You are on page 1of 17

Violence and Victims, Volume 22, Number 3, 2007

Child Maltreatment,
Revictimization, and Violent
Behavior
Daniela Hosser, PhD
Stefan Raddatz, MA
Michael Windzio, PhD
Criminological Research Institute of Lower Saxony
Hanover, Germany

The study investigates the cumulative impact of child maltreatment and victimization in
adolescence on violent behavior in young adulthood in a nonclinical high-risk sample. The
sample consists of 1,526 incarcerated young men (14 to 24 years) who were interviewed
with standardized instruments during their prison term. Violent and nonviolent offend-
ers with and without repeated victimization experiences throughout the life cycle were
compared. Results show that child maltreatment doubles the risk for violent victimization
in adolescence. Repeated victimization experiences in adolescence heighten the risk for
later violent offending. This is the case for officially registered violence and self-reported
violent behavior. In addition, child maltreatment increased the probability of self-reported
violence as well. However, the interaction effect of victimization in childhood and vic-
timization in early adolescence counteracted the main effects. Being repeatedly victim-
ized throughout the early life cycle slightly reduced the probability of being a frequent
offender.

Keywords: child abuse; victimization; cumulative impact; prisoners

W
hile the simple black-and-white image of passive, innocent crime victims
on the one hand and active, unscrupulous, violently acting perpetrators on
the other is still promoted in public (Loeber, Kalb, & Huizinga, 2001), early
systematical analysis of perpetrator–victim interactions (Hentig, 1948; Wolfgang, 1958)
already showed that the protagonists are often perpetrator and victim in one person.
Today, a considerable amount of research documents a clear relationship between vic-
timization experiences and violent and criminal acting-out behavior (Greenwald, 2002).
Victimization and trauma exposure seem to play a central role in the development and
persistence of violence. This assumption is supported by cross-sectional and longitudinal
studies on clinical and nonclinical samples as well as by representative studies. Strong
evidence also stems from research with young convicts and prisoners showing that
young (violent) offenders experience multiple victimization and repeated trauma more
frequently than their nonoffending counterparts (McMackin, Leisen, Sattler, Krinsley,
& Riggs, 2002). However, concerning delinquent or violent behavior later in life, still

