Multi-Level Effects on Behaviour Outcomes in Canadian Children - May 2001

2.1. Introduction

Previous

Contents

Next

Sociological crime and delinquency research has been restrictive in focusing on direct predatory forms of physical aggression among adolescents and adults. In contrast, psychological research (Lagerspetz, Bjorkqvist, & Peltonen, 1988; Pepler & Sedighdeilami, 1998; Tremblay, 1991, 1999; Tremblay et al., 1996) has examined distinctions in the forms of childhood aggression, differentiating direct physical aggression (e.g., childhood hitting, kicking, and biting) from indirect non-physical aggression (e.g., the intentional exclusion of others from play, rumors, and gossip). Meanwhile, a range of sociological perspectives have drawn attention to the need for research to examine variability in individual level outcomes in relation to a broader range of factors measured at the family or neighbourhood levels of analysis. This report has the objective of blending psychological and sociological perspectives to examine how multi-leveled risk factors including individual, family, and neighbourhood effects influence both direct physical and indirect forms of aggression in childhood.

Building on prior research that has partitioned the variance in children's physical and indirect aggression into between-family and within-child variability (Tremblay et al., 1996), this research will investigate whether the variance in aggression can additionally be explained by between-neighbourhood variability. The inclusion of neighbourhood features in this research will broaden the consideration of structural risk factors for both indirect and direct physical aggression. Both "objective" (e.g., census characteristics) and "subjective" neighbourhood characteristics (e.g., perceived neighbourhood problems) are included to assess contextual influences on childhood aggression (see Upchurch, Aneshensel, Sucoff, & Levy-Storms, 1999). This study also examines how, or through what model form, neighbourhood characteristics affect aggression.

The current research also adds to a growing body of research investigating the impact of neighbourhoods on children's well-being. Since Canadian studies have found patterns of increasing neighbourhood disadvantage (Hajnal, 1995; Hatfield, 1997; Myles, Picot, & Pyper, 2000), the implications of neighbourhood features for children in Canada warrant further investigation (see also Boyle & Lipman, 1998; Kohen, Hertzman, & Brooks-Gunn, 1998). Some of the policy implications of the research findings on neighbourhood and family disadvantages for children are also considered.

Structural approaches in sociology share a concern with the implications of social inequalities for individuals. A range of criminological theories and research examines and explains the patterning of crime and delinquency by age, gender, race/ethnicity, family structure, and social class position (Hagan et al., 1989; Hagan & Peterson, 1995). These socio-demographic factors also operationalize structural inequalities in mental health research, yielding differential levels of stress exposure and levels of coping resources, which combine to affect symptoms of psychiatric disorders and psychological distress (Aneshensel, 1992; Avison & Gotlib, 1995; Horwitz & Scheid, 1999; Mirowsky & Ross, 1989; Pearlin, Menaghan, Lieberamn, & Mullan, 1981; Turner, Wheaton, & Lloyd, 1995). Both of these sociological fields reveal that as disadvantages increase in individuals' lives, so do the likelihood of problematic outcomes including decreased well-being and increased deviant behaviour. These areas have emphasized the need to further examine the role of neighbourhoods as sources of social inequality on childhood outcomes (Avison, 1999b; Cohen, Slomkowski, & Robins, 1999; McLeod & Nonnemaker, 1999; Pearlin, 1999; Sampson, 1997). These perspectives in sociology are then linked in this research to approaches in developmental psychology to further examine how structural inequalities affect younger children (see also McLeod & Shanahan, 1993; Menaghan, 1999; Pepler & Sedighdeilami, 1998; Tremblay et al., 1996).

2.1.1 Gendered Outcomes

Aneshensel and colleagues (1991) raised the issue of including multiple outcomes across domains in mental health research. In their research, a model examining structural effects on substance use/dependence (which is more common among males) was compared with the results of a second model examining affective or anxiety disorders (more common to females). These results showed that the effect of being female was to increase affective or anxiety disorders, but was to decrease substance abuse/dependence. They conclude from these results that "men and women appear to be similarly affected by the types of stressful events and circumstances considered here, but these effects are manifest as different types of disorder" (p. 174). We suggest that considering gendered manifestations of behaviour more broadly could expand the significance of this framework. This focus would clarify whether social structural effects across levels of analysis are common across males and females or have gender-specific effects. While research on child and youth well-being in sociology has investigated how environmental features may affect more composite outcomes (e.g. externalizing and internalizing behaviour problems) or more specific outcomes like predatory delinquency, qualitative distinctions in aggression have received little attention.

