Policy Implications from Stathis Kalyvas’ Concepts of Attitudinal and Behavioural Support
by James Khalil
It is often claimed that local ‘support’ is a critical asset during insurgent-COIN campaigns. Below this apparent consensus, however, authorities such as Lawrence ‘of Arabia’ and Mao refer primarily to voluntary preferences (i.e. ‘attitudinal support’), on the usually implicit assumption that this asset delivers sufficient tangible benefits (i.e. ‘collaboration’ or ‘behavioural support’) to the belligerents in question. In contrast, the likes of Kalyvas and Kilcullen focus directly upon behavioural support, placing emphasis upon the role of control and authority. Taking issue with both of these stances, a core point of this paper is that the drivers of collaboration are diverse, with the role of attitudinal support varying substantially between and within cases. Furthermore, the above authorities are all equally guilty of overlooking that the strength of this relationship also depends upon the form of collaboration in question. For instance, while insurgents may be able to coerce food from a population, their ability to gain information upon the whereabouts of COIN forces is likely to depend to a greater extent upon supportive attitudes. Thus, it is necessary to focus research upon attitudinal and behavioral support as two distinct, but interrelated, assets. Rather than representing academic navel-gazing, the insights from such investigations have key implications for policy, determining the extent to which COIN forces should privilege efforts to undermine sympathy for the insurgents, provide security to coerced populations, target insurgent channels of funding, and so on, within each specific environment in question.
The Importance of Conceptual Clarity
It is frequently asserted that local support is a critical asset during insurgent-counterinsurgent (COIN) campaigns, but, as expressed by Stathis Kalyvas, ‘below this unanimity lies extreme confusion, for there are two very different ways of thinking about support.’ Specifically, the term may be used to refer to voluntary preferences gained from the community (subsequently referred to as ‘attitudinal support’), or actions and inactions that provide tangible benefits to specific belligerents (subsequently referred to as ‘behavioural support’ or ‘collaboration’). The latter encompasses the supply of various goods (food, clothing, finances, and so on), and services (from provisions of shelter and information, up to and including enlistment). It is possible, or even common, for individuals to provide behavioral support to multiple belligerents almost simultaneously.
‘[Guerrillas] must have a friendly population, not actively friendly, but sympathetic to the point of not betraying rebel movements to the enemy.’ (T. E. Lawrence ‘of Arabia’)
‘Because guerrilla warfare basically derives from the masses and is supported by them, it can neither exist nor flourish if it separates itself from their sympathies and cooperation.’ (Mao Tse-tung)
‘The guerrilla’s major asset is non-military and without it he is helpless: he must have the sympathy and support, active and passive, of the local population. Any Robin Hood who loses it is dead, and so is any guerrilla.’ (Eric Hobsbawm)
Box 1: Accounts privileging attitudinal support
A number of prominent academics and practitioners have stressed the need for insurgents and COIN forces to obtain attitudinal support on the often implicit assumption that behavioral support is largely driven by attitudes that are sympathetic, or, at least, non-hostile (see Box 1). In contrast, other authorities (see Box 2) focus directly upon collaboration through emphasizing the extent to which behavioral support may be gained from unsympathetic or even hostile populations, often highlighting the role of authority and coercion. Taking issue with both of these stances, a core point of this paper is that the drivers of collaboration are diverse, and that the role played by attitudinal support varies significantly between and within cases. Furthermore, these authorities are all equally guilty of overlooking that the strength of this relationship also depends upon the form of collaboration in question. As noted by Elisabeth Jean Wood regarding the case of El Salvador, for instance, ‘sustained flows of high-quality information are much more difficult to extract coercively than tortillas or water,’ suggesting that attitudinal support is less critical for the latter.
‘Emphasis should be placed upon behavior, rather than attitudes and sympathies. Attitudes, in the sense of preferences, affect behavior but are not identical with it; nor in most cases are they the primary influence on it.’ (Nathan Leites & Charles Wolf Jr.)