318 © 2007 Springer Publishing Company


Child Maltreatment, Revictimization, and Violent Behavior 319

little is known about the cumulative impact of the different and recurring victimization
experiences throughout the life cycle.
Up to now, most research has concentrated on the relationship between child abuse
and violence. Numerous empirical studies show that the experience of violence in child-
hood predates a higher degree of delinquency in adolescence and adulthood (Heck &
Walsh, 2000; Pfeiffer, Wetzels, & Enzmann, 1999; Smith & Thornberry, 1995; Wetzels,
Enzmann, Mecklenburg, & Pfeiffer, 2001; Widom & Maxfield, 2001). These coherencies
were detected for recorded criminal offences (Zingraff, Leiter, Myers, & Johnson, 1993) as
well as for self-reported delinquency (Smith & Thornberry, 1995). Widom’s (1989, 1992)
prospective longitudinal study led to the result that maltreated and neglected children have
a higher risk to become delinquent in adolescence. Maltreatment or neglect in childhood
increased the likelihood of being imprisoned to about 59% during adolescence and to 28%
during adulthood (Widom & Maxfield, 2001). The likelihood of committing an act of
violence was 30% higher in the group of maltreated and neglected children. Furthermore,
they were imprisoned at a younger age, committed almost twice as many offenses, and
were recidivists more often than those who were neither maltreated nor neglected by their
parents. Other longitudinal studies found similar evidence for strong developmental effects
of child abuse and neglect on violence. In the study of Kelley, Thornberry, and Smith
(1997), 70% of child victims of abuse reported engaging in violent behavior as teenagers
compared to 56% of nonabused children. Maltreated youths also gave higher self-reports
of having been involved (at any time) in each of the following delinquency categories:
minor, moderate, serious, violent, and general delinquency. Furthermore, persons abused
in childhood more often had an official record of delinquency (45% vs. 32%) (see also
Krisberg, 2005). A representative German survey conducted by Wetzels et al. (2001) indi-
cated that with increasing frequency and intensity of interfamilial violence in childhood,
rates of juvenile delinquency rose.
Thereby, child abuse and maltreatment seem also to heighten the risk of revictimization
in adolescence and adulthood. Heitmeyer and colleagues (1996) found that juveniles who
were maltreated in childhood approved of violence to a larger extent and had higher rates of
revictimization and an increased probability of acting violently in adolescence. Other stud-
ies also found evidence for the assumption that a history of child maltreatment increases the
risk of further victimization (Desai, Arias, Thompson, & Basile, 2002; Messman-Moore,
Long, & Siegfried, 2000). Reported rates of physical victimization range from 24% to 51%
for adults with childhood abuse histories (Dietrich, 2003). Furthermore the negative conse-
quences of (re)victimization seem to be more profound than those of a single victimization
in adolescence or adulthood (Arata, 2000; Classen, Field, Koopman, Nevill-Manning, &
Spiegel, 2001; Follette, Polusny, Bechtle, & Naugle, 1996). A combined prospective and
retrospective study of Becker-Lausen, Sanders, and Chinsky (1995) found significant links
between a range of child maltreatment experiences and later victimization experiences.
Persons who had experienced some form of physical or sexual child abuse had a two to four
times greater risk for adult physical, sexual, or psychological victimization (Desai et al.,
2002). In sum, a childhood trauma can be the starting point of a chain reaction of victimiza-
tion across the life cycle from childhood through adolescence into adulthood.
The repeated experience of victimization in adolescence or adulthood is regarded
as a further risk factor for violent behavior (Loeber et al., 2001). Several theories exist
about the way in which child abuse and revictimization are related to violent behavior.
Patterson, Reid, and Dishion (1992) argued that parental violence intensifies behavioral
320 Hosser et al.

problems of the child, in turn promoting violent behavior through school failure, associa-
tion with delinquent peers, and deviant behavior, such as the consumption of alcohol and
illegal drugs. This creates a social context that further increases the risk of revictimiza-
tion and violence and diminishes the chance of more adaptive developmental trajectories.
Greenwald (2002) formulated a model of a trauma-informed offense cycle that is based
on the work of Dodge (for a summary of results, see Dodge, 2003; Dodge & Schwartz,
1997) and Chemtob, Roitblat, Hamada, Carlson, and Twentyman (1988). An initial trauma
violates the victim’s sense of personal safety and the belief in the world as a safe and just
place. As a consequence, the victim shifts from an offensive to a defensive posture, which
is called “survival mode.” In such a state of high sensitivity to potential threats, minor or
even neutral stimuli could be misinterpreted as a threat. Trauma-related hypersensitivity
can promote the “misinterpretation of social cues, heightened arousal and anger, and inap-
propriate aggression in the perceived cause of self-protection” (Greenwald, 2002, p. 112).
Worse, such a hostile information processing style is self-perpetuating. The perception of
threat increases the likelihood of aggression, engendering hostility in others and thus con-
firming one’s perception of others of being dangerous. Trauma-related affect dysregulation
may also trigger reactivity to a variety of situations and stimuli perceived as thematically
related to the trauma. While acting-out behavior can provide immediate relief (anger may
cover feelings of fear and helplessness), the consequences of such a behavior foster fur-
ther alertness and mistrust, thus reinforcing reactivity and aggression. Likewise, persons
who act violently have a higher risk of becoming a victim themselves if their victim tries
to defend him- or herself (Fishman, Mesch, & Eisikovits, 2002). Unfortunately, most
research studies investigated either the impact of child abuse on violence or the influence
of victimization in adolescence. They thereby failed, however, to investigate the interplay
between different forms of victimization and violence in different developmental stages.
Furthermore, they often miss controlling for the possible confounding effect of the other
factor (Maker, Kemmelmeier, & Peterson, 2001). Thus, there is a risk to overstress the
impact of child abuse on violent behavior and to neglect the impact of more recent victim-
ization. Besides, the relationship and the cumulative impact of child abuse and revictimiza-
tion on later violent behavior has been studied primarily by examining sexual abuse and
sexually assaultive revictimization (Classen et al., 2001; Messman-Moore et al., 2000). In
this context, Banyard, Williams, and Siegel (2001) found a mediating effect of revictim-
ization in the relationship between childhood sexual abuse and psychological and social
problems in adulthood. However, the question of whether these findings can be set equal to
physical abuse in childhood and revictimization has not been examined yet. Moreover, the
role of aggressiveness in the relationship between victimization and violence is unclear.
According to theory, traumatic events can foster a hostile information processing style that
promotes aggressive reactions that in turn intensify the risk of revictimization. However,
if and to what extent aggression mediates the relationship between child maltreatment
and revictimization in adolescence is still an open empirical question. In sum, a closer
examination of the cumulative impact of violent victimization that occurs in childhood and
reoccurs in adolescence may contribute to a deeper understanding of cycles of violence
across the life span that can be useful for the development of more effective prevention and
intervention strategies and programs.
For this reason, the current study examines the cumulative impact of child maltreatment
and (re)victimization in adolescence on violent behavior in young adulthood. The sample
consists of young prisoners, a high-risk group for which research reported extremely high
rates of child abuse and victimization experiences. Violent behavior will be measured
Child Maltreatment, Revictimization, and Violent Behavior 321

