Youth Violence and Juvenile
Justice
http://yvj.sagepub.com/
Protective Factors for Youth Exposed to Violence: Role of Developmental Assets in
Building Emotional Resilience
Sonia Jain, Stephen L. Buka, S. V. Subramanian and Beth E. Molnar
Youth Violence and Juvenile Justice 2012 10: 107
DOI: 10.1177/1541204011424735
The online version of this article can be found at:
http://yvj.sagepub.com/content/10/1/107
Published by:
http://www.sagepublications.com
On behalf of:
Academy of Criminal Justice Sciences
Additional services and information for Youth Violence and Juvenile Justice can be found at:
Email Alerts: http://yvj.sagepub.com/cgi/alerts
Subscriptions: http://yvj.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://yvj.sagepub.com/content/10/1/107.refs.html
>> Version of Record - Dec 22, 2011
What is This?
Downloaded from yvj.sagepub.com by guest on October 11, 2013
Protective Factors for Youth
Exposed to Violence: Role of
Developmental Assets in
Building Emotional Resilience
Youth Violence and Juvenile Justice
10(1) 107-129
ª The Author(s) 2012
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1541204011424735
http://yvj.sagepub.com
Sonia Jain1, Stephen L. Buka2,
S. V. Subramanian3, and Beth E. Molnar3
Abstract
There is compelling evidence that many youth exposed to community violence manage to adapt
successfully over time. Developmental assets have been deemed salient for positive youth
development, though limited longitudinal studies have examined their relevance for high-risk
youth. Using the Developmental Assets framework, the authors test whether supportive relationships, high expectations, and opportunities build emotional resilience directly or indirectly
via interaction with risk. Further, the authors examine the effect of neighborhood collective
efficacy on resilience. The authors use multiwave data from 1,166 youth aged 11–16 years and
data about their neighborhoods from the Project on Human Development in Chicago Neighborhoods (PHDCN). Generalized estimating equations (GEE) were used to examine whether baseline protective factors in subjects’ home, peer, and neighborhood environments predicted log
odds of emotional resilience at Waves 2 and 3 among youth ETV. Over 7 years, 60–85% were
emotionally resilient. Positive peers and supportive relationships with parents and other adults
had significant main effects. Positive peers and family support were particularly protective for witnesses and victims. Structured activities and collective efficacy influenced change in resilience differentially among ETV groups. Strengths-based policies and systems should focus on building
developmental assets within the family, peer, and community environments for high-risk youth
who have been exposed to violence (ETV).
Keywords
resilience, developmental assets, violence prevention, protective factors, mental health
1
Health and Human Development Program, WestEd, Oakland, CA, USA
Center for Population Health and Clinical Epidemiology, Brown University, Providence, RI, USA
3
Department of Society, Human Development and Health, Harvard School of Public Health, Boston, MA, USA
2
Corresponding Author:
Sonia Jain, Health and Human Development, WestEd, 300 Lakeside Drive, 25th Floor, Oakland, CA 94612, USA
Email: sjain@wested.org
108
Youth Violence and Juvenile Justice 10(1)
Introduction
Over the last three decades, a substantial body of research has focused on the increased risk of
psychosocial problems or deviant behaviors among youth who grow up in violent communities.
Indeed, exposure to community violence (ETV) increases one’s risk of psychosocial, behavioral,
and academic problems (Kliewer, Lepore, Oskin, & Johnson, 1998; Osofsky, 1995); however, it
is not deterministic (Gorman-Smith & Tolan, 2003). In fact, the majority of youth witnesses and victims of violence subsequently develop into healthy, caring, and confident adults (Benard, 2004;
Werner & Smith, 2001). The impetus of this multidisciplinary study was that there must be developmentally appropriate protective factors that youth are able to tap into, to tip the balance from vulnerability in favor of resilience. Few longitudinal studies have empirically examined the
significance of developmental assets, that is, stage-salient protective factors that have shown to
be fundamental for positive youth development, in building resilience among at or high-risk youth
(Taylor et al., 2002; Werner, 2005).
Need for a ‘‘Resilience’’ Perspective to Combat Community Violence
Psychological health outcomes have been the most common sequelae of violence exposure
observed, that is, posttraumatic stress disorder, depression, or anxiety (Gorman-Smith & Tolan,
1998; Lynch, 2003; Overstreet & Braun, 1999). A resilience perspective suggests that being resilient
does not mean a person is invulnerable to stress, but rather that youth may bounce back, cope, and
recover constructively toward ‘‘normal’’ health in a few years (Luthar, 1993). Resilience researchers
have consistently shown that 50–70% of children under adverse conditions generally fare well
(Benard, 2004; Rhodes & Brown, 1991; Vigil, 1990; Werner & Smith, 2001). Traditional deficitfocused research tends to focus on predicting negative outcomes, revealing that approximately
30% (20–49%) of the youth do not fare well (Rutter, 1987, 2000; Werner & Smith, 2001), rather
than predicting positive outcomes among a greater percentage of youth who are ‘‘successful’’
(40–80%, Benard, 2004) within the same environment. A resilience perspective suggests a shift
in perspective of the researcher to hone in on elements of positive development that also occurs over
time among youth who may have been ETV. In fact, researchers have hypothesized that protective
factors are more predictive of positive development than risks are to negative outcomes (Rutter,
1987; Werner & Smith, 2001). And the ecologically-based protective factors are equally amenable
to intervention, as risk is, and may contribute to reducing risk exposure as well.
Selection of Protective Factors: Theoretical and Empirical Considerations
Since resilience researchers have generally been critiqued for testing a sundry list of protective factors (Luthar & Zelano, 2003), we explicitly identified the most robust factors that have been conceptually and empirically shown to be robust across disciplines for resilient and nonresilient youth.
Also, we found that few studies have explored protective factors relevant to late adolescent through
early adulthood years; hence, we also relied on developmental theory and youth development literature to guide our selection. Further, we focused on extrapolating factors at the ecological levels that
are amenable to change.
Previous studies. Resilience research within the context of community violence is in its beginning
stages. Although numerous longitudinal studies have been undertaken in the last 30 years to better
understand resilient trajectories of children exposed to chronic poverty (Garmezy, 1985), parental
psychopathology (Rutter, 1985; Werner & Smith, 1992), and child abuse and neglect (Garbarino,
Dubrow, Kostelny, & Pardo, 1992), little has been done to document resilience in the face of community violence—despite its high prevalence, persistence, and magnitude.
Jain et al.
109
Recently, several investigators have examined protective factors relevant for emotional and
behavioral health in the face of community violence (Bowen & Chapman, 1996; Gorman-Smith,
Henry, & Tolan, 2004; Hammack, Richards, Luo, Edlynn, & Roy, 2004; Kliewer et al., 2004; Lynch
& Cicchetti, 1998). The majority have focused on different dimensions of family structure and functioning, finding that parental support (Bowen & Chapman, 1996; Kliewer et al., 1998; Kuther &
Fisher, 1998), family cohesion (Gorman-Smith & Tolan, 1998; 2003), parental attachment (Lynch
& Cicchetti, 1998), or simply presence of a parent (Fitzpatrick & Boldizar, 1993; Overstreet &
Braun, 1999) are protective against adverse outcomes—suggesting that resilience-based interventions should focus on improving quality of parent–child relationships and other family-level factors.
However, growing evidence suggests that parents may not be able to compensate for the negative
effects of ETV beyond a certain threshold level of risk (Hammack et al., 2004; Kliewer et al., 2004;
Luthar & Goldstein, 2004; Sullivan, Kung, & Farrell, 2004), partly because family functioning may
also be compromised in most dangerous neighborhoods or those exposed to high rates of community
violence over time (Krenichyn, Saegert, & Evans, 2001; Osofsky, 1995; Richters & Martinez,
1993). Others have suggested that parent–adolescent relationships are generally at their worst during teen years and that families become of lesser importance as children develop into adolescents
(O’Donnell, Schwab-Stone, & Muyeed, 2002). This suggests that other external sources of support
and resources such as in schools, peer groups, and neighborhoods in parallel deserve greater consideration. However, scarce evidence exists documenting the salience of community, schools, and
peers to mitigate the effects of violence (Schwartz & Proctor, 2000).
Studies that have examined protective factors for ETV have (a) lacked a clear conceptual framework to guide their selection of protective factors. For example, taking an ecological perspective, a
focus solely on relationships would miss resources and opportunities in the schools and/or communities that may be equally relevant to overcome obstacles for at-risk youth; (b) focused largely on
earlier development years absent consideration of expanding exposures postadolescence; (c) relied
on cross-sectional or short longitudinal data, not measuring long-term impacts of protective factors;
and (d) have not accounted for neighborhood-level differences in crime to explicitly account for the
individual risk of ETV.
The Developmental Assets and Ecological Framework
This strengths-based study is guided by several interdisciplinary individual and ecological level
frameworks. Many researchers concur that to fully examine the issues related to youth ETV, an
ecological–transactional framework is required (Cicchetti & Lynch, 1993; Dawes & Donald,
2000), which places the developing child within the dynamic distal context of their families, communities, and societies at large. The Developmental Assets framework (Lerner, Taylor, & von
Eye, 2002; Benson, Leffert, Scales, & Blyth, 1998; Leffert et al., 1998) from the youth development literature offers a promising conceptual model for the study of resilience. Going beyond the
prevention of high-risk behaviors and into enhancement of resilience, assets reflect core developmental processes operating at multiple levels (Scales & Leffert, 1999). Search Institute highlights
four external developmental assets including supportive relationships, empowerment, boundaries
and expectations, and constructive use of time. They suggest that by means of positive experiences
that meaningful opportunities and relationships with adults offer, reinforced by systems and policies, has tremendous benefits to protect youth from high-risk behaviors, and enhance positive
developmental outcomes. The developmental assets framework is in sync with the ecological–
transactional framework and has tremendous potential to complement, strengthen, and expand
existing resilience research and practice.
Moreover, the growing youth development and positive psychology movement contends that
there are external factors fundamental for positive development of all adolescents into adulthood,
110
Youth Violence and Juvenile Justice 10(1)
including for at-risk youth (Luthar & Zelano, 2003; Masten, 2001). Focused on youth as assets who
develop within the context of their families, school, and communities simultaneously, the youth
development perspective (Lerner & Galambos, 1998; Pittman, Wilson-Ahlstrom, & Yohalem,
2003) offers great insight for the study of resilience. At-risk youth themselves have voiced the
importance of having positive forces such as educational and job opportunities, connection with
adults, and meaningful uses of their time as key for countering ‘‘the draw of the streets’’ (Ginburg,
Alexander, Hunt, Sullivan, & Cnaan, 2002), and not just the absence of risk factors in their lives
(Lerner, Taylor, & von Eye, 2002; Ungar, 2004).
Protective Factors and Resilience
Literature documenting the salience of developmental assets for an array of developmental outcomes is slowly accumulating (Benard, 1991, 2004; Lerner, Taylor, & von Eye, 2002; Benson
et al., 1998). Presence of one caring adult whether in the community, home, or school (Luthar &
Zelano, 2003; Werner & Smith, 2001) structured opportunities to participate in meaningful activities
that provide leadership, sense of responsibility, and decision making, as well as high expectations
from parents or other adults have shown to improve mental health for all youth (Benard, 2004; Larson, 2000; Resnick et al., 1997).
