Violent and Antisocial Behaviour at Football Events

Review of interventions

Jirka Taylor, Sara-Laure Faraji, Sashka Dimova, Alex Sutherland, Lucy Strang For more information on this publication, visit www.rand.org/t/RR2532

RAND Europe is a not-for-profit research organisation that helps to improve policy and decision making through research and analysis. RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors.

Limited Print and Electronic Distribution Rights This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited. Permission is given to duplicate this document for personal use only, as long as it is unaltered and complete. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial use. For information on reprint and linking permissions, please visit www.rand.org/pubs/permissions.

Support RAND Make a tax-deductible charitable contribution at www.rand.org/giving/contribute www.rand.org www.randeurope.org III

Preface

This report presents the findings of a study to improve policy and decision making in on approaches that have been taken to the public interest through research and prevent and respond to antisocial and violent analysis. This report has been peer reviewed behaviours among populations watching and in accordance with RAND’s quality assurance attending football events, and the extent to standards. which these approaches proved effective. For more information about RAND Europe or The report has been prepared for Qatar this document, please contact: University, to inform that country’s preparations Christian van Stolk towards hosting the 2022 FIFA World Cup, but RAND Europe is intended to be of interest and relevance to Westbrook Centre, practitioners, policy-makers, academics and Milton Road people interested in the field of sport spectator Cambridge CB4 1YG safety in general. United Kingdom RAND Europe is an independent not-for- Tel. +44 1223 353 329 profit policy research organisation that aims [email protected]

Summary

Disorder and violence at football matches are To answer these questions, the research team well-recognised issues that have attracted undertook a rapid review of available evidence. considerable attention in the media as well This took the form of a literature review built as among policymakers and practitioners. on rigorous and systematic methodological In an effort to prevent and respond to the approaches, the parameters of which were phenomenon, numerous strategies and tightly defined to allow for the examination interventions have been implemented by of available evidence within existing time relevant stakeholder groups, including police constraints. and other security professionals, football The review observed a multitude of strategies clubs and associations, fan organisations and that can be and have been implemented local and national governments. However, to counter antisocial behaviour at football despite the wide range of tools available in the matches. They can broadly be grouped into fight against antisocial and violent behaviour four categories: 1) organisation of the venue at football matches, there are gaps in the (e.g. equipping stadiums with cameras, seating- understanding of the current state of practice only arrangements), 2) organisation of the and its effectiveness. events (e.g. arranging transport for away fans; This rapid evidence review responded to this setting up early kick-off times); 3) approaches gap through a focused, structured literature to policing (e.g. dialogue-based policing, search and aimed to provide a critical police liaison teams); and 4) laws, policies assessment of previous research into these and partnerships (e.g. higher penalties for issues. It addressed the following research hooliganism, cooperation with fan associations). questions: With respect to the effectiveness of these 1. What approaches have been taken to interventions, the review found that the existing prevent and respond to antisocial and evidence base (at least as captured by the violent behaviours among populations parameters of the review) is underdeveloped. watching and attending major sporting The review found evidence pertaining only to a events, in particular international football subset of interventions. Of the studies offering matches? an effectiveness assessment, the majority faced notable methodological limitations. Only 2. To what extent have these approaches four studies identified in the review offered proved effective, insofar as those planning evidence based on a comparatively robust major international football tournaments methodological design. can draw on a body of good practice? ii Violent and Antisocial Behaviour at Football Events

Among interventions for which there prohibitions on entry and match day travel is positive evidence, the utilisation of by high-risk supporters. Evidence from security cameras and mandatory transport stakeholder interviews (primarily with police arrangements for visiting fans were found officers) shows they are seen as very effective by methodologically robust studies to be in reducing disorder, although quantitative effective in reducing disorder, although each analyses of existing data have not established intervention was discussed by only one study. a firm link. Similarly, one study on the use of Similarly, early kick-off times were found by mounted police provided tentative evidence two studies to result in reductions in antisocial based on stakeholder testimonies and behaviour. A large number of studies also quantitative analyses, but stressed the results found evidence of effectiveness for a variety need to be interpreted with caution. of policing approaches aimed at establishing Among interventions that do not appear to dialogue and lines of communication with be effective, two methodologically robust fans in an effort to improve intergroup studies failed to find any positive effect for fan relations. However, while pointing uniformly registration schemes as a precondition for a in the same direction, this evidence is largely ticket purchase. Similarly, none of the three based on observations and qualitative studies examining the effect of alcohol bans interviews, typically building on a limited found any positive results. These included number of matches and consultations. bans within stadiums, city-wide bans and Mixed evidence is available for the bans on consumption while in transit to the effectiveness of banning orders, i.e. stadium. iii

Table of contents

Preface III Summary i Table of contents iii List of tables v Abbreviations vi Acknowledgements vii 1. Introduction 1 1.1. The phenomenon of spectator violence in football 1 1.2. Preventing and responding to antisocial and violent behaviour 1 1.3. How the evidence review was conducted 2 1.4. Transferability and how to use the information in this report 2 1.5. Structure of this report 3 2. Key findings on the quality of the identified evidence and the evidence base 5 2.1. The quality and quantity of identified evidence 5 2.2. The evidence base 7 3. Detailed findings: interventions regarding rules of attendance 11 3.1. Evidence summary 11 3.2. Football banning orders 12 3.3. Mandatory spectator registration 14 4. Detailed findings: logistical organisation of matches 17 4.1. Evidence summary 17 4.2. Earlier starting times and daylight kick-off times 18 4.3. Mandatory transport for away fans 19 4.4. City-wide bans on alcohol 19 4.5. Prohibiting alcohol on transport 20 5. Detailed findings: in-stadium equipment, features, services 21 5.1. Evidence summary 21 5.2. Alcohol ban inside stadiums 21 5.3. Security cameras 23 iv Violent and Antisocial Behaviour at Football Events

6. Detailed findings: policing interventions 25 6.1. Evidence summary 25 6.2. Dialogue-based policing 26 6.3. Covert and intelligence-based policing 35 6.4. Use of mounted police 36 7. Discussion 39 References 41 Annex A. Methodology 45 Methods 45 Inclusion criteria 46 Database searches 46 Snowball search 48 Combined count 48 Appendix B. Overview of implemented strategies 49 Annex C. Data extraction template 57 v

List of tables

Table 1. Overview of strategies and interventions to counter antisocial and violent behaviour 6 Table 2. Studies assessing the effectiveness of interventions related to rules of attendance 11 Table 3. Studies assessing the effectiveness of interventions related to logistical organisation of matches 17 Table 4. Studies assessing the effectiveness of in-stadium interventions 21 Table 5. Studies assessing the effectiveness of policing interventions 25 Table 6. Overview of interventions to counter antisocial and violent behaviour 49 Table 7. Data extraction template – Part 1 57 Table 8. Data extraction template – Part 2 58 Table 9. Data extraction template – Part 3 58 vi Violent and Antisocial Behaviour at Football Events

Abbreviations

CIV Dutch Hooliganism Registration Institute (Centraal Informatiepunt Voetbalvandalisme)

ESIM Elaborated Social Identity Model

FBOs Football Banning Orders

FIOs Football Intelligence Officers

MWG Mounted Working Group

PLOs Police Liaison Officers

PLTs Police Liaison Teams

PSUs Police Support Units

SLOs Supporter Liaison Officers

SPT Special Police Tactic

SWP South Wales Police

UKFPU UK Football Policing Unit vii

Acknowledgements

The authors would like to thank Professor Baruch from RAND Europe for their thoughtful Mariam AlMaadeed of Qatar University for feedback on early drafts of the report. her support during the research process. The views presented in this report are the In addition, we would also like to thank the authors’ and remaining errors are also our own. quality-assurance reviewers Elta Smith and Ben

1 Introduction

Hooliganism, spectator violence and antisocial the issue of hooliganism and football- behaviour at football matches are well- related disorder continues to raise questions recognised issues (see, e.g. Spaaij 2005). regarding its causes and exacerbating and Public authorities, clubs and fan associations mitigating factors. These factors may include have implemented strategies and interventions a wide variety of variables, ranging from intended to prevent and respond to instances socioeconomic factors and alcohol use to of antisocial behaviour and violence connected weather to local cultural context and rivalries. to large sporting events, and football matches The focus on this question is not surprising in particular, with the aim of ensuring the – understanding what influences spectator safety and security of spectators, players and violence could help further inform strategies to members of staff. counter the phenomenon. To that end, a rapid evidence review conducted in parallel with the This report sets out findings of an evidence present review set out to take stock of existing review into approaches and strategies employed evidence on factors associated with football- to counter hooliganism and problematic related disorder. The results of this review spectator violence at football matches, with a are available in a parallel report prepared by particular emphasis on international football. It RAND Europe, Violent and antisocial behaviour addresses two research questions: at football events and factors associated with 1. What approaches have been taken to these behaviours (forthcoming.) prevent and respond to antisocial and violent behaviours among populations watching and 1.2. Preventing and responding to attending major sporting events, in particular international football matches? antisocial and violent behaviour 2. To what extent have these approaches The following chapters of this report describe proved effective, insofar as those planning a multitude of strategies and interventions major international football tournaments implemented with the objective to counter can draw on a body of good practice? antisocial and violent behaviour at and around football matches. We capture both 1.1. The phenomenon of spectator preventative (e.g. banning spectators with a record of violence, designing stadiums) violence in football and reactive measures (e.g. arrests made In spite of its prominence as a societal by the police on the day of the match); challenge (see, e.g. Council of the EU 2006), measures which are coercive and prohibitive 2 Violent and Antisocial Behaviour at Football Events

(e.g. removing disruptive spectators or editorials and features, as well as research banning alcohol); and measures which aim focusing exclusively on sports other than to encourage and reward peaceful behaviour football, were excluded. (such as marketing and information provision). The quality of identified evidence was assessed The approaches examined include those which utilising an adjusted version of the Maryland try to control how crowds move geographically Scientific Methods Scale (Madaleno and and spatially through features of stadium Waights 2014). The scale ranks studies from 1 design, and those which aim to effect rational (least robust) to 5 (most robust) and assigns decision making by individuals. two scores to each study: one based on the The actors involved in these interventions robustness of the study design and one based often, but by no means always, include the on how the design was implemented. The police. Police and other security professionals purpose of the scale is to quickly summarise attend almost all football matches (national the adequacy of research for answering and international) to deter and respond questions about effectiveness. These scores to antisocial and violent behaviour. Other are presented in the evidence summary tables stakeholders that are important in helping presented at the beginning of each chapter on ensure an orderly conduct of football matches detailed findings. include football clubs, football associations and An important limitation of this review’s supporters’ clubs; supporters themselves and methodology is that, while systematic, its their self-policing efforts also play a key role. parameters and inclusion/exclusion criteria necessarily constrained its scope. It is possible 1.3. How the evidence review was that further applicable lessons might be conducted identified in literature that did not meet the criteria for inclusion in this review. Annex A provides a detailed account of the methodology employed for this review. The approach taken was a ‘rapid evidence 1.4. Transferability and how to use assessment’ – an approach to identifying and the information in this report critically assessing a body of research that is The findings presented in this report are methodological, rigorous and repeatable but predominantly based on studies authored compromises on some of the breadth of a full in Europe, which raises the question of their systematic review. transferability to other contexts. This is The research team conducted a search of particularly the case for papers discussing several academic (Web of Science, Scopus, strategies to manage crowd behaviour during PsycInfo and Social Sciences Abstracts) and domestic football leagues, which may be grey literature (OpenGrey, IssueLab, OAISTER responding to context-specific challenges and Advanced Google search) databases and and issues. Nevertheless, while the context snowball searches in order to find as many of each football match and league may differ, relevant studies as possible in the time available. the findings in this report provide a relatively The initial inclusion criteria were any articles consistent cross-national overview of what reporting primary research published in English type of interventions and strategies have after 2005 reporting on the implementation been considered and utilised in the context of strategies to counter antisocial and violent of football matches. The evidence on the behaviour at football matches. Commentaries, effectiveness of these strategies tentatively 3

demonstrates which interventions may be 1.5. Structure of this report more likely than not to be successful in achieving their desired outcomes. In addition, The remainder of this report is structured as a valuable contribution of some of the more follows. Chapter 2 provides an overview of the qualitative and ethnographic studies included key findings from the review and sets out the in this review lies in the fact that they offer main messages for policymakers and planners lessons on the practical implementation involved in all aspects of planning a football and functioning of interventions that extend tournament. Chapters 3–6 provide more detail beyond an analysis of their impacts. Lastly, on what we found in relation to four types of information in this report also discusses the interventions, describing the previous studies data and data sources utilised in existing that were included in our review and our critical studies to examine the effectiveness of the assessment of them. Chapter 7 provides a interventions employed. This may represent a summary and draws out broader lessons and useful resource for others wishing to undertake ideas for building the evidence base. similar effectiveness assessments.

