SHARON F. LAMBERT, CRYSTAL L. BARKSDALE, AND VON E. NEBBITTIT HAS BEEN WELL ESTABLISHED that high-risk behaviors among adolescents, such as substance use and unprotected or early sexual behavior, place adolescents at increased risk for several negative outcomes. Although current rates of adolescent alcohol and substance use are significantly lower than the peak rates in the 1970s and 1980s (Mulye et al. 2009), trends remain troubling. It is estimated that 26 percent of youth are considered to be heavy drinkers (i.e., they consume five or more drinks within several hours), and 20 percent of youth currently use marijuana (Centers for Disease Control and Prevention 2008). Similarly, although more than one-third of youth are currently sexually active, it is estimated that only 62 percent of them regularly use condoms (Centers for Disease Control and Prevention 2008). In addition, it is estimated that almost half of the annual 20 million new cases of sexually transmitted infections (STIs) occur among 15- to 24-year-olds (Centers for Disease Control and Prevention 2013; Weinstock, Berman, & Cates 2004). High-risk adolescent behaviors, such as substance use and risky sexual behavior, are significantly related to adverse outcomes, including depression and anxiety (e.g., Hallfors et al. 2005; Mason & Korpela 2009), poor self-esteem (e.g., Ethier et al. 2006), increased difficulties with self-control and emotion regulation (e.g., Fishbein et al. 2006), and increased risk of STIs and human immunodeficiency virus (HIV; e.g., Taylor-Seehafer, & Rew 2007). These adverse outcomes highlight the need to better understand the etiology of health-risk behaviors in order to inform preventive interventions. However, contextual determinants beyond the family and peer domains are not well understood.
Ecological models highlight the neighborhood environment as an important context for understanding developmental outcomes. These models suggest that the effects of the neighborhood may be direct and indirect, and they interact with other contexts important for development. Only recently, however, has the neighborhood context been integrated into studies examining the etiology of health-risk behaviors. This research has found that neighborhoods with a high concentration of poverty, high disorganization, and low social cohesion generally have higher rates of adolescent problem behaviors (Lanctot & Smith 2001; Seidman et al. 1998; South & Baumer 2000).
Neighborhood institutional resource models, collective socialization models, and contagion or epidemic models provide frameworks for understanding the observed associations between neighborhood characteristics and youth problem behaviors (Jencks & Mayer 1990). Resource models highlight police controls, as well as the availability and accessibility of neighborhood resources that promote healthy developmental outcomes (e.g., libraries, hospitals, community centers). Collective socialization models emphasize the importance of adult role models, supervision and monitoring of youth activities by neighborhood adults, and social organization for promoting positive youth adjustment. Contagion or epidemic models suggest that the negative behavior of neighborhood adults and peers spreads and can affect youth problem behavior, in part through social learning (Leventhal & Brooks-Gunn 2000). Although much of the research examining neighborhood and community context effects on adolescent development has focused on structural characteristics of neighborhoods such as income, employment rates, and residential instability (Leventhal & Brooks-Gunn 2000), youth perceptions of their neighborhoods are increasingly recognized as a valid indicator of the neighborhood context (e.g., Bass & Lambert 2004), with important implications for adolescent health-risk behaviors.
NEIGHBORHOODS AND SUBSTANCE USE
Neighborhood poverty, crime, and social disorganization, as assessed using census-based indicators, have been linked with increased substance use (for a review, see Scheier et al. 2001). In addition, adolescents’ subjective experience of their neighborhood environments, such as the perception of danger or threat, perception of drug use in the neighborhood, and perception of crime and violence, has been linked with adolescent substance use outcomes. For example, perceived neighborhood stress (Scheier et al. 2001) and perceived drug use in the neighborhood (Blount & Dembo 1984; Dembo et al. 1985) have been found to predict alcohol and substance use among racial and ethnic minority youth. Similarly, Lambert et al. (2004, 2005) found that African American adolescents who perceived more violence and drug activity in their neighborhoods were more likely to use substances than African American adolescents without such negative neighborhood perceptions.
