity has an even greater effect when the student has a like-minded friendship group. Moreover, if the adolescent’s friendship group (hypothetically) approves of such discriminatory behaviour, the likelihood of acting in a politically violent manner is increased, even for adolescents who do not support group-focused enmity.


Introduction
Politically motivated crime against a person is characterised by the fact that the victims are interchangeable in the sense that individuals are seen as representatives of their social or ethnic group. In Germany, the Federal Ministry of the Interior records criminally relevant acts related to politically motivated crime, a statistic that has been increasing for years (Bundesministerium des Innern, für Bau und Heimat 2021). All prejudice-led behaviour against individuals poses a particular threat to our society. Such behaviour threatens the basic democratic values of our community and respect for human rights as laid out in the German constitution (Bundesministerium des Innern, für Bau und Heimat 2020a). Both the aim and outcome of such behaviour is to perpetuate the oppression of minorities. Persons who are discriminated against have fewer social opportunities to be regarded as equals and to have fair chances in the competition for important positions. Many of these discriminated persons thus experience a disadvantaged start (e.g. at school, in social and occupational settings) because of the prevailing prejudices against them. They are systematically rejected by unfair means, which is why this type of crime always represents an attack on democracy (Quent 2019, pp. 222-223).
One characteristic of politically motivated crime against a person of a certain social or ethnic group is that it can also send anxiety-inducing messages to other members of the affected group, which may significantly undermine the sense of security of those affected. Victims are thus left with hardly any opportunity to reduce their risk of victimisation by changing their own behaviour or habits given that the group labels applied by potential perpetrators can be assessed as largely unchangeable (e.g. origin, skin colour or sexual orientation) or have an identityforming function for the persons concerned (e.g. religious values, gender identity). Furthermore, it must be assumed that the psycho-traumatic consequences are more serious for victims of prejudice than for victims of comparable experiences that have a similar trauma potential but lack the prejudice component (Herek et al. 1999;Iganski 2001). Despite the increased trauma following experiences of victimisation, victims of prejudice are less likely to report such crimes. Concerns regarding retribution or the fear of unequal treatment by the responsible law enforcement authorities will most likely have an impact here (Craig-Henderson and Sloan 2003;Herek et al. 2002;Kite and Whitley 2016).
The present study is first aimed at shedding light on the frequency of politically motivated crime perpetration among adolescents in Schleswig-Holstein. In this context, adolescence constitutes a particularly important life phase for research purposes as adolescents are most politically impressionable (Sears and Levy 2003), begin to engage with politics (Eckstein 2018) and form political attitudes during this stage of life (Jaschke 2012). The federal state of Schleswig-Holstein is of relevance here because, as in Germany as a whole, the number of politically right-wing motivated crimes has increased in recent years. According to the Ministry of the Interior, Rural Areas, Integration and Equality of Schleswig-Holstein, a total of 709 right-wing politically motivated crimes were committed in 2019, of which 40 were violent acts. Regarding violent offences, this corresponds to an increase of almost 40% compared with the previous year (Ministerium für Inneres, ländliche Räume, Integration und Gleichstellung 2020). However, it must be noted that on the one hand, only crimes that come to the attention of the police are recorded and, on the other hand, crimes are only interpreted as politically motivated if this information existed when the police investigation began. Cases in which the political motivation is subsequently determined are not included in the official statistics. In addition, in 2019, a total of 1060 individuals were identified as belonging to the far-right scene in Schleswig-Holstein, of whom 360 were classified as violently inclined (Ministerium für Inneres, ländliche Räume, Integration und Gleichstellung 2020).
Besides the descriptive examination of politically motivated crime, another focus of the study is the analysis of motivations behind the perpetration of these crimes, for which the theory of planned behaviour (Ajzen 1985) is applied. Although vari-ous survey studies show connections between group-focused enmity (GFE) towards social minorities and discriminatory behavioural intentions (Asbrock et al. 2007;Küpper et al. 2016;Wagner et al. 2008;Zick et al. 2014), apart from experiments, there is less knowledge available on whether attitudes can also be regarded as potential risk factors for actual behaviour: 1. How is group-focused enmity towards social minorities related to the perpetration of politically motivated crime? Another important focus concerns the role of the social environment in this regard: 2. How are the attitudes of the social environment related to the perpetration of politically motivated crime? Moreover, little is known about how the influence of one's personal groupfocused enmity and the (hypothetical) reaction of the social environment to such behaviour interrelate: 3. Does group-focused enmity towards social minorities have a greater effect on politically motivated crime if the social environment endorses this? 4. Do the (hypothetical) positive reactions of the social environment have a greater effect on politically motivated crime if the person themselves supports group-focused enmity?
Heitmeyer's definition (2002a) of the Bielefeld disintegration approach, according to which various dimensions of misanthropic attitudes can be integrated into a common concept of group-focused enmity, was largely used as a basis for the measurement of group-focused misanthropic attitudes. According to Heitmeyer (2002a), an ideology of inequality forms a core feature of the concept of group-focused enmity, whereby the liberal-democratic principle of human fundamental equality (open or concealed) is suspended. The present study examines group-focused enmity that may relate to (alleged) ancestry (anti-immigrant attitudes), sexual orientation (antigay attitudes), or the impairment of social and economic participation (ableist and anti-homeless attitudes). It should be noted that anti-immigrant attitudes are often directed against individuals who are German as well.
Politically (right-wing) motivated crime is defined in accordance with the definition of the Federal Ministry of the Interior, Building and Community (Bundesministerium des Innern, für Bau und Heimat 2020b), which defines politically motivated crime as general crimes, but for which, under consideration of the attitudes of the perpetrator, there are indications that these crimes are directed against a person because of their political attitudes, nationality, ethnicity, race, skin colour, religion, ideology, origin, or because of their external appearance, disability, sexual orientation, or social status. It is important to note that the relevance of the criminal law regarding the various politically motivated crimes cannot be examined on a caseby-case basis in this study. For example, group-directed verbal abuse is unlikely to have constituted a criminal offence in all cases. However, owing to the lack of information on the criminal liability of each individual case of adolescent behaviour, these cases are also referred to as politically motivated crime.
In the next section, the theoretical background with regard to the theory of planned behaviour and the current state of research is explained, followed by the formulation of hypotheses. A description of the student survey and the sample is given in Sect. 3, whereas the operationalization of the concepts is presented in Sect. 4. This is followed by a description of the analytic strategy and a presentation of the results in Sect. 5. In addition to the purely descriptive presentation of the prevalence of politically motivated behaviours, logistic regression results and the conditional average K marginal effects are then presented. The final section summarises and discusses the results, implications, and limitations of the study.