However, the relevance of developmental assets for high-risk youth has been rarely tested. Only
one exploratory study to our knowledge (Taylor et al., 2002) has found a positive association
between the number of assets and competencies among gang members. Considering only the numbers of assets (0–10, 11–20) however, and not specific association of each asset to an outcome
(Price, Dake, & Ruthie, 2001), undermines the importance of the few assets available to the highest
risk individuals. Moreover, asset studies have not methodologically accounted for the context of violence or other risks that assets inevitably interact with. Low reliability and validity of the Search
Institute assets instrument among ethnically diverse inner-city youth has also hindered such investigations (Price et al., 2001; Price, Spence, Sheffield, & Donovan, 2002).
Simply having a set of protective factors does not ensure resilience over time particularly among
youth disproportionately exposed to violence (Mazza & Overstreet, 2000). Youth exposed to violence may display average or better-than-expected functioning as a result of variation in ‘‘actual’’
risk exposure (Luthar, Cicchetti, & Becker, 2000), differences in individual characteristics including
genetic variability, or differential exposure to assets in the environment. Benson (2002) found that
youth under adversity who had higher numbers of assets were 7 times less likely to have high-risk
behavior (33%) compared to those with an average number of assets (5%). They found that 40 assets
explained 47–54% of the variance in thriving for all youth, over and above demographics; however,
among youth with one or more risk factors, only 10% of ‘‘thriving’’ was explained.
Neighborhood-Level Collective Efficacy and Resilience
In a review, Wandersman and Nation (1998) noted that, ‘‘research associating resilience to neighborhood factors is sparse.’’ Growing evidence suggests that neighborhoods matter for adolescent
development, though most have focused on examining negative effects of living in poor neighborhoods (Leventhal & Brooks-Gunn, 2000). Few studies have explored how communities may come
together to build resilience (Garbarino, 1995). If neighborhoods, via institutional and social conditions, have the power to affect development negatively, similarly, they may have the power to
influence positive development (Connell & Aber, 1995). This is in sync with others who recognize
the importance of positive social processes within disordered neighborhoods such as collective
efficacy (Jain, Buka, Subramanian, & Molnar, 2010; Molnar, Cerda, Roberts, & Buka, 2008;
Molnar, Miller, Azrael, & Buka, 2004; Sampson, Morenoff, & Gannon-Rowley, 2002; Sampson,
Jain et al.
111
Raudenbush, & Earls, 1997) and social networks (Garbarino et al., 1992) in preventing peer or
parent-to-child violence.
The Present Study
The present strengths-based study enhances prior knowledge across disciplines by testing the
relevance of theoretically and empirically based developmental assets for high-risk youth into early
adulthood, controlling for individual and neighborhood-level risks. This study aims to (a) understand
the main effects of protective factors on emotional resilience longitudinally, controlling for individual and neighborhood-level covariates; (b) determine whether protective factors moderate the
association between exposure to violence and emotional resilience; and (c) examine whether
neighborhood-level collective efficacy is associated with emotional resilience and whether it modifies the effect of assets on building resilience among youth exposed to violence.
Method
Study Design and the Sample
Data for this study come from the Project on Human Development in Chicago Neighborhoods
(PHDCN), a community-based multilevel longitudinal study conducted in 1994–2002 of adolescents, their caregivers, and their neighborhoods. Sampling began by defining 343 neighborhood
clusters (NCs) based on aggregated census tracts, representing every dwelling unit within the city.
NCs were geographically sensible and homogenous in terms of race/ethnicity, socioeconomic status,
family structure, and housing density. In 1994–1995, a community survey was conducted in which
an independent sample of 8,872 residents (>18 years) were randomly sampled from the 343 NCs
with a 75% response rate (PHDCN, 1998). A random sample of 6,226 children and youth within
6 months of ages 0 (in utero), 3, 6, 9, 12, 15, and 18 were also selected from a random sample of
80 NCs at baseline using a multistage probability design, for the Longitudinal Cohort Study (LCS).
About 25 youth per NC were interviewed three times. A detailed description of the sampling procedures used in the PHDCN has been reported elsewhere (Earls & Buka, 1997). For the present
study, the community survey, 1990 U.S. Census, and the LCS served as the primary data sources.
The final sample included subjects with nonmissing data at baseline, and assuming data were
missing at random, the longitudinal models estimated values for the missing responses in subsequent
waves. Values were imputed only if one wave was missing data. If more than one wave was missing data, then no imputation was done. Of the total 1,517 youth who participated in Cohorts 12 and
15 at Wave 1, 1,238 had complete data on ETV at Wave 2, 47 had missing data on outcome at
either Wave 2 or 3, and 25 were missing data on at least one covariate. Thus, the final sample
included 1,166 youth in 78 neighborhoods for analysis. Subjects dropped from the analysis
(n ¼ 351) were more likely to be Black, from single-parent families, and have fewer assets (family
boundaries, collective efficacy, other adult support) but similar internalizing scores and ETV compared to the other respondents.
Measures
Primary dependent variable. Using a reduced 28-item versions of the Youth Self Report and Young
Adult Self Report scales (Achenbach, 1991), a continuous internalizing problem score (0–53) was
calculated. The scale included 15 items on anxiety/depressive symptoms that captured feelings of
loneliness, worthlessness, unhappiness, or whether the subject cries or worries a lot; 9 items on
somatic symptoms on feelings of dizziness/being overtired or having physical problems such as
headache or nausea; and 4 items on withdrawal symptoms that captured whether the subject rather
112
Youth Violence and Juvenile Justice 10(1)
be alone, is shy or secretive, or refuses to talk. If fewer than 6 items were missing, then 0, 1, 2
responses were averaged to create the three subscales that were then summed to create an internalizing scale. Coefficient as ranged from .86 to .89 at each wave.
For the purposes of this study, we categorized the number of youth who met the criteria for
emotional resilience versus not, as adequate positive adaptation in the context of risk. We considered
positive adaptation to be better than or average mental health functioning but not exceptional
functioning, since we are considering exposure to significant adversity, that is, witness or victim
to violence (Luthar & Zelano, 2003). Hence, the internalizing scale score was dichotomized into
1 ¼ resilient youth with scores 0.50 standard deviation below the sample median and 0 ¼ nonresilient youth with scores 0.50 standard deviation above the median. We used gender-specific medians
for all youth as the cutoffs versus T-scores to better classify emotional ‘‘well-being’’ as typically
found in a nonclinical setting (Tedeschi & Kilmer, 2005).
Primary independent variables. Exposure to community violence. Subject’s exposure to 18 different
violent events in the community in the past year was measured using the My ETV scale (Buka,
Selner-O’Hagan, Kindlon, & Earls, 1997; Kindlon, Wright, Raudenbush, & Earls, 1996;
Selner-O’Hagan, Buka, Kindlon, Raudenbush, & Earls, 1998) at Wave 2. This did not account for
any violence the adolescent might be experiencing at home. Three subscales of (a) witnessing (7
items; a ¼ .74), (b) victimization (7 items; a ¼ .57), and (c) heard of (3 items, a ¼ .37) were developed, as a sum of yes/no responses (Brennan, Molnar, & Earls, 2007). The psychometric properties
of these scales have been tested in diverse populations using item-response theory and Rasch modeling (Brennan et al., 2007; Selner-O’Hagan et al., 1998). Based on the continuous scales, a categorical ETV group variable was also created to allow group-specific comparisons with 0 ¼ nonexposed
group who scored 0 on witnessing and victimization scale but may have heard of violence (1 or more
acts); a ¼ witness group had witnessed at least one act of violence in the past year, b ¼ victim group
that had been a victim of at least one act of violence and had witnessed one act or not. The size of the
‘‘heard of’’ group was too small (n < 20) to stratify separately so they were grouped with the unexposed youth. Both continuous and categorical variables were tested, to account for the frequency and
severity of violence (Buka, Stitchick, Birdthistle, & Earls, 2001).
Protective factors. Items from the PHDCN that corroborated with Search Institute external assets
(Benson & Leffert, 1999), that is, support, opportunities, boundaries and expectations, and empowerment, and the California Healthy Kids Survey Resilience module (WestEd) were identified at all
waves. Since only parts of reliable scales were available at Waves 2 and 3, baseline data were used.
Under support, family support (6 items; a ¼ .73), friend support (8 items; a ¼ .71), and other adult
support (4 items, a ¼ .53) emerged from the Provision of Social Relations instrument (Turner,
Frankel, & Levin, 1983) per factor analysis and item deletion reliability tests. If more than half the
items were not missing, then an average score was calculated based on very/somewhat/not true
responses. Note, previously validated and reliable scales were used as much as possible and placed
within the developmental assets theoretical domains. For measures where new scales were developed,
extensive exploratory and confirmatory factor analysis, item deletion reliability tests, other psychometric analyses, and theory were used (unpublished to date). Under expectations and boundaries, positive peer influence (10 items from Deviance of Peers; Huizinga, Esbenson, & Weihar, 1991; a ¼ .62)
captured whether friends model responsible behavior, for example, the number involved in sports/
community/religious/family/after-school activities, considered good students or good citizens. Family
boundaries and expectations scale (13 items from Home; Caldwell & Bradley, 1984; a ¼ .63) captured items on parental monitoring, and having clear rules and consequences at home. Under opportunities, sum of time spent in structured activities per week in school or after-school was calculated
based on 2 items from the school questionnaire (Furstenburg, 1990). All scales were individually
Jain et al.
113
standardized to have a mean of zero and standard deviation of one. Only significant interactions are
shown in the results.
Neighborhood-level predictors. Neighborhood social cohesion. (Sampson et al., 1997; a ¼ .80) was a
sum of 5 items from the community survey (strongly disagree to strongly agree) about residents’
willingness to help, trust each other, get along, share the same values, and perceive the community
as close-knit. Neighborhood social control (5 items from community survey; a ¼ .80; Sampson
et al., 1997) captured perception of neighborhood boundaries, that is, neighbors will intervene if
children are skipping school, hanging out on a street corner, or spray-painting graffiti. Collective
efficacy (Sampson et al., 1997) was a sum of these two subscales based on aggregated independent
resident responses from the community survey. Internal consistency of the scale was high, with
Cronbach’s coefficient a of .89; higher scores representing greater collective efficacy in a neighborhood. Organizations and services index included 8 items on the presence of various local organizations and programs such as parks, block group, neighborhood watch group, mental health
center, and 6 items on youth services such as recreational programs, after-school programs, mentoring/counseling services.
Neighborhood-level confounders. Concentrated poverty was calculated using the first principal component of three U.S. Census items: percentage of persons unemployed, receiving public assistance,
and living below the federal poverty line in 1990. Perceived violence in the community was a sum of
5 items on the community survey assessing how often the respondent had witnessed a robbery or
mugging, a fight among neighbors, a fight with weapon, sexual assault or rape, or a gang fight in
the last 6 months. The responses ranged from 1 ¼ often to 4 ¼ never; higher score representing
greater perceived violence at baseline.
Individual-level confounders. Sociodemographics of youth included age (centered at the mean), gender (female ¼ reference group), family socioeconomic position (composite of parental income, education and occupational code; maximum of either parent was used), family structure (two biological
parents ¼ reference, biological/one nonbiological, one biological, and other/ two nonbiological),
and race/ethnicity (White, Asian/PI, and other race was the reference group, vs. Black and Hispanic
groups). Continuous measures at individual and neighborhood levels were grand-mean centered for
ease of interpretation. For missing responses to these covariates, the mean value was imputed and a
variable indicating imputation was added to all models.
