The Rise and Fall of the English Defence League: Self-Governance, Marginal Members and the Far Right

What determines the success or failure of far-right organisations? This article uses new qualitative data to explain the sudden rise and subsequent decline of the English Defence League, an anti-Islamic, street protest organisation established in the UK in 2009. We explain the rise and fall of the English Defence League through the lens of the theory of collective action to show that the English Defence League initially motivated activism by supplying selective incentives that were enhanced by the participation of others. The pursuit of ‘participatory crowding’ led to indiscriminate recruitment into the organisation that enabled numbers to expand into the thousands, but ultimately caused the English Defence League’s downfall because it resulted in the presence of large numbers of ‘marginal members’ with low levels of commitment whose subsequent exit was decisively destructive. Self-governance mechanisms to ensure greater loyalty from members could have prevented the English Defence League’s decline but would also have limited its initial success.

success (Fielding, 1981;Walker, 1977), and in the 2000s, the British National Party (BNP) won a number of local authority seats and close to a million votes and two Members of the European Parliament (MEPs) at the 2009 European Parliament elections (Goodwin, 2010(Goodwin, , 2011, but both of these successes proved short-lived -each organisation was effectively moribund a decade after the peak of its success. In 2010 and 2011, the EDL attracted thousands of participants to its public demonstrations, but it then also experienced a steep decline which saw numbers dwindle to a few hundred participants and the end of its capacity for mass mobilisation. This article uses qualitative research evidence to explain the trajectory of the EDL and thereby meets Goodwin's (2011: 50) challenge of getting inside 'the black box of right-wing extremist organisations' to understand how activists are recruited and commitment is sustained (or not).
We explain the rise and fall of the EDL through the lens of the theory of collective action (Olson, 1965;Ostrom, 1998;Tullock, 1971) to show that the EDL initially motivated activism by supplying selective incentives that were enhanced by the participation of others -what Iannaccone (1992) termed 'participatory crowding'. The pursuit of participatory crowding led to indiscriminate recruitment into the organisation that enabled numbers to expand into the thousands, but ultimately caused the EDL's downfall because it resulted in the presence of large numbers of 'marginal members' with low levels of commitment whose subsequent exit was decisively destructive. Self-governance mechanisms to ensure greater loyalty from members could have prevented this decline but would also have limited the group's initial success.
Most studies of right-wing extremism utilise either demand-side or supply-side explanations. Demand-side explanations suggest far-right groups will appeal to individuals with low socioeconomic status in times of social and economic uncertainty (Betz, 1998;Falter and Schumann, 1988), or show that far-right activism may satisfy participant demands for dignity and fraternity (Ezekiel, 1995;Fryer and Levitt, 2012). In the context of the EDL, scholars have identified the salience of the group's Islamophobic rhetoric to young, White, working-class men lacking educational or employment opportunities (Goodwin et al., 2016;Kassimeris and Jackson, 2015;Winlow et al., 2017), and the emotional appeal of EDL activism to these individuals (Busher, 2016;Pilkington, 2016).
Supply-side approaches emphasise the role of political organisations in providing opportunities for political participation and suggest that the organisation of parties and groups will be a key determinant of their success or failure (Dinas et al., 2016;Kitschelt, 1995;Mudde, 2007). In respect of the EDL, scholars have identified organisational ineptitude and internal conflict over the ends of the group and the means to achieve them (Busher, 2016), and the resignation of its founder Tommy Robinson (Goodwin et al., 2016: 5;Pilkington, 2016: 40-41) as important determinants of the group's fortunes.
However, demand-side explanations may struggle to explain why some people with particular socioeconomic characteristics become involved in far-right activism while others do not. Demand-side approaches may therefore treat extreme right success as 'a passive consequence of macro-level socioeconomic developments' with little regard to the agency of leaders or members (Mudde, 2007: 4). In addition, in analysing supply-side factors, it is important to identify the relevant institutional factors, not their consequences. Hence, we believe that previous explanations of the EDL's failure have wrongly attributed causal weight to symptoms of the group's decline, whereas the real explanation concerns the failure of the group to put in place appropriate self-governance mechanisms.
We concur with Norris (2005) that a satisfactory explanation of far-right electoral support or, in our case, activism must understand the interaction of supply and demand. A recent study of the initial success and subsequent failure of the Ku Klux Klan in the 1920s shows how supply and demand interactions may determine the fortunes of a far-right organisation. Fryer and Levitt (2012) identify the tremendous demand for hatred, religious intolerance and, also, fraternal membership in the US at this time. Klan membership peaked at around four million people in 1924, but following the conviction of a high-profile Klan member for rape and murder, membership declined to approximately 1% of its peak by 1930. The authors argue that this volatility can be explained by the readiness of the Klan to supply membership to any supporter who demanded it: many Klan members were recruited in the years immediately prior to the conviction by sales agents who received financial rewards for recruiting new members, regardless of their loyalty to the group. Many of these recruits also belonged to other fraternal organisations that offered an alternative source of many of the benefits of Klan membership. The Klan's failure to put in place mechanisms to ensure the loyalty of its recruits resulted in a membership base with low levels of attachment who exited the Klan following the high-profile conviction. Fryer and Levitt use the term 'marginal members' to identity members with low levels of attachment who will exit a group when the utility associated with their membership is reduced. Our study of the EDL's rise and decline similarly identified marginal members as an important factor in the organisation's collapse.
