Introduction

In May 2010, hundreds of inner-city residents in Kingston, Jamaica marched to Gordon House, the island's seat of parliament, demonstrating against the extradition of Christopher "Dudus" Coke. Dudus, one of the island's most prominent "dons" (as area leaders associated with criminal organizations are commonly known), was wanted by the United States on drugs and weapons trafficking charges. Soon after the protest march, gunmen associated with Dudus took a different strategy to protect their leader and barricaded the entrances to Tivoli Gardens, his neighborhood strong-hold. A state of emergency was declared, and the Jamaican security forces killed at least 73 civilians in the process of "recapturing" the neighborhood, while Dudus was only arrested a month later and is now serving 23 years in a US federal jail. Similarly, as Brazil prepares for the Olympic Games and previously for the World Cup, campaigns by Police Pacification Units (UPPS) to recapture favelas and reestablish state control feature regularly in the international news media. Reports on confrontations between security forces and criminal organizations tend to stress that the challenge for security forces is not only to conquer the territory by force, but also to win over the population. Commentators frequently explain the support that inner-city residents display for criminal leaders such as Dudus by pointing towards the public and private goods they provide. Indeed, in many cities across the world, criminal organizations have taken on many of the functions traditionally associated with the state, providing crucial public goods such as social welfare, security and dispute resolution to the urban poor. In this article, we seek to understand the citizens' perceptions of criminal organizations in relationship to their perceptions of state actors and institutions, comparing support for "irregular" and "formal" governance actors.

Over the last decade or two, researchers have increasingly begun to pay attention to the role that non-state armed actors play in urban governance (e.g.^{Davis, 2010}; ^{Jaffe, 2013}; ^{Arias and Goldstein, 2010}). While political scientists had previously tended to focus on rural guerilla movements and insurgents, there is now a wealth of empirical evidence that groups wielding "coercive capacity that either parallels or challenges that held by the state" exist in many Latin American and Caribbean cities (Davis, 2010, p. 398). The appearance of such actors in urban areas has often been associated with the changing role of the state in the aftermath of neoliberal reforms and they are often seen as emerging in contexts in which the state fails to provide adequate public goods and services (^{O'Donnell, 1993}; ^{Goldstein, 2010}; Jaffe and Aguiar, 2012). Where the justice system is inaccessible to residents confronted with rampant criminal violence, non-state armed groups are often the only actors able and willing to provide a modicum of security and dispute resolution. In addition to wielding coercive capacity, however, these groups also provide basic social services, such as access to welfare, the organization of solid waste management and even the construction of public green spaces. The relationship between non-state armed actors and the formal state often oscillates between competition and collusion (^{Auyero, 2007}). On the one hand, non-state armed actors contribute to maintaining order and therefore take on a valued if not always overt role as partners in urban governance. On the other hand, however, groups financing their activities at least in part through an involvement in organized, often transnational crime pursue goals that put them at odds with the formal state (^{Heyman, 1999}; Arias, 2006).

Building on ^{Davis' (2010)} conceptualization of "irregular armed forces", this article conceptualizes gangs, militias and other extra-state armed groups engaged in organized violence, that provide public and private goods in specific territories yet do not seek to replace or overthrow the formal state apparatus, as "irregular governance actors". These irregular governance actors rely on the use of violence in maintaining power, and indeed are the source of much conflict, but they also play an important role in governing urban territories and populations. While our understanding of the complex empirical relationship between irregular and formal actors in urban governance has grown in recent years, we know relatively little about the way in which citizens view these arrangements, as public opinion research on different types of governance actors has remained scarce.

Journalistic accounts, but academic literature in security studies and anthropology as well, frequently highlight the support that the provision of much-needed public and private goods by irregular governance actors generates among residents. Such accounts portray the state as not only unable to exercise core functions effectively within pockets of its territory, but also suggest that, as a result, the state is confronted with a loss of residents' "hearts and minds" (^{Davis, 2010}; ^{Flanigan, 2014}; ^{Koonings and Kruijt, 2007}). This interpretation implies a competitive relationship in terms of loyalty and support for the state and irregular governance actors. In this paper, we question this "zero-sum" understanding of the relationship between the state and irregular governance actors, and investigate public opinion in inner-city neighborhoods empirically.

Conceptually, we distinguish two broad dimensions of governance. The material dimension captures the degree to which the state (or an alternative governance actor) plays a role in the daily life of citizens by providing crucial public goods. These range from core state functions, such as upholding the rule of law and protecting citizens from bodily harm, to the provision of services and infrastructure in health, education or environmental management. The subjective dimension of governance, by contrast, reflects the degree to which citizens are supportive of and hold positive attitudes towards the state or other governance actors.

