Sie sind auf Seite 1von 12

Unraveling the Link between Trauma and

Male Delinquency: The Cumulative Versus


Differential Risk Perspectives
Tina Maschi

This study examined how the cumulative (additive) versus differential (individual) effects of
trauma influenced male delinquency. Using a comprehensive measure of trauma, a secondary
data analysis was conducted on a nationally representative sample of male youths between the
ages of 12 and 17. Logistic regression analyses revealed that all three types of stressors had an
effect on male delinquency, even after controlling for potential confounding factors. These
findings underscore the need to adopt a broader view of the types of trauma linked to juvenile
delinquency. Implications for social work practice and intervention research are discussed.

KEY WORDS: cumulative risk; exposure to violence; juvenile delinquency; stressful life events; trauma

W
ithin the social work, mental health, and Although both differential and cumulative ef-
criminal justice fields, researchers gen- fects of trauma have been linked to youths’ antiso-
erally agree that trauma places youths cial behavior, researchers generally have investigated
at risk of juvenile delinquency. Proponents of the these effects independently. Broadly defined, trauma
developmental and cumulative risk theoretical in this regard refers to being a victim of violence,
models (Cichetti & Carlson, 1989; Rutter, 1985, being a witness to violence, or experiencing stress-
1987) disagree, however, about whether it is a spe- ful life events. Of the three types of trauma, re-
cific risk factor or an accumulation of risk factors search on child maltreatment and witnessing vio-
that produces an adverse outcome, such as delin- lence has mainly adopted a developmental approach,
quency. Developmental theory suggests that youths and research on stressful life events has adopted a
who have experienced a traumatic event of suffi- cumulative risk approach. Child maltreatment, often
cient magnitude, such as child physical abuse, may defined as being a victim of physical abuse, sexual
experience long-term, negative consequences— abuse, and neglect that occurs before age 12, is
psychological, social, and behavioral—that continue positively related to adolescent and adult criminal-
and sometimes worsen in adolescence and adult- ity (Falshaw, Browne, & Hollin, 1996; Haapsalo &
hood (Dodge, Bates, & Pettit, 1990; Lemmon, 1999; Pokela, 1999; Lemmon, 1999; Widom, 1989a;
Perez, 2001; Widom, 1989a). Research has shown Zingraff, Leiter, & Myers, & Johnsen, 1993). First-
that 50 percent to 79 percent of male victims of generation studies on the child maltreatment–de-
child maltreatment, whose abuse occurred before linquency connection, conducted between 1950
age 12, later became involved in serious juvenile and 1988, generally found that youths who were
delinquency (Lemmon; Stouthamer-Loeber, Wei, maltreated in childhood were at greater risk of
Homish, & Loeber, 2002). In contrast, cumulative engaging in illegal behavior, especially violent be-
risk theory suggests that youths who experience an havior (Widom, 1989a). A causal link, however,
accumulation of negative or stressful life events, such cannot be drawn from these early studies because
as parental divorce or school suspension, increase of their methodological limitations, such as the use
their risk of engaging in juvenile delinquency of cross-sectional retrospective designs, small sample
(Agnew, 2002; Eitle & Turner, 2002; Garmezy & sizes, a lack of comparison groups, differing defini-
Masten, 1994; Hoffman & Cerbone, 1999; Ireland, tions and data collection methods, and failure to
Smith, & Thornberry, 2002). control for confounding variables (Widom, 1989b).

Maschi
CCC / Unraveling
Code: 0037-8046/06 the ©2006
$3.00 Link between
National Trauma and
Association of Male Delinquency: The
Social Workers Cumulative Versus Differential Risk Perspectives 59
Later research, however, provided stronger evi- acting out aggressively or committing delinquent
dence of a causal relationship between being abused acts (Hoffman & Su, 1997; Ireland & Widom, 1994).
as a child and adolescent delinquency. Second-gen- Events such as parental divorce or separation, mov-
eration studies conducted after 1988 used prospec- ing to a new home, chronic illness or death in the
tive longitudinal designs, comparison groups, and family, failing a class or being suspended from school,
control variables. For example, in Widom’s (1989b) among others, may engender a sense of loss or un-
classic study, the maltreated group was more likely certainty. Also, many of these experiences are un-
to have had official juvenile and adult criminal of- changeable and can block the achievement of de-
fenses, including violent offenses, than the com- sired goals (Agnew, 1992). Research exploring the
parison group. Of all the subtypes of maltreatment, impact of the combined effects of recent stressful
physical abuse had the strongest relationship to vio- life events and exposure to violence on juvenile
lent crime later in life. Although advancing our delinquency has been minimal. However, prelimi-
understanding of the problem, many second-gen- nary results support a cumulative risk model. Using
eration studies were still confounded by other longitudinal data, Eitle and Turner (2002) found
methodological issues. For example, they did not that the risk of antisocial behavior among youths
control for other adverse experiences, such as wit- increased with recent exposure to community vio-
nessing violence or experiencing stressful life lence and stressful life events, as well as having re-
events, or victimization that occurred after age 12. cently received traumatic news.
An exception, Ireland and colleagues (2002) con- Traumatic experiences linked to juvenile delin-
trolled for lifetime adverse experiences and found quency may range from minor to severe and have
a relationship between youths who were victim- an independent or additive effect that may span
ized in both childhood and adolescence or ado- from childhood to adolescence. What remains to
lescence only and delinquency. However, no sig- be understood is whether the magnitude of spe-
nificant relationship was found between youths cific or differential risk factors or the accumulation
who experienced childhood maltreatment only of risk factors increases the risk of juvenile delin-
and delinquency. quency. The present study extends earlier research
Witnessing violence is another traumatic expe- by using a comprehensive measure of trauma, and
rience linked to youths’ illegal behavior (Edleson, it examines the impact of the cumulative versus
1999; Jaffe,Wolfe,Wilson, & Zak, 1986; Johnson et differential risks on serious delinquency in adoles-
al., 2002). A high degree of exposure to family and cent boys.The components of the trauma measure
community violence has been found among the include (1) being a victim of violence (for example,
vast majority of adjudicated youths (Erwin, physically abusive punishment, physical assault, and
Newman, McMackin, Morrissey, & Kaloupek, sexual assault), (2) witnessing family and commu-
2000; Martin, Sigda, & Kupersmidt, 1998). Al- nity violence, and (3) experiencing stressful life
though witnessing violence in the home and com- events (for example, parental separation or divorce,
munity often occurs in conjunction with being a death of a loved one, and being suspended from
victim of violence, there have been very few in- school). In addition, this research controlled for the
vestigations of their combined effects. Cross-sec- common correlates of delinquency, including age,
tional and longitudinal analyses of exposure to vio- race and ethnicity, social class, family structure, geo-
lence have found that youths who have been graphic location, delinquent peer exposure, nega-
victims of or witnesses to family or community tive affect, and social support.
violence or both are more likely to engage in ag- Using a developmental perspective, I hypoth-
gressive and antisocial behavior (Johnson et al.; esized that prior victimization experiences, such
Kruttschnitt & Dornfeld, 1993). as being a victim of physical or sexual assault, in-
In addition, youths who experience an accumu- crease the likelihood of juvenile delinquency, es-
lation or clustering of stressful life events are more pecially violent delinquency, in adolescent boys,
likely to engage in delinquent acts than their peers even when controlling for confounding variables.
who do not experience these events (Agnew & Adopting a cumulative risk perspective, I counter-
White, 1992; Eitle & Turner, 2002; Hoffman & hypothesized that the additive effects of exposure
Cerbone, 1999). Adolescent boys tend to external- to violence and stressful life events greatly increase
ize their reactions to these stressful experiences by the odds of adolescent boys engaging in juvenile