It should be noted, however, that drug activity and the visibility of drug sales in a neighborhood do not necessarily indicate a high rate of substance use (Saxe et al. 2001). For example, drug sales may be a reliable source of income for some youth, but drug use may be considered unacceptable (Feigelman, Stanton, & Ricardo 1993). Thus, neighborhoods may provide access to substances, but substance use likely depends on norms and beliefs about use, and the presence, absence, or quality of resources that promote healthy adjustment and discourage use (Allison et al. 1999).
NEIGHBORHOODS AND SEXUAL RISK-TAKING
Neighborhood disadvantage has been linked with several risky sexual behaviors and outcomes, such as having sex more frequently and with several partners (e.g., South & Baumer 2001; Ramirez-Valles, Zimmerman, & Newcomb 1998), inconsistent contraceptive use (Baumer & South 2001), and teenage childbearing (e.g., South & Baumer 2001). Research examining neighborhood disadvantage and the timing of sexual activity, however, has produced mixed results (Browning et al. 2008; Dupere et al. 2008). There are specific elements of the neighborhood that have been associated with an adolescent’s sexual-risk behavior, including the presence of community-based sexual health services, the demography of the community (Billy, Brewster, & Grady 1994), and collective efficacy (Browning et al. 2008). While there are an increasing number of studies that have attempted to identify how these specific community elements, or mechanisms, are related to sexual-risk behavior (e.g., South & Baumer 2001; Browning et al. 2008; Cubbin et al. 2005), there has not been a systematic investigation into how youth-level factors, which may be affected by neighborhood factors, affect the relationship between neighborhood disadvantage and sexual-risk behaviors.
DEPRESSION AS A MEDIATOR OF THE LINK BETWEEN NEIGHBORHOOD AND HEALTH-RISK BEHAVIOR
The structural and social characteristics of the neighborhood are increasingly being recognized as having implications on many types of mental health problems among children and adolescents, including internalizing problems such as depression (Gutman & Sameroff 2004; Leventhal & Brooks-Gunn 2000; Xue et al. 2005). Characteristics of neighborhoods may influence the number and intensity of stressors and negative life events that individuals experience, increasing their vulnerability to depression (Cutrona, Wallace, & Wesner 2006). For example, resource-poor neighborhoods may be characterized by social and physical stressors such as crime, violence, drug sales and activity, graffiti, and vandalism—each of which has been linked with increased levels of psychological distress among residents (Aneshensel & Sucoff 1996; Latkin & Curry 2003). Similarly, fear of crime/violence and low perceived neighborhood safety have been linked with depressive symptoms, even after adjusting for known correlates of depression (Zule et al. 2008). Deteriorating physical conditions in the neighborhood, poor housing quality, noise, and crowding also have been identified as predictors of psychological distress. To the extent that these types of neighborhood disorder are experienced as uncontrollable, individuals may endorse symptoms of learned helplessness, feel hopeless, or endorse other symptoms of depression (Latkin et al. 2007). Neighborhoods with few institutional supports or networks of informal supports that can ameliorate feelings of depression may have higher rates of depression-related concerns.
Of concern are the well-documented associations between adolescents’ depressive symptoms and their involvement in high-risk behaviors. For example, depressive symptoms have been linked with adolescents’ increased delinquency (Leas & Mellor 2000), physical fighting (Pesa et al. 1997), risk for community violence exposure (Borowsky & Ireland 2004; Lambert et al. 2005), sexual-risk behaviors (Lehrer et al. 2006), and STI- and HIV-associated attitudes and sexual behaviors (DiClemente et al. 2001). Similarly, hopelessness, a significant correlate of depression, has been linked with adolescent participation in high-risk behaviors, including violence, substance use, and sexual behavior (Bolland 2006; Harris, Duncan, & Boisjoly 2002). These links between depression and health-risk behaviors suggest that depression may mediate observed associations between neighborhood risk and adolescent health-risk behaviors.