‘Contrary to a widespread perception that irregular wars are merely contests for ‘hearts and minds’, they can be seen primarily as a competition for territorial control, where violence is used to challenge and to create order. To be sure, both incumbents and insurgent actors must generate popular ‘collaboration’ and deter ‘defection’. However, the extent of collaboration they can achieve hinges largely on the degree of control they are able to exercise.’ (Stathis Kalyvas & Mathew Kocher)
‘Field experience in both Afghanistan and Iraq … have shown that insurgent intimidation easily overcomes any residual gratitude effect, while historical studies have shown that in civil wars and insurgencies, popular support tends to accrue to locally powerful actors rather than to those actors the population sees as more congenial: the more organized, locally present, and better armed a group is, the more likely it is to be able to enforce a system of rules and sanctions.’ (David Kilcullen)
Box 2: Accounts privileging behavioural support
Rather than representing academic navel-gazing, the insights from research into the relationship between these two forms of support have significant COIN policy implications. For instance, if a given population is sympathetic towards an insurgent organization, and individuals collaborate largely to gain status, respect and honor from the community, then the primary COIN focus must be upon undermining this attitudinal support. However, if it is found that insurgents tend to coerce behavioral support from a specific group, provision of security will also be of key importance. Alternatively, in locations where individuals are often enticed to collaborate through material incentives (e.g. payments for IED emplacements, salaries for cadre, and so on), specific efforts must be made to cut incoming insurgent finances, and livelihood initiatives should simultaneously be prioritized. Much of the evidence cited in this paper is drawn from a specifically designed field research program undertaken by the author into the relationship between attitudinal and behavioural support during the ‘People’s War’ in Nepal, with secondary evidence gained from locations as distinct as Iraq, Afghanistan, Lebanon and Peru.
How Insurgents Generate Attitudinal Support
With insurgent approaches to gaining attitudinal support varying considerably between and within cases, the purpose of this first section is to briefly review a limited number of frequently reoccurring themes. The focus is primarily upon support for insurgents, rather than COIN forces, as the bulk of the empirical evidence lies with the former. However, many of the underlying principles are the same, for instance, with the insurgents and COIN forces equally capable of losing attitudinal support through the use of excessive force, with the Iraqi Al Qa’ida franchise and various instances of COIN ‘collateral damage’ (that despicable term) in Afghanistan coming readily to mind. Sympathy for the insurgents is often driven by a range of ‘structural’ factors (i.e. ones that can loosely be considered to be beyond the immediate control of the belligerent actors involved), such as poverty, inequality between religious, ethnic, tribal or clan groups, economic crises, landholding patterns, a lack of political ‘voice’, urbanization, ‘youth bulges’, and so on. While such factors are indisputably critical, they can only ever provide a partial explanation, and a specific focus must also be placed upon how insurgents successfully channel the associated grievances.
Many insurgent groups gain considerable attitudinal support through what amounts to alternative governance in locations where the state has limited presence. Numerous organizations provide local justice, for instance, and while such systems are invariably grounded on the specific worldview of the group in question, they often serve to generate support from certain sectors through being more accessible, more efficient, less expensive, and often less corrupt than state provisions. In the case of Peru, for instance, authorities maintain that this alternative justice ‘gave a certain legitimacy’ to the Shining Path, provoked ‘a certain diffuse sympathy,’ and was ‘viewed positively by the majority.’ Many insurgents also deliver welfare and development to local populations. For instance, David Kilcullen notes that Hezbollah has ‘charities that will help you if you are poor, and they can get you a job, and teach your children in theirs schools, and treat you the in their hospital if you are sick.’ Similarly, while much media attention focuses upon the violence perpetrated by Hamas, prior to their 2006 election victory it is reported that most of their resources were channeled towards ‘the social and welfare programs that the movement provides to the Palestinians.’