by officially recorded delinquency and by self-report data to avoid an underestimation


of the impact of (re)victimization on violent behavior. In detail, the study deals with the
following interrelations of victimization and violent behavior: (a) First, it will be exam-
ined whether child maltreatment is associated with a higher risk of violent behavior in
young adulthood, thus replicating findings from previous research. (b) In addition, it is
assumed that maltreated children will report higher rates of revictimization through ado-
lescence. Thereby, the influence of aggressiveness on victimization will be examined as
well. (c) Third, it is hypothesized that repeated victimization during adolescence leads to
an increased risk for violent offending. (d) Finally, it is expected that persons who have
experienced both maltreatment in childhood and revictimization in adolescence will have
the highest risk of violent offending in young adulthood.

METHOD

Participants and Procedures


The present data was sampled as part of an ongoing longitudinal study (the Hanover
Prison Study) conducted in five youth prisons in northern Germany. All inmates meeting
the sample criteria were asked to take part in the project as soon as they were admitted to
one of the cooperating prisons. Sample criteria consisted of being male and of German
nationality, between the ages 14 and 24 and serving a prison term for the first time.
Participants were interviewed three times during their prison term. The first interview (t1)
was conducted as soon as possible after the beginning of the prison term (usually within
the first 4 weeks), the second interview (t2) took place about 8 weeks after the first one,
and a final interview (t3) was carried out shortly before release (usually within the last
4 weeks). The recruiting process was extended over several years, from January 1998 to
December 2001. Standardized oral interviews were conducted with each participant by
trained interviewers (mostly students of psychology). The average interview took no lon-
ger than 2 hours. Before the interview, participants were informed verbally that (a) their
participation in this study was voluntary, (b) they could withdraw at any time or refuse
to answer the questions, and (c) their answers would be treated confidentially. They were
asked to give written consent stating they had been informed about these arrangements.
The interviewees received €10 (approximately $10) for taking part in each interview. The
participation rate was 88.7%.

Sample
The following analysis includes data of 1,526 prisoners interviewed at least twice
(t1, t2) during their prison term. The attrition rate at t2 was at an acceptable low level
(10.8%). The mean age of the participants at t1 was 20.0 years (median = 20.0, SD = 1.9).
The average length of their prison terms was 1.86 years (self-report; median = 1.58,
SD = 1.24). Nearly 50% of the participants were sentenced because of a violent offense:
18.5% because of serious acts of violence (murder, homicide, sexual offences, serious
[armed] robbery, serious [armed] bodily harm) and 31.3% because of less severe acts
of violence (robbery without weapons, unarmed bodily harm); 34.9% were sentenced
because of theft or burglary, 5.0% were convicted of a drug offense, and 10.3% were
sentenced because of other criminal offenses, such as like driving offenses, vandalism,
or fraud.
322 Hosser et al.