60 Social Work Volume 51, Number 1 January 2006


delinquency. Gaining a better understanding of the Adolescents were asked about their past experi-
trauma–delinquency connection can help research- ences of direct and indirect violence, past-year stress-
ers and practitioners develop more effective pre- ful life events, delinquent activity, emotional and
vention and treatment responses targeted at pro- psychological states, and peer relations.This analy-
tecting and maintaining the well-being of youths sis mainly used the adolescents’ self-report infor-
and their communities. mation for the variables under investigation, ex-
cept for family sociodemographic variables, which
METHOD were based on parent or guardian reporting.
Sample
The male adolescent sample, ages 12 to 17 Measures
(unweighted N = 2, 018 and weighted N = 2,065), Juvenile Delinquency. The NSA Juvenile Delin-
was drawn from the 1995 National Survey of Ado- quency Module, a self-report delinquency measure,
lescents (NSA). The 1995 NSA consists of a na- has two scales of serious juvenile delinquent acts:
tionally representative sample of 4,023 adolescents property offending and violent offending. Accord-
ages 12 to 17, of which roughly half are boys. The ing to official statistics, property offending and vio-
sample was designed to reflect the general popula- lent offending are the two most common delin-
tion of adolescents from 1988 U.S. Census Bureau quent activities for which male youths are arrested
statistics for geographic location, age, gender, and (Snyder, 2003). The validity of adolescents’ self-re-
race and ethnicity.The sampling design used was a ports for past delinquent behavior, especially when
stratified random sample, and the male population asked during a home phone interview is of some
consisted of two subsamples, a national probability concer n (Brown, 1999; Far rington, Loeber,
household sample (n = 1,599) and a probability Stouthamer-Loeber, Van Kammen, & Schmidt,
oversample (n = 419) from central city areas in the 1996). The measure used in this study only asked
United States (Kilpatrick & Saunders, 1995). respondents to remember their delinquent events
over the past year, a time frame in which memory
Procedures recall is much more dependable (Williams &
Self-report information was gathered in 1995 from Uchiyama, 1989).The survey questions were phrased
adolescents and their parents or guardians during in a way in which the adolescent respondents only
one-hour structured telephone interviews. House- had to respond with a yes or no. Therefore, if any
holds were reached using random digit dialing. other people (such as parents or guardians) were in
Depending on the respondent’s preference, a tele- listening range, the content of the interview would
phone interview was conducted in either English have remained confidential (Kilpatrick & Saunders,
or Spanish. Permission was obtained first from a 1995).
parent or guardian to interview the respondent Property offending included three items: (1) hav-
adolescent in the home. Interviewers used com- ing stolen or attempted to steal an item worth
puter-assisted telephone interviewing, which of- more than $100, (2) having stolen or attempted to
fers more advanced capabilities such as complex steal a motor vehicle (car or motorcycle), and (3)
skip patterns and question ordering. The inter- having broken into or attempted to break into a
viewers were provided by the New York survey building or vehicle. Property offending was coded
research team Schulman, Ronca, and Bucuvalas, as a binary variable (1 = past-year property of-
Incorporated. fending, 0 = no past-year property offending). Vio-
The response rate for the telephone interviews lent delinquency or violent offending had four items:
was high. Ninety percent of parents and caregivers (1) being involved in gang fights, (2) using force or
completed interviews, of which the majority (78.9 strong-arm methods to get money or things from
percent, n = 4,235) gave permission for an adoles- people, (3) having or attempting to have sexual
cent in the home to be interviewed. This resulted relations with someone against his or her will,
in a 75 percent response rate of adolescents from and (4) attacking someone with the idea of seri-
eligible households, of which 95.0 percent (n = ously hurting or killing them. Violent delin-
4,023) had parental permission and 83.2 percent (n quency also was coded as a binary variable (1 =
= 3,347) had a completed parent interview past-year violent delinquency, 0 = no past-year vio-
(Kilpatrick & Saunders, 1995). lent delinquency). Cochran’s chi-square statistic was

