You are on page 1of 23

Social Movement Studies, 2014

Vol. 13, No. 1, 70–91, http://dx.doi.org/10.1080/14742837.2013.832622

Change in Covert Social Movement


Networks: The ‘Inner Circle’ of the
Provisional Irish Republican Army
RACHEL STEVENSON & NICK CROSSLEY
Department of Sociology, University of Manchester, Manchester, UK

ABSTRACT In this paper, we contribute to both the growing body of literature on social movement
networks and the growing body of literature on change in networks by exploring patterns and
mechanisms of change within the network of the ‘inner circle’ of the Provisional Irish Republican
Army. Specifically, we focus upon the period between 1969 (when it formed) and 1988 (the last point
for which we have been able to gather good data). Our primary aims are substantive. We want to
know how this network changed over time. In addition, however, our analysis identifies changes
which other analysts might look for in their networks and offers methodological suggestions for
those who, like us, find that their networks do not meet the assumptions of mainstream approaches to
modelling network dynamics. There is a further dimension to the paper, however. We are studying a
covert social movement network. This is a special type of movement network whose organisation and
dynamics are predicted to vary from other movement networks. Some have suggested that they are
inclined to be relatively static because the need for trust within them is so great and the risk to
whatever they hold secret so considerable when new ties are formed that their members tend only to
recruit within the pool of their pre-existing ties and actively seek to minimise recruitment and the
formation of new ties. One of the aims of our paper was to determine whether and to what extent this
is so.

KEY WORDS : Covert networks, Provisional Irish Republican Army, network dynamics, network
change, social movement networks

The importance of social networks and the usefulness of social network analysis (SNA) as
a research method are both widely acknowledged in social movement studies, as this
special issue attests (see also Crossley, 2007; Diani & McAdam, 2003). Networks
facilitate communication between and therefore coordination of activists and groups
(Marwell & Oliver, 1993); they often generate trust, mutual support and a willingness to
cooperate (Coleman, 1988; Gould, 1991, 1993, 1995; Pfaff, 1996); they are channels of
recruitment at both the individual and the collective level (Crossley & Ibrahim, 2012;
Fernandez & McAdam, 1988; Kitts, 2000; McAdam, 1982, 1986; McAdam & Paulsen,
1993; Passy, 2001, 2003; Snow, Zurcher, & Ekland-Olson, 1980; but see Biggs, 2006;
Jasper & Poulson, 1995); they affect outflow of activists from a movement (Sandel, 1999)

Correspondence Address: Rachel Stevenson, Department of Sociology, University of Manchester, Manchester,


UK. Email: rachel.stevenson@postgrad.manchester.ac.uk
q 2013 Taylor & Francis
Change in covert social movement networks 71

and they provide channels for the diffusion of movement frames, repertoires and triggers
(Andrews & Biggs, 2006; Hedström, 1994; Hedström, Sandel, & Stern, 2000;
Ohlemacher, 1996; Oliver & Myers, 2003).
None of these effects are automatic. All are mediated by the inter-agency of those
involved. On this point, we agree with those who argue that the graph theoretic and
statistical methods of SNA sometimes fail to engage sufficiently with issues of meaning
and identity and that, to overcome these limitations and give due regard to agency, it is
important to combine SNA with other, particularly qualitative methods, in a mixed method
approach (Crossley, 2010; Edwards & Crossley, 2009; Emirbayer & Goodwin, 1994;
Mische, 2003). These are qualifications, however, not reasons to doubt the importance of
either networks for social movements or formal SNA for movement research.
In this paper, we contribute to this growing body of literature by exploring patterns and
mechanisms of change in a social movement network. Specifically, we analyse change
within the network of the ‘inner circle’ of the Provisional Irish Republican Army (PIRA)
between 1969, when it formed, following a split with the original Irish Republican Army
(IRA), and 1988, which is the last point for which we have been able to gather good data.
Our primary aims are substantive. We want to know how this network changed over time.
In addition, however, we hope that our analysis both identifies changes which other
analysts might look for in their networks and suggests some methodological hints for those
who, like us, find that their networks do not meet the assumptions of mainstream
approaches to modelling network dynamics.
In terms of research, PIRA have received less attention from movement scholars than
one might expect, given the longevity and intensity of their campaign. However, there is
some good sociological work on the movement (e.g. Alonso, 2006; Bosi, 2008, 2012;
O’Dochartaigh & Bosi, 2010; White, 1989, 1993, 2006; White & Falkenberg-White,
1991), and we have drawn upon it where relevant in this paper. We do not directly engage
with this work here, however, as its analytic focus is different from our own. Our focus, to
reiterate, is networks and, more specifically, change in social movement networks.
Very little work has been done on change within movement networks (important
exceptions are Diani, 1990, 1995; Saunders, 2007). Most network analytic studies of social
movements, like most studies of social movements more generally, tend to focus upon
snapshots, capturing the structure of a network at a specific moment in time (Stohl, 2008).
This is problematic because social networks and social movements are inherently
dynamic, always evolving as a function of the interactions of their participants. New
relations are formed, established relations acquire new dimensions and meanings, and some
relations either become latent, as parties drift apart, or dissolve following disagreements.
In addition, new participants may join and existing participants leave a network, as an
effect of birth, death, recruitment, geographical mobility, retirement, resignation,
excommunication, etc.
Furthermore, these ‘mechanisms of change’ are often themselves activated by wider
changes, both endogenous and exogenous to the network. In the case of social movement
networks, for example, successes and defeats can affect in and outflow of activists, as can
shifts in tactics, goals and ideologies (McAdam, 1983). Moreover, change to a network can
itself be a cause of further change, as when an inflow of new participants changes the ‘vibe’
of a movement, attracting further new members but also prompting some members to leave
(Crossley, 2006, p. 185). Our aim in this paper was to identify and explore some of these
mechanisms.
72 R. Stevenson & N. Crossley

There is a further twist to this issue of change in relation to our case study, however. We
are studying a covert social movement network (CSMN; on CSMNs see Crossley,
Edwards, Harries, & Stevenson, 2012). CSMNs are a special case of movement networks
whose organisation and dynamics are predicted to vary from other movement networks in
key respects (Crossley et al., 2012; Enders & Su, 2007; Lindelauf, Borm, & Hamers, 2009;
Morselli, Giguère, & Petit, 2007). In particular, some have suggested that they are inclined
to be relatively static because the need for trust within them is so great and the risk to
whatever they hold secret so great when new ties are formed that their members, first, tend
only to recruit within the pool of their pre-existing ties, and second, actively seek to
minimise recruitment and the formation of new ties (Crenshaw, 1992; Della Porta, 1992,
2006; Erikson, 1981). One of the aims of our paper was to determine whether and to what
extent this is so.

Modelling Change?
A number of methods for statistically modelling network dynamics have been developed
in recent years, affording social network analysts a much greater understanding of such
dynamics and a range of sophisticated tools for analysing it empirically (Doreian &
Snijders, 2010, 2012; Doreian & Stokman, 2005; Snijders, 2011). Unfortunately, however,
each approach rests upon assumptions that constrain the types of data they are able to
model. The most widely used modelling approach, Simulation Investigation for Empirical
Network Analysis (SIENA), for example, assumes a relatively stable population of nodes
and cannot be used in instances where there are large changes in that population over time
(Snijders, 2011). SIENA models the making and breaking of ties within a (more or less)
stable of population of nodes. Other approaches, by contrast, allow for a significant inflow
of actors but tend to assume that ties, once formed, are not subsequently broken and cannot
handle significant outflow of nodes from a network (Snijders, 2011; Bollobás, 1985). They
model entrance to a network and subsequent growth of the network but not turnover.
These assumptions are problematic in relation to our data. As we discuss in more detail
later, the turnover in the node set of our network is very large. This is an important finding.
It suggests that we are right to worry about the neglect of change in network studies of
social movements because they do change markedly. Also, it challenges the
abovementioned theories which predict minimal change in CSMNs, but it precludes our
use of existing modelling techniques and requires that we find an alternative to the
conventional modelling approaches for our analysis.
In addition, modelling techniques commonly assume full knowledge of the network or at
least of relevant alters on behalf of each node, such that nodes can ‘choose’ who to form ties
with. They also assume that nodes are able to form/maintain ties openly and freely. This is a
problematic assumption in relation to many networks but especially CSMNs, whose
members seek to maintain high levels of secrecy with respect to their activities and contacts.
Furthermore, CSMNs often form and reproduce in conditions in which external
agencies seek to damage and render them ineffective. General models of network
dynamics inevitably neglect to account for such specificities (but see Carley, Dombroski,
Tsvetovat, Reminga, & Kamneva, 2003). Indeed, most models are focused upon factors
endogenous to a network, when social movement networks, particularly covert and
paramilitary networks like PIRA, are shaped to a significant extent by their interaction
with a challenging, changing and unpredictable exogenous environment.
Change in covert social movement networks 73