One of the forms of research addressing the implications of structural inequalities concerns circumstances that produce gendered outcomes through differential exposure or vulnerability to social stressors (i.e., a "structural strain" perspective; see Aneshensel, 1992; Kessler & McLeod, 1984; Mirowsky & Ross, 1995; Turner & Avison, 1989; Turner et al., 1995). Research shows a consistent pattern of higher levels of psychological distress among women than men. Further analyses have examined whether this gender difference may in fact be due to two rival hypotheses outlined in the "gendered response" and "response bias" theories (Mirowsky & Ross, 1995). The gendered response theory predicts that men and women may have different emotional responses; therefore men may be as likely to be distressed as women, but may show this distress through anger rather than depressive symptoms. In contrast to this prediction, empirical research has found that the effect of being female is to increase both depressive symptomatology (e.g., sadness) and anger. By any of the six measures assessed, women experience more distress than men (Mirowsky & Ross, 1995, p. 463). A second potential explanation for the gender gap in distress involves response biases to survey items where women may be more expressive of their emotions than men. This perspective suggests that differences between male and female levels of distress are artifactual and due to response styles. However, the observed gender gap in distress when measured by either distress or anger holds net of a measure of differential expressiveness (Mirowsky & Ross, 1995).

A summary of issues related to interpreting gender differences in outcomes considers how a "...disorder may take behavioral forms as well as emotional ones" (Mirowsky & Ross, 1995, p. 465). Although the behavioural manifestations of alcoholism, drug abuse, and antisocial behaviour may be more prevalent among males than females, these outcomes are not themselves necessarily indices of psychological/emotional distress according to this view (see Mirowsky & Ross, 1995). Rather, these outcomes may be indicative of differential behavioural responses to societal strains by gender.

Within unequal social structures, behavioural responses may be considered as adaptations to environmental constraints and opportunities (Patillo-McCoy, 1999; Suttles, 1968). Gender role socialization occurs within social structure, and as Heimer (1995) also indicates in research on violence, "...the specific form of these acts varies with position in the power-structure" (p. 143). Rather than focusing on a range of disorders within mental health research per se (see Aneshensel et al., 1991; Rosenfield, 1999), the gendering of violent and delinquent behaviour more specifically (Heimer & DeCoster, 1999), or on the co-occurrences of multiple problems (Kessler et al., 1994; Loeber, Farrington, Stouthamer-Lober, & Van Kammen, 1997), this research examines distinctions within a specific form of childhood behaviour at an early life stage: childhood aggression.

This report examines how neighbourhood and family conditions affect the potentially gendered manifestations of behaviour in the forms of both indirect and physical aggression. Including only one form of aggression that may be more sensitive to the behaviours of males or females may potentially obscure conclusions regarding contextual influences. Policy implications of neighbourhood effects research may be clarified through the inclusion of both forms of aggression. Furthermore, research may not fully capture the effects of social structural locations on children and youth by operationalizing only more mainstream sociological outcomes, such as physical aggression.

2.1.2 Childhood Aggression

Physical aggression has been defined by Loeber and Hay (1997) as "...a category of behaviour that causes or threatens physical harm to others" (p. 371). Common features of aggressive acts include the potential for harm, that they are intentional, and they must be aversive to the victim (Coie & Dodge, 1998, p. 783-784). Consistent with the definition of physical aggression, indirect aggression involves anger and actual or potential harm in the forms of a damaged reputation or relationships, and psychological harm (see Buss, 1961; Coie & Dodge, 1998, p. 791; Crick, Bigbee, & Howes, 1996; Feshbach, 1969; Lagerspetz et al., 1988; Rutter, Giller, & Hagel, 1998, p. 148). While earlier definitions of indirect aggression included overt behaviours, more recent research examines a set of covert behaviours that include subtle differences among indirect, social, and relational aggression. These behaviours commonly involve the manipulation of the social structure with the intent to control others and induce harm (see also Rys & Bear, 1997; Verlaan, 1995; Xie, Swift, Cairns, & Cairns, 2000).

