A man ran after [a] young man [carrying a purse] and grabbed him. Soon young men emerged from houses, business and connecting streets, many running with an enthusiasm usually reserved for the football pitch. Some carried large sticks. Suddenly, the young man with the purse was being slapped and punched […]. As he walked punches came from different angles, landing on his face, back, legs and arms. […] A boy, not older than 10, followed the procession. “He won't steal again,” he said smiling. A middle aged man laughed: “He deserves it.”
(J. Moresco, “Mob ‘Justice’ on the Rise”)This description, capturing the reaction of people in a suburb of Accra, Ghana, when they heard cries to catch a young man suspected of broad-daylight robbery, typifies violent self-help in urban Ghana. The phenomenon of violent self-help is, however, by no means unique to Ghana; it is part of a global pattern, from Lagos to New York, Cape Town to Sao Paulo, and La Paz to Lahore (Reference Huggins and HugginsHuggins 1991; Reference BakerBaker 2002; Reference BurrBurr 2006; Reference MeagherMeagher 2007; Reference SmithSmith 2004; Reference GoldsteinGoldstein 2003; Reference HandyHandy 2004). As a mechanism of social control (Reference BlackBlack 1983), significant public recourse to violent self-help speaks to broader issues of the rule of law, and suggests that the performance of the state and its legal institutions is being questioned. In most cases of vigilantism, the aim is to “even the score” against a proven or suspected wrongdoer (Reference BlackBlack 1989).Footnote 1 Thus violent self-help hinges directly on the transhistorical problem of order in all human societies (Reference WrongWrong 1994), and there is significant variability in how it is suppressed.
There is a strong scholarly consensus about violent self-help, which suggests that its incidence is inversely correlated with effective intervention by law (E. Reference AndersonAnderson 1999; Reference BlackBlack 1983; Reference Black and BaumgartnerBlack & Baumgartner 1987; Reference GoldsteinGoldstein 2003; Reference BrownBrown 1975). Thus E. Anderson has located the antecedents of violent self-help in a sociopolitical context in which effective intervention by the state has atrophied and personal responsibility for one's safety becomes paramount, “resulting in a kind of ‘people's law,’ based on ‘street justice’” (E. Reference AndersonAnderson 1999:10). In such social circumstances, sometimes described as frontier communities, the ability to “take care of oneself” is met with social approbation, which translates into a sense of physical and social security. The attribution of the genesis of vigilante self-help to a lack of effective state intervention implies, of course, a deterrent-based foundation for the problem of order. As some researchers have pointed out, however, the presence or absence of state intervention may not be the central consideration in diminishing recourse to violent self-help (Reference Black and BaumgartnerBlack & Baumgartner 1987:34; Reference WacquantWacquant 1993; Reference WeisburdWeisburd 1988). The prevalence and persistence of violent self-help speaks, unquestionably, to the primary necessity of state substantive performance, measured by the provision of security against victimization. But an important and often neglected issue may be the potential role of normative considerations about state recognition of people's rights to fair and dignified treatment. In other words, both the fact and the character of state intervention may be crucial to an understanding of the persistence of violent self-help. This article is concerned with discovering the conditions that may explain public support for vigilante violence.
One important approach to the problem of order is the normative-based notion of procedural fairness, as developed by Tyler and his colleagues (see, e.g., Reference TylerTyler 1990, Reference Tyler and Tonry2003; Reference Tyler and HuoTyler & Huo 2002; Reference Sunshine and TylerSunshine & Tyler 2003; Reference Tyler and WakslakTyler & Wakslak 2004). Unlike deterrence-based regulatory strategies, which view people simply as instrumentally rational beings, procedural fairness research argues that people are moral beings who are concerned with the procedural propriety and the fairness and trustworthiness of legal authorities (see also Reference MacCormickMacCormick 2007; Reference BeethamBeetham 1991; Reference Bottoms, Bottoms and TonryBottoms 2002). There is now abundant evidence that when people believe that procedures are fairly administered, and that legal authorities are trustworthy, they are more likely to comply with the law and cooperate with legal authorities. Conversely, perceptions of unjust procedures and untrustworthiness can lead to public unwillingness to cooperate with the police, to resistance to vesting the police with additional powers, and to general noncompliance with legal decisions. What remains unexplored is whether and how perceptions of police violations of their normative obligations to treat people with procedural fairness might affect people's expressions of support for violent self-help mechanisms such as vigilantism, especially within a context of apparent police ineffectiveness in the provision of security.
In this current research, I explore the issue of vigilante self-help within the context of the Tylerian model. Of particular interest is whether a relationship exists between people's assessments of police procedural fairness and their expressions of support for vigilantism. I begin this article by presenting a discussion of the theoretical model that seeks to link procedural injustice to vigilante self-help. Next I offer a very brief context-setting schematic discussion of policing and crime in Ghana—where the study was conducted—and explain the data collection methods for the present study. Finally, I present the survey data that test the specified model, and I conclude with a discussion of the results. Although expressions of support for vigilantism do not, ipso facto, set people on a course of action, there is evidence that such expressions may often lead to active involvement in vigilante violence (see, e.g., Reference WeisburdWeisburd 1988). In any case, perhaps the greatest significance of support for vigilantism is the ability to create an environment that condones and justifies the practice. It is also important to recognize from the outset that, in a broader context, a fuller understanding of vigilante justice is connected to Hobbesian questions of the sovereignty of law (see, e.g., Reference LenzLenz 1988; Reference AbrahamsAbrahams 1998; Reference Rosenbaum and SederbergRosenbaum & Sederberg 1974). However, the present data, and indeed the Tylerian theory of procedural justice itself, do not allow one to venture into that territory.