Instruments
Violence. At t1, prisoners were asked about the offenses leading to their incarcera-
tion. The most serious offense was coded as the “index-offense” Inmates were classified
as violent offenders (nonviolent = 0, violent = 1) when they were convicted of one of
the following criminal offenses: murder, homicide, sexual offenses, robbery, or bodily
harm. Furthermore, prisoners were queried at t1 about their criminal activities before
incarceration regardless whether these activities had been indicated or officially recorded.
Self-reported violence was measured with four items, asking for the lifetime incidence
(“never,” “once or twice,” “approximately five times,” “approximately 10 times,” “more
than 10 times”) of different acts of violence (robbery, armed or unarmed bodily harm,
rape/sexual abuse). Persons who admitted committing violent offenses “once or twice” or
“approximately five times” were categorized as “recurrent violent offender.” If someone
committed one of these acts “approximately 10 times” or “more than 10 times,” he was
categorized as “frequently recurrent” violent offender.
Victimization. The lifetime incidence of victim experiences (robbery, threat, armed or
unarmed bodily harm, sexual offenses) was collected at t1 on the basis of gradated fre-
quencies (“never,” “once or twice,” “approximately five times,” “approximately 10 times,”
“more than 10 times”). A three-value variable “victim experience” was computed: “no vic-
tim experience,” “few victim experiences” (“once or twice” or “approximately five times”
victim of one of the mentioned offenses), and “frequent victim experiences” (“approxi-
mately 10 times” or “more than 10 times” victim of one of the crimes mentioned).
Child Maltreatment. Maltreatment by parents before the age of 10 was raised at t2 with
an shortened German version of the Conflict Tactic Scales (Straus, 1979; see also Pfeiffer
et al., 1999). Six different forms of behavior are rated on a four-step scale according to
their frequency of occurrence: “never,” “rarely,” “sometimes,” “frequently”: (1) “threw an
object at me,” (2) “pushed or tackled me hard,” (3) “hit me,” (4) “beat me with an object,”
(5) “battered me or beat me up,” and (6) “punched or kicked me.” With this information, an
index of parental violence was composed that combined frequency and severity of parental
violence. Both factors have proved to be equally important for the prediction of the long-
term consequences of child abuse (Pfeiffer et al., 1999). The index consists of the values
“never” (experienced none of these treatments), “light corporal punishment” (treatments 1
to 3 maximally “sometimes” or 4 at least rarely but never 5 or 6), “serious corporal punish-
ment” (treatments 1 to 3 frequently, 4 at least rarely), or “maltreatment” (behavior 5 or 6).
Early Aggressiveness. The scale “antisocial personality disorder” from the screening
questionnaire for the structured interview for axis II of the Diagnostic and Statistical
Manual of Mental Disorders (3rd ed., revised) (SKID II) behavioral disorders was used
at t2. The dichotomous items assessing aggressive behavior (“started physical conflicts,”
“used weapons in fights,” “forced somebody to get in sexual contact,” “intentionally hurt
somebody” [except in a fight], “mugged and robbed somebody,” “hurt an animal intention-
ally”) before the 14th birthday were added up to a mean score for “early aggression.”
Drug and Alcohol Addiction (Before Imprisonment). Persons who in the first inter-
view labeled themselves as being addicted to drugs or who stated using heroin or cocaine
“frequently” or “regularly” were categorized as drug addicts. Persons who appraised
themselves to be addicted to alcohol or admitted drinking alcohol several times a day were
classified as being addicted to alcohol.
Control Variables. To control the influence of the social context measured at t1 on
the development of violent behavior, the following variables were used in the regression
Child Maltreatment, Revictimization, and Violent Behavior 323

analysis (Table 1): Educational level (special school for low achievers, lower secondary
school, secondary/grammar school), family structure (children reared only by a single par-
ent), living in a children’s home (in Germany, children have to stay in such state-controlled
institutions if they do not have anyone else who cares for them adequately. Persons living
in a children’s home for at least a few years during childhood were coded as 1) social status
of the family (according to the “standard international socio-economic index of occupa-
tional status” from Ganzeboom, de Graaf, Treiman, & de Leeuw, 1992), membership in
a delinquent peer group (if the prisoner reported a membership in a group of delinquent
peers before his imprisonment, he was coded as with 1).