Attitudinal support may also be gained if the violence perpetrated by the insurgents, against repressive state forces, members of other religious or ethnic groups, local ‘exploitative’ elites, and so on, provides a sense of retribution. Commenting upon the Philippine case, for instance, Richard Kessler claims that a critical strength of the New People’s Army was ‘in its ability to carry out carefully chosen assassinations, which, until 1985, won it considerable support.’ In contrast, however, various respondents in Nepal asserted that only a fraction of the populace supported similar violence against the targeted ‘enemies of the people’. Not all attitudinal support is generated through actions, however, in the sense that rhetoric alone often plays a prominent role. For instance, the Maoists were viewed in a positive light by many through being the most prominent advocates of republicanism, particularly towards the latter stages of their conflict as palace rule became increasingly dictatorial and repressive. Anti-US rhetoric has also often played an important role, in cases as distinct as Nepal, Vietnam and Afghanistan.
How Insurgents Generate Behavioural Support
In shifting the attention from attitudinal support to collaboration it is worth reiterating the key point that the extent to which the latter is driven by the former is contested, with authorities such as Lawrence and Mao implicitly disagreeing with Kalyvas and Kilcullen. Attributing importance to the drivers of behavioral support is a complex task, and it is necessary to draw insight from rational choice theory (RCT). At the center of RCT-influenced research into armed conflict is the free-rider problem and its core conclusion that ‘rational’ individuals abstain from participation in such endeavors as they are equally able to benefit from the rewards irrespective of their actions. On this basis the focus shifts from collective benefits (e.g. revolution, independence, redressing grievances of religious or tribal groups, and so on) towards incentives that serve to motivate on an individual basis. For the purposes of the subsequent discussion the identified drivers have been clustered under the headings of economic, security and socio-psychological (as summarized in Figure 1).
The application of RCT is not universally welcomed, and a common critique targets the core assumption that individuals are self-interested. However, while narrow versions of the RCT framework are restricted to the realms of economics and security, ‘thick’ variants also incorporate socio-psychological factors. Thus, critically, individuals acting in accordance with perceived group aims may be treated as being ‘rational’ if their motives in doing so are to gain respect or to avoid ostracism. Arguably, this proves to be of particular relevance for insurgency given that such conflicts often occur in ‘traditional’ societies where individuals have a greater tendency to act in pursuit of community ends. However, as shall become apparent below, the incorporation of socio-psychological factors does serve to stretch RCT in that such variables are often extremely difficult to define and measure in such environments.
Figure 1: Motivators often associated with behavioral support for insurgents
It is also frequently observed that the interests pursued by ‘rational’ individuals are not universal, but are contingent upon context and culture. For instance, the extent to which material gain is pursued varies considerably between and within communities. Indeed, at a broader level this represents a significant limitation as RCT in its present form can only tentatively explain why drivers from security through to status inconsistently motivate. While this does undermine efforts to reach a universal understanding of human motivation, however, it does not prevent comparative conclusion of immediate practical value to COIN policy. For instance, research may find that specific religious, ethnic or tribal groups are more consistently incentivized by Taliban financial enticements than others, suggesting a need to undermine this driver through investing in economic programs in their areas of residence. The following subsection focuses upon such material incentives, and this is followed by a consideration of security and socio-psychological drivers.
The very act of providing behavioral support to insurgents often entails an economic cost in the form of food or finances, or at least an ‘opportunity cost’ associated with contributions of time. On the flip side of the coin, however, the historical record shows that a wide variety of selective material benefits have been used by insurgents to gain behavioral support. For instance, during their respective conflicts the Chinese Community Party, the Peruvian Shining Path, and the Nepali Maoists all reportedly provided grain specifically to entice ‘good behavior’. Perhaps the most well-known example, however, is observed by Jeffrey Race, who claims that in his research locations in Vietnam ‘each beneficiary of land redistribution retained his land only as long as he did not oppose the revolutionary movement, and indeed only as long as he assisted in required ways.’ A more recent and ‘kinetic’ example is offered by Antonio Giustozzi from the London School of Economics, who reports that the Taliban pay villagers US$15-55 to conduct targeted assassinations or to fire rockets at enemy bases.