Procedural Injustice and Violent Self-Help
A number of criminological researchers have recently highlighted the need to focus on normative considerations that shape people's orientation to law and legal authorities such as the police (see, e.g., Reference Sunshine and TylerSunshine & Tyler 2003; Reference Bottoms, Bottoms and TonryBottoms 2002; Reference TylerTyler 1990). This interest is based on the recognition that “human beings are norm-users, whose interactions with each other depend on mutually recognizable patterns that can be articulated in terms of right versus wrong conduct, or of what one ought to do in a certain setting” (Reference MacCormickMacCormick 2007:20). Thus Reference TylerTyler (1990) has proposed a “normative” perspective of justice, in which citizens' conceptions of justice and their compliance with the law are based on perceptions of fairness and equity. This normative perspective suggests that citizens are concerned with justice, which encompasses both distributive and procedural dimensions; but the empirical evidence has focused especially on procedural fairness, which in some contexts at least is more important than distributive fairness. Legal authorities are seen as procedurally fair if citizens rate the qualities of decisionmaking and treatment processes positively. Paternoster and his colleagues have identified the key elements here to be consistency, neutrality, objectivity, lack of capriciousness, impartiality—all aspects of “fairness of decision-making”—and treating people as human beings—“fairness of treatment”—(Reference PaternosterPaternoster et al. 1997; see also Reference TylerTyler 1990).
This framework has been used to study different aspects of people's encounters with agents of the criminal justice system (Reference Tyler and HuoTyler & Huo 2002; Reference Sunshine and TylerSunshine & Tyler 2003; Reference OhbuchiOhbuchi et al. 2005; Reference PaternosterPaternoster et al. 1997; Reference SparksSparks et al. 1996). An unstudied aspect of perceived police procedural injustice is, however, the extent to which it might contribute to public support for violent self-help. Unlike the criminological literature, there is a corpus of evidence from researchers in work organizations that has linked judgments about perceived injustice to behaviors such as worker protest (Reference VermuntVermunt et al. 1996), sabotage (Reference Giacalone, Giacalone and GreenbergGiacalone et al. 1997), theft (Reference GreenbergGreenberg 1993), and violence (Reference Folger and SkarlickiFolger & Skarlicki 1998). At the heart of judgments about procedural justice is the conception of people as rights-bearing individuals who demand to be treated fairly and with dignity and respect (Reference HerbertHerbert 2006; Reference FukuyamaFukuyama 1992). The possession of rights is thus the realization of socially recognized claims made on others and accepted by them (Reference BarbaletBarbalet 2001). The main argument of this article is that public recourse to, or expressions of support for, vigilante violence reflects, at in least in part, public resentment against the perceived procedural injustices of the nation-state and its legal apparatus.
Barbalet's discussion of the twin notions of resentment and vengefulness are relevant here. According to Barbalet, resentment is an emotional apprehension directed against the advantage gained by others at the expense, and in contravention, of established societal norms regarding “acceptable, desirable, proper, and rightful outcomes and procedures” (Reference BarbaletBarbalet 2001:138); it is therefore, at root, a normative emotion. Vengefulness, by contrast, is “an emotion of power relations” in the sense that it is the consequence of “an exercise of power in which the subject may lose what they regard, in the context of social interactions, as a self-defining attribute”; it is therefore a power-based retaliation against an abrogation of rights and a reassertion of human dignity or worth (Reference BarbaletBarbalet 2001:136). Barbalet further argues that both of these emotions have regular points of convergence, with resentment often reinforcing vengefulness. Both emotions are clearly relevant to the issue of violent self-help.
Ordinarily, police professionalism translates into public expectations that police officers will perform their duties within a set of fair and accountable guidelines (Reference SeronSeron et al. 2004). In too many states, however, dehumanizing treatment, injustice, and abuse have characterized the experiences of some sections of society with legal authorities. Such experiences can be felt as socially injurious, which in turn can engender moral anger as people seek to reassert themselves and lay claim to the social recognition denied them. The police treat people procedurally unfairly when they deny them the opportunity to state their point of view, or when the police display bias and capriciousness. At worst, the police can sometimes show contempt for people as citizens when, for example, they go to a police station to report a crime. Indeed, such concerns may sometimes aggravate the significance of the police's lack of effectiveness because while the public may, in some situations, be sympathetic to genuine reasons for police ineffectiveness, they may be unsympathetic if and when police disregard what they consider to be their basic right to be treated as citizens with genuine issues to raise. McDougall has argued that “the act that, more certainly than other, provokes vengeful emotion is the public insult, which […] lowers one in the eyes of one's fellows” (cited in Reference BarbaletBarbalet 2001:135). Of course insults and undignified treatments need not be public since, as Scott argues, “one can experience indignity at the hands of another despite the fact that no one else sees or hears about it” (1990:113). The public nature of an insult, however, exacerbates the injury it inflicts: “an insult, a look of contempt, a physical humiliation, an insult on one's character and standing, a rudeness is nearly always far more injurious when it is inflicted before an audience” (Reference ScottScott 1990:113).