This article explains the rise and fall of the EDL by uniting supply and demand factors -we identify the supply-side factors that meant the EDL was initially able, but then became unable, to meet the demands of its members. This explanation is grounded in the theory of collective action set out by Olson (1965), Tullock (1971) and Ostrom (1998), that identifies a fundamental problem of politics that the benefits of political action accrue to both participants and non-participants, whereas the costs are borne by participants alone, so it is in each individual's self-interest to freeride on the activism of others. It is generally understood that successful groups solve this collective action problem by providing benefits exclusively to participants that act as selective incentives to motivate participation. These benefits are supplied as club goods -goods that are neither purely public nor private but can be supplied exclusively to many people -and may include welfare, entertainment, friendship networks and access to public office (Buchanan, 1965;Olson, 1965).
We seek to explain far-right activism using the theory of collective action and use the case of far-right activism to test the theory of collective action. Specifically, we test the hypothesis that far-right activism faces an inherent collective action problem and that significant features of the organisation, operation and fortunes of far-right groups are explained by the way that they solve this problem.

Methodology
Our data were principally collected from ethnographic fieldwork and formal and informal interviews. Between February 2013 and June 2016, we attended a total of 12 demonstrations held by the EDL and Pegida UK (the anti-Islam protest group briefly led by Robinson after his departure from the EDL) in English towns and cities, including London, Manchester, Newcastle and Rotherham. Through attending EDL demonstrations, a rapport was developed with regional and divisional leaders, which resulted in invitations to two 'meet-and-greet' sessions for new and prospective EDL members, social gatherings at pubs and membership of a closed EDL Facebook group. During fieldwork, informal, unstructured interviews were undertaken with 17 activists and many more were engaged in short conversations about their activism. Eight of these activists were interviewed on more than one occasion, enabling their views to be tracked over time. Field notes -including details of informal interviews -were recorded in situ or immediately after the event. Fieldwork was supplemented by semi-structured, formal interviews with key informants, in which five participants, including EDL founder and original leader Tommy Robinson, were approached in 2016 and invited to reflect on their experience as activists. These formal interviews were shaped by a question guide that addressed entry into the group, reasons for maintaining or ceasing activism and perceived reasons for the EDL's decline, while also allowing flexibility to discuss other issues that arose. As is common in much field research, the sample recruited was nonprobabilistic and purposive, not aiming for statistical representativeness, but nevertheless seeking to select respondents with a range of sociological characteristics; our respondents ranged from 16 to 54 years of age; were based in the Greater London, the Midlands and Norfolk; and comprised 15 men and 5 women. In total, approximately 70 hours were spent directly interacting with present or former EDL activists in the form of participant observation at events and formal and informal interviews, and 36 interviews were conducted with 20 individuals. At this point, our field notes and interviews began to replicate the same data, indicating that saturation point had been reached. This saturation point is consistent with that found by other qualitative studies (Guest et al., 2006).
Ethical approval for the research (participant observation and formal and informal interviews) was obtained from our university Research Ethics Panel prior to commencement. All activists involved in direct interactions during fieldwork were aware of the researchers' status. However, the largely anonymous nature of demonstrations -and the crowds they attracted -meant that it was impossible for the researchers to identify themselves to every activist.
Our data analysis involved a process of descriptive inference from data that principally comprised field notes of observations and informal interviews and transcripts of formal interviews. We also had transcripts of formal speeches, EDL mission statements and the group's public and private social media output. To facilitate our search for observable implications and to make sense of this large quantity of data -field notes and transcripts alone amounted to 86,000 words -we used a deductively and inductively derived coding framework. It is widely accepted that the coding process will be informed by prior theoretical interests, and researchers should explicitly acknowledge these when recounting their approach to data analysis and transparently record their coding procedures to boost the reliability of their research (Benaquisto, 2008;Layder, 1998;Merriam and Tisdell, 2016;Saldana, 2009). We first deductively identified the concepts that reflected our hypothesis that far-right activism faces an inherent collective action problem and that significant features of the organisation, operation and fortunes of far-right groups are explained by the way that they solve this problem: selective incentives, operational features, organisational features, rise and decline. We then turned to our data and identified segments relevant to these concepts (and the hypothesis being tested) and coded these segments with a word or short description that captured the salient features of the text or image, such as, 'drug usage', 'violence', 'war analogy', 'Muslim grooming gangs', 'loyalty', 'in-fighting' and 'indiscriminate recruitment' (see Saldana, 2009: 14). Some of these labels came from us (e.g. 'war analogy') whereas others represented the language used by our respondents (e.g. 'Muslim grooming gang'). We then grouped like segments together based on patterns of similarity. These grouped patterns became our inductively derived categories. For example, we identified a similarity between 'war analogies' and 'Muslim grooming gangs' because the coded data referred to the purported protection of country and family by EDL members. This pattern was used to develop the inductive category of increased self-worth that, in turn, we understood to be a selective incentive and an observable implication of our hypothesis. Similarly, we identified a resemblance between 'dual Facebook accounts' and 'face coverings' because the coded data reflected an avoidance of stigmatising behaviour. This commonality enabled us to develop the inductive category of commitment mechanisms, an organisational feature of the EDL that we believe reflected its approach to solving the collective action problem. As is often the case with qualitative research, our data analysis was an iterative process that was at times both inductive and deductive (Merriam and Tisdell, 2016: 202).