Maschi / Unraveling the Link between Trauma and Male Delinquency: The Cumulative Versus Differential Risk Perspectives 61
significant (p < .01) for both property offending • serious illness or injury of a family member,
and violent delinquency. close friend, or self.
Trauma. The NSA’s 20-item multivictimization
scale was used to measure lifetime exposure to vio- The blockage of positively valued goals was
lence for youths ages 12 to 17. Exposure to vio- measured by three school-related items: (1) repeat-
lence consisted of four types of differential trauma ing a school grade, (2) getting suspended from
that may have occurred in the home, school, or school, and (3) receiving at least one failing grade.
community: (1) physical abuse, (2) physical assault, According to general strain theory, any blockages
(3) sexual assault, and (4) witnessed violence. Physi- to the achievement of success or status, such as in
cal abuse was defined as an act in which a youth the case of educational goals for adolescents, in-
received physical punishment from a caregiver, creases the risk of juvenile delinquency (Agnew,
which consisted of spanking (resulting in bad 1985). Cochran’s chi-square statistic was significant
marks, bruises, cuts, or welts), burning, cutting, or (p < .01) for both stressful life events subscales.
being tied up. Physical assault was defined as the
youth being hit, attacked, or beat up either in the Control Variables
neighborhood, school, or home. Sexual assault in- Some of the common correlates of delinquency
cluded sexual acts in which a youth was forced to were included in the analysis as covariates to ad-
give or receive fondling, oral sex, or anal penetra- just for their effect on the relationships that are of
tion. Witnessed violence was defined as having ob- central interest. The following variables are risk or
served someone being shot, stabbed, sexually as- protective factors commonly associated with juve-
saulted, mugged, robbed, or threatened with a nile delinquency and commonly controlled for in
weapon. For cumulative trauma, each reported studies examining the link between trauma and de-
event of exposure to violence was equivalent to linquency: age, socioeconomic status, race and eth-
one point. The sum of these points comprised the nicity, family structure, geographic location, nega-
cumulative exposure to violence score. Because tive affect, delinquent peer exposure, and social
each of the scale items was dichotomous, the support (Agnew & White, 1992; Brezina, 1998;
Cochran chi-square was calculated. It was signifi- Brownfield, 1986; Capowich, Mazerolle, & Piquero,
cant (p < .01) for both the differential and cumu- 2001; Esposito & Clum, 2002; Heck & Walsh, 2000;
lative scales of exposure to violence. Jang, 1999). Age was measured by the adolescent’s
The NSA Stressful Life Events Module was used current age at the time of the interview. Race and
to measure such events during the preceding year. ethnicity was dichotomized into two groups based
Each of the 14 stressful life events reported equaled on perceived societal status (1 = ethnic minority
one point. The sum of points represented the total status and 0 = majority status). The racial majority
cumulative stressful life events score for that year. group consisted of white, non-Hispanics, and the
The differential effects of stressful life events were racial and ethnic minority group consisted of Af-
measured by subdividing the scale into two sub- rican American, Hispanic, Native American, and
categories: (1) the loss of positively valued stimuli Asian youths. Socioeconomic status was measured
and (2) the blockage of positively valued goals, by total family income and parent education.Total
which according to general strain theory are the family income (yearly salary earned by one or more
two major categories of stressful life events that household members) was a binary variable catego-
are linked to juvenile delinquency (Agnew, 1985). rized as above or below the 1995 estimated pov-
The loss of positively valued stimuli was repre- erty level (1 = $20,000 and higher and 0 = less
sented by than $20,000). Parent education was measured by
whether the person designated as the head of
• parents separated or divorced household had obtained a high school diploma or
• parental unemployment GED (1 = yes and 0 = no). Family structure was
• new stepmother or stepfather defined as a married couple (two-parent biologi-
• death of a family member cal or adoptive family or stepfamily); and other
• loss of a close friend family categories included parents or caregivers
• move to a new home who were separated, living as an unmarried couple,
• change to a new school divorced, widowed, or single and never married (1

62 Social Work Volume 51, Number 1 January 2006


= intact and 0 = nonintact). Geographic location Cochran chi-square was significant (p < .01) for
described urban (1 = city and city suburb) or rural this measure.
(0 = small town or rural) location of the family
residence. ANALYSIS
Negative affect, which has positively correlated to To evaluate the cumulative versus differential ef-
juvenile delinquency in earlier research (Agnew & fects of trauma on male delinquency, logistic re-
White, 1992; Brezina, 1998), was represented by gression analyses examined the dichotomous out-
two measures: anger and depression. Anger was as- come variables: past-year property offending and
sessed by the following questions: “Have little things violent offending. All analyses were weighted based
bothered you a lot or could they make you very on 1988 U.S. Census Bureau statistics and controlled
angry?” (1= yes, 0 = no), and “When was the most for potential confounding variables: age, race and
recent time you felt that way?” (0 = never angry, 1 ethnicity, socioeconomic status, family structure,
= angry more than six months ago, 2 = angry within geographic location, negative affect, delinquent peer
the past six months, and 3 = angry within the past exposure, and social support. Cases with missing
month). Cronbach’s alpha for this measure was .79 data were dropped from the analyses.
and the Cochran chi-square was significant (p <
.01). Depression was measured by the 12-item NSA FINDINGS
Depression Module. This scale is based on DSM- Sample Characteristics
IV criteria for depression and includes items such The mean age of the male adolescents was 14.47
as “depressed most of the day for at least two weeks,” (SD = .17). Most of the male youths were white
“felt worthless,” “had trouble/difficulty concentrat- and resided in two-parent families in small town or
ing,” or “had thoughts of hurting oneself.” Each rural locations (Table 1). About 15 percent (n =
affirmative response equaled one point, with a pos- 301) of the sample reported having committed at
sible total score of 12. The depression scale con- least one type of delinquent act during the past
sisted of dichotomous items (α = .58) and the year. Relative to property offending (7.4 percent, n
Cochran chi-square was significant for this sample = 153), violent offending (10.6 percent, n = 219)
(p < .01). was more commonly reported by the sample.
Delinquent peer exposure was represented by the The prevalence and type of trauma experienced
13-item NSA Delinquent Peer Exposure Module varied among sample members (N = 2,065). The
(α = .88) and measured the proportion of friends majority reported being exposed to violence (77.8
who were reported as engaging in delinquent acts percent, n = 1,607) or having experienced past-
during the past year. These acts included theft or year stressful life events (88.3 percent, n = 1,823).
burglary, alcohol or drug use, alcohol or drug sales, Witnessing violence (76.3 percent, n = 1,576) was
or physical or sexual assault. Youths were first asked the most common type of exposure to violence
whether they had friends who engaged in any of reported. For direct victimization experiences, re-
these activities. If they answered yes, they were asked spondents reported being victims of physical as-
what proportion of their friends engaged in that sault (21.3 percent, n = 439) much more than be-
activity. A five-item Likert scale determined the ing victims of physically abusive punishment (8.5
proportion of delinquent friends who committed percent, n = 176), or sexual assault (3.4 percent, n =
any of these delinquent acts with responses ranging 71). As for past-year stressful life events, more of the
from 4 = all of them to 0 = none of them. youths (80.7 percent, n = 1,666) reported at least
Social support was defined as the presence of a one past-year incident of the loss of positively val-
caring individual in a youth’s life, operationalized ued stimuli than the blockage of positively valued
by two survey questions: “Was there anyone in your goals (51.0 percent, n = 1,054).
childhood that you knew you could count on or
depend on to be there when you needed them?” Effects of Trauma
and “Was this true throughout your whole child- To determine the relative contribution of cumula-
hood?” These two items were added together to tive trauma and differential trauma to property of-
represent the youth’s level of social support (2 = fending and violent delinquency, hierarchical lo-
always true, 1 = sometimes true, 0 = never true). gistic regression analyses were conducted. For each
Cronbach’s alpha (.39) was low; however, the of the analyses, the control variables were entered