It is important to note that feelings of resentment against socially injurious treatment, which often result in cries of injustice, “[are] never fully laid to rest except by a public reply” (Reference ScottScott 1990:115). The reply expressed in the form of vigilante violence often contains two messages. First, as an expression of self-help—violent or benign—vigilantism is a form of crime control (Reference BlackBlack 1983, Reference Black1989, Reference Black1998; Reference Black and BaumgartnerBlack & Baumgartner 1987; Reference BakerBaker 2002). As Reference Rosenbaum and SederbergRosenbaum and Sederberg (1974) have argued, vigilante actions are an attempt to defend a given social order by resorting to means that contravene normative rules meant to safeguard that order in the first instance. In this sense, vigilantism is commonly seen as the direct outcome of chronic police failure in the provision of physical security. A vivid illustration is furnished in Reference HarnischfegerHarnischfeger's (2003) account of the antecedents of the Bakassi Boys' vigilante group in southeastern Nigeria. Amidst an unprecedented upsurge in violent crime, intimidation from gangs, and apparent police lack of effectiveness, local traders mobilized the youth in the neighborhood to procure security to salvage the local economy, which had been shattered by widespread violence. Popular support for the Bakassi Boys was almost universal, and within a short period, residents reported that the Boys had transformed “Onitsha […] into the safest place in Nigeria, a place where one could walk through the town centre at midnight with a case full of money” (Reference HarnischfegerHarnischfeger 2003:25). This is a clear demonstration of the primacy of the effective provision of security to researchers' attempts to understand vigilante violence, and this is the dimension of the explanatory mechanism of vigilantism that has been emphasized strongly in the self-help literature (see, e.g., Reference AdinkrahAdinkrah 2005; Reference BakerBaker 2002; Reference GoldsteinGoldstein 2003; Reference MeagherMeagher 2007; Reference SmithSmith 2004).
Accounts of African vigilante self-help also illustrate the importance of police corruption as an explanatory factor. In most cases, corruption means police neglect of their duties, with, sometimes, law enforcement agencies colluding with criminal elements for personal gain (see, e.g., Reference AdinkrahAdinkrah 2005; Reference SmithSmith 2004). Again, an extract from Harnischfeger's accounts proves instructive. The police had to tolerate the existence of a vigilante group—a powerful challenge to their claims to legitimacy—because the group had provided “a list of policemen who collaborated directly with armed robbers,” which the police leadership did not want to be made public (Reference HarnischfegerHarnischfeger 2003:24).
The second message often contained in vigilante violence, which is frequently neglected in the literature on self-help, is an analogous symbolic injury or punishment targeted at law enforcement agencies (see Reference BarbaletBarbalet 2001). Vigilante violence in this sense becomes a form of “symbolic defiance” and protestation against “daily indignities” (Reference ScottScott 1990:113). This is often the case when police failure to be consistent with the normative expectations of fair and equitable treatment culminates in the advantage of one actor over another (see Reference BarbaletBarbalet 2001). Police violation of socially established norms of proper, acceptable, and rightful procedures can take many different forms. For instance, vigilante violence may result from attitudes and behavioral patterns demonstrative of Reference BerkiBerki's (1986) notion of “malevolent indifference.” Here, the police claim, often militantly, the exclusive right to deal with the problem of order while at the same time failing or refusing to respond to the security needs of some sections of society (see E. Reference AndersonAnderson 1999). In such circumstances, violent self-help becomes an attempt to compel the police to bridge the gap between what people might consider to be their socially established entitlement to procedurally fair treatment, and the abusive and/or neglectful treatment that characterizes their encounters with the police. In this symbolic cast, vigilante violence can be an expression of contempt for police claims, in peacetime, to be the sole legitimate legal dispensers of violence against wrongdoers.Footnote 2 Thus, Reference WeberWeber has argued that, in the particular context of privilege, when the legitimation claims of the powerful unravel, and this becomes “unambiguously and openly visible to everyone,” they become one of the most passionately hated objects of attack (1978:953–4). It is not fanciful to transfer such comments to the context of policing. Such attacks can often be physical, such as assaults on police officers, or symbolic and hence directed at the authority of the police.
The preceding arguments suggest a model in which procedural justice, corruption, and effectiveness may be conceptualized as direct and independent explanatory factors for vigilante support. Figure 1 provides a simple heuristic diagram that summarizes the direction of the impacts of these variables.
Figure 1. Support for Vigilante Self-Help
But while the model specified in Figure 1 is important and useful, it may not capture what can often be a complex relationship between police corruption, effectiveness and police procedural justice, and support for vigilantism. It would certainly be a leap of logic to assume that the mere perception of procedural injustice or poor police performance automatically translates into the development of attitudes supportive of vigilantism. It is therefore important to consider an alternative and potentially more useful model. Most often, the salient role of these concerns is rather to undermine public trust in legal institutions. Reference Tyler and HuoTyler and Huo (2002), for instance, have established a strong relationship between procedural justice and the trustworthiness of law enforcement agencies (see also Reference TankebeTankebe 2008a). Other researchers have highlighted the various ways through which corruption and poor performance may weaken public trust in legal authorities (see, e.g., Reference GoldsmithGoldsmith 2005; Reference RothsteinRothstein 2000; Reference Rose-AckermanRose-Ackerman 1999; J. C. Reference Anderson and TverdovaAnderson & Tverdova 2003). A long tradition of debate among political theorists also makes clear that poor regime performance—defined in terms of both tyrannical and abusive exercise of power, and failure to meet the physical or material needs of citizens—can often erode feelings of obligation to obey legal directives, leading to rebellion and violence (see, e.g., Reference GurrGurr 1970; Reference MooreMoore 1978).
