Recent advances in criminological research have revealed that social context can influence delinquent behaviour separately from the characteristics of individuals residing in a given neighbourhood. However, the link between structural conditions and behavioural outcomes is still an undeveloped area of investigation. Social disadvantage, while linked historically in many studies to high rates of youthful offending, does not operate consistently. The concept of collective efficacy has been introduced to explain why some disadvantaged neighbourhoods do not demonstrate excessive criminality. The key question is what social mechanisms help to distinguish adolescent offending in more and less disadvantaged social contexts. We identify alcohol consumption, with its strong empirical association with violent crime, as a key factor in understanding the complex relationship between different forms of collective efficacy and the occurrence of violent crime within neighbourhoods of differing levels of disadvantage.
Since youthful violent crime, while relatively uncommon in the context of the entire panoply of delinquent behaviours, gives rise to much social and political concern, approaches to prevention also consider what other protective factors might constrain the most serious forms of youthful offending. Since the primary institutions of social control, the family and the school, are also embedded in neighbourhood contexts, we will extend the concept of collective efficacy to create new measures -family and school efficacy - which will serve as important controls in our models. These will be examined, in conjunction with individual factors and alcohol use, to test a complex model of variation in violence across neighbourhoods with differing levels of social disadvantage.
While many studies elsewhere have looked at individual characteristics and neighbourhood effects on crime rates, this is the first Canadian study to apply the concepts of social disorganization and collective efficacy and novel measures of family efficacy and school environment in an explanation of youthful violence (see Browning 2012 for effects on alcohol use and Browning and Erickson 2009 for effects on violent victimization). Moreover, this study is unique in providing a representative sample of adolescents, with variation in age, gender, and ethnicity as well as in the consumption of alcohol, and a wide range of violent occurrences. Understanding the complex nature of youthful violence is essential for the creation of effective prevention and intervention programs that suit both youth and neighbourhoods.
The importance of alcohol
The correlation between alcohol use and violent crime, involving both victims and perpetrators, has been well documented in the literature (Harrison, Erickson, Adlaf, and Freeman 2001; Welte, Barnes, Hoffman, Wieczorek, and Zhang 2005; Mulvey, Odgers, Skeem, Gardner, Schuber, and Lidz 2006; Browning and Erickson 2009). Even when other substances (both licit and illicit) are considered, alcohol has the strongest association (Boles and Miotto 2003; Felson, Teasdale and Burchfield 2008; Erickson, MacDonald and Hathaway 2009). This association holds at both the individual and aggregate levels and is particularly strong among adolescents and younger adults (Lurigio and Davis 1992; Harrison and Gfroerer 1992; Menard, Mihalic, and Huizinga 2001). Alcohol is the most commonly used substance among high school students and allows for more reliable estimates than do less commonly used substances. However, there is little research on the variance of the alcohol/violence nexus by neighbourhood context. If neighbourhoods differ in the levels of youth substance use, then it is reasonable to assume that these macro-level differences may affect the strength or very existence of the link between alcohol and violence. Moreover, since most alcohol consumption does not lead to violence, attention must be paid to other conditions that constrain its expression (Pernanen 1991). Including the neighbourhood context allows us to examine the impact of neighbourhood-level differences on the ability of social institutions, such as schools and families, to curb violent offending among adolescents, while considering different rates of alcohol consumption.
While most studies report a positive association between substance use and violent offending, different causal explanations are offered (Kaplan, Tolle, and Yoshida 2001; McNulty and Bellair 2003). Some argue that substance use only provides the context in which violence occurs, and thus the relationship between substance use and violence is not necessarily causal (Fagan 1993). Others have found evidence that the particular pharmacological effects of a drug make violence more likely (Fagan 1993; Boles and Miotto 2003; McNulty and Bellair 2003). In a rare study of this issue using a non-U.S, sample (Finland), Felson et al. (2008) found at least partial evidence of a causal relationship between alcohol use and violent offending among delinquent youth. At the city level, the strong relationship between the use of alcohol and violent offending persists even after controlling for socio-economic status (SES), heroin use, and autocorrelation (Martin, Maxwell, White, and Zhang, 2004). However, other elements of the neighbourhood or social context may moderate the association (Fagan 1993).