Maschi / Unraveling the Link between Trauma and Male Delinquency: The Cumulative Versus Differential Risk Perspectives 63
Table 1: Sample Demographic quency, when controlling for potential confound-
Characteristics (Weighted N = 2,065) ing factors.
Variable % N
Results indicate that both cumulative trauma and
differential trauma explained adolescent male de-
Age
linquency. The model chi-square statistic for cu-
12–14 50.6 1,045
mulative trauma was significant for property of-
15–16 49.2 1,017
fending [χ2(N = 1,862) = 179.70, p < .01] and
Unknown 0.2 3
violent offending [χ2(N = 1,862) = 71.65, p < .01)
Total 100.0 2,065
(Table 2), suggesting that both models fit the data.
Race/ethnicity
The analyses revealed that the cumulative effects of
White 71.0 1,466
exposure to stressful life events significantly increased
African American 14.2 293
the odds of youths engaging in property offending
Hispanic 7.5 155
but not the cumulative effects of exposure to vio-
Native American 3.4 70
lence. The odds that delinquent youths had com-
Asian 1.0 22
mitted property offending were 1.31 times greater
Other 0.8 17
for adolescent males who had experienced past-
Unknown 2.0 42
year stressful life events than adolescent males who
Total 100.0 2,065
had not. In comparison, exposure to both violence
Family structure
and stressful life events significantly predicted vio-
Non-intact 21.2 438
lent delinquency.The odds ratio was slightly greater
Intact 78.6 1,623
for cumulative exposure to violence (OR = 1.40, p
Unknown 0.2 4
< .01) than the experience of stressful life events
Total 100.0 2,065
(OR = 1.17, p < .01). The control variables, age
Total family income
(OR = .78, p < .01) and parent education (OR =
Less than $20,000 16.1 332
2.37, p < .05) significantly influenced property of-
More than $20,000 77.9 1,609
fending. Family income (OR = 1.84, p < .05), mi-
Unknown 6.0 124
nority status (OR = 2.30, p < .01), and family struc-
Total 100.0 2,065
ture (OR = .59, p < .05) significantly influenced
Parent/head of household education
violent offending using the cumulative trauma
8th grade or less 2.2 45
model.
Some high school 7.4 153
The chi-square statistic for differential trauma
High school graduate 31.1 643
reveals an overall fit of the model for the data for
Some college 27.6 570
property offending [χ2(N = 1,862) = 159.24, p <
College degree 22.0 455
.01] and violent delinquency [χ2(N = 1,862) =
Graduate degree 9.5 197
81.58, p < .01) (Table 3). A positive and significant
Unknown 0.1 2
effect was found for the blockage of positively val-
Total 100.0 2,065
ued goals. Male youths who experienced some type
Geographic location
of school failure in the past year, such as failing a
Rural 65.0 1,342
class, being left back a grade, or being suspended,
Urban 34.8 719
were 2.31 times more likely to commit property
Unknown 0.2 3
offenses than male youths who had not. Unexpect-
Total 100.0 2,065
edly, having been a victim of sexual assault signifi-
cantly decreased the odds of engaging in property
offending (OR = .39, p < .05).This result could be
first as a block, followed by the set of trauma vari- attributed to the relatively small sample size (n =
ables in the second block. The chi-square statistic, 35) of male youths who reported a history of sexual
the unstandardized logistic regression coefficient, abuse. Having been a victim of physical assault,
the standard error of b, the odds ratio, and the con- witnessing violence, and experiencing the block-
fidence intervals are presented for the impact of age of positively valued goals were all positively
cumulative trauma (Table 2) and differential trauma and significantly related to committing violent acts.
(Table 3) on property offending and violent delin- The odds that respondents had done so were 1.52

64 Social Work Volume 51, Number 1 January 2006


Table 2: Cumulative Effects of Trauma on Property Offending
and Violent Offending (Weighted N = 1,862)
Model 1 Model 2
Property Offending Violent Offending
Variables b SEB Expb CI (95%) b SEB Expb CI (95%)
Exposure to violence –.03 .05 0.97 (0.87, 1.08) .34 .05 1.40*** (1.27, 1.53)
Stressful events .27 .05 1.31*** (1.18, 1.45) .16 .05 1.17*** (1.07, 1.28)
Anger .26 .10 1.30** (1.08, 1.58) .25 .09 1.28** (1.09, 1.51)
Depression –.03 .04 0.97 (0.89, 1.06) –.01 .04 0.99 (0.92, 1.07)
Delinquent peers .19 .02 1.21*** (1.17, 1.25) .12 .01 1.12*** (1.09, 1.15)
Social support –.45 .16 0.64 (0.39, 1.04) –.36 .19 0.70 (0.49, 1.01)
Age –.25 .08 0.78*** (0.67, .906) .01 .06 1.01 (0.89, 1.14)
Income –.53 .29 0.59 (0.34, 1.03) .61 .28 1.84* (1.08, 3.16)
P/HH education .86 .41 2.37* (1.07, 5.25) –.55 .28 0.58 (0.33, 1.01)
Race/ethnicity –.13 .25 0.88 (0.54, 1.43) .83 .21 2.30*** (1.52, 3.47)
Family structure –.13 .26 0.88 (0.53, 1.46) –.52 .23 0.59* (0.38, 0.92)
Geographical location .03 .23 1.03 (0.66, 1.60) .17 .20 1.19 (0.80, 1.75)
Constant –1.24 1.15 0.29 –4.37 1.01 0.01***
–2LL (Initial) 827.70 855.71
–2LL (Ending) 648.00 784.07
χ2 179.70 71.65
Sig .01 .01
Note: P/HH = parent/head of household.
*p < .05. **p < .01. ***p < .001.