We do not intend these comments as a critique of either specific modelling techniques or


modelling per se. Huge advances have been made in recent years, existing approaches are
very impressive and we look forward to a time when techniques more appropriate to our
case study have been devised. Furthermore, our understanding of network dynamics has
benefited hugely from the literature on these techniques. We mean only to specify why, in
this paper, we have not used any of the existing modelling approaches and have opted,
rather, for a more open-ended and inductive approach.
Our approach comprised an initial, exploratory phase, where network composition and
basic structural properties were compared at four time points, in an effort to identify
notable changes, which we then used as a springboard, in phase two, for developing
localised theories of change, which were then tested through further analysis. Acting upon
our preference for mixed method approaches, furthermore, we sought to weave an
extensive reading of historical accounts into phase two, using the knowledge gleaned to
both generate further questions for our analysis and to help us understand/explain the
changes observed. Context and qualitative detail are both crucial to an adequate
understanding of social network dynamics and we sought to achieve this by way of
secondary sources.

Introducing PIRA
PIRA formed after a split in the forerunning IRA in December 1969. The points of
contention were the interrelated issues of abstentionism and armed struggle. The then IRA
leadership was running-down the use of physical force in their campaign against British
imperialism and wished to drop the policy of abstentionism, allowing volunteers who were
successful in political campaigns to take up their seats in the parliaments of Belfast and
Dublin. This course of action was an anathema to traditional Republicans as recognition of
these parliaments was thought a de facto recognition of British rule in Ireland, and the
cessation of violence was regarded as a forfeiture of their birthright to defend their
homeland by force. Consequently, a number of longstanding figures broke away from the
IRA and formed the ‘Provisionals’. This split was echoed in the political wing of the
movement, Sinn Fein, the following January (Bowyer-Bell, 1997; Coogan, 2000; Taylor,
1998). PIRA soon became the larger of the two movements and the old IRA, now known
as the ‘Officials’, formally demilitarised in 1972.
The Provisionals continued their violent campaign against the British government until
1998, when, following a drawn-out peace process, the Good Friday agreement was
signed. However, prior to the peace process in 1986, a faction had broken away from PIRA
the same year, when the policy of abstentionism was again brought into question and this
time successfully removed from the Republican constitution. This group of activists
formed Republican Sinn Fein and later, their armed wing, the Continuity IRA.
Furthermore, the signing of the peace agreement in 1998 provoked another split, with
ex-Provisionals who refused to renounce the use of force, short of their goal of a
unified and autonomous Ireland, forming the Real IRA (Coogan, 2000; English, 2005;
Moloney, 2002).
Throughout its history, PIRA maintained a relatively rigid command structure.
The formal leadership of the movement consisted of a seven-man Army Council, which
included a Chief-of Staff, an Adjutant General and a Chairperson. The Quartermaster
General also usually sat on the council but, depending upon the discretion of the Chief of
74 R. Stevenson & N. Crossley

Staff, this may have been in an ‘eighth man’ non-voting position (Moloney, 2002). The
Army Council was elected by the Army Executive, who, in turn, was elected by rank and
file delegates at the Army Conventions.
The logistics of PIRA were controlled by the General Headquarters (GHQ) staff.
A number of directorates contributed to GHQ such as the Quarter Masters Department
(weapons), the Finance Department (money) and the Engineering Department (weapons
innovations). Each branch had a director and an indeterminate, but usually small number
of staff (Horgan & Taylor, 1997, p. 10). In the late 1970s, GHQ was separated into two
geographical areas. The six counties of Northern Ireland plus the five border counties
became Northern Command and the remaining counties of the Republic of Ireland,
Southern Command. (Moloney, 2002, p. 377). As PIRA banned armed attacks in the
Republic, Southern Command provided logistical back-up or a safety retreat. Northern
Command was the main operational area (Collins, 1997; O’Callaghan, 1999).
The movement was further separated into geographically distinct Brigades, each with
its own internal command hierarchy. A Brigade usually consisted of an Officer
Commanding and his Adjutant, an Operations officer and Brigade Quarter Master,
although there are other possible positions. Brigades were subdivided into smaller groups,
delineated along the lines of pre-existing communities which, prior to the late 1970s, were
arranged into Battalions and Companies, much like a standard army. After the attempted
introduction of a cell structure around 1978, however, the Battalions and Companies were
replaced by smaller, self-contained Active Service Units (ASUs). Each ASU had its own
Officer Commanding who controlled the day-to-day activities of the group, answering to
his/her superiors only when initiatives were dictated from above, or a planned operation
was of particular complexity or significance to the broader campaign (Harnden, 1998;
Moloney, 2002; O’Callaghan, 1999). The total command structure, after reorganisation, is
visualised in Figure 1.
All accounts suggest that this formal organisational structure was respected by many
PIRA members for much of the time and was important [with the exception of the
implementation of ASUs in the deeply insular areas of South Armagh and Tyrone

Army General Convention

Army Executive

Army Council

Northern Command GHQ Southern Command

Brigade Staff Brigade Staff Brigade Staff Brigade Staff

ASU ASU ASU ASU ASU ASU ASU ASU

Figure 1. PIRA’s formal command structure.


Change in covert social movement networks 75

(Harnden, 1998)]. However, like any formal structure of any organisation, it offers only a
faint and incomplete outline of the network of concrete interactions that ‘got the business
done’ on a day-to-day basis. Formal structures are always embedded in networks of
informal ties (positive and negative) which affect their functioning. It is well documented
that PIRA operatives were connected via community, friendship, employment and family
ties, as well as those forged in activism (Coogan, 2000; Harnden, 1998; Moloney, 2002;
Sharrock & Devenport, 1997). Indeed, English argues that ‘Informal status often
transcended official title, when it came to influence within the movement’ (2005, p. 213);
whether this be exercised in the role of trusted friend to an Army Council member, or that
of well-respected veteran without an official leadership position. Thus, in and around the
official PIRA leadership, we find a network of actors whose interactivities constitute an
‘inner circle’ of the movement. It is this inner circle that we focus upon in what follows.

Data
PIRA membership was secret during the period that we are analysing, often only being
confirmed upon death; likewise position within the official command structure (Toolis,
2000). In recent years, however, first-hand accounts have begun to emerge in the public
domain, particularly for the earlier period of the campaign. These accounts have provided
our source. In particular, we found usable data in 9 auto/biographical accounts by key
participants,1 10 journalistic, academic and historical texts (which in all cases drew upon
in-depth interviews with movement members)2 and 1 archival website (CAIN, 1968­
Present). The data from these 20 sources were cross-referenced where possible, and 18%
of the events upon which our networks are based (see below) were referenced by more
than one source. A further five3 sources were also reviewed but found to lack the detailed
information necessary for conversion into valid and reliable (inner circle) network data.
They were, however, useful for contributing to the ‘story’ of the network.
Network data were obtained by systematically surveying the abovementioned sources in
search of references to meetings which: (1) had a strategic significance in terms of PIRA
activities and (2) involved one or more members of the official command structure.
For example, all events relating to important tactical decisions, resource provision and
propaganda coups were included where names of attendees were available. ‘Official
command structure’, for these purposes, refers to positions in the Army Council and
Executive through GHQ directorates, Prison Commanders and Sinn Fein leaders. As such,
this network not only includes those who were part of the official leadership, but also those
who may have been of a lesser organisational position but with whom the leaders regularly
interacted. We refer to this group of people as the ‘inner circle’.
The data, which were all processed and analysed using Ucinet (Borgatti, Everett, &
Freeman, 2002), were collected in the form of two-mode actor-events (incidence)
matrices, which were then affiliated in order to derive actor-to-actor ties and subsequently
dichotomised. As structural change and turnover of membership is our key focus here,
only the presence or absence of a relationship in each time period was analysed, rather than
its strength. A total of 330 events and 255 actors were identified for the 20-year period and
they were divided into four periods, t1 – 4: 1969 –1973 (t1), 1974– 1978 (t2), 1979 – 1983
(t3) and 1984 – 1988 (t4).
The rationale for this division was, in some part, to have even lengths of time to
compare and a sufficient number of phases to track trends. However, our phases loosely
76 R. Stevenson & N. Crossley