We also considered alternative explanations for the rapid collapse of the EDL and purposefully looked for data that supported these explanations. Our alertness to the possibility that data may be inconsistent with our theory, and our consideration of alternative explanations, meant that we reduced the risk that our theoretical perspective enforced 'a pre-determined explanatory grid on the data' and increased the validity of our research (Layder, 1998: 113;Merriam and Tisdell, 2016: 248).
The great strength of qualitative research is the depth and richness of the data that can be generated and the insights thereby offered into the subjective perceptions of individuals. While quantitative research can provide useful information about the socioeconomic profile of far-right activists (e.g. Goodwin et al., 2016) and textual analysis of far-right publications can generate important data about their public ideology (e.g. Kassimeris and Jackson, 2015), qualitative research generates data about the subjective, personal experiences of activists. However, a notable feature of our data was the consistency of responses. While personal circumstances or characteristics differed, every activist gave an account of the direct, personal benefits of EDL activism consistent with the theory of collective action.
Descriptive inference can be made with some, but never complete, certainty. First, the process of inference involves judgements about what data are important or not. Second, new data may be discovered that do not fit the theory. Nevertheless, we do not believe there are observations within our data that are inconsistent with our conclusions. In this article, we consider alternative explanations and we welcome attempts by other scholars to replicate our results from our data or to show that our data more accurately fits an alternative theory.
We recognise that a single case study cannot claim generalizability. However, we trust that in having provided a detailed account of our fieldwork, the descriptive data it generated and our data analysis processes, our study's concepts and findings can be used by other researchers who wish to conduct a comparison between the trajectory of the EDL and other far-right organisations.

The EDL: Rise and fall
The EDL was formed in Luton in March 2009 following a provocative demonstration by the Islamist group al-Muhajiroun. The organisation's mission statement (EDL, 2016) suggests that the EDL engages in peaceful and non-violent activities as part of its 'struggle against global Islamification'. Although it claims to want to utilise the courts and legislature to protect human rights, democracy and the traditions of England, the EDL favours direct action in the form of confrontational demonstrations. Our own and other studies have found EDL members to be predominantly White, male and working-class -like other far-right parties and organisations (Bartlett and Littler, 2011;Busher, 2016;Kassimeris and Jackson, 2015;Meadowcroft and Morrow, 2017;Pilkington, 2016;Treadwell and Garland, 2011).
However, whereas far-right groups such as the National Front and the BNP advocated a White supremacist ideology, the EDL does not espouse biological racism. Instead, the leadership of the EDL insists it is not a racist organisation. Indeed, one of the EDL's first 'stunts' was to invite journalists to film members burning a swastika flag (Kassimeris and Jackson, 2015: 172). While the EDL has denied racism, and this denial may be genuine, the group nevertheless promotes 'a culturally racist discourse of Islamophobia' that 'involves the demarcation of an in-group and an out-group, where the in-group considers itself superior' (Kassimeris and Jackson, 2015: 184). In this sense, its rejection of pluralism and minority rights means it can properly be considered 'far right' (Mudde, 2014: 98). We will show that the EDL's formal rejection of biological racism was an important factor in its initial success but that this rejection became compromised.
EDL membership is a fluid concept. Busher (2016: 6) noted that 'among activists, there were differences of opinion over what it meant to be part of the group'. Bartlett and Littler's (2011: 28-29) survey of online EDL supporters found that 88% of respondents who had attended a demonstration self-identified as members. We saw a similar view expressed by the administrator of a closed Facebook group following a demonstration in Bradford in October 2013: 'Ty for those of u who attended yesterday n those of u who didnt shame on u as ur not true edl members' (emphasis added). Herein, we assume that a member is someone who regularly attends EDL demonstrations.
The EDL quickly grew into a national movement capable of drawing significant numbers to its protests. Figure 1 shows the numbers of participants at EDL demonstrations from October 2009 to April 2015. Initially, EDL protests were well-attended, with more than 2000 protestors regularly present. However, from late-2011, there was a marked decline in numbers that continued throughout the next 18 months, with some protests The murder of British soldier Lee Rigby by Islamic extremists in Woolwich, South London, on 22 May 2013 produced a sudden upturn in the EDL's fortunes. An (alreadyplanned) demonstration in Newcastle 3 days after the murder was attended by 1500 people. The two subsequent demonstrations attracted similar numbers -1000 people were present in London on 27 May 2013 and 2000 protestors were in Birmingham in July 2013. We discuss below why an event like the murder of Lee Rigby can have a dramatic impact on people's willingness to engage in activism, albeit for a limited time, but in the present context, the salient point is that the increased numbers were short-lived.