The available evidence regarding processes that account for the observed associations between neighborhood conditions and health-risk behaviors confirms that psychological distress is an important mechanism linking neighborhood characteristics and individual behavior. For example, research with adult samples has found that the effects of neighborhood disadvantage and disorder on substance use are indirect, operating through psychological distress (Boardman et al. 2001; Latkin et al. 2007). Similarly, Hill and Angel (2005) found that anxiety and depression partially mediated the association between neighborhood disadvantage and heavy drinking, providing support for tension reduction and self-medication hypotheses (Greeley & Oei 1999) that individuals use substances to reduce stress or negative affect. A similar process may operate for sexual risk-taking behaviors.
MODERATORS OF THE RELATIONSHIP BETWEEN NEIGHBORHOOD, DEPRESSION, AND HEALTH RISK
It is important to recognize that numerous factors may determine whether and the degree to which exposure to neighborhood stress is associated with adolescents’ depressive symptoms and health-risk behavior, as evidenced by resilience among youth who reside in challenging environments (Fergus & Zimmerman 2005). Moreover, conceptualizations of the tension reduction hypothesis propose that only some individuals will use substances to manage their negative affect (Hussong et al. 2001) and highlight the importance of considering aspects of the social context that may moderate the associations between stress, depression, and health-risk behaviors. Prior research has highlighted gender, age, parenting behavior, and peer affiliations as factors that may moderate the effects of the neighborhood environment on adolescent developmental outcomes.
Gender Differences
Male adolescents may have earlier and more frequent unsupervised exposure to the neighborhood environment (Leventhal & Brooks-Gunn 2000) and therefore more opportunities to engage in problem behaviors. Relatedly, males generally report more exposure to community violence than females (e.g., Buka et al. 2001) and may be similarly exposed to other neighborhood risks more often than females. Some research has shown that the neighborhood environment may have stronger effects on males than females (Leventhal & Brooks-Gunn 2000). For example, Ramirez-Valles, Zimmerman, and Juarez (2002) found that neighborhood poverty was associated with timing of first intercourse for adolescent males but not females.
Parent Supervision
Considerable research has documented the significant role of parental monitoring and supervision in curbing youth problem behaviors (e.g., Dishion & McMahon 1998). In addition, these parenting behaviors may mitigate the effects of the neighborhood environment on youth adjustment when parents limit their adolescents’ exposure to the neighborhood and neighborhood activities. For example, Browning, Leventhal, and Brooks-Gunn (2005) found that inconsistent supervision was associated with adolescent early sexual activity, particularly for adolescents living in disadvantaged neighborhoods. In contrast, parental supervision and monitoring of adolescent activities have been associated with adolescents not engaging in sexual intercourse; among sexually active adolescents, these parenting behaviors have been linked with older age at first intercourse, using protection, having fewer partners, and avoiding adolescent pregnancy (Miller, Benson, & Galbraith 2001). Still, it should be noted that the utility of parent management strategies may vary across neighborhoods (Howard et al. 2003).
Considerable research has linked adolescents’ health-risk behaviors with their friends’ deviant or risky behaviors (e.g., French & Dishion 2003; Prinstein, Boergers, & Spirito 2001). The consistent links between deviant peer affiliation and adolescent problem behaviors, such as substance use, violent offending, and early and high-risk sexual behaviors (Gifford-Smith et al. 2005), may exist because adolescents often are reinforced for behaviors that conform to peer expectations and pressures (Prinstein, Boergers, & Spirito 2001). Of relevance to substance use, adolescents may learn how to manage emotional states from their peers (Brown, Dolcini, & Leventhal 1997), and it has been proposed that self-medication may be more common in settings where substance use is reinforced as a coping strategy (Hussong et al. 2001).
The Present Study
Understanding how neighborhood and community-level risk may influence adolescent health-risk behavior is important to inform interventions to mitigate these risks and promote healthy developmental outcomes. Thus, the present study examines whether neighborhood risk is associated with substance use and sexual risk-taking behaviors, and whether these associations are accounted for, at least in part, by adolescents’ depressive symptoms. In keeping with ecological models highlighting the importance of interactions between the many contexts in which youth develop, individual, family, and peer factors are examined as possible moderators of the linkages between neighborhood risk, depressive symptoms, and health-risk behaviors.
