and qualitative approach (Ragin 1994). It is in many respects ethnographic in character, in that it necessitates a ¶multi-‐method qualitative approach· in order to achieve its aims and in large part seeks to ¶XQGHUVWDQGWKHLQVLGHUV·YLHZRI their own world· (Flick 2007 p149). It broadly follows the approach described by Angrosino as ¶the collection of information about the material products, social relationships, beliefs and values of a community· (2007 pXV) using ethnographic methods that ¶have been adapted by scholars from many academic disciplines and professional fields· (Angrosino 2007 p19). The nature of this study is such that it
32 ZDV DQWLFLSDWHG WKDW WKH UHVHDUFK PHWKRGRORJ\ ZRXOG EH D ¶OLYLQJ· SURFHVV WKDW needed to ¶discover a ´methodologyµ for itself· (Clough + Nutbrown 2007 p xi) and so consequently some refinement of the research plan was expected and indeed welcomed when changes were necessary. In order to further explain the research methodology it is necessary to justify the choice oIWKH¶)ODVKSRLQWV·PRGHO
:K\¶)ODVKSRLQWV·
As detailed in the literature review in the next chapter, there are a number of theories that could be applied to any given public order or disorder scenario. It has also been VKRZQ WKDW WKH ¶)ODVKSRLQWV· PRGHO has been successfully used to analyse a number of different types of disorder events including urban disorders, football violence and single issue protest (Waddington D 2007). The six interdependent levels specified by the Flashpoints model, assessing the structural, political/ideological, cultural, contextual, situational and interactional factors behind such incidents, provide an effective mechanism for FDSWXULQJPDQ\RIWKH¶VXUURXQGLQJ·FDXVDOIDFWRUVWKDWQHHGWREHDFFRXQWHGIRU in explaining how and why a particular public order incident occurred.
In its original form, the Flashpoints model gives little consideration to police traditions, culture or doctrines. As such, it was necessary to add another level of structuration to take account of this area. David Waddington has suggested that an additional seventh level would enhance the model:
focusing on such recognised influences on police behaviour as national traditions of policing, formal systems of accountability and the particular philosophies, ¶PLVVLRQV· RU VWDQGDUG RSHUDWLQJ SURFHGXUHV JRYHUQLQJ aspects of strategy and tactics in relation to public order situations (Waddington D 2010 p346).
The areas he lists here as integral to this additional level of analysis amount to doctrine. For British public order policing, this is set out in the ACPO manual of Guidance on Keeping the Peace (KtP 2007) the first edition of which was published
33 in 2001. As will be discussed in more detail in due course, inclusion of this seventh level provides better contextual understanding of police agency at the ¶LQWHUDFWLRQDO·OHYHORIWKHPRGHO
Hence there are seven levels of analysis available for consideration which, when addressed with due regard to the advantages of understanding the issue of identities in crowd management (Reicher 1987; Adang 2002; Stott 2005), add meaning to this study.
The broad theoretical framework sketched out above, has ¶influenced the design, orientation and character· of this study (Cooper 2008 p9). Having selected a predominantly qualitative approach, the specific field methods of participant observation linked with informal interviews and conversations, elite interviewing and case studies chosen for the data collection need to be explained and justified.
7KH¶VDPSOLQJ·SURFHVVXVHd for deciding which sources would be the most productive and effective was straightforward. The respondents interviewed depended on the nature of the organisations and people being studied, as well as on the (very) ¶legitimate limitations on [researcher] time, mobility, access and so forth· and of course, the research aim (Ambrosino 2007 p48). As far as public order policing in Britain was concerned, choices were driven by what was DFKLHYDEOH EDVHG XSRQ WKH DXWKRU·V SRVLWLRQ DV DQ ¶LQVLGH LQVLGHU· DQG WKH numerous possible sources available to him due to his length of police service and existing network of contacts.