map on to accepted phases of PIRA’s campaign and this was a further reason for adopting
them. t1 is often deemed a phase of defence and violent reprisal, which resulted in a form
of ‘tit for tat’ warfare between the Provisionals, the British Army and Loyalist paramilitary
groups. t2 has been described as a one of attrition, where both sides, Republican and
British, worked to wear the other down, boosting support and breaking enemy morale.
t3 was characterised by reorganisation and strategic change, including a decision to pursue
movement goals by institutional-political as well as and in combination with paramilitary
means – popularly referred to as the ‘armalite and ballot box’ or ‘long war’ strategy.
Finally, t4 was characterised by increased institutional-political participation, a resultant
split in the movement and the beginnings of a dialogue towards peace.
This time frame involves most of the key events in PIRA history, as commonly
understood, including: Internment, Bloody Sunday, the 1975 Truce, the 1981 hunger
strikes and the 1986 split. Events for 1969 begin with a meeting at the Casement Park
Social Club in Belfast in August, at which a number of leading IRA men decided to break
away from the Southern leadership and take control of the IRA in their city. Most
commentators agree that this was a significant event in the run up to the formal split at the
December 1969 Convention and that these men formed the organisational nucleus from
which the new Provisional wing would grow (Bowyer-Bell, 1997; Coogan, 2000; Taylor,
1998). Events end in 1988 because the quality of the data matching our inclusion criteria
notably deteriorates after this point. This is perhaps unsurprising as the desire for secrecy
remains stronger for more recent events.
We believe that constructing two-mode data-sets of key meetings from secondary
sources, as outlined above, is not only a practical way to capture the leadership of a covert
social movement, but also affords great possibility for researchers who are interested in
where power and influence actually lie, rather than in official portraits of such. Moreover,
this form of SNA might prove particularly useful when applied to contemporary social
movements which have a more diffuse organisational structure, allowing key players to
emerge through analysis of the empirical social network data.
We are mindful of the limits of our sources, however. They are, in most cases,
retrospective accounts of once covert activities in a struggle which, even if now peaceful
and legal, formally involved violence, illegality and still involves very strong feelings on
all sides – factors which may prejudice recall. Similar problems attend to all studies of
covert movements, though, and are inevitable if we are to study them empirically, by any
methodology. Having said this, we believe that cross-referencing improves reliability and
raises our confidence in the data. Likewise, although the auto/biographical nature of some
of our sources probably biased the data we gleaned in terms of the prominence and activity
of the authors – who are key figures in their own life stories – this too was tackled by
cross-referencing with various and varied auto/biographical narratives (those of leadership
as well as rank and file members, geographically disparate volunteers, loyal veterans,
drop-outs and informers), as well as a number of more general texts written by embedded
journalists, historians and social scientists. Each source was used to scrutinise
inconsistency and possible bias in the others, particularly in the more personal accounts.
Although interviews with movement members and the consultation of government
sources may have been desirable as a triangulation technique, they were not feasible for
this project. The nature of the information necessary to map the PIRA network precludes
the use of government documents (for examples of research which uses these, see Bosi,
2012; Horgan & Taylor, 1997; White & Falkenberg-White, 1991). Despite the passing of
Change in covert social movement networks 77

the minimum 30-year embargo on the release of documents such as those from the early
years of the ‘troubles’, official records which include names and activities of suspected
PIRA members – still regarded as potentially sensitive to national security – have not
been released and are unlikely to be so (Bowyer-Bell, 1997).
Other researchers have interviewed former PIRA members regarding their recruitment
and motivations (for examples, see Alonso, 2006; Bosi, 2012; White, 1993; White &
Falkenberg-White, 1991), but they did not directly seek specific details about dates,
comrades and illegal activities, nor could they have. Indeed, White states that interviews
were only granted to him because he ‘made clear that respondents would not be asked
about illegal activities that they may have participated in’ (1993, p. 180).
Moreover, interviews in previous PIRA research were conducted prior to the recent
controversy surrounding ‘the Belfast Project’, where a court ruled that eight in-depth
interviews with ex-Provisionals concerning their armed activism should be released to the
Police Service of Northern Ireland with a view to bringing convictions in a murder case. An
appeal is still in process as to the release of this information, which was supposed to remain
sealed until the death of participants (Martin, 2012). It is very unlikely that ex-PIRA volunteers
would volunteer for interview in this legal environment. As Anthony McIntyre, former PIRA
member and researcher on the Belfast Project states, ‘who will now come forward and reveal
anything?’ (PSNI, 2012) Indeed, even prominent figures whose former PIRA membership is
well known are said to be ‘very nervous’ about the situation (O’Leary, 2012).
Nevertheless, it must be reiterated that the secondary sources used in this project were
all either first-hand accounts of PIRA activity during our time period, or were based on
interviews with movement members. Coogan, for example, estimates that over his decades
of research into the group he has (in)formally interviewed over 500 people ‘either in or
concerned with the IRA’ (2000, p. 777). Similarly, Bowyer-Bell (1997) puts his figure at
over 100 interviews, and so respected was his writing that PIRA placed one of his books
about the movement (Bowyer-Bell, 1970) on a list of recommended reading for new
recruits (Horgan, 2008). Indeed, we believe that the long-standing and sometimes warm
relationships between the authors that we have consulted and the key Republican figures
they interviewed (Bowyer-Bell, 1997; Coogan, 2000; White, 2006) serve to strengthen our
data and take it beyond that which could have been collected by us, neither of whom have
ties to the movement. Moreover, on a more practical level, many of the interviews
conducted by these authors were with important figures from the early years of ‘the
troubles’ that are now dead. The use of secondary sources, therefore, not only gives us
access to first-hand information from Provisionals, politicians and members of the security
services we would not otherwise have, but also to a greater amount of information than we
would have been able to accrue within our time restrictions.
Whilst doubt has been cast upon the accuracy of the information in some of the
autobiographies that we have drawn upon, corroboration has been offered from different
quarters, often in equal measure.4 Indeed, one might argue that no account of such a
contentious subject is likely to be produced without debate. And where doubts still might
exist about the soundness of this information, we contend that more valid data are unlikely
to be collected by other means. In-depth interviews conducted by ourselves, for example,
would involve the same potential problems with (in)accurate ‘presentations of self’
(or indeed others) as in the written auto/biographical texts.
As in most research, it is likely that data are missing from our study. This is in the form
of participants whom we may not have picked up, or missing connections between those
78 R. Stevenson & N. Crossley

that we have. This could potentially affect the structure of our network, most obviously in
terms of density for example, as well as our data on the turnover of personnel. Indeed, the
Provisional movement was of course a covert one, which does provide greater challenges
for data collection than with an overt movement. However, as there is a good consensus
between our 20 sources regarding the main figures in the PIRA leadership, across our time
periods, it is likely that any missing participants would have been peripheral. Our focus, to
reiterate, is the leadership of the movement. Indeed, we have the qualitative data to chart
the occupants of many official positions, such as those on the Army Council, in every year
of our study, all of whom are included in our network. Furthermore, as the involvement
even of those who now publically deny their participation in the movement has been well
documented by the authors we consulted, there is no reason to believe that any significant
figures have been left out of their accounts.
The potential for connections between Provisional operatives to be missing from our
network is perhaps greater than the potential for people to be missing. Again, however, we
believe that the significant connections between the most important people have been
captured. As we are not analysing the strength of relationships between actors in this
paper, we have concentrated only on whether there is, or is not, a link between operatives
in each time period, measured by co-participation in a PIRA event. Hence, we only need to
capture one incident of co-participation, per pair, per period, in order to register a
connection between them; not an exhaustive list of all events where connections between
two people were made. Again, to return to the above point, as the leadership figures are
quite well documented in our sources, it is unlikely that a link between key players would
not be mentioned where historically there was one. Moreover, our sources, which are
largely historical narratives, are written in such a manor that connections between actors
are rarely mentioned without grounding them in specific circumstances. We feel that our
events criteria are sufficiently broad to capture most relationships, ranging from relatively
large scale formal meetings to one-on-one social occasions.
Even if there are ties missing from our data, moreover, robustness testing suggests
that our findings are unlikely to be significantly affected as a consequence. Iterated
re-measurement of the network with an added 10% of random ties revealed no significant
structural changes.