Data Analysis
All analyses were done using SAS version 8.0 (SAS Institute, 1999). First, among the final sample of
1,166 youth, differences in protective factors, individual, and neighborhood level characteristics
were examined by ETV group. Chi-square tests and t tests were used to assess whether differences
in categorical and continuous covariates between groups were significant. Bivariate Pearson correlations were examined to assess the magnitude and significance of the correlations between the primary outcome, risk of ETV, and protective factors. Systematic differences between respondents and
nonrespondents were also examined.
Next, Generalized Estimating Equations (GEE) with a logit function were estimated regressing
intercept at Wave 2 and change in log odds of emotional resilience between Waves 2 and 3 onto individual and neighborhood-level predictors at baseline (Bryk & Raudenbush, 1987; Liang & Zeger,
1986; Subramanian, Jones, & Duncan, 2003). Unstructured within-subject correlations of binary
response between Waves 2 and 3 were modeled, partly to account for the temporal association
between predictors and outcome, and to adjust for clustering. GEE was the preferred method of analysis as this technique provides a statistically robust model that adequately accounts for variation in the
114
Youth Violence and Juvenile Justice 10(1)
outcome that exists at multiple levels, as well as adjustments for expected autocorrelation across time
(within-subjects) and space (between subjects within neighborhoods; Fitzmaurice, Laird, & James,
2004). GEE further (a) accommodates missing data at various time points, (b) does not assume comparable growth across all subjects, (c) allows for inconsistent timing of data collection, and (d) in comparison to nlmixed, estimates group-specific parameters, not subject-specific parameters in relation to
prototypical neighborhoods (Hanley, Negassa, Edwardes, & Forrester, 2003; SAS Institute, 1999;
Wolfinger & Chang, 1998). First, a person-period data set was created in which each person had three
records, one for each wave (Singer & Willett, 2003). Next, data were structured with 2,332 repeated
observations at Level 1 nested within 1,166 individuals at Level 2, nested within 78 neighborhoods at
Level 3. Depending on the covariates included in the model, the analytic sample varied.
To test specific hypotheses, multilevel models were sequentially built starting with the null
model with no predictors, adding time (age), primary risk variable (ETV group), Level 2 controls
(sex, race, socio-economic position [SEP], family structure), and Level 3 controls (neighborhood
perceived violence and concentrated poverty). Victimization (continuous) was also kept in the
models since it changed the coefficient of ETV group significantly, and remained significant (p
< .05) even after inclusion of assets. To test for the main effects of assets on the log odds of emotional resilience at Wave 2 (intercept; Aim 1), individual assets were added to the fully conditional
model; for the main effect of an asset on the rate of change between Waves 2 and 3 (slope), a twoway interaction asset age was included in the above model. To test for the interactive effect of
each asset with ETV (Aim 2), a two-way interaction term between an asset and ETV was included
in the intercept model above, and a three-way interaction term was included in the slope model
above to assess the differential effects of assets on slopes by ETV groups. Interactive slope models
also controlled for age ETV term but not age sex as slope did not vary by sex (p > .05). Note
ETV was kept as two dummy variables with witness versus not and victim versus not to allow for
comparisons across groups; and separate models were run for each asset. Age was centered at
Wave 2 thus the intercept terms in the models estimated the log odds of emotional resilience at
Wave 2 associated with a one standard deviation increase in the asset, controlling for covariates.
Slope terms estimated the change in the log odds of emotional resilience between Waves 2 and 3
for each SD increase in an asset, controlling for covariates. Finally, the main and interactive
effects of neighborhood-level collective efficacy and organizational services, separately with ETV
and each asset, on emotional resilience were tested.
Results
Sample Characteristics
Table 1 presents the individual and neighborhood characteristics at baseline of 1,166 youth in 78
Chicago neighborhoods, stratified by ETV group, that is, unexposed (n ¼ 255; 22%), witnesses
(n ¼ 519, 45%), and victims (n ¼ 392, 34%). The average age of subjects in all three groups was
13.5 years (11–16) at Wave 1, 15.5 years (12–20) at Wave 2, and 18.1 years (15–22) at Wave 3.
Blacks were overrepresented in the witness and victim groups, compared to the unexposed (37% and
41% vs. 21%, p < .05). Victims were more likely to be male (55%) and living in single family households (33%) compared to witnesses and unexposed. In terms of neighborhood characteristics, witnesses and victims lived in neighborhoods of higher mean concentrated poverty compared to the
unexposed group (p < .05). In terms of the distribution of protective factors, the unexposed group
had significantly higher levels of family support and positive peers compared to the other two
groups; and victims reported significantly lower positive peer influence, family boundaries, and
friend support than other ETV groups. Hours in structured activities, other adult support, neighborhood cohesion, and neighborhood control were similar across all ETV groups (p > .05).
Jain et al.
115
Table 1. Selected Sample Characteristics by Exposure to Violence Groupa, N ¼ 1,166 Youth in 78 Neighborhoods, PHDCN Cohorts 12 and 15.
Individual-Level Covariates
Age at baseline (range: 11–16)
Socioeconomic positionb (3.0, 3.5)
Sex
Male
Female
Race
Black
Hispanic
White and othersc
Family structure
Two biological parents
One biol—one nonbiol
One biological parent
Two nonbiological parent
Protective factors at baseline
Support
Family support (4.7, 0.9)
Friend support (3.8, 1.3)
Other adult support (2.7, 1.3)
Neighborhood cohesion (2.2, 2.8)
Opportunities
Hrs structured activities (0.9, 12.1)
Expectations and boundaries
Positive peer influence (3.6, 3.3)
Family boundaries (5.3, 0.9)
Neighborhood control (2.6, 2.2)
Neighborhood-level at baseline
Collective efficacy (1.9, 2.6)
Organizational services (.05, 0.5)
Concentrated povertyb (1.1–2.7)
Perceived violence (1.3–2.9)
Unexposed,
n ¼ 255
Witness group,
n ¼ 519
Victim group,
n ¼ 392
M (SD)
13.2 (1.4)
0.07 (1.4)
Percent (N)
40.4% (103)2
59.6% (1532)2
M (SD)
13.5 (1.5)
0.18 (1.4)
Percent (N)
46.8% (243)2
53.2% (276)2
M (SD)
13.7 (1.5)
0.06 (1.5)
Percent (N)
55.2% (217)
44.8% (175)
20.8% (53)1,2
52.5% (134)
27.7% (68)1,2
37.0% (192)
46.6% (242)
16.4% (85)
41.1% (161)
44.1% (173)
14.8% (58)
56.9% (145)1,2
17.7% (45)2
21.6% (55)1,2
3.9% (10)1
46.4% (241)2
17.2% (89)2
28.9% (150)
7.5% (39)
38.8% (152)
22.4% (88)
33.2% (130)
5.6% (22)
M (SD)
0.20 (0.84)1,2
0.04 (1.05)
0.03 (1.01)
0.00 (1.03)
M (SD)
0.00 (0.97)
0.04 (0.96)
0.04 (0.97)
0.06 (0.95)
M (SD)
0.04 (1.03)
0.02 (0.97)
0.03 (0.98)
0.03 (1.03)
0.01 (1.01)
0.06 (0.88)
0.13 (1.09)2
0.15 (0.91)2
0.02 (1.06)
0.02 (0.96)
0.05 (0.94)
0.01 (0.95)
0.06 (1.01)
0.05 (1.00)
0.00 (1.01)
0.01 (1.06)
0.17 (0.10)
0.17 (0.72)1,2
0.01 (0.37)
0.04 (0.94)
0.16 (0.09)
0.06 (0.76)
0.01 (0.34)
0.01 (1.03)
0.17 (0.10)
0.03 (0.78)
0.01 (0.35)
0.00 (1.04)
Note. PHDCN ¼ Project on Human Development in Chicago Neighborhoods; SD ¼ standard deviation.
a
Sample size is based on complete data for Cohorts 12 and 15 at Wave 2 for ETV, all covariates and nonmissings for both
Waves 2 and 3 outcome. The witness group includes youth who had witnessed at least one act of violence in the past year
(¼1), victim group (¼1) includes youth who had been a victim of at least 1 act of violence and had witnessed one act or not.
The unexposed group had witnessed or been a victim of no act of violence in the past year.
b
Socioeconomic status is based on principal component of parental income, education, and occupation. Neighborhood concentrated poverty is principal component of % poverty, % unemployed, and % on public assistance.
c
Other race includes Asian, Pacific Islanders, and Native Americans.
1
p < .05 versus witness group. 2p < .05 versus victim group.
The percentage of youth who met the criteria for emotional resilience varied by the level of risk
(ETV), ranging from 60% to 85% at any time point (see Table 2). Across time, 59% of all youth were
resilient at both waves, 11% nonresilient at either wave, and 30% crossed-over, that is, nonresilient
became resilient or resilient became nonresilient. Victims were least likely to be emotionally resilient
at both waves (50%), followed by witnesses (62%). Emotional resilience dropped among the unexposed
over time, whereas it increased among the witnesses and victims emulating the unexposed by Wave 3.
116
Youth Violence and Juvenile Justice 10(1)
Table 2. Percent Youth Who Are Emotionally Resilienta Across Waves by Exposure to Violence
Single Time Point
No ETV (n ¼ 264)
Witness group
(n ¼ 525)
Victim group (n ¼ 406)
All youth (1,195)
Resilience Over Time
Wave
2 (%)
Wave
3 (%)
Resilient
at Both (%)
Nonresilient
(%)
R
NR (%)
NR
R (%)
851,2
752
781,2
772
691,2
622
62
92
162
142
91,2
152
60
72
74
76
50
59
17
11
10
13
23
17
Note. ETV ¼ exposed to community violence; R ¼ resilience; NR ¼ nonresilient.
a
Percent emotionally resilient is based on (a) risk and (b) adapted or not; Resilient youth (¼1) are those with internalizing
score less than .50 standard deviation of the sample median, nonresilient (¼0) have internalizing scores greater than .50 standard deviation. Cutoffs are based on gender-specific medians for total sample at each wave.
1
p < .05 versus witness group.
2
p < .05 versus victim group.
Multivariate correlations (see Table 3) revealed that internalizing scores at Waves 2 and 3 were
significantly correlated (p < .05) with risk, that is, witnessing, victimization, as well as with several
protective factors, that is, family support, friend support, other adult, positive peer influence, and
family boundaries. Of all the protective factors, positive peers, family support, and family boundaries had significant correlations with both the outcome and the ETV variables. Collective efficacy
was positively correlated with hours in activities, friend support, positive peers, and organizational
services, but not with internalizing scores.
Multilevel Models
Developmental Assets and Emotional Resilience
Main effects of exposure to violence. Next, Tables 4 and 5 display the final conditional models of
GEEs, showing the association between assets at baseline and the odds of emotional resilience at
Wave 2 (intercept) and over time from Waves 2 to 3 (slope), controlling for individual and
neighborhood-level confounders. Note, only the fixed effects are shown, as random effects are not
estimated per marginal linear models. The unexposed group had the highest odds of emotional resilience at Wave 2, that is, 3.10 (95% CI [2.25, 4.26]), compared to 2.25 (95% CI [1.22, 4.15]) for
witnesses, and 1.64 (95% CI [0.79, 3.41]) for victims, conditional on individual characteristics and
perceived violence in the neighborhood and poverty. Hence, witnesses (OR ¼ 0.70, 95% CI [0.54,
0.97]) and victims (OR ¼ 0.55, 95% CI [0.35, 0.80]) had 30% and 45% lower odds of being emotionally resilient compared to the unexposed group (p < .05). Males had higher odds of emotional
resilience, though sex did not interact with any of the assets; suggesting that all youth regardless
of gender benefited similarly from access to assets.