RESULTS

Victimization in Childhood and Adolescence


In line with previous research, an above-average number of inmates stated that they were
beaten and maltreated by their parents before the age of 10 (see Table 1). Overall, 23.9%

TABLE 1. Descriptive Statistics


Mean SD Minimum Maximum
Age at imprisonment 20.02 1.89 15.18 26.89
Parental violence
Never 0.266 — 0 1
Light 0.260 — 0 1
Serious 0.238 — 0 1
Maltreatment 0.234 — 0 1
Victimization in adolescence
None 0.138 — 0 1
Few 0.538 — 0 1
Frequent 0.323 — 0 1
Interaction: frequency of victimization
× maltreatment 0.108 — 0 1
Early aggression 1.444 1.28 0 6
Delinquent peers 0.416 — 0 1
Social status 35.272 11.46 16 88
Single parent 0.248 — 0 1
Graduation from school
Special school for low achievers 0.078 — 0 1
Lower secondary school 0.307 — 0 1
Secondary/grammar school 0.086 — 0 1
Lived in a children’s home 0.306 — 0 1
Drug addicted 0.248 — 0 1
Addicted to alcohol 0.136 — 0 1
Self-reported violence
None 0.154 — 0 1
Few 0.495 — 0 1
Frequent 0.349 — 0 1
Violent index offense (1 = yes) 0.498 — 0 1
324 Hosser et al.

said they were seriously punished, and 23.4% reported maltreatment. In contrast, using
the same measures as the study at hand, a representative German school survey (Wetzels
et al., 2001, p. 231) concerning 14- to 17-year-old pupils documented data for a nonde-
linquent group of youths (serious physical punishment: 19.5%; maltreatment: 10.4%). A
comparison of the two studies showed that twice as many of the inmates were confronted
with maltreatment in their childhood.
Violence in childhood was positively correlated with the likelihood of victimization in
adolescence (τc = .14; p < .01). According to self-reports, more than half the interviewees
were at least once a victim of violence in adolescence, and nearly a third were victims 10
times or more. Figure 1 demonstrates that the group of inmates who were maltreated in
childhood with 33.5% also shows the highest rate of frequent victimization in adolescence.
In the group of inmates who had never been corporally punished 20.3% were frequently
victimized in adolescence (χ2[6] = 50.75, p < .001). The data also showed that parental
violence was associated with early aggressiveness of the child (τc = .12; p < .01). According
to theory, aggressiveness in childhood is often associated with a deviant lifestyle in early
and late adolescence that in turn heightens the risk for victimization.
Ordinal logistic regression gives information about the dependency of an ordinal scale
variable on several explanatory variables (Table 2). Therefore, it was used in the present
study to determine the influence of child abuse on the frequency of victimization (none,
few, frequent) in adolescence and adulthood. Besides parental violence, other factors
were included in the analysis to predict victimization in adolescence: educational level,
family structure, living in a children’s home, social status of the family, membership in a
delinquent peer group, early aggressiveness, early deviant behavior, and drug and alcohol
addiction. The regression coefficient b gives information about the direction of the asso-
ciation between independent and dependent variables, as it does in linear regression.
Ordinal logistic regression analysis showed that serious parental violence and maltreat-
ment heightened the risk for victimization in adolescence and early adulthood. The risk of

Figure 1. Parental violence and victimization in adolescence.


Child Maltreatment, Revictimization, and Violent Behavior 325

TABLE 2. Ordinal Logistic Regression: Effects of Maltreatment and Further Risk


Factors on Victimization (Incidence Classes)
Victimization
b exp(b)
Age at imprisonment –0.025 0.975
Parental violence
Never Reference Group
Light 0.245 1.278
Serious 0.305* 1.356*
Maltreatment 0.847*** 2.333***
Early aggression 0.242*** 1.274***
Delinquent peers 0.068 1.071
Social status 0.013** 1.014**
Graduation from school
School dropout Reference Group
Special school for low 0.023 1.023
achievers
Lower secondary school –0.041 0.959
Secondary or grammar school 0.183 1.201
Lived in a children’s home 0.124 1.131
Single parent 0.049 1.050
τ1 –1.077 —
τ2 1.600 —
LR χ²(df) 104.73(12)
Pseudo-R2 (Nagelkerke) 0.077
Note. N = 1,526. 1 = no, 2 = few, 3 = frequent. Brant test of proportionality of odds
across response categories: LR χ2(df) = 11.26(12); p = .507.
* p < .05. ** p < .01. *** p < .001.