The important point here, for the purpose of the present study, is that once police corruption, ineffectiveness, or procedural injustice have fatally undermined the perceived trustworthiness of the police (and indeed the criminal justice system), very little can be done to sustain or justify police legitimation claims, thereby freeing some sections of society to challenge those claims in the form of vigilante violence. When the public do not trust the police, they will be unwilling to defer to the police and to accept police claims, in peacetime, to exercise exclusive authority in the provision of security and in dealing with lawbreakers. Zimring has put it aptly: “[t]he vigilante is by definition suspicious of his government, and that is one reason why vigilantes are willing to arrogate the power to punish crime to community groups” (cited in Reference MessnerMessner et al. 2006:562).
Thus a more useful approach to understanding support for vigilantism requires a model that takes into account this potential mediating role of police trustworthiness. Figure 2 provides such a heuristic model. In this model, the influences of procedural justice, police effectiveness, and corruption are seen as contingent on the extent to which they shape public perceptions of police trustworthiness. Neither of these hypothetical models—that is, the direct and the “mediated through trust” models—has so far been examined empirically. Therefore in the present study I attempt to examine whether and how procedural injustice and other aspects of police behavior influence the willingness of Ghanaians to support vigilantism. The following hypotheses are tested in this study.
Figure 2. Support for Vigilante Self-Help, Mediated Through Trustworthiness
Hypothesis I
People who perceive the police as being ineffective in the provision of security and those who experience police corruption are more likely to express support for vigilante self-help than those who consider the police to be effective and have not experienced police corruption. This is one of the common hypotheses about how police behavior may affect public support for or recourse to vigilantism (Reference SmithSmith 2004; Reference BakerBaker 2002; Reference AdinkrahAdinkrah 2005). Yet there are almost no empirical studies that systematically investigate this repeatedly mentioned hypothesis in the literature on violent self-help. The present study offers a preliminary test of this hypothesis.
Hypothesis II
People who perceive the police as treating people procedurally unjustly are more likely to report support for vigilante self-help than those who perceive the police as treating people in a procedurally just manner. The discussion thus far should make this hypothesis self-explanatory. Suffice it to note here that this hypothesis is grounded in the argument that, as people are moral beings, their deference to legal authorities is often shaped by the perception that the actions and behavior of these authorities are consistent with normative expectations about fairness and equity. Consistent with this line of reasoning, I focus on whether perceptions of procedural justice affect the likelihood that people will hold views supportive of vigilantism.
Hypothesis III
People who perceive the police as being trustworthy are less likely to support vigilante violence than people who perceive the police not to be trustworthy. When people trust legal institutions, they are likely to develop normative commitments and attachments to these institutions and therefore be less inclined to challenge the legitimacy of these institutions.
Vigilantism emanates from a process of evaluation and judgment about the state's exercise of legitimate power; it is an attempt to exercise power by actors who do not have the official legitimacy to do so. But as some political scientists have argued, the judgments that people express about the legitimacy of political and legal institutions are linked to the status they occupy in a particular society (Reference CoicaudCoicaud 2002; Reference GilleyGilley 2006). Among the vast array of descriptive factors one might consider, education is perhaps the most relevant here. The key argument in most of these studies is not simply that academic education serves to generate greater participation in the political process (see, e.g., Reference Rosenstone and HansenRosenstone & Hansen 1993; Reference Shields and GoidelShields & Goidel 1997; Reference VerbaVerba et al. 1995). Education also assists the cultivation of democratic values; it broadens citizens' outlook and enables them to understand the need for “norms of tolerance” (Reference LipsetLipset 1959:79) and the rule of law rather than resort to violence. Lipset further argues that “the higher one's education the more likely one is to believe in democratic values and support democratic practices” (Reference LipsetLipset 1959:79).Footnote 3 Vigilantism is a subversion of the principles that underpin liberal democracy, especially the rule of law. The evidence on the link between education and democracy therefore leads me to a fourth hypothesis.
Hypothesis IV
People who have higher educational attainment are less likely to report support for vigilantism than people who have less educational attainment. Higher educational attainment will translate into an understanding that law enforcement is the primary responsibility of the state. Beneficiaries of academic education will therefore consider it unacceptable for nonstate actors to arrogate that responsibility to themselves; rather, if the state law enforcement is deficient, they will press for its reform by democratic means.
Policing in Ghana: A Brief Background
Since the colonial period, the Ghana police have faced various challenges. Colonial policing was conducted with little concern for procedural propriety and democratic accountability because these were not prerequisites for the gunpoint extraction of agricultural and mineral resources, which were used mainly for the development of British society. There have been few serious efforts at the legitimation and redefinition of the ideology of the police in the postcolonial period. General public expectations that the return to multiparty democratic governance in 1993 after decades of military rule would bring improvements in the quality of policing have remained a distant dream (Reference TankebeTankebe 2008a; Reference Aning and KarikariAning 2002). Abusive and intimidating policing practices and widespread corruption have generated an underlying sense of public mistrust and discomfort with the police (Reference Aning and KarikariAning 2002; Reference Appiahene-GyamfiAppiahene-Gyamfi 2002). The Ghana police force, then, appears to be an archaic institution, incorrigibly and unrepentantly allied to its historical colonial ideology and unresponsive to demands of the current liberal democratic political environment.