ANALYTIC STRATEGY
Path analysis was used to examine the hypothesized relationships among the constructs. These analyses were conducted using Mplus 5.21 (Muthén & Muthén 2009), using full information maximum likelihood estimation, which allows for missing data under missing at random assumptions (Little & Rubin 1990; Rubin 1987), where participants who have data on at least one study variable are included in the analysis. Multiple indicators of fit were used to evaluate the models: chi-square, the comparative fit index (CFI), and the root mean square error of approximation (RMSEA). Hu and Bentler (1999) suggested that CFI values greater than .95 and RMSEA values less than .08 represent an acceptable fit; RMSEA values equal to or less than .05 represent a good fit (Browne & Cudek 1993).
Mediation was tested with three models according to guidelines outlined by Holmbeck (1997). First, the direct effect of neighborhood risk on the health-risk behaviors was assessed. Next, the indirect effects were assessed. Specifically, the fit of a model with paths from neighborhood risk to the mediator, depressive symptoms, and from depressive symptoms to the health-risk behaviors was tested. A third model including the indirect and direct effects was tested to determine whether the direct effect was reduced with the hypothesized mediators in the model. A reduction in the direct effect suggests mediation. To provide an additional test of mediation, the confidence interval–based test of mediation recommended by MacKinnon et al. (2002) was performed to determine the significance of the indirect effect; if the confidence interval for the indirect effect does not contain zero, the indirect effect is considered significant.
Because participants were nested within housing developments, housing development was specified as a cluster variable. Using this strategy, standard errors were adjusted to account for the nonindependence of observations within the cluster (i.e., housing development). The effect of participants’ school on the outcomes was controlled in each analysis.
To examine the hypothesized moderators, gender, parental supervision, and deviant peer affiliation, multiple group analyses were performed. For these analyses, the continuous moderators were dichotomized. A median split was used to dichotomize parental supervision and deviant peer affiliation into high and low. Each moderator was examined in a separate model. Models in which paths were freely estimated for each value of the dichotomous moderator were compared with models in which paths were constrained to be equal for the different levels of the moderator. A significant decrement in chi-square model fit for the constrained model provides evidence of significant moderation in the path that was constrained.
Descriptives
For this study, substance use was operationalized as the mean number of times participants reported they had used tobacco, alcohol, or marijuana in the past year. Approximately 40 percent of participants reported that they had used alcohol at least once, whereas approximately 30 percent and 20 percent of adolescents reported using marijuana and tobacco, respectively, at least once. Males reported using marijuana more than females (χ2(6) = 15.16, p < .05) and they reported slightly more tobacco use than females (χ2(5) = 9.43, p = .09). Adolescents’ sexual-risk behavior was based on the sum of two items about their last sexual experience: whether or not participants drank alcohol or used drugs before or during sex, and whether or not they had used a form of protection. Possible scores ranged from 0 to 2. Because only four participants reported engaging in two sexual-risk behaviors, scores were dichotomized such that a score of 1 indicated that participants had engaged in one or two sexual-risk behaviors.
Path Analysis
Path coefficients and fit statistics for path models are presented in table 7.1. The first model examining the direct effect of neighborhood risk on substance use and sexual-risk behaviors revealed a significant association between neighborhood risk and sexual-risk behaviors (standardized path coefficient = .17, p < .001). Neighborhood risk was not associated with substance use. The indirect model (model 2) revealed significant associations between neighborhood risk and depressive symptoms, as well as between depressive symptoms and each health-risk behavior; however, this model provided a poor fit to the data, as evidenced by the comparative fit index (CFI) and Tucker Lewis index (TLI) (see table 7.1). In the third model, neighborhood risk was associated with increased depressive symptoms, but depressive symptoms were not associated with sexual risk. Moreover, the direct path from neighborhood risk to sexual risk remained significant while controlling for the indirect paths. Thus, depressive symptoms did not mediate the association between neighborhood risk and sexual-risk behavior. There was a significant indirect effect of neighborhood risk on substance use (neighborhood risk → depressive symptoms → substance use indirect pathway, z = 2.43, p < .001) in the third model.
TABLE 7.1 Path Estimates and Fit Statistics for Path Analytic Models
Coefficients in bold are significant at p < .05. All coefficients are standardized. NA, not available.