Research in the United States has suggested that the tendency in the literature on aggression to find higher levels among boys than girls may be due to the focus on physical aggression (see Crick & Grotpeter, 1995). A finding of a "lack" of aggression in girls may obscure attention from potentially harmful behaviours that girls may be more likely to engage in, and be the recipients of than boys (Crick & Grotpeter, 1995, p. 710). While males are seen as more likely to value instrumentality and physical dominance, girls may be more likely to value relational issues, including establishing intimate connections with others (Crick & Grotpeter, 1995; Taylor, Gilligan, & Sullivan, 1995). Children's strategies may be chosen on the basis of the likelihood that they will inflict harm (or aggress) by disrupting valued goals (Crick et al., 1996). Therefore, girls would be more likely than boys to engage in relational harm, while boys would be more overt in their strategies. Further, child socialization practices may differentially reinforce aggressive responses by gender (see Cairns & Kroll, 1994; Rosenfield, 1999).

Gender differences in peer reported relational and indirect aggression have been observed by Crick and colleagues in the United States, and in Finland by Lagerspetz et al. (1988) and Bjorkqvist, Lagerspetz, and Kaukiainen (1992). This research has found that females in middle childhood and early adolescence have higher levels of relational or indirect aggression than males. However, males show higher levels of physical and overt aggression than females. Recent Canadian research by Tremblay and colleagues (1996) used the person most knowledgeable about the child's ratings of indirect aggression. Similar to the peer assessed results, females at each age had higher levels of indirect aggression than boys, and boys had higher levels of physical aggression than girls. Although a range of structural effects on physical aggression and other externalizing behaviours have been examined, fewer environmental linkages have been drawn to indirect aggression. Two Canadian studies using the National Longitudinal Survey of Youth data have included social structural and processual effects on both forms of aggression including familial socioeconomic status and parenting effects (Pepler & Sedighdeilami, 1998; Tremblay et al., 1996). Social structure is expanded in this research to include aspects of children's social context including neighbourhood characteristics and additional family characteristics.

2.1.3 Multi-Level Risk Factors

A broad range of interdisciplinary research has raised the implications of contextual environments for individuals (Aaronsen, 1997; Anderson, 1990, 1997, 1999; Aneshensel & Sucoff, 1996a; Boyle & Lipman, 1998; Brooks-Gunn, Duncan, Klebanov, & Sealand, 1993; Brooks-Gunn, Duncan, & Aber, 1997a,b; Elliott et al., 1996; Garbarino, Dubrow, Kostelny, & Pardo, 1992; Garner & Raudenbush, 1991; Gephart, 1997; Jencks & Mayer, 1990; Kohen et al., 1998; Kowaleski-Jones, 2000; Leventhal & Brooks-Gunn, 2000; McLeod & Nonnemaker, 2000; Peeples & Loeber, 1994; Ross, 2000; Sampson, 1992, 1997; Sampson, Raudenbush, & Earls, 1997; Sucoff & Upchurch, 1998; Upchurch et al., 1999; Willms, 1986; Wilson, 1987). Although research has considered the role of context for physical aggression, the more subtle linkage to indirect aggression has emerged recently from qualitative research (Anderson, 1997).

Ethnographic research on neighbourhood disadvantage has provided some insight into gendered manifestations of the linkages between the "code of the streets" and the necessity of aggression to protect one's reputation as a requirement for daily negotiation of the environment. Among males, this is accomplished through physical aggression. Among females however, this negotiation of the environment may include more indirect means. As Anderson (1997) observes, "A major cause of conflicts among girls is 'he say, she say.' This practice begins in the early school years and continues through high school. It occurs when 'people,' particularly girls, talk about others, thus putting their 'business in the streets.' Usually one girl will say something negative about another in the group, most often behind the person's back. The remarks will then get back to the person talked about. She may retaliate or her friends may feel required to 'take up for' her. In essence, this is a form of group gossiping in which individuals are negatively assessed and evaluated. As with much gossip, the things said may or may not be true, but the point is such imputations can cast aspersions on a person's good name. The accused is required to defend herself against the slander, which can result in arguments and fights, often over little of real substance. Here again is the problem of low self-esteem, which encourages youngsters to be highly sensitive to slights and to be vulnerable to feeling dissed. To avenge the dissing, a fight is usually necessary" (p. 26).