Additionally, the very specific act of enlisting into the insurgent ranks is commonly driven by two further material enticements. Firstly, insurgent cadre may be provided with salaries (or, in certain cases, ‘payments in kind’), for instance, as was offered by the Provisional Irish Republican Army. Marcella Ribetti also observes this phenomenon in Colombia, adding that ‘material incentives for the lower ranking guerrillas are more often necessities, rather than luxuries.’ Yet, this author additionally claims that ‘joining the guerrillas seemed a good option, because it offered to provide for all their material needs in exchange for seemingly easy tasks.’ Maoist cadre in Nepal reportedly also earned in the region of 300 or 500 Rupees per month, although given that many of their contemporaries sought and often gained superior economic opportunities in India, for instance, the extent to which this drove enlistment is debatable.
Secondly, individuals may be encouraged to enlist because once within the insurgent ranks they are frequently provided with the opportunity to embezzle money intended for the campaign, or to otherwise make exploitative gains, as Ribetti also notes with regard to the Colombian case. Unsurprisingly, the Nepali Maoists interviewed on the subject tended to focus upon the internal monitoring systems utilized to prevent such activities during their ‘People’s War’, and maintained that the individuals caught were punished. However, many respondents from the opposing political parties and those lacking affiliations contradicted these claims, often referring to specific cases of reported embezzlement. Similar motives have been reported in cases from Iraq to Sierra Leone, and the reported ‘criminalization’ of insurgency forms a prominent line of enquiry in recent conflict studies literature. This is perhaps best expressed by David Keen, who maintains that ‘to paraphrase Carl von Clausewitz, war has increasingly become the continuation of economics by other means,’ adding that ‘war is not simply a breakdown in a particular system, but a way of creating an alternative system of profit.’
In turning the attention to security matters, it is necessary to comment upon three main selective motivators. Firstly, the threat of apprehension or physical targeting by COIN forces offers an actual or potential cost that may serve to dissuade individuals from providing behavioral support to insurgents. These COIN acts may be contingent upon the behaviours of the targeted individuals (referred to by Stathis Kalyvas as ‘selective violence’), and thus dissuade directly. In practice, however, they are also often based upon simple ‘guilt by association’ or ‘an unwillingness or failure to discriminate’ (referred to as ‘indiscriminate violence’), in which case they work, theoretically at least, through socio-psychological disincentives (as discussed shortly). In general terms, indiscriminate violence is considered to be more common amongst COIN forces, as insurgents tend to have improved local knowledge through which they are able to make their violence selective.
Secondly, and on the benefits side of the calculation, insurgents may be able to offer protection selectively to their supporters against this very violence perpetrated by the state. In practice, however, this is intimately connected to the relative capabilities of the armed adversaries in the location in question, and failures to provide protection of any form have been reported in cases as distinct as Zimbabwe, Peru and Nepal. Referring to Farabundo Marti de Liberación Nacional in El Salvador, Elisabeth Jean Wood similarly notes that ‘during most of the war, the FMLN offered little protection from government forces in the case-study areas.’ Indeed, specifically with the free-rider hurdle in mind, Wood goes on to maintain that ‘protection per se does not explain the ongoing participation of those who continued to support the insurgency.’ However, individuals may also be ‘protected’ from COIN violence through absconding with the insurgents, and in certain contexts this may function as a selective incentive for the specific act of enlistment.
Finally, insurgents may provide a benefit in the sense of abstaining from violence against collaborators. Or, put another way, insurgents ‘encourage’ participation though targeting those who do not provide this support. For instance, Ismet Imset’s notes that the locals in contact with the Partiya Karkerên Kurdistan (PKK) were aware that the ‘attacks were directed not at ordinary people but villagers with state connections, who agreed to collaborate against the Kurds although they themselves were Kurds.’ This author continues that ‘the message was that any family who dealt with the state would be destroyed.’ Similarly, the Nepali Maoists relied upon threats and actual violence to combat ‘anti-people activities’ such as a refusal to ‘contribute’ to the cause. Over half of the 700 sampled respondents from the provinces of Pyuthan, Dang and Gorkha asserted that they collaborated on the bases of this coercion, representing a particularly high figure given that the Maoists retained their influence in these locations at the time of the investigation.