Based on original survey data collected amongst residents of Jamaican inner-city neighborhoods, this article analyzes public opinion toward formal and irregular governance actors in neighborhoods where crucial public and private goods are provided by so-called dons, rather than the state. Principal factor analysis, a technique to identify clusters of variables and map the contours of public opinion, indicates that support for formal institutions is a separate dimension from support for the non-formal institutions headed by dons. Our results thus show that attitudes toward both types of governance actors are independent of each other. This indicates that formal and irregular governance actors do not necessarily compete for citizen loyalties, with positive attitudes toward one directly translating into negative attitudes towards the other. The complex relationship of competition and collusion between formal and irregular governance actors in the provision of services also seems to translate into more nuanced attitudes among citizens than the simple "zero-sum" logic would suggest. These findings have important implications for the way we view citizens in zones of weak material state governance. They challenge widespread notions of competition for "hearts and minds" between the state and irregular governance actors.

The first section of the article reviews the literature on the uneven territorial reach of the state and on irregular governance actors, critically assessing assumptions about the material and the subjective dimensions of governance. Next, we provide background to the case of the Jamaican dons and their role in inner-city governance. Following an explanation of the methodology used in researching this case, we present the results of our public opinion study on citizen attitudes, drawing on principal factor analysis to explore perceptions of formal and irregular governance actors.

Uneven state governance and irregular governance actors

In contexts across the world, but perhaps especially in the global South, neoliberal reforms have been associated with a profound and multifaceted crisis of the state. The "roll-back reforms" that accompanied structural adjustment programs affected not only the functioning of the state bureaucracy and its ability to provide public services. They also undermined public confidence in what ^{O'Donnell (1993, pp. 1357-1358)} refers to as the ideological dimension of the state, that is the belief that the state is "not just an arena for the pursuit of particularistic interests", but that its decisions are normally oriented "in terms of some conception of the public good". This rolling back of the state was not a spatially uniform process, however. Contemporary states in Latin America and the Caribbean are characterized by a substantial degree of socio-spatial differentiation in the way public institutions relate to and are experienced by citizens (e.g.^{Harbers, 2015}; ^{Pribble, 2015}). To capture this dynamic, O'Donnell proposes a conceptual map where "blue", "green" and "brown" zones, respectively, indicate declining performance of public institutions. Even though brown zones with limited state presence may not be entirely new, the radical implementation of neoliberal policies contributed to a rapid increase in "browning". Moreover, while traditionally ineffective state institutions were regarded primarily as a characteristic of remote and inaccessible hinterlands, in the aftermath of structural adjustment zones of obvious weak state governance emerged even in capital cities (O'Donnell, 1993). Low-income urban neighborhoods, which had sometimes benefited significantly from the largesse of the developmental state, were hit particularly hard by cutbacks in the public sector (^{Rodgers, 2006}; ^{Jaffe and Aguiar, 2012}).

The issue of unequal state presence plays a prominent role in two bodies of scholarship. First, research in anthropology on neoliberal democracy has analyzed the implications of the unequal access to state protection and social provisioning for citizens (^{Auyero, 2007}; ^{Auyero and Swistun, 2009}; ^{Goldstein, 2012}).¹ Much of this literature has focused on negative consequences for residents in zones characterized by weak material state governance. It has generated the important insight that alternative forms of political order and control may emerge in such areas. The alternative to effective state institutions and public service provision by the state is thus generally not anarchy, although it is often associated with diminished access to democratic rights. Various case studies show that irregular governance actors, such as local strongmen, paramilitaries or criminal gangs, often take over a range of state functions (^{Jackson, 2003}; ^{Arias, 2006}; ^{Davis, 2010}; ^{Arjona, 2014}; ^{Flanigan, 2014}; ^{Lilyblad, 2014}). Within bounded territories, these irregular governance actors maintain public order by establishing norms and punishing transgressors; they regulate access to public services such as utilities and employment; and they often produce a sense of political community.

A second body of research that has begun to focus on the political dynamics that emerge in these "brown zones" is based in security studies and international relations. Drawing on the experience from various regions in the global South, these scholars have taken up the insight that the retreat of the state may open up spaces for alternative forms of political order. However, their focus is not primarily on the consequences for citizens living in such areas. Rather, their concern is that dynamics developing there may ultimately pose a threat that extends far beyond local communities. The lack of effective state control is perceived as providing an opening for transnational organized crime and terrorist networks, so that such areas may become sanctuaries for outlaws and "breeding grounds for terrorism and criminal activities" (^{RAND, 2007: XV}).² The Counterinsurgency Field Manual, published by the U.S. Army and the Marine Corps, highlights that effective counterinsurgency campaigns must therefore "work to eliminate all sanctuaries" and to reestablish state control ^{(U.S. Army/Marine Corps, 2009, pp. 1-17)}.