victimization in adolescence was 2.33 times higher for children who experienced maltreat-
ment as for persons who did not make such an experience. Light parental violence did not
increase the risk of victimization. Persons who had already shown aggressive behavior
in early childhood had an increased risk of victimization. Somewhat contradictorily, the
risk of being victimized was also increased by the social status of the family, whereby
it is necessary to consider that the social status of the entire sample was below average
(Enzmann & Greve, 2001). Overall, the rate of explained variance of the regression model
was around 8%, which refers to further important explanatory variables that have not been
considered here.
In ordinal logistic regression, conditional effect plots can visualize the effects of indepen-
dent variables on the probability of each category of the dependent variable. Figure 2 shows
predicted probabilities for each incidence class of victimization in dependence on early
aggression and maltreatment. All control variables were held constant at their mean value.
Figure 2 clarifies two circumstances. For those who suffered from maltreatment,
the probability of frequent victimizations is around 20 percentage points higher than
for juveniles having not being maltreated (short-dashed lines, marked with triangles).
Furthermore, the probability of frequent victimization rose with the magnitude of early
326 Hosser et al.

Figure 2. Effects of early aggression score and parental maltreatment on the probability of
victimization category: results from ordinal logistic regression.

aggressiveness. The probability for adolescents who had not been maltreated in childhood
and who were not showing early aggressive behavior was 18.6%. Highly aggressive but
not maltreated children had a more than doubled victimization probability (49.6%). The
risk of victimization rose up to 69.6% if these highly aggressive children additionally were
seriously maltreated.

Victimization and Violence


As shown in Table 1 and according to the index offense, 49.8% of the inmates could be
classified as violent offenders. Because of self-report data, 34.9% of the inmates were
classified as “frequently recurrent violent offenders” and 49.5% as “occasionally vio-
lent offenders,” adding up to a total of 74.4% of violent offenses. Thus, the discrepancy
between official recorded (49.8%) and self-reported violence was 34.6 percentage points.
Parental violence was associated with later violent behavior (Figure 3). The portion of
frequently recurrent violent offenders was 22.3% in the group of those who never expe-
rienced parental violence and 31.1% in the group of maltreated inmates (χ2[6] = 32.62;
p < .001; τc = .10, p < .01; Figure 3).
Next, the cumulative impact of parental violence and victimization experiences in
predicting violent actions was examined. The influence of victimization on self-reported
(unrecorded) violent delinquency was analyzed as well as the influence of victimization
in relation to the index offense (act of violence: yes/no). Concerning the index offense,
there was a chronological sequence of variables: victimization happened before the index
offense.
As a dependent variable, self-reported violence consisted of three categories. For statisti-
cal reasons, the multinomial logistic regression model was applied instead of the ordinal
model.1 In the multinomial model, effects of explanatory variables reflect changes in log
Child Maltreatment, Revictimization, and Violent Behavior 327

Figure 3. Parental violence and violent behavior.