Key findings on the quality of the identified 2 evidence and the evidence base

2.1. The quality and quantity of 2.1.2. The studies cover a breadth of identified evidence strategies and interventions to counter antisocial and violent behaviour 2.1.1. Twenty-six studies were selected Table 6 in Annex B presents an overview of all for inclusion in this review interventions mentioned in reviewed studies: their aims, who employs the strategy, where A total of 26 studies were included following and when it is employed and who it targets. the application of the screening and inclusion The table overleaf summarises the key criteria: interventions. • 980 (non-unique) sources identified • 141 retained after applying inclusion 2.1.3. There are some potentially criteria and removing duplicates promising practices but little systematic evidence on the effectiveness of • 63 retained after title and abstract search approaches used to minimise crowd • 26 included after full-text review. violence associated with football matches There were several principal reasons for While a number of interventions to attrition in the sample of identified literature, all tackle antisocial and violent behaviour at of which were linked to the inclusion criteria. football matches have been described For the most part, studies that were excluded in the literature, assessments of their from the review fell into one or more of the effectiveness are comparatively rare. This following categories: 1) studies were not does not necessarily mean that the tactics published in English; 2) studies focused on identified are not effective; it may be rather a sport other than football; and/or 3) studies that there is relatively little robust evidence focused on amateur/youth football. Studies demonstrating effectiveness, at least within were further excluded from the discussion of the parameters of our review. The studies effectiveness if they 4) did not examine effects reviewed cover approaches to preventing, of interventions on antisocial and violent reducing or managing violence at football behaviour and related outcomes; and/or 5) did matches. Few studies, however, rigorously not include any empirical data. tested interventions in a way that allow firm conclusions to be drawn about effectiveness in such a way that we were able to extract clear advice and guidance for those planning 6 Violent and Antisocial Behaviour at Football Events

Table 1. Overview of strategies and interventions to counter antisocial and violent behaviour

Venue organisation Event organisation

CCTV cameras Matchday alcohol restrictions in host town

Segregation of home and away fans Identity card/fan registration schemes

Strategic placement of stages, sound equipment Refusing entrance to inebriated spectators and screens Earlier/daylight kick-off times All-seater stadiums Mandatory transport to stadium Establishment of processing and holding areas for spectators who are arrested or refuse to leave Restriction/banning orders the premises Removal of disruptive spectators In-stadium alcohol restrictions Screening of items brought into the stadium Posting signs to convey safety information to Reducing situational instigators of violence (e.g. spectators offensive T-shirts, banners) Changing venues for ‘high-risk’ events Control of the dispersal process Physical security features (e.g. Plexiglas barriers) Ticket policies, ban on touting

Policing and other staff Laws, policies, partnerships

Intelligence-led policing Coordinating platforms for relevant stakeholders

Community/low-intensity/dialogue-based Zero-tolerance of hooliganism policies from policing/police liaison teams football clubs; information campaigns

Use of riot gear Fan projects (e.g. liaison between clubs and fan associations) Establishment of an effective command post Support to anti-hooliganism opposition from Use of different security ‘levels’ football fans

Increased visibility of security/police; show of Certification and permits to host events force Adopting/advertising legislation and penalties for Relying on reactive tactics violent and/or hate behaviour

Standby officers; relief officers Encouragement of marketing to gender- and age- diverse crowds Deployment of mounted police

Use of hand-held cameras 7

international football events. For example, effectiveness is evidence-supported have most studies did not incorporate any rigorous not been extensively evaluated for policing analysis of the intervention in contrast with football crowds. This is in spite of the fact that a comparison condition (e.g. matches where such approaches are routinely used to police a given intervention was not in place or football matches, in particular international matches played before the introduction of the football matches (Spaaij 2005). intervention in question). Of those that did Further complicating the attribution of any make a comparison between the intervention effect to a given intervention is the fact that and another type of circumstance, none interventions are rarely implemented in were randomised controlled trials (which are isolation from other strategies but rather as regarded as providing higher-quality evidence). part of a broader approach, rendering the The few quasi-experimental designs varied in assessment of individual components very both design and quality; there were only four difficult. More broadly, any observed decreases studies that were assessed as at least 3 on the in disorder and hooliganism may be at least adjusted Maryland Scientific Methods Scale partly the result of wider socio-economic and provide credible evidence of effectiveness factors not connected with any concrete anti- (Di Domizio and Caruso, 2015; Giacomantonio hooliganism strategies. For instance, Veuthey et al. 2015; Priks 2014; Schaap et al. 2015). and Freeburn (2015) noted that factors such Much of the research reviewed was small- as general long-term decline in crime may scale, interview-based or observational in have contributed to the recorded decrease in nature, typically conducted with one group, hooliganism in England, although they added during a single match or over a short time that international data do not necessarily frame. Although such studies can generate support a strong relationship between overall evidence and knowledge that can inform crime levels and football-related disorder. policy development, they typically do not include a comparison condition and do not 2.2. The evidence base allow for meaningful assessments of whether reductions in violence (or antisocial behaviour) 2.2.1. There is some evidence that were due to the tactics discussed or observed. football banning orders can be effective in This is because we cannot reliably assert ‘what reducing antisocial and violent behaviour would have happened otherwise’. Four reviewed studies included an examination In addition, the review also identified a number of the impact of football banning orders of descriptive papers that mention the (FBOs) in the UK and offered some tentative existence or implementation of interventions evidence of their effectiveness (see section without providing an assessment of their 3.2). Banning orders are police-initiated (and effectiveness. Some of these papers claim judge-issued) restrictions on match attendance that the interventions are effective but do and matchday travel imposed on individuals not provide any supporting evidence. These considered high-risk. Stakeholders (both studies are also listed in Table 6 in Annex B. police and fans) tended to view the orders as Frequently, these descriptive papers focused effective in changing fan behaviour; however, on the topics of crowd behaviour and police authors who examined official police data were tactics. What the review also discovered was unable to conclude that observed decreases that some commonly used police practices (such as intelligence-led policing) whose 8 Violent and Antisocial Behaviour at Football Events

in indicators of interest (such as arrests) were to purchase a ticket need to be registered attributable to using banning orders. with the football club and provide some information about themselves. The conditions 2.2.2. Dialogue-based or low-intensity for registration may vary across schemes and tactics are shown to be promising, if not may include differential conditions for different proven types of matches (e.g. more information needs to be provide Fifteen studies examined policing approaches, d for away-match tickets). which predominantly incorporated some Neither study found the programme in question form of dialogue-based or low-intensity effective in achieving the intended results. tactics aimed at improving communication and relations between the police and fans 2.2.4. Starting matches earlier in the day (see section 6.2). Under this approach, police might displace antisocial and violent behaviour officers focus on establishing communication and positive relations with fans, with the Two studies assessed a series of interventions objective to improve intergroup relations related to the logistical organisation of and marginalise antisocial behaviour. The domestic and some international matches studies spanned international tournaments (see section 4.2). Both found that early kick-off and three European national football leagues. times led to positive outcomes at the time of While all of the studies in this group drew on the match but diverged in their assessment of observational data and qualitative testimonies risks outside stadiums. A quantitative study and covered only a small number of matches, from the Netherlands observed a reduction in the evidence they provided pointed uniformly the number of incidents outside stadiums while in the direction of positive results. Dialogue- an ethnographic study from the UK suggested based policing appeared to have the potential early starting times may lead to increased risks to lead to positive outcomes such as improved post-match. The Dutch study also observed intergroup relations, enhanced legitimacy that daylight kick-off times led to increased of the police and greater self-policing by risks inside stadiums. fan crowds, all of which can contribute to greater likelihood of preventing disorder and 2.2.5. Cameras and mandatory transport deescalating conflict situations. One study on arrangements for fans both appear to lead the use of mounted police utilised a relatively to positive outcomes robust design and offered preliminary evidence One quasi-experimental study from Sweden that the deployment of horses could yield observed that the use of cameras led to positive results, although it stressed the need reductions in disorder inside stadiums for further research (see section 6.4). during domestic league matches without any associated spill over effect to adjacent 2.2.3. Fan registration programmes areas (see section 5.3). A Dutch study also have not been shown to achieve their examined the effect of mandatory transport hypothesised results requirements, i.e. a system whereby supporters Two methodologically robust studies assessed of visiting teams were required to use the impact of fan registration schemes in the designated modes of transport to reach the UK and Italy as a prerequisite for a domestic stadium. The study found that this system league ticket purchase (see section 3.3). The had positive results both inside and outside idea behind such a scheme is that fans wishing stadiums (see section 4.3). 9

2.2.6. Prohibiting alcohol has not been although one of these studies found evidence shown to reduce violence and antisocial of increases in the number of incidents behaviour overall outside the match venue. One of the studies also assessed the effects of city-wide alcohol None of the three studies (primarily focused on bans (see section 4.4) and bans on alcohol domestic matches in Brazil, the Netherlands consumption on match day transport (see and the UK) examining the effect of alcohol section 4.5) but did not find either intervention bans inside stadiums found any effect on effective. violence during matches (see section 5.2),

Detailed findings: interventions 3 regarding rules of attendance

3.1. Evidence summary grouped together by individual interventions. For each study, the table presents the country Six reviewed sources commented on the in focus, its scores on the adjusted Maryland effectiveness of two strategies related to Scientific Methods Scale (one for study design conditions under which individuals can or and one for execution) and a brief summary of cannot attend football matches: FBOs and its results. Studies that are based on the same mandatory supporter registration. These data and analysis are listed in the same row. studies are summarised in Table 2 below,

Table 2. Studies assessing the effectiveness of interventions related to rules of attendance

Study Country Study quality Results

Football banning orders

Hopkins and England 1,1 Decrease in arrests but unclear if linked Hamilton-Smith (domestic) to FBOs; some evidence of behaviour (2014) modification

Hopkins (2014) England 1,1 Decline in arrests not linked to FBOs; FBOs (domestic) seen as effective by police

Hamilton-Smith et England, Wales, 1,1 Some evidence of positive effect on al. (2011); Hamilton- Scotland severity of future reoffending; FBOs seen as Smith and Hopkins (domestic) effective by police (2013)

Stead and Rookwood UK (domestic) 1,1 FBOs seen as effective by police and fans (2007)

Mandatory supporter registration

Di Domizio and Italy (domestic) 3,3 Failure to increase the share of casual fans Caruso (2015) in attendance

Schaap et al. (2014) Netherlands 3,3 Not effective in reducing violence (domestic) 12 Violent and Antisocial Behaviour at Football Events

3.2. Football considered a threat by the police at the time banning orders of the order’s expiration (Home Office 2013, referenced in Hopkins and Hamilton-Smith 2014). However, the authors noted that the 3.2.1. What is the intervention and where Home Office claims were not supported by has it been used? any evidence and therefore, in the absence of Most studies in this group focused on the use a rigorous impact evaluation, it was difficult of FBOs in the UK. Under an FBO, a court can to draw any firm conclusions regarding the restrict the ability of individuals to travel to orders’ effectiveness.1 certain areas (stadiums and their vicinity) for Arrest data presented in the study also specific periods of time in relation to football demonstrated that instances of disorder and matches with the objective of reducing the resulting high numbers of arrests of England likelihood of football-related disorder. FBOs can supporters were more common prior to the apply both domestically and internationally. In introduction of the FBOs. Tournaments held the latter example, individuals under an FBO after 2000 with large numbers of FBOs in place can be asked to surrender their passport prior took place largely without major disturbance, to a match that is scheduled to take place although the authors noted that a number of abroad. An application for an FBO can be made arrests had been made at the 2006 World Cup in by a chief police officer on the basis of police Germany. However, the authors urged caution in intelligence if there is ‘reasonable grounds to interpreting the data and noted that the results believe that making a banning order would of several studies suggesting high numbers of help to prevent violence or disorder’ (Hopkins arrests prior to 2001 may have been attributable and Hamilton-Smith 2014, 282). Orders can be to policing tactics in the host countries (see, e.g. applied for a period of three to ten years and Frosdick and Marsh 2005, referenced in Hopkins have been implemented in their current form in and Hamilton-Marsh 2014). England and Wales since 2000 and in Scotland since 2006. The authors also conducted an online survey of football supporters to examine whether 3.2.2. What have studies found? FBOs influenced domestic behaviour of risk supporters. The survey received 199 Hopkins and Hamilton-Smith (2014) reviewed responses, including 63 from individuals the available evidence of FBOs’ effectiveness. who had previously been subject to an order. They drew on Home Office reports and data The majority (59 per cent) of respondents on arrests resulting from disorder involving believed that hooligans had become more England supporters at six international selective in deciding when to turn out in tournaments preceding the introduction of the order to avoid undesired police attention. By FBOs, and from five tournaments held after contrast, nearly a third of respondents (29 their introduction. The authors noted that the per cent) felt that banning orders had made Home Office considered that the FBOs resulted little or no difference to hooligans’ behaviour. in the transformation of the behaviour of risk Still, 12 per cent of respondents opined that supporters, with the vast majority (92 per cent) banning orders had effectively ended hooligan of individuals subjected to an FBO no longer

1 Similarly, banning orders were considered very successful in an earlier Home Office report (2005), which did not offer any supporting evidence beyond statistics on their use. 13

organisations in their clubs. Specifically with officers were generally positive about the respect to individuals previously subjected to impact of the banning orders and claimed they FBOs, their reactions were split in three roughly had observed modifications in the behaviour equal groups. The largest group, 37 per cent, of football hooligans. According to the admitted to having changed their behaviour as interviewees, these modifications manifested a result of having been subjected to an FBO. themselves in two principal ways. First, in These respondents continued their involvement line with the survey findings by Hopkins and in hooliganism but were more selective in Hamilton-Smith (2014), some risk supporters deciding when and where to engage in such were seen as more selective in choosing which activities. The next group, 33 per cent, indicated matches to attend in an attempt to reduce the that banning orders had made no difference possibility of the police constructing a profile in their behaviour. The remainder, 30 per cent, on them. In other words, the possibility that reported having desisted from hooliganism. an FBO would be imposed on them rendered This group either stopped going to football attending football matches regularly too matches altogether or refrained from any risky for this group of supporters. Second, hooligan activities while attending football. The interviewees felt that the most determined results of the survey need to be interpreted supporters under FBOs would attempt to work with caution as they cover a small, non- around the temporal and spatial restrictions representative group of fans. imposed on them so that they could continue their involvement in hooliganism to the extent A study by Hopkins (2014) aimed to shed possible. One suggested way supporters light on the effectiveness of FBOs. This paper could do this was by congregating in a pub examined the number of arrests at domestic just outside the exclusion zone specified by football matches in England and Wales the FBO. A notable limitation of this interview- since the 1992–1993 season to investigate based evidence is the relatively small number whether there was any change following the of officers involved. In addition, some of the introduction of the banning orders. The author interviewees were responsible for submitting found that the number of football-related applications for FBOs, which may have made arrests had been falling steadily in the decade them more positively predisposed towards the before banning orders were introduced and use of the instrument. continued to fall at a similar pace after their introduction. This led the author to conclude Similar observations were made in an earlier that the reduction in football-related arrests report by Hamilton-Smith et al. (2011) and post-2001 appears to be a part of a long-term a follow-up paper by Hamilton-Smith and trend rather than a result of the implementation Hopkins (2013) evaluating the implementation of the banning orders. of FBOs in Scotland. The evaluation drew on administrative data on convictions and case In addition to the analysis of arrest data, files and on focus groups and interviews Hopkins also conducted 27 semi-structured (involving over 50 individuals) with operational interviews with three types of police officers: 1) and strategic stakeholders. These stakeholders Match Commanders (officers responsible for included police officers, procurators, policing management), 2) Football Intelligence football-club security managers and football Officers (FIOs; officers responsible for association representatives. With respect collecting intelligence) and 3) Force Banning to the situation in Scotland, the evaluation Officers (officers responsible for submitting concluded that while banning orders did not order applications to court). The interviewed 14 Violent and Antisocial Behaviour at Football Events