By late-2013, it was clear the EDL could no longer attract thousands of people to its demonstrations. Its original leaders, Tommy Robinson and his cousin Kevin Carroll, resigned from the organisation in October 2013, claiming that they could 'no longer keep extremist elements at bay' (Quilliam Foundation, 2013). In 2015, Robinson became the leader of Pegida UK, the British counterpart of the German anti-Islamist protest movement. But this organisation failed to replicate the EDL's early success -only 150 people joined the group's demonstration in Rotherham in June 2016, for example -and it appears to have been defunct since 2017.
The conundrum that faced the EDL was that while many people shared the group's ideological concerns -for example, a YouGov (2015) survey showed that over half of UK voters thought there was a fundamental clash between Islam and British values -for each of those individuals, it was nevertheless rational to freeride on the participation of others given that any public goods that the group supplied (e.g. influence over government policy or increased media attention on the issues members considered important) would be provided to participants and non-participants alike. Meadowcroft and Morrow (2017) have shown that the EDL solved the collective action problem by supplying the benefits of access to violence, group solidarity and increased self-worth, and that these benefits outweighed the costs of time, money, unwanted police attention and stigma for sufficient people to enable the group to hold well-attended demonstrations. This article develops and extends Meadowcroft and Morrow's analysis by identifying the importance of 'participatory crowding' in the EDL's solution of the collective action problem and explaining the group's decline, as well as its rise.

Collective Action, Participatory Crowding and Marginal Members
In a classic contribution to the economics of religion, Iannaccone (1992) observed that the satisfaction an individual derives from participation in organised religious activities will depend both on his or her own inputs and also the inputs of others; many facets of religious participation such as hymn singing, communion and even speaking in tongues are more satisfying and enjoyable when experienced collectively. This phenomenonthat individual enjoyment is increased by the participation of others -is known as 'participatory crowding'.
However, Iannaccone (1992) argued that participatory crowding could lead to a particular form of freeriding: ostensible participants could freeride on the greater contribution of more committed members. Iannaccone (1992: 281) argued that this was problematic because '[l]ess committed members threaten to swamp groups that would otherwise have high levels of participation' and, if this happened, the overall quality of club goods such as warm greetings, tuneful singing and religious enthusiasm would decline, threatening an exodus of members that might challenge the group's viability.
Moreover, relating this analysis to Fryer and Levitt's (2012) study of the Klan, lesscommitted, or, as the authors put it, marginal members may undermine collective action if they leave a group in response to a change to the internal or external environment and thereby compromise the group's ability to supply benefits to its remaining members. If we take seriously the injunction from behavioural economics that people weigh losses more heavily than gains -the endowment effect (Kahneman et al., 1990) -then it follows that the exit of marginal members harms a group more than if they had not joined at all. Being part of a small organisation that was once large is surely less satisfying than being a member of a small organisation that was always small.
However, collectives can establish mechanisms to prevent the recruitment of marginal members. Iannaccone (1992) and Iannaccone and Berman (2006) suggest that organisations can require members to engage in stigmatising behaviour that will inhibit participation in alternative contexts and penalise members for participating in non-collective activities.
In this article, we argue that the club goods supplied by EDL participation were enhanced through participatory crowding -the more participants in attendance at demonstrations, the greater the supply of club goods. Like Iannaccone's collective, the EDL was also vulnerable to attracting members without a high level of commitment to the organisation but did not implement the measures identified by Iannaccone and Berman to boost levels of commitment. Consequently, EDL numbers swelled into the thousands, but many of these members had low levels of commitment which left the organisation vulnerable to rapid collapse.

Participatory Crowding and the Mobilisation of the EDL
We found no evidence that EDL leaders (or members) had a coherent plan of how the group's street-based protests would achieve its stated goals. Similarly, Pilkington (2016: 38) observed that while 'awareness raising' was an ostensible reason for the EDL's demonstrations, in reality, 'there was little discussion or knowledge of the particular issue about which awareness was being raised in advance of the national demos'. Despite the low efficacy of street-based protests, mass mobilisation was nevertheless important to the EDL. Busher (2016: 128) observed that 'EDL activists … were all acutely aware that the success of the EDL, whatever that might look like, depended to a large extent on the number of people it was able to attract to its events' (emphasis added). Similarly, in interview, Robinson maintained that if the EDL had presented itself as a traditional right-wing extremist group, 'it wouldn't have worked, it wouldn't have got big numbers …' Although activists often lacked knowledge about the issues that they were purportedly trying to raise awareness of and rarely thought their participation would effect change, we believe mass mobilisation was nevertheless important to the EDL because the effects of participatory crowding enhanced the three key benefits supplied to its members: access to violence, increased self-worth and group solidarity.
Previous studies establish that the opportunities for violent conflict provided by EDL demonstrations played an important role in attracting members, particularly those with a football hooligan background (Busher, 2016;Pilkington, 2016;Treadwell and Garland, 2011). These findings cohere with our study. During fieldwork, a young male member told a meet-and-greet audience that being in the 'lion's den' of counter-protesters was 'the best part' of EDL activism, and members were seen using drugs before aggressively taunting law enforcement and counter-protesters. In interview, Robinson described the EDL's hooligan members as men 'who will fight, and want to fight', and Darren, a Leicester-based factory worker, admitted that demonstration violence 'got me hooked'.