While RCT analysts may gravitate towards the above economic and security-based explanations, the ‘father’ of the free-rider problem, Mancur Olson, himself notes that socio-psychological stimuli ‘are among the kinds of incentives that may be used to mobilize a latent group.’ As previously observed, however, the difficulty is often that such variables prove to be particularly difficult to adequately define and measure, especially in the context of armed conflict. While only a few such drivers are considered here for the sake of brevity, Stathis Kalyvas identifies ‘curiosity and the prospect of excitement and adventure, the lure of danger, the acquisition of a new and more rewarding individual identity or moral worldview, [and] the pleasure of acting as one’s own agent,’ alongside negative motivators such as ‘anger, moral outrage caused by public humiliation, and the desire to take revenge.’ As with the material and security incentives discussed above, these socio-psychological drivers must motivate selectively, and in this sense ‘revenge’ is distinct from ‘retribution’ (as this term was previously applied in the context of attitudinal support) in that the latter may be gained by an entire community, while the former is experienced only by participants.
Focusing, firstly, upon ‘adventure’ as a driver, Marcella Ribetti asserts that in the case of Colombia ‘combat appears to have been an exciting experience form many.’ This factor is also identified by Kilcullen, who maintains that local farmers assisted the Taliban during a 2006 confrontation on the grounds that ‘this was the most exciting thing that had happened in their valley in years.’ With the opportunity to benefit from such ‘adventures’ open to many, however, such explanations fail to separate the collaborators from those remaining on the side-lines. Yet, as observed above, this limitation also applies to other drivers, posing questions as to why only certain individuals provided with the opportunity elect to plant IEDs in exchange for cash, collaborate in exchange for land, or chose to abscond with insurgents in order to benefit from this ‘protection’. While this limitation is significant it does not prevent conclusions that are of immediate value to COIN policy-making.
Perhaps more centrally to the current discussion, however, are community provisions of status, and related concepts such as honor and prestige. In the case of Northern Ireland, for instance, Eamon Collins observed that amongst certain communities ‘IRA men have considerable status, and for those Provos who look for sexual advantage from it, there is no shortage of women willing to give more than the time of day to IRA volunteers.’ Conversely, Roger Petersen notes that ‘when a majority is cooperating in a resistance effort, those who remain on the sidelines will receive the contempt of their fellows.’ It is at this critical juncture that the concept of attitudinal support remerges, providing an important determinant of whether communities elect to bestow such status and contempt. Put another way, while attitudinal support may not drive collaboration directly (failing on the free-rider hurdle), it encourages such activities indirectly through provoking communities to provide status, honor and prestige.
The concept of indiscriminate violence may also return at this point, however, as if applied by COIN forces it may trump the effects of attitudinal support, and actually prevent communities from encouraging such acts. As expressed by Kalyvas ‘the logic of indiscriminate violence assumes civilians to be able to lobby armed actors to decrease the level of their activities,’ and ‘this requires that civilians have access and influence on armed actors and, conversely, armed actors care about civilians.’ However, Kalyvas goes on to suggest that indiscriminate violence largely fails to achieve its objectives, amongst other reasons, because it overestimates the above linkage. Indeed, while this is an area in particular need of additional research, this violence may even backfire through driving communities to bestow additional status and prestige upon the collaborators who seek to provide retribution. Seemingly on the basis of this dynamic, Arjun Karki and David Seddon claim that indiscriminate police violence ‘resulted in a substantial proportion of the local population making common cause with the Maoists, and the mid-west was effectively confirmed as a Maoist heartland.’
So, who were right? Were Lawrence, Mao and Hobsbawm correct to focus attention upon attitudes, implying that sympathy translates with sufficient consistency into collaboration? Or, were Kalyvas and Kocher, and Kilcullen correct to place the emphasis directly upon behaviors, stressing the role of control and authority? Unfortunately there are no simple answers to such questions, firstly, because the motives are very diverse, and vary considerably between locations and over time. And, secondly, as previously observed, attempts to draw such broad conclusions are undermined by a reality that it depends to a large extent upon the type of collaboration in question. Put another way, attitudinal support may be critical for obtaining certain forms of behavioral support, such as information on the whereabouts of the COIN forces, but much less so for others, including food and shelter. Thus, in each field location is it necessary to focus upon attitudinal and behavioral support as distinct, but interrelated, variables.