eliminate football-related offending, there 3.2.4. Key messages was tentative evidence that they contributed Banning orders are considered by police to a reduction in the severity of future officers and some other stakeholders as a key reoffending by individuals who had previously intervention that has helped reduce or prevent received an order. The study also consulted violence at UK stadiums. Other sources of with stakeholders in England and Wales to evidence offer a less clear picture on their benchmark the Scottish application of FBOs effectiveness. to that in England and Wales. Interviewees in England and Wales offered testimonies similar to those reported in Hopkins (2014). According 3.3. Mandatory to the stakeholders, in England and Wales spectator registration FBOs were seen as very effective with respect both to the individuals subjected to them 3.3.1. What is the intervention and where and those supporters who wished to avoid has it been used? receiving one. Similarly, FBOs were credited Another type of strategy evaluated in the with playing a key role in reducing hooliganism reviewed literature is the requiring of various in the UK by all three FIOs and all three fans forms of spectator registration in order to gain interviewed in a paper by Stead and Rookwood admittance to a football match. An example (2007), discussed in greater detail here in the of such a scheme is the so-called fidelity chapter on policing interventions (Chapter 6). card, implemented in the Italian Serie A in Echoing the above-mentioned lack of clarity May 2008 (Ferrari 2012). The fidelity card is regarding the effectiveness of FBOs, an essentially a fan identification card, and since earlier paper by Stott and Pearson (2006) the 2010–2011 season, possession of one has also noted that the impact of the instrument been required for holders of season tickets was not clear. According to the authors, one and away supporters to gain admittance to contributing factor was that FBOs were often stadiums. The objective of the fidelity card was deployed in conjunction with other strategies, to keep violent supporters away from matches such as low-intensity policing, which may be and encourage the attendance of casual more instrumental in bringing about desired fans. Its introduction followed the adoption outcomes. In addition, in support of other of a package of counter-violence measures studies (see, e.g. Pearson 2005), the authors adopted in Italy in 2008 that included the pointed out potential problems with the use of following elements: 1) closed-door matches FBOs related to civil rights and the instrument’s where necessary, 2) a ban on cumulative sales proportionality (or lack thereof). of away tickets, 3) increased punishment for football-related hooliganism and 4) sanctions 3.2.3. Main strengths and limitations of for football clubs supporting fan clubs involved the evidence in violent events (Ferrari 2012). The evidence on the effectiveness of Similarly, in the Netherlands, clubs may restrict banning orders comes primarily from the ticket sales to only individuals who have testimonies of police officers and some fan previously registered with the club and are in representatives (all studies scored 1,1). The possession of corresponding documentation strength of qualitative evidence contrasts (e.g. a season card or a club card). The with inconclusiveness of findings based on requirements for obtaining the away-match quantitative analyses. card necessary to buy away-match tickets may 15

be even stricter and may include the furnishing approximately 1,300 more casual fans, reducing of additional information or photographs. The the overall effect on casual attendance to 900 cards can also be scanned during entry to fewer spectators. Building on these results, the stadium, verifying the identity of the ticket the authors concluded that the hypothesised holder again at the time of the match (Schaap substitution effect between committed and et al. 2014). uncommitted fans produced by the fidelity card failed to materialise. While this paper had one 3.3.2. What have studies found? of the most robust methodological designs among those identified, it focused on an Di Domizio and Caruso (2015) assessed the outcome that is only indirectly presumed to be impact of the above-mentioned counter- linked with football-related disorder. hooliganism policies, with a particular focus on the fidelity card initiative. Reductions in various An evaluation of strategies to combat indicators of interest were recorded in Italy in hooliganism in the Netherlands (Schaap et al. the aftermath of the introduction of the counter- 2015) also covered mandatory registration hooliganism policies. Between the 2006–2007 of spectators among the interventions it and 2011–2012 seasons, there was a decrease reviewed. The analysis utilised data provided in the number of matches with injuries (from by the Dutch Hooliganism Registration Institute 101 to 60), injured civilians (from 134 to 83), (Centraal Informatiepunt Voetbalvandalisme injured policemen (from 316 to 37), people [CIV]) on professional football matches in the arrested (from 246 to 75) and people charged Netherlands over five seasons (2006–2007 with violent behaviour (from 848 to 504). to 2010–2011). The data indicated whether and which specific anti-hooliganism measures Specifically with respect to the fidelity card, the were in place for a given match and whether authors analysed its impact on ticket sales on specific types of hooliganism-related incidents casual fans, testing whether its introduction occurred either inside or outside the stadium. might induce the attendance of more A limitation of this measure was its binary supporters unaffiliated formally with either character – it indicated whether a given type club. To that end, they examined effects on of incident occurred or not but did not provide ticket sales for matches with entry restrictions information about its extent or severity. The in place for away fans (those matches were method employed by the authors was a multi- designated as high-risk) and for matches level logistic regression. The results of the where fidelity card owners were eligible for model showed that the registration requirement an exemption from these restrictions. The in the form of a home-match card or an away- analysis drew on match-level data covering match card was not significantly related to five seasons (2007–2008 to 2011–2012) the risk of incidents. This observation was and utilised a series of regression models. true for incidents both inside and outside The results of the models showed that entry the stadium. As such, the authors concluded restrictions applicable to away fans resulted that the mandatory cards did not appear to in a reduction of casual attendance by about be effective in combating hooliganism. As a 2,200 people per match, suggesting that some possible explanation, the authors hypothesised occasional spectators were turned off by that potential hooligans may still have been able matches designated as risky. This effect was to obtain tickets despite the card requirements. compensated in instances where holders of While this study design was among the most fidelity cards were afforded an exemption to robust in the reviewed group of papers, the the entry restrictions. Such matches attracted 16 Violent and Antisocial Behaviour at Football Events

authors noted it was not possible to rule out 3.3.3. Main strengths and limitations of alternative explanations for the observed results. the evidence In addition to the studies discussed above, Both Di Domizio and Caruso’s (2015) and banning orders and supporter cards are also Schaap et al.’s (2015) studies discussed in this discussed, along with other interventions section are among the most robust included adopted in Italy,2 in a paper by Ferrari (2012). in this review (scored 3,3). A limitation of Di The author does not discuss the effectiveness Domizio and Caruso’s (2015) study of the of individual interventions but, similarly to Di Italian context is that it did not directly examine Domizio and Caruso (2015), notes that the changes in violence or disorder but rather size adoption of a package of anti-hooliganism and composition of match spectator crowds. strategies was followed by a notable reduction in several relevant indicators. Between the 3.3.4. Key messages seasons 2005–2006 and 2010–2011, the Two comparatively robust studies failed to number of riots fell by 56 per cent, injuries identify any positive effect of mandatory among supporters by 58 per cent, injuries registration schemes. among police officers by 81 per cent, arrests by 48 per cent and number of police force deployed by 35 per cent.

2 These include 1) stadium safety requirements, requiring each stadium with a capacity of at least 10,000 seats to have an operational security group consisting of key stakeholders such as local police, state police, fire rescue and venue safety management; 2) establishment of a National Centre for the Monitoring of Sports Events with the authority to declare selected matches as high-risk; 3) increased punishments for violent crimes committed inside or in the vicinity of stadiums or linked to sporting events. 17

Detailed findings: logistical 4 organisation of matches

4.1. Evidence summary studies are summarised in Table 3 below, organised by individual interventions so each Two studies assessed the effectiveness of five study is listed more than once. As in the strategies related to the logistical organisation previous section, for each study, the table of football matches: 1) early starting times, 2) presents the country in focus, its scores on the daily kick-off times, 3) mandatory transport adjusted Maryland Scientific Methods Scale arrangements for visiting fans, 4) city-wide (one for study design and one for execution) bans on alcohol and 5) bans on alcohol and a brief summary of its results. consumption on match day transport. These

Table 3. Studies assessing the effectiveness of interventions related to logistical organisation of matches

Study Country Study quality Results

Early starting times

Schaap et al. Netherlands 3,3 Reduction in incidents both inside and outside (2014) (domestic) stadiums

Pearson and England (domestic 1,1 Reduction in risk during matches, potential Sale (2014) and international) increase in post-match risk

Daylight kick-off times

Schaap et al. Netherlands 3,3 Increase in incidents inside stadiums, no (2014) (domestic) change outside

Mandatory transport arrangements

Schaap et al. Netherlands 3,3 Reduction in incidents both inside and outside (2014) (domestic) stadiums

City-wide bans on alcohol

Pearson and England (domestic 1,1 Not effective in preventing consumption Sale (2014) and international)

Ban on alcohol consumption on matchday transport

Pearson and England (domestic 1,1 Not very effective in preventing consumption Sale (2014) and international) 18 Violent and Antisocial Behaviour at Football Events

4.2. Earlier starting and football spectators. The paper drew on a times and daylight series of ethnographic studies covering fans kick-off times of three English teams (Blackpool, Manchester United and the England national team) during the period 1995–2009. Over the course of the 4.2.1. What is the intervention and where data collection, the researchers observed in has it been used? excess of 200 matches (including matches Two studies (one from the Netherlands and held outside of England) with more than 2,000 one from the UK) examined two strategies hours of observations (including travel time to related to the scheduling of football matches foreign matches). The paper also incorporated and their possible impact on antisocial and data from direct observations of police violent incidents. These strategies were: 1) operations during English Premier League scheduling matches earlier in the day, thereby match days in the 2008–2009 season and limiting opportunities for the consumption from an unspecified number of interviews with of alcohol or drugs in the run-up to a match; British and Italian senior police officers. 2) scheduling matches so that they take Assessing the impact of early kick-off times, place during daylight hours, thereby avoiding the authors concluded that this strategy was darkness around the stadium. effective in reducing the amount of alcohol drunk by supporters before matches, in 4.2.2. What have studies found? particular at weekends. By extension, assuming The first study was conducted by Schaap et al. a link between alcohol consumption and risk (2015). As explained in Chapter 3, the analysis of violence, early kick-off times were seen took the form of a multilevel logistic regression as effective in decreasing the potential for model utilising data provided by the CIV. violent incidents during the match. At the Regarding the time of the match, the results of same time, according to the authors, matches the model confirmed the authors’ hypothesis: with early kick-off times could be followed by matches played later in the day were found prolonged periods of drinking and increased to be more likely to experience incidents both risk of violence after the match, particularly if inside and outside the stadium. However, supporters of the away team do not leave the contrary to expectations, playing at night-time, match area immediately (as is the case with i.e. during hours of darkness, was found to be derbies between teams from the same city). related to an increased likelihood of incidents This was confirmed by authors’ observations of occurring inside the stadium but not outside. a number of Manchester United vs. Manchester The authors also found that the impact of the City fixtures with early kick-off times, which starting time of match was not affected by were not found to have reduced the overall whether or not the match is played in darkness amount of alcohol consumed and instead led to (i.e. a 7 p.m. starting time in the Netherlands in increases in alcohol consumption among some January would be mean the match would be supporters. Therefore, the authors concluded played in darkness, whereas the same starting that while early kick-off times may be effective time in May would mean the match would be in reducing football-related disorder at the time played in daylight). of the match, they may potentially lead to an increased likelihood of issues later in the day. Earlier kick-off times were also among the Nevertheless, based on interview testimonies, strategies examined in an ethnographic paper the authors suggested that early kick-off times by Pearson and Sale (2011) focused on alcohol 19

for high-risk matches may still be preferable of incidents both outside and, to a lesser as they help avoid holding matches during the degree, inside the stadium. However, the hours of darkness, which may have a strong authors noted that the mandatory transport effect on the probability of incidents occurring. strategy was seen as intrusive and costly, was unpopular with supporters and was likely to be 4.2.3. Main strengths and limitations of discontinued (Ferwerda et al. 2011; Van der Aa the evidence 2011, referenced in Schaap et al. 2015). One study in this section employed a comparatively robust design (scored 3,3). 4.3.3. Main strengths and limitations of the evidence The other study is based only on a series of observations, albeit covering a relatively large The evidence for this type of intervention is number of matches (scored 1,1). based on only one study from only one country, albeit with a relatively robust research design 4.2.4. Key messages (scored 3,3). Both studies that examined the effect of early 4.3.4. Key messages starting times found positive results, although one study suggested the possibility of increased Mandatory transport arrangements were found risk post-match. Moving kick-off times to daylight to be effective in reducing disorder both inside hours was found to have a counterintuitively and outside stadiums. negative effect on disorder inside stadiums. 4.4. City-wide 4.3. Mandatory transport bans on alcohol for away fans 4.4.1. What is the intervention and where 4.3.1. What is the intervention and where has it been used? has it been used? Pearson and Sale (2011) also examined the The Dutch study by Schaap et al. (2015) also impact on alcohol consumption by football examined the effectiveness of mandatory supporters of city-wide bans on alcohol transport arrangements for away fans sales. The authors observed the application deployed at some domestic matches in the of city-wide bans during six trips to different Dutch league. Under these arrangements, Italian venues that hosted matches involving fans of the visiting team had to utilise a an English team. The bans typically took the predetermined trajectory to arrive at the form of a prohibition on the sale of alcohol by stadium on the day of the match and were restaurants, bars and supermarkets and on not allowed to deviate from the route. The public drinking during the 24 hours surrounding idea behind this intervention was to minimise a given match. contact with home-team supporters. 4.4.2. What have studies found? 4.3.2. What have studies found? The authors concluded that the introduction According to the authors, mandatory transport of city-wide bans was not effective in arrangements for away-team fans were found preventing alcohol consumption. According to be effective in reducing the likelihood to the study’s findings, there were three 20 Violent and Antisocial Behaviour at Football Events

principal ways through which fans were able using modes of transit that had been officially to procure alcohol in contravention of the designated to carry football supporters ban. First, some local upscale restaurants (typically buses or trains). would continue to serve alcohol, sometimes at the request of locals who did not wish to 4.5.2. What have studies found? be deprived of a drink to accompany their Based on the authors’ observations, this meals. Second, local youth would purchase strategy was not effective in reducing the risk alcohol from supermarkets a day in advance of alcohol-related violence for two principal and resell to English fans on the day of the reasons. First, some supporters ignored match. Third, fans continued to be served by the prohibition and found a way to continue some bars located in less central streets and consuming alcohol while in transit; second, thus less visible to the police. As a conclusion, some supporters were already drunk before the authors suggested that bans on alcohol they began their formal transit to the stadium. consumption may even have had undesired consequences by engendering an ‘us vs. 4.5.3. Main strengths and limitations of them’ mentality vis-à-vis the police among the evidence football supporters and by encouraging fans to congregate away from the police, who The evidence underpinning this intervention were thus in a worse position to monitor the comes from an ethnographic study based only situation (e.g. in the instance of a confrontation on match day observations (scored 1,1). with home-team supporters). 4.5.4. Key messages 4.4.3. Main strengths and limitations of Transport-related alcohol bans were not found the evidence to be effective in preventing consumption of The evidence underpinning this intervention alcohol on match days. comes from an ethnographic study employing a very limited number of observations (scored 1,1).