Crime and insecurity remain widespread, particularly since 1993, generating unparalleled levels of public anxiety and outrage about police ineffectiveness (Reference Appiahene-GyamfiAppiahene-Gyamfi 1998; Reference TankebeTankebe 2008b; Reference Karikari and KarikariKarikari 2002). Some sections of Ghanaian society have turned to private security firms to create fortresses and islands of security, while others have resorted to vigilantism and other forms of self-protection. Vigilante violence in Ghana is disproportionately concentrated in the major cities of Accra, Kumasi, and Tema (Reference AdinkrahAdinkrah 2005:417). It is significant, for the purposes of the argument of this article, that these cities have the strongest presence of the state and its legal apparatus. Collectively, these cities have approximately 20 percent of police personnel in Ghana (Reference TankebeTankebe 2007), a situation typical of policing in sub-Saharan Africa (Reference MareninMarenin 1982; Reference HillsHills 2000; Reference HerbstHerbst 2000).
Is recourse to vigilante self-help merely the product of perceived police corruption and ineffectiveness? May police adherence to normative requirements of procedural fairness shed light on the reasons for public expressions of support for vigilante self-help?
Data and Methods
The current study was conducted in summer 2006. The sample was selected through a multistage sampling technique involving the random selection of 450 households from Census Enumeration Areas (CEAs) in Accra, Ghana. The sampling occurred at three main stages. The first stage involved the use of CEAs with population and housing information from the 2000 Population and Housing Census as the primary sampling units. Seven CEAs were selected to represent the 1,724 CEAs (Ghana Statistical Service 2005:17). The second stage involved the selection of 60 households from the records relating to each CEA using a systematic random sampling technique. These CEAs did not, however, contain information on the individuals living in such households, and thus a third stage became necessary in the sampling process. From each selected household, the researcher compiled a full list of all individuals aged 18 years and above at the time of the study using surnames in alphabetical order. In the overwhelming number of cases, this involved three to four people. A person was then randomly selected from this sampling frame as a research participant. Where a randomly sampled house had more than one household, one household person was randomly selected to be the research participant (United Nations Statistics Division 2005).
The questions used to operationalize each of the variables in the study are shown in the Appendix. Of the 450 questionnaires given to the selected research participants, 374 were returned. This represents a response rate of 83.1 percent. The age of the research participants ranged from 18 to 70 (mean of 31). The sample was 39.3 percent female, and all participants had at least some education (20 percent had a college-level education). Approximately 53 percent had suffered personally as victims of crime within the 12 months preceding the survey. The incidence of victimization is rather high compared to the findings reported by Reference Adu-MirekuAdu-Mireku (2002). In a survey of 208 residents of three communities in Accra, Adu-Mireku found that only 33.9 percent of the respondents in the study had been victims of a crime in the 12 months preceding the study (2002:160). The risk of victimization reported in the present study was even higher when compared to the result from the British Crime Survey, which was 23 percent, in the last 12 months preceding the survey (Reference WalkerWalker et al. 2006:14). This may be less surprising if one considers the growing public anxiety about crime in Ghana, which official statistics, despite their flaws, do not repudiate (see Reference TankebeTankebe 2008a; Reference Appiahene-GyamfiAppiahene-Gyamfi 1998). Finally, 34.5 percent of the research participants had had a recent direct encounter with the police.
In considering the findings presented in this study, it is important to recognize possible biases in the results that might be ascribed to nonresponse bias. Because the questionnaires were in some cases left with the research participants to be collected at a later agreed date, it is possible that the selected participants might not have answered the questionnaires themselves. This appears to have resulted in the oversampling of male research participants and the undersampling of less well-educated persons. To correct the gender bias, I weighted the data to reflect more accurately the gender of the population of the Accra metropolitan area as measured in the 2000 Housing and Population Census (Ghana Statistical Service 2005). This weighting produced an adjusted sample of 51 percent female participants. It is important to note that the weighted data altered only the findings about age; that is, age was not a significant predictor of support for vigilantism when the analysis was conducted with the unweighted data, although—as shown later—it was significant when using weighted data.
Findings
The data were analyzed using ordinary least squares (OLS) regression. The purpose of the regression analysis was twofold: first, to establish the unique effects that each independent or explanatory variable had on the outcome variable (i.e., support for vigilantism) while controlling for other variables in the model, and second, to ensure that conclusions about the significant impacts of any of the explanatory variables were not dependent upon the impacts of other variables in the equation (Reference Hutcheson and SofronniouHutcheson & Sofronniou 1999). Before proceeding with the regression analysis, I conducted a correlation analysis to determine whether there were cases of very high relationships among the explanatory variables to suggest problems of multicollinearity. There were no such cases. Results for the Variance Inflation Factor (VIF) showed that the tolerance statistics scores were all far greater than 0.2. As a general rule, Reference MenardMenard (1995) recommends that values below 0.2 should be a cause for concern because they may suggest that multicollinearity has unduly influenced OLS estimates.