To determine whether gender, parental supervision, or deviant peer affiliation moderated the indirect pathways, multiple group analyses were performed. In terms of gender differences, the association between depressive symptoms and substance use was slightly stronger for males than for females (Δχ2 = 3.54, p = .06), and the association between depressive symptoms and risky sexual behavior was slightly stronger for males than for females (Δχ2 = 3.23, p = .07). Model comparisons revealed that the association between neighborhood risk and depressive symptoms was stronger for adolescents with high paternal supervision than for adolescents with low paternal supervision (Δχ2 = 5.22, p < .05). Because of this difference, the pathway of neighborhood risk to depressive symptoms to substance use was stronger for adolescents with high paternal supervision. Contrary to expectation, the association between neighborhood risk and sexual-risk behavior was stronger for adolescents with low deviant peer affiliation than for adolescents with high deviant peer affiliation (Δχ2 = 5.71, p < .05). The association between depression and substance use was marginally stronger for adolescents with high deviant peer affiliation than for adolescents with low deviant peer affiliation (Δχ2 = 3.65, p = .06).
DISCUSSION
Ecological models highlight the significance of the neighborhood context for understanding youth health outcomes. However, only recently has empirical research examined which aspects of the neighborhood context may influence health-risk behaviors. The current research examined depressive symptoms as a possible mechanism linking adolescents’ perceptions of neighborhood disorder with sexual-risk behavior and substance use, and whether individual, family, and peer influences moderated the association between neighborhood disorder and these health-risk behaviors. Results revealed an indirect effect of neighborhood disorder, such that neighborhood risk was associated with depressive symptoms, which in turn were associated with substance use. However, neighborhood risk remained associated with sexual-risk behavior after accounting for depressive symptoms, suggesting that other mechanisms are responsible for the association between neighborhood disorder and adolescents’ sexual-risk behavior.
Neighborhood Disorder, Depressive Symptoms, and Health-Risk Behaviors
Findings of this research highlight psychological distress as one mechanism that may link neighborhood disorder with increased adolescent substance use. Although conclusions regarding causality are not appropriate given the cross-sectional data, findings suggest that neighborhood disorder and depressive symptoms may co-vary in ways that increase adolescents’ risk for substance use, consistent with prior research documenting longitudinal associations between neighborhood risk and later substance use (e.g., Lambert et al. 2004). Nonetheless, it is important to note that depressive symptoms are just one of several possible psychological and physical responses to neighborhood disorder. If neighborhood disorder is not perceived as stressful or youth with chronic exposures to neighborhood risks become desensitized, other types of adaptations may be observed. In addition, many adolescents show resilience in the face of neighborhood risk and psychological distress; this could serve to protect them from adverse outcomes.
That the association between depressive symptoms and substance was stronger for adolescents with high deviant peer affiliations was not surprising. Peers who exhibit deviant behavior may have greater access to substances, thereby increasing opportunities for adolescents to model this behavior or succumb to pressure to engage in behaviors consistent with their peers. In addition, the accepted norms and expectations for engaging in high-risk behavior may be particularly salient among such peer groups. Youth who participate in deviant or delinquent activities may do so, in part, because they have limited skills for adaptive coping; thus, they may be more likely to manage their negative affect with substances. Similarly, adolescents with fewer competences for prosocial behavior may drift toward peer contexts where behaviors such as substance use are encouraged and reinforced.
Different processes appear to link neighborhood disorder to substance use and sexual-risk behavior. While research suggests that depressed youth engage in more risky behaviors, including substance use and risky sexual behaviors, several other factors such as impulsivity and community norms for adolescent sexual behavior may better explain adolescents’ involvement in sexual risk behaviors (Donovan 2004; Kahn et al. 2002) than depressive symptoms. Alternatively, it may be that consideration of heterogeneity in how adolescent depressive symptoms are expressed and experienced is important for understanding the link between neighborhood disorder and sexual-risk behavior. For example, youth with depressive symptoms that co-occur with risk-taking tendencies may experience different outcomes than youth with more internalized depression, who may be more likely to self-medicate with substances.