Main effects of assets on emotional resilience at Wave 2. As shown in Table 4, four developmental assets had positive main effects on odds of emotional resilience, that is, they were protective for
all youth regardless of violence exposure. Friend support, family support, other adult support, and
positive peers increased the odds of emotional resilience significantly (p < .01), above and beyond
individual and neighborhood-level confounders. For instance, an increase of 1 SD in positive
peer influence increased the odds of emotional resilience by 22% (OR ¼ 1.22, 95% CI [1.10,
1.36], p < .001) for the reference group. The intercept term remained significant even after inclusion of all assets, thus additional factors not considered in the study are likely contributing to emotional resilience.
117
1.00
0.08*
0.13***
0.24***
0.08*
0.10**
0.03
0.16***
0.07*
0.04
0.05
0.03
0.03
1.00
0.55***
0.13***
0.19***
0.26***
0.05y
0.07*
0.04
0.12***
0.08*
0.00
0.08*
0.02
0.07*
2
0.12***
0.01
0.02
0.02
0.12***
0.17***
0.01
0.05y
0.02
0.06y
1.00
0.47***
3
0.07*
0.02
0.02
0.04
0.08**
0.07*
0.01
0.03
0.02
0.01
1.00
4
1.00
0.28***
0.34***
0.01
0.28***
0.12***
0.02
0.00
0.02
0.01
5
1.00
0.21***
0.06y
0.25***
0.06y
0.08**
0.02
0.07*
0.00
6
1.00
0.02
0.25***
0.14***
0.03
0.05y
0.01
0.01
7
1.00
0.08**
0.00
0.77***
0.11***
0.93***
0.25***
8
1.00
0.12***
0.08**
0.11***
0.08**
0.07*
9
1.00
0.01
0.06*
0.01
0.05y
10
1.00
0.09**
0.95***
0.25***
11
1.00
0.11***
0.12***
12
1.00
0.26***
13
Note. PHDCN ¼ Project on Human Development in Chicago Neighborhoods.
All measures are continuous so Pearson correlations were used to test significance. Internalizing scores are measured at Waves 2 and 3, ETV at Wave 2, and protective factors at baseline. All
protective factors are standardized to a mean of 0 and standard deviation of 1.
yp < .10. *p < .05. **p < .01. ***p < .00.
Resilient outcomes
Internalizing score w2
Internalizing score w3
Risk
Witnessing
Victimization
Protective factors
Family support
Friend support
Other adult support
Neighborhood cohesion
Positive peers
Family boundaries
NC boundaries
Hours in activities
Collective efficacy
Organizational services
1
Table 3. Multivariate Correlations Among Study Variables, PHDCN Cohorts 12 and 15
118
1.18 [0.97, 1.44]y
1.66 [1.29, 2.13]***
1.27 [1.00, 1.63]y
ns
1.44 [1.15, 1.79]**
1.1 [0.90, 1.37]
ns
1.33 [1.01, 1.75]*
ns
ns
1.22 [1.10, 1.36]***
0.98 [0.86, 1.12]
1.09 [0.98, 1.22]
1.07 [0.94, 1.21]
1.01 [0.87, 1.16]
0.88 [0.28, 2.77]
3.10 [2.25, 4.26]
[1.15, 1.41]***
[1.23, 1.52]***
[1.02, 1.27]***
[0.90, 1.19]
1.28
1.37
1.14
1.03
0.72 [0.59, 0.89]**
—
ns
ns
ns
1.21 [1.02, 1.43]*
1.00 [0.83, 1.21]
ns
1.31 [1.08, 1.12]***
1.39 [1.18, 1.64]***
1.15 [0.98, 1.36]y
ns
2.25 [1.22, 4.15]*
ns
ns
ns
1.14 [0.96, 1.35]
0.91 [0.76, 1.10]
ns
1.30 [1.10, 1.55]**
1.27 [1.09, 1.48]**
ns
ns
1.64 [0.79, 3.41]**
Victims
ns
ns
ns
Victim: OR ¼ 0.79 [0.60, 1.04]y
ns
ns
ns
victim: OR ¼ 0.77 [0.57, 1.03]y
ns
ns
Witness OR ¼ 0.73 [0.54, 0.97]*
Victim: OR ¼ 0.53 [0.35, 0.80]**
Model Cd
Interaction of
Asset and ETV
Difference Between
Groups
Note. OR = odds ratio; CI ¼ confidence intervals; ETV ¼ exposed to community violence; GEE ¼ generalized estimating equations; ns ¼ nonsignificant coefficient; PHDCN ¼ Project on Human
Development in Chicago Neighborhoods; SD ¼ standard deviation.
a
All assets are continuous measures at baseline standardized to a mean of 0 and standard deviation of 1. Dependent variable is the log odds of emotional resilience (proportion of youth with internalizing score within 0.50 SD above the sample median (¼0) vs. ones above the 0.50 cutoff ¼ 0). Log odds coefficients were converted to odds ratio and 95% confidence intervals by taking natural log
of each coefficient.
b
The odds ratio shown is an estimate of the odds of emotional resilience associated with 1 SD increase in the asset for the unexposed group, controlling for covariates. Model A controls for sex,
race/ ethnicity, family socioeconomic position, family structure, age (centered at Wave 2), ETV group (0, 1, 2), and frequency of victimization.
c
The odds ratio shown is an estimate of the odds of emotional resilience associated with 1 SD increase in the asset within that group, controlling for covariates. Model B includes an interaction term
between asset and categorical ETV with dummies witness or not; victim or not, without main effect of asset. The significant p values reflect that odds ratio changed significantly with the addition of
asset for that group, compared to the odds ratio at base.
d
Model C includes the main effect of asset in the models, thereby providing estimates of whether the difference in the interactions is significant compared to the unexposed group.
yp < .10. *p < .05. **p < .01. ***p < .001.
Supportive relationships
Friend support
Family support
Other adult support
Neighborhood cohesion
Boundaries and expectations
Positive peers
Neighborhood control
Family boundaries
Opportunities for meaningful participation
Hours in activities
Collective efficacy
Organizational services
Base model
ETV
ETV with dummies
Main Effect of Asset
on Each Group
Witnesses
Main Effect of Asset
on Each Group
þ ETV/Controls
Unexposed
Model Bc
Model Ab
Table 4. GEE Modelsa Predicting Emotional Resilience at Wave 2 by Baseline Developmental Assetsa: Main and Interactive Effects PHDCN, 1,166 Youth Nested in 78 Chicago
Neighborhoods (OR and 95% CI)
Jain et al.
119
Interactive effects of assets with exposure to violence. Positive peers and family support had
borderline interactive effects (p < .10) with being a victim (see Table 4). More positive peers resulted
in 21% lower odds of emotional resilience for victims (OR ¼ 0.79, 95% CI [0.60, 1.04]) compared to
the unexposed. Family support was also marginally less protective for victims compared to the unexposed group (OR ¼ 0.77, 95% CI [0.57, 1.03]). Notably though, when we look at significance of
assets within each ETV group: for witnesses for instance, each unit increase in positive peers
(OR ¼ 1.21, 95% CI [1.02, 1.43]) and other adult support (OR ¼ 1.15, 95% CI [0.98, 1.36]) significantly increased the odds of resilience. The nonsignificant interaction terms imply that these supports were equally likely to increase resilience for victims and witnesses compared to the unexposed
group. Friend support was more beneficial for witnesses and victims, resulting in 30% higher odds of
resilience with each unit increase in asset, compared to an 18% increase among the unexposed.
Main and interactive effect on rate of change in resilience. Besides examining the effect of assets on
likelihood of resilience at a single time point, we further assessed whether assets predicted the rate of
change in emotional resilience across time, as displayed in Table 5. Conditional on all covariates,
emotional resilience increased marginally by Wave 3 in the base model (OR ¼ 1.06, 95% CI
[0.92, 1.13]). By Wave 3, emotional resilience decreased by 28% within the unexposed and 1% for
witnesses yet increased by 27% among the victims. Thus, emotional resilience changed differentially by ETV group (p < .05), that is, compared to the unexposed group, emotional resilience
increased significantly for witnesses (OR ¼ 1.21, 95% CI [1.01, 1.46]) and victims (OR ¼ 1.55,
95% CI [1.29, 1.86]).
Greater friend support, positive peers, and hours in structured activities alter the slope marginally
differentially for ETV groups. For instance, friend support at baseline inversely affected the slope
for witnesses versus unexposed (OR ¼ 0.84, 95% CI [0.73, 0.99]), that is, each unit increase in friend
support increased resilience 7% among the unexposed; yet, decreased the rate of resilience among
witnesses by 8%.
Within the witness and victim groups, the level of family support, neighborhood cohesion, and
control also changed the odds of resilience from Waves 2 to 3, similar to unexposed groups. Thus
victims with greater family support had less of an increase in resilience from Waves 2 to 3 (OR ¼
.91, 95% CI [0.83, 0.99]), compared to victims with no family support. Similarly, for witnesses, each
unit increase in baseline neighborhood-level protective factors was associated with 15% lower emotional resilience by Wave 3 suggesting that lower ETV group may benefit more initially from higher
neighborhood cohesion and control, but the protective effects do not last.
Collective Efficacy and Emotional Resilience
Main and interactive effects. Neighborhood collective efficacy at baseline did not influence the
odds of emotional resilience at Waves 2 or 3, above and beyond inclusion of all individual and
neighborhood-level confounders including ETV and individual-level assets (Table 4). Collective
efficacy however was a significant predictor of the rate of change in emotional resilience for witnesses; the effect of collective efficacy on the slope was robust and significant even after inclusion
of individual assets and ETV group; though the effect on rate of change was not significantly different by ETV groups (Table 5). The decrease in emotional resilience was greater for the unexposed
(OR ¼ 0.95 (95% CI [0.81, 1.11]) and witnesses (OR ¼ 0.86 (95% CI [0.77, 0.95]) who had higher
levels of collective efficacy at baseline, whereas the increase in resilience among victims (OR ¼
1.03 (95% CI [0.93, 1.12]) did not vary by collective efficacy.
None of the cross-level interactions between individual-level assets and collective efficacy were
significant; in fact, the more proximal assets (those with main effects originally), remained significant in most cases above and beyond inclusion of collective efficacy and the neighborhood-level
confounders. Organizations and services had no main or interactive effects.