Dynamics of group membership may increase aggressive acts and violence, partly due to the perception of personal anonymity provided by a crowd and partly because it is more difficult for the police to impose sanctions on individuals in a large group (Drury and Reicher, 2009). Robinson understood the benefits conferred by crowd behaviour in facilitating violent conflict from his years as a member of the Luton Town football hooligan firm. In interview he revealed that he used similar tactics to mobilise EDL supporters and football hooligans. Robinson appeared to understand the importance of anonymity and how large crowds could impede the police in penalising offenders, when explaining why he supplied balaclavas to participants at a chaotic demonstration in 2009: 'If [the police] see one person wearing a balaclava they're stopping him. If they see everyone wearing balaclavas, they ain't stopping them'.
EDL participation also increased its members' self-worth, something understood by Lamont (2000) in her study of working-class men to include dignity affirmation, selfexpression, upholding moral codes and displaying competence. Meadowcroft and Morrow (2017) suggest that EDL participation bolstered members' self-worth by purportedly giving them an opportunity to protect their families and country from radical Islam's alleged threat. We would add that EDL participation bestowed an additional sense of self-worth on its primarily White, working-class members by providing them with an opportunity to celebrate their identity. This was recognised by Robinson in interview: You have to understand that in Luton my whole life as White English I'm a minority in town there's never been anything that celebrates who you are … St Lucian Day … we have a massive festival, yeah? Paddy's Day, massive festival; Eid we have fairground rides, massive, massive celebrations, so, there's a void to be filled … English kids are already walking with their heads down, like they should be ashamed of who they are … Now the English Defence League filled this massive void for so many people who come and just said, 'yes, fuck it, this is who I am, this is filling my fucking life' (emphasis added).
Attracting large numbers to demonstrations would therefore enhance participants' feelings of pride and celebration to a greater degree than if the EDL had attracted fewer participants.
Finally, EDL participation provided its members with solidarity: unity, oneness of purpose and group identification (Wintrobe, 2006: 26). Previous studies have shown that EDL members likened their participation to being part of a family and that feelings of unity motivated activism (Pilkington, 2016: 179-180;Meadowcroft and Morrow, 2017). Self-worth and solidarity may of course overlap: Lamont's (2000: 49-50) study suggests that racial solidarity can be a source of self-worth, and this view is reinforced by Robinson's analysis of why EDL participation was uniquely appealing to its White members.
The role of participatory crowding in supplying group solidarity may be more complex than its role in facilitating violence and increased self-worth because a large group may ostensibly offer less solidarity than a small and insular organisation. But it seems clear that members did experience a strong sense of solidarity when attending large EDL events. The emotional benefit of participating within a crowd was recognised by Davie, an older EDL member with a history of political activism in mainstream parties: I'm quite a loner but it does feel good to be with a lot of people who have a common cause and I think the one time I did have tears in my eyes was when we was marching, I think it was behind the Houses of Parliament, you could see the England flag, you could see Big Ben … I did feel a little tearful at that point (emphasis added).
The experience of being one of many like-minded individuals sharing a common experience elicited solidarity which was amplified through participatory crowding. When reflecting on the speeches given by EDL leaders, including Robinson, Darren stated: The adrenaline and the buzz that you used to feel on these speeches was different. It was like a sensation you've never felt. It's weird, it was spine-tingling … because you're not on your own then, you understand? You're surrounded by people with like-minded mentality (emphasis added).
We contend that the amplification of solidarity by participatory crowding can also explain the EDL's short-lived resurgence following the murder of Lee Rigby. As noted, 1500 people attended an EDL demonstration in Newcastle 3 days after the murder -the largest EDL event for almost 2 years. But within a few months, EDL participation returned to the levels prior to the murder.
People may benefit from participatory crowding following dramatic events like the murder of Lee Rigby because emotions of shock and anger are more powerfully expressed collectively -they do not take to the streets alone, but rather seek solidarity with others. In the case of the murder of Lee Rigby, the already-planned EDL demonstration targeting Islamic extremism provided an obvious forum for people to come together. Fieldwork conducted at the demonstration found that many of these additional participants were not EDL members who had returned to the fold, but people from outside the EDL who used this event to obtain solidarity enhanced by participatory crowding. Once their immediate emotional reaction had faded and solidarity was no longer required, these people did not continue their EDL involvement. A similar view was put by Darren in interview when asked why this surge in EDL activism was not sustained: I think it was like an instant reaction … Whether they believed in the organisation or not, there was people there that day who didn't believe in the organisation, but that was the only platform … once things died down, it don't affect them no more, and it goes to the back of their mind.
Consequently, the spike in EDL numbers was not sustained.

The Problem of Marginal Members
When collective action is motivated by club goods supplied to participants, then the immediate benefits of activism may become more important than the long-term strategic goals that activism is supposed to achieve. Mobilisation may become an end in itself, so that, as Jasper (2011: 296) wrote, 'The satisfactions of action … [may] become a motivation every bit as important as a movement's stated goals'. Indeed, we believe that for many EDL participants, the immediate benefits of activism enhanced by mass mobilisation became more important than any long-term goals.