Only limited support was found for individual, family, and peer moderators of the link between neighborhood disorder, depressive symptoms and health-risk behavior. Prior research also has found limited support for maternal and paternal supervision as a protective factor against the adverse effects of neighborhood and community variables (e.g., community violence exposure). It has been suggested that the protective effects of these variables are only apparent in contexts of low risk (Ceballo et al. 2003; Sullivan, Kung, & Farrell 2004). In very high-risk neighborhoods, community- and peer-level norms about acceptable and unacceptable behavior may be particularly salient for youth, making effective parental monitoring and supervision more difficult to achieve. Interestingly, results suggested that the association between depressive symptoms and the health-risk behaviors was somewhat stronger for males. Although these results should be interpreted with caution because the gender difference was only marginal (p = .06), this finding may reflect differences in norms for how males and females are expected to manage their negative affect. For example, it may be more acceptable for males to express their internal distress and negative affect via engaging in externalizing behaviors. In addition, females may have a larger repertoire of adaptive skills for managing their depressive symptoms.
Implications, Limitations, and Directions for Future Research
Our findings highlight the importance of the neighborhood context for adolescent health-risk behaviors. The direct association between neighborhood risk and sexual risk, as well as the indirect association linking neighborhood risk, depressive symptoms, and substance use, generally persisted across gender, parental supervision, and deviant peer affiliation, thus suggesting that interventions to prevent or reduce adolescent health-risk behaviors would benefit from an assessment of how adolescents experience and manage their neighborhoods. Additionally, comprehensive interventions targeting youth health-risk behaviors should consider community norms for expressing and managing feelings of distress, which include, but are not limited to, depressive symptoms. It also will be important for research and practice to identify the types of strengths and protective factors that youth and their families possess despite neighborhood risk, and how the presence of these factors may differentiate between youth who engage in health-risk behaviors from those who do not.
These results should be considered in the context of study limitations and suggest several directions for future research. The cross-sectional data examined here allowed a preliminary examination of associations between neighborhood disorder, depressive symptoms, and health-risk behaviors. However, longitudinal assessment of these constructs is necessary to determine whether neighborhood disorder predicts increases in depressive symptoms and whether depressive symptoms predict greater involvement in health-risk behaviors. Reciprocal associations also should be examined in light of some research showing that adolescent substance use and sexual behavior predicts increased depression (e.g., Hallfors et al. 2005).
In addition, future research should consider whether the moderating effects of parent supervision and deviant peer affiliation vary by gender. For example, regarding parental supervision, it has been suggested that the quantity of behavioral controls parents exert varies for male and female adolescents, with females being monitored more than males (Leventhal & Brooks-Gunn 2000) and males being granted earlier and more unsupervised time in the neighborhood (Beyers et al. 2003). This gender difference in socialization suggests that the protective effects of parental regulatory behavior may be stronger for females than for males (Browning et al. 2005). Similarly, gender differences in socialization also can affect peer affiliation and expression of distress. This may be particularly relevant for health-risk behaviors for which the social and societal consequences are perceived to vary between males and females. For example, parents may monitor their female adolescents more because the consequences of early sexual activity (e.g., pregnancy) may affect females more strongly than males (Browning et al. 2005).
Neighborhood influences on adolescent outcomes are likely moderated by age because youth exposure to the neighborhood, participation in neighborhood activities, and perceptions of the neighborhood change with increasing age (Aber et al. 1997; Leventhal & Brooks-Gunn 2000). Future research should examine this possibility, and whether younger children still experience increased supervision if parents and caregivers work out of the home for long hours. For example, in some cases, younger children under the care of older siblings might be at greater risk for exposure to neighborhood risks, depending on the nature and amount of their siblings’ involvement in unsupervised neighborhood activities.
Although this research examined parent and peer variables as possible moderators of the effect of neighborhood risk on health-risk behavior, these also may mediate the effect of neighborhood risk (e.g., Leventhal & Brooks-Gunn 2000). Thus, future research should consider both the moderating and mediating effects of parents and peers. Finally, in future research, it will be important to assess community norms about substance use and adolescent sexual activity because these interact with neighborhood disorder to predict health-risk behaviors.