Theoretical Background and Current State of Research
In recent decades, extremism research in Germany has been strongly influenced by the methodological approach of attitude research (e.g. Heitmeyer 2002b) or the authoritarianism studies of Leipzig University (Decker and Brähler 2018;2020) or Bielefeld University (Zick and Küpper 2021). The relationship between such attitudes and political violence or discriminatory behaviour as a preliminary stage of politically motivated crime has so far also been investigated in a few studies. On the basis of data from a student survey in Lower Saxony, Germany, Kliem et al. (2018) showed that a right-wing extremist world view can be associated with an increased risk of committing various crimes against "non-German" citizens. Even when further correlates of right-wing extremism were included, anti-immigrant attitudes, anti-Muslim attitudes, chauvinism, anti-Semitism and the advocacy of a rightwing authoritarian dictatorship were related to the perpetration of right-wing crimes, as well as violence against foreign persons and persons with left-wing political attitudes . Several meta-analyses have been devoted to this topic as well (Dovidio et al. 1996;Jones et al. 2017;Schütz and Six 1996;Talaska et al. 2008). Talaska et al. (2008) pointed to a moderate connection between prejudiced attitudes and behaviour. In their study, most of the explanatory power was attributed to the affective facet of the attitudinal dimension (Talaska et al. 2008). A prior metaanalysis based on 60 independent studies, carried out by Schütz and Six (1996), reported a moderate relationship, with a mean correlation of r = 0.36 between prejudice and behaviour and of r = 0.45 between prejudice and behavioural intentions (Schütz and Six 1996). Dovidio et al. (1996) analysed 23 studies and also found a medium correlation between white people's racial attitudes and discrimination (r = 0.32) (Dovidio et al. 1996). The meta-analysis by Jones et al. (2017), which examined this relationship in an occupational environment, found evidence that racism and ageism predict discriminatory selection and opposition to organizational policies that are designed to benefit the respective (out-)group. Beyond that, racism was a predictor for discriminatory performance evaluation, whereas sexism did not relate to any of the forms of workplace discrimination examined (Jones et al. 2017). However, most of the studies included in the meta-analyses were laboratory experiments, the results of which cannot be easily applied to the real world. External validity constitutes a common issue within psychological experiments, mainly because the subjects are often (psychology) students whose voluntary participation is generally related to their interest in science or their interest in the monetary reward often given for participation. In addition, psychology courses often require their students to serve as subjects for a predetermined number of hours. This selection of subjects represents an opportunity sample and is thus systematically different from the overall population. Such participants are not even representative of the subpopulation of students because of the lack of random selection. The results cannot be easily generalized to other groups of people. A study by Carlsson and Eriksson (2017) provides an exam-ple of a study that tested the relationship between attitudes and behaviour outside the laboratory. In their study, they found evidence that landlords were more likely to discriminate in regions with more negative attitudes towards ethnic minorities. They concluded that survey evidence of people's attitudes, a relatively simple and inexpensive way to obtain information, is a useful predictor of ethnic discrimination (Carlsson and Eriksson 2017, p. 14).
The studies conducted thus far have, for example, been based on the situational action theory, which puts moral attitudes in the foreground (Wikström and Bouhana 2016;Gali et al., 2018;Pauwels and Svensson 2017;Pauwels et al. 2021;Schils and Pauwels 2016) or the justification-suppression model (Jones et al. 2017). Others consider this link to be a classic example of the attitude-behaviour relationship (Schütz and Six 1996). This relationship is subject to the theory of reasoned action (Ajzen and Fishbein 1980), as well as its further development as the theory of planned behaviour (Ajzen 1985), and forms the theoretical foundation for this study. According to the theory of reasoned action, intention is the best predictor of behaviour providing there is sufficient motivation, time and mental capacity. The intention in turn depends on the attitudes towards the behaviour. According to Ajzen and Fishbein (1980), a person will perform a behaviour if they evaluate it positively. In this study, group-focused enmity represents the attitude towards the behaviour. This addresses the question of whether the behaviours considered in this study can truly be regarded as politically motivated. In this context, the hypothesis is: H 1 The more pronounced the adolescents' personal group-focused enmity towards social minorities is, the greater the likelihood of the perpetration of a politically motivated crime.
The intention may also depend on the subjective norms that the potential perpetrator perceives in their social environment. A person will perform a behaviour if they believe that individuals of importance to them would also evaluate this behaviour positively (Ajzen and Fishbein 1980). The role of the social environment in relation to prejudice or other forms of group-focused enmity towards social minorities has thus far been studied in many experiments. For example, Stangor et al. (2001) show that learning about the racial beliefs of others has the potential to either promote or inhibit changes in personal stereotypes. Another experiment goes even further by concluding that the realisation that one's own beliefs are shared with others influences both attitudes and behaviours, as well as the strength of the attitude-behaviour relationship . Besides experiments, there is little evidence available on the influence of the social environment on discriminatory behaviour. In the context of prejudiced attitudes, correlational studies find that people adjust their perceived norms regarding prejudice, evaluations of discriminatory scenarios and reactions to misanthropic jokes in line with others (Crandall et al. 2002). Regarding adolescents, more recent longitudinal studies conclude that, over time, the level of prejudice among peers affects personal prejudice (Hjerm et al. 2018; see also Mitchell 2019;Miklikowska et al. 2019a) and that adolescents adjust their attitudes in line with classmates whom they perceive to be dominant, but not towards their teachers (Váradi et al. 2021). Moreover, there is evidence of a parent-child corre-lation regarding intergroup and right-wing extremist attitudes (Degner and Dalege 2013;Haselbach et al. 2021;Jugert et al. 2016;Meeusen and Dhont 2015). Differentiated analyses show that mothers are the primary influence for prejudice against HIV/AIDS, obesity and race, whereas fathers are more influential regarding sexism and attitudes aimed at homosexuals (O'Bryan et al. 2004).
In this study, the subjectively assessed group-focused enmity of the adolescents' parents, teachers and friends, as well as individual assessments of how the social environment would judge politically motivated crime, are used to measure the subjective norm. A distinction is made between these two constructs, as the intensity of rejection is affected if a person "merely" has a negative attitude towards a group of people in general (group-focused enmity) or if they already approve of a specific, politically motivated crime. In this way, it can be distinguished whether the general negative attitude in the social environment or the advocacy of a crime in the social environment alone leads to the perpetration of a politically motivated crime. The hypotheses for the subjective norm are: H 2 The more pronounced the group-focused enmity towards social minorities is in the social environment, the more likely the adolescent is to commit a politically motivated crime.
H 3 If the social environment endorses politically motivated crime, it becomes more likely that an adolescent commits such a crime.
Because according to the theory of planned behaviour both one's personal attitudes and the assessment of the social environment have a positive effect on behaviour, the hypotheses are: H 4 The relationship between adolescents' personal group-focused enmity towards social minorities and politically motivated crime is stronger if the social environment endorses these crimes.
H 5 The relationship between the endorsement of politically motivated crime in the social environment and the perpetration of politically motivated crime is stronger if an adolescent themselves supports group-focused enmity towards social minorities.
Hypotheses H 4 and H 5 should not be considered as competing with Hypotheses H 1, 2 and 3. For most people, it will be true that a behaviour will only be performed when the personal attitude towards the behaviour, as well as the subjective norm, is positively inclined towards the behaviour. However, if the person has no relevant significant individuals, the respective personal attitudes will be of sole importance. Apart from that, it is also possible that an individual is so strongly anchored in their group that the subjectively experienced social pressure is the only determinant of their behaviour, and the significance of their personal attitudes recedes into the back-ground (Ajzen 1985). According to Ajzen (1985), the relationship can be described with the following formula 1 : The theory of reasoned action can, however, only be applied if the behaviour is under volitional control (Ajzen 1985). For this reason, Ajzen (1985) has further developed the theory into the theory of planned behaviour. This theory contains the same components as the theory of reasoned action but adds a third dependent factor to account for barriers outside of one's control: the perceived behavioural control, which determines a person's perception of how easy or difficult it is for them to perform the behaviour. If the attitude towards the behaviour, the subjective norm and the perceived control of the behaviour all lead to a positive result, then a person will most likely formulate a behavioural intention and the behaviour can be predicted relatively well (Ajzen and Fishbein 1980). As the study has no method of measuring the perceived control of the behaviour, it is assumed that the politically motivated crime is completely volitional. It must also be pointed out that attitudes and the subjective norm are considered in relation to behaviour and not in relation to the underlying intention. For this reason, somewhat smaller effects are expected than if the relationship were to be examined in terms of intention.