times greater for youths who had been physically Study Limitations
assaulted, had witnessed violence, or had experi- Several limitations of the research design warrant
enced the loss of positively valued goals, compared discussion. First, although self-report studies are
with youths who experienced other types of trauma. generally considered more reliable than official re-
Anger and delinquent peer exposure were the only ports for capturing undetected cases of trauma and
two control variables that were positive and signifi- delinquency, they are also subject to recall bias, er-
cant for both the differential and cumulative risk ror, or validity issues. Factors possibly affecting the
models.Youths with higher levels of anger and ex- accuracy of reporting for topics as sensitive as trauma,
posure to delinquent peers were at increased odds especially sexual abuse for male adolescents, include
of engaging in property offending and violent of- unconscious denial, repression of past events, the
fending, respectively. Age (OR = .78, p < .01) re- redefinition of past events in light of present expe-
mained a significant control variable for property rience, and avoidance of stigma (Brown, 1999;
offending and race and ethnicity (OR = 2.23, p < Finkelhor, Hotaling, Lewis, & Smith, 1990; Huizinga
.01), income (OR = 1.84, p < .05), and family struc- & Elliott, 1986). In addition, the study did not use
ture (OR = .60, p < .05) were significant for vio- standardized measures of trauma or distinguish
lent offending using the differential effects of trauma. whether exposure to violence occurred in the home
Younger adolescents were at a decreased risk of or in the community. Furthermore, although this
engaging in property offending. Ethnic and racial survey is one of the only data sets available in the
minority youths had a more than two times greater public domain that has a comprehensive measure
likelihood of engaging in violent offending com- of trauma and a delinquency measure, it is almost
pared with white youths. Although male youths 10 years old, possibly limiting its representativeness
from homes with incomes higher than $20,000 had and generalizability to contemporary adolescents.
increased odds of engaging in violent offending, Moreover, although this study uses a nationally rep-
male youths from two-parent families did not. resentative sample of male adolescents, its findings

Maschi / Unraveling the Link between Trauma and Male Delinquency: The Cumulative Versus Differential Risk Perspectives 65
Table 3: Differential Effects of Trauma on Property Offending
and Violent Offending (Weighted N = 1,862)
Model 3 Model 4
Property Offending Violent Offending
Variables b SEB Expb CI (95%) b SEB Expb CI (95%)
Physical abuse .48 .27 1.55 (.907, 2.64) .42 .24 1.52 (0.95, 2.44)
Physical assault .13 .11 1.14 (.913, 1.42) .33 .10 1.40*** (1.16, 1.69)
Sexual assault –.94 .31 0.39** (.212, .715) .19 .15 1.21 (0.91, 1.62)
Witness violence –.07 .11 0.94 (.760, 1.15) .39 .09 1.48*** (1.25, 1.75)
Loss + stimuli .12 .07 1.13 (.987, 1.29) .08 .06 1.09 (0.97, 1.22)
Blockage + goals .84 .14 2.31*** (1.76, 3.02) .36 .12 1.44** (1.14, 1.81)
Anger .29 .10 1.33** (1.10, 1.62) .24 .09 1.28** (1.08, 1.51)
Depression –.02 .05 0.98 (.894, 1.07) .01 .04 1.01 (0.93, 1.09)
Delinquent peers .18 .02 1.20*** (1.16, 1.24) .11 .01 1.11*** (1.08, 1.15)
Social support –.29 .21 0.75 (.497, 1.14) –.32 .19 0.72 (0.50, 1.05)
Age –.25 .08 0.78** (.670, .915) .01 .06 1.01 (0.89, 1.15)
Income –.32 .30 0.72 (.406, 1.29) .61 .28 1.84* (1.07, 3.16)
P/HH education .78 .41 2.19 (.987, 4.86) –.46 .29 0.63 (0.36, 1.12)
Race/Ethnicity –.19 .27 0.83 (.494, 1.40) .80 .22 2.23*** (1.46, 3.41)
Family structure –.16 .26 0.85 (.507, 1.43) –.51 .23 0.60* (0.38, 0.93)
Geographical location –.02 .24 0.98 (.613, 1.56) .17 .20 1.18 (0.79, 1.76)
Constant –1.77 1.23 0.17 –4.70 1.04 0.01***
–2LL (Initial) 768.60 850.48
–2LL (Ending) 609.36 768.89
χ2 159.24 81.58
Sig .01 .01
Note: P/HH = parent/head of household.
*p < .05. **p < .01. ***p < .001.

