Planned hooligan fights: Contributing factors and significance for individuals who take part

Several European countries have recently been confronted with mutually arranged confrontations between hooligan groups in a predesignated setting. This article explores the significance of this form of collective violence for those involved and how this relates to existing collective violence theory. In addition to international and national questionnaires and subsequent in-depth interviews with police officials, two case studies were conducted and compared with a ‘regular’ (not mutually arranged) hooligan confrontation. We also assessed the criminal history and psychological traits of individuals participating in mutually arranged fights (n = 38) and individuals taking part in a regular confrontation (n = 76). Our results indicate that the meaning of mutually arranged confrontations differs importantly from that of spontaneous collective violence. Furthermore, data indicate that criminal career measures differ between individuals who are involved in mutually arranged confrontations and spontaneous collective violence. Theoretical implications are discussed.


Introduction
Since the 1980s, violent confrontations between supporter groups in and around football stadiums, endangering both those directly involved as well as other spectators, have been a source of concern (Dunning et al., 1986). In broad terms, three explanations have been put forward to explain this phenomenon. Taylor (1971) focused on the crumbling traditional relationship between football clubs and the working-class community from the 1960s onwards, which led to resistance from this particular fan base, who subsequently engaged in confrontations with rival fans and the police. Marsh (1978) focused upon the intragroup dynamics and the rituals and performance of violence within hooligan groups, whereas the Leicester School emphasized the emergence of a workingclass subculture in which behaviour was underpinned by uncivilized moral standards (Dunning et al., 1988).
Explanations for violent confrontations around football that take a class approach are widely rejected nowadays (Dunning, 2000;Taylor, 1987). This also goes for so-called convergence explanations of collective violence in general, which focus on the criminal and deviant character of those involved (Ball and Drury, 2012;Reicher, 2001). Instead, explanations of collective violence have taken a group-dynamic discourse, to which the concept of social identity is central (Reicher, 2001). From this perspective, precipitating incidents (trigger events) may result in the emergence of an individual's self-understanding as a member of a group. Subsequently, an accentuation of perceived similarities (with the in-group) and differences (from the out-group) may serve as a catalyst for violent collective action, with situation-specific norms guiding group and individual behaviour (Postmes and Spears, 1998;Reicher, 2001). The social identity perspective, then, provides meaning to collective violence by pointing to specific circumstances in which a social identity may become salient, making escalation against another group whose behaviour is seen as illegitimate more likely. A role of individual characteristics of those involved in collective violence is explicitly dismissed (Reicher, 2001).
In line with this group-dynamic discourse, the long-term declining trend of footballrelated disorder has been explained by improvements in the policing of crowds (Cleland and Cashmore, 2016). At the same time, various authors have pointed out an apparent increase in mutually arranged violent confrontations between hooligan groups away from the stadium and outside of match days (Cleland and Cashmore, 2016;Jewell et al., 2014). These fights are characterized by mutual consultation between the two parties involved, at a minimum about time and place. Furthermore, arrangements may be made with regard to the wearing of distinctive clothing and informal codes of legitimate action, such as group size and the fighting rules participants are required or expected to adhere to (Adang, 1999;Spaaij, 2007). Also, the fight may be recorded by individuals who are part of the participating groups but who are not involved in the fight itself. 1 Prior studies indicate that hooligan groups consist of individuals who are regularly involved in violence, both in groups and alone (Piquero et al., 2015;Lösel and Bliesener, 2003). Furthermore, individuals belonging to such groups have been observed to seek out opportunities to behave violently around football matches (for example, Adang, 2011). This may lead to the assumption that it is particularly individuals inclined to behave violently who participate in mutually arranged confrontations. Thus far, prior empirical work has an exclusive focus on 'regular' confrontations. Consequently, the question remains whether mutually arranged confrontations merely reflect a difference on a continuum from completely spontaneous to fully planned collective violence, or whether the contributing factors to mutually arranged confrontations differ from the antagonistic relationships around which 'regular' confrontations revolve. Therefore, the current article aims to provide insight into whether and to what extent (individuals participating in) mutually arranged confrontations differ from (those participating in) 'regular' confrontations and how this fits with existing theories on football violence, specifically the social identity perspective.

Theoretical considerations
From the 1960s onwards, scholars aimed to explain collective violence by providing a link between individuals, their actions in crowds and the context in which these actions arise (Reicher, 2001). The social identity perspective on collective violence emerged as a counterpart to then popular convergence explanations stressing the role of individual characteristics. Within the social identity perspective, trigger events and their resulting intergroup dynamics are assumed to facilitate group forming and the emerging of antagonistic relationships between groups. More specifically, it is argued that, in an associated context, an individual thinks, feels and behaves in accordance with the group he/she identifies with (Reicher, 2001).