120
1.06 [0.92, 1.13]y
0.82 [0.70, 0.95]**
0.97 [0.88, 1.07]
0.91 [0.83, 0.99]*
ns
1.02 [0.93, 1.11]
0.99 [0.90, 1.08]*
1.02 [0.93, 1.12]
ns
0.93 [0.84, 1.04]
1.03 [0.93, 1.12]
ns
0.96 [0.86, 1.08]
0.89 [0.80, 0.98]*
ns
0.93 [0.84, 1.03]
0.86 [0.77, 0.95]**
ns
1.27 [1.15, 1.40]***
Victims
0.91 [0.82, 1.01]y
1.05 [0.96, 1.15]
ns
0.85 [0.76, 0.95]**
0.99 [0.90, 0.91]
Witnesses
Witness and victim: OR ¼ 0.81 [0.64, 1.03]y
ns
ns
Witness OR ¼ 0.84 [0.70, 1.02]y
ns
ns
Witness OR ¼ 0.84 [0.73, 0.99]*
ns
ns
ns
Witness: OR ¼ 1.21 [1.01, 1.46]*
Victim: OR ¼ 1.55 [1.29, 1.86]**
Model Cc
Interaction of Asset
and ETV Difference
Between Groups
Note. OR = odds ratio; CI ¼ confidence intervals; ETV ¼ exposed to community violence; GEE ¼ generalized estimating equations; ns ¼ nonsignificant coefficient; PHDCN ¼ Project on Human
Development in Chicago Neighborhoods; SD ¼ standard deviation.
a
All assets are continuous measures at baseline standardized to a mean of 0 and standard deviation of 1.
b
The odds ratio shown is an estimate of the rate of change in emotional resilience associated with 1 SD increase in the protective factor within that group, controlling for covariates. All models
include protective factor age, ETV age, and main protective factor ETV interaction terms using dummy ETV variable. Model B includes three-way interaction term between protective
factor age ETV, without the main effect of asset. Thus, the significant p values reflect that slope changed significantly with the addition of asset for that group, compared to the rate at base.
c
Model C includes the main effect of asset in the models, thereby providing estimates of whether the difference in the interactions is significant compared to the unexposed group.
d
Base model controls for sex, race/ethnicity, family socioeconomic position, family structure, age, ETV group (0, 1, 2), and frequency of victimization.
yp < .10. *p < .05. **p < .01. ***p < .001.
Supportive relationships
Friend support
1.07 [0.951.21]
Family support
1.01 [0.84, 1.21]
Other adult support
1.05 [0.91, 1.21]
Neighborhood cohesion
0.93 [0.80, 1.08]y
Boundaries and expectations
Positive peers
1.14 [0.98, 1.32]y
Neighborhood control
0.96 [0.83, 1.12]
Family boundaries
1.09 [0.98, 1.22]
Opportunities for meaningful participation
Hours in activities
1.14 [0.92, 1.42]
Collective efficacy
0.95 [0.81, 1.11]y
Organizational services
0.88 [0.28, 2.77]
Base modeld
Age
ETV Age
Unexposed
Model Bb
Main Effect of Asset
on Each Group
Table 5. GEE Modelsa Predicting Change in Emotional Resilience Between Waves 2 and 3 by Baseline Developmental Assetsa: Main and Interactive Effects PHDCN, 1,166
Youth Nested in 78 Chicago Neighborhoods (OR and 95% CI)
Jain et al.
121
Discussion
This longitudinal, strengths-based study explored whether multilevel protective factors build
emotional resilience among an ethnically diverse sample of at-risk youth. Specifically, we examined
whether developmental assets deemed salient for all adolescents, were protective for adolescents
exposed to violence above and beyond individual and neighborhood confounders. Reliable and valid
measures of assets tested among an ethnically diverse sample were used.
Supportive Relationships
Indeed, the role of supportive relationships in the positive development of children is well documented for health and mental health (Wight, Botticello, & Aneshenel, 2006) though for youth exposed to
violence, evidence is limited. While family support was protective against Post-traumatic Stress Disorder (PTSD) in one study, (Ozer & Weinstein, 2004), the impact of support by friends was not in
another and needs to be further examined (O’Donnell et al., 2002). In our sample, supportive relationships were particularly strong predictors of emotional resilience for all youth including witnesses and
victims across time, beyond individual and neighborhood confounders (Gorman-Smith, Tolan, &
Henry, 2000; Hammack et al., 2004; Kliewer et al., 2004; Krenichyn et al., 2001; O’Donnell et al., 2002).
Family support remained highly protective for all ETV groups at Wave 2 though less so for victims, as has been suggested (Kliewer et al., 2004). It remained equally protective over time for the
lowest and highest ETV groups. Among victims for instance, family support seemed to stabilize
emotional resilience, not enhance it, as has been suggested previously (Hammack et al., 2004; Ozer
& Weinstein, 2004). This implies that support from family has strong and stable effects on emotional
well-being even for victims, as has been documented for victims of physical abuse (Lansford et al.,
2006). This contradicts with others who have found limited protective effects of family cohesion or
caring relationships with a parent on internalizing symptoms (Kliewer et al., 2004). Although we
accounted for both family structure and function and controlled for perceived violence in the neighborhood, it is possible that family and friends functioning is also compromised due to exposure to
community violence (Lynch & Cicchetti, 2002). How violence impacts family functioning and peer
interactions needs to be better understood.
As suggested by Bronfenbrenner (1979), the family and peer microsystems serve as highly influential factors for youth development. And per Cauce, Felner, and Primavera (1982) and O’Donnell,
Schwab-Stone, and Muyeed (2002), the impact of support varies by the source of social support—so
we looked at friend versus family versus other adult support separately. In our study, friend support
was more protective for witnesses and victims than the unexposed group initially; though, positive
effects of friends did not last across time for witnesses or victims (Schwartz & Proctor, 2000). In
fact, having friend support at baseline, and becoming a witness to violence, seemed to have a negative effect on the mental health of the witness group. Interestingly, having ‘‘other’’ adult support
and neighborhood support/cohesion at baseline did not influence emotional resilience in later years.
Opportunities for Meaningful Participation
Numerous investigators have noted the benefits of participating in meaningful activities such as
sports, drama, arts on mental health and related outcomes (Bell & Suggs, 1998; McNeal, 1998);
however, few have explored the benefits for at-risk youth longitudinally. We found that hours spent
in structured activities during early adolescent years had a significant effect on building emotional
resilience among the unexposed group only. Structured activities did not buffer the effects of being a
witness or victim at Wave 2; but they did influence the rate of change in development of resilience
differentially by ETV groups. Thus, victims who spent more hours per week in school-based or
122
Youth Violence and Juvenile Justice 10(1)
after-school activities at baseline had a slower increase in emotional resilience after Wave 2 and
unexposed had a slower decrease. Since the unexposed group had significantly higher odds of
resilience at Wave 2, some tendency toward the ‘‘mean’’ is expected. However, it also suggests that
having participated in structured activities at baseline does not buffer the effects of subsequently
becoming a witness or a victim to violence. Future studies should consider the benefits of participation in activities after exposure to violence, accounting for the decrease in opportunities for 18- to
24-year-olds (Pittman et al., 2003), and cover a broader range of outcomes.
Boundaries and Expectations
Research shows that high expectations and boundaries set by parents, teachers, and peers can have
both positive and negative effects on the child (Crosnoe, 2000; Leffert et al., 1998). In this study,
we found that having positive peers increased the odds of emotional resilience for all youth,
beyond individual and neighborhood risks. This finding is consistent with others that have shown
the importance of having positive friends particularly during adolescent years (Crosnoe, 2000;
Leffert et al., 1998).
The effects of positive peers on both emotional resilience at Wave 2 or across time varied by the
ETV group such that victims with positive peers benefited less at baseline than the unexposed group.
In terms of the rate of change, resilience dropped further for witnesses with each unit increase in
positive peers, whereas the unexposed group had slower decline longitudinally. O’Donnell et al.
(2002) also found that peer support surprisingly increased the odds of depression among witnesses.
Indeed, it is possible that having positive peers who are good students, good citizens, and honest
individuals, could have a negative psychological effect on youth who become witnesses to violence
as they might feel relatively more anxiety and distress while trying to meet the expectations of
‘‘well-functioning’’ peers, especially in the context of a school and community which they now perceive to be violent. Another explanation may be that deviant friends (who may also have been
exposed to violence) may provide more positive emotional support for at-risk youth.
We also found that victims benefited the most from having assets in their lives, for example,
friend support. Given that youth with highest ETV (and generally other risks) also have the lowest
average number of assets (Lerner, Taylor, & von Eye, 2002), it would be particularly useful to build
external assets in families and peers for the highest risk youth as O’Donnell (2002) notes that ‘‘they
not only need these services the most but also will benefit the most from them.’’
There may be a number of mechanisms by which assets might translate into emotional resilience
within the context of community violence (Rutter, 1987, 1995). For instance, assets might (a) reduce
actual exposure to violence, for example, participating in sports would prevent ETV during afterschool hours; (b) reduce the impact or trauma associated with violence or enhance coping, for example, talking to caring adults may provide a venue to disclose distress; (c) reduce the negative chain of
events from ETV, for example, youth who have witnessed violence may not associate with deviant
peers; or (d) enhance self-esteem via positive peer relations or access to meaningful opportunities,
due to involvement in extracurricular activities or volunteer work.
Neighborhood-Level Collective Efficacy
Finally, neighborhood-level cohesion and control, individually or as a composite of collective
efficacy (Sampson et al., 1997), did not influence emotional resilience at any single time point but
increased resilience over time, especially for victims. Other longitudinal studies have found
that neighborhood cohesion or quality does not protect against the effects of violence to influence
adjustment among at-risk youth (Furstenburg & Hughes, 1995; Kliewer et al., 2004). Though
slightly protective (2–8%) for all groups at Wave 2, by Wave 3 odds of emotional resilience dropped
Jain et al.
123
significantly among witnesses and unexposed living in a cohesive community; whereas victims in a
cohesive community continued to have increased odds of emotional resilience. This suggests that
living in a caring community improves emotional well-being of a lower risk group initially, whereas
trust and cohesion in a community protects victims against increase in vulnerability or embodiment
of stress over time (Luthar et al., 2000). Given that limited factors have been found to be protective
for the highest risk individuals, this study suggests that community capacity building efforts may
be particularly useful interventions in sustaining emotional resilience for victims (Wolkow &
Ferguson, 2001). This study furthers the robustness of the collective efficacy construct toward
building resilience, an unexplored area of research deserving greater attention (Wandersman &
Nation, 1998). Future neighborhood inquiries on resilience would benefit from capturing changes
in collective efficacy at subsequent time points, and assessing positive developmental trajectories
over a longer life span.
Moreover, despite widespread recognition that direct neighborhood effects are rather small and
that they largely operate through proximal forces, little is known about the specific ways by which
neighborhoods influence resilience. Many communities throughout the United States, in an
attempt to combat violence, are turning to organizing residents and building cohesion, yet how
a distal community process translates into resilience among its youth is largely unknown. Understanding how collective efficacy works through proximal forces would further the salience of this
powerful construct.
Our study findings suggest that as youth in urban neighborhoods negotiate healthy
development, within the context of violence and other risks, there are factors in the expanding
social spheres such as support from family, friends, or other adults, having positive peers, neighborhood control, cohesion, and time spent in structured activities that may buffer the effects of
violence and subsequently build emotional resilience. Peers and communities may become more
important as youth age past adolescence. Family support also carries potential to protect against
the effects of witnessing greater acts of violence longitudinally. Identifying interactive processes
by which schools, families, and community together influence positive development also
deserves greater attention.