4.4.4. Key messages City-wide bans on alcohol were not found to be effective in preventing consumption of alcohol on match days.

4.5. Prohibiting alcohol on transport

4.5.1. What is the intervention and where has it been used? Pearson and Sale (2011) also assessed situations in the UK in which the prohibition of alcohol consumption was applied to supporters’ transport to and from the stadium 21

Detailed findings: in-stadium 5 equipment, features, services

5.1. Evidence summary These studies are summarised in Table 4 below, organised by individual interventions. As in the Four studies assessed the effectiveness of previous sections, for each study, the table two strategies deployed inside stadiums presents the country in focus, its scores on the related to security features and (limitations on) adjusted Maryland Scientific Methods Scale the provision of alcohol: alcohol bans inside (one for study design and one for execution) and stadiums and the utilisation of security cameras. a brief summary of its results.

Table 4. Studies assessing the effectiveness of in-stadium interventions

Study Country Study quality Results

Alcohol ban inside stadiums

Pearson and Sale England (domestic 1,1 No changes in alcohol-related violence (2014) and international)

Nepomuceno et al. Brazil (domestic) 2,1 Little effect on violent behaviour (2017)

Schaap et al. Netherlands 3,3 No change in incidents inside stadiums, (2014) (domestic) increase outside

Security cameras

Priks (2014) Sweden 4,2 Reduction in incidents inside stadiums, no (domestic) change outside

5.2. Alcohol ban examined the impact of bans on alcohol inside inside stadiums the stadium.

5.2.2. What have studies found? 5.2.1. What is the intervention and where has it been used? Looking at national and international matches involving English teams, the ethnographic study Three studies (covering primarily domestic by Pearson and Sale (2011) did not observe any matches in the UK, Brazil and the Netherlands) changes in alcohol-related violence that could 22 Violent and Antisocial Behaviour at Football Events

be connected to the use of bans on alcohol of the ban, utilising a non-parametric model. inside the stadium. The authors did observe, The results of the analysis showed that the however, several unintended consequences of alcohol ban had little effect on hooligans and such bans. Most fundamentally, supporters their violent behaviour. In the period preceding rushed to consume as much alcohol as the introduction of the ban, there were 139 possible before entering the stadium, which matches with at least one incident occurring worked against the ban’s objective to reduce (608 incidents in total). After the introduction the level of fans’ drunkenness at the time of of the ban, there were 170 matches with at the match. Efforts to maximise consumption least one incident occurring (755 incidents before the start of the matches also led to last- in total). An explicit limitation of the study minute crushes at turnstiles immediately before acknowledged by the authors is the fact that kick-off times, particularly at away matches. the underlying dataset captured only officially Where some consumption was allowed in recorded hooliganism incidents, which left out a restricted areas (e.g. in UK stadiums not within potentially large number of instances of violent sight of the pitch),3 these locations tended to behaviour where perpetrators were not caught become crowded, with potential for jostling, by the police. In addition, while the paper does beer spillage and subsequent tensions. The not clarify this, it appears that the model used lack of alcohol provision inside stadiums also did not include a dummy variable to capture the increased the possibility that fans would end up existence or absence of the new policy. drinking alongside supporters of the opposing The evaluation of Dutch anti-hooliganism team in pubs and near off-licences outside the strategies by Schaap et al. (2015) also stadium in an unsegregated and uncontrolled examined the effects of a ban on alcohol fashion. Lastly, a small group of supporters sales inside the stadium. The results indicated ignored the ban and managed to bring alcohol that such bans were not associated with the inside the stadium. likelihood of incidents occurring inside the A quantitative study by Nepomuceno et al. stadium (after accounting for heterogeneity (2017) assessed the impact of a ban on alcohol between matches). Alcohol bans appeared to sales inside stadiums that was in place in the increase the likelihood of incidents occurring Brazilian state of Pernambuco between 2009 outside the stadium, thereby having the and 2015. The authors analysed a dataset opposite of their intended effect. However, covering matches of the states’ three biggest the authors urged caution in interpreting the football clubs, which included criminal justice model’s results, since the underlying dataset information on any hooliganism incidents that included a large number of missing values with may have occurred during the matches (the respect to whether alcohol was sold during the total number of matches in the sample was 375, match – this information was not available for incidents occurring at 309 of these). The dataset 37 per cent of matches included in the dataset. covered the period 2005–2015 and therefore included four years not covered by the alcohol 5.2.3. Main strengths and limitations of ban. The authors ran a pre/post comparison the evidence of the number of hooliganism incidents that The evidence underpinning this intervention took place before and after the introduction comes from three studies with different

3 In UEFA competitions, sale of alcohol inside stadiums is completely prohibited, with the exception of executive sections. 23

research designs (one of which was relatively held at three stadiums which had installed robust – the studies were scored 3,3, 2,1 and cameras before they became mandatory 1,1 respectively) and from three different served as a control group. countries, spanning multiple contexts. Results from the study’s ordinary least squares regression models indicated that matches 5.2.4. Key messages played under surveillance cameras saw a 64 All three studies conclude that bans on per cent reduction in incidents from an initial alcohol inside stadiums had little to no effect average of 0.26 incidents per match. The on disruptive and violent behaviour inside significance of these results was confirmed in stadiums. Two studies suggested such bans a series of robustness checks, including the may lead to negative outcomes outside addition of month-fixed effects and stadium- stadiums. specific linear trends. They also remained valid after controlling for other possible explanations 5.3. Security cameras such as size of attendance or number of officers present at the match (derived from data from the Swedish National Police Force). Therefore, 5.3.1. What is the intervention and where the study concluded that the observed reduction has it been used? in disorderly behaviour could be linked to the use A study by Priks (2014) evaluated the effect of surveillance cameras and was not attributable of surveillance cameras installed in Swedish to any endogenous factors or parallel policing football stadiums on unruly behaviour. The interventions. Furthermore, the author found no author took advantage of a decision in 2000 significant differences between matches with by the Swedish Football Association that all and without cameras in the number of incidents stadiums hosting matches in Allsvenskan occurring outside of stadiums, suggesting there (the highest Swedish league) needed to be was no displacement effect of unruly behaviour. equipped with cameras within two years. The The study demonstrated that findings remain actual date of the camera installation varied valid under various model specifications. across individual teams due to differences in While the study did not discuss explicitly its the time it took to obtain necessary permits assumptions behind the chosen independent and procure the equipment. variables, this does not appear to be a constraint as the timing of the intervention (installation 5.3.2. What have studies found? of security cameras) was exogenous to prior The study examined referee reports which disruptive behaviour at football matches. recorded the number of incidents during matches held in the period 1999–2005 to 5.3.3. Main strengths and limitations of examine any changes in the volume of unruly the evidence behaviour in the aftermath of the introduction The evidence underpinning this intervention is of the surveillance cameras. The data covered based on only one study, albeit with a robust 1,273 matches, of which 211 were played in methodological design (scored 4,2). stadiums without cameras. To check for any possible displacement effect of hooliganism 5.3.4. Key messages from inside the stadium to its vicinity, the The introduction of cameras was found to lead study also examined Swedish police data on to a reduction in incidents inside stadiums, incidents outside football stadiums. Matches without any spill over effects to adjacent areas.

6 Detailed findings: policing interventions

6.1. Evidence summary are summarised in Table 5 below, organised by individual interventions. As in the previous The review identified 17 studies that sections, for each study, the table presents the examined the effectiveness of various policing country in focus, its scores on the adjusted interventions. Of these, 15 studies examined Maryland Scientific Methods Scale (one for a dialogue-based approach to policing study design and one for execution) and a brief (although the used terminology differed across summary of its results. Studies that are based studies), one study focused on covert and on the same data and analysis are listed in the intelligence policing and one study assessed same row. the utilisation of mounted police. These studies

Table 5. Studies assessing the effectiveness of policing interventions

Study Country Study quality Results

Dialogue-based policing – international football competitions

Hylander et al. (2010); Germany (World 1,1 No major incidents; tactics seen as Rosander and Guva (2012) Cup 2006) successful

Schreiber and Adang Germany (World 2,1 Tactics seen as successful in (2010a, b) Cup 2006) comparison with high-profile policing

Schreiber and Stott (2012) Portugal (Euro 1,1 Tactics seen as successful in 2004) improving police–fan relations

Stott et al. (2007) Portugal (Euro 1,1 Tactics seen as successful in 2004) comparison with high-profile policing

Stott et al. (2008a) Portugal (Euro 2,1 Lower visible police presence 2004) and number of arrests, improved intergroup relations and identification with the police 26 Violent and Antisocial Behaviour at Football Events

Study Country Study quality Results

Dialogue-based policing – domestic football leagues (UK)

College of Policing (2015) UK (domestic and 1,1 Tactics seen as successful in some international) improving police–fan relations

Stott et al. (2008b) UK (domestic) 1,1 Tactics seen as successful in improving police–fan relations

Stott et al. (2012); Stott UK (domestic) 1,1 Tactics seen as successful in (2014) improving police–fan relations

Stott et al. (2016a) UK (domestic) 1,1 Tactics seen as successful in improving police–fan relations

Dialogue-based policing – domestic football leagues (other European countries)

Beedholm Laursen (2017) Denmark 1,1 Tactics seen as successful in (domestic) improving police–fan relations

Havelund et al. (2016) Denmark 1,1 Tactics seen as successful in (domestic) improving police–fan relations

Stott et al. (2016b) Sweden (domestic) 1,1 Tactics seen as successful in improving police–fan relations

Covert and intelligence-based policing

Stead and Rookwood UK (domestic) 1,1 Intelligence-based policing seen (2007) as more effective than more overt tactics

Deployment of mounted police

Giacomantonio et al. UK (domestic and 3,2 Mixed results; horses seen as (2015) some international) effective by police officers

6.2. Dialogue-based describe the policing approach differed (e.g. policing low-key policing, low-intensity policing, event policing) the basic tenets and objectives of the assessed approaches were very similar. Five of 6.2.1. What is the intervention and where these papers explicitly focused on international has it been used? football tournaments while the rest focused on Most of the studies reviewed (15 of 17) domestic leagues (UK, Denmark and Sweden). examined some form of dialogue-based approach to policing football events, aimed 6.2.2. What have studies found? at fostering communication and intergroup International tournaments relations between the police, other authorities A paper by Hylander and Granström (2010) and fans. While the terminology used to conducted a case study of policing and 27

organising strategies employed during a (such as food and drink vendors, public match between Germany and Poland at the restrooms) and physical infrastructure (e.g. 2006 World Cup held in Dortmund, Germany. carpet on the ground signposting directions to In the run-up to the match, the police utilised most important venues, spatial organisation several strategies to prevent and reduce of places where supporters congregate). As violence (described in detail by Hau 2008) the study concluded, ‘the covert police strategy with the overall objective to be friendly and was to behave in a friendly way and to treat communicative, avoid any action that might the fans as festival participants rather than have been seen as provocative and respond rival supporter groups.… The most successful professionally to any conflict situations. To disarming, however, seemed to be all the illustrate, supporters (predominantly Polish [event] arrangements’ (p.20). A subsequent fans) were welcomed at the train station by a paper by Rosander and Guva (2012), analysing human avenue of police officers to help fans the behaviour of police at the same match navigate their way from the station to other based on the same observational data, destinations, such as the locations of large reiterated the conclusion that the ‘friendly spectator screens and the stadium. The police but firm’ approach (p.67) was successful in were unarmed, dressed in uniforms but with no keeping the match day in the city peaceful. helmets or visors, and were available to answer In another paper examining policing tactics at questions. Fans of the opposing teams were the 2006 World Cup (and comparing it to earlier not separated during the day. tournaments), Schreiber and Adang (2010a) The authors collected data via two methods. conducted structured observations of ten The first was a participant observation matches and associated activities held in three undertaken by three pairs of senior host cities in North Rhine-Westphalia. On three researchers. The observations took the form occasions, the researchers also conducted of semi-structured, multipoint observations observations either the day before or the day and lasted ten hours. In the second, the after the match. During the observations, research team conducted 38 interviews with focused on crowded areas, the researchers supporters of both nationalities, police officers, took samples every 15 minutes on the number match organisers and other service agents of people and the interactions between fans as well as a follow-up interview with a police as well as between fans and the police. commissioner. The field interviews were Any incidents and associated interventions conducted in the streets and at the locations were also recorded. The team of observers of the large screens before and after the match consisted of 23 graduate course participants and during half-time. (comprising postgraduate students and police officers) trained in advance by the Based on the collected data, the authors authors of the paper. In total, the observers concluded that overall the match day was recorded 1,020 samples. The methodology peaceful and without major issues. The for structured observations was in line with approach of the police was seen by interviewed that used by a paper on policing tactics during fans as welcoming and not at all provocative. the 2004 UEFA European Championship In addition to policing tactics, the authors (Euro) (Stott et al. 2008a, discussed below), noted that other organisational features likely enabling a direct comparison between the played a substantial role in keeping the event two tournaments. In addition, the research largely incident-free. These included good team conducted a series of interviews with the accessibility of facilities for visiting spectators 28 Violent and Antisocial Behaviour at Football Events