Prior empirical work supports this line of thought. For instance, individual behaviour in violent crowds usually is aimed at and limited to the party held responsible for a certain course of events, and violent behaviour tends to remain within the boundaries defined as appropriate by the groups with which individuals identify Reicher, 1999, 2000;Reicher, 1984Reicher, , 1987Reicher, , 1996. The specific strengths of social identity theory are its ability to explain the rapidity with which consensus within crowds can arise and why any person may come to act violently in crowds (Reicher, 2001;Terry et al., 1999). However, even in the face of trigger events and intergroup dynamics gravitating towards collective violence, only a minority of the individuals present actually involve themselves in violent behaviour (Adang, 2011). Such variations in the willingness to participate in collective violence, as well as actively seeking opportunities for violence in collective settings, remain unaccounted for by social identity theory.
Prior empirical work shows that differences in offensive action tendencies in crowds can be traced back to the social categories with which individuals identify, the content of these categories, and the persons prototypical thereof (Herrera and Reicher, 1998;Reicher and Hopkins, 1996a, b;Reicher and Sani, 1998;Reicher, 1998, 1999), as well as the extent to which individuals experience anger (Levy et al., 2017;Mackie et al., 2000;Yzerbyt et al., 2003). This indicates that cognitive processes, especially those that deal with the processing of social cues, and their resultant emotions may play a key role in collective violence involvement (also see Van Stekelenburg and Klandermans, 2017). Consequently, a common ground between the social identity perspective and the convergence explanations it dismisses may be found in the way group identification dynamics and personal characteristics interact.
Previously, the so-called 'young male' syndrome has been linked to collective violence involvement (Adang, 2011). This syndrome refers to a high prevalence of risk-taking and criminal behaviours among males between 15 and 35 years of age, particularly in the presence of like-minded peers (Tamás et al., 2019). At the same time, pointing towards demographic characteristics (sex: male, age: young) is too unspecific, particularly because by far the larger part of young males do not engage in this type of risk-taking and antisocial behaviour. Given the relatively high prevalence among hard-core hooligans of, in particular, traits associated with violent behaviour -such as attention deficit hyperactivity disorder (ADHD), emotion-regulation deficits, heightened impulsivity, sensation-seeking behaviour and antisocial tendencies -such traits have been implicated in repeated and continuous participation in collective violence (Lösel and Bliesener, 2003;Piquero et al., 2015;Russell, 2004). In addition, the presence of antisocial tendencies has been found to contribute to positive attitudes towards violent behaviour, such as the belief that violence is acceptable or that violence contributes to self-esteem and social image (Huesmann and Guerra, 1997;Slaby and Guerra, 1988). The inclusion of individual psychological characteristics in explanations of involvement in collective violence therefore may provide a more detailed picture of who participates in collective violence and why.
Spectator violence around football matches manifests itself in the form of competitive violence between socially organized fan groups (Spaaij, 2007). Among such groups, a degree of shared identity exists both within countries and across societies, which revolves around 'their explicit interest in violent confrontation with rival fan groups' (Spaaij, 2008: 373). However, improvements in stadium management and the increased risks of arrest, mean there are limited opportunities to behave violently around football matches and at stadium grounds. Arguably, this has led hooligan groups to seek alternatives in confronting one another (Cleland and Cashmore, 2016;Jewell et al., 2014). Given the specific context in which mutually arranged confrontations occur and the actions required from individuals to organize these fights, a process of self-selection may occur. This suggests that individual characteristics may have value in explaining this type of collective violence. At the same time, prior studies demonstrate that mutually arranged confrontations are more than just an outlet for the violent tendencies of those taking part. In 1997, a pre-arranged and well-organized confrontation between rival hard-core sides (Ajax and Feyenoord) took place in the Netherlands; this confrontation resulted in the death of one Ajax-supporting participant and in several participants on both sides getting seriously wounded. A case study of this confrontation by Kerr and De Kock (2002) indicated that specific events further increased an already existing antagonistic relationship between the hard-core sides involved, which subsequently served as a motive for a mutually arranged confrontation between these groups. Although not mentioned as such by Kerr and De Kock (2002), their account of these events fits social identity theory by referring to trigger events and intergroup dynamics to explain why these groups confronted one another. Groups that participate in mutually arranged confrontations seem to adhere to a priori set rules (Adang, 1999), and the death of an Ajax-supporting participant in the case studied by Kerr and De Kock (2002) was recognized by participants as a clear and unwelcome transgression of these rules. Because violent behaviour during mutually arranged confrontations tends to remain within predefined boundaries, this indicates another link between individuals' behaviour and the group the individual identifies with, which is also in line with social identity theory.