As established above, one of the immediate benefits of EDL activism -access to violent conflict -was particularly salient for the organisation's football hooligans. Indeed, an important factor in the early success of the EDL was its ability to recruit from hooligan networks, as Robinson revealed in interview: 'I was going on to football messaging boards … I was going on all of them … saying "look at this, look at this, look at this"'. For many of these hooligans, the EDL served the purpose of providing new opportunities for fighting. Pilkington (2016: 79) similarly found evidence that, for some hooligans, EDL demonstrations replaced football violence, with one respondent revealing, 'just as I got banned from football the EDL started so guys that I knew that were banned as well says "come along Rob, it's the same sort of thing"'.
However, the EDL's ability to supply violence was compromised by changed policing strategies. Treadwell's (2014: 128) analysis shows that the policing of the EDL was initially heavy-handed and reactive; the challenge of keeping the peace between the EDL and counter-demonstrators led to public disorder and a high number of arrests. Then, policing became 'premised largely on preventing any real contact between rival groups' and reducing arrests and disorder at demonstrations. A number of demonstrations from 2009 to 2011 resulted in EDL members breaking through police lines, throwing projectiles and engaging in street fights, but by 2012, this disorder had diminished as the second policing strategy took hold. The police more effectively kept EDL and counter-demonstrations apart, and routed protests away from potential flashpoints such as Islamic centres, thereby reducing the potential for violent conflict (Treadwell, 2014: 134-135).
At interview, Robinson explained that the numbers at EDL demonstrations declined from 2011 because: The police were successful in what they were doing, the days become quite boring, intentionallywise … Tower Hamlets they held everyone in the road for six hours -it's boring as fuck. Do you think I've just travelled down from Newcastle to stand in the road for six hours without having beer?
Similarly, Darren stated that police action that made EDL protests uneventful with little prospect of violence had discouraged attendance: … you'd have the lads who'd ring me and say, 'do you think it's going to kick-off?' And dependent on my answer would be whether they'd turn up or not … that's why the numbers dwindled, because people was just going there, the Old Bill was kettling us in, there was no trouble, there was no fighting, and that was it.
The EDL's football hooligans were particularly vulnerable to assuming the role of marginal members and deserting the organisation en masse because they retained their ties to another organisation -football firms -that they could return to once EDL activism ceased to be enjoyable. In interview, Samantha, a Wolverhampton member, suggested that the EDL's hooligans returned to their football roots once activism became unappealing: '… you just don't get numbers, they're not as exciting as what they first were, so they've gone back to the football …' Pilkington (2016: 183) similarly found that some hooligans returned to their football networks once EDL violence reduced, with one respondent stating: '… you come to the EDL, you aren't getting no violence … But if you go to a football match, you've got 50 times more chance of having a kick-off …' The modification of police tactics reduced violence and made EDL events boring, thereby leading to an exodus of marginal members -a significant proportion of the EDL given that Busher (2016: 38) estimated hooligans comprised between 30% and 40% of members. The changed police tactics may be understood as an exogenous shock that reduced a key benefit for many members.
The exit of these marginal members was significant to the EDL's downfall for three reasons. First, it reduced the participatory crowding essential to the consumption of the other club goods supplied. As the organisation became smaller, the benefits of self-worth and solidarity for those who remained were reduced.
Second, because the organisation measured its success by the numbers in attendance at its events, the decline in participation convinced members that the group was failing and this led to internal recriminations that further undermined the supply of self-worth and solidarity. Indeed, many respondents identified in-fighting or 'backstabbing' as a principal reason for the group's collapse. In-fighting made EDL membership less enjoyable.
Third, the exit of marginal members increased the relative salience of the EDL's biological racists who adhered to a White supremacist ideology. There is evidence from our own research, and that of Busher (2016) and Pilkington (2016), that many EDL activists had a history of activism in overtly racist organisations, such as the National Front and BNP. These biological racists had two negative effects on the organisation. First, as Robinson explained in interview, some of these members formed 'inner circles' within the EDL and tried to attract others to their White supremacist sub-groups. Robinson recounted that the presence of biological racists 'became such a big burden on me, constantly battling them' and precipitated his decision to resign.
Second, biologically racist members reduced the self-worth and group solidarity other members derived from their participation. We found evidence that members had withdrawn from the EDL because of the presence of open racists. As Davie recounted: Things started to get a bit sinister after the [2013] Newcastle demo. I went to a meet-and-greet in Huntingdon and there were a bunch of people in the middle of the pub doing Hitler salutes to the time of the song, you know, the full thing, from the chest up, which is something I'd never seen before … I left [the EDL] after that. I was quite disgusted.
Someone who did not hold racist views could not obtain self-worth or group solidarity from participation in an organisation in which people openly espoused biological racism and used Nazi symbolism. Moreover, given that one of the most important costs associated with EDL membership was stigma -particularly, the stigma of belonging to what many outside the EDL believed was a racist organisation -the presence of open racism increased the costs of EDL activism. Marginal members may have effectively masked the number of committed racists within the EDL, so that their exit left an organisation with a significant proportion of racist members.
When these negative consequences are placed in the context of the endowment effect noted above, the transformation of the EDL from what members believed was a non-racist mass movement to a small and fractious group riddled with racists would have had a devastating effect on people's willingness to participate and hence the group's fortunes.