Data Preparation
A preliminary analysis revealed a single component at both t1 and t2, but a disconnected
dyad at t3 and a slightly larger secondary component (n ¼ 8) at t4. We elected to remove
these secondary components, focusing upon main components only, and believe that we
had good reasons to do so. The dyadic component in t3 represents two activists who were
joint OCs of the maze prison in 1979. Initially their status as prison leaders qualified them
for inclusion in the ‘inner circle’ as they were in charge of detained volunteers during the
final stages of the ‘Blanket’ and ‘Dirty Protests’, both of which could be regarded as
successful enterprises in gaining attention and support for the Republican cause.
Moreover, they were significant in events leading up to the hunger strikes of 1980 and
1981 (one led the first group of strikers in 1980). However, because of their involvement in
the hunger strikes and the expected mental and physical toll it would take, both men then
lost their shared OC position and were replaced by another volunteer. Although engaged in
the blanket and dirty protests, these men did occasionally wash and put on the prison
Change in covert social movement networks 79

uniform in order to receive visits. However, no mention of face-to-face contact between


them and the ‘inner circle’ was found in our sources – which is necessary to meet our
inclusion criteria. In addition, the Republican inmates’ decision to go on hunger strike was
very unpopular with the wider leadership outside of the prison (Bowyer-Bell, 1997,
p. 502), which suggests not just a physical but also an ideological separation between the
prisoners, particularly the men in this dyad, as figureheads of this ‘rebellion’, and the
participants in the main ‘inner circle’ component. For these reasons, we felt justified in
excluding them from this time period.
The smaller component in t4 was removed for similar reasons. All participants in this
subgroup were active in the England Department in 1984 and thus were members of GHQ
staff. However, they were all arrested subsequent to the Brighton bombing of October that
year and remanded in custody, preventing further involvement with the leadership group
unless visited in prison by a member of the ‘inner circle’. No evidence of such contact was
found during the remaining 4 years of activity in this network. Moreover, many commentators
describe the England Department as an insular group, necessary to help maintain the ‘sleeper’
status of its operatives on the British mainland (Bowyer-Bell, 1997; O’Callaghan, 1999).
Indeed, we found no evidence of connection between the actors in the main component of the
‘inner circle’ and this peripheral and detached group, prior to their arrests in 1985.

Results
Our preliminary, exploratory analysis of the network t1 – 4 identified three key findings:
1. There is considerable turnover in the nodes populating the network.
2. Despite this, most key structural properties remain relatively stable.
3. Apart from network centralisation, which steadily rises.
In what follows, we elaborate both upon these findings and upon our efforts to explain and
further explore them. A key element in this, moreover, is our observation from the
secondary literature of an emergent subgroup within the network seeking and eventually
succeeding in achieving leadership of PIRA. Our network analysis identifies this group
and we seek to bring our findings on this into dialogue with those from qualitative-
historical literature.

Population Turnover
As mentioned earlier, an important finding of the study was the extent of the turnover in
network membership. A total of 128 participants were involved in the inner circle in t1. By
t2, however, 89 of these appear to have dropped out, with the remaining 39 being joined by
20 newcomers. Similarly, by t3 only 29 of the 59 inner circle participants from t2 remain,
with 55 new entrants and 14 returnees from t1 who were not involved during t2. Continuing
this pattern, 34 of the 98 involved at t3 continue into t4, where they are joined by 51 new
comers, a further 4 returning (for the first time) from t1 and 4 returning (after a lapse) from t2.
Looking at this information from another angle: of the 2495 actors who participated in
the inner circle at any time, only 16 were active in all four phases; 19 in 3 (11 in
consecutive phases); 37 in 2 (23 consecutive) and 177 in 1 only (see Figure 2). Given the
tendencies for some activists to return after a period of absence, we must be careful how
we interpret these figures, especially in relation to the ‘one onlys’ from later periods (who
80 R. Stevenson & N. Crossley

4 Periods
(6%)
3 Periods
(8%)

2 Periods
(15%)

1 Period
(71%)

Figure 2. Number of phases individual activists were involved in the inner circle.

may have returned in the period after our study finishes) and in particular we must
acknowledge that we do not know whether those joining the inner circle in t4 continued
beyond it. However, there is a fairly clear tendency towards short-term involvement and
only a tiny proportion of those involved remain involved across all four or even three of
our periods. This challenges the abovementioned theories of CSMNs which predict stasis
in their composition, and it begs the question of why turnover is so extensive.
We can approach this question on three levels. (1) general factors affecting many
activists; (2) specific factors which we know to have affected particular activists; (3)
network factors and, in particular, the centrality of activists in the network. We can identify a
number of factors, beyond ageing (both coming of age and becoming too old) and death
(more common amongst paramilitaries than the general population), which served variously
to pull new recruits into the Republican movement, and by extension into its inner circle, and
to push them out. The movement itself was in a constant process of change, for example
growing out of the old IRA but becoming something new, shifting its geographical centre of
gravity from south to north and constantly revising its policies. Each change made PIRA
more attractive for some and less for others. Likewise, we can identify a series of major
events (e.g. Internment, the Falls Curfew, Bloody Sunday and the 1981 hunger strikes)
which served both to a generate huge ‘moral shocks’ (Jasper, 1998) boosting recruitment
(Coogan, 2000; English, 2005; Taylor, 1998) but also to add to the costs of involvement,
especially for those in senior positions. In addition, echoing McAdam (1983), narrative
accounts suggest that perceived movement successes/failures contributed significantly to
in/outflow from the movement and its inner circle. Contrary to McAdam, however, there are
indications that perceived successes sometimes prompted an outflow as weary activists
believed that the struggle was ending. As senior Provisional, Brendan Hughes, said of a
1972 ceasefire (in the wake of the fall of the Stormont government, a major PIRA success):

People thought ‘this is it’. No one before had fought the British government to a
position where they were negotiating with IRA people so there was an expectation
that the war was over. I think there was a need to believe it: a wish to believe it and a
Change in covert social movement networks 81

desire for it to happen [ . . . ] people wanted it to end then. They wanted back to their
wives and they wanted back to their kids. (cited in Taylor, 1998, p. 139)