It has been argued that the predefined boundaries around mutually arranged confrontations provide a relatively safe domain for displaying aggression while at the same time maintaining an image of the hooligan culture as exciting (Adang, 1999;Kerr and De Kock, 2002;Spaaij, 2006). This suggests that specific values of the hooligan culture may add significance for those who participate in mutually arranged confrontations. Spaaij (2008) has identified six fundamental features of the hooligan subculture. First, violence in the football context is associated with individual peak experiences, such as excitement and pleasurable emotional arousal. This may equally apply to mutually arranged confrontations (see Kerr and De Kock, 2002). Second, individual and collective reputations may be managed by putting oneself in dangerous situations regardless of potential physical injuries. In this regard, it is of particular relevance that mutually arranged confrontations may be more often concerned with establishing or maintaining a collective reputation than with resolving ongoing conflicts (Newson, 2017). Third, territorial identifications are considered a fundamental feature of the hooligan culture. This refers to defending territory such as the stadium or surrounding grounds and invading rivals' territory. However, since mutually arranged confrontations usually take place outside of football grounds (Giulianotti and Armstrong, 2002), this feature of the hooligan culture appears to be of less relevance. Fourth, the hooligan culture has been described as providing a sense of solidarity and belonging owing to the shared collective experiences. Indeed, peer groups may accommodate an individual's need for prestige, status or sense of security (Crosnoe and McNeely, 2008;Megens and Weerman, 2010). At the same time, this results in interdependence between individual interest and group membership, which may manifest itself in peer pressure to take part in violent confrontations (King, 2001). Fifth, hard masculinity -a social construct that is primarily based on physical prowess, fighting ability and physical health -is considered a fundamental feature of the hooligan culture (Spaaij, 2008). Sixth and finally, representations of sovereignty and autonomy are reflected in attempts to manipulate, disrupt or circumvent security regimes (Spaaij, 2008). From this perspective, mutually arranged confrontations may be seen as an unintended consequence of police measures aimed at reducing football-related violence.
Taken together, the current literature seems to indicate that -in addition to satisfying individual needs for excitement and sensation -mutually arranged confrontations are a way of establishing, maintaining and increasing status within the hooligan subculture. A shared social identity on the one hand appears to provide a platform for violence while on the other hand restricting it. Violence, within certain limits, may be experienced as something good and desirable. First, this suggests that self-selection into mutually arranged confrontations may be stronger compared with that into spontaneous acts of collective violence and that, consequently, the individual characteristics of those participating may diverge depending on the context in which collective violence comes to be. Second, the current literature suggests that, unlike in spontaneous acts of collective violence, although a trigger event and antagonistic relations between groups may provoke mutually arranged confrontations they are no prerequisite. This implies that contributing factors at the individual (psychological characteristics) and contextual level (trigger events, intergroup relationships) around mutually arranged confrontations diverge from those around spontaneous collective violence. Contrary to social identity theory (Reicher, 2001), this implies that offender characteristics may not be completely irrelevant when explaining collective violence.

Current study
Despite regular references to mutually arranged confrontations (Adang, 1999;Giulianotti and Armstrong, 2002;Newson, 2017;Spaaij, 2007), to our knowledge only one empirical study into this matter has been conducted (Kerr and De Kock, 2002). This study, however, has several limitations. First, the data available for analysis were limited to newspaper reports and a few minutes of grainy publicly available camera footage that recorded part of the events from a distance of more than 100 metres. Second, their case study dates back to an event that occurred in 1997, leaving open the option that there have been changes in the hooligan subculture. Third, no attempt was made to systematically compare the confrontation under scrutiny with 'regular' collective violence between hooligan groups. In the current study, the context in which mutually arranged confrontations occur is addressed on the basis of more recent data and compared with a 'regular' not mutually arranged confrontation around a football match.

Data
There was no realistic possibility of conducting participative observation or to interview individuals involved in mutually arranged confrontations. Although potential respondents were approached for interviews, the media attention on this topic at the time of the study resulted in restraint among potential respondents. Consequently, information provided by the police and related parties was the only reasonable alternative. An implication is that data are based on information from secondary sources and not on information from participants themselves. This means that no first-hand information was available about participants' experience of mutually arranged confrontations and their motives for taking part (for example, excitement, peer pressure). As a result, in this study we were dependent on the impressions and experiences of expert respondents, which possibly may have led to bias with regard to participant experiences and motives. Figure 1 displays the various data sources used for the study. In the following paragraphs, these data sources are further explained.

International and national questionnaires
In 2016, an international questionnaire was sent out by the Dutch National Football Information Point (NFIP) of the Dutch National Police to officials from other European NFIPs to assess whether other European countries also face mutually arranged confrontations. Of the 15 officials from 13 countries who responded, seven reported experiences with mutually arranged confrontations. Also in 2016, a questionnaire was sent out to Dutch police officers who had security and public order around professional football within their remit and to police professionals involved in gathering covert information on these topics. A total of 35 national questionnaires -of which 11 were from police officers involved in gathering covert information -were returned. The number of football clubs about which information was obtained was assessed as a measure of representativeness. This showed that (at least one) questionnaire was returned for 31 out of the then 35 football clubs (89 percent) playing professional football in the Netherlands. The national questionnaires provided information on involvement in mutually arranged confrontation of supporter groups linked to 25 of those football clubs. In both the international and national questionnaires, respondents were asked about observed trends, developments and implications for public order policing. Following the international and national surveys, semi-structured interviews were held with six foreign and nine Dutch police officials (of whom six worked covertly). Topics that were discussed in the interviews aligned with the topics addressed in the questionnaires. Interviews therefore mainly served to provide additional information on the answers respondents had already provided in writing.