Undermining Attitudinal Support for Insurgents
While insurgent-COIN conflicts are commonly characterized as a competition between two (or more) sides for support from the populace, in practice they often more closely resemble a struggle to lure reluctant community members ‘off the fence,’ and thus loses in attitudinal support for one set of belligerents do not necessarily imply gains for their competitors. Nevertheless, reducing attitudinal support for the insurgents is of clear benefit for the COIN forces, and this necessitates targeting the specific channels used to generate this asset. For instance, if a specific group gains sympathy through provisions of welfare, this may be undermined through targeting their sources of funds, or enabling the state to absorb greater responsibility for such provisions. If they generate attitudinal support through providing alternative systems of justice, then the COIN forces should invest in making the formal processes more accessible, affordable, efficient, and / or less corrupt. If sympathy is gained as the insurgents seek to redress ethnic or tribal grievances, then, to the extent that these are genuine, COIN policy should aim to achieve the same.
Undermining Behavioral Support for Insurgents
Policy-decisions must also reflect realities regarding behavioral support, and in the first instance this involves understanding local motivators. If revenge is a substantial driver, then this points to a need to avoid future instances of ‘collateral damage’ and other forms of excessive violence. If it is found that individuals are motivated largely by gains in status, respect and honor, it is necessary to target the extent to which communities bestow such rewards, through prioritizing efforts to reduce attitudinal support for the insurgents, as described above. Or, if individuals are often coerced into these behaviors, then the COIN forces must additionally focus their attention upon security provisions. Where economic incentives (salaries for cadre, payments for IED emplacements, and so on) provide a common motivator, livelihood programs should be prioritized, with parallel attempts being made to undermine the insurgent ability to offer such funding. While COIN campaigns routinely involve each of the above, shifts in the prioritization of these activities are likely to make a substantial difference, particularly given that motivations vary considerably between cases, provinces, villages, and individuals.
 Stathis Kalyvas, Logic of Violence in Civil War (New York: Cambridge University Press, 2006), 92
 In contrast to the US-led forces in Iraq and Afghanistan, in other environments the COIN actors often draw upon all of the forms of behavioral support identified here.
 Lawrence (1928) cited in Richard Stubbs, Hearts and Minds in Guerrilla Warfare (Singapore: Eastern Universities Press, 2004) 2; Mao Tse-tung (1937), On Guerrilla Warfare (Chicago: University of Illinois Press, 1961), 44. Note that the term ‘guerrilla’ overlaps considerably with ‘insurgent’, with the former relating more directly to kinetic matters (i.e. guerrilla warfare) and generally implying a rural focus.
 Elisabeth Jean Wood, Insurgent Collective Action and Insurgency in El Salvador (Cambridge: Cambridge University Press, 2003), 156.
 Nathan Leites and Charles Wolf Jr., Rebellion and Authority (Chicago: Markham, 1970), 45; Stathis Kalyvas and Matthew A. Kocher, “The Dynamics of Violence in Vietnam,” Journal of Peace Research, 46:3 (2009) 339; David Kilcullen, Accidental Guerrilla (Oxford: Oxford University Press, 2009), 67-68
 The findings from this research are published in James Khalil, “Insurgent-Populace Relations in Nepal,” Small Wars and Insurgencies, 23:2 (2012) 221-244
 Ton de Wit and Vera Gianotten, “The Center’s Multiple Failures’”, in David Scott Palmer, Shining Path of Peru (New York: St. Martin’s Press Inc., 1992), 72; Ponciano del Pino, “Family, Culture, and Revolution” in Steve Stern (ed.) Shining and Other Paths (Durham and London: Duke University Press, 1998), 161; Lewis Taylor, Shining Path (Liverpool: Liverpool University Press, 2006), 128
 David Kilcullen, Counter Insurgency (London: Hurst & Company, 2010), 153
 Khaled Hroub, Hamas: A Beginner’s Guide (London: Pluto Press, 2006), 136-137
 Richard Kessler, Rebellion and Repression in the Philippines (New Haven and London: Yale University Press, 1989), 77
 Whilst outside of the scope of this brief paper, there is undoubtedly variance in the extent to which insurgents are reliant upon local collaboration. This largely depends upon the extent to which insurgents are able to gain sustenance from external state or non-state sources, or through market activities (most notoriously involving trade or manufacture of narcotics in certain cases). However, the local reaction against the Al Qa’ida franchise in Iraq during the ‘Awakening’ demonstrates the risk associated with creating attitudinal opposition.