Limitations and Strengths
This study has several limitations. First, protective factors were measured at baseline only, thus they
do not account for changes in the neighborhoods, friends, or family over the length of the observation period. Future studies should consider additive effects of protective factors over time, and
measuring all theoretically relevant assets if possible. Also, school-level factors were not considered partly due to limited availability of data. Next, ETV was measured comprehensively at Wave
2 only, limiting our ability to measure the benefits of protective factors after a youth was exposed
to violence. An ideal study design would capture time-varying ETV and exclude abuse in the family. Third, all data regarding individuals, peers, and families were based on youth self-reports that
may be subject to recall bias or social desirability (neighborhood measures came from an independent sample of adults). Future studies may consider triangulating the measures of protective factors with secondary sources of data from schools, and/or parental, peer, and teacher assessments.
Next, the results are limited to urban youth in one city, thus may not be generalizable nationally or
to suburban or rural areas.
The study’s strengths include its multilevel design that allowed an ecological–developmental theory analysis. The use of longitudinal data allowed for accounting of some of the temporal ambiguity
between exposure and outcomes. In most cases, standardized, conceptually relevant measures were
used. Covariates were used at each level, including neighborhood-level risk.
124
Youth Violence and Juvenile Justice 10(1)
Implications for Practice and Research
This prospective study identified multiple environmental factors associated with positive development for at-risk youth. Future studies should continue to combine knowledge from multiple disciplines to better conceptualize and test multidimensional competence particularly among victims,
using a broader array of protective factors at multiple levels. Researchers should follow youth into adulthood, accounting for the dynamic changes in risk, protection, and resilient functioning over a longer life
span. Modeling trajectories or person-focused analysis will possibly capture the process of recovery and
adaptation from violence (Crockett, Moilanen, Raffaelli, & Randall, 2006; Masten, 2001; Obradovic,
Burt, & Masten, 2006). Use of rigorous qualitative and quantitative methods to operationalize and measure positive stage-salient outcomes should be employed, and analyses should be stratified by race and
gender to better represent population-specific exposures and outcomes.
In sum, researchers, practitioners, and society as a whole need to acknowledge the achievements
and successes attained by most youth in urban neighborhoods who despite overwhelming adversities
manage to develop into ‘‘caring, confident, and contributing adults’’ (Werner & Smith, 2001). There
are indeed events, characteristics, and environments that can protect youth from harm and guide
them toward positive development. In addition to prevention of underlying root causes of violence,
public health interventions should focus on building assets in the schools, families, and communities
in urban neighborhoods.
Authors’ Note
S. Jain conceptualized and designed the study, led the data analysis, interpretation, and wrote the
manuscript. Buka and Subramanian provided input on the study methodology, analytic plan and
results, and reviewed earlier drafts of the manuscript. Molnar supervised all aspects of the study
design, analysis, interpretation, and reviewed manuscript drafts. Human Participant Protection: The
study was reviewed and approved by the institutional review board of the Harvard School of Public
Health. A parent or guardian provided written consent before each assessment, and each young person assented as well.
Acknowledgments
The authors are especially thankful to all the families, youth, administrators, and those involved in
PHDCN for participating in, and providing access to the data set.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/
or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this
article: This research was supported by the Maternal and Child Health Training grant
2T76MC00001-51 and the Harvard Injury Control Research Center grant R49/CCR115279-04.
This project was also supported by Award No. 2009-IJ-CX-0103 by the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. The opinions, findings, and conclusions or recommendations expressed in this publication are those of the authors(s) and do not
necessarily reflect those of the Department of Justice. SV Subramanian was supported by the
National Institutes of Health Career Development Award (NHLBI 1 K25 HL081275). Funding for
the Project on Human Development in Chicago Neighborhoods (PHDCN) was provided by the
John D. and Catherine T. MacArthur Foundation, the National Institute of Mental Health, and the
National Institute of Justice.
Jain et al.
125
References
Achenbach, T. M. (1991). Manual for the youth self-report and 1991 profile. Burlington, VT: Department of
Psychiatry, University of Vermont.
Bell, C. C., & Suggs, H. (1998). Using sports to strengthen resiliency in childrenTraining ‘‘heart.’’ Child and
Adolescent Psychiatric Clinics of North America, 7, 859–865.
Benard, B. (1991). Fostering resiliency in kids: Protective factors in the family, school, and community.
Portland, OR: Western Center for Drug-Free Schools and Communities.
Benard, B. (Ed.). (2004). Resiliency: What we have learned. San Francisco, CA: WestEd.
Benson, P. L. (2002). Search institute chapters. In New directions for youth development. Wiley Periodicals.
Benson, P. L., & Leffert, N. (1999). Developmental assets: A synthesis of the scientific research on development. Minneapolis, MN: Search Institute.
Benson, P. L., Leffert, N., Scales, P., & Blyth, D. (1998). Beyond the ‘‘village’’ rhetoric: Creating healthy communities for children and adolescents. Applied Developmental Science, 2, 138–159.
Bowen, G. L., & Chapman, M. V. (1996). Poverty, neighborhood danger, social support, and the individual
adaptation among at-risk youth in urban areas. Journal of Family Issues, 17, 641–666.
Brennan, R. T., Molnar, B. E., & Earls, F. (2007). Refining the measurement of exposure to violence (ETV) in
urban youth. Journal of Community Psychology, 35, 603–618.
Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA: Harvard University Press.
Bryk, A. S., & Raudenbush, S. W. (1987). Application of hierarchical linear models to assessing change. Psychological Bulletin, 101, 147–158.
Buka, S. L., Selner-O’Hagan, M. B., Kindlon, D. J., & Earls, F. J. (1997). The ‘‘My exposure to violence interview’’ administration and scoring manual, Version 3. Boston, MA: Harvard School of Public Health.
Buka, S. L., Stitchick, T. L., Birdthistle, I., & Earls, F. J. (2001). Youth exposure to violence; prevalence, risks
and consequences. American Journal of Orthopsychiatry, 71, 298–310.
Caldwell, B., & Bradley, R. (1984). Home observation for measurement of the environment (HOME)Revised
edition. Little Rock: University of Arkansas.
Cauce, A. M., Felner, R. D., & Primavera, J. (1982). Social support in high-risk adolescents: Structural
components and adaptive impact. American Journal of Community Psychology, 10, 417–428.
Cicchetti, D., & Lynch, M. (1993). Toward an ecological/ transactional model of community violence and
child maltreatment: Consequences for children’s development. Psychiatry: Interpersonal and Biological
Processes, 56, 96–118.
Connell, J. P., & Aber, J. L. (1995). How do urban communities affect youth? Using social science research to
inform the design and evaluation of comprehensive community initiatives. In J. P. Connell, A. C. Kubish, L.
B. Schorr, & C. H. Weiss (Eds.), New approaches to evaluating community initiatives: Concepts, methods,
and context (pp. 93–125). Washington, DC: Aspen Institute.
Crockett, L. J., Moilanen, K. L., Raffaelli, M., & Randall, B. A. (2006). Psychological profiles and adolescent
adjustment: A person-centered approach. Development and Psychopathology, 18, 195–214.
Crosnoe, R. (2000). Friendships in childhood and adolescence: The life course and new directions. Social
Psychology Quarterly, 63, 377–391.
Dawes, A., & Donald, D. (2000). Improving children’s chances: Developmental theory and effective interventions in community contexts. In D. Donald, A. Dawes, & J. Louw (Eds.), Addressing childhood adversity
(pp. 1–25). Cape Town, South Africa: David Philip.
Earls, F., & Buka, S. L. (Eds.). (1997). Project on human development in Chicago neighborhoods: Technical
report. Rockville, MD: National Institute of Justice.
Fitzmaurice, G. M., Laird, N. M., & James, H. W. (2004). Applied longitudinal analysis. Hoboken, NJ: John
Wiley.
Fitzpatrick, K. M., & Boldizar, J. P. (1993). The prevalence and consequences of exposure to violence among
African American youth. Journal of the American Academic of Child Adolescent Psychiatry, 32, 424–430.
126
Youth Violence and Juvenile Justice 10(1)
Furstenburg, F. (1990). Philadelphia family management study. Youth Interview Schedule. Philadelphia, PA:
University of Pennsylvania, Population Studies Center.
Furstenburg, F. F., & Hughes, M. E. (1995). Social capital and successful development among at-risk youth.
Journal of Marriage and the Family, 57, 580–592.
Garbarino, J. (1995). Raising Children in a Socially Toxic Environment. San Francisco: Jossey-Bass Publishers.
Garbarino, J., Dubrow, N., Kostelny, K., & Pardo, C. (Ed.). (1992). Children in danger. San Francisco, CA:
Jossey-Bass.
Garmezy, N. (1985). Stress-resistant children: the search for protective factors. In Davids A. Elmsford, recent
research in Developmental psychopathology (pp. 213–33). NY: Pergamon Press.
Ginburg, K. R., Alexander, P. M, Hunt, J., Sullivan, M., & Cnaan, A. (2002). Enhancing their likelihood for a
positive future: Focus groups reveal the voice of inner-city youth. Pediatrics. 109, e95.
Gorman-Smith, D., Henry, D. B., & Tolan, P. H. (2004). Exposure to community violence and violence perpetration:
The protective effects of family functioning. Journal of Clinical Child and Adolescent Psychology, 33, 439–449.
Gorman-Smith, D., & Tolan, P. (1998). The prevalence and consequences of exposure to violence among
African-American youth. Development and Psychopathology, 10, 101–116.
Gorman-Smith, D., & Tolan, P. H. (2003). Positive adaptation among youth exposed to community violence. In
S. S. Luthar (Ed.), Resilience and vulnerability: Adaptation in the context of childhood adversities (pp. 392–
413). New York, NY: Cambridge University Press.
Gorman-Smith, D., Tolan, P. H., & Henry, D. B. (2000). A developmental-ecological model of the relation of
family functioning to patterns of delinquency. Journal of Quantitative Criminology, 16, 169–198.
Hammack, P. L., Richards, M. H., Luo, Z. P., Edlynn, E. S., & Roy, K. (2004). Social support factors as moderators of community violence exposure among inner-city African American young adolescents. Journal of
Clinical Child and Adolescent Psychology, 33, 450–462.
Hanley, J. A., Negassa, A., Edwardes, M. D., & Forrester, J. E. (2003). Statistical analysis of correlated data
using generalized estimating equations: An orientation. American Journal of Epidemiology, 157, 364–375.
Huizinga, D., Esbenson, F., & Weihar, A. W. (1991). Are there multiple paths to delinquency? The Journal of
Criminal Law and Criminology, 82, 83–118.
Jain, S., Buka, S. L., Subramanian, S. V., & Molnar, B. E. (2010). Neighborhood predictors of dating violence
victimization and perpetration in young adulthood: A multi-level study. American Journal of Public Health,
100(9), 1737–1744.
Kindlon, D., Wright, B. D., Raudenbush, S., & Earls, F. (1996). The measurement of children’s exposure to
violence: A Rasch analysis. International Journal of Methods in Psychiatric Research, 6, 187–194.
Kliewer, W., Cunningham, J. N., Diehl, R., Walker, J. M., Atiyeh, C., Neace, B., . . . Mejia, R. (2004). Violence
exposure and adjustment in inner-city youth: Child and caregiver emotion regulation skills, caregiver-child
relationship quality, and neighborhood cohesion as protective factors. Journal of Clinical Child and Adolescent Psychology, 33, 477–487.