This article contributes to the research agenda on the uneven state presence and the impact of irregular governance actors by questioning the relationship between public goods provision and public support. To do so, we distinguish two broad dimensions of governance. The material dimension of governance, as outlined in the introduction, captures the degree to which a governance actor plays a role in the daily life of citizens by providing crucial public goods. These range from core state functions, such as upholding law and order and protecting citizens from bodily harm, to the provision of public services and infrastructure. The subjective dimension of governance refers to the Weberian notion of authority, more specifically the likelihood that commands from a given source will be obeyed because the source is considered legitimate (Weber, 1947). States need loyalty and rely on at least tacit support from citizens for their role in society. To govern effectively, states need to be able to capture what is often referred to as "the hearts and minds" of a significant share of the population within their territory (^{Lukes, 2005}).

Scholarship in security studies and international relations has tended to assume that these two dimensions co-vary. Primacy has been attached to the material dimension, however. Even though political theorists from ^{Weber (1947)} to Gramsci (1971) to ^{Lukes (2005)} have argued that authority and legitimacy derive from much more than narrow material interests, the assumption in this literature has so far tended to be that the provision of goods and services plays a crucial role in generating citizen support and loyalty. Thus, absent or poorly functioning state institutions are assumed to undermine citizen allegiance to the state, which creates an opening for irregular governance actors to move in and capture citizen loyalties. This logic emerges particularly forcefully from the policy-oriented literature on security risks arising from "weak" and "failing" states. The RAND report (2007, pp. 7-8) on "ungoverned territories" emphasizes the material dimension when pointing out that "state health and welfare institutions may not reach into a substantial portion of the state's rural areas or inner cities." This material dimension is assumed to have a direct effect on the subjective dimension, because this "lack of presence allows other organizations to take precedence in determining the rules of everyday life. Thus, individuals may look to warlords, mullahs, or tribal leaders rather than state entities for judicial processes. Or insurgent groups may offer the only health care or other social services available to individuals residing in ungoverned territories". Ultimately, this literature views the existence of "competing local allegiances" (^{RAND, 2007, p. 8}) as having a corrosive effect on existing state institutions, thus creating a vicious circle.

Even a critical discussion of the concept of "ungoverned territories", edited by ^{Stanislawski (2008)} and published in the International Studies Review, shares many of these underlying assumptions. Building on the work of ^{Bunker and Sullivan (2003)}, Stanislawski conceptualizes so-called "black spots" as territories within formally sovereign states in which illicit actors or transnational criminal organizations have effectively taken over the role of the state. They are "forgotten islands of international disorder and most of their inhabitants usually prefer them to remain such" (Stanislawski, 2008, p. 366). One of the reasons why even civilians in these areas are disaffected from the state is that such areas "are characterized by ... mass public profiting from illicit activity" (Stanislawski, 2008, p. 367). Allegiance is thus assumed to be bought with goods and services. The understanding that citizen loyalties can be captured with goods and services also underpins the U.S. military strategy for countering security risks as articulated in the Counterinsurgency Field Manual, which highlights that effective counterinsurgency requires not just establishing control over territory, but that "the decisive battle" is the one "for the people's minds" ^{(U.S. Army/U.S. Marine Corps, 2009, pp. 1-29)}. In 2004, according to the manual, the "lack of adequate sewer, water, electricity, and trash services" in several Baghdad neighborhoods allowed hostile forces to recruit fighters and launch a violent campaign against U.S. forces (1-9).

While the bellicose language of counterinsurgency strategy, developed in relation to war-torn Iraq and Afghanistan, may appear inapplicable to more democratic contexts such as Jamaica, the notion that the material and the subjective dimension of governance are causally linked also underpins scholarship on Latin American and Caribbean democracies. The literature on structural adjustment frequently assumes that the material "rolling-back" of the state is associated with the decline of the state as a category of belonging and political identity. The associated emergence of other forms of political communities is often conceptualized as "parallel polities" (^{Leeds 1996)} or "fragmented sovereignty" (^{Davis, 2010}; ^{Richani, 2007}). Here, the state's competitors for citizen loyalty are generally drug cartels and criminal gangs, rather than terrorists and insurgents.

^{Davis (2010)}, for instance, highlights the ineffectiveness of state institutions in cities in the global South. She argues that "many urban residents become less connected to national states as a source of political allegiance" and consequently are "more prone to identify with alternative 'imagined communities' or networks of loyalties built on locally-based but spatially-circumscribed allegiances" (Davis, 2010, pp. 400-401). She expresses concern about a vicious cycle, where disappointment with the ability of the state to provide goods and services creates opportunities for alternative networks to become the framework for citizen loyalties. This, in turn, contributes to the further erosion of the state in both the material and subjective sense. Similarly, ^{Leeds (1996)} speaks of drug gangs in the favelas of Rio de Janeiro as "parallel polities" that use various strategies to legitimize their power and, in so doing, undermine local-level democratization in Brazil.