following groups: 184 interviews with fans; nine findings of the Schreiber and Adang (2010b) with employees of security companies; 40 with paper were similar to observations made in police officers, of which 31 took place during their wider study (2010a), i.e. that low-intensity the tournament and nine in its aftermath; and and welcoming policing was conducive to two with FIFA volunteers. The data was further positive intergroup relations. However, while complemented by an analysis of media data acknowledging that the match day was from newspapers, TV and online sources. free of large-scale hooligan confrontations, the authors noted an increased number of The comparison of the observational data incidents and arrests. In contrast with Hylander from the 2004 Euro and the 2006 World Cup and Granström (2010), the authors observed offered mixed results. The number of incidents a worsening in the situation towards the end recorded in observational data from Germany of the day and suggested that the police may was higher than that recorded in Portugal have abandoned somewhat their original low- (incidents recorded in four per cent and one intensity concept. This may have contributed per cent of the respective samples.). At the to what the authors termed a ‘self-fulfilling same time, the frequency of recorded positive prophecy around violent behaviour and fan police–fan and fan–fan interactions was higher disorder’ (p.486). in Germany than in Portugal. The authors also noted that the concept of low-intensity policing This review also identified a series of papers had not been applied uniformly during the World produced as part of a larger research project Cup. The police appeared to implement some of examining the use of low-intensity policing in its features but did not implement a full ‘friendly the context of the 2004 Euro held in Portugal. A but firm’ approach. For instance, in contrast paper by Schreiber and Stott (2012) examined with the tactics in Portugal, the majority of the utilisation of low- intensity policing policing was done by units in riot gear, making techniques during the tournament and their it more difficult to provide a differentiated effect on the security situation surrounding response to any arising conflict situations. group-stage matches involving Germany. The Where implemented, the authors observed study’s focus on German fans reflects the fact positive outcomes resulting from the low- that, in the run-up to the tournament, German intensity approach. These included improved fans were identified as a high-risk group. facilitation of fan and legitimate intergroup Under the low-intensity policing approach, the relations as well as conflict de-escalation and Portuguese police decided to offer multiple greater marginalisation of disorderly behaviour. levels of response. In standard situations, the These findings led the authors to conclude police would deploy in small teams or pairs that where the German police was able to offer with the objective of conveying a sentiment of a targeted and context-dependent form of security. Their portfolio would include proactive policing, positive outcomes ensued. By contrast, communication, provision of advice to fans issues occurred where this was not the case. to help their enjoyment of the tournament, and monitoring and addressing small-scale Drawing on the same data as used in incidents. Situations that escalated would Schreiber and Adang (2010a), Schreiber and be resolved by specialised intervention units, Adang (2010b) examined in greater depth which would not be visible until necessary. the policing tactics surrounding the match between Germany and Poland at the 2006 The study collected data in two principal World Cup, the same match as that analysed ways. First, the authors conducted semi- by Hylander and Granström (2010). The structured observations over the course of 29

the tournament covering events involving of Albufeira, while the rest of the tournament German fans. Second, the research team also and all other match locations saw no collective conducted ad hoc interviews with 173 German violence involving England fans. All major cities and 12 Dutch fans (the Netherlands was one in Portugal (and therefore all match venues) of Germany’s opponents at the group stage). were policed by the national police (Policia In addition, the team undertook unstructured de Segurança Pública [PSP]), which adopted interviews with other stakeholders, including the low-intensity approach described in the members of the German, Dutch and Czech fan previously discussed paper. By contrast, rural embassies4 (the Czech team was also in the areas were served by the Portuguese National same group) and members of the NGO FARE Guard (Guarda Nacional Republicana [GNR]), (Football Against Racism in Europe). Lastly, the which adopted a more reactive, high-profile research team also conducted a series of semi- approach. The authors set out to examine the structured consultations with several police relationship between public policing strategies, commanders. Based on the collected data, social identity and social disorder. the authors concluded that the low-intensity As with the previous paper, this study utilised approach to policing helped foster non- two principal data collection methods. First, violence and good relations between fans from the research team conducted semi-structured different nations. Interviewees largely credited field observations at match venues on all the low-intensity policing tactics with having matchdays involving the England team and contributed to a calm and positive atmosphere, during the second night of riots in the town particularly in comparison with their of Albufeira. Second, the research team experiences from other contexts. In addition, consulted with England fans. The researchers according to the authors’ observations, the conducted interviews with ‘approximately 74’ low-intensity approach was successful in (p.80) England fans during the tournament. deescalating situations of emerging conflict. In addition, the study team ran a web-based As such, the authors argued that the results of survey of England fans before and after the study lent support to the Elaborated Social the tournament (recruited via articles and Identity Model (ESIM), developed in previous advertisements, fan organisations’ distribution literature (Drury and Reicher 2000; Reicher lists and flyers distributed during the 1996, 2001; Stott and Reicher 1998), as an tournament). The online questionnaire elicited account of processes determining crowds’ 102 responses, 39 of them received before collective action. the tournament and 63 after the tournament. As part of the same research project, a paper Lastly, the research team reached out for by Stott et al. (2007) also examined policing testimonies to 20 fans engaged in previous employed by the Portuguese police during research and heard back from 14 individuals. the 2004 Euro. In contrast with the previous Based on the data collected, the research team paper, this study focused on the fans of the noted differences in the policing strategies England team and their behaviour during the employed in Albufeira and elsewhere. In tournament. In this regard, important focal particular, incidents observed in Lisbon were points of the study were two riots involving deescalated while the situation in Albufeira England fans in the southern Portuguese town

4 Fan embassies are stationary services offered in host cities for visiting team fans. They may provide guide materials, goodwill events and other forms of support and advice. 30 Violent and Antisocial Behaviour at Football Events

failed to be contained. Based on fans’ number of fans and police present in their testimonies, the low-intensity approach to areas, their interactions and the occurrence policing was seen as enhancing the legitimacy of any incidents and corresponding police of the police while the opposite was true for interventions. In total, the observers took high-profile policing. The authors concluded 1,896 samples, which were divided roughly that low-intensity policing contributed to a equally between seven increased-risk matches cultural change on the part of England fans. and seven standard-risk matches (899 and As part of this change, hooliganism seemed 977 samples, respectively). In addition to the to have become more marginalised and observations, the study drew on responses to isolated. These findings, however, need to be an online questionnaire distributed to fans of interpreted with caution given the limitations the England team, the same as the one used to of the data. As the authors noted, greater inform the paper by Stott et al. (2007). certainty could have been achieved if they had Quantitative observation data revealed had more quantifiable data on any changes that visible police presence was observed in the perception of intergroup relations and in approximately half (56 per cent) of the if they had been able to follow the same fans samples. This was significantly less than throughout the tournament (thereby allowing a during the 2000 Euro (p< .001). Positive within-subject study design). contacts with the police were observed in Stott et al. (2008a) also examine low-intensity five per cent of all observational samples, policing tactics at the 2004 Euro in Portugal a significantly lower share than during the and build on the previous paper by Stott 2000 Euro (p< .001). The observational data et al. (2007). This study was a quantitative was also consistent with arrest records in attempt to identify reasons for the successful that it indicated a significantly lower number maintenance of public order in host cities of disorder incidents during the 2004 Euro during the tournament. The data used for in comparison with the pervious tournament the analysis also drew on a combination of (p< .001). According to the authors, this observations and consultations with fans. confirms the low profile of the deployment Unlike in the study by Stott et al. (2007), of the Portuguese police and the ensuing observations informing this paper were low levels of disruption and disorder. Data structured and covered 14 matches at the on fans’ perceptions further corroborated tournament covering all matches played by positive impacts of this form of policing. the following teams: Portugal, the Netherlands, Post-tournament measures of intergroup Germany and England. The method of these relations were significantly more positive observations was identical to that employed than the recorded pre-tournament values by a previous study (Adang and Cuvelier 2001) for all comparison groups (local population, focusing on the 2000 Euro, enabling a direct opposition fans and Portuguese police), comparison between the two tournaments. suggesting fans viewed their actual intergroup The observations were undertaken by English- relations more positively than expected before speaking Portuguese observers, split equally the tournament. Furthermore, intergroup between university students and police relations measures and in-group identification academy cadets. Observations covered the measures were both significantly and night before the match and the afternoon positively correlated with perceived similarity and night of the match day. Observers to the Portuguese police. This suggests that recorded in regular intervals data on the once fans viewed as legitimate the policing 31

approach they experienced, this changed the Football Club, a team in the English Football relationship between perceived similarity to League (promoted to the second tier of the police and in-group identification. While English professional football, in 2002–2003). before the tournament, identifying strongly Following two high-profile incidents involving as an England fan was linked to perceived Cardiff City fans in 2001 and 2002, the South dissimilarity to police, the opposite was true Wales Police (SWP – the agency with local following the tournament. jurisdiction), the club, its supporters and the local authority put in place initiatives designed Domestic football leagues – United to tackle hooliganism. A key component of this Kingdom process was the development and adoption of A study by Stott et al. (2008b) examined dialogue-based policing on the part of the SWP, policing operations (including aspects of moving away from a previous concept based dialogue-based or community policing) in the on deterrence. context of domestic matches in the English The authors collected data on the Premier League and its impact on crowd implementation of the new policing approach dynamics and disorder. The study relied on 19 via a series of semi-structured observations at semi-structured observations conducted during 23 Cardiff City matches, with researchers either the 2004–2005 season, primarily in the form of interspersed among fans or in the company shadowing commanders, spotters and Police of the club’s safety and security team. During Support Units (PSUs). In some instances, the the observations, the researchers recorded research team also attended briefings, pre- the chronology of events and their impression event planning meetings and training events. of aspects such as fan and police behaviour The team also interviewed stakeholders and their interactions. While attending the (number of interviews not specified), including matches, researchers also conducted semi- police, club representatives and supporters. structured interviews (number not specified) Specifically with respect to dialogue-based with fans, police officers and club officials. policing, the study suggested that proactive For several matches, interviews were also communication and engagement on the part conducted before and after the match and the of the PSUs had beneficial results in terms of team attended pre-match planning meetings. building positive relations between the police In addition, the authors were in regular and fans, providing accurate risk assessments consultation with the SWP, who discussed and using resources efficiently. Interactions their strategy and tactics and provided with fans were also found to be effective documentation for review to the research team. in making indiscriminate tactics, such as Lastly, the research team also held a series containment exercises, more acceptable to of meetings and other consultations with fan them. The authors noted a potential tension representatives, including people known to the between intelligence- and community-based police as having engaged in hooliganism. policing models but suggested that by adopting An analysis of the data revealed findings a more community-based policing approach, similar to those in previous papers on policing intelligence objectives might also be satisfied strategies. Collected evidence indicated through increases in perceptions of legitimacy. that fans’ relationships with the police and The ESIM model was also the basis for a other authorities were marked by a higher study by Stott et al. (2012), which examined degree of perceived legitimacy. In addition, the policing of matches involving Cardiff City supporters seemed to have exercised a notable 32 Violent and Antisocial Behaviour at Football Events

degree of self-regulation in situations that football in England) in 2014. To police the match, could potentially lead to intergroup conflict, the West Yorkshire Police utilised Police Liaison with hooligan components of the fan base Teams (PLTs) consisting of Police Liaison increasingly marginalised. Furthermore, the Officers (PLOs). The role of PLOs is to serve as adoption of the dialogue-based approach a link between event attendees, organisers and enabled the SWP to gather better intelligence. public order commanders. They use negotiation All these factors combined allowed the police and communication skills to defuse situations to withdraw a substantial portion of resources and resolve minor problems. Their objective dedicated previously to football policing. By is to build relationships and trust between contrast, in away matches involving Cardiff authorities and crowd members and also to City fans, sporadic incidents of major collective prevent disproportionate responses from the conflict occurred. According to the authors, police. According to the authors, this case study this was the case in situations where the host represented the first examination of the use of police appeared to rely more on a deterrence- PLTs in the context of football policing, as their based approach and seemed less ready or able use had hitherto been limited almost exclusively to engage in a dialogue. to policing political demonstrations. Arrest data cited by the authors also confirmed The study followed principles of Participant the observation of a decrease in intergroup Action Research (PAR), in which practitioners conflict. Between the 2002–2003 and 2004– engage with research – one of the authors 2005 seasons, the number of arrests of Cardiff was directly involved in the match as a fans (at both home and away matches) fell public order commander. Data for the case from 194 to 90 and the number of ‘significant study was collected through a combination incidents’ (as categorised by the SWP) fell of researchers’ observations, opportunistic from 14 to five, all of which happened at away interviews with police officers and fans and matches. The authors noted that this period analysis of official documentation, including also coincided with the implementation of transcripts of the police radio log, arrest FBOs, of which Cardiff City had the highest reports, etc. In addition, after the match the number of all English and Welsh clubs in 2002. research team held two debrief sessions However, the authors contended that FBOs by with PLOs and force spotters tasked with the themselves were an insufficient explanation identification of problem fans. for the observed decrease, pointing at the Based on collected data, the authors concluded continued (albeit less frequent) violence at that the PLT was successful in opening a line Cardiff’s away matches. of dialogue with fans, which in turn promoted These conclusions based on data from Cardiff self-regulation among supporters. This appears were also reiterated in a book chapter by Stott to have led to greater perception of police (2014), which stressed that heavy-handed legitimacy and helped avoid conflict situations. tactics could generate or escalate conflict The PLT was also seen as a source of timely situations while strategies based on facilitation and rich information on risks during the day, and dialogue could help reduce tensions. providing a more nuanced assessment than would have been available from FIOs and police More recently, Stott et al. (2016a) conducted a spotters. The PLOs did encounter, however, case study of a dialogue-based policing tactic some resistance from their police colleagues deployed by the West Yorkshire Police at a (including senior staff), some of whom felt match between Oldham Athletic and Bradford that PLTs were not an appropriate tool for City in League One (the third tier of professional 33