Table 1 presents results from four OLS regression models. These cumulatively included what might be described as control, performance, procedural justice, and trustworthiness variables. Considering the control variables (see Model 1), the results showed that age and educational attainment were each significant predictors of support for vigilante self-help. More specifically, people who were older (β=0.153, p<0.05) and those who were less educated (β=−0.196, p<0.01) were more likely to express support for vigilante practices than younger and better educated people. Gender, previous encounters with the police, victimization, and political affiliation were, however, not statistically significant predictors of support for vigilantism. Age and education together accounted for only 7 percent of the variance in public support for vigilante practices.
Table 1. Determinants of Public Support for Vigilantism in Ghana
Note: Standard errors are in parentheses. n=374.
* p<0.05,
** p<0.01,
*** p<0.001
In the second model, performance-related variables were incorporated into the model. The results showed that neither perceived police effectiveness in dealing with crime nor respondents' experience of police corruption had a significant effect on public support for vigilante self-help. This result is noteworthy, given that much of the attempt to explain violent self-help, particularly in sub-Saharan Africa, has tended to suggest that perceptions of police ineffectiveness ipso facto lead to public support for violent self-help. The only performance-related variable that was a significant predictor of vigilante support was judgments about police anti-corruption measures (β=−0.162, p<0.05); that is, the more the participants thought the police leadership was doing well in tackling police corruption, the less willing they were to express support for vigilante violence. The effects of both age (β=0.127, p<0.05) and education (β=−0.233, p<0.001) remained statistically significant when the performance factors were controlled. Overall this second model explained approximately 11 percent of the variance in vigilante self-help.
Model 3 incorporated all the control, performance, and procedural justice variables in the equation. Quality of treatment was the only procedural justice variable that had a significant effect on support for vigilantism (β=−0.177, p<0.05). Specifically, participants who rated the quality of police treatment of citizens highly were less likely to express support for vigilante violence. Contrary to expectations, neither distributive fairness nor quality of police decisionmaking was a significant predictor of vigilantism. Procedural justice researchers have often tested the combined effects of quality of treatment and quality of decisionmaking to determine the effects on public compliance and cooperation (see, e.g., Reference Sunshine and TylerSunshine & Tyler 2003; Reference Tyler and TonryTyler 2003). Adopting this approach, both of these subscales were combined to form an overall procedural fairness index. The results, not presented here, showed that the combined effects were significant (β=0.170, p<0.05); that is, participants who perceived the police to be procedurally fair were less likely to support vigilantism. As shown further in Table 1, the effects of age (β=0.145, p<0.05), education (β=−0.251, p<0.001), and corruption reforms (β=−0.141, p<0.05) remained statistically significant, even when procedural justice variables were entered in the model.
Finally, in Model 4, police trustworthiness was introduced. As hypothesized in Figure 2, this transcended the effects of aspects of police performance and procedural justice that had previously been significant predictors of vigilantism. This suggests that the importance of those variables in the explanation of support for vigilante self-help may be that they lead to perceived police trustworthiness, which is the key variable. In this final regression model, the strongest predictor of vigilante support was police trustworthiness (β=−0.395, p<0.001). In other words, people who judged the police to be less trustworthy were more likely to express support for vigilante violence. Interestingly, even with all the other variables (including trustworthiness) in the model, both age (β=0.127, p<0.05) and educational attainment (β=−0.221, p<0.01) remained significant predictors of vigilante support. Put differently, among the sample surveyed, the typical profile of a vigilante supporter was an older person with low educational attainment who did not trust the police. These three variables explained a total of 20 percent of the variance in support for vigilantism.
Discussion and Conclusion
Farrington and his colleagues have argued that “[i]n order to draw conclusions about individuals, it is necessary to carry out research based on individuals” (Reference FarringtonFarrington et al. 1986:336). This is what the present study aimed to achieve through a self-reported study that sought to explore the factors that explained expressions of support for vigilantism in Ghana. There are three particularly important findings from the Ghana study. First, and most important, public support for vigilante self-help was fundamentally linked to people's judgments about the trustworthiness of the police; people who perceived that the police were not trustworthy were more likely to support vigilantism than people who perceived the police to be trustworthy. Reference Mishler and RoseMishler and Rose (2001) have argued that trust links ordinary citizens to political and legal institutions, thereby enhancing the legitimacy of these institutions. It follows that a widespread perception of police untrustworthiness can undermine the police's exclusive claims to legitimate power to deal with lawbreaking behavior. Yet due to historical factors—such as Ghana's colonial heritage and police brutality, particularly during prolonged periods of military rule—and subsequent poor police performance, both procedurally and substantively (including widespread corruption), many Ghanaians remain highly distrustful of the police; hence the widespread reports of vigilante violence in Ghana (Reference Amoakohene and KarikariAmoakohene 2002; Reference AdinkrahAdinkrah 2005; Reference Karikari and KarikariKarikari 2002).
The second noteworthy finding concerns police performance and procedural justice. When perceived police trustworthiness was not controlled for, the findings showed that, among variables other than demographic variables, only “satisfaction with police corruption reforms” and “perceived quality of treatment in encounters with police” were significant predictors of support for vigilante violence. The former appears consistent with the argument that public satisfaction with the performance of a police department in tackling corruption can often be more important to its moral standing than the fact that people encounter instances of police corruption (Reference Tankebe and CheliotisTankebe in press). That is, if police managers can convey a message of effectiveness in attempted corruption reforms, this should persuade members of the public that an instance of police misbehavior that causes much resentment and injustice is being seriously attended to by the relevant authorities.