On the basis of the previous discussion, we can identify two assumptions that characterize contemporary theorizing about citizens' perceptions of formal and irregular governance actors. First, the majority of studies associate the emergence of irregular governance actors and their provision of alternative forms of political order with a loss of the "hearts and minds" for the state. Second, and more fundamentally, loyalty is conceived of as a zero-sum game. In this view, the state and irregular governance actors, such as criminal organizations, are seen as fundamentally antagonistic forces that compete for citizens' allegiance. According to this logic, citizens define their political identities either in terms of the state or a competing political order. This bipolar understanding involves the assumption that it is logically impossible to simultaneously support criminals and the police, to be on the side of justice and on the side of crime. Citizens have to choose, and a positive attitude towards one pole directly translates into a negative attitude towards the other pole.

The idea that support for the state and support for irregular governance actors are mutually exclusive is rarely questioned in either the literature on neoliberal democracy or on security. Yet, it is surprising in light of empirical scholarship on the collusive relationships that often exist between irregular governance and state actors (^{Auyero, 2007}). This either/or assumption not only deserves closer attention given its theoretical implications, it also has significant consequences for human rights. The question whether casualties of military action should be classified as insurgents or civilian bystanders is notoriously contentious. The idea that the civilian population either actively or passively supports insurgents contributes to blurring this line. In the extended conflict between the military, FARC (Fuerzas Armadas Revolucionarias de Colombia) and AUC (Autodefensas Unidas de Colombia) in Colombia, for instance, the "fact that the civilian population is believed to always line up with one or another of the armed groups in the conflict, and the belief that it is impossible to pry them away, is sometimes used to justify the violation of their human rights by the army" (^{Ramírez, 2010, p. 96}).

Yet, empirically, the link between the subjective and the material dimension of governance has remained underexplored. More specifically, we know relatively little about the way in which citizens in zones of weak material state governance view the state and irregular governance actors. Do they indeed support an antagonistic reading of the state and irregular governance actors? Do their perceptions line up with the unidimensional distinction drawn by policy-makers and social scientists? Our article seeks to contribute to a better understanding of this link through an analysis of the contours of public opinion in downtown Kingston, Jamaica.

Governing the inner-city in Jamaica

In media as well as academic accounts, dons such as Christopher "Dudus" Coke are often referred to as druglords or kingpins. However, in addition to their criminal activities, they are important governance actors in Jamaica's inner-city neighborhoods. This role developed historically through their centrality to a specific form of collective clientelism known as "garrison politics". The original source of the dons' power resided in their role as clientelist brokers between politicians and impoverished urban populations in the post-independence system of garrison politics. In the 1960s and 1970s, the two main political parties -the People's National Party (PNP) and the Jamaica Labour Party (JLP)- created party-loyal "garrisons" in Downtown Kingston. Politicians concentrated supporters in new housing developments and distributed money, jobs and weapons through the dons. During these decades, dons not only played a critical role in the violent maintenance of these political strongholds, they were also important actors in both facilitating and limiting the post-independence state's aspirations to roll out its presence in society (^{Gray, 2004}).

This garrison system began to change in the 1980s, as dons' role shifted from brokers to co-rulers. Economic recession, foreign debt and structural adjustment reduced politicians' capacity to use public funds as political favors. More broadly, structural adjustment was accompanied by a decrease in effective service provision, i.e. a weakening of the material dimension of governance outlined above. The contraction of the state's material presence or efficacy was especially evident in inner-city neighborhoods in the urban areas of Kingston, Spanish Town and Montego Bay. While the commitment of the state to inner-city neighborhoods in the pre-structural adjustment period had often been limited and uneven, strong cuts in public spending resulted in diminished material state efficacy in providing urban infrastructure, social services and social facilities. In addition, the decline in the state's material dimension in these areas was especially evident in terms of residents' restricted access to public security and rule-of-law dispute resolution (^{Gopaul, 1996}; ^{Clarke and Howard, 2006}).