A Failure of Self-Governance
The recruitment of marginal members within a political organisation is not inevitable; rather, organisational failures left the EDL particularly vulnerable to such recruitment. Iannaccone (1992) and Iannaccone and Berman (2006) contend that collectives can impose requirements of deviant behaviour and penalise activities that compete for member resources, as mechanisms to ensure commitment. When implemented effectively, these mechanisms can increase the demand for internal substitutes and the utility of group membership. We consider each mechanism, and the EDL's failure to implement it, in turn. Iannaccone (1992) suggests that bizarre dress standards -such as shaved heads or robes -can invite scorn and stigma on a group member, thereby inhibiting participation in other, more mainstream, contexts. Although Iannaccone's examples are derived from religion, there is evidence that some far-right groups make similar demands of members: studies into neo-Nazism show that being labelled as a 'deviant' may be a group joining requirement, with members being required to appear in public wearing a 'nationalist' uniform. This leads to isolation and disassociation from existing social networks, thereby boosting reliance on -and commitment to -the organisation (Blee, 2009;Hundeide, 2003: 110-112).
Conversely, the EDL did not enforce dress codes or require behaviour that would deviate from standard social norms. While stigma is a cost of EDL participation, members could reduce it. Numerous participants at EDL demonstrations wore face coverings, such as balaclavas or pig masks, to preserve their anonymity. Indeed, Robinson sought to hide his identity by wearing balaclavas at early protests and adopting a pseudonym. During a national demonstration in Manchester, members sought to avoid counter-demonstrators who were trying to photograph them because, as Samantha explained, 'some members are trying to keep their identities hidden'. Similarly, a divisional Facebook administrator expressed concern that she would not be able to attend a local demonstration because it would be at her place of work -a university where she supervised cafeteria workers. Other members revealed that they had two Facebook accounts: one used for EDL activities and another for general use.
Far from imposing stigmatising behavioural standards, the EDL allowed members to hide their organisational affiliation and avoid the stigma that may otherwise follow, thereby differentiating it from other far-right organisations and Iannaccone and Berman's collectives. That the EDL was willing to shield members from stigma also separates it from Iannaccone's (1992: 276) collectives where '[p]otential members are forced to choose: participate fully or not at all'. Had the EDL required more stigmatising behaviour of its members -for example, banning face coverings, imposing a uniform or requiring members to share EDL content on social media -it may have screened out marginal members and boosted commitment levels among members willing to adhere to the organisation's demands. Of course, this would almost certainly have prevented the EDL from attracting large numbers of people to its early demonstrations.
The other mechanism identified by Iannaccone (1992) and Iannaccone and Berman (2006) to guarantee commitment to the collective is to prohibit or penalise activities that compete for member resources, such as friendships within secular society. A similar process has been observed in some far-right organisations: Bjorgo (2002: 10-14) notes that the process of joining a neo-Nazi group often involves reducing contact with family and friends; if a neo-Nazi's relationships with friends and family are harmed through his or her participation in the group, then that individual may find it difficult to sever organisational ties.
Conversely, EDL members were not required to reduce pre-existing social connections to non-EDL friends and family, and to football firms. We never observed any attempts made by EDL leaders to sanction external socialising, and indeed, members openly referred to friends or spouses who were not part of the organisation. Similarly, football hooligan firm involvement co-existed alongside EDL participation for many members, with Busher (2016: 137) noting that hooligan members tended to identify, and be identified, as 'football lads'. Robinson additionally revealed at interview that it was easy for him to leave the EDL because he had retained his ties with the Luton Town football firm, explaining that, 'if you take the EDL away I've still got a massive social circle through football really'. The EDL often did not become the centre of a member's social life, thereby distinguishing it from Iannaccone's collectives and other, more restrictive, farright organisations.
Fieldwork suggested that members whose outside relationships were harmed by EDL participation were more long-standing. For example, during a conversation in a pub in April 2013, one Leicester-based bouncer who remained part of his division as numbers declined recounted that when he publicly declared his EDL involvement, he lost friends and people avoided him; ultimately his social circle principally consisted of other EDL members. Another long-term member admitted that his mother had not spoken to him for 4 years because of his EDL involvement. For these members, it was difficult to return to their old social circles or find new social connections outside the EDL. Therefore, they remained loyal to the organisation.
Although the EDL could have sought to guard against marginal members by prohibiting activities that competed for member resources, doing so would have increased the costs of participation and alienated many of the EDL's initial members, especially those drawn from football hooligan networks. Like stigmatising behaviour, this would surely have limited its initial ability to solve the collective action problem and produced a smaller organisation.