Case studies
At the time the study was conducted, the Dutch NFIP had gained knowledge of 47 mutually arranged confrontations in the Netherlands concerning the years 2009-13. Of these confrontations, 31 had actually materialized and 16 were either prevented by the police or cancelled by the groups involved themselves. However, to our knowledge, only one of the fights that materialized -between supporters of PSV and Roda JC Kerkrade (2012) -resulted in arrests and a police investigation. Therefore, this case was included in the current study. Information on more recent mutually arranged confrontations was provided in May 2016 by the Dutch police. Information covered 5 out of 35 (14 percent) Dutch clubs playing professional football and led to information on 17 additional mutually arranged confrontations, of which eight were prevented or did not occur. To our knowledge only one of the other nine cases, a mutually arranged confrontation between supporters of SC Heerenveen and FC Groningen (2015), resulted in arrests and a police investigation. Consequently, this case was also included for case analysis.
Several criteria led to these cases being labelled as mutually arranged confrontations. In both cases, police investigations revealed contacts between the leaders of the groups involved prior to the confrontation. Furthermore, information was found in these case files on rules that were set prior to the confrontations, including group size, clothing to distinguish groups from one another and 'rules of engagement'. However, with only two police investigations conducted into a total of 40 confrontations that actually occurred, available recent empirical data are limited. Various suspects in the case files that were studied refused to make a statement and invoked their right to remain silent, further illustrating the challenge of gathering empirical data on this topic. To add to the material from the criminal investigations that were conducted (which contained transcripts of police interviews with suspects and witnesses, wiretap reports of suspects' phone conversations, digital data from seized mobile phones -such as WhatsApp and text messagesand/or camera footage), we also conducted semi-structured interviews with police personnel involved in these criminal investigations (n = 5). 2

Comparison case
During the match between FC Utrecht and FC Twente on 4 December 2011, fireworks were thrown from the visitors' section towards the FC Utrecht family section. A number of FC Utrecht supporters reacted furiously to this event and tried to reach the stand holding the visiting FC Twente supporters. Owing to the stewards in the stadium, they did not succeed. In a second attempt, about 80-100 FC Utrecht supporters then tried to reach the visitors' stand from outside the stadium, resulting in a violent confrontation with police officers. No specific information that supporter groups would seek the confrontation with each other had been present prior to the match and generally the fireworks thrown towards the family section were considered to be the trigger event. For the comparison case, police were confident the individuals arrested represented those responsible and/or involved in the violence. Furthermore, the comparison case was extensively investigated by an external commission that conducted a document analysis in relation to this match, held 26 interviews with the parties involved in football safety around this specific match and reviewed police documentation of arrested suspects (Auditteam Voetbal en Veiligheid, 2012).

Offender analysis
In both the 2012 and 2015 mutually arranged confrontations, about 40-50 individuals were involved. However, not all individuals were arrested for their involvement. Based on the information available in the police files, we, as researchers, established the involvement of 38 people (mutually arranged confrontation sample). Our data are limited to these individuals. Involvement was determined based on their being caught redhanded, reading digital communications and telecommunication history, viewing camera footage, obtaining covert information and recording suspects' communication by secretly placed equipment. In addition, information on 76 individuals involved in the comparison case was collected (comparison sample).
To assess an individual's criminal career history, the police registration system HKS (Dutch: Herkenningsdienstsysteem) was consulted. The HKS contains information on every suspect detained by the Dutch police and the indictable offences involved, and it has been suitable for scientific research since 1996 (Bijleveld, 2007). 3 Given that the minimum age for legal responsibility is 12 in the Netherlands, criminal career information may be incomplete for the nine subjects who were born before 1984. In order to assess psychological characteristics, we turned to available police, Probation Service and forensic psychological data. These data largely rely on information provided by the individuals themselves, by their family or by professionals aware of the individual's personal situation, and they do not entail validated clinical assessments of psychological characteristics. Consequently, the current study uses behavioural indicators as proxies for psychological characteristics. However, because not all behavioural indicators may be recorded as such, this strategy implies a risk of false negatives. Table 1 presents the behavioural indicators focused upon and references indicative of their presence. We explored whether significant differences emerged in relation to criminal career and psychological trait measures between the mutually arranged confrontation and comparison samples by utilizing Chi-square and t-tests.