The above quotation links gender and neighbourhood conditions. Context is implicated in this quotation with the structural conditions promoting cultural codes that reinforce "respect" on the streets, accomplished by physical means for males, and by reputation, interpersonal relations, and potentially physical means for females. These observations suggest expanding current sociological research on children's and youth's outcomes to include more sensitive measures of aggression that may better assess how behaviour is manifest by gender. Both relational and physical aggression may be seen as "self-protective" strategies in disadvantaged contexts, or as adaptations to structural circumstances (see also Suttles, 1968). In terms of the prevention of serious physical violence, it may also be advantageous to consider indirect aggression as a precursor, possibly identifying a point of intervention.

2.1.4 Neighbourhood Effects

Research has emphasized the continued need to examine how neighbourhoods affect children and youth (Aneshensel & Sucoff, 1996a,b; Boyle & Lipman, 1998; Brooks-Gunn et al., 1997a,b; Burton & Jarrett, 2000; Cook, Shagle, & Degirmencioglu, 1997; Duncan & Aber, 1997; Furstenberg, Cook, Eccles, Elder, & Sameroff, 1999; Kupersmidt, Griesler, DeRosier, Patterson, & Davis, 1995; Leventhal & Brooks-Gunn, 2000; McLeod & Edwards, 1995; Sucoff & Upchurch, 1998). A range of theoretical models specifying neighbourhood influences on children and youth outcomes have been described in the literature. These models are listed in Figure 2.1. As the articles and literature reviews referenced above provide a comprehensive overview of the empirical findings of neighbourhood effects on children and youth, this section will provide only a brief summary. Specific studies categorized by the type of model tested in the research are also reviewed in Foster (2000).

Figure 2.1
Theoretical Models of Neighbourhood Effects on Childhood Aggression

Structural	Mediational	Ecologial	Composite Risk
Neighbourhood SES Collective Socialization Institutional Resources Epidemic/ Contagion Relative Deprivation Competition Other Neighbourhood Dimensions	Familial Mediators Home Environment Community Mediators Norms/Collective Efficacy	Protective Other Potentiator Person-Environment Fit	Composite Score of Risks Including Neighbourhood Conditions

The models considered in this research draw on both the structural tradition emphasizing main effects (see Sucoff & Upchurch, 1998) and indirect mediational model forms (see Kowaleski-Jones, 2000; McLeod & Nonnemaker, 2000). The structural tradition suggests that neighbourhood risk factors exert "uniform" harmful effects on youth. Structural models emphasize the comparative effects of different neighbourhood factors, or the effect of a combination of factors measured by composite indices derived from census data (e.g., neighbourhood disadvantage). Approaches vary in specifying the model as a set of general risk conditions on a range of adjustment outcomes, or whether neighbourhood risk factors have been selected as potentially influential on a particular outcome (e.g. behaviour problems). The theoretical models have evolved from a focus on neighbourhood socioeconomic status to include a broader range of neighbourhood conditions (see also Duncan & Aber, 1997).

In column one of Figure 2.1, the five models identified by Jencks and Mayer (1990) are listed that have been tested in the literature to examine the effects of neighbourhood income levels on child and youth outcomes (Boyle & Lipman, 1998; Brooks-Gunn et al., 1997a,b; Kohen et al., 1998). The models specify which mechanisms in the neighbourhood environment affect children. Peers are emphasized as socializing agents in "epidemic" models, socialization by adults in the community are central to "collective socialization" models, and the influences of other adults not living in the community are included in "institutional" models. This set of models predicts that the presence of affluent neighbours will foster child development. In contrast, "relative deprivation" and "competition" models emphasize social comparisons and access to limited resources. They predict that affluent neighbours will hinder disadvantaged children. Finally, an additional "no effect" model predicts that affluent neighbours have no influence on children's outcomes (Mayer & Jencks, 1989).