Imprisonment often contributed to this weariness and to the costs of activism, encouraging
exit. Also, in some cases it brought isolation, cutting activists off from the movement and/
or its inner circle. In other cases, however, prisons served as network ‘foci’ (Feld, 1981),
bringing activists together and bringing some into the inner circle. It is widely
documented, for example, that cages 9 and 11 of the Maze prison brought together a group
of activists who would serve, when in prison, as an important ‘think tank’ for the
movement, and when on the outside, as an important faction pursuing control of the
movement and the implementation of their think tank ideas (Moloney, 2002; Sharrock &
Devenport, 1997). We return to this emerging leadership group.
At a more specific level, we can account for the lapses of 49 of the 56 activists who are
present in only two or three of our phases. For 3 of the 19 threes and 5 of the 37 twos, all of
whom participate in continuous time phases, gaps can be explained in terms of their not
entering the network until t2 or t3. For some, such as participant (‘P’ hereafter) 169, this is
due to his young age and radicalisation by ‘later’ events such as the hunger strikes
(Collins, 1997). For others, such as P130, we believe it is a consequence of being a
member of a more peripheral brigade, outside of the movement’s geographical centres: i.e.
Belfast, Dublin and, to a lesser extent, Derry. For those external to the original core, it
would have taken time to build up the trust necessary for the PIRA leadership to engage
with them. This might be proven by continued successful activity at the rank and file level.
Access to resources was also important, however. P130 is known to have contributed a lot
of money, as leader of a rural Brigade, to PIRA central command upon entrance to the
inner circle (Harnden, 1998; Moloney, 2002).
Similarly, some participation patterns can be explained by ‘retirement’. This is the case
for 4 threes and 2 twos. For P22 and P93, we know that retirement was due to poor health.
P22 was one of the ‘hooded men’ who were subject to torture during interrogation after
they were interned in the early 1970s. This was believed to have lasting mental and
physical consequences and he subsequently relinquished his position as a senior figure in
the Quarter Masters Department (Moloney, 2002). Again, age figures here, with a number
of participants ‘dropping out’ in the later phases as they reach their more senior years. For
some, however, retirement was not complete. P65, for example, returns at t4 as an ‘elder
statesman’ at the 1986 convention. His status was apparently such that his views on
abstentionism assumed considerable importance (White, 2006). Likewise, P17 drops out
of militant activity but returns in a political role as a local Sinn Fein (SF) Councillor
(Sharrock & Devenport, 1997).
Of course, there are often multiple push and pull factors at work. P2, who was involved
in t1 and t2, formally retired on grounds of age and associated poor health, but it is also
claimed that there was an internal campaign against him, which laid blame for the
disastrous 1975 ceasefire and resulting sectarian violence firmly at his feet (Moloney,
2002). Such political manoeuvring, on behalf of a newly emerging leadership subgroup,
along with his participation in the truce talks, also resulted in the sidelining of P21, a close
associate of P2 (White, 2006).
Political dynamics, in terms of both internal rivalries and ideological disagreements are
a major cause of change more widely, and 4 threes and 3 twos – plus the two men
discussed in the previous paragraph – cease participation due to this. As with P2 and P21,
82 R. Stevenson & N. Crossley

P55 was also pushed out of the movement. The official reason was that he had an
extramarital affair, but such behaviour was not entirely uncommon in the movement and
was irrelevant to the achievement of Republican goals, leading some to allege that his
expulsion was a ruse by the aforementioned emerging leadership subgroup, to take control
of a PIRA newspaper, which P55 was then editing (Moloney, 2002).
The remaining six participants dropped out of interaction with the inner circle when
they stood down from positions in SF. Again this was a consequence of a take over
strategy implemented by the emerging leadership subgroup, this time featuring the long
held policy of federalism to which PIRA then subscribed. After a sustained period of
campaigning, and the kudos of managing the political aspects of the hunger strikes in front
of the worlds’ media, the subgroup had the ‘Eire Nua’ policy removed at the 1982 SF
conference. Subsequently, these six figures, many of whom were involved with the
authorship of the policy, resigned or refused to stand for re-election to the SF officer board
(White, 2006). The removal of this group from the upper echelons of the Provisional
movement was cemented a year later when a senior member of the ascending subgroup
took over presidency of SF (Bowyer-Bell, 1997).
Rather than being pushed out as a result of group dynamics and rivalries P86, a former
British soldier, is said to have become disillusioned with PIRA after participation at the
leadership level during t1 and t2. He defected, allegedly of his own accord, to America in
1979 (Coogan, 2000).
Promotion too is an important factor. 4 threes and 1 two figure in the early phases not
because they hold a formal position, or even much sway within the movement, but because
they have interacted, perhaps only once, with somebody else who does. They appear again
later, however, after achieving a formal leadership role. P97 and P98, for example,
participate in t1 only as attendees at a meeting held by a very senior Belfast figure
regarding reorganisation of PIRA units in the Republic. In t3, however, these men both
return as senior figures in Southern Command, from which P97, in particular, advanced
further up the PIRA hierarchy (O’Callaghan, 1999).
These cases differ from others, where contact with the inner circle was contingent and
short lived. P26, for example, was a resident in the USA and involved with a weapons
procurement ring which transported much needed arms across the Atlantic and with Irish-
American support body, NORAID (Clarke & Johnston, 2007; English, 2005). His
geographical separation as well as resource procurement role meant that contact with other
inner circle members was always limited. Likewise for P27, who was described only as
part-time Director of Publicity for SF (Moloney, 2002; White, 2006).
Another significant factor in turnover is imprisonment. This affects 3 threes and 10
twos. Its effect is not always interruption, however. Imprisonment can lead an activist to
join our inner circle if it brings them into contact with high status alters or with other
operatives involved in significant activity or, for similar reasons, maintain participation in
the leadership group. For example, P87 is only part of our networks whilst imprisoned, as
he was involved in a number of protests which received publicity and support in Portlaoise
jail (Coogan, 2000; MacStiofain, 1975). And P105 earned his place by helping three senior
figures escape from Mountjoy prison, rather famously, by helicopter in 1973 (Anderson,
2003). However, for others, particularly after the removal of special category status in
1976, imprisonment did suspend involvement. P118, for example, was interned in the
infamous huts of Long Kesh with many senior men, including some of those who were to
form the aforementioned emergent leadership subgroup (Adams, 1997). Despite this early
Change in covert social movement networks 83

interaction, however, no evidence of connection or formal PIRA position could be found


for this man in the subsequent time periods. Prison, in other words, can be both an entrance
to and an exit from the inner circle, depending upon whom one is imprisoned with, how
one copes and what one becomes involved in when inside.
Finally, for 1 three and 4 twos, death puts a stop to involvement. For P42 (White, 2006)
and P57 (Bishop & Mallie, 1988), this is due to natural causes, but for the remaining 3, loss
of life was movement related. P223 and P23 (Adams, 1997; Anderson, 2003) are both
believed to have been shot in sectarian killings, whilst P113 died during the 1981 hunger
strike (Coogan, 2000).

Network Centrality and Participation Longevity


Biographical contingencies are clearly important in determining the length of
involvement. Pushing our analysis one step further, however, we hypothesised that
those activists who enjoyed a more central and thus integrated position within the network
were more likely to have remained active within it over time. There are three reasons to
believe this. First, centrality and integration are rewards which incentivise continued
involvement. Actors receive greater recognition within the network and are likely to
derive a greater sense of importance from their involvement. Second, integration within a
network increases the controlling influence of the network upon the actor. Their exposure
to collective situational/identity definitions is likely to be greater, their behaviour is under
a greater level of (informal) surveillance by others and their opportunities for drifting or
deviating are therefore reduced. Third, the other side of this is that they receive greater
support for their involvement, which reduces their likelihood of burning out or becoming
disillusioned. Centrality and integration within a network support commitment to the aims
of any collective action in which it is engaged.
To test this hypothesis, we created a binary variable indicating whether network
members were involved in ‘at least three’ of our periods and then, for each period,
selecting only those involved in that period, we compared average degree and closeness
scores for those who were, with those who were not, using Ucinet’s ‘network friendly’6
T-test to determine statistical significance. We selected degree centrality, specifically,
because we believe that higher levels of contact equate, in this case, to higher levels of
positive reinforcement and support for involvement, and also greater surveillance and a
greater threat of punishment for deviation. Closeness is a less obvious choice but still
relevant because it suggests that an actor’s indirect contacts are relatively close, and
indirect contacts, where close, may exert a similar effect upon participants. The esteem
(or disdain) in which one is held by ‘friends of friends’ is important and the thought that
they are concerned for one’s welfare may be an important support, but this is less so for
‘friends of friends of friends’ and so on. Our results are presented in Table 1.
These results support our hypothesis. As a network ages, we might expect those with a
longer history of involvement to be more central within it, because time allows them to
accumulate contacts and significance. However, we have found that those who ‘stayed the
course’ were more central and integrated in the network from the start, which suggests that
centrality and integration help to maintain commitment.
It might be argued, against this, that centrality and longevity are each an effect of levels
of ambition and commitment which are independent of and unaffected by them. We agree
that ambition and commitment are important and cannot be reduced to network position.
84 R. Stevenson & N. Crossley

Table 1. Centrality and longevity of involvement.

Mean degree centrality Mean closeness centrality


1 – 2 Periods 3 –4 Periods p 1 – 2 Periods 3 – 4 Periods p
t1 9.5 22 0.00 40.7 47.9 0.00
t2 5.7 8.4 0.02 38.5 42.8 0.01
t3 9.8 19.6 0.00 42.4 50.2 0.00
t4 8.5 17 0.00 35.7 43 0.00

However, we believe that, for the reasons given, network centrality and integration do
contribute to commitment and we take our data as evidence in support of their positive
association. That is not to say that this relationship is one directional, however.
Commitment may equally generate increased centrality. We discuss narrative evidence for
this below.