Kliewer, W., Lepore, S., Oskin, D., & Johnson, P. (1998). The role of social and cognitive processes in children’s adjustment to community violence. Journal of Consulting and Clinical Psychology, 66, 199–209.
Krenichyn, K., Saegert, S., & Evans, G. W. (2001). Parents as moderators of psychological and physiological
correlates of inner-city children’s exposure to violence. Applied Developmental Psychology, 22, 581–602.
Kuther, T. L., & Fisher, C. B. (1998).Victimization by community violence in young adolescents from a suburban city. Journal of Early Adolescence, 18, 53–76.
Lansford, J. E., Malone, P. S., Stevens, K. I., Dodge, K. A., Bates, J. E., & Pettit, G. S. (2006). Developmental
trajectories of externalizing and internalizing behaviors: Factors underlying resilience in physically abused
children. Development and Psychopathology, 18, 35–55.
Larson, R. W. (2000). Toward a psychology of positive youth development. American Psychologist, 55, 170–183.
Leffert, N., Benson, P. L., Scales, P. C., Sharma, A. R., Drake, D. R., & Blyth, D. A. (1998). Developmental
assets: Measurement and prediction of risk behaviors among adolescents. Applied Developmental Science,
2, 209–230.
Jain et al.
127
Lerner, R. M., & Galambos, N. L. (1998). Adolescent development: Challenges and opportunities for research,
programs, and policies. Annual Review of Psychology, 49, 413–446.
Lerner, R. M., Taylor, C. S., & von Eye, A. (Eds.). (2002). Pathways to positive development among diverse
youth. New directions for youth development: Theory, practice, and research. San Francisco: Jossey-Bass.
Leventhal, T., & Brooks-Gunn, J. (2000). The neighborhoods they live in: The effects of neighborhood
residence on child and adolescent outcomes. Psychological Bulletin, 126, 309–337.
Liang, K. Y., & Zeger, S. L. (1986). Longitudinal data analysis using generalized linear models. Biometrika, 73,
13–22.
Luthar, S. S. (1993). Annotation: Methodological and conceptual issues in the study of resilience. Journal of
Child Psychology and Psychiatry, 34, 441–453.
Luthar, S. S., Cicchetti, D., & Becker, B. (2000). The construct of resilience: A critical evaluation and
guidelines for future work. Child Development, 71, 543–562.
Luthar, S. S., & Goldstein, A. (2004). Children’s exposure to community violence: Implications for understanding risk and resilience. Journal of Clinical Child and Adolescent Psychology, 33, 499–505.
Luthar, S. S., & Zelano, L. B. (2003). Research on resilience: An integrative review. In S. S. Luthar (Ed.),
Resilience and vulnerability: Adaptation in the context of childhood adversities (pp. 510–549). New York,
NY: Cambridge University Press.
Lynch, M. (2003). Consequences of community violence. Clinical Child and Family Psychology Review, 6,
265–274.
Lynch, M., & Cicchetti, D. (1998). An ecological-transactional analysis of children and contexts: The
longitudinal interplay among child maltreatment, community violence, and children’s symptomatology.
Development and Psychopathology, 10, 235–257.
Lynch, M., & Cicchetti, D. (2002). Links between community violence and the family system: Evidence from
children’s feelings of relatedness and perceptions of parent behavior. Family Process, 41, 519–532.
Masten, A. S. (2001). Ordinary magic: Resilience processes in development. American Psychologist, 56,
227–238.
Mazza, J. J., & Overstreet, S. (2000). Children and adolescents exposed to community violence: A mental
health perspective for school psychologists. School Psychology Review, 29, 86–101.
McNeal, R. (1998). High school extracurricular activities: Closed structures and stratifying patterns of
participation. Journal of Educational Research, 91, 183–191.
Molnar, B. E., Cerda, M., Roberts, A. L., & Buka, S. L. (2008). Effects of neighborhood resources on aggressive
and delinquent behaviors among urban youths. American Journal of Public Health, 98, 1086–1093.
Molnar, B. E., Miller, M., Azrael, D., & Buka, S. L. (2004). Neighborhood predictors of concealed firearm
carrying: Results from the project on human development in Chicago neighborhoods. Archives of Pediatrics
& Adolescent Medicine, 158, 657–664.
Obradovic, J., Burt, K. B., & Masten, A. S. (2006). Pathways of adaptation from adolescence to young
adulthood. Annals of New York Academy of Sciences, 1094, 340–344.
O’Donnell, D. A., Schwab-Stone, M. E., & Muyeed, A. Z. (2002). Multidimensional resilience in urban
children exposed to community violence. Child Development, 73, 1265–1282.
Osofsky, J. D. (1995). The effects of violence exposure on young children. American Psychologist, 50, 782–788.
Overstreet, S., & Braun, S. (1999). A preliminary examination of the relationship between exposure to community violence and academic functioning. School Psychology Quarterly, 14, 380–396.
Ozer, E. J., & Weinstein, R. S. (2004). Urban adolescents’ exposure to community violence: The role of support, school safety, and social constraints in a school-based sample of boys and girls. Journal of Clinical
Child and Adolescent Psychology, 33, 463–376.
Pittman, K., Wilson-Ahlstrom, A., & Yohalem, N. (2003). Out-of-school-time policy commentary #4: afterschool for all? Exploring access and equity in after-school programs. Washington, DC: The Forum for
Youth Investment, Impact Strategies, Inc. Retrieved from http://www.forumforyouthinvestment.org/files/
OSTPC4.pdf
128
Youth Violence and Juvenile Justice 10(1)
Price, C. S., Spence, S. H., Sheffield, J., & Donovan, C. (2002). The development and psychometric properties
of a measure of social and adaptive functioning for children and adolescents. Journal of Clinical Child and
Adolescent Psychology, 31, 111–122.
Price, J. H., Dake, J. A., & Ruthie, K. (2001). Assets as predictors of suicide attempts in African American
inner-city youths. American Journal of Health Behavior, 25, 367–375.
Project no Human Development in Chicago Neighborhoods (1998). Developmental Epidemiology Research
Unit, Harvard School of Public Health. Wave One Manual.
Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris, K. M., Jones, J., . . . Udry, J. R. (1997).
Protecting adolescents from harm: Findings from the National Longitudinal Study on Adolescent Health.
Journal of the American Medical Association, 278, 823–832.
Rhodes, W. A., & Brown, W. K. (1991). Why some children succeed despite the odds. New York, NY: Praeger.
Richters, J. E., & Martinez, P. E. (1993). Violent communities, family choices, and children’s chances: An
algorithm for improving the odds. Development & Psychopathology, 5, 609–627.
Rutter, M. (1985). Resilience in the face of adversity: Protective factors and resistance to psychiatric disorder.
British Journal of Psychiatry, 147, 598–611.
Rutter, M. (1987). Psychosocial resilience and protective mechanisms. American Journal of Orthopsychiatry,
57, 316–331.
Rutter, M. (1995). Psychosocial adversity: Risk, resilience and recovery. Southern African Journal of Child and
Adolescent Psychiatry, 7, 75–88.
Rutter, M. (2000). Resilience reconsidered: Conceptual considerations, empirical findings, and policy
implications. In J. P. Shonkoff & S. J. Meisels (Eds.), Handbook of early childhood intervention (Vol. 2,
pp. 651–682). New York, NY: Cambridge University Press.
Sampson, R. J., Morenoff, J. D., & Gannon-Rowley, T. (2002). Assessing ‘neighborhood effects’’: Social
processes and new directions in research. Annual Review of Sociology, 28, 443–478.
Sampson, R. J., Raudenbush, S. W., & Earls, F. (1997). Neighborhoods and violent crime: A multilevel study of
collective efficacy. Science, 277, 918–924.
SAS Institute. (1999). SAS user’s guide, version 8. Cary, NC: Author.
Scales, P. C., & Leffert, N. (1999). Developmental assets: A synthesis of the scientific research on adolescent
development. Minneapolis, MN: Search Institute.
Schwartz, D., & Proctor, L. J. (2000). Community violence exposure and children’s adjustment in the school
peer group: The mediating roles of emotion regulation and social cognition. Journal of Consulting and
Clinical Psychology, 68, 670–683.
Selner-O’Hagan, M., Buka, S., Kindlon, D., Raudenbush, S., & Earls, F. (1998). Urban youth exposure to
violence. Journal of Child Psychology & Psychiatry & Allied Disciplines, 39, 215–224.
Singer, J. D., & Willett, J. B. (Eds.). (2003). Applied longitudinal data analysis: Modeling change and event
occurrence. New York, NY: Oxford University Press.
Subramanian, S. V., Jones, K., & Duncan, C. (2003). Multilevel methods for public health research. In I.
Kawachi & L. F. Berkman (Eds.), Neighborhoods and health (pp. 65–111). New York, NY: Oxford University Press.
Sullivan, T. N., Kung, E. M., & Farrell, A. D. (2004). Relation between witnessing violence and drug use initiation among rural adolescents: Parental monitoring and family support as protective factors. Journal of
Clinical Child and Adolescent Psychology, 33, 488–498.
Taylor, C. S., Lerner, R. M., Von Eye, A., Balsano, A. B., Dowling, E. M., Anderson, P. M., . . . Bjelobrk, D.
(2002). Individual and ecological assets and positive developmental trajectories among gang and
community-based organization youth. New Directions for Youth Development: Pathways to Positive Youth
Development, 95, 57–72.
Tedeschi, R. G., & Kilmer, R. P. (2005). Assessing strengths, resilience and growth to guide clinical interventions. Professional Psychology: Research and Practice, 36, 230–237.
Jain et al.
129
Turner, R., Frankel, B., & Levin, D. (1983). Social support: Conceptualization, measurement, and implications
for mental health. In J. R. Greeley (Ed.), Research in community and mental health (Vol. III, pp. 67–111).
Greenwich, Greater London: JAI Press.
Ungar, M. A. (2004). Constructionist discourse on resilience: Multiple contexts, multiple realities among at-risk
children and youth. Youth & Society, 35, 341–365.
Vigil, J. D. (1990). Cholos and gangs: Culture change and street youth in Los Angeles. In R. Huff (Ed.), Gangs
in America: Diffusion, diversity, and public policy (pp. 142–162). Thousand Oaks, CA: SAGE.
Wandersman, A., & Nation, M. (1998). Urban neighborhoods and mental health: Psychological contributions to
understanding toxicity, resilience, and interventions. American Psychologist, 43, 647–656.
Werner, E. E. (2005). What can we learn about resilience from large-scale longitudinal studies? In S. Goldstein,
& R. B. Brooks (Eds.), Handbook of resilience in children. New York, NY: Springer.
Werner, E. E., & Smith, R. S. (1992). Overcoming the odds: High risk children from birth to adulthood. New
York, NY: Cornell University Press.
Werner, E. E., & Smith, R. S. (2001). Journeys from childhood to midlife: Risk, resilience and recovery. Ithaca,
NY: Cornell University Press.
WestEd, Health and Human Development. California healthy kids survey resilience & youth development
supplement. Retrieved from http://www.wested.org/chks/pdf/h11Full_rydm_0809.pdf
Wight, R. G., Botticello, A. L., & Aneshensel, C. S. (2006). Socioeconomic context, social support, and
adolescent mental health: A multilevel investigation. Journal of Youth and Adolescence, 35, 115–126.
Wolfinger, R., & Chang, M. (1998). Comparing the SAS GLM and MIXED procedures for repeated measures.