Contrary to many current understandings of vigilantism, neither experience of police corruption nor perceptions of police (in)effectiveness were significant predictors of support for vigilantism in the present study. Of course, it would be wrong to suggest that people will not support or even resort to vigilante violence when there is widespread insecurity. The challenges that violent crimes pose in Ghana are undoubtedly enormous; hence the need for a well-equipped and effective police force to provide people with a minimum of security. For a militaristic police organization used to authoritarian and abusive policing strategies, there is the danger that the police will take advantage of people's security fears and respond to crime in a manner that undermines democratic practices and civil rights (Reference Karikari and KarikariKarikari 2002). What the present findings suggest, however, is that vigilante support has its roots in normative (“trustworthiness”) rather than instrumental (“effectiveness”) assessments of policing. The finding about aspects of police procedural fairness underscores the importance of police adherence to the normative expectations of fair and humane treatment of citizens. Reference FukuyamaFukuyama (1992) aptly observes that the demands that citizens make in the modern state hinge on the need to be treated with sufficient personal dignity by figures of authority. Just treatment is an essential condition under which people identify with the state, “feel one with it, and accept its ruling as [their] own” (Reference LucasLucas 1980:1). Thus what poor police treatment of citizens does is to weaken moral identification with the police institutions, which then allows citizens to arrogate police powers to themselves. But as the overall findings demonstrated, the effects of performance and procedural justice variables on vigilante support appeared to be embraced within perceived police trustworthiness.
The third noteworthy result concerned the role of academic education. The data showed that the higher people's level of educational attainment, the less likely they were to express support for vigilantism. This finding appears supportive of the broader empirical analyses, which have established a strong, positive relationship between formal education and behavior and attitudes favorable to democracy. Education equips people with the requisite skills to understand what may often appear as the abstract subject of politics and the rule of law (Reference Rosenstone and HansenRosenstone & Hansen 1993:136). The Ghana data seem to show that increased literacy leads to the development of greater awareness, understanding, and tolerance of due process of criminal justice practice and rule of law, the essential principles of democratic governance. As academic education levels rise across Ghanaian society, the positive effects of increases in knowledge and understanding of rule of law may work to suppress the current alacrity with which people resort to vigilante violence. It is perhaps appropriate to sum my argument here with a variation of Reference LipsetLipset's (1959:80) conclusion on the education-democracy hypothesis: if we cannot say that greater educational attainment is a sufficient condition for suppression of vigilante violence, the findings reported here do suggest that it comes close to being a necessary condition for the suppression of attitudes that are supportive of vigilante violence.
I do not suggest that mere expressions of support automatically translate into actual behavior; numerous situational variables become decisive in translating attitudes into vigilante participation. Attitudes favorable to vigilantism can, however, hinder a person's readiness to intervene, if he or she has the power to stop a vigilante act. This is where the positive relationship between age and vigilante attitudes becomes important and helps us understand why acts of vigilantism in sub-Saharan Africa are mainly carried out by youth (see Reference SmithSmith 2007) even though the findings reported in the present study appear to suggest that they hold less favorable attitudes to vigilantism. In a sociopolitical setting where there remains strong respect for old age, could active or passive encouragement by older citizens be a contributory factor to the current widespread recourse to vigilantism? The data seem to demonstrate that, while vigilantism may undoubtedly express “the anger of youth about inequality and injustice” (Reference SmithSmith 2007:169), it is the ideological support of adult members of Ghanaian society that emboldens them to do so.
Undoubtedly, vigilante acts are to be strongly excoriated, but to assume that such excoriation is, by itself, enough to stem these acts, while failing to grasp the roots of vigilantism, misses the point. It achieves nothing to remind the public of the legal and constitutional provisions that prohibit the extralegal exercise of powers otherwise reserved exclusively for the criminal justice system, since, as Reference BeethamBeetham (1991) and Reference CoicaudCoicaud (2002) argue, legality alone constitutes an insufficient premise for successful legitimation claims. Indeed, many members of the public are likely to receive these reminders with contempt because it seems inconceivable to them that the police should be afforded a monopoly in the exercise of force while failing in their own obligations. Nor are the current exhortations by some commentators to punish, severely, vigilante perpetrators the best way to address the problem. Certainly, and considering that vigilantism is a form of defiance directed at the nature of police exercise of power, “the decisive assertion of symbolic territory by public retribution discourages others from venturing [similar] public defiance” (Reference ScottScott 1990:197). The success of such a policy is doubtful, precisely because it does not address the problem of trust and the poor treatment of citizens, which seem to underpin public support for vigilante activity. Addressing the vigilante problem requires the implementation of democratic policing mechanisms with the aim to ensure that policing not only is responsive to public safety needs, but also adheres to democratic principles of respect for the dignity and civil liberties of citizens in order to build public trust in the police. As Reference YoungYoung has argued persuasively, however, the state of a police force is “a reflection of the way of life, standard of probity, as well as a measure of political responsibility of the community” (1951:1). Thus democratic reforms of the Ghana police are unlikely to succeed if corruption and impunity continue to be widespread features of public life.