Contributing factors
Of the covertly working police officials, only two respondents mention the need to settle a score or extreme rivalry between hooligan groups supporting different teams from the Often acting impulsively Needing to learn 'to count to 10' Acting before thinking (especially in stressful situations) Emotion-regulation deficits Having followed training to control aggression Suffering from tantrums or outbursts Being of explosive or angry character Sensation-seeking features Showing an increased need for excitement Looking for exciting situations Getting a kick out of or loving exciting situations same region as a significant contributing factor in arranged confrontations. The majority indicate that they do not consider trigger events or antagonistic relationships between the groups involved to be the main reason for the occurrence of these fights. Subsequent interviews with six respondents working covertly further illustrate this point of view. They report that mutually arranged confrontations also occur between hooligan groups whose clubs have never played against one another and between whom no troubles are known. 4 Respondents mention social media in this respect, which allows hooligan groups to easily contact each other without having to have ever physically met during football matches.
Survey data of the Dutch police professionals responsible for security and public order around professional football and European NFIP representatives further emphasize the role of the group's reputation. One respondent notes that taking on a bigger club especially adds to the reputation of the group: 'one does not want to lose to the little one, while the other gains status by taking on that specific big one.' Respondents further elaborate on this matter by referring to the existence of an informal ranking revolving around the collective reputation of hooligan groups. Collective reputation in this regard is dependent on whether groups are able to mobilize enough people, to show up and fight, adhere to the rules and come out the winner. In addition, some point to the posting of footage of the confrontation in (private) social media communities as a way to enhance a group's status. Their answers furthermore indicate that confrontations may be arranged by leading members of the older hard cores to provide the young cores with an opportunity to gain status, or that these confrontations may be organized by young hard-core groups themselves with the same objective in mind.
The foregoing indicates that mutually arranged confrontations predominantly are not the result of antagonistic relationships or trigger events. Most often, survey data indicate that a need for excitement (n = 6) or to defend the collective reputation of the hooligan group involved or the town from which they originate (n = 4) are perceived to underlie the occurrence of a mutually arranged confrontation. Rather than being spontaneous and prompted by feelings of hostility and anger, the motivations of arranged confrontations appear to be more distant, not only fulfilling participants' current need for excitement but also securing or increasing the long-term subcultural standing of the group as a whole. At times these motivations coincide. This is illustrated by hooligan groups recruiting outsiders to participate in arranged confrontations in order to be able to mobilize a fighting group of respectable size and with sufficient fighting ability to increase the likelihood of winning confrontations, with the purpose of safeguarding or increasing the group's collective reputation. According to the respondents interviewed, these outside recruits usually are individuals who have shown that they are willing and capable to fight, for instance those skilled in martial arts.
Our case study information also illustrates the importance of excitement and collective reputation. Audio recordings made secretly by the police in the aftermath of the arranged confrontation in the 2015 case file indicate that participants derive pleasurable feelings from participating in the arranged confrontation, revealing that suspects regarded the confrontation as 'a top fight' and labelled it as 'cool'. During subsequent police interrogations, one suspect explicitly acknowledged that he likes to fight: 'I think it's a kick. It may not be the right word, but it feels good. I love it.' Furthermore, in both the 2012 and 2015 cases, no current or historical troubles between the supporter groups involved were known of. For the 2015 case, WhatsApp conversations obtained in the criminal investigation indicated that an individual with a leading role in an older generation firm feels that 'the youngsters should earn a reputation'. To this end, he arranged fights with other groups, including the 2015 confrontation. The relevance attributed to collective reputation is also underscored by the reaction of FC Groningen hard-core members to the fact that, contrary to agreements, at least one individual in their group had used a weapon.
A police-recorded conversation shows that some of those involved -including the group leader, who had been making arrangements -felt the use of a weapon might result in damage to the collective reputation of their group and in a decrease in the willingness of other groups to confront them. Therefore, they are determined to identify and punish those who had transgressed the collective norm.
Finally, our comparison case shows that -despite some prior incidents (that is, violence against FC Utrecht players by FC Twente supporters) -there seemed to be no current tensions between the supporter groups involved. Disturbances appeared to be triggered by FC Twente supporters repeatedly throwing fireworks towards FC Utrecht supporters in the family stand. A spontaneously formed group of enraged FC Utrecht supporters then tried to reach the area where the FC Twente supporters were located and from where the fireworks had been thrown. The view that the hostilities were generated by the circumstances is further substantiated by the outcome of the police investigation, which indicated that the violence could not be related to specific, already existing groups of problematic supporters or the presence of identifiable 'leaders' or so-called 'key hooligans' who played an important connecting or leading role in orchestrating the violence.