Control Variables
Prejudice is not the only necessary condition for the perpetration of politically motivated crime against social minorities (Allport 1954). It is almost certain that this conclusion can also be applied to other forms of group-focused enmity given that there is no monocausal theory that can comprehensively explain the perpetration of such crimes. Rather, the emergence of these behaviours can be understood in the context of an interplay of several explanatory approaches (regarding the problem of complexity within explanatory approaches in the field of right-wing extremism research, see Eicker 2021). If the analyses were limited to the consideration of an individual's group-focused enmity and their social environment alone, the relationship between these attitudes and politically motivated behaviours may be overestimated. Therefore, this study draws on a risk and protective factor approach to identifying other relevant determinants of the perpetration of these crimes, which are to be used as control variables (for meta-analyses regarding several risk and protective factors, see for example Wolfowicz et al. 2019;Jahnke et al. 2021). Beelmann (2020), who compiles various risk and protective factors in his social-developmental model of radicalization, is of particular relevance here.
In particular, the control variables include parental violence that is associated with anti-immigrant attitudes and extreme right-wing behaviour and crimes (Baier et al. 2009(Baier et al. , 2010Wetzels and Greve 2001). If individuals have contact experiences with the extreme right-wing scene, this represents another situation in the social environment that makes the adoption of the prevailing attitudes and behaviours more likely. Interethnic contact, on the other hand, reduces prejudices against foreign groups, according to Allport's (1954) contact hypothesis. Empirical studies indicate that individuals who maintain a friendship with a member of a foreign group or engage in other forms of social contact with foreign groups show fewer prejudices and right-wing extremist attitudes (Fuchs 2003;Hamberger and Hewstone 1997;Maresch and Bliesener 2015;Pettigrew 1997;Pettigrew et al. 2011;Pettigrew and Tropp 2006;Wagner et al. 1989;Wagner and van Dick 2001), which could in turn be related to engaging in less discriminatory behaviour.
Other variables that are controlled for can be considered as personality factors such as self-control, which has been discussed within the framework of hate crime literature for a considerable period of time (Walters 2011). Various studies link risk-seeking behaviour, as a dimension of self-control, to membership of rightwing groups (Bjørgo 2002;Watts 2001), the perpetration of right-wing extremist crime , and indirectly to political violence (Pauwels and Heylen 2017). Another component of personality that is included in the analysis of this study are aggressive tendencies. The perpetrators' general propensity for violence often leads to the adoption of right-wing extremist attitudes, which is used to legitimise right-wing crimes (Heitmeyer 1994;Krüger 2008;Maresch and Bliesener 2015). Previous studies show connections between the affinity for violence and right-wing perpetration of violence (Fuchs 2003;. A further characteristic that is negatively related to right-wing extremism is the capacity for empathy. A poorly developed ability to feel compassion for others and to adopt the perspective of others can lead to intolerant and misanthropic attitudes Bäckström and Björklund 2007). Previous studies show that low levels of empathy can thus be accompanied by devaluations of certain groups (Heyder 2005) and the perpetration of right-wing extremist crime . Another personality trait that is used to explain right-wing extremism is hierarchical selfinterest, which refers to an expression of the logic of the market and the individual striving to be "better than others"; the term can thus be understood as an internalized elbow mentality (Hadjar 2005). Hierarchical self-interest is primarily a construct of values that is reflected at the individual level within the dimensions of performance orientation, Machiavellianism, competitive thinking and individualism. Economic change and economic crises can increase the importance of personal success and retention of one's status, and lead to a "survival of the fittest" mindset (Hagan et al. 1999). The maintenance of one's own status in society can lead to the denial of rank to other groups of people and ultimately to their complete devaluation, which has been demonstrated in previous studies (e.g. Maresch and Bliesener 2015;Hagan et al. 1999).
Social structural factors, which are discussed to explain the genesis of rightwing extremist attitudes and behaviour, particularly within the framework of deprivation theories, are also controlled for. Groups identified as "other," "different" or "abnormal" are scapegoated as being responsible for the subjectively perceived gap between the (non-)fulfilment of certain wishes and expectations (mostly of an economic nature). In many studies, individual and collective deprivation is linked to right-wing extremism (Baier et al. 2016;Doosje et al. 2012;Heyder and Gaßner 2012;Rippl and Baier 2005;Zick and Wagner 1998). As various indicators of deprivation (i.e. subjectively assessed family financial situation, individual relative deprivation, as well as collective absolute and relative deprivation) were identified as insignificant in the context of discriminatory behaviour among students in previous analyses relating to this study , only the welfare status, the school type and social deprivation are used as correlates of politically motivated crime in the current study. In contrast to the other subjectively assessed indicators of deprivation, which primarily relate to economic disadvantage, social deprivation refers to disadvantage in terms of social integration opportunities and enquires about the level of social support in the adolescents' close community.