Finally, even if the EDL had imposed the measures contemplated by Iannaccone and Berman, the organisation may have still been powerless to stop the infiltration by biological racists. This is because the previous studies considered herein suggest that individuals who participate in White supremacist organisations are willing to comply with stigmatising and alienating requirements. To prevent the attendance of biological racists, the EDL could have sought to implement some of the mechanisms for radical right longevity contemplated by De Lange and Art (2011), particularly, the careful vetting of activists. However, to do so would have had two consequences. First, the vetting of activists would have imposed an administrative burden on EDL leaders. Former Regional Organiser Luke admitted in interview that no rigorous attempts were made to screen prospective members because it was 'hard work' looking at their Facebook profile. Second, the EDL would have needed to make attendance at demonstrations invite-only; the open nature of its events meant that anyone -including biological racists -could attend, with Luke noting: '… you can't stop people going, that's the trouble, 'cos it's an open event and anybody can turn up. You used to … throw the odd person out but you'd never see 'em in all them people'. If all activists had been vetted and attendance at demonstrations made invite-only, then it is likely that the EDL could have prevented infiltration, but these measures would also have hindered its capacity for mass mobilisation.
The collapse of the EDL was caused by the internal supply-side factor of weak selfgovernance that meant it became unable to meet member demands. The organisation expanded rapidly after its launch in 2009, relying on the intuitive understanding of how to recruit members held by its leading figures. However, this fast, almost spontaneous, growth meant that the organisation did not have the governance structures necessary to ensure its long-term sustainability. Of course, if such mechanisms had been in place, then the organisation would almost certainly not have grown so rapidly. Hence, the basis of the organisation's early success was also the cause of its eventual demise.
When Robinson launched Pegida UK in 2015, he sought to implement some of the self-governance mechanisms suggested above, a decision we assume was informed by his experience of the failure of the EDL. Face coverings were banned, ostensibly so participants did not appear thuggish, but this also meant that participants could not avoid the possibility of being recognised. Importantly, Pegida UK held silent demonstrations with its first two events staged at a Birmingham industrial estate, thereby signalling the unlikelihood of violent confrontation. In interview, Robinson stated that Pegida UK demonstrations were 'silent, no chanting, teetotal, so people understand … I'm not coming for a fight. If we wanna come for a fight … we wouldn't be marching in the middle of nowhere …' As the account set out here would lead one to expect, Pegida UK's uneventful, peaceful demonstrations attracted relatively few participants and a silent march through Rotherham in 2016 was its last public event.

Alternative Explanations
As noted in the introduction, two principal alternative explanations for the rise and fall of the EDL have been proposed. First, the leadership of Tommy Robinson has been cited as an important determinant of the group's fortunes (Goodwin et al., 2016: 5;Pilkington, 2016: 40-41). Here, the resignation of Robinson is understood to have humiliated and demoralised supporters, and triggered the fragmentation of the group. We do not believe the evidence supports this explanation of the EDL's fortunes. As discussed above, at interview, Robinson described the mounting problems he encountered running the organisation, which he came to view as insurmountable, convincing him that the group had little long-term future and precipitating his resignation. We therefore believe Robinson's resignation from the EDL was a symptom, not a cause, of its failure. In addition, the decline in EDL numbers began in late-2011, 2 years before Robinson resigned. Although Robinson was a popular figure, we found no evidence that he commanded loyalty or devotion from members. Discussing Robinson's role in the EDL, Davie explained, 'I like listening to him because he expresses himself well, but I don't feel drawn to him', and that even if Robinson had remained leader, 'after I saw the people doing the Hitler salute in the pub I would have left [anyway]'. It is also salient that Robinson's post-EDL venture, Pegida UK, did not replicate the EDL's success, despite his presence at its head, suggesting he alone was not a powerful draw for activists.
Second, it has been suggested that the EDL collapsed because of in-fighting and 'factioning' that created tension within the group and led to the formation of White supremacist splinter organisations such as the North-West Infidels (Busher, 2016: Chapter 5;Pilkington, 2016: 54-55). Indeed, a number of respondents attributed the decline of the EDL to in-fighting and 'backstabbing'. However, our research suggests that weak selfgovernance encouraged these conflicts; Robinson's powerlessness to prevent rival groups affiliating themselves with the EDL and seeking to recruit their members to its ranks created factions. It is also clear that in-fighting became a significant problem as the organisation failed. Respondents repeatedly described recriminations taking place after poorly attended demonstrations, with members accusing one another and the leadership of not doing what was required to make the organisation successful. While the decline of the organisation was accompanied by in-fighting and factioning, we believe the evidence shows this too was a symptom rather than a cause. In-fighting may have accelerated the process of decline because it made membership less enjoyable, but at root in-fighting was caused by weak self-governance mechanisms and was a response to the decline of the organisation as members blamed one another and the leadership as the organisation faltered.

Conclusion
This article has used the theory of collective action to understand the rise and subsequent decline of the EDL using new data that open the 'black box' of a far-right organisation. This approach has enabled the identification of the relevant demand-and supply-side factors: The EDL supplied benefits that satisfied activists' demands for violence, self-worth and solidarity and were enhanced by participatory crowding. This enabled the group to not only grow quickly and put thousands of people on the streets but also fuelled indiscriminate recruitment that increased the presence of marginal members with low levels of commitment whose subsequent exit reduced participatory crowding and increased the relative salience of biological racists within the organisation.
The EDL developed quickly and relatively spontaneously, driven by its leading members' intuitive grasp of collective action dynamics but consequently did not have in place the self-governance mechanisms that might have ensured the commitment of members and screened out biological racists. Our analysis suggests that if far-right organisations do not establish effective self-governance mechanisms then their long-term viability will be compromised.