 While the key inference remains that non-participation is the default option, the concept of ‘free-riding’ proves to be a misnomer in the context of conflict through implying universal attitudinal support. In most cases numerous individuals from the communities that the insurgents supposedly represent oppose the aims or methods applied by these belligerents, and as such their preference to withhold behavioural support is not an attempt to free-ride. Thus, the discussion in this essay is framed around the costs and benefits to individuals of providing behavioural support, rather than ‘solutions’ to the free-rider problem.
 Many critics are seemingly unaware that RCT is sufficiently elastic to incorporate such variables. As with all variables, however, it is necessary to identify and define these ex ante in order to avoid tautology. This is described, for instance, in Margaret Levi, “A Model, a Method, and a Map”, in Mark Irving Lichbach and Alan Zuckerman (eds.), Comparative Politics (Cambridge: Cambridge University Press, 1997), 19-41
 Critically, this translates into an inability to reliably understand, for instance, why certain individuals accept cash to place an IED, whereas others do not. However, important insights have emerged through, amongst other lines of enquiry, social identity theory and locus of control theory (with the latter overlapping considerably with the so-called ‘efficacy solution’ to the free-rider problem).
 Jeffery Race, War Comes to Long An (Berkeley and Los Angeles: University of California Press, 1972), 174
 Antonio Giustozzi, Koran, Kalashnikov, and Laptop (New York: Columbia University Press, 2008), 14 and 41
 Marcella Ribetti, “The Unveiled Motivations of Violence in Intra-State Conflicts”, Small Wars and Insurgencies 18/4 (2007), 710
 Ribetti, Unveiled Motivations, 711
 David Keen, Economic Functions of Violence in Civil Wars (USA: Adelphi Series, International Institute of Strategic Studies 1998), 11
 The exclusive reference to state violence in this section is for simplicity. Such campaigns are not always ‘two-player games’, and other belligerent groups often use violence in an effort to reduce collaboration with the insurgents in question.
 Stathis Kalyvas, “The Paradox of Terrorism in Civil War”, Journal of Ethics, 8 (2004) 101-103
 Wood, Insurgent Collective Action, 13
 This is discussed at length in Kalyvas and Kocher, Dynamics of Violence
 Cited in David Romano, Kurdish Nationalist Movement (Cambridge: Cambridge University Press, 2006), 86
 Information on research methods is provided in James Khalil, Insurgent-Populace Relations in Nepal: Attitudinal and Behavioural Support for the Maoists during the People's War, PhD Thesis, University of Leeds, UK.
 Mancur Olson, Logic of Collective Action (Cambridge: Harvard University Press, 1965), 61
 Kalyvas, Logic of Violence, 96-97
 Ribetti, Unveiled Motivations, 712
 Kilcullen, Accidental Guerrilla, 40-41
 Eamon Collins, cited in Kalyvas, Logic of Violence, 97
 Roger Petersen, Resistances and Rebellion (Cambridge: Cambridge University Press, 2001), 41-42
 Of course, this is not the only channel through which supportive attitudes may drive collaboration, as individuals may gain personal (internal) benefits through ‘doing the right thing’ (refered to by some as ‘personal normative rewards’). It is certainly possible to treat such individuals as being ‘rational’, but this motive stretches the RCT framework further in the sense of being even more difficult to define and measure than the socio-psychological variables in the text.
 Kalyvas, Logic of Violence, 158
 Arjun Karki and David Seddon, People’s War in Nepal (Delhi: Adroit Publishers 2003) 23
 See, for instance, Khalil, “Insurgent-Populace Relations”, 227