football policing. According to the authors, to policing football matches employed by the this pushback was similar to what PLTs had Danish police. Under this approach, specially encountered in other contexts, e.g. the policing trained police officers reach out to those of political protests. supporters of the visiting team amenable to a dialogue with the police in order to help ensure The use of PLTs was also reviewed in a all fans’ behaviour remains legitimate. Such an report by the UK College of Policing (2015), approach is typically implemented before the although the review focused on the practice match and can continue during it. In Denmark, of their deployment at large events in general, the approach was introduced in Aarhus in rather than solely in the context of football 2010 and, following a positive evaluation matches. The report drew on 26 interviews by the Danish national police (Diderichsen with the following groups of stakeholders: 2011, referenced in Havelund et al. 2016) police commanders who had deployed was subsequently expanded nationwide. PLTs (ten interviewees), PLT members Based on an analysis of official police data on (nine interviewees), protestors and event arrests, the authors observed that there was participants who had liaised with PLTs (five a notable decrease in the number of arrests interviewees) and staff of local businesses related to matches in the Danish Superliga (the who had liaised with PLTs (two interviewees). highest league) following the introduction of Overall, there was a general agreement among dialogue-based policing – from 714 arrests interviewed stakeholders that the use of PLTs in the season 2008–2009 to 99 in the season had the potential to contribute to reductions 2014–2015. According to the authors, the in conflicts and disorder. They were thought Danish police viewed the implementation of the to be able to achieve this outcome through concept of dialogue policing as one of the main greater facilitation of communication and reasons behind this decrease. engagement as well as through relationship building. Specifically with respect to football, The study by Havelund et al. (2016) aimed to while one interviewed commander did not feel add the perspective of football supporters to the need to use PLTs at every football match, these observations. They ran an online survey this interviewee, like others, offered examples of members of Brondby Support, a club of of their value. For instance, according to supporters of Brondby (a Danish football club) the offered testimonies, PLTs were seen as with approximately 3,600 members, to ask valuable before European fixtures since away them about their perceptions of the dialogue- fans may arrive and gather early before these based approach to policing. The survey matches, opening an opportunity for PLTs to received 623 responses, which were analysed engage with them. Along similar lines, PLTs and used to construct cross-tabulations could facilitate engagement with organisers of and frequency tables. The vast majority fan-based fora well in advance of a match. As of respondents (82 per cent) felt that that such, the report concluded that the utilisation dialogue-based policing was conducive to a of PLTs in the run-up to and during specific more peaceful atmosphere at and surrounding football fixtures should be considered. football matches. A slightly smaller group (74 per cent) also indicated that the policing Domestic football leagues – other approach had contributed to the strengthening European countries of dialogue between supporters and the Havelund et al. (2016) examined the perceived police. Furthermore, approximately two-fifths effectiveness of a dialogue-based approach of respondents (43 per cent) reported that 34 Violent and Antisocial Behaviour at Football Events

dialogue with the police had helped them better the Special Police Tactic (SPT), a national understand the work police do in connection approach to crowd policing adopted in Sweden with football matches. These results led the (not limited to football). The SPT utilises authors to conclude that dialogue-based small, highly mobile squads of officers (called policing can have positive impacts on the DELTA units) trained to use high-level force overarching police objectives of preventing and and equipped accordingly with the objective resolving conflict situations at football matches to achieve tactical flexibility. Complementing and preserving a calm atmosphere. However, the DELTA units are units of Dialogue Police, this conclusion is based on a relatively small designed to build confidence and trust among survey limited to organised supporters of one crowd members and resolve issues though club in the Danish Superliga. communication and negotiation. Specifically in the context of football, the operationalisation Beedholm Laursen (2017) also examined of the Dialogue Police differs slightly between football policing tactics in Denmark in Swedish regions. The Stockholm region connection with Brondby. The author introduced a football-specific unit of the conducted participant observation on 25 match Dialogue Police, divided into two subunits. days during the 2012–2013 and 2013–2014 The first, deployed in civilian clothes but with seasons. As part of these observations, the clear yellow identification tabards, focuses author conducted a series of ethnographic on building trust and does not make arrests interviews (number not specified) with or use coercion (where possible). The second members of Fraktion64, a small group of subunit, deployed in civilian clothes without Brondby ultras consisting of 10–20 members. any markings, undertakes intelligence tasks, Similarly to Havelund et al. (2016), Beedholm similarly to UK spotters. In other regions, Laursen concluded that dialogue-based policing there is no distinction between liaison and can lead to positive outcomes. In particular, intelligence, and units focusing on supporter it may incentivise supporters to exercise a engagement (called Supporter Police) wear greater degree of self-policing if the way the police uniforms and tabards. Furthermore, crowd is being policed is perceived as fair and in one of the regions outside Stockholm, legitimate. Failure to achieve this may result the police deployed an experimental police in crowd members sympathising with small unit (called EVENT police) configured in the groups of confrontational fans. However, the same way as the DELTA units (albeit with a author also noted that the use of targeted different uniform) and tasked with proactive coercive force remained important for providing communication with fans, focusing on dialogue a flexible and differentiated response, stressing and facilitation as opposed to the use of force. that successful event policing requires At the time of writing of the report, the EVENT implementing dialogue and coercive policing police concept was being trialled in other approaches in a complementary way based on locations as well. The approach is modelled on an informed differentiation of fan behaviour. the Danish concept of dialogue-based policing Policing interventions were also assessed in discussed above in Havelund et al. (2016) a report by Stott et al. (2016b), developed as (and described in detail in Havelund et al. part of a research project aimed at identifying 2013). Important counterparts in the dialogue and developing good practices in the context of with police are Supporter Liaison Officers crowd management at matches in the highest (SLOs), appointed by football clubs to facilitate Swedish football league. The report examined constructive communication with their fans. the implementation and effectiveness of 35

The authors collected data via observations that reason, they are all invariably assessed of 12 fixtures in the highest Swedish league. as 1,1 on the study quality scale. Furthermore, The observations followed the principles of they generally cover either single or a limited participatory research and involved researchers number of matches, raising questions about as well as practitioners (UK police officers, the generalisability of their findings. SLOs from Sweden, Denmark and the UK, stadium safety and security managers from 6.2.4. Key messages Sweden, Denmark and the UK) as members While provided by studies with methodological of the observation team. Each observation limitations, the evidence on the effectiveness was followed by a workshop during which the of dialogue-based policing points uniformly in research team discussed collected data. the same direction. The intervention is seen Reflecting on the collected data, the authors as effective in engendering better intergroup noted that the variation observed between relations, particularly in comparison with high- approaches adopted in various localities can profile tactics. be confusing and contrasts with the notion of a unified national approach. They observed that 6.3. Covert and officers in uniforms faced greater difficulties intelligence-based opening lines of dialogue with fans, which limited their contribution to risk management. policing The Stockholm model appeared to be effective in building trust and confidence between 6.3.1. What is the intervention and where fans and the police and improved mutual has it been used? communication. By contrast, in areas where the Intelligence-led, or intelligence-based, policing roles of spotting and liaison were not formally refers to a policing approach based on the separated, the authors noted comparably collection of intelligence data which can be worse information flows, risk assessment and used to assess risks to public order. Stead relationships between police and fans. With and Rookwood (2007) examined intelligence- respect to EVENT units, the authors found led policing as part of their paper on policing these can add to overall police ability to offer a tactics in the UK. Intelligence-based policing graded tactical response and work with crowds can also be understood as informing other in a fashion that is perceived as legitimate. strategies discussed earlier in this report. For Their impact, however, had been limited due to instance, the imposition of FBOs relies heavily their early stage of development and the lack of on football intelligence data. a clear organisational framework. 6.3.2. What have studies found? 6.2.3. Main strengths and limitations of the evidence Stead and Rookwood (2007) conducted three interviews with fans of three Premier League All studies in this group share the same clubs and with three UK FIOs. Testimonies methodological limitations. They are all by fans and police officers suggested that, in based on either participant observations comparison with approaches in continental or stakeholder testimonies (obtained via Europe, policing in the UK is seen as more interviews or surveys), or a combination covert and intelligence-led, and more effective. thereof and, with the exception of one study, Relatedly, while interviewed FIOs felt that the offer no statistical test of their findings. For use of protective clothing or riot gear increased 36 Violent and Antisocial Behaviour at Football Events

the chances of preventing disorder, interviewed provided by the UK Football Policing Unit fans tended to express the opposite sentiment. (UKFPU), primarily based on reports by FIOs Reflecting on another aspect of UK football and data provided by the Mounted Working policing, interviewed FIOs stressed the utility Group (MWG) of the UK Association of Chief of using intelligence data in policing matches Police Officers, a now-defunct body replaced and fans generally expressed understanding of by the National Police Chiefs’ Council. The its merits (and the necessity of underlying data UKFPU dataset covered 2,804 fixtures that collection). took place at stadiums of Premier League and Championship teams (i.e. in the two highest 6.3.3. Main strengths and limitations of tiers of English football) in the 2010–2011, the evidence 2011–2012 and 2012–2013 seasons in the English domestic league or European club The one study included in this section has competitions, or as friendly matches. This major methodological limitations (scored 1,1). dataset yielded information on attendance, Furthermore, in assessing the effectiveness of match category, number of ejections and intelligence-led policing, it may be impossible number of arrests as well as a binary indication to isolate a specific tactic from others which of whether an incident report had been filed by build on football intelligence data, such as an FIO. The MWG data indicated the number of banning orders. horses (if any) deployed at each match. 6.3.4. Key messages Assessing the bivariate relationship between the two datasets, the authors noted that the Similarly to banning orders, intelligence-led presence of police horses at matches was policing is seen as very effective by the police, associated with higher numbers of arrests although this finding is based on very limited and ejections and an increased probability of evidence. disorder. However, these results are largely a reflection of the fact that the higher risk 6.4. Use of a match was considered to be, the more mounted police likely police horses were to be deployed. To account for possible confounding factors 6.4.1. What is the intervention and where such as crowd size or match risk, the research has it been used? team also conducted a series of multivariate analyses. These revealed that the deployment This intervention takes the form of deploying of mounted officers was not significantly police with horses at or in the vicinity of football associated with ejections or whether reports stadiums. Its use by some police forces in the of disorder were filed. They were, however, UK was the subject of a study by Giacomantonio positively associated with the number of et al. (2015), which assessed the value of arrests. Based on a review of police reports, mounted police units in the UK and specifically the authors suggested that the presence of examined the deployment of police with horses officers on horseback may facilitate arrests, in the context of football matches in England. for instance by providing greater visibility of ticket touting in a large crowd. Examining 6.4.2. What have studies found? only matches where horses were present, Giacomantonio et al. (2015) undertook the multivariate models suggested that the a statistical analysis of quantitative data number of horses deployed was weakly (at 37

the ten per cent level) associated with the and because police leadership may decide to number of arrests and had a significant (at send horses predominantly to matches where the five per cent level) negative association more disorder is anticipated. with the likelihood of an incident report The authors also conducted two focus groups being filed. Again, a review of police reports with football fans (19 participants in total) to offered suggestions of possible mechanisms complement the analysis presented above through which these associations manifest with an account of the public’s perception. themselves. In one instance, mounted officers All participants were involved with fan were able to position horses to form a cordon organisations and were serious and lifelong separating home and away fans and prevent football fans; at least two participants had any confrontation. been previously subjected to a banning order. The researchers also ran a survey of police Overall, focus group participants held a officers deployed to football matches during negative view of football policing but tended to the 2013–2014 season, asking respondents regard mounted police separately from other to rate the quality of policing and other police forces. The presence or absence of related variables. Data obtained through the mounted police did not seem to play a role in survey was complemented with a review of how fans viewed football policing, with other operational orders, documents establishing factors (such as use of force) playing a much policing plans for the match and detailing more important role. In interpreting the results the number of police deployed. In total, 128 of the focus groups, it is necessary to keep in valid responses were obtained covering 49 mind that these represent the views of a very unique matches. Approximately two-thirds of small number of fans and not including any responses came from match commanders casual football supporters. and FIOs, with the rest coming from officers in other roles such as PSU commanders or 6.4.3. Main strengths and limitations of tactical advisers. Most of the matches covered the evidence by the responses took place in the Premiership, The evidence on the effectiveness of mounted the Championship and League One (the three police is provided by only one study, albeit of top tiers of English football). According to the a relatively robust design (scored 3,2). As the survey results, the presence of horses was study noted, the results should be understood negatively associated with levels of disorder as preliminary and further work is needed to reported at individual matches (significant determine the chain of causality. at the ten per cent level). In addition, survey responses indicated that the presence (as 6.4.4. Key messages well as the number) of horses at matches was positively linked to how officers perceived The presence of horses was found to be the overall level of positive interactions with associated with a higher number of arrests, members of the public (significant at the ten and the number of horses was negatively per cent level). Overall, however, the research associated with the likelihood of an incident team stressed that it was impossible to draw report being filed. One possible explanation causal conclusions regarding the associations is that horses may make arrests easier described above, primarily because mounted and allow for more effective prevention of police were more likely to be present at disorder. Positive outcomes associated with matches with higher overall levels of policing the presence of horses were also reported by police officers deployed at football matches.

7 Discussion

The results of this review indicate that while a our understanding of what works. Notable range of interventions and strategies to combat persisting challenges in this regard are the football-related antisocial and violent behaviour isolation of the effect of individual interventions have been described in the literature, a much adopted as part of a whole package of smaller number have been subjected to an measures and the transferability of evidence assessment of their effectiveness. The evidence across contexts. provided by studies included in this review Nevertheless, the evidence presented in offers a mixed picture. For some interventions, this report offers useful insights for the such as early kick-off times or the utilisation of organisation of football matches as well as security cameras, results from a small number future efforts to assess the effectiveness of of methodologically robust studies suggest they different organisational approaches. First and can achieve positive outcomes. A larger number foremost, in a few areas, the reviewed evidence of less robust studies also uniformly appear tentatively points in the direction of positive to support the implementation of dialogue- outcomes associated with clearly defined based approaches to policing football events. types of interventions. Further, the literature By contrast, no evidence of effectiveness was included in the review also provides examples found for other interventions, such as various of how different interventions have been put forms of bans on alcohol. into practice and considerations surrounding Overall, the findings of the review demonstrate their implementation. Lastly, the discussion of that the evidence base underpinning existing assessments and what the reviewed interventions and tactics to tackle football- studies found suggests possible data sources related disorder is underdeveloped. Further and methodological approaches for those research, particularly in the form of rigorous wishing to replicate these assessments in their evaluation studies, will be necessary to develop respective contexts.