Data and Participants
The study (Regional analyses of right-wing extremism in Schleswig-Holstein) is based on a representative student survey that focuses on right-wing extremism in the seventh and ninth grades of general schools in Schleswig-Holstein, Germany . The study was approved by the Ministry of Education and Science of Schleswig-Holstein and was carried out between February and June 2018. It was previously conducted in a comparable form in 2013 and 2014 (Maresch and Bliesener 2015). In order to ensure representative sampling, school classes were randomly selected on the basis of class lists provided by the Statistical Office for Hamburg and Schleswig-Holstein. The sample was stratified according to the three predominant school types in Schleswig-Holstein (upper secondary school (Gymnasium) and lower and intermediate secondary school (Gemeinschaftsschule, Regionalschule)), the adolescents' age group (7th and 9th grade), and the four regional court districts of Schleswig-Holstein (Lübeck, Kiel, Itzehoe, Flensburg). The survey was designed in the form of a computer-administrated group survey, which was conducted in the computer rooms of the respective schools in the presence of at least one teacher as well as a trained test leader. The test leaders were responsible for accessing the survey link on the computers and were also the main contact for any inquiries. The students completed the questionnaire independently, but remained seated together in the classroom. The duration of the survey was approximately one school lesson (45 min). Prior to the conduction of the survey, the adolescents' parents were informed about the aims of the study, including the survey contents, by means of information leaflets. Only those students whose parents had explicitly provided consent to their child's participation in the study were able to take part in the survey. The students were also informed that (a) survey participation was voluntary, (b) the answers were processed anonymously, (c) they could withdraw from the survey at any time, (d) individual questions could remain unanswered and (e) non-participation would not result in any disadvantages. A survey response rate of 43.5% was determined at the school level. The main reasons for refusal to partic-K ipate in the survey given by the school principals included limited time, a shortage of teaching staff and problems within the school. At the student level, the response rate was 69.2%, with the reasons for non-participation mostly referring to the illness of a pupil or the lack of parental consent (including students who forgot to return the consent form to their teachers). Overall, this resulted in a combined response rate of approximately 30.1%. Overall, n = 1337 pupils in the seventh and n = 1487 pupils in the ninth grade were successfully surveyed, which corresponds to a total sample of N = 2824. Approximately the same number of male (50.7%) and female (49.3%) adolescents participated in the survey. The average age was 14.7 years (SD = 1.2), with a range of 12 to 19 years. Approximately 6.5% of the students' families were dependent on receiving welfare benefits. Almost a quarter of young people had a migration background (23.2%). The majority of young people with a migration background had a Turkish (3.5%), Polish (3.4%) or Russian background (3.0%). Table 1 provides an overview of the sample. Table 2 shows the survey instruments, response options, the source of the respective items, and Cronbach's α as a measure of internal consistency.