Appendix
Procedural Justice
Tyler and his colleagues have argued that procedural justice encapsulates two principal dimensions: procedural fairness and distributive fairness (see, e.g., Reference Tyler and HuoTyler & Huo 2002; Reference Tyler and BladerTyler & Blader 2000; Reference Tyler and TonryTyler 2003). Procedural fairness was measured with items developed by Reference Sunshine and TylerSunshine and Tyler (2003) in their study in New York. A factor-analytic technique known as principal component analysis (PCA) was employed to determine whether all items adapted from the Sunshine and Tyler study would load on the same scale in Ghana. Although there are various techniques available for this exercise, Spector has argued that PCA “would seem a reasonable factor analytic model to use” in social science research (cited in Reference ReisigReisig et al. 2007:1015). The results showed that the items loaded on two subscales: Quality of Treatment and Quality of Decisionmaking. The questions used to measure both scales had a Likert-type response set: 1=strongly disagree to 5=strongly agree.
For Quality of Treatment, the research participants were presented with the following statements: (1) The police treat everyone with respect. (2) The police treat everyone with dignity. (3) The police treat everyone equally. (4) The police respect people's rights. (5) The police follow through on their decisions and promises they make (alpha value=0.83, mean=2.57, s.d.=1.02).
For Quality of Decisionmaking the items were: (1) The Ghana police always act within the law. (2) The police take account of the needs and concerns of people they deal with. (3) The police sincerely try to help people with their problems. (4) The police clearly explain the reasons for their actions. (5) The police try to find the best solutions for people's problems. (6) The police provide opportunity for unfair decisions to be corrected. (7) The police use rules and procedures that are fair to everyone. Each question had a Likert-type response set: 1=strongly disagree to 5=strongly agree. All items were combined into an overall index of quality of decisionmaking (alpha value=0.84, mean=3.02, s.d.=0.91).
Distributive Fairness
Distributive fairness was measured by a simple two-item Likert-type scale to indicate people's satisfaction with the fairness of both specific and general outcomes in transactions with the police. The items were as follows: (1) How fair are the outcomes you or someone you know received from the police? (2) In general, how often do you think the outcomes people receive from the police are what they deserve under the law? Both items were combined to create a summary index of distributive fairness (alpha=0.73, mean=1.51, s.d.=0.77).
Trustworthiness
The items used to measure police trustworthiness were adapted from Reference Sunshine and TylerSunshine and Tyler (2003) and included the following: (1) The Ghana police are trustworthy. (2) I am proud of the police in Ghana. (3) I have confidence in the Ghana police. (4) The Ghana police are often dishonest (reversed). (5) The Ghana police are usually honest. (6) The Ghana police always act within the law. Each question had a Likert-type response set: 1=strongly disagree to 5=strongly agree. The items were combined to create a summary index of police trustworthiness (alpha value=0.80, mean=2.55, s.d.=1.08).
Police Effectiveness
The items used to measure public perceptions of police effectiveness, adapted from Reference Sunshine and TylerSunshine and Tyler (2003), encapsulated concerns about police responsiveness and neighborhood safety. There were 10 items in all: (1) The police respond promptly to calls about crimes (e.g., robbery, assault). (2) The police are always ready to provide satisfactory assistance to victims of crime. (3) The police are always able to provide the assistance the public need from them. (4) The police are doing well in controlling violent crime (e.g., armed robbery). (5) Crime levels in my neighborhood have changed for the better in the last year. (6) There are not many instances of crime in my neighborhood. (7) I feel safe walking in my neighborhood in the night. (8) Overall my neighborhood is a good place to live in terms of security. (9) Overall the police are doing a good job in my neighborhood. (10) When the police stop people they usually handle the situation well. Each question had a Likert-type response set: 1=strongly disagree to 5=strongly agree. These were then combined to create a police effectiveness index (alpha value=0.86, mean=3.35, s.d.=0.84).
Corruption
Public experience of police corruption was measured with five items: (1) Have you ever paid money to a police officer or promised the officer a favor to overlook your unlawful behavior (e.g., speeding, assault)? (2) Have you ever witnessed somebody pay money to a police officer or promise the officer a favor for the officer to overlook their unlawful behavior (e.g., speeding, assault, theft)? (3) Have the police ever refused to investigate, arrest, charge, or prosecute you because you are related to a police officer? (4) Have you ever made use of somebody related to a police officer to prevent a case being pursued against you? (5) Have you ever personally known a situation where the police decide not to investigate, arrest, charge, or prosecute somebody because that person is a relative or a friend? Each question had a Likert-type response set: 1=never, 4=many times. The items were combined to create an index of police corruption (alpha value=0.71, mean=1.28, s.d.=1.20).
Tankebe (in press) has argued that while public experiences of police corruption may be important in undermining police legitimacy, the perception that the police leadership is not doing enough to deal with the problem can be even more significant. The present study therefore sought to measure public assessments of police anti-corruption reforms performance. One survey question was used to measure this: “The police leadership is doing enough in tackling corruption with the police service.” There were four responses ranging from 1=quite poor to 5=very good (mean=2.68, s.d.=1.48).
Vigilantism
Public attitudes toward vigilantism were measured using a Likert-like scale with responses ranging from strongly disagree to strongly agree. The items included: (1) It is all right for members of the public to beat up crime suspects. (2) People who kill armed robbers should not be blamed. (3) It is sometimes OK for people to take the law into their own hands if they feel the police are unable to protect them. (4) It is pointless to hand over a suspected criminal to the police because they won't bring the offender to justice. (5) Each community should organize itself to provide it with security against criminals even if the police disagree with that. Responses to these questions were summed to create an index of support for vigilantism (alpha=0.70, mean=3.15, s.d.=0.85).