References

Adang, O. & C. Cuvelier. 2001. Policing Di Domizio, M. & R. Caruso. 2015. ‘Hooliganism Euro2000: International police co-operation, and Demand for Football in Italy: Attendance information management and police and counterviolence policy evaluation.’ German deployment. Ubbergen, NL: Tandem-Felix. Economic Review 16(2): 123–37. doi:10.1111/ geer.12051 Beedholm Laursen, R. 2017. ‘Danish Police Practice and National Football Fan Crowd Diderichsen, A. 2011. Etik for politifolk Behaviour. Dialogue or coercive force?’ Soccer [Ethics for Police Officers]. Frederiksberg, DK: and Society 18: 1–16. doi:10.1080/14660970.2 Samfundslitteratur. 017.1302933 Drury, J. & S. Reicher. 2000. ‘Collective Action College of Policing. 2015. The Use of Police and Psychological Change: The emergence of Liaison Teams in the Policing of Events: A new social identities.’ British Journal of Social review of practice. Available at: Psychology 39: 579–604. http://library.college.police.uk/docs/college-of- Ferrari, L. 2012. ‘Stadiums: Italian measures to policing/Use-of-police-liaison-teams-2015.pdf. combat stadium violence and disorder.’ World Date accessed: 22 May 2018. Sports Law Report 10(6). Connors, E. 2007. Planning and Managing Ferwerda, H.B., S. Barlagen, J. Bos, B. Security for Major Special Events: Guidelines for Bremmers, T. Van Ham & E. Van der Torre. law enforcement. Alexandria, Virginia: Institute 2011. Uitsupporters Centraal. Een Onderzoek for Law and Justice. Available at: naar het Bezoeken van Uitwedstrijden in het https://www.hsdl.org/?view&did=482649. Date Betaald Voetbal [Supporters central - A study of accessed: 22 May 2018. visiting away games in professional football]. Council of the EU. 2006. Council Resolution The Netherlands: Auditteam Voetbal en of 4 December 2006 concerning an Updated Veiligheid. Handbook with Recommendations for Frosdick, S. & P. Marsh. 2005. Football International Police Cooperation and Hooliganism. Cullompton, UK: Willan Measures to Prevent and Control Violence Publishing. and Disturbances in Connection with Football Matches with an International Dimension, in which At Least One Member State is Involved (2006/C 322/ 01). 42 Violent and Antisocial Behaviour at Football Events

Giacomantonio, C., B. Bradford, M. Davies & ———. 2013. Football Banning Orders. Cited in R. Martin. 2015. Making and Breaking Barriers: Hopkins & Hamilton-Smith. 2014 Assessing the value of mounted police units in Hopkins, M. 2014. ‘Ten Seasons of the Football the UK, RR-830/1-ACPO. Santa Monica, Calif.: Banning Order: Police officer narratives on the RAND Corporation. Available at: operation of banning orders and the impact on https://www.rand.org/pubs/research_reports/ the behaviour of “risk supporters”.’ Policing and RR830.html. Accessed 22 May 2018. Society 24(3): 285–301. doi:10.1080/10439463 Hamilton-Smith, N., B. Bradford, M. Hopkins, .2013.784293 J. Kurland, C. Lightowler, D. McArdle & N. Hopkins, M. & N. Hamilton-Smith. 2014. Tilley. 2011. An Evaluation of Football Banning ‘Football Banning Orders: The highly effective Orders in Scotland. Edinburgh, UK: Scottish cornerstone of a preventive strategy.’ In Government Social Research. Football Hooliganism, Fan Behaviour and Crime: Hamilton-Smith, N. & M. Hopkins. 2013. ‘The Contemporary issues, edited by M. Hopkins Transfer of English Legislation to the Scottish and J. Treadwell. Basingstoke, UK: Palgrave Context: Lessons from the implementation Macmillan. of the Football Banning Order in Scotland.’ Hylander, I. & K. Granström. 2010. ‘Organizing Criminology and Criminal Justice 13(3): for a Peaceful Crowd: An example of a football 279–97. match.’ Forum: Qualitative Sozialforschung/ Hau, S. 2008. Communication as the Most Forum: Qualitative Social Research 11(2). Important Police Strategy at the Football World Available at: Cup Final 2006. FOG report no. 61. Linköping: http://www.qualitative-research.net/index.php/ Department of Behavioural Sciences and fqs/article/view/1462/2968. Date accessed: Learning (IBL), Linköping University 22 May 2018. doi:http://dx.doi.org/10.17169/ fqs-11.2.1462. Havelund, J., M. Lauritsen, L. Joern & K. Rasmussen. 2016. ‘Not All Cops are Bastards: Madaleno, M. & S. Waights. 2014. Guide to Danish football supporters’ perception of Scoring Methods Using the Maryland Scientific dialogue policing.’ European Police Science and Methods Scale. What Works Centre for Local Research Bulletin 15: 42–47. Economic Growth. As of 18 May 2018: http://www.whatworksgrowth.org/public/files/ Havelund, J., C. Stott, K. Rasmussen & L. Joern. Scoring-Guide.pdf 2013. ‘Policing Football – From ‘hooliganism’ to inter-groups dynamics.’ Journal of Education, Madensen, T. & J.E. Eck. 2008. Spectator Psychology and Social Sciences 1(2): 36–41. Violence in Stadiums. US Department of Justice, Office of Community Oriented Policing Home Office. 2005. Football (Disorder) Act Services. 2000: Report to Parliament. November. As of 18 May 2018: Mojet, H. 2005. ‘The European Union and http://webarchive.nationalarchives. Football Hooliganism.’ The International Sports gov.uk/20100408143048/http://www. Law Journal Issue 1-2. homeoffice.gov.uk/documents/Football- Disorder-20062835.pdf?view=Binary 43

Nepomuceno, T.C.C., J.A. de Moura, L. Câmara Schaap, D., M. Postma, L. Jansen & J. e Silva & A.P.C.S. Costa. 2017. ‘Alcohol and Tolsma. 2015. ‘Combating Hooliganism in Violent Behavior Among Football Spectators: the Netherlands: An evaluation of measures An empirical assessment of Brazilian’s to combat hooliganism with longitudinal criminalization’. International Journal of Law, registration data.’ European Journal on Criminal Crime and Justice 51: 34–44. Policy and Research 21(1): 83–97. Pearson, G. 2005. The regulation of football Schreiber, M. & O. Adang. 2010a. ‘Fiction, Facts crowds – Preventing Violence. Paper presented and a Summer’s Fairy Tale – Mixed messages at the WHO National Conference. Liverpool. at the World Cup 2006.’ Policing and Society 20(2): 237–55. Pearson, G. & A. Sale. 2011. ‘“On the Lash” – Revisiting the effectiveness of alcohol controls ———. 2010b. ‘The Poles are Coming! Fan at football matches.’ Policing and Society 21(2): behaviour and police tactics around the World 150–66. Cup match Germany vs. Poland.’ Sport in Society 13(3): 470–88. Pearson, M.J.G. 2006. ‘Football Banning Orders: Analysing their use in court.’ Journal of Schreiber, M. & C. Stott. 2012. ‘Policing Criminal Law 70(6). International Football Tournaments and the Cross-cultural Relevance of the Social Identity Priks, M. 2014. ‘Do Surveillance Cameras Affect Approach to Crowd Behaviour.’ Police Practice Unruly Behavior? A close look at grandstands.’ and Research 13(5): 407–20. doi:10.1080/1561 Scandinavian Journal of Economics 116(4): 4263.2011.627742 1,160–79. doi:10.1111/sjoe.12075 Spaaij, R. 2005. The Prevention of Football Raspaud, M. & F. da Cunha Bastos. 2013. Hooliganism: A transnational perspective. In: ‘Torcedores de Futebol: Violence and public Actas del X Congreso Internacional de Historia policies in Brazil before the 2014 FIFA World del Deporte. Aquesolo, J, ed. CESH, Seville, pp. Cup.’ Sport in Society 16(2): 192–204. doi:10.10 1-10. 80/17430437.2013.776251 Stead, D. & J. Rookwood. 2007. ‘Responding to Reicher, S. 1996. ‘“The Battle of Westminster”: Football Disorder: Policing the British football Developing the social identity model of crowd fan.’ Journal of Qualitative Research in Sports behaviour in order to explain the initiation and Studies 1(1): 33–41. development of collective conflict.’ European Journal of Social Psychology 26: 115–34. Stott, C. 2014. ‘Policing Football “Hooliganism”.’ In Football Hooliganism, Fan Behaviour and ———. 2001. ‘The St. Pauls’ Riot: An explanation Crime, edited by M. Hopkins and J. Treadwell. of the limits of crowd action in terms of a Basingstoke, UK: Palgrave Macmillan. social identity model.’ In Intergroup Relations: Essential readings, edited by M.A. Hogg and D. Stott, C., O. Adang, A. Livingstone & M. Abrams. Philadelphia, PA: Psychology Press. Schreiber. 2007. ‘Variability in the Collective Behaviour of England Fans at Euro2004: Rosander, M. & G. Guva. 2012. ‘Keeping the “Hooliganism”, public order policing and social Peace – Police behaviour at a mass event.’ change.’ European Journal of Social Psychology Journal of Investigative Psychology and 37(1): 75–100. doi:10.1002/ejsp.338 Offender Profiling 9(1): 52–68. doi:10.1002/ jip.1352 44 Violent and Antisocial Behaviour at Football Events

———. 2008a. ‘Tackling Football Hooliganism Stott, C. & S. Reicher. 1998. ‘How Conflict – A quantitative study of public order, policing Escalates: The inter-group dynamics of and crowd psychology.’ Psychology Public collective football crowd “violence”.’ Sociology Policy and Law 14(2): 115–41. doi:10.1037/ 32: 353–77. a0013419 Stott, C., O. West & M. Radburn. 2016a. ‘Policing Stott, C., J. Havelund, F. Lundberg, S. Khan, Football “Risk”? A participant action research L. Joern, J. Hoggett, K. Rasmussen & S. case study of a liaison-based approach to Vestergren. 2016b. Policing Football in Sweden. “public order”.’ Policing and Society 18(3): 1–16. Syddansk Universitet. As of 18 May 2018: doi:10.1080/10439463.2015.1126267 https://core.ac.uk/download/pdf/50621412.pdf Tsoukala, A. 2013. ‘Controlling Football-related Stott, C., J. Hoggett & G. Pearson. 2012. Violence in France: Law and order versus the ‘“Keeping the Peace”: Social Identity, procedural rule of law.’ Sport in Society 16(2): 140–50. doi: justice and the policing of football crowds.’ 10.1080/17430437.2013.776247 British Journal of Criminology 52(2): 381–99. Van der Aa, E. 2011. ‘Tijd voor afschaffing doi:10.1093/bjc/azr076 combiregeling [Time for the abolition of combi- Stott, C., A. Livingstone & J. Hoggett. regulation]’ De Telegraaf, 27 January. 2008b. ‘Policing Football Crowds in England Veuthey, A. & L. Freeburn. 2015. ‘The Fight and Wales: A model of “good practice”?’ Against Hooliganism in England: Insights for Policing and Society 18(3): 258–81. other jurisdictions?’ Melbourne Journal of doi:10.1080/10439460802091641 International Law 16(1). Stott, C. & G. Pearson. 2006. ‘Football Warren, I. & R. Hay. 2009. ‘“Fencing them in”: Banning Orders, Proportionality, and The A-League, policing and the dilemma of Public Order Policing.’ Howard Journal public order.’ Soccer and Society 10(1): 124–41. of Criminal Justice 45(3): 241–54. doi:10.1080/14660970802472726 doi:10.1111/j.1468-2311.2006.00419.x 45

Annex A. Methodology

To address the research questions discussed The purpose of these searches was to in Chapter 1, a rapid evidence assessment identify search terms producing the most was carried out to identify and review the relevant results for this review. literature. Unlike a full systematic review, 3. Identification of three pieces of literature which aims to search the entire evidence base of relevance to this review. Final search comprehensively, the scope and coverage of strings were then tested by checking that the rapid evidence assessment was restricted these papers would be captured in the through search and screening criteria selected results. to focus on the most relevant literature. This Annex sets out the methods employed in the 4. A systematic search using a targeted review. search of criminal justice and behavioural science databases. We liaised with Methods research experts from RAND’s Knowledge Services team, composed of professional We used several different ways of searching librarians and information specialists, for relevant evidence. These included the to focus on the most comprehensive following: databases. For academic literature, the following databases were searched in 1. Three interviews with experts on crowd the final review: Web of Science, Scopus, behaviour and safety from European law PsycInfo and Social Sciences Abstracts. In enforcement agencies, to help define the addition, grey literature was also searched scope of the review and discuss relevant using the following databases: OpenGrey, literature and key themes emerging from IssueLab, OAISTER and Advanced Google research in this field. search. These databases were searched 2. A preliminary search in Google Scholar, simultaneously by RAND Knowledge Web of Science and JSTOR informed by Services, who collated the results from these early discussions, using a number these databases before transmitting them of iterations of the following search terms: to the research team. (football OR soccer) AND (crowd OR 5. ‘Snowball’ searching. All papers captured spectator OR fan) AND (behav* OR violen* by the search were then analysed by OR antisocial OR alcohol* OR hooligan*) the research team. Bibliographies and AND (strateg* OR prevent* OR respon* OR references cited in the literature that met measure* OR effect* OR polic* OR reduc*). the inclusion criteria were followed up on 46 Violent and Antisocial Behaviour at Football Events

and identified for inclusion in this review. Database searches Through this snowballing approach, seven pieces of literature were added to the Search terms review. Search terms and synonyms were as follows: All search results were screened by members of the research team against the inclusion criteria. Screeners consulted with each other (football OR soccer) AND during the early stages of the review to ensure consistency in the process. For the papers (crowd* OR spectator* OR fan OR fans) which were included in the review, the evidence AND was analysed and synthesised and the findings (behaviour* OR behavior* OR violen* OR interpreted and presented in this report (see antisocial OR anti-social OR “anti social” OR below). alcohol* OR hooligan*) AND

Inclusion criteria (strateg* OR prevent* OR respon* OR measure* OR effect* OR polic* OR reduc* OR assess* OR The inclusion criteria were as follows: evaluat*)

Included in the rapid evidence assessment Databases searched and numbers of studies found Articles in English The research team liaised with research experts from RAND’s Knowledge Services team to Articles reporting primary research focus on the most comprehensive and relevant Grey literature databases. Initially, we proposed searching Web of Science, Scopus, PsycInfo, Social Sciences Articles published after 2005 Abstracts, JSTOR and Google Scholar for peer- Research from or about any country reviewed research. It was subsequently decided to exclude JSTOR from the searches, as the Systematic reviews/rapid evidence assessments database produced thousands of irrelevant results. Furthermore, Knowledge Services The exclusion criteria were as follows: advised excluding Google Scholar, as the search engine cannot support sophisticated search strings, does not allow the exportation of more Excluded from the rapid evidence assessment than one citation at a time into EndNote and returns results of which most, if not all, would Commentaries, editorials and features be found in the other academic databases. In Research relating to sports other than football addition, grey literature was also searched for using OpenGrey, IssueLab, OAISTER and an Narrative reviews Advanced Google search. Search results were imported into a single EndNote file and combined with the literature identified in the previous review. After removal of duplicate studies, the titles and abstracts 47

were screened by two researchers (LS and JP). Data from relevant studies was extracted Full papers of potentially relevant studies were by three researchers (SD, LSF, JT) into a retrieved and a final judgement on eligibility standardised template (see Annex 3). was made by three researchers (SD, LSF, JT).