Dependent Variable
Politically Motivated Crime. The politically motivated crimes constitute the dependent variable within this study. This was measured by asking the students whether they had ever verbally abused an individual who was foreign, homosexual, homeless, or had a disability. Furthermore, the adolescents were asked to report whether   (-): The items were reversed for the analyses K they had ever verbally threatened or damaged the property of a person belonging to one of the aforementioned groups. In addition, the students were also asked whether they had ever committed bodily harm against an individual belonging to these social minority groups or threatened one of these individuals with a weapon. For the analysis, a maximum value scale was formed for the politically motivated crimes (0 = no politically motivated crime; 1 = at least one count of politically motivated crime). If the students had not committed any of the five offences, they were assigned the value 0, whereas, in cases where the students had committed at least one or more of the five offences, they were assigned the value 1. For example, if a student had verbally abused someone, but not threatened them with a weapon, the response to the verbal abuse question was included in the index. This type of operationalisation was chosen because this study does not have differential assumptions regarding the relationship between group-focused enmity and minor offences (e.g. verbal abuse), serious offences (e.g. assault), or the number of crimes committed. It is assumed that the independent variables can explain the different degrees of serious crime to the same extent.

1) I have previously taken part in actions of right-wing organisations (distributing flyers, demonstrations, commemoration marches) 2) I have previously read a flyer from a right-wing organisation/group 3) I have previously been on the Internet on the site of a right-wing organisation/group 4) I have previously received an e-mail from a right-wing organisation/group 5) I have listened to music by at least one of the following groups
In order to do justice to the explanation of the crimes perpetrated against the various social groups, a different model is calculated for each of the four groups. Furthermore, I formed an overall index referring to the perpetration of at least one politically motivated crime against at least one of the social minority groups. Thus, only those students who had not committed any of the politically motivated crimes against any of the four groups are given the value 0. Students who had committed at least one offence against at least one group are given the value 1.

Attitudes
In order to operationalise anti-immigrant, anti-gay, ableist and anti-homeless attitudes, the adolescents were asked to rate between two and three different statements per dimension on a four-point response scale (1 = do not agree at all; 4 = fully agree). The exact item formulations can be found in Table 2. Using these two or three items (depending on the dimension), mean value scales were formed for each dimension. In addition, a mean value scale was also used to create an overall scale for all four dimensions.

Subjective Norm
Two forms of operationalisation are chosen for the measurement of the subjective norm (for explanation see Sect. 2). Introduced by the question "How much would the following person agree with these statements?", the students were asked to indicate the extent to which their mother, father, best friend and their class teacher would each agree to statements concerning anti-immigrant, anti-Muslim and antigay attitudes on a four-point response scale (1 = do not agree at all; 4 = fully agree). For the analysis, I formed four mean scales relating to the extent of group-focused K enmity measured using all three statements for the mother, father, best friend and class teacher.
In addition, using a three-point response scale, the students were asked to indicate how their mother, father, teacher or friendship group had reacted, or would react, to the adolescents' perpetration of one of the politically motivated crimes. Four dummy variables were then formed for the mother, father, teacher and friendship group, which compare the positive reaction (=1) with the negative and neutral reaction (=0). The comparison of the positive reaction to the negative and neutral reaction is based on the theory of planned behaviour, in which it is assumed that a person's actions are influenced by the fact that they believe that important people in their social environment would evaluate the behaviour positively. The theory thus formulates an effect of a positive (hypothetical) reaction to the behaviour, which is taken into account by this type of operationalisation.