Rituals surrounding collective violence
In the survey and subsequent interviews, respondents specifically mentioned the involvement of hooligan groups or groups of fanatical supporters in mutually arranged confrontations. Some, in addition, explicitly stated that only a very small proportion of the hooligans known to the police participated in these fights. From the respondents' answers it can be deduced that individuals participating in mutually arranged confrontations often frequent pubs known to be patronized by hard-core supporter groups, have identifying tattoos linking them to the hard core or are designated by the police as hard-core members. Furthermore, several respondents stated that international bonds of friendship between hooligan groups may result in several hooligan groups confronting others together. For instance, friendship ties between supporters of Dutch and Belgian football clubs were mentioned. One of the respondents explicitly stated that they considered mutually arranged fights and the individuals participating in them to belong to a specific hooligan subculture to which many unwritten rules apply.
According to survey and interview data, mutually arranged confrontations mainly take place at remote 'neutral' places (for example, forests, industrial sites and large parking lots). Thus, mutually arranged confrontations mostly cannot be traced back to territorial identifications. Survey and interview data, however, also indicate that -in addition to agreement about time and location -clear terms are set in advance about the number of people (per group) that will participate and the use of weapons. Our data suggest that the use of weapons is generally not accepted and that other 'rules of engagement' are used (for example, when someone is lying on the ground they are to be left alone). The 2012 case file contains footage indicating that, prior to the actual fighting, group sizes were checked, which was followed by shouting 'it's fair, it's fair, we'll take you on, we'll take you on' and by the two groups physically confronting one another. In addition, the police investigation revealed that, when planning the confrontation, both parties agreed that the fight would stop if 'things go too far'. Furthermore, although suspects in the 2015 case investigation invoked their right to remain silent with regard to their involvement, they did make statements about the illicit use of a weapon by one of their own, illustrating their disapproval: 'I think it is outrageous that someone has been stabbed. I am strongly against this'; 'stabbing is a cowardly deed that does not fit'; and 'I heard that someone has been stabbed with a knife, I am quite shocked'. Although these answers may have been socially desirable in an interrogation setting, also from the secret recording in a suspect's car it becomes clear that, by using a weapon, a collective norm had been transgressed: 'Using a knife goes way too far. Together we should talk about that.' Rather than being impulsive and uncontrolled, mutually arranged confrontations are thus characterized by strict norms in the run-up to the confrontation, in the actual confrontation and in its aftermath. The provision of a relatively safe domain in which to display aggression represents sovereignty and autonomy. These aspects of the hooligan subculture are also reflected in the fact that the police, according to respondents, generally are not aware of when and where these confrontations will take place.
Survey and interview data also relate to the behaviour that is expected from individuals who associate with hooligan groups: 'Individuals have to be there, and they can't say that they don't want to fight'. Although survey and interview data with regard to expected behaviour may be subject to bias owing to their second-hand nature, they are substantiated by case file information that shows intergroup solidarity and the mutual agreement of groups and the individuals who are a part of it to fight. For instance, camera footage of the 2012 confrontation showed that, once the fighting had stopped, individuals belonging to both groups gave one another high-fives and briefly embraced before walking back in the direction they had come from. Furthermore, one suspect in the 2015 case stated to the police officers first to arrive on the scene -thus outside of an interrogation setting -that mutually arranged confrontations are something 'that happen between people wanting the same: just taking each other on, and then it's over'. In addition, the police investigation reveals that, after the 2015 confrontation, leaders of both cores expressed their satisfaction. A microphone secretly placed in a suspect's car recorded a participant in the confrontation saying 'I shook hands with one of the leaders and a few more boys shook hands. And then it was just nicely finished'. Survey and interview data additionally suggest that taking part in arranged confrontations may result in individuals being 'rewarded' -for example, by allowing them to visibly identify with the hooligan group from then on by wearing a particular tattoo -and as such contribute to the individual's standing and reputation in the group.
Collective and individual reputation management, as well as excitement, seem to underlie mutually arranged confrontations, but case file information also allows for the hypothesis that individuals may feel pressured to participate. Although suspects in interrogations may try to reduce their responsibility vis-à-vis the police, their statements suggest that resorting to violence may sometimes be perceived as necessary: 'I drove them there, while I knew they wanted to fight. When I was in the car, turning back was not an option.' In addition, anecdotal stories from police officers imply that individuals whose performance during the confrontation is considered insufficient may be punished (for example, being beaten up, being outcast or not being allowed to watch matches from the same stand in the stadium). Given the anecdotal nature of the data gathered in the current study, the data do not permit statements about the frequency of such punishments, or whether punishments may result from other behaviour (such as the leaking of information about confrontations).
Taken together, the survey data, interviews and case studies illustrate that individuals may be intrinsically motivated to participate in mutually arranged confrontations. At the same time, these data give room to the hypothesis that perceived pressure to meet norms of hooligan culture may play a role in an individual's decision to participate. The lack of first-hand information about participants' experiences and motives for participating, however, necessitates caution with regard to this matter. At the same time, the findings in our comparison case are starkly different, because, in their attempt to reach the visiting supporters, provoked FC Utrecht supporters used violence against the authoritiesnotably stewards (in the stadium) and the police (outside the stadium). Suspects in the comparison case also reacted less rationally when confronted with footage of their behaviour and some appeared genuinely shocked. Asked about their motives and thoughts at the time by the police judge, they stated they had 'just been angry with the FC Twente supporters' and they had 'not thought things through'. This strongly contrasts with the premeditated and regulated nature of the violence displayed during mutually arranged confrontations.