This decrease in the material dimension coincided with dons' turn to alternative sources of income, in drug trafficking and extortion as well as in various legal enterprises. The advent of structural adjustment and their increased financial independence from politicians enabled the more powerful dons to replace members of parliament (MPS) as community patrons (^{Sives, 2002}, ²⁰¹⁰). While dons remain entangled with state actors, the more successful ones have gone beyond being brokers and local patrons to being partners-in-governance. Dons' role and impact do vary significantly across inner-city communities, depending on, among other factors, their economic base, the nature of their organizations, their attitude toward politics, and their political, social and business connections (^{Figueroa et al., 2008}). Drawing on their own funds and their access to the means of violence, the most powerful dons offer a broad range of "public" services within the socio-political space of inner-city neighborhoods. Residents rely on them for services ranging from welfare provision and employment facilitation to the maintenance of green spaces (^{Jaffe, 2013}). In addition, dons provide security and conflict resolution, known locally as "community justice" (^{Duncan-Waite and Woolcock, 2008}). The dons' role in providing services that would otherwise be unavailable to inner-city residents is frequently seen as legitimizing their position in the community.

This subjective dimension of support is often understood as incompatible or competitive with loyalty to the Jamaican state (e.g.^{Charles, 2002}; ^{Price, 2004}). A number of studies on Jamaica's dons, and certainly many reports in popular media, imply that the inner-city neighborhoods suffer from a form of political pathology, in which entire populations and territories are isolated from "normal" political institutions and values. This resonates with a view of swathes of the urban landscape as standing in opposition to mainstream Jamaican society, as is evident in the characterization of inner-city neighborhoods as "counter societies" (Charles, 2002). These neighborhoods are often depicted as "the most likely sources of crime and violence, generalized lawlessness, indiscipline and urban revolt" (^{Johnson, 2005, p. 589}), as criminogenic spaces that harbor residents whose morals are deviant and corrupt, and who have turned their backs on the formal political system.

Methods

The study on which this article relies took the form of public opinion research amongst inner-city residents, seeking to identify attitudes visávis both dons and the state. Our analysis draws on an original survey of 400 residents in eight inner-city communities in Kingston. This survey was developed as a quantitative component within a larger research project on dons, which also involved extensive ethnographic research in Kingston's inner-city neighborhoods. To the best of our knowledge, it is the first survey on formal and irregular governance specifically targeting the urban poor in Jamaica. The lack of public opinion data about this group is in part due to the methodological challenges inherent in conducting surveys in such settings (^{Bulmer and Warwick, 1983}). The majority of existing studies on comparable irregular governance actors tend be ethnographic and based on long-term fieldwork in one area, for instance focusing on one community controlled by a gang or a broker (^{Auyero, 2000}; ^{Rodgers, 2006}; ^{Arias, 2006}). In Jamaica, a number of ethnographic studies based in particular Jamaican neighborhoods have engaged with the topic (^{Harrison, 1988}; ^{Jaffe, 2013}). The quantitative survey, in contrast, sought to shed light on the political attitudes and perceptions of inner-city residents drawing on a larger geographic sample.

Quantitative data that enable the study of public opinion in inner-city areas are scarce. Even though public opinion data are now available for many countries in the global South, national probability samples (such as data gathered by the Latin American Public Opinion Project, LAPOP) tend not to include a sufficiently high number of low-income urban residents.³ Survey-based research amongst inner-city populations and on irregular governance actors has been limited for a number of different reasons. First, inner-city residents are difficult to target with conventional sampling methods, since these methods require extensive information about the population to compile a sampling frame (e.g. telephone directories, electoral registers, census tracts with formal streets and official home addresses). Knowledge about marginalized and vulnerable populations is often limited, which is why the ability to collect accurate information about citizens by itself is considered an indicator of state capacity (^{Soifer, 2013}). An inherent characteristic of areas where the material presence of the state is weak is therefore that inhabitants often remain "unseen" (^{Scott, 1998}; ^{Lee and Zhang, 2013}). There are no "comprehensive directories of who's where" (^{Bulmer, 1983, p. 91}) because residents in these neighborhoods are "hard-to-count" (^{Ericksen and Kadane, 1985}). Second, inner-city residents are particularly reluctant to answer the questions of outside interviewers. Thus, even when they are included in the sample, they are less likely to actually participate (^{Ward, 1983}). As ^{Goldstein (2004, pp. 41-45)} has argued, residents of low-income neighborhoods may be especially predisposed to distort or withhold information in quantitative survey research. This refusal to cooperate is connected to a distrust of (government) researchers' intentions and to hopes that specific representations of neighborhood life may elicit a positive government or NGO response. In addition, while national surveys may attempt to include inner-city neighborhoods, issues of insecurity in violence-prone neighborhoods can inhibit public opinion research and census-taking efforts more generally. Survey interviewers risk facing overt hostility, all the more so if they broach sensitive topics related to illegal activities. In 2011, census takers collecting population data in Jamaica had dogs set on them, were threatened with rape, and had their lives threatened (^{Virtue, 2011}).