Control Variables
Furthermore, control variables such as sex (0 = male; 1 = female), migration background (0 = no migration background; 1 = migration background) 2 , age (12-19), school type (0 = upper secondary school; 1 = lower/intermediate secondary school) and welfare status were included in the analyses. Self-control, hostility/aggressive tendencies, empathy, hierarchical self-interest, social deprivation and interethnic contact were all assessed by multiple items that were formed into mean value scales (see Table 2 for further information). The frequency of parental violence was recorded using a six-point scale (0 = never; 1 = 1 or 2 times; 2 = 3-12 times; 3 = several times per month; 4 = once per week; 5 = several times per week), whereby the statements for mothers and fathers were asked separately (see Table 2). For the analyses, the answers for mothers and fathers were combined. For example, if a student had only experienced violence at the hands of the father but not the mother, only the information relating to the father would be considered. Following this, I formed a mean value scale, which indicates the average frequency with which pupils had experienced parental violence. The adolescents' contact with the right-wing extremist scene was measured by asking about 11 different contact possibilities, such as wearing certain right-wing brands or listening to right-wing music (see Table 2 for all items). I then formed a sum scale of all 11 items.

Analytical Strategy
As an initial step, the prevalence rates for the politically motivated crimes are presented. For the further analysis of the relationship between group-focused enmity and the politically motivated crimes perpetrated against social groups, I calculated logistic regression models, which control for various other correlates of politically right-wing motivated violence, using Stata 14.2 (StataCorp 2015). To address the hierarchical data structure of students nested within classrooms, I employed clustered robust standard errors. To account for missing data, I applied multiple imputation by chained equations to estimate the missing data for all variables, apart from the dependent variables (proportion of missing values of the items: 1.0-17.2%). In total, the regression models are based on n = 2741 pupils from 171 classes. Owing to the large sample sizes and the associated pronounced statistical power, even marginal group differences are able to reach statistical significance. For this reason, one should look at the strength of the effects, especially for the findings that only become significant at the 5% level. I calculated average marginal effects (AMEs) for the different politically motivated crimes. AMEs indicate the mean change of the likelihood of the politically motivated crime in percentage points if the relevant explanatory variable changes by one unit. In addition, conditional average marginal effects, which take different combinations of the independent variables into account, are presented. With these analyses, it can be shown whether personal group-focused enmity towards social minorities is more decisive for the perpetration of politically motivated crimes when the social environment endorses these crimes and whether the social environment can increase the likelihood of politically motivated crime even further if the student themselves supports group-focused enmity towards social minorities. Table 3 shows the number and proportion of pupils who reported having perpetrated the different politically motivated crimes. It also displays the proportion of adolescents who had engaged in at least one of the types of behaviour. The last column presents an index of politically motivated crimes perpetrated against at least one of the social minority groups. As expected, it can be stated that the less serious acts (verbal abuse (any group: 10.6%) and verbal threats (any group: 5.9%)) are more prevalent than the more serious crimes (e.g. assault (any group: 3.9%) or threats using a weapon (any group: 1.8%)). Property damage, which can also be regarded as somewhat less serious, only has a low prevalence rate (any group: 2.3%). At least one offence (unspecific to the group) was reported by 13.5% of the surveyed adoles- K cents. Overall, most politically motivated crimes were carried out against foreigners (8.8%), followed by people with disabilities (5.6%), homosexual persons (3.3%) and the homeless (1.8%).

Results from Logistic Regression Models
The analyses (Table 4) show that anti-immigrant, anti-gay, ableist and anti-homeless attitudes are related to the perpetration of politically motivated crime against the respective group. The greater the extent of group-focused enmity towards the specific group, the greater the likelihood of the student reporting at least one type of politically motivated crime against foreign people (AME = 0.049, p < 0.001), the homeless (AME = 0.009, p = 0.007), people with disabilities (AME = 0.021, p < 0.001), homosexual persons (AME = 0.022, p < 0.001) or at least one of the social minority groups (AME = 0.088, p < 0.001). Hypothesis 1 can therefore be confirmed. With regard to the attitudes of the adolescents' social environment, I found that the attitudes of the adolescents' best friend are related to the perpetration of politically motivated crime against foreign persons and at least one of the social minority groups. The more the students' best friend supported anti-gay, anti-immigrant and anti-Muslim attitudes, the more likely the students were to perpetrate politically motivated crimes against foreign people (AME = 0.027, p < 0.001) and at least one of the social minority groups (AME = 0.023, p = 0.016). In addition, a positive assessment (versus a negative/neutral assessment) of how the friendship group (AME = 0.062, p < 0.001) would, or did, react to an offence increased the adolescents' likelihood of engaging in politically motivated crime against foreigners. The positive reaction of the friendship group is also related to the perpetration of politically motivated crime against people with disabilities (AME = 0.049, p = 0.003) and at least one of the social minority groups (AME = 0.099, p < 0.001). Overall, it appears that the general negative attitude of the best friend towards a group of people (group-focused enmity) already increases the probability of committing an offence, although the specific positive reaction is the greater factor when the strength of the effects is compared. The following analyses in Sect. 5.4 therefore only refer to the (hypothetical) positive reaction as the stronger variable. Hypotheses 2 and 3 are thus confirmed for the friendship group aspect of the adolescents' social environment.
Furthermore, the greater the teacher's group-focused enmity, the more likely the student was to engage in the aforementioned behaviours against people with disabilities (AME = 0.011, p = 0.040). In contrast, a (hypothetical) positive reaction by the teacher reduced the probability of the adolescents perpetrating a criminal act against a foreign (AME = -0.042, p = 0.024) or homeless person (AME = -0.015, p> = 0.009). Given that only 2.4% of the teachers evaluated this behaviour positively, and thus the cell is very sparsely filled, this finding is not given any greater significance and may therefore be more of a statistical artefact. In contrast to the theoretical expectations, more anti-gay, anti-Muslim and anti-immigrant attitudes held by the adolescents' mother was associated with fewer politically motivated behaviours perpetrated against homosexual persons (AME = -0.016, p = 0.023). Owing to these differing results and their small effect sizes, in combination with the significance levels that were scarcely reached, no definitive statements regarding the K effect of the adolescents' social environment, other than for the effect of the friends, can be made. Hypotheses 2 and 3 are only confirmed for the friendship group aspect of the adolescents' social environment, but not for the other aspects of the social environment.
With regard to the control variables, I found that gender, migration background, school type, the receipt of welfare benefits, self-control, empathy, hostility/ aggressive tendencies, empathy, social deprivation, parental violence and contact with the right-wing extremist scene are related to politically motivated crime against some of the social minority groups. When considering the explained variance of the different models, explanatory power between 20.2 and 35.2% was obtained. The level of the explained variance thus allows the conclusion that relevant explanatory factors of politically motivated crime were taken into account.