are not generalizable to the female adolescent become delinquent and others remain resilient to
population. Finally, evidence for a causal pathway its adverse effects. These explanations can inform
between trauma and delinquency was based on a how effective prevention and intervention efforts
cross-sectional, retrospective design in which it is must be structured to improve health, mental
determined the proper temporal sequencing of the health, and social justice outcomes. These findings
variables. are particularly relevant for social workers who
touch the lives of at-risk youths, including those
DISCUSSION who work in schools, juvenile justice, substance
This investigation compared explanations for the abuse agencies, and child welfare and other fam-
developmental and cumulative risk models regard- ily-serving agencies.
ing the effect of trauma on juvenile delinquency Partial support was found for both opposing
using a nationally representative sample of male perspectives. Overall, some evidence was found for
adolescents. The relative contribution of cumula- the cumulative risk hypothesis that the additive ef-
tive and differential trauma were examined to dis- fects of exposure to violence and stressful life events
cern how well trauma predicted serious delin- increase the odds of juvenile delinquency among
quency (property offending and violent offending) adolescent boys. In addition, there is support for
among adolescent males. A better understanding the hypothesis on the differential effects of
of the types of trauma linked can help identify trau- trauma that past victimization experiences signifi-
matized youths most at risk of delinquency as well cantly increase the likelihood of later delinquency,
as help to explain why some traumatized youths especially violent delinquency. A series of analyses

66 Social Work Volume 51, Number 1 January 2006


revealed that being a victim of or witness to vio- the type of trauma experienced. The pathway to
lence and experiencing stressful life events increases violent delinquency for adolescent boys was sig-
the likelihood that male youths would engage in nificantly linked to being a victim of physical as-
serious acts of delinquency. sault and witnessing violence. Male youths who
In general, these findings suggest that there are were victims of or witnesses to violence were sig-
multiple pathways from trauma to delinquency nificantly more likely to have perpetrated violence.
among adolescent boys. Being a victim of trauma In contrast, the effect of the differential effects of
was both a reoccurring additive effect of less severe exposure to violence was not positively or signifi-
stressors and the result of one major stressor. Gen- cantly related to property offending. In fact, among
eral strain theory is a theoretical framework that is adolescent boys, having been a victim of sexual
particularly suited for use by social work practitio- abuse inversely affected property offending. That
ners attempting to conceptually ground their in- is, male youths who had been sexually abused were
terventions with troubled youths. It posits a causal significantly less likely to engage in property of-
link between prior experiences of strain (that is, fending than male youths who had not been sexu-
adverse life experiences, such as trauma) and juve- ally abused. No significant relationship was found
nile delinquency. This connection is argued to be between being a victim of sexual abuse and vio-
mediated through an adverse emotional response lent offending.
(that is, negative affect, often in the form of anger) The link between being a victim of physical as-
to the trauma (Agnew, 1985). Research has shown sault and violent offending is not surprising. Male
negative affect to be both a short- and long-term youths are overwhelmingly the victims of nonsexual
consequence of trauma as well as a risk factor for crimes, such as homicide, aggravated assault, simple
juvenile delinquency (Brezina, 1998). assault, robbery, larceny, and vandalism (Finkelhor
This study’s findings are consistent with a gen- & Ormrod, 2000; Snyder & Sickmund, 1999). Male
eral strain theory perspective, which emphasizes the youths also are more likely to be victimized by
heightened risk of delinquency associated with any friends and acquaintances, as opposed to family
of the following combination of stressful factors: members and strangers (Finkelhor & Ormrod;
the magnitude, recency, duration and frequency, and Snyder & Sickmund).This finding underscores the
clustering of negative life events (Agnew & White, need for social workers to assess and treat not only
1992).Therefore, it is plausible that one major or a the impact of family violence, but also the impact
series of minor events can lead to delinquency that of community violence, especially among adoles-
is exacerbated by an adverse emotional response to cent boys and their peers. The effect of peer vio-
trauma. lence among friends and rival peer groups may be
The findings also reveal that the relationship a strong indicator of why adolescent boys engage
between trauma and delinquency varied by the type in violence.
of trauma and the type of offending. In the cumu- The differential type of trauma (that is, the block-
lative trauma model, the effect of exposure to vio- age of positively valued goals) is linked to both prop-
lence significantly increased the odds that male youths erty offending and violent offending. Male youths
would engage in violent delinquency, but not prop- who experienced school failure or blocked goals
erty offending. In contrast, stressful life events were were at significantly greater risk of engaging in
positively related to serious juvenile delinquency. property offending, such as stealing or damaging
Adolescent boys who had experienced stressful life the property of others, as well as engaging in vio-
events in the preceding year were at increased risk lence.The link between the blockage of positively
of both property offending and violent delinquency, valued goals and male delinquency has both con-
a finding consistent with the literature (Agnew & ceptual and empirical support in the literature
White, 1992; Hoffman & Cerbone, 1999). These (Agnew, 1985; Hoffman & Cerbone, 1999;
findings underscore the need for practitioners to Mazerolle & Maahs, 2000). This finding supports
conduct assessments on the occurrence of proximal the general strain theory perspective that the fail-
(recent) and distal (not so recent) minor stressors ure to achieve legitimate social status is an instru-
and coping abilities to deal with such events. mental factor as to why some adolescent boys seek
Exploring the differential effects of trauma re- illegitimate social status through delinquency
vealed a pattern of offending that largely reflected (Agnew, 1985).