Network Structure
Moving now to the basic structural properties of the network (see Table 2), we note, firstly,
that, notwithstanding turnover, density remains relatively stable. Caution is necessary
when comparing density scores, at least where the number of nodes (order) varies. The
greater the number of nodes in a social network the more difficult it is for everybody to
enjoy a tie to everybody else such that density tends to drop. However, variations in order
are not great across our periods and we get a similar picture if we use the more
comparison-friendly measure, ‘average degree’. There is some variation in average
degree, with a notable drop in t2 and a slight increase at t3. On the whole, however, the
figure is relatively stable. Similarly, average path lengths are relatively consistent. In other
words, the network remains relatively stable in terms of the level of interaction between
participants, in spite of a considerable turnover of personnel.
It is difficult to explain this stability with the data that are available to us. However, we
believe that it may reflect consistent constraints upon network formation/reproduction and
also perhaps a local optimum of functionality. That is to say, the level of interconnection
remains the same because the constraints of maintaining secrecy, whilst contributing to the
continued functioning of the movement, also remain the same. Network structure is an
effect of the activities participants are engaged in and the constraints they experience,
including but not exclusively those effected by the network itself.

Table 2. Basic network properties.

Degree Betweeness Closeness


Av. Av. centralisation centralisation centralisation
Period Order Density distance degree (%) (%) (%)
t1 128 0.09 2.5 (3) 11 (SD 11) 35.17 28.44 40.61
t2 59 0.12 2.5 (3) 7 (SD 5) 37.33 42.17 43.05
t3 98 0.13 2.7 (3) 13 (SD 11) 39.99 35.21 45.26
t4 85 0.14 2.3 (2) 11 (SD 10) 45.91 21.94 47.58
Change in covert social movement networks 85

Centralisation and the Emerging Leadership Group


Notwithstanding this basic structural stability, Table 2 shows a steady increase in both
degree and closeness centralisation.7 This suggests that, over time, the ‘inner circle’ was
becoming more centred upon a relatively small group of activists who, we assume, were
gaining in importance.
Building upon our earlier analysis, we hypothesise that this reflects the increasing
importance assumed by those who had been in the network over a longer period and who,
consequently, had strengthened their position within it, perhaps whilst also interacting
with one another in an informal core of senior figures. If this were so, it would shed further
light, at a network level, on a claim common in the literature and alluded to above, of a
new, would-be leadership group taking shape within PIRA during the 1980s.
On a narrative level, several examples have already been discussed of the strategies that
this new leadership group used in order to wrestle control of the PIRA movement from
other long-standing figures. For example, the tactic of ruining a person’s symbolic
reputation in order to affect their structural position, as in the case of P2 and 55, where
blame for violent sectarian feuding and personal controversy were used against them in
order to suggest that these men were supportive of behaviour contrary to Republican
ideals. Other examples include creating new layers of or alternative power structures, such
as the Northern Command or Revolutionary Council, in order to circumvent existing ones
which were still dominated by members of the ‘old leadership’.
Moreover, the position of the emerging group was further strengthened by ‘geo-politics’.
For example, Northern Command meetings were always held in the ‘six counties’, where
many members of the established leadership, who were ‘on the run’, could not go for risk of
arrest. The largely northern profile of the newly emerging clique meant that no such
obstacles existed for them (Moloney, 2002; O’Callaghan, 1999; Sharrock & Devenport,
1997). These examples accord with one PIRA operative’s observation that the senior
members of this new group preferred to ‘isolate and marginalise’ their competition rather
than confront them openly about matters they disagreed upon, or directly challenge them for
their positions within the movement (Moloney, 2002, p. 178).
Once power had been consolidated by the new leadership group, senior organisational
positions were quickly taken up by them. And where positions were limited, such as on the
Army Council, for example, others were co-opted into supposedly temporary positions to
ensure that any remaining rivals from the ‘old leadership’ were convincingly outnumbered
(Moloney, 2002). This would add to the increasing centrality of an emerging leadership
group.
That is not to say that all previous leadership figures were replaced by new comers,
however. Indeed, commentators argue that a good deal of political skill was exercised in
convincing some veteran figures to support the direction of the new leadership and remain
in position, thereby providing continuity for volunteers who were uneasy with such
change and demonstrating the reverence for tradition which is so important in Republican
culture (Bishop & Mallie, 1988). In addition, these veteran leadership figures were often
used by the new leadership to float new or innovative policy ideas for this very reason, as
they were more likely to get sceptics on their side – a tactic which also had the benefit of
protecting the senior members and ideologues of the new leadership group from any
negative backlash. Indeed, more junior participants were also used to float potentially very
86 R. Stevenson & N. Crossley

unpopular policy so as also to provide this protective buffer to those now at the centre of
the ‘inner circle’ (Moloney, 2002; Sharrock & Devenport, 1997).
One further and perhaps final measure taken by the new leadership group to ensure their
central positions and control over the movement was to introduce the idea of a permanent
leadership. Previously, if volunteers were arrested or had to go ‘on the run’, they would
lose their position within the Provisional hierarchy. The new commanders felt that this
system caused significant instability in leadership as well as the loss of important
knowledge and talent. It was thus ordered that anybody who held a senior position was to
distance themselves from any involvement, beyond the managerial, in armed operations.
This meant that the new leadership group held on to their leadership positions by avoiding
death or arrest. It also meant that challengers could not emerge for vacated command
positions, a tactic they themselves had used during their ascent.
One commentator argues that after the introduction of permanent leadership, those at
the head of the movement became a ‘self-perpetuating elite’ rewarded as much, perhaps
more, for their loyalty as for particular skills (Moloney, 2002, p. 157). The relative
distance created between the leadership and the military side of things, including relative
safety from arrest, eased the leadership’s increasing participation in constitutional political
affairs, which in turn opened links with new, especially political operatives. This course of
events also paved the way for the dropping of abstentionist policy which eventually led to
the 1986 split in the movement (Coogan, 2000; Taylor, 1998).

Core and Periphery


To further explore the idea of an emerging leadership group, we examined core – periphery
structures (using Ucinet’s categorical algorithm). Our first step was to ascertain whether
our network manifests evidence of strong core – periphery structures across time.
As Table 3 indicates, it does.
In a ‘classic’ core – periphery structure, members of the core have a higher density of
ties to one another than to members of the periphery, and members of the periphery have a
higher density of ties to the core than to one another. We find exactly this pattern in three
of our four phases. The exception is t3. Even in t3, however, there is, as in the other periods,
a considerable difference in the internal density of the core (0.84) compared to that of the
periphery (0.08). Indeed, in t3 the core is particularly cohesive.
We also note that in each period, with the exception of period two, where the network is
much smaller in any case, the size of the core is similar (between 29 and 35). This suggests
that this core size may have proved optimum. But who was in the core and, more
specifically, were they those who had been in the inner circle for longer?
To test this, we cross-tabulated, for each period, core/periphery membership with a
binary variable indicating whether or not the members of the network were involved in
three or more of our periods. In addition, we calculated the odds ratio for involvement in
the core for those who are involved in the network in three or more periods (see Table 4).
The figure given, in each case, indicates how many times more likely activists who are
involved in three or more of our periods are to figure in the core for the particular period in
question.
We see that, from t2 onwards, activists who were involved across three or more periods
form a majority in the core. And from the third period onwards, the majority of those
involved in three or more periods are found in the core. In other words, as the movement
Change in covert social movement networks 87

Table 3. Core – periphery structure across the four periods.

t1 t2 t3 t4
Core Periphery Core Periphery Core Periphery Core Periphery
(n ¼ 35) (n ¼ 93) (n ¼ 14) (n ¼ 45) (n ¼ 31) (N ¼ 67) (n ¼ 29) (n ¼ 56)
Core 0.57 0.07 0.63 0.1 0.84 0.03 0.65 0.08

Periphery 0.07 0.04 0.1 0.09 0.03 0.08 0.08 0.06

Note: All scores are density scores ranging between 0 (no ties are present) and 1 (all possible ties are
present)

Table 4. Core/periphery membership by time spent in inner circle.

t1 t2 t3 t4
Periods of
involvement Periphery Core Periphery Core Periphery Core Periphery Core
2 or fewer 73 24 24 4 51 15 43 14
3 or more 20 11 20 10 15 16 13 15
Odds ratioa 1.67 3 3.63 3.54
a
Greater likelihood of activists with three or more phases of involvement being in the core.

and its ‘inner circle’ ages, those who have spent more time in that inner circle achieve a
greater chance of belonging to its core – chi-square probability values are not strictly
appropriate in this context but it is noteworthy, as additional support for this trend, that
they become significant for the final two periods.8 This is also supported by the odds ratios.
Although those who would go on to belong to the inner circle for either three or four
periods were only 1.67 times more likely to belong to the core in the first period, they were
3.54 times more likely by the final period (and this figure was slightly higher for the third
period).
This supports our contention that an elite was beginning to form within the network, and
that this elite were drawn predominantly from those who had enjoyed a longer period of
involvement in the network, which, in turn, was a reflection of their centrality at the earlier
stages of the network’s development. Of course, the initial higher centrality of this group
would also have been an advantage for them if and insofar as they sought to establish
themselves as a core or, indeed, even if the emergence of a durable core was an unintended
outcome of other projects.