Conditional Average Marginal Effects
To answer the question of whether adolescents' personal group-focused enmity has a greater effect on the perpetration of politically motivated crimes in cases where the social environment endorses these, conditional average marginal effects are presented. Figure 1 shows the average marginal effects of adolescents' personal groupfocused enmity on perpetrating a politically motivated offence against one of the social minority groups according to the different reactions of the social environment (i.e. mother, father, teacher, friendship group) to such behaviour. The results showed that regardless of whether the social environment would support such an act, adolescents' personal group-focused enmity was always significantly related to the perpetration of such a behaviour. However, the results indicate an amplification effect in relation to the adolescents' friendship group, as a positive evaluation of such an act by friends solidifies the effect of adolescents' group-focused enmity on politically motivated crime. If the adolescents' friendship group reacts negatively or neutrally to one of the aforementioned crimes, one more point on the adolescents' personal group-focused enmity scale increases the average probability of perpetrating a politically motivated crime by 8.3 percentage points (p < 0.001). If the adolescents' friendship group reacts positively, one more point on the adolescents' personal group-focused enmity scale increases the average probability of perpetrating such an act by 13.1 percentage points (p < 0.001). Hypothesis 4 can therefore be confirmed in relation to the friendship group aspect of the social environment. For the evaluation of the adolescents' mother, father and teacher, no reinforcing effects of their reactions were observed. The effect of adolescents' personal group-focused enmity on their behaviour hardly changes when these three persons react differently, which is why H 4 cannot be confirmed for these aspects of the social environment. Furthermore, the results can answer the question of whether the social environment's (hypothetical) positive reactions to a politically motivated crime have a greater effect on such behaviour if the adolescent supports group-focused enmity themselves. Figure 2 shows the average marginal effects of the friends group's positive reaction to an offence on perpetrating a politically motivated offence against one of the social minority groups according to different levels of adolescents' personal group-focused enmity. The results show that even if the adolescent themselves holds no personal group-focused enmity, the positive reaction of their friendship group increases the average probability of the perpetration of politically motivated crime by 6.0 percentage points (scale point 1; p < 0.001). However, this average probability is even higher the more pronounced adolescents' personal group-focused enmity towards social minorities is. For example, a positive response by the friendship group can increase the average probability of engaging in such behaviour by 19.7 percent- do not agree at all (***) do rather not agree (***) do rather agree (***) fully agree (***) Fig. 2 Average marginal effects of a positive reaction by the adolescents' friendship group on at least one politically motivated offence against at least one of the social minority groups according to the different levels of adolescents' personal group focused enmity (with 95% CIs) age points in cases where the adolescent holds strong group-focused enmity (scale point 4; p < 0.001). The positive reaction of the adolescents' friendship group is thus always decisive. Even if the respondent does not support any group-focused enmity, the reaction of their friendship group can lead to the perpetration of politically motivated crime, although the effect is greater if the adolescent themselves holds these attitudes. Hypothesis 5 can therefore be confirmed in relation to the friendship group aspect of the social environment. The reactions of the adolescents' mother, father, and teacher were not related to politically motivated crime against at least one of the social minority groups, not even in cases where the respondent themselves strongly supports group-focused enmity towards social minorities.