Maschi / Unraveling the Link between Trauma and Male Delinquency: The Cumulative Versus Differential Risk Perspectives 67
IMPLICATIONS FOR SOCIAL would benefit from adopting anger management
WORK PRACTICE and prosocial peer mentoring programs and school
Many social work practitioners are intricately in- retention policies for high-risk youths.
volved with youths who have been affected by
trauma and those who engage in delinquency. FUTURE RESEARCH
School social workers, child welfare workers, juve- The positive and significant associations found be-
nile and criminal justice social workers, substance tween trauma and delinquency warrant future lon-
abuse social workers, and community mental health gitudinal investigations into a causal pathway model
social workers are positioned to have considerable linking trauma and delinquency. Future research
influence on the lives of youths and their families that would be useful for social work practitioners
dealing with adverse behavioral responses to trauma. should use reliable and valid multivictimization
Providing empirical evidence of the differing path- measures geared toward adolescents. The measures
ways from being a victim to becoming an offender should differentiate between family and commu-
is important information for practitioners who as- nity violence and explore the objective and subjec-
sess and develop treatment interventions for this tive effect of trauma on male and female adoles-
vulnerable and often misread population. This in- cents. In addition, based on the significance of
vestigation sought to provide this empirically based control variables in this study, future research on
evidence. For interventions to be effective in pre- this topic should explore potential intervening fac-
venting or mitigating the effect of trauma on juve- tors, such as anger and delinquent peer exposure
niles, it is essential to use multifaceted interven- on the trauma delinquency link. In addition, the
tions. Such interventions that prevent or reduce significance of control variables, such as socioeco-
juvenile delinquency are supported in the litera- nomic status (as measured by family income and
ture (Catalano, Arthur, Hawkins, & Berglund, & parental education), race and ethnicity, and family
Olson, 1998; Henggeler, Schoenwald, Borduin, structure, may be representative of additional types
Rowland, & Cunningham, 1998; Howell, 2001). of stressors linked to delinquency. Gaining a better
Yoshikawa (1994) refers to this multilevel approach understanding of any mediating or moderating in-
as ensuring the “cumulative protection” of our fluences that link trauma and delinquency can help
youths and communities. to inform prevention and intervention efforts geared
Targeted interventions should become part of toward interrupting the pathway between being a
the routine activities that occur in school, mental victim and becoming an offender.
health, juvenile justice, and community settings and
should include comprehensive trauma assessment CONCLUSION
and treatment. Comprehensive trauma assessment This study was distinct and advanced the literature
necessitates a broad assessment tool that measures on delinquency in that it adopted a comprehen-
minor to severe stressors that negatively affect youths. sive measure of trauma and compared the cumula-
Although child abuse is currently the only major tive versus differential effects of trauma on male
type of trauma that is legally mandated to be re- adolescent delinquency, while controlling for po-
ported and that may result in an intervention, other tential confounding factors. The conclusion that
types of trauma may be left unidentified and if iden- different types of trauma have a positive and sig-
tified, left untreated. Hence, social workers, teach- nificant impact on serious delinquency has broad
ers, mental health professionals, and community and implications for the assessment and treatment of
correctional staff members need trauma awareness trauma, especially among high-risk male youths.
training that not only includes the identification of Social work, mental health, and criminal justice
child maltreatment, but also other stressors that professionals need to recognize and address a wide
negatively affect adolescents, such as family and variety of minor to severe stressors that affect
community violence, academic problems, parental youths to reach the common goal of delinquency
divorce, and the loss of a loved one. Furthermore, prevention and intervention. Although some nega-
because risk factors such as anger, delinquent peer tive life events may be unavoidable, increasing
exposure, and school failure or expulsion signifi- youths’ coping skills is an important avenue to cir-
cantly increase the risk of juvenile delinquency, cumvent both the cumulative and differential im-
schools and community and correctional settings pact of trauma.