Our survey and questionnaire were developed in collaboration with Jamaica's University of Technology. Neighborhoods were chosen on the basis of two criteria. First, we selected neighborhoods in Kingston that were located geographically in the low-income area of the city known as Downtown and in which the presence of dons and their position as irregular governance actors had been verified through the ethnographic component of the research project, through media reports, and in discussions with research assistants. Thus, we deliberately targeted "hard-to-count" areas that often fall outside the purview of quantitative studies. The second criterion for neighborhood selection was access and safety. To overcome some of the challenges of doing public opinion research in inner-city neighborhoods, our survey was carried out by research assistants who resided in and/or were well acquainted with the neighborhoods included in the study, and who had been trained as interviewers. Surveyors were recruited via an NGO⁴ and included students at Jamaica's University of Technology. Only neighborhoods in which surveyors were willing to carry out the survey and where they felt safe enough to do so could be included. On the basis of these criteria, the following eight neighborhoods, distributed across Downtown Kingston and different electoral constituencies, were selected: Fletcher's Land, Hannah Town, Maxfield Ave, Duhaney Park, Allman Town, Mountain View, Swallowfield, and Jones Town.⁵

The unit of observation for the study is the individual, as questions relate to respondents' attitudes. Within the study neighborhoods, conditions on the ground did not allow for probability sampling. Purposive sampling was therefore used to ensure that the target population, i.e. residents of inner-city neighborhoods with don presence, was reached effectively. Surveyors approached residents in public places within the neighborhood and asked for their cooperation. As the experience of official census takers in Jamaica indicates (^{Virtue, 2011}), knocking on random doors is generally not welcomed or considered in these contexts. While surveyors verified that respondents lived in the respective neighborhood, a common vulnerability with purposive sampling is that -within the target population- a large share of respondents come from more accessible subgroups. For instance, surveys collected during the day may include a large number of unemployed respondents. To minimize the over-inclusion of accessible populations, two steps were taken. First, the survey was conducted at different times of the day and the week to ensure access to both employed and unemployed residents.⁶ Second, quota sampling was used to create a gender and age balance in each neighborhood. Of the 400 survey respondents, 51 per cent were male and 49 per cent, female. The age distribution was as follows: 27 per cent were aged 15-25 years; 25 per cent were 25-35 years; 21 per cent were 35-45 years; 16 per cent were 45-55 years, and 11 per cent were 55 and older. Fieldwork was carried out between April and July 2010 on 26 different days.

Questionnaires consisted of 27 statements on multiple topics including security, politics and justice, to which responses were sought on a Likert scale ranging from "strongly agree" (= 1) to "strongly disagree" (= 5). While the survey had a broader scope than measuring the subjective dimension of governance, it provides a unique opportunity to explore and compare citizen attitudes towards the state and irregular governance actors. Table 1 reports the survey items capturing attitudes towards governance actors and their descriptive statistics.

Table 1 Descriptive statistics for variables included in the analysis 

Source: Authors' Survey. Note: Responses ranged from 1= strongly agree to 5 = strongly disagree.

Understanding the subjective dimension of governance: public opinion in inner-city Kingston

The results of our survey of public opinion on formal and irregular governance actors yield three important findings. First, comparing our survey respondents to the voting age population, as captured by a survey of the Latin American Public Opinion Project, suggests that inner-city residents tend to hold more negative views of the state, though the difference is not as stark as we might expect. Second, we find that attitudes towards dons, on the one hand, and attitudes towards the state, on the other, are clustered. Attitudes towards some aspects of the formal system, such as the prime minister, are thus not independent of attitudes toward other aspects of the formal system, such as the government and law enforcement. This suggests that the analytical distinction between different types of governance actors made by policy-makers and social scientists reflects public perceptions. Third, while citizens do distinguish between these two broad types of governance actors, the existence of two dimensions shows that citizens do not necessarily see these two systems as competing. We now discuss each of these results in more detail.

As outlined above, the inner-city neighborhoods in Kingston where the survey was conducted are characterized by weak material state governance and the literature suggests that citizens in the inner city are particularly disaffected with the state. We might therefore expect them to express more negative attitudes about formal political institutions, such as the police or the head of government, than the general public. To explore how inner-city residents compare to the overall voting age population, we relate the results of our survey to Jamaican data from the 2010 wave of the Latin American Public Opinion Project (LAPOP).⁷ Because the wording of questions and scales of responses differ between the two datasets, results of this comparison should be viewed as preliminary and suggestive, rather than definitive. Nevertheless, a clear trend emerges: inner-city residents tend to be more negative about state institutions than the general population, though the difference is not as large as we might expect.

Our survey asks respondents to what extent they agree with the statement "I trust the Prime Minister". LAPOP includes a comparable question: "To what extent do you trust the prime minister?"⁸ In our survey, 62 per cent of respondents disagree with the statement, while 54 per cent of respondents in the LAPOP survey fall in the low trust category. With regard to trust in the police, there is almost no noticeable difference: 65 per cent of our respondents express negative attitudes, compared to 63 per cent of LAPOP's respondents.