Summary and Discussion
Group-focused enmity is not a marginal social phenomenon. On the one hand, the reduction of anti-immigrant, anti-homeless, ableist and anti-gay attitudes in Germany means that all members of society can live in a tolerant and diverse country, can orient themselves to a liberal-democratic basic order, and can build on the basis of a strong civil society. On the other hand, such pejorative attitudes are strongly related to politically motivated crimes. Using a sample of students from Schleswig-Holstein, Germany, the aim of the present study was to investigate this relationship and further study the role of the social environment in this context. With regard to the prevalence of politically motivated crimes, the present study shows, as expected, that serious criminal offences (i.e. assault or threats using a weapon) are reported less frequently than more minor offences (i.e. verbal abuse and verbal threats). Results of the logistic regression models show that group-focused enmity towards the respective social minority group show a significant correlation with politically motivated crimes perpetrated against the group, even after controlling for further influencing variables, which confirms H 1. Given that this strong relationship was found, the behaviours considered in this study can indeed be generally seen as politically motivated.
Beyond the aforementioned studies, this analysis not only examined the classic attitude-behaviour relationship but also included the role of the social environment. The results showed that the extent of group-focused enmity attitudes towards social minorities held by the adolescents' best friend and the friendship group's (hypothetical) reaction to an offence were also associated with the adolescents' perpetration of politically motivated crime, although the latter has the greater impact. Hypotheses 2 and 3 can therefore be confirmed for the friends (both best friend and friendship group) aspect of the social environment, but not for the adolescents' parents or teacher.
Furthermore, the analyses show that adolescents' personal group-focused enmity is nevertheless significantly related to the perpetration of politically motivated crime, even if the social environment does not endorse the offence. However, reinforcing effects were found in relation to a positive reaction by the adolescents' friendship group, which confirms H 4 for this type of social environment. The current study also reveals that even if the respondent does not support group-focused enmity towards social minorities, the friendship group's supportive response to such an act increases the likelihood of perpetrating politically motivated crime. However, the relationship between the friendship group's positive reaction and the perpetration of politically motivated crime becomes even stronger when the adolescent themselves holds anti-immigrant, anti-homeless, ableist or anti-gay attitudes, which confirms H 5 in relation to the friendship group. The current results clearly demonstrate the importance of the social context's influence, which is why further studies should examine other context factors, e.g. the proportion of migrants in the school class or the proportion of adolescents whose parents receive social welfare. The peer context appears to be especially relevant, meaning that social pressure seems to play an important role.

Implications
In order to reduce politically motivated crime in the future, individuals who, for example, experience the perpetration of discrimination, become aware of resentment towards certain social groups, or overhear right-wing extremist remarks in their peer group, should address these issues with the individual(s) concerned and attempt to start a dialogue regarding the background of these attitudes and behaviours. The aim is to encourage young people to reflect on their own attitudes and behaviour.
The fact that people in the social environment can reduce the level of politically motivated crime was only found in relation to the best friend and the friendship group, but not for the other social environments, which could be because peers constitute the main source of identification, especially in adolescence. Peers thus have an immense impact on the formation of personal attitudes (Nesdale 2001), which can thereby also affect behaviour. One explanation could be that adolescents spend more time with their peers than with their parents (Larson et al. 1996). Moreover, the influence of parents decreases during the adolescence stage, the influence of friendship groups increases, confirming the idea that friends become even more important in the lives of adolescents (De Goede et al. 2009). Whether parental attitudes can influence their children's behaviour may be more dependent on factors such as the level of children's identification with their parents (Sinclair et al. 2005), parental support (Miklikowska 2016) or on how salient the attitudes are at home (Levy and Hughes 2009). How much influence the adolescents' teacher has on the perpetration of politically motivated behaviours could also depend on these factors. Furthermore, the results highlight the importance of intervening as early as the adoption of group-focused enmity in order to reduce politically motivated crime. Given that schools are an important socialisation factor regarding the formation of political attitudes in adolescence (Rippl 2005) and because extensive and universal field access can be easily attained in this setting, specific prevention approaches tailored to discriminatory and politically motivated behaviours should be further developed within the school context. According to the present study, teachers' group-focused enmity is somewhat unrelated to students' politically motivated behaviours. However, studies indicate that general social support provided by teachers can counteract the development of prejudice and facilitate social trust among adolescents (Miklikowska et al. 2019b). In addition, the use of "student leaders", who are trained to challenge instances of inter-group prejudice, shows promising results (Paluck 2011). A stronger curriculum focus on current socio-political issues, as well as the importance of freedom and an open society, also appears to be relevant. For example, various authors (Kuhn and Buhl 2011;Watermann 2003;Westle 2012;) point out that political knowledge is in a contradictory relationship to group-focused enmity or right-wing extremist attitudes.

Limitations
It must be noted that the design of the study only allowed for the surveying of students in the seventh and ninth grades, which means that reliable statements are restricted to this age group. Future surveys should, if possible, also record prevalence rates for different age and/or class levels. Such differentiated analyses appear to be of great relevance, particularly in relation to suitable age phases regarding the implementation of prevention programmes . As in all survey studies, the data may have been falsified by the students owing to processes of social desirability or deliberate deception. A study by Köllisch and Oberwittler (2004), which compared students' subjective reports of their police contact with actual police records, analyses how reliable student reports are. In general, they found a high validity of the adolescents' self-reports. However, there seems to be a tendency to underreport police contact among respondents with a high educational status. Therefore, the results of survey studies on self-reported juvenile delinquency must be interpreted with caution. In relation to the present study, underreporting may pose a potential issue for a sensitive topic such as politically motivated crime. Especially among adolescents with a high educational status, the connection between groupfocused enmity towards social minorities and politically motivated crime could be underestimated if students concealed their crimes. We attempted to limit processes of social desirability by explaining the underlying data protection mechanisms (e.g. anonymous survey, no feedback to parents, school, or teachers) in detail to the participating students.
The current student survey is a cross-sectional survey, meaning that statements regarding causal relationships cannot ultimately be made. However, a longitudinal analysis also supports the causal hypothesis that prejudice leads to discriminatory intentions (Wagner et al. 2008). Future studies should shed further light on the connection between group-focused enmity and actual politically motivated behaviours within the framework of longitudinal research.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4. 0/.