Offender characteristics
Mutually arranged confrontations appear to stem from a specific subculture of which violence is a core aspect. Furthermore, to strengthen collective reputation, individuals with a known status of being capable of violence are recruited to participate in mutually arranged confrontations. Therefore, we expect that individuals participating in such fights more often have a violent criminal history (both solo and in groups) compared with individuals involved in 'regular' hooligan confrontations. By extension, we also expect higher levels of behavioural indicators associated with violent offending for the mutually arranged confrontation sample. Table 2 provides a comparative analysis with regard to criminal career measures of the mutually arranged confrontation and comparison samples. From Table 1 it can be seen that the mean age of the mutually arranged confrontation sample is significantly higher (p < .03) than that of the comparison sample. Chi-square analyses indicated that the prevalence of general as well as violent offending was significantly higher (p < .01) in the mutually arranged confrontation sample than in the comparison sample. At the same time, however, prior violent offending in the comparison sample (39 percent) proved not to be an exception. The higher level of prior violent offending in the mutually arranged confrontation sample appeared to be related to their significantly (p < .01) higher level of collective violence offending rather than to individual violent offending. 5 Finally, contrary to our expectations, the prevalence of individual traits associated with violent behaviour did not differ between the two samples.
Results from the conducted analyses fit with the image that emerges from the surveys and interviews, indicating that it is particularly individuals who are used to committing violence -whether alone or in a group -who participate in mutually arranged confrontations. The absence of differences between the mutually arranged confrontation sample and the comparison sample with regard to the presence of psychological traits may be explained by the fact that, in the comparison case, arrested individuals were located in the same stand as hard-core supporters. This may be linked to an interest in hooligan culture to say the least, a line of thinking that is underscored by the comparison case report. This report notes that some of those arrested may be described as the new generation of hooligans. This may also explain why a large proportion of those arrested (39 percent) had been involved in violence prior to taking part in this specific incident.

Discussion
The current article has examined to what extent the meanings attributed to the participants' violence overlap when spontaneous and mutually organized acts of collective violence are compared, who participates in these different types of collective violence, and how this relates to the currently dominant theoretical views on this topic. In line with the social identity perspective, our data show that the groups involved in mutually arranged confrontations have common norms that are actively enforced when these are transgressed by individuals participating in these fights. At the same time, in contrast to our comparison case and a wide body of literature on spontaneous collective violence (see Reicher, 2001), our data indicate that trigger events and antagonistic relationships between the groups involved do not seem to be primary drivers of mutually arranged confrontations. Instead, the meaning of mutually arranged confrontationscompared with spontaneous collective violence -may be traced back to rituals of violence, which are characteristic of the hooligan culture (Spaaij, 2008).
The taking place of mutually arranged confrontations particularly relates to the need to maintain or establish a collective reputation, although the excitement, fun and pleasure derived from participating in these fights may also play an important role. In addition, based on the data gathered, it may be hypothesized that individuals feel pressured to conform to hooligan group norms. Although individuals who do not belong to hooligan groups also get involved in mutually arranged confrontations, survey and interview data indicate that they are purposefully approached by leading hooligan group members because of their skills in martial arts, which may serve to benefit the hooligan group's standing. Furthermore, our data indicate that those involved in mutually arranged confrontations are characterized by a more elaborate history of registered violent offences and are older at the time of involvement compared with the individuals involved in regular hooligan violence. In addition, our data suggest the presence of behavioural problems among a significant portion of both the mutually arranged sample and the comparison sample. Taking into consideration that some of the arrestees in the comparison case were described as 'the new generation' of hard-core hooligans, our data are in line with prior findings that violent hooligan groups are made up of both young adolescent males and generally violent individuals who have outgrown adolescence and suffer from problems on a range of psychological traits associated with violent behaviour (Lösel and Bliesener, 2003;Russell, 2004). At the same time, because psychological traits were not assessed with validated psychological measurements, caution in interpreting this finding is required.
Taken together, our results indicate that the main motivations for mutually arranged confrontations relate to social dominance and excitement-seeking. Our findings therefore largely differ from the perceived injustice and efficacy around which spontaneous acts of collective violence revolve (Reicher, 2001) but fit the recently proposed quadripartite violence typology, which differentiates between excitement-seeking, greed for social dominance or goods, revenge and self-defence as the main motivations for violence (Howard, 2015).