In addition to attitudes about specific institutions, our survey also includes the more general statement: "The current system of government works just fine". In response to this statement, 22 per cent of respondents express positive attitudes about the system of government. While no item in the LAPOP survey is entirely comparable, the question: "In general, would you say that you are very satisfied, satisfied, dissatisfied or very dissatisfied with the way democracy works in Jamaica?" also captures broad system support; 44 per cent of respondents are satisfied or very satisfied with the way democracy works in Jamaica. Overall, then, we observe somewhat more negative attitudes about formal state institutions among inner-city residents than among the general voting age population.

To understand the subjective dimension of governance, we explore the contours of public opinion towards different types of governance actors with principal factor analysis, a technique to identify clusters of variables. The aim of factor analysis is to reveal underlying or "latent variables". Latent variables reflect concepts that are so complex or abstract that they are difficult to measure or observe, because no single indicator can capture them adequately. Factor analysis is particularly useful to study "etic" analytical concepts, such as "irregular governance actors", which may not resonate with respondents. Even though these concepts are part of the scholarly vocabulary, respondents may not be familiar with them. Including such concepts as items in surveys is therefore ineffective and runs the risk of provoking invalid responses. To capture them, public opinion scholars commonly employ a series of variables that tap into different aspects of the phenomenon of interest (e.g.^{Wlezien, 2004}; van der Brug and van Spanje, 2009). Studies of racial resentment, for instance, generally rely on factor analysis to mitigate social desirability bias (e.g.^{Wilson and Davis, 2011}). Rather than asking respondents to place themselves on a "racial resentment scale", surveyors ask if they agree with statements such as "For African Americans to succeed they need to stop using racism as an excuse" and "African Americans should not need any special privileges when slavery and racism are things of the past". In brief, the idea of factor analysis is that -while the latent variable cannot be observed directly- it can be uncovered because it correlates with other variables that better lend themselves to measurement.

In this case, analyzing public opinion with principal factor analysis allows us to study attitudes without making assumptions about dimensionality. In other words, the method does not assume that positive attitudes towards the dons are directly related to negative attitudes about formal governance actors. The method is inductive, in the sense that the nature of this relationship is treated as an open question, because co-variance of the observed variables is used to uncover information about the existence and dimensionality of latent variables (^{Blalock, 1982}).

Multiple items included in our survey tap into attitudes towards dons and state institutions. Table 2a reports the results of principal factor analysis after rotation. Even though 13 variables were included in the analysis, not more than six factors are possible. Of these, only two are significant following ^{Kaiser's (1960)} criterion (eigenvalues >1).⁹ Table 2b reports factor loadings for statistically significant factors. The analysis suggests two main findings. First, the dimensionality of attitudes towards governance structures is low. Even though 13 variables were included in the analysis, only two factors are statistically significant. These factors respectively account for 50 and 41 per cent of the variance in the matrix. This suggests that there is considerable co-variance among the items, which implies the existence of latent variables. Second, the factors capture attitudes towards two types of governance actors. Table 2b shows factor loadings, which reflect the correlation between the variable and the factor. Higher correlations indicate more commonality. Following ^{Stevens (1992)}, factor loadings larger than 0.3 may be considered statistically significant for our sample. They are reported in bold.¹⁰

Table 2 Factor analysis of legitimacy variables 

Source: Authors' Survey.

The first factor represents attitudes towards formal governance actors and the state. It strongly correlates with trust in the prime minister and satisfaction with the current system of government. It also mildly correlates with support for the statements: "The MP cares about what happens to us", "Most police can be trusted" and "In Jamaica, the law treats everyone the same". The second factor reflects attitudes towards the dons. This factor correlates with the statement: "If this community had a strong don, crime would be less" and "The media should be more positive about dons". It also correlates highly with trust in dons and mildly with support for the statement: "Community justice is more effective than the formal system".

What are the substantive implications of these results? And what do they tell us about the way citizens conceive of the relationship between different types of governance? First, the factor analysis shows that attitudes towards dons, on the one hand, and attitudes towards formal governance, on the other, are clustered. Attitudes towards some aspects of the formal system of governance, such as the prime minister, are thus not independent of attitudes toward other aspects of the formal system, such as the government and law enforcement. This suggests that the analytical distinction between different types of governance actors made by policy-makers and social scientists to some extent reflects public opinion and citizen perceptions. Contrary to the commonly held assumption, however, legitimacy is not a zero-sum game. If this were the case, we would have found only one dimension that correlated positively with variables reflecting attitudes about formal governance and negatively with variables reflecting attitudes about the dons (or vice versa). Instead, our results reveal a more complex picture, where attitudes about formal and irregular governance actors are associated with distinct dimensions.