Search results (academic and grey databases)

Database Search terms Number of results

Web of TOPIC: (football OR soccer) 209 (citations Science AND about American Indexes=SCI- TOPIC: (crowd* OR spectator* OR fan OR fans) football) = 191 – EXPANDED, AND duplicates with R1 SSCI, A&HCI TOPIC: (behaviour* OR behavior* OR violen* OR antisocial OR anti- = 26 social OR “anti social” OR alcohol* OR hooligan*) AND Strateg* OR prevent* OR respon* OR measure* OR effect* OR polic* OR reduc* OR assess* OR evaluat*

Scopus TITLE-ABS-KEY ((football OR soccer)) AND TITLE-ABS-KEY 273 – American Article, ((crowd* OR spectator* OR fan OR fans)) AND TITLE-ABS-KEY football and Review, ((behaviour* OR behavior* OR violen* OR antisocial OR anti-social duplicates = 35 Article in OR “anti social” OR alcohol* OR hooligan*)) AND TITLE-ABS-KEY Press (strateg* OR prevent* OR respon* OR measure* OR effect* OR polic* OR reduc* OR assess* OR evaluat*))

PsycInfo football OR soccer 170 – duplicates Academic AND and American Journals and crowd* OR spectator* OR fan OR fans football = 9 Reviews AND behaviour* OR behavior* OR violen* OR antisocial OR anti-social OR “anti social” OR alcohol* OR hooligan* AND strateg* OR prevent* OR respon* OR measure* OR effect* OR polic* OR reduc* OR assess* OR evaluat*

Social football OR soccer 38 – duplicates Sciences AND and American Abstracts crowd* OR spectator* OR fan OR fans football = 1 Academic AND Journals and behaviour* OR behavior* OR violen* OR antisocial OR anti-social OR Reviews “anti social” OR alcohol* OR hooligan* AND strateg* OR prevent* OR respon* OR measure* OR effect* OR polic* OR reduc* OR assess* OR evaluat* 48 Violent and Antisocial Behaviour at Football Events

Database Search terms Number of results

OpenGrey, football crowd* OR spectator* OR fan OR fans “behaviour “ site:.org. 24 IssueLab, uk OAISTER, football hooligan* “behav*” site:.org.uk Advanced football “behav*” spectator site:.gov.uk filetype:pdf Google soccer behav* crowd site:.gov.uk search football anti-social OR antisocial OR alcohol OR hooligan* PDF files “spectator*” site:.org only soccer anti-social OR antisocial OR alcohol OR hooligan* “spectator*” site:.org anti social behavior football spectators europe

Summary results application of inclusion and exclusion criteria, title and abstract screen, full-text review and • Total number of results from academic data extraction. This process resulted in the databases: 70 following: • Total number of results from grey literature • 266 (non-unique) sources identified search: 24 • 47 retained after applying inclusion criteria • Total number of studies retained after title and removing duplicates and abstract search: 39 • 24 retained after title and abstract search • Total number of studies included in the review: 19. • 7 included in the review.

Snowball search Combined count

The bibliographies of studies identified via In total, we identified 141 publications across the search of academic and grey literature the searches after removing duplicates and databases retained for full-text review were applying inclusion/exclusion criteria. Following also searched for potentially relevant sources. a review of titles and abstracts, 63 studies were The research team compiled the list of all retained and, following a full-text review, 26 potentially relevant sources and undertook studies were identified as relevant for inclusion. the same steps as those for the results of the In addition, 26 articles informed the review database search, i.e. removal of duplicates, indirectly and are cited within this report. 49

Appendix B. Overview of implemented strategies

In addition to the basic characteristics of practice without any discussion of supporting individual interventions, the table below evidence. As the last column demonstrates, summarises which reviewed papers discuss this literature review identified studies that the intervention in question and the last offered evidence of effectiveness (albeit column indicates whether the review identified of a varying degree of strength) for only a an assessment of its effectiveness. In multiple small subset of interventions listed in Table instances, interventions were presented in 6. This evidence is discussed in detail in the the reviewed literature as effective or best subsequent section.

Table 6. Overview of interventions to counter antisocial and violent behaviour

Empirical Undertaken Undertaken Intervention assessment of Intervention when and Relevant literature by whom? target effectiveness where? identified?

Venue organisation

CCTV Football Inside All fans Council of the EU Yes cameras club stadium (2006); Home Office before (2005); Mojet (2005); match Priks (2014); Spaaij (2005); Veuthey and Freeburn (2015)

Segregation Football Inside All fans Council of the EU No of home and club stadium (2006); Di Domizio and away fans during Caruso (2015); Ferrari match (2012); Home Office (2005); Madensen and Eck (2008); Mojet (2005); Raspaud and da Cunha Bastos (2013); Spaaij (2005); Stead and Rookwood (2007); Veuthey and Freeburn (2015); Warren and Hay, (2009); 50 Violent and Antisocial Behaviour at Football Events

Empirical Undertaken Undertaken Intervention assessment of Intervention when and Relevant literature by whom? target effectiveness where? identified?

Strategic Football Inside All fans Madensen and Eck No placement of club stadium (2008); Mojet (2005) stages, sound before equipment, match and screens

All-seater Football Inside All fans Home Office (2005); No stadiums club stadium Mojet (2005); Spaaij before (2005); Madensen and match Eck (2008); Veuthey and Freeburn (2015)

Establishment Football Inside All fans Madensen and Eck No of processing club stadium (2008) and holding before areas for match spectators who are arrested or refuse to leave the premises

In-stadium Football Inside All fans Connors (2007); Yes alcohol club (could stadium Madensen and Eck restrictions be legally during (2008); Mojet (2005); mandated) match Nepomuceno et al. (2017); Pearson and Sale (2014);Schaap et al. (2014); Council of the EU (2006); Veuthey and Freeburn (2015);

Posting Football Inside All fans Council of the EU No signs to club stadium (2006); Madensen and convey safety before Eck (2008) information to match spectators

Changing Football N/A All fans Madensen and Eck No venues for association (2008) ‘high-risk’ events 51

Empirical Undertaken Undertaken Intervention assessment of Intervention when and Relevant literature by whom? target effectiveness where? identified?

Physical Football Inside All fans Council of the EU No security club stadium (2006); Ferrari (2012); features (e.g. before Madensen and Eck Plexiglas match (2008); Mojet (2005); barriers) Raspaud and da Cunha Bastos (2013); Veuthey and Freeburn (2015);

Event organisation

Matchday Local Inside All fans Pearson and Sale Yes alcohol government stadium (2014)Veuthey and restrictions in Freeburn (2015); host town

Identity Football Outside All fans Di Domizio and Caruso Yes card/fan club stadium (2015); Ferrari (2012); registration before Home Office (2005); schemes match Schaap et al. (2014); Spaaij (2005); Veuthey and Freeburn (2015)

Refusing Football Outside Drunk fans Council of the EU No entrance to club stadium (2006); Madensen and inebriated before Eck (2008); Veuthey spectators match and Freeburn (2015)

Earlier/ Football Outside All fans Pearson and Sale Yes daylight kick- association stadium (2014); Schaap et al. off times before (2014) match

Mandatory Football Outside Away team Schaap et al. (2014) Yes transport to club stadium fans stadium before match 52 Violent and Antisocial Behaviour at Football Events

Empirical Undertaken Undertaken Intervention assessment of Intervention when and Relevant literature by whom? target effectiveness where? identified?

Restriction/ Courts/ Outside High-risk Council of the EU Yes banning football club stadium fans (2006); Di Domizio orders before and Caruso (2015); match Ferrari (2012); Hopkins and Hamilton-Smith (2014); Hamilton et al. (2013); Hamilton- Smith et al. (2011); Home Office (2005); Hopkins (2014); Pearson (2006); Mojet (2005); Stead and Rookwood (2007); Tsoukala (2013); Watson (2013); Veuthey and Freeburn (2015)

Removal of Police, Inside Disruptive Connors (2007); No disruptive security stadium fans Madensen and Eck spectators staff, event during (2008) staff match

Screening of Security Outside All fans Council of the EU No items brought staff, event stadium (2006); Madensen into the staff before and Eck (2008); Mojet stadium match (2005); Veuthey and Freeburn (2015); Warren and Hay (2009)

Reducing Police, Inside and All fans Council of the EU No situational security outside (2006); Ferrari (2012); instigators staff, event stadium Madensen and Eck of violence staff before (2008); Warren and (e.g. offensive and during Hay (2009) T-shirts, match banners)

Control of Police, Inside and All fans Council of the EU No the dispersal security outside (2006); Madensen and process staff, event stadium Eck (2008) staff after match

Ticket policies, Football Outside the All fans Council of the EU No ban on touting clubs stadium (2006); Mojet (2005); before the Veuthey and Freeburn match (2015) 53

Empirical Undertaken Undertaken Intervention assessment of Intervention when and Relevant literature by whom? target effectiveness where? identified?

Policing and other staff

Intelligence- Police Primarily All fans Council of the EU Very little on led policing outside (2006); Council of the its own but stadium EU (2010); Connors intelligence before (2007); Havelund et policing feeds match al. (2013); Hylander into other et al. (2010); Raspaud assessed and da Cunha Bastos interventions (2013); Rosander and Guva (2012); Spaaij (2005); Stead and Rookwood (2007)

Community/ Police Primarily All fans Council of the EU Yes low-intensity/ outside (2010); Hau (2008); dialogue- stadium Schreiber and Adang based before (2010a, b); Schreiber policing/police match and Stott (2012); liaison teams Stott et al. (2007); Stott et al. (2008a,b); Stott et al. (2012); Stott (2014); Stott et al. (2016a,b); Watson (2013); Warren and Hay (2009); Veuthey et al. (2015)

Use of riot Police Primarily All fans Schreiber and Adang Not on its own gear outside (2010a, b); Stead and but as part of stadium Rookwood (2007) assessments before surrounding match low-intensity policing

Establishment Police Inside and All fans Council of the EU No of an effective outside (2006); Madensen and command stadium Eck (2008) post before and during match

Use of Police Inside and All fans Madensen and Eck No different outside (2008) security ‘levels’ stadium before and during match 54 Violent and Antisocial Behaviour at Football Events

Empirical Undertaken Undertaken Intervention assessment of Intervention when and Relevant literature by whom? target effectiveness where? identified?

Increased Police Inside and All fans Connors (2007); Not on its own visibility of outside Madensen and Eck but assessed security/ stadium (2008); Mojet (2005); in contrast with police; show before Stead and Rookwood low-intensity of force and during (2007) policing match

Relying on Police Inside and All fans Connors (2007); Not on its own reactive outside Madensen and Eck but assessed tactics stadium (2008) in contrast with before low-intensity and during policing match

Standby Police Inside and All fans Connors (2007); No officers; relief outside Giacomantonio et al. officers stadium (2015) before and during match

Deployment Police Inside and All fans Giacomantonio et al. Yes of mounted outside (2015) police stadium before and during match

Use of hand- Police Primarily All fans Stead and Rookwood Not on its own held cameras outside (2007) but as part of stadium intelligence-led before policing and during match

Laws, policies, partnerships

Coordinating Football Ongoing All fans Connors (2007); No platforms clubs, intervention Council of the EU for relevant police, local (2006); Council of the stakeholders government, EU (2010); Di Domizio media and Caruso (2015); Ferrari (2012); Mojet (2005); Warren and Hay (2009); 55

Empirical Undertaken Undertaken Intervention assessment of Intervention when and Relevant literature by whom? target effectiveness where? identified?

Zero-tolerance Football Ongoing Fan Mojet (2005); Spaaij No of hooliganism clubs intervention associations (2005); Veuthey and policies from Freeburn (2015) football clubs; information campaigns

Fan projects Football Ongoing All fans Home Office (2005); No (e.g. liaison clubs intervention Spaaij (2005); Veuthey between and Freeburn (2015) clubs and fan associations)

Support Football Ongoing All fans Home Office (2005); No, although to anti- fans intervention Spaaij (2005) fans’ self- hooliganism policing is opposition discussed in from football studies on fans low-intensity policing

Certification Local Ongoing Football Madensen and Eck No and permits to and state intervention clubs (2008); Veuthey and host events government Freeburn (2015)

Adopting/ Local Ongoing All fans Di Domizio and Caruso No advertising and state intervention (2015); Ferrari (2012); legislation and government Home Office (2005); penalties for Madensen and Eck violent and/or (2008); Mojet (2005); hate behaviour Raspaud and da Cunha Bastos (2013); Tsoukala (2013); Watson (2013); Warren and Hay (2009); Veuthey and Freeburn (2015)

Encouragement Football Ongoing All fans Madensen and Eck No of marketing club intervention (2008) to gender- and age-diverse crowds

Annex C. Data extraction template

Table 7. Data extraction template – Part 1

Coder

General Full reference

First author

Year

Brief summary

Include? Yes/no/yes + reason for studies that would otherwise be excluded

Format of document Review/synthesis, empirical study, letter to editor…

Type of data used Quantitative – primary data collection through surveys Quantitative – using secondary data Qualitative – interviews/focus groups Mixed methods Other

Type of sporting event (World Cup, Euro, domestic…)

Year of event

Country of event

Country of origin of perpetrators

Specific groups of perpetrators

Country origin of victims 58 Violent and Antisocial Behaviour at Football Events

Table 8. Data extraction template – Part 2

Interventions Name/Title

Objective(s) What it aimed to achieve

Actors undertaking the Who was implementing it? Who was otherwise engaged measures and how? E.g. police, private sector, supporters’ groups

Target of measure Who it was targeted at (fans, others)’ individual vs. group level, or a specific place

Timeframe When it was implemented

Where were the measures undertaken

Inputs What were the costs? Any information on resources involved?

Activities/Measures What was planned or actually delivered?

Outputs and outcomes Expected or evidence on actual outputs and outcomes

Impacts Expected or evidence on actual impacts

Table 9. Data extraction template – Part 3

Assessments Comparison condition Was there a comparison condition (does not have to be a comparison group)?

Type of evaluation E.g. control/comparison groups, pre/post, etc.; describe statistical methods used

Sample size (initial)

Sample size (analysis)

Limitations Explicitly reported

Any other coder comments

Quality assessment Topic relevance

Quality of study design and conduct

Quality of analysis and interpretation

Any other comments

Further literature to review