The strategy employed for selecting respondents and settings for FRQGXFWLQJ REVHUYDWLRQV ZDV RQH RI ¶SXUSRVHIXO VDPSOLQJ· 3DWWRQ 7KLV DSSURDFK LV IRFXVHG RQ ¶VHOHFWLQJ information-‐rich cases for study in-‐GHSWK· (Patton 2002p230), and was also used to select the case studies described later. Patton outlines fifteen purposeful sampling strategies and this thesis was guided,
34 in particular, by two of these, ¶VQRZEDOO RU FKDLQ VDPSOLQJ·, DQG ¶FULWHULRQ VDPSOLQJ·Patton 2002 S¶6QRZEDOOVDPSOLQJ·DPRXQWVWRLGHQWLI\LQJ¶well situated people· RULQFLGHQWVWKDWKDYHEHHQKLJKOLJKWHGUHSHDWHGO\DQG¶FULWHULRQ VDPSOLQJ· ZKHUH ¶cases... meet some predetermined criterion· Patton 2002). A total of sixteen semi-‐structured interviews were conducted with police officers who had several years of experience in a command role, encompassing the complete range of strategic, tactical and operational functions (see chart below).
The data gathering phase had in effect commenced many years before this study and the knowledge already acquired was supplemented in a directed manner using a variety of methods and sources. It was felt that this would ensure an appropriate amount of relevant data allowing the construction of the kind of ¶WKLFN GHVFULSWLRQV· XSRQ ZKLFK TXDOLWDWLYH UHVHDUFK GHSHQGV *UD\ S Identifying the subjects for interview was again based on knowledge of the subject area, and knowing who the policy decision makers were in the British police service.
Rank Experience Formal/Informal
Chief Constable Strategic, tactical and operational command
F
Deputy Ch. Constable x 2 Strategic, tactical and operational command
F
ACC/Commander x 2 Strategic, tactical and operational command
F
Ch/Supt. x 3 Strategic, tactical and operational command
35 Superintendent x 3 Strategic, tactical and operational
command
F
Ch/Inspector x 2 Tactical and operational command F
Inspector Tactical and operational command F
Sergeant Operational Command Strategic, tactical and operational advisor
F
Commissaire and command team
Strategic, tactical and operational command
I
Colonel x 2 Strategic, tactical and operational command
I
Lt Colonel x 3 Strategic, tactical and operational command
I
Major/Commandant x 2 Tactical and operational command I
Capitaine x 3 Tactical and operational command I
Lieutenant x 2 Operational Command I
Fig. 1 Interviews: subjects rank, role, experience and whether interview formal or informal.
The strategy in this area was, where possible, to take advantage of existing relationships. Two senior and influential ACPO officers already knew the researcher, ZKRIHOWWKDWLIWZRVXFKKLJKSURILOHSROLFHRIILFHUVZHUH¶RQERDUG· the rest would follow, a tried and tested route to success that is suggested by Arksey and Knight (1999 p122/3). This approach did work and in every case the respondent was happy to help and gave up significant time from their busy schedules. More importantly, their enthusiasm for their work was evident and
36 their candour appreciated. All the British interviews were conducted according to a semi-‐standardised interview schedule (Fielding and Thomas 2008 p246) so that the same questions were asked of each respondent but with allowance for VXSSOHPHQWDU\TXHVWLRQVWKDWFRXOGSUREHDQVZHUVJLYHQRUHYHQJR¶RIISLVWH·LILW was felt that this would benefit the research and enable the examination of the context of thought, feeling and action· (Arksey and Knight 1999 p32); additionally and importantly qualitative interviews especially ¶allow for understanding and meanings to be explored in depth· (Arksey and Knight 1999 ).
The questions were designed to address the key issues of the research and WKHLQWHUYLHZVFKHGXOHORRVHO\IROORZHGWKHVWUXFWXUHRIWKH¶)ODVKSRLQWV0RGHO·LQ order to ensure consistency and relevance to the research question. Supplementary questions were also included to be asked where appropriate. All interviewees consented to their interviews being recorded on a digital voice recorder for later transcription thus ensuring a ¶permanent record that captures the whole of the conversation· (Arksey and Knight 1999 p 105). This was helpful LQWKDWWKHLQWHUYLHZVFRXOGEHODWHUWUDQVFULEHGLQWR¶:RUG·GRFXPHQWVLQRUGHU that analysis of the texts could be carried out. This assisted greatly, providing a complete record of what was said. Contemporaneous field notes of key points were also made during the interviews in order to have a summary of the main issues that when considered alongside the transcripts would allow for a thorough analysis.