Cary, NC: SAS Institute.
Wolkow, K. E., & Ferguson, H. B. (2001). Community factors in the development of resiliency: Considerations
and future directions. Community Mental Health Journal, 37, 489–498.
Bios
Sonia Jain is a senior researcher at WestEd Health and Human Development Program. A social developmental
epidemiologist, Jain conducts research, evaluation and planning in the areas of youth development and resilience, the link between health and education, violence, mental health, and better understanding social determinants of youth health inequities. At the time of this research, Dr. Jain was a doctoral student at the Harvard
School of Public Health, Department of Society, Human Development and Health.
Stephen L. Buka is an epidemiologist and developmental psychologist whose work focuses on the causes,
development and prevention of major psychiatric and cognitive disorders. Current studies include investigations of prenatal risks for schizophrenia, attention deficit disorder and addictive disorders, including neuroimaging and molecular genetics techniques; and community-level influences on youth substance use and
delinquency. Dr Buka is currently a professor and epidemiology section head at the Department of Community
Health at Brown University.
S.V. Subramanian, (‘Subu’ or ‘Subra’) is a professor of population health and geography in the Department of
Society, Human Development and Health at the Harvard School of Public Health. He is also a faculty associate
at the Institute of Quantitative Social Sciences at Harvard University and is a member of the Steering Committee for the Harvard Center for Population and Development Studies. Subu has published over 250 original articles and book chapters in the field of social epidemiology, applied multilevel methods, health inequalities in
India, and on cross-comparative assessment of social determinants of health.
Beth E. Molnar is a social and psychiatric epidemiologist, studying causes and consequences of violence
against children and adolescents, particularly mental health and behavioral sequelae of child abuse, neglect, and
exposure to community violence. She is also studying the effects of neighborhood social processes on child
abuse and neglect, and on high-risk behaviors of adolescents. Dr. Molnar is an assistant professor of Society,
Human Development and Health at the Harvard School of Public Health.
Journal of Urban Health: Bulletin of the New York Academy of Medicine
doi:10.1007/s11524-016-0060-y
* 2016 The New York Academy of Medicine
Changes in Attitudes toward Guns and Shootings
following Implementation of the Baltimore Safe
Streets Intervention
Adam J. Milam, Shani A. Buggs, C. Debra M. Furr-Holden,
Philip J. Leaf, Catherine P. Bradshaw, and Daniel Webster
ABSTRACT Among youth 15 to 24 years of age, homicide and nonfatal shootings are the
leading causes of mortality and morbidity. Urban youth’s attitudes and perceptions
about the use of gun violence to resolve conflict present a major barrier to efforts to
reduce gun homicides and nonfatal shootings. The current investigation extends the
existing literature on attitudes toward guns and shootings among high-risk youth ages
18 to 24 by measuring perceived norms and viewpoints regarding gun violence in two
analogous Baltimore City neighborhoods pre-implementation and 1-year post-implementation of the Safe Streets intervention (adapted from the CeaseFire/Cure Violence
intervention). The Safe Streets intervention is designed for communities with high rates
of gun violence and utilizes outreach workers to identify and build trusting relationships
with youth ages 15 to 24 who are at greatest risk of being involved in gun violence. The
outreach workers also position themselves in the community so that they can rapidly
intervene in disputes that have the potential to lead to gun violence. Chi-squared tests
and exploratory structural equation modeling (ESEM) were used to examine changes in
attitudes toward gun violence 1 year after the implementation of the Safe Streets
intervention. There was a statistically significantly improvement in 43 % of the attitudes
assessed in the intervention community post-intervention compared to 13 % of the
attitudes in the control community. There was a statistically significant improvement in
the violent attitudes toward personal conflict resolution scale after implementation of
the intervention in both the intervention (b = −0.522, p G 0.001) and control community
(b = −0.204, p G 0.032). Exposure to the intervention (e.g., seeing stop shooting signs in
your neighborhood) was also associated with the nonviolent attitudes toward conflict
scale. Overall, the study found greater improvement in attitudes toward violence in the
intervention community following the implementation of the Safe Streets program.
These findings offer promising insights into future community violence prevention
efforts.
KEYWORDS Violence, Young adult, Attitudes, Gun violence
Milam, Furr-Holden, Leaf, Bradshaw, and Webster are with the Department of Mental Health, Center for
the Prevention of Youth Violence, Johns Hopkins University, Bloomberg School of Public Health, 624 N.
Broadway, 8th floor, Baltimore, MD 21205, USA; Milam and Buggs are with the Department of Health
Policy and Management, Johns Hopkins University, Bloomberg School of Public Health, 624 N.
Broadway, 5th floor, Baltimore, MD 21205, USA.
Correspondence: Adam J. Milam, Department of Mental Health, Center for the Prevention of Youth
Violence, Johns Hopkins University, Bloomberg School of Public Health, 624 N. Broadway, 8th floor,
Baltimore, MD 21205, USA. (E-mail: amilam3@jhu.edu)
MILAM ET AL.
INTRODUCTION
Despite an overall national reduction in fatal and nonfatal shootings since the
1990s, gun violence continues to have an enormous impact on the lives and wellbeing of many youth in urban areas of the USA. According to the US Department of
Justice’s Bureau of Justice, firearm homicides decreased nearly 40 % between 1993
and 2011, and nonfatal gun assaults against people 12 years old and older decreased
70 % during the same time period.1 The FBI reports continued year-over-year
decreases in gun homicides between 2011 and 2014.2 However, among youth 15 to
24 years of age, homicide remains the leading cause of death for black males and the
third leading cause of death for white males, with the crude rate of gun homicides
for black males being over 10 times that of their white counterparts. Furthermore,
the devastating effect of firearm violence on youth in America does not just involve
homicides; for every young person killed with a gun, there are about four other
youths who are victims of nonfatal gun assaults (Fig. 1).3
The majority of all firearm homicides and nonfatal shootings in the USA are
committed with a handgun.1 Although federal law prohibits the possession of a
handgun for any person under the age of 18, research has shown that many
juveniles and young men who are restricted due to disqualifying convictions do still
possess and carry handguns.3 Studies assessing gun possession and ownership
among urban youth have found that witnessing violence and expressing violenceprone attitudes seem to predict ownership of guns, particularly handguns.4
Ethnographic investigations have found that many urban youth believe that gun
carrying, particularly in neighborhoods with high levels of crime, is a normal
occurrence and that the social norm or Bcode of the street^ is to be ready and willing
to respond to threats with lethal violence.5, 6 Many young males in high-crime
neighborhoods also believe that an act of blatant disrespect requires a response
1
2
13
Violent Attitudes
toward Personal
Conflict Resolution
Seem Stop
Shooting Signs
in neighborhood
14
15
Personal & Peer
Violent Attitudes
toward Conflict
Resolution
Time
16
29
30
Non-violent
Attitudes toward
Conflict Resolution
Seen Stop
Shooting Signs
in
neighborhood
FIG. 1 Structural model examining attitudes toward gun violence after implementation of the Safe
Streets intervention.
CHANGES IN ATTITUDES TOWARD GUNS AND SHOOTINGS
potentially involving deadly force.6, 7 Previous research proposes that violent
retaliation is driven by the desire to restore one’s perceived reputation and social
status after an incident of disrespect, assault, or victimization.7 Additionally, male
youth in highly violent neighborhoods frequently believe that a failure to retaliate
may affect one’s status and even one’s safety.6–9
Urban youth’s attitudes and perceptions about the use of gun violence to resolve
conflict present a major barrier to efforts to reduce gun homicides and nonfatal
shootings among this population. However, violence prevention interventions aimed
at shifting youth’s views on violence as an acceptable means of retaliation or conflict
resolution have demonstrated some promise.10, 11 For example, one study of urban
youth ages 7 to 17 found that approximately 1 month after touring a hospital and
seeing a presentation on gun violence, followed by discussions with a police officer,
the youth expressed significantly reduced scores on a scale measuring beliefs
supporting aggressive behaviors. There was also a trend toward reduced scoring on
a scale measuring likelihood of violence, yet no significant changes in scores on
attitudes toward conflict or violent intentions.11 Another study examined scores on
a survey on attitudes regarding the use of violence in conflict before and after a
cohort of mostly eighth and ninth grade urban school students participated in a 2-h
hospital-based program that simulated the final living moments of a youth killed by
a gunshot.12 The validated Attitudes Toward Guns and Violence Questionnaire
(AGVQ)13 was given to the study participants 2 weeks prior and 4 weeks after the
intervention, and the pre-post comparison in scores revealed significant decreases in
public and charter students’ scores on questions related to aggressive responses to
shame and in total AGVQ score for public school students. However, no measurable
change in total AGVQ score was found among charter school students.13 These two
studies suggest that the youth’s attitudes involving guns and violence might be
malleable, but they do not directly answer questions about whether a larger-scale,
community-based intervention could alter perceptions and beliefs about the
appropriateness of resorting to gun violence to settle disputes.
Cure Violence, formerly known as CeaseFire and referred to as Safe Streets in
Baltimore and in the current paper, is a public health violence preventive
intervention designed to prevent shootings among adolescents and young men by
changing attitudes, behaviors, and social norms most directly related to gun
violence. Following lessons learned from public health efforts to prevent the spread
of infectious diseases, the program is designed for communities with high rates of
gun violence and utilizes outreach workers to identify and build trusting
relationships with youth ages 15 to 24 who are at greatest risk of being involved
in gun violence, based on a history of violence or current involvement in risky street
activities such as gang affiliation or illicit drug selling. In addition to serving as
positive role models and resources for connecting youth to job and educational
opportunities to steer them from actions that might heighten violence risk, the
outreach workers also position themselves in the community so that they can rapidly
intervene in disputes that have the potential to lead to gun violence. When
encountering or being informed of potential conflicts, the workers act quickly to
intervene with one or both parties in the dispute depending upon the circumstances.
One component of the intervention is to get the involved individual(s) to identify
that they could or would experience negative consequences to a violent response and
to help identify alternatives other than violence for getting relief from the causes of
the dispute(s). Ideally, this allows the parties to utilize alternative dispute resolution
and reconciliations not involving the potential of death. Consistent with the Cure
MILAM ET AL.
Violence model, Safe Streets also seeks to shape community norms that renounce
violence and the actions of many in the neighborhood who are not Bclients^ through
public events and campaigns that bring the community together, provide positive
activities for youth and families, and promote nonviolent behavior.14 The Cure
Violence intervention has been replicated in dozens of cities across the world and has
been associated through multiple independent evaluations with significant reductions in shootings in implementation areas.12 For example, prior research has shown
that Safe Streets is largely effective at reaching its primary goal of reducing shootings
and homicides in high-violence communities in Baltimore City.15.
However, to date, no evaluation has measured Cure Violence’s (or Safe Streets’)
impact on attitudes about the acceptability of using guns to settle conflicts, which
are hypothesized to be an important mediator of the effects of the intervention on
violence-related outcomes. Further exploration into the social norms related to gun
violence among young men living in violent urban communities, as well as increased
knowledge of the effectiveness of interventions aimed at shaping those norms, would
not only inform our understanding of the potential effects of Safe Streets but also
help to tailor gun violence prevention efforts for this high-risk population. The
current investigation extends the existing literature on attitudes toward guns and
shootings among high-risk youth by m...
Purchase answer to see full
attachment