The question then remains to what extent inner-city residents express positive attitudes towards formal governance institutions and the dons. Table 3 classifies respondents into a 2 x 2 table. For each dimension, respondents with factor scores above the mean are classified as high and those with factor scores below the mean as low. If support for formal and irregular governance actors were a zero-sum proposition, we would observe a clustering in the top-right and bottom-left quadrant, with the top-right quadrant representing those inner-city residents whose loyalties have been "captured" by dons and the bottom-left quadrant representing those emotionally attached to the state.

Table 3 Classification of respondents 

Source: Authors' Survey.

Instead, the table shows that respondents are distributed across all four cells fairly evenly. How can we understand these results? Half of respondents express positive attitudes towards irregular governance run by dons (49.8%). However, a majority of respondents (59.3%) expresses positive attitudes about formal governance. This is an important result, as it contradicts assumptions that inner-city neighborhoods are zones where the majority of citizens are disaffected and "counter societies" have emerged (^{Charles, 2002}). The results refute the idea that residents have turned their back on the state, as only a minority of respondents (19.5%, top-right quadrant) display high trust in don-related institutions and low trust in formal governance. Another minority (21.2%, top-left quadrant) are citizens that hold negative views about both formal and irregular governance actors. A sizable group, almost a third of respondents (30.3%), express positive attitudes about both formal governance and the dons. These respondents place trust in the police as well as in community justice, in the prime minister as well as the don. The contours of citizen attitudes towards dons and formal governance actors are therefore more complex than scholarship and policy-makers have assumed thus far.

Conclusion

Jamaica, like many other countries in the global South, has been subject to decades of neoliberal structural adjustment policies. "Roll back" reforms in particular have been associated with declining public services and, ultimately, a weakening material presence of the state in marginalized parts of the country. Based on original survey data collected amongst residents of inner-city neighborhoods in Jamaica, we have explored the subjective dimensions of governance, which reflects the extent to which the state succeeds in winning hearts and minds, in neighborhoods where its material presence is weak. The underperformance of the state in these areas is partially compensated by irregular governance actors, known locally as dons. They offer a range of "public" services within their territories, ranging from welfare provision and employment facilitation to the maintenance of green spaces and an alternative system of conflict resolution, known locally as "community justice". So far, the bulk of the literature on governance has assumed that the material and the subjective dimension of state-presence co-vary and that formal and irregular governance actors compete for citizen loyalties, with positive attitudes toward one directly linked to negative attitudes towards the competitor.

Contrary to the commonly held assumption, we found that citizen attitudes do not follow a zero-sum logic. Our results suggest that understanding the subjective dimension of governance as an either/or proposition does not adequately reflect the situation on the ground. From the perspective of inner-city residents, positive attitudes towards dons, and the services they provide, are not experienced as necessarily competitive with positive attitudes towards formal governance actors. The fact that many citizens do not apprehend dons and formal governance as competing systems suggests that conceptualizing the subjective dimension of governance in either/or terms might not necessarily be the most fruitful approach. While our findings run counter to the assumptions that have so far tended to underpin theorizing on the subjective dimension of governance, the finding itself should not necessarily be surprising. The "either/or" approach to the subjective dimension of governance appears to be rooted, at least partly, in the European experience of centralized state organization, which only very incompletely captures the reality of postcolonial states. In contrast to the historically unique situation in Europe, the relationship between citizens and postcolonial states tended to be mediated to a much greater degree by local brokers and strongmen (^{Migdal, 1994}). In Jamaica, dons assumed this role of brokers. In light of this history of mediated relationships and the entanglement between formal and irregular governance actors (^{Jaffe, 2013}), it is perhaps not so surprising that citizens do not view different types of governance actors as competing.

This finding has profound policy implications. A positive reading of the findings recognizes that the situation in inner-city neighborhoods is not as bleak as has been suggested. The notion of inner-city neighborhoods as deviant spaces disconnected from the modern nation is not confirmed by our results. A more pessimistic interpretation of the results, by contrast, might focus on the top left quadrant of Table 3 and highlight that one fifth of survey respondents hold negative attitudes towards both formal and irregular governance actors. Loyalty towards governance actors -even if they provide services- can therefore not be taken for granted. The Counterinsurgency Field Manual recommends winning the hearts and minds by, among other things, discrediting competitors as criminals ^{(U.S. Army/Marine Corps 2009)}. Yet, citizens may become disaffected with irregular governance actors without simultaneously developing more positive views towards formal governance actors. Overall, our results call for a much more nuanced view of citizen attitudes towards formal and irregular governance actors, as these are more complicated than previous theorizing has acknowledged.