Due consideration was given to whether the use of Computer Assisted Qualitative Data Analysis (CAQDAS) would be of benefit and it was decided not to use it on the grounds that the amount of data gathered from the interviews would be such that a paper based analysis would be achievable. Using the CAQDAS SURJUDPPHV¶$WODVWL·DQG¶1YLYR·WKHRQO\RQHVthat were available, was viewed as unlikely to add any value. Furthermore, access to the hardware containing
37 these programmes would be problematic and to become comfortable with the software would be vitally important (Lewins 2008 p417) but time constraints meant that this necessary level of comfort would not be achievable. Added to this was DGHJUHHRISHUVRQDOSUHIHUHQFHIRUEHLQJ¶LQWRXFK·ZLWKWKHGDWDDQG ¶the personal nature of the relationship between data and the researcher· (Lewins 2008 p417). This was considered to be a rational decision based on an informed position (Delamont 2010).
The review of existing literature has provided some essential theories for XVHLQWKHDQDO\VLVQRWDEO\'DYLG:DGGLQJWRQ·V¶)ODVKSRLQWV·PRGHO:DGGLQJWRQ 2007 p49). It has also revealed a number of sources of contextual data and the fact that there have been numerous studies into public order policing, including a comparative study of rioting in France and Britain (Waddington et al 2010). Notwithstanding, it is still the case that there has not been a comparative study of the policing of public order in France and Britain, nor one that was designed to show how the respective police forces manage public order and disorder nor, importantly, why they do what they do. In order to achieve the necessary broad understanding of the issues it was decided that as well as the contextual knowledge provided by a historical approach, it was necessary to compare case studies from both countries.
A case study is an in-‐depth study of a particular instance or of a few carefully selected cases (Gilbert 2008). Care needs to be taken with generalisation of findings but it is possible to be far more detailed than when using a large sample (Gilbert 2008). Case studies ¶add depth of understanding to individual, group, organisational, social, political and related phenomena· (Yin 2009 p4) and can ¶explain the presumed causal links in real-‐life interventions that are too complex for the survey or experimental strategies· (Yin 2009 p19). For this thesis the aim was to be able to draw some conclusions through a comparative
38 analysis (Gomm et al 2002). $SSOLFDWLRQRIWKH¶)ODVKSRLQWV·DQDO\Vis framework to the data provided the necessary rigour. It also ensured that a systematic procedure was followed (Yin 2009 p14). Embedding the case studies in this appropriate theoretical framework meant that they could be used analytically with the rich detail that emerged providing illuminating insights, ¶making formerly opaque connections suddenly pellucid· (Mitchell 2002 p183).
Deciding which case studies were most suitable was based on their being both interesting for their own sake and that they were exceptional in some way (Gilbert 2008 p36). The relevance and currency of the events in question was also a consideration. It was necessary to select cases ¶in which the experiences or processes to be studied [would] become especially clear· (Flick 2007 p28). Selection of the public disorder events for study was based upon the following main considerations:
x How recent were the events and the availability of evidence
x The seriousness of the disorder and that sustained police agency was necessary to deal with it
x The involvement of a wide range of police assets drawn from a large geographical area
Andrew Bennett·VZRUNRQcomparative methodology shows us how comparison of case studies helps with identification of additional variables and hypotheses, making inferences regarding causal mechanisms and developing historical explanations. Such comparison should achieve effective
operationalisation of variables, or ¶construct validity·%HQQHWW. The ability to identify new hypotheses is important and case studies do so by deduction and induction. There were some considerations necessary regarding the potential for bias in case selection, however it was thought that for the method of this thesis the choices of which events to study were well justified and the ¶tension between
39 parsimony and richness·%HQQHWWSachieved through application of the ¶)ODVKSRLQWV·PRGHO, as adapted, to the data proved to be beneficial. The limitations of case study methods including any lack of representativeness or ¶weak capability·LQWKH ¶causal weight·%HQQHWWS of variables were also catered for by ¶)ODVKSRLQWV·. This thesis illuminates a number of policy
implications; case studies contribute to theory building (Bennett 2002) and such theory development can be useful in policy making (George & Bennett 2005).
Decisions as to where best to go for participant observation in France were taken by the French Embassy, as already explained, and their choices proved to be appropriate, relevant, and informative. The conclusions, set out at the end of the thesis, are enriched by the empirical data derived from the fieldwork. The structure of the thesis is detailed in the next section.