Research findings on collective efficacy and social disorganization
First, we will consider the development of these concepts and then consider their relationship to adolescent violence as found in the literature. Our theoretical framework combines themes from Shaw and McKay's social disorganization, Hirschi's social control, and Sampson and various colleagues' collective efficacy. Generally, social disorganization theory links neighbourhood disadvantage, in the form of low-income concentration or high concentrations of ethnic minorities/ immigrants, with high rates of crime. Shaw and McKay (1942) did not believe this link was direct but operated through disrupted informal social control. Essentially, social disorganization theory operates as a social control theory at the neighbourhood level. Later, Sampson and his colleagues used the term collective efficacy to refer to the informal social control used by neighbourhoods to control crime within their boundaries (Sampson 1985; Sampson and Groves 1989; Morenoff, Sampson, and Raudenbush 2001). It refers to a shared set of values and sense of trust among neighbours with regard to ensuring neighbourhood safety (Sampson, Raudenbush, and Earls 1997).
Many American as well as recent studies in Europe and Montreal have been consistent in finding a positive relationship between neighbourhood disadvantage and violent offending (see Peeples and Loeber 1994; Kupersmidt, Griesler, DeRosier, Patterson, and Davis 1995; Ouimet 2000; Oberwittler 2004). Less conclusively, the majority of previous studies of collective efficacy have found it is inversely related to violent offending (see Morenoff et al. 2001; Almgren 2005). However, these studies tell us little about how the neighbourhood context may affect the behaviour of individuals and nothing about how such contexts may affect the strength of various individual-level protective and risk factors. Since the neighbourhood provides the context in which individual-level behaviour takes place, it is important to determine how variations in context may translate into variations in individual behaviour. For instance, Hay, Fortson, Hollist, Altheimer, and Schaible (2006) found that neighbourhood disadvantage amplified the negative impact of a composite measure of overall family problems on violent offending by adolescents; however, other than physical punishment, individual components of this measure were not significant. Moreover, a recent Dutch study that included several cities as well as neighbourhoods found that including cities added to the amount of variance explained beyond the neighbourhood level (Weijters, Scheepers and Gerris 2007).
A large body of prior research has concluded that neighbourhood disadvantage (specifically immigrant concentration, concentrated poverty, minority concentration, and residential mobility) accounts for much of the individual-level differences in violent offending (Peeples and Loeber 1994; Kupersmidt et al. 1995; Silver 2000; Wikstrom and Loeber 2000; McNulty and Bellair 2003; Sampson, Morenoff, and Raudenbush 2005; Haynie, Silver, and Teasdale 2006). Peeples and Loeber (1994) found that blacks committed more frequent violent acts of a serious nature than did whites. However, this apparent racial effect disappeared once they accounted for neighbourhood economic disadvantage. McNulty and Bellair (2003) reached the same conclusion through their use of the U.S. Adolescent Health Data.
In a study that included both neighbourhood disadvantage and neighbourhood social ties, researchers found that social ties did not mediate the direct effects of ethnic heterogeneity or neighbourhood poverty on the likelihood of assault (Warner and Rountree 1997). Wikstrom and Loeber (2000) found that, in the most disadvantaged neighbourhoods, there was no difference in male aggressive offending based on a wide array of risk and protective factors, including school, family, and peers. For all other neighbourhoods, the likelihood of aggressive offending increased with increases in risk factors and decreases in protective factors. However, when the balance of risk and protective factors was held constant, increases in neighbourhood disadvantage were positively associated with increases in aggressive offending (Wikstrom and Loeber 2000).
In a spatial analysis study, Morenoff et al. (2001) discovered that high levels of collective efficacy were associated with low rates of homicide. They also determined that collective efficacy mediated some, but not all, of the effects of neighbourhood disadvantage on homicide rates...