68 Social Work Volume 51, Number 1 January 2006


REFERENCES men and women: Prevalence, characteristics, and
Agnew, R. (1985). A revised strain theory of delinquency. risk factors. Child Abuse & Neglect, 14, 19–28.
Social Forces, 64, 151–167. Garmezy, N., & Masten, A. S. (1994). Chronic adversities.
Agnew, R. (1992). Foundation for a general strain theory In M. Rutter, E. Taylor, & L. Hersov (Eds.), Child
of crime and delinquency. Criminology, 30, 47–88. and adolescent psychiatry: Modern approaches (pp. 32–
Agnew, R. (2002). Experienced, vicarious, and anticipated 47). Cambridge, MA: Blackwell Science.
strain: An exploratory study on physical victimiza- Haapsalo, J., & Pokela, E. (1999). Child-rearing and child
tion and delinquency. Justice Quarterly, 19, 603–632. abuse antecedents of criminality. Aggression and
Agnew, R., & White, H. R. (1992). An empirical test of Violent Behavior, 4, 107–127.
general strain theory. Criminology, 30, 475–500. Heck, C., & Walsh, A. (2000). The effects of maltreatment
Brezina, T. (1998). Adolescent maltreatment and and family structure on minor and serious
delinquency: The question of intervening processes. delinquency. International Journal of Offender Therapy
Journal of Research in Crime and Delinquency, 35, 71– and Comparative Criminology, 44, 178–193.
99. Henggeler, S. W., Schoenwald, S. K., Borduin, C. M.,
Brown, R. A. (1999). Assessing attitudes and behaviors of Rowland, M. D., & Cunningham, P. B. (1998).
high-risk adolescents: An evaluation of the self- Multisystemic treatment of antisocial behavior in children
report method. Adolescence, 34, 25–32. and adolescents. New York: Guilford Press.
Brownfield, D. (1986). Social class and violent behavior. Hoffman, J. P., & Cerbone, F. G. (1999). Stressful life
Criminology, 24, 421–438. events and delinquency escalation in early adoles-
Capowich, G. E., Mazerolle, P., & Piquero, A. (2001). cence. Criminology, 37, 343–373.
General strain theory, situational anger, and social Hoffman, J. P., & Su, S. (1997). The conditional effects of
networks: An assessment of conditioning influences. stress on delinquency and drug use: A strain theory
Journal of Criminal Justice, 29, 445–461. assessment of sex differences. Journal of Research in
Catalano, R. F., Arthur, M.W., Hawkins, J. D., Berglund, Crime and Delinquency, 34, 46–78.
L., & Olson, J. J. (1998). Comprehensive community Howell, J. C. (2001). Juvenile justice programs and
and school-based interventions to prevent antisocial strategies. In R. Loeber & D. Farrington (Eds.),
behavior. In R. Loeber, & D. Farrington (Eds.), Child delinquents: Development, intervention, and service
Serious & violent juvenile offenders: Risk factors and needs (pp. 305–322). Thousand Oaks, CA: Sage
successful interventions (pp. 248–283). Thousand Oaks, Publications.
CA: Sage Publications. Huizinga, D., & Elliott, D. S. (1986). Reassessing the
Cichetti, D., & Carlson,V. (Eds.). (1989). Child maltreat- reliability and validity of self-report delinquency
ment: Theory and research on the causes and consequences measures. Journal of Quantitative Criminology, 2, 293–
of child abuse and neglect. New York: Cambridge 327.
University Press. Ireland, T. O., Smith, C. A., & Thornberry, T. P. (2002).
Dodge, K. A., Bates, J. E., & Pettit, G. S. (1990). Mecha- Developmental issues in the impact of child
nisms in the cycle of violence. Science, 250, 1678– maltreatment on later delinquency and drug use.
1683. Criminology, 40, 359–399.
Edleson, J. L. (1999). Children’s witnessing of adult Ireland, T., & Widom, C. S. (1994). Childhood victimiza-
domestic violence. Journal of Interpersonal Violence, 14, tion and risk for alcohol and drug arrests. Interna-
839–870. tional Journal of the Addictions, 29, 235–274.
Eitle, D., & Turner, R. J. (2002). Exposure to community Jaffe, P., Wolfe, D., Wilson, S., & Zak, I. (1986). Similarities
violence and young adult crime: The effects of in behavioral and social maladjustment among child
witnessing violence, traumatic victimization, and victims and witnesses to family violence. American
other stressful life events. Journal of Research in Crime Journal of Orthopsychiatry, 56, 142–146.
and Delinquency, 39, 214–237. Jang, S. J. (1999). Age-varying effects of family, school, and
Erwin, B. A., Newman, E., McMackin, R. A., Morrissey, peers on delinquency: A multilevel modeling test of
C., & Kaloupek, D. G. (2000). PTSD, malevolent interactional theory. Criminology, 37, 643–686.
environment, and criminality among criminally Johnson, R. M., Kotch, J. B., Catellier, D J., Winsor, J. R.,
involved male adolescents. Criminal Justice and Dufort,V., Hunter, W., & Amaya-Jackson, L. (2002).
Behavior, 27, 196–215. Adverse behavioral and emotional outcomes from
Esposito, C. L., & Clum, G. A. (2002). Social support and child abuse and witnessed violence. Child Maltreat-
problem-solving as moderators of the relationship ment, 7, 179–186.
between childhood abuse and suicidality: Applica- Kilpatrick, D. G., & Saunders, B. E. (1995). National
tions to a delinquent population. Journal of Traumatic Survey of Adolescents in the United States, 1995.
Stress, 15, 137–146. Washington, DC: National Institute of Justice and
Falshaw, L., Browne, K. D., & Hollin, C. R. (1996).Victim the Inter-university Consortium for Political and
to offender: A review. Aggression and Violent Behavior, Social Research, Institute for Social Research,
1, 389–404. University of Michigan. (ICPSR 2833)
Farrington, D. P., Loeber, R., Stouthamer-Loeber, M.,Van Kruttschnitt, C., & Dornfeld, M. (1993). Exposure to
Kammen, W. B., & Schmidt, L. (1996). Self-reported family violence: A partial explanation for initial and
delinquency and a combined delinquency serious- subsequent levels of delinquency. Criminal Behaviour
ness scale based on boys, mothers, and teachers: and Mental Health, 3, 61–75.
Concurrent and predictive validity for African- Lemmon, J. H. (1999). How child maltreatment affects
Americans and Caucasians, Criminology, 34, 493–517. dimensions of juvenile delinquency in a cohort of
Finkelhor, D., & Ormrod, R. (2000, June). Characteristics of low-income urban youths. Justice Quarterly, 16,
crimes against juveniles [Juvenile Justice Bulletin]. 357–376.
Washington, DC: U.S. Department of Justice, Office Martin, S. L., Sigda, K. B., & Kupersmidt, J. B. (1998).
of Justice Programs, Office of Juvenile Justice and Family and neighborhood violence: Predictors of
Delinquency Prevention. depressive symptomatology among incarcerated
Finkelhor, D., Hotaling, G., Lewis, I. A., & Smith, C. youth. Prison Journal, 78, 423–438.
(1990). Sexual abuse in a national survey of adult Mazerolle, P., & Maahs, J. (2000). General strain and
delinquency: An alternative examination of

Maschi / Unraveling the Link between Trauma and Male Delinquency: The Cumulative Versus Differential Risk Perspectives 69
conditioning influences. Justice Quarterly, 17, 753–
778.
Perez, D. M. (2001). Ethnic differences in property,
violent, and sex offending for abused and nonabused
adolescents. Journal of Criminal Justice, 29, 407–417.
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.
Snyder, H. N. (2003). Juvenile arrests 2001 [Juvenile Justice
Bulletin]. Washington, DC: U.S. Department of
Justice, Office of Juvenile Justice and Delinquency
Prevention.
Snyder, H. N., & Sickmund, M. (1999). Juvenile offenders
and victims: 1999 national report. Washington, DC:
U.S. Department of Justice, Office of Juvenile Justice
and Delinquency Prevention.
Stouthamer-Loeber, M., Wei, E. H., Homish, D. L., &
Loeber, R. (2002). Which family and demographic
factors are related to both maltreatment and
persistent serious juvenile delinquency? Children’s
Services: Social Policy, Research, and Practice, 5, 261–
272.
Widom, C. (1989a). Child abuse, neglect, and violent
criminal behavior. Criminology, 27, 251–272.
Widom, C. (1989b). Does violence beget violence? A
critical examination of the literature. Psychological
Bulletin, 106, 3–28.
Williams, C. L., & Uchiyama, C. (1989). Assessment of life
events during adolescence: The use of self-report
inventories. Adolescence, 24, 95–118.
Yoshikawa, H. (1994). Prevention as cumulative protec-
tion: Effects of early family support and education
on chronic delinquency and its risks. Psychological
Bulletin, 115, 28–54.
Zingraff, M. T., Leiter, J., Myers, K. A., & Johnsen, M. C.
(1993). Child maltreatment and youthful problem
behavior. Criminology, 31, 173–202.

Tina Maschi, PhD, LCSW, ACSW, is assistant professor,


Department of Social Work, Monmouth University, 400
Cedar Avenue,West Long Branch, NJ 07764; e-mail:
tmaschi@monmouth.edu. Dr. Maschi thanks Drs. Helene
Raskin White, William Fisher, Mark Schmitz, Nancy Scotto
Rosato, and the anonymous reviewers for their helpful
comments.
Original manuscript received November 17, 2003
Final revision received March 1, 2005
Accepted April 11, 2005

70 Social Work Volume 51, Number 1 January 2006

Das könnte Ihnen auch gefallen