odds or odds compared with the omitted category of the dependent variable. Hence, inter-
preting them in terms of increasing or decreasing probabilities can be misleading (Long
1997). Following from this, Table 3 shows the marginal effects of each explanatory variable
if all control variables are held constant at their mean values. Since in our model marginal
effects closely resembled discrete unit effects, changes in the probabilities of each category
of self-reported violence can be read directly from differences in percentage points. The
multinominal model showed that parental violence under simultaneous consideration of
victimization in adolescence predicted later violence. The main effect of being maltreated
during childhood increased the probability of frequent violent behavior in young adulthood
by 15.9 percentage points. This result persisted after controlling for the influence of victim-
ization experiences in adolescence, early aggressiveness, and delinquent peers. Additionally,
persons experiencing frequent victimization in adolescence had a 32.3-percentage-point
higher risk of belonging to the group of frequent violent offenders than those without
(Table 3). In line with this result, we found a positive effect on recorded violence. However,
the magnitude of effects was substantially higher for frequent unrecorded violence. In
Table 3, according to the marginal effect, the probability of recorded violence was 11.2 per-
centage points higher for frequently victimized juveniles than for nonvictimized persons.
With respect to the cumulative impact of childhood abuse and revictimization, the result
gives insight into a rather differentiated pattern: according to the interaction term “frequent
victimization × maltreatment,” being both frequently victimized in early adolescence and
maltreated during childhood decreased the probability of frequently committing violent
offenses by 17.0 percentage points. At the same time, it increased the probability of having
committed few violent offenses by 13.9 percentage points. Hence, juveniles who cumu-
late victimization experiences over their life course slightly tend to be casual offenders
rather than frequent offenders. Nevertheless, compared with the reference group (neither
parental violence nor adolescent victimization), maltreated and frequently victimized
juveniles still had a 31.2-percentage-point higher probability of being frequent offenders
(0.159 + 0.323 – 0.170 = 0.312).
328 Hosser et al.

TABLE 3. Multinomial and Binary Logistic Regression: Marginal Effects of


Parental Violence and Further Risk Factors on the Probability of Self-Reported
Violence and Index Offense
Violent Index
Self-Reported Violence Offense
Multinomial Logit Binary Logit
Marginal Effects Marginal Effects
None Few Frequent 0 = No 1 = Yes
Age at imprisonment 0.001 0.002 – 0.003 – 0.013
Parental violence
Never Reference Group Reference Group
Light – 0.024 – 0.032 0.057 0.060
Serious – 0.004 0.001 0.002 – 0.007
Maltreatment – 0.033 – 0.125** 0.159** 0.016
Adolescent victimization
Victimization: none Reference Group Reference Group
Victimization: few – 0.071*** 0.071 – 0.000 0.027
Victimization: frequent – 0.121*** – 0.201*** 0.323*** 0.112*
Frequent victimization
× maltreatment 0.030 0.139* – 0.170*** – 0.076
Early aggression –0.078*** – 0.039** 0.118*** 0.073***
Delinquent peers –0.060*** – 0.055* 0.116*** 0.064*
Social status 0.000 – 0.000 – 0.000 – 0.000
Graduation from school
School dropout Reference Group Reference Group
Special school for low 0.005 0.039 – 0.044 0.015
achievers
Lower secondary school 0.006 0.001 – 0.008 0.007
Secondary/grammar – 0.007 0.031 – 0.024 – 0.021
school
Lived in a children’s home 0.030 – 0.013 – 0.017 – 0.056
Single parent – 0.009 0.009 0.000 0.033
Drug addicted 0.032 – 0.006 – 0.026 – 0.089**
Addicted to alcohol – 0.034 – 0.068 0.102* 0.172***
LR χ²(df) 345.27(28) 122.54(17)
Pseudo-R2 (Nagelkerke) 0.307 0.103
Note. N = 1,526. dy/dx is for change of dummy variable from 0 to 1.
*p < .05. **p < .01. ***p < .001.

The probability of frequent violent offending increased about 11.8 percentage points
along with every further point on the scale of early aggressiveness. Belonging to a deviant
peer group raised the probability of frequent violent offending by 11.6 percentage points—
all other variables held constant at their mean value. Addiction to alcohol heightened a
person’s tendency to show violent behavior frequently (+10.2 percentage points) as well.
Child Maltreatment, Revictimization, and Violent Behavior 329

The probability of recorded violence (“violent index offense,” 0 = no, 1 = yes) has
been estimated by using binary logistic regression. Again, marginal effects were reported,
and overall the model showed results that were consistent with the estimation of effects
on self-reported violence in the multinomial model, even if the interaction effect was not
significant. In contrast to self-reported violence, the index offense was also predicted by
illegal drug addiction, although the absolute value of the effect was far lower than the
effect of alcohol.