Collective socialization and institutional resource models have been expanded to include the level of male joblessness in the community and ethnic diversity. Competition models have also included the concentration of families in the environment. Other key structural features of neighbourhoods include residential stability, immigrant concentration, the adult/child ratio, the population density, and neighbourhood family structure (Boyle & Lipman, 1998; Sampson, Morenoff, & Earls, 1999).

Findings have supported theories of collective socialization and neighbourhood resources to explain neighbourhood effects in early childhood, and additionally support competition theories for early school aged children. In an extensive review of the literature on neighbourhood effects on children and youth, recent research concludes: (a) across all of the outcomes, SES appeared to matter most, although the particular indicator of SES that mattered most varied by outcome. The strongest evidence was provided for the importance of high-SES neighbourhoods for achievement outcomes among both children and adolescents. Low-SES neighbourhoods and residential stability mattered most for adolescent juvenile delinquency. Low-SES neighbourhoods also seemed to be associated with young children's externalizing behaviour problems (Leventhal & Brooks-Gunn, 2000, p. 328).

The structural effects models assessed in this report address the following research questions: (1) Which features of the neighbourhood environment affect childhood physical and indirect aggression? (2) Are both objective and subjective environments influential on childhood aggression? Mediational models then expand the structural effects approach by investigating the processes through which neighbourhoods affect residents. The inclusion of processual variables operationalizes the intervening explanatory mechanisms (Cook et al., 1997; Leventhal & Brooks-Gunn, 2000, p. 321). Studies testing mediational models within the neighbourhood effects literature include four approaches: (1) aggregate level neighbourhood process mediators of neighbourhood structural effects on aggregate neighbourhood rate outcomes; (2) multilevel research with aggregate neighbourhood process mediators and individual level outcomes; (3) individual subjective perceptions of neighbourhoods on individual level outcomes; and (4) family level mediating processes with individual level outcomes. These models vary by the level of analysis of the mediators and outcomes, and the analytic techniques used to test the models. The mediating variables also vary by the life stage of those studied, and the theoretical orientation of the research. In research on children, theoretical perspectives have expanded to include a mediating role for family mechanisms (Klebanov, Brooks-Gunn, Chase-Lansdale, & Gordon, 1997, p. 121-122; Klebanov, Brooks-Gunn, & Duncan, 1994, p. 443; Furstenberg et al., 1999; Sampson, 1992, 1997). In their recent review article Leventhal and Brooks-Gunn (2000, p. 322) reformulate the main predictions of the Jencks and Mayer (1990) models to include three main potential mechanisms by which neighbourhoods affect children and youth: (1) institutional resources; (2) relationships; (3) norms/collective efficacy. The latter two mechanisms are suggested to be particularly important for delinquency.

A number of mediational models have been considered with different processes theorized as mediating. One set of these models includes collective subjective neighbourhood processes, aggregated to community level indices, to mediate the effects of neighbourhood structural characteristics on community and individual level outcomes (Sampson et al., 1997). At the individual level, some research considers how subjective dimensions of neighbourhoods may mediate the effects of antecedent neighbourhood conditions on a range of outcomes. It has also been suggested that these mediators of neighbourhood structural effects are themselves another form of neighbourhood effects on youth outcomes (Furstenberg et al., 1999).

Other mediating processes in recent research with children have included family mechanisms that transmit neighbourhood effects (Sampson, 1997). These familial mediators can be distinguished as within-home and extra-familial processes (Furstenberg et al., 1999). However, for young children, the most common and salient mediators include the within-home mediators including parental behaviour and the home environment (Leventhal & Brooks-Gunn, 2000, p. 324). Several reasons have been suggested in the literature for the lack of research on family mediating mechanisms (Cook et al., 1997), including that the traditional focus of the neighbourhood effects literature has been on adolescents (Klebanov et al., 1997, p. 120).

The main research questions used to investigate the mediational models include: (1) Does the perceived environment explain the effects of the structural environment on aggression? (2) Is the effect of neighbourhood conditions on both forms of childhood aggression reduced by indicators of the home environment?

Previous

Contents

Next