Conclusion
In this paper, we have opened up the question of the ways in which social movement
networks change and the ways in which we might analyse these changes, particularly
when they do not conform to the assumptions of mainstream methods of modelling
network change. At a substantive level, we have four key findings: (1) there is a high rate
of turnover within the network (activists leaving and entering), with the majority of
activists being present for one phase only; (2) those who do stay in the network for
longer tend to be those who enjoy a more central position within it (for reasons which we
discuss); (3) in spite of this large turnover, key network properties tend to remain
88 R. Stevenson & N. Crossley

relatively stable over time, suggesting that they are the product of either constraints that
network members’ are working under or those members preferences and (4)
notwithstanding this stability, we see a steady increase in degree and closeness
centralisation and a tendency for those who have been in the network over a longer
period of time to form a core within the network.
These findings contribute to our understanding of the history and organisation or PIRA.
In addition, however, we hope that they generate questions for further analysis both of
PIRA networks and of movement networks more generally. Furthermore, we hope that
they serve as an example of how network dynamics might be analysed in cases where the
assumptions of conventional modelling methods are not met. In part, our methodological
approach has involved a simple comparison of network properties over time, using this as
a way to generate hypotheses which we have then been able to test. We have, in other
words and in contrast to the usual (deductive) models, approached change inductively.
Furthermore, qualitative and substantive understanding of both the biographical
trajectories of actors in the networks and the wider sociopolitical context within which
the network was operating have been crucial. Real world historical networks, centred upon
collective action, are subject to multiple constraints and opportunities, at various levels,
from the personal lives of each of those involved to the historical stage upon which they
collectively act. Much that impacts upon the network happens ‘outside’ of it, so to speak,
and we must have a strong qualitative understanding of this environing context if we are to
analyse the network adequately. In conjunction with this approach, we advocate the use of
events participation data collected from secondary sources as a way of identifying and
assessing the leadership of a covert social movement group.

Notes
1. Adams (1997), Anderson (2003), Clarke and Johnston (2007), Collins (with McGovern) (1997), MacStiofain
(1975), Maguire (1973), O’Callaghan (1999), Sharrock and Devenport (1997) and White (2006).
2. Bowyer-Bell (1997), Bishop and Mallie (1988), Coogan (2000), Dillon (1991), English (2005), Harnden
(1998), Moloney (2002), Taylor (1998), Toolis (2000) and White (1993).
3. Bradley and Feeney (2009), Fulton with Nally and Gallagher (2008), Gilmour (1998), Maguire (2006) and
McGartland (2009).
4. Sharrock and Devenport (1997, p. 201) state that events from Sean O’Callaghan’s autobiography, refuted by
leading Republican’s, were later corroborated by other movement members and even, in one case, the former
Irish Prime Minister, Garrett Fitzgerald.
5. Two hundred and fifty-five actors initially met the inclusion criteria, and minus six of the actors in the separate
component in t4, who were only present in this time period and no other, leaves 249 actors in the network. The
remaining two actors in the t4 component and the two in the t3 dyad, which was also removed from analysis,
were present in other time periods and thus remain in the network.
6. Network data do not conform to the assumptions of many contemporary statistical methods (especially case-
wise independence) but Ucinet has a range of alternatives which circumvent these problems.
7. Betweeness centralisation varies in an erratic manner that we are unable to explain.
8. At the 0.05 (t3) and 0.01 (t4) levels, respectively.

References
Adams, G. (1997). Before the Dawn. London: Vintage.

Anderson, B. (2003). Joe Cahill: A life in the IRA. Dublin: O Brien Press.

Andrews, K., & Biggs, M. (2006). The dynamics of protest diffusion. American Sociological Review, 71,

752 – 777.
Change in covert social movement networks 89

Alonso, R. (2006). The IRA and armed struggle. London: (Cass series on political violence) Routledge.
Biggs, M. (2006). Who joined the sit ins and why? Mobilization, 11, 241 – 256.
Bishop, P., & Mallie, E. (1988). The Provisional IRA. London: Corgi.
Bollobás, B. (1985). Random graphs. New York: Academic Press.
Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002). Ucinet for Windows. Harvard, MA: Analytic
Technologies.
Bosi, L. (2008). Explaining the emergence process of the civil rights protest in Northern Ireland. Journal of
Historical Sociology, 2, 342 – 371.
Bosi, L. (2012). Explaining pathways to armed activism in the Provisional Irish Republican Army, 1969 – 1972.
Social Science History, 36, 347 – 390.
Bowyer-Bell, J. B. (1970). The secret army: A history of the IRA, 1916 – 1970. London: Blond.
Bowyer-Bell, J. (1997). The secret army: The IRA. New Brunswick, NJ: Transaction.
Bradley, G., & Feeney, B. (2009). Insider: Gerry Bradley’s life in the IRA. Dublin: O Brien Press.
CAIN (Conflict Archive on the Internet). (1968 – Present). Conflict and politics in Northern Ireland. Retrieved
from http://cain.ulst.ac.uk
Carley, K., Dombroski, M., Tsvetovat, M., Reminga, J., & Kamneva, N. (2003). Destabilising dynamic covert
networks. Proceedings of the 8th International Command and Control Research and Technology
Symposium. Conference held at the National Defense War College, Washington, DC. Evidence Based
Research, Vienna, VA.
Clarke, L., & Johnston, K. (2007). Martin McGuinness: From guns to government. Edinburgh: Mainstream
Publishing Company.
Coogan, T. P. (2000). The I.R.A. London: Harper Collins.
Coleman, J. (1988). Free riders and zealots. Sociological Theory, 6, 52–57.
Collins, E. (with McGovern, M.) (1997). Killing rage. London: Granta.
Crenshaw, M. (1992). Decisions to use terrorism: Psychological constraints on instrumental reasoning. In
D. Della Porta (Ed.), Social movements and violence: Participation in underground organisations
(pp. 29 –42). Bingley: JAI Press.
Crossley, N. (2006). Contesting psychiatry. Routledge: London.
Crossley, N. (2007). Social networks and extra-parliamentary politics. Sociology Compass, 1, 222– 236.
Crossley, N. (2010). The social world of the network: Qualitative aspects of network analysis. Sociologica,
Retrieved from http://www.sociologica.mulino.it/main
Crossley, N., Edwards, G., Harries, E., & Stevenson, R. (2012). Covert social movement networks and the
secrecy-efficiency trade off: The case of the UK Suffragettes (1906 – 1914). Social Networks, 34, 634– 644.
Crossley, N., & Ibrahim, J. (2012). Critical mass, social networks and collective action: The case of student
political worlds. Sociology, 46, 596 – 612.
Della Porta, D. (Ed.). (1992). Social movements and violence: Participation in underground organisations.
Bingley: JAI Press.
Della Porta, D. (2006). Social movements, political violence and the state. Cambridge: Cambridge University
Press.
Diani, M. (1990). The network structure of the Italian ecology movement. Social Science Information, 29, 5–31.
Diani, M. (1995). Green networks. Edinburgh: Edinburgh University Press.
Diani, M., & McAdam, D. (2003). Social movements and networks. Oxford: Oxford University Press.
Dillon, M. (1991). The Dirty War. London: Arrow.
Doreian, P., & Snijders, T. (2010). Dynamics of social networks I, Special Issue of Social Networks, 32(1).
Doreian, P., & Snijders, T. (2012). Dynamics of social networks II, Special Issue of Social Networks, 34(3).
Doreian, P., & Stokman, F. (Eds.). (2005). Evolution of social networks. London: Routledge.
Edwards, G., & Crossley, N. (2009). Measures and meanings: Exploring the ego-net of Helen Kirkpatrick Watts,
militant suffragette. Methodological Innovations Online, 4.
Emirbayer, M., & Goodwin, J. (1994). Network analysis, culture and the problem of agency. American Journal of
Sociology, 99, 1411 – 1454.
Enders, W., & Su, X. (2007). Rational terrorists and optimal network structure. The Journal of Conflict
Resolution, 51, 33–57.
English, R. (2005). Armed struggle: The history of the IRA. Oxford: Oxford University Press.
Erikson, B. (1981). Secret societies and social structure. Social Forces, 60, 188 – 210.
Feld, S. L. (1981). The focused organization of social ties. American journal of sociology, 86, 1015 – 1035.
90 R. Stevenson & N. Crossley