In the case of mutually arranged confrontations, the unwritten norms and rules of the hooligan culture appear to ensure that the violence can take place in a controlled and relatively safe environment that prevents predetermined boundaries being crossed in the heat of the moment. This raises the question of who might be attracted to such opportunities for controlled violence. Our data suggest a self-selection process with regard to the individuals who do and do not participate, particularly concerning the interest in and level of prior violent offending. Given the prevalence of personal characteristics associated with violence -such as heightened impulsivity and emotion-regulation deficits -for a significant proportion of those taking part in mutually arranged confrontations, these traits may be of importance in this regard, facilitating or contributing to the observed self-selection process. The sample sizes, however, may have been too small to detect statistically significant differences between the mutually arranged confrontation sample and the comparison sample. This study, then, suggests that individual characteristics are not to be disregarded -as maintained by the social identity perspective -at least not in the case of mutually arranged confrontations. Our findings thereby support the inclusion of an individual-oriented approach to the study of mutually arranged confrontations -a stance further underscored by a typical lack of the type of antagonistic group dynamics in mutually arranged confrontations that are held to be characteristic of spontaneous collective violence. However, processes of self-selection also appear to exist in cases of spontaneous acts of collective violence, with the majority of individuals present not involving themselves in actual violence (Adang, 2011). The findings of the current study, then, may serve as a starting point for future studies assessing the relative contribution of individual characteristics to participation in collective violence in general.
When considering the current findings, it is important to keep in mind the limitations of our data. First, given the secrecy of mutually arranged confrontations, this study relies almost exclusively on second-hand sources on participants' experiences and motives. The information from our police respondents is limited to the fights that have become known to them. In addition to this, police respondents may be biased in their understanding with regard to participants' experiences and motives for participating in mutually arranged confrontations. Second, the aim of police investigations as laid down in the case files is to inform and convince a judge, which also creates a certain bias. However, given the fact that we had access to the original transcripts of interrogations and subjects' communications (telephone, WhatsApp), we were able to triangulate the findings extensively. Despite their obvious downsides and not having been able to gain direct access to individuals (formerly) involved in mutually arranged confrontations, the police investigations offered a unique perspective on the backgrounds of mutually arranged confrontations. Third, most individuals involved in the selected cases were not arrested. The 2012 and 2015 confrontations resulted in fewer arrests than might have been expected given the reported number of individuals involved (about 40-50 in each case) and we have no way of knowing to what extent those arrested were representative of the section of the hooligan population involved in arranged confrontations. Fourth, with regard to criminal career measures, in general when committing a crime the chances of arrest (and thus being registered in the HKS) are low. Reliance on official data is therefore likely to underestimate an individual's actual criminal behaviour; at the same time, this might give more significance to our finding of enhanced involvement in violence among participants in mutually arranged confrontations. Fifth and finally, information on individual traits was assessed on the basis of Dutch National Police and Probation Service data. Because specific diagnostics were not available in these data, the current findings should primarily be seen as indicators of behavioural traits. Future research may seek to address the personal characteristics of those involved in different types of collective violence using validated clinical instruments. Summarizing, in this study we relied on key informants such as police professionals with many years of experience in football and security. Interviewing and surveying this group of respondents enabled us to place findings from the police investigation files and offender data into a broader context. By combining a criminological perspective, incorporating a reference group and triangulating data sources, we have sought to fill a lacuna in the current collective violence literature.

Conclusion
In this study, qualitative and quantitative data were used to explore the contributing factors of mutually arranged confrontations and the significance of these fights for those involved. For comparison purposes, the data were matched to data on a 'regular' spontaneous football-related confrontation. An important limitation of our data was their restriction to secondary instead of first-hand sources. Consequently, bias may have occurred with regard to participants' experience of mutually arranged confrontations and their motives for participating in these fights.
The results of our study indicate that antagonistic relationships between the hooligan groups involved are not a prerequisite for mutually arranged confrontations to occur. Instead, there needs to be a basis of mutual understanding, respect and trust. In addition, mutually arranged confrontations appear to predominantly revolve around establishing or maintaining a 'tough' collective reputation, with motivations for taking part being linked to excitement-seeking, positive attitudes towards violent behaviour and establishing or maintaining social goods or dominance. This largely contrasts with spontaneous acts of collective violence, which are the result of feelings of anger in response to specific events. Analyses furthermore show that individuals in the mutually arranged confrontation sample are older and have a more extensive history of violent offending than the comparison sample. With regard to individual traits associated with violent offending, no statistically significant differences between the mutually arranged confrontation sample and the comparison sample were found. This fits an interpretation of hooligan culture as a whole being attractive to individuals with personal characteristics that fit the hard masculinity and frequent violence displayed by these groups. Taken together, the results of our study suggest that the various, and seemingly conflicting, explanations that thus far have been offered for football-related disorder and collective violence all have relevance in understanding this phenomenon. More specifically, convergence explanations for collective violence stressing individual characteristics appear applicable to instances of collective violence that occur in an organized manner and in a predesignated setting. Inclusion of an individual-oriented approach alongside the current context-oriented approach, may therefore prove useful in future studies on mutually arranged confrontations as well as on collective violence in general.