Fernandez, R., & McAdam, D. (1988). Social networks and social movements. Sociological Forum, 3,
357 – 382.
Fulton, K., with Nally, J., & Gallagher, I. (2008). Unsung hero. London: John Blake Publishing.
Gilmour, R. (1998). Dead ground: Infiltrating the IRA. London: Little, Brown and Company.
Gould, R. (1991). Multiple networks and mobilisation in the Paris commune, 1871. American Sociological
Review, 56, 716– 729.
Gould, R. (1993). Collective action and network structure. American Sociological Review, 58, 182– 196.
Gould, R. (1995). Insurgent identities. Chicago, IL: Chicago University Press.
Harnden, T. (1998). Bandit country: The IRA in South Armagh. London: Hodder and Stoughton.
Hedström, P. (1994). Contagious collectivities. American Journal of Sociology, 99, 1157 – 1179.
Hedström, P., Sandel, R., & Stern, C. (2000). Mesolevel networks and the diffusion of social movements.
American Journal of Sociology, 106, 145 – 172.
Horgan, J. (2008). Interviewing terrorists: A case for primary research. In H. Chen, E. Reid, J. Sinai, A. Silke,
& B. Ganor (Eds.), Terrorism informatics: Knowledge management and data mining for homeland security
(Integrated series in information systems). (Chap. 4, pp. 73 – 100) New York, NY: Springer.
Horgan, J., & Taylor, M. (1997). The Provisional Irish Republican Army: Command and functional structure.
Terrorism and Political Violence, 9(3), 1 – 32.
Jasper, J. M. (1998). The art of moral protest. Chicago, IL: Chicago University Press.
Jasper, M. J., & Poulsen, J. D. (1995). Recruiting strangers and friends: Moral shocks and social networks in
Animal Rights and Anti-Nuclear protests. Social Problems, 42, 493– 512.
Kitts, J. (2000). Mobilizing in black boxes. Mobilization, 5, 241– 258.
Lindelauf, R., Borm, P., & Hamers, H. (2009). The influence of secrecy on the communication structure of covert
networks. Social Networks, 31, 126– 137.
MacStiofain, S. (1975). Memoirs of a revolutionary. London: Gordon Cremonesi.
Maguire, E. (2006). Enemy of the empire. Life as an international undercover IRA activist. Dublin: O Brien Press.
Maguire, M. (1973). To take arms: A year in the Provisional IRA. London: Quartet Books.
Martin, A. (2012). Explosive troubles interviews set to surface. BBC News Online, April 4. Retrieved from http://
www.bbc.co.uk/news/uk-northern-ireland-17604972
Marwell, G., & Oliver, P. (1993). The critical mass in collective action. Cambridge: Cambridge University Press.
McAdam, D. (1982). Political process and the development of black insurgency. Chicago, IL: University of
Chicago Press.
McAdam, D. (1983). Tactical innovation and the pace of insurgency. American Sociological Review, 48,
735 – 754.
McAdam, D. (1986). Recruitment to high risk activism. American Journal of Sociology, 92, 64–90.
McAdam, D., & Paulsen, R. (1993). Specifying the relationship between ties and activism. American Journal of
Sociology, 99, 640– 667.
McGartland, M. (2009). Fifty dead men walking. London: John Blake Publishing.
Mische, A. (2003). Cross-talk in movements. In M. Diani & D. McAdam (Eds.), Social movements and networks
(pp. 258 – 280). Oxford: Oxford University Press.
Moloney, E. (2002). A secret history of the IRA. London: Norton and Company.
Morselli, C., Giguère, C., & Petit, K. (2007). The efficiency/security trade-off in criminal networks. Social
Networks, 29, 143– 153.
O’Callaghan, S. (1999). The informer. London: Corgi.
O’Dochartaigh, N., & Bosi, L. (2010). Territoriality and mobilisation. Mobilisation, 15, 404– 425.
Ohlemacher, T. (1996). Bridging people and protest. Social Problems, 43, 197– 218.
O’Leary, J. (2012). Martin McGuinness disturbed by Boston College project. BBC News Online. January 13.
Retrieved from http://www.bbc.co.uk/news/uk-northern-ireland-16553214
Oliver, P. E., & Myers, D. J. (2003). Networks, diffusion, and cycles of collective action. In M. Diani & D.
McAdam (Eds.), Social movements and networks: Relational approaches to collective action (pp. 173 – 203).
New York: Oxford University Press.
Passy, F. (2001). Socialisation, connection and the structure/agency gap. Mobilization, 6, 173 – 192.
Passy, F. (2003). Social networks matter. But how? In M. Diani & D. McAdam (Eds.), Social movements and
networks (pp. 21 –46). Oxford: Oxford University Press.
Pfaff, S. (1996). Collective identity and informal groups in revolutionary mobilisation. Social Forces, 75,
91– 117.
Change in covert social movement networks 91

PSNI Legal Bid for Boston College Tapes – “Not the Way to the Truth”. (2102). BBC News Online. January 15.
Retrieved from http://www.bbc.co.uk/news/uk-northern-ireland-16566598
Sandel, R. (1999). Organisational life aboard a moving Bandwagon. Acta Sociologica, 42, 3 –15.
Saunders, C. (2007). Comparing environmental movement networks in periods of latency and visibility.
Graduate Journal of Social Science, 4, 109– 139.
Sharrock, D., & Devenport, M. (1997). Man of war, man of peace. The unauthorised biography of Gerry Adams.
London: Macmillan.
Snijders, T. (2011). Network dynamics. In J. Scott & P. Carrington (Eds.), Sage handbook of social network
analysis (pp. 501 – 513). London: Sage.
Snow, D., Zurcher, L., & Ekland-Olson, S. (1980). Social networks and social movements. American
Sociological Review, 45, 787 – 801.
Stohl, M. (2008). Networks, terrorists and criminals: The implications for community policing. Crime, Law,
Social Change, 50, 59–72.
Taylor, P. (1998). Provos: The IRA and Sinn Fein. London: Bloomsbury.
Toolis, K. (2000). Rebel Hearts: Journeys within the IRA’s soul. London: Picador.
White, R. (1989). From peaceful protest to Guerrilla War. American Journal of Sociology, 94, 1277 – 1302.
White, R. (1993). Provisional Irish Republicans: An oral and interpretive history. Westport, CT: Greenwood
Press.
White, R., & Falkenberg-White, T. (1991). Revolution in the city. Terrorism and Political Violence, 3, 100– 132.
White, R. W. (2006). Ruairi O’Bradaigh: The life and politics of an Irish revolutionary. Bloomington, IN:
Indiana University Press.

Rachel Stevenson is a PhD student in the Department of Sociology at the University of


Manchester. Her Doctoral research explores the structures of covert social movement
networks.

Nick Crossley is Professor of Sociology at the University of Manchester. His most recent
book is Towards Relational Sociology (Routledge 2011). He is currently completing a
book on punk and post-punk networks in the UK and is about to embark on a Leverhulme
funded project on covert networks.
Copyright of Social Movement Studies is the property of Routledge and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.

You might also like