DISCUSSION AND CONCLUSIONS

The aim of the present study was to examine if child maltreatment is associated with a
higher risk of (re)victimization in adolescence and if, in addition, both child maltreatment
and violent revictimization increase the risk for later violent offending.
In line with previous research, young incarcerated offenders were victims of parental
violence in childhood with above-average frequency. More than 20% of the interviewees
reported maltreatment by their parents during childhood, which is twice as high as the
rate of the general population. Additionally, around 24% said that as a child they had
received serious physical punishment. Confirming our first research question, 46% of the
maltreated children became recorded violent offenders later on, which is a significantly
higher than the proportion of violent offenders in the group not maltreated. However,
nearly 30% of those who were not confronted with violence in their childhood committed
acts of violence as well.
If the influence of (re)victimization in adolescence is controlled for, maltreated
children still have a 16% higher probability to become frequent violent offenders,
according to self-report data. Albeit under control of victimization in adolescence,
maltreatment has no significant effect on recorded violence. Thus, our study replicates
findings from previous research where a mediating effect of sexual revictimization
was found (Banyard et al., 2001). Consequently, the impact of child abuse on recorded
violence can be overestimated if the impact of later victimization or trauma is not
controlled for.
Those seriously punished or maltreated in childhood also became victims of violence
in adolescence more often, confirming our second assumption. This effect remained even
after other potential influencing factors had been controlled for in the analysis. However,
the number of inmates who were not maltreated in childhood and yet had experienced
frequent victimization is large enough (20.3%) to encourage further research to look
for other factors leading to victimization. Furthermore, child maltreatment seems to be
closely related to symptoms of early aggressiveness in childhood. According to the idea
of a trauma-induced offense cycle (Greenwald, 2002), one might assume that maltreat-
ment promotes a (stable) affective and cognitive response pattern that is characterized by
a sense of danger, anger, helplessness, and a hostile attribution bias. In situations trigger-
ing trauma-related affects, aggressive acting-out behavior is often shown, heightening the
risk of violent (re)actions and revictimization and further increasing sensitivity to trigger
situations. In line with such assumptions, maltreated children in our sample who showed
aggressive behavior at an early age had a much higher probability of revictimization in
adolescence than maltreated children without early aggressiveness. Delinquent peers do
not increase the risk of revictimization. However, in a cross-sectional study, the causal
structure of the variables remains unclear.
330 Hosser et al.

Violent victimization in adolescence and later violent behavior seem to correspond


closely to each other. Frequent violent victimization significantly contributes to the
amount of officially recorded as well as self-reported violent offending whereby the influ-
ence on the last one is higher. A deviant surrounding (in this case delinquent peers) and
excessive consumption of alcohol intensify this vicious circle. Thus, our third research
question was also confirmed.
Moreover, we found a significant impact of cumulative victimization over the life
course, in contrast to our assumption. Whereas the main effects of maltreatment during
childhood and frequent victimization during adolescence increased the probability of
being a frequent violent offender, the interaction term between maltreatment and frequent
victimization decreased this probability. It increased the probability of being a casual
offender instead. Hence, being maltreated and frequently victimized slightly counteracts
the main effects of maltreatment and frequent victimization on being a frequent violent
offender. As a possible explanation for this somewhat contradictory result, one might
assume that for this group of persons, violence is more an expression of reactive aggres-
sion than of proactive aggression. Possibly, they adhere to or are pressed into a cycle of
victimization, and their “few” violent offenses result mainly from a continual involvement
in violent conflicts. Further research is needed to examine this in more detail, studying the
sort of violent offenses and the criminal context in which violence occurs. Presumably,
this group of maltreated and multiply revictimized persons is of special interest for inter-
vention and offender treatment because here the intrinsic motivation to change one’s own
behavior and/or social context could be expected to be especially high.
It is self-evident that these results apply only to a group of young prison inmates and
thus cannot be generalized for the general population. Further research is needed to con-
firm the findings with female prisoner samples as well as with larger community samples.
Limitations of the study result from the fact that victimization experiences in childhood
and adolescence were asked for retrospectively and that only standardized self-report data
were used. Because of this, it is conceivable that (violent) offenders exaggerated their
victimization experiences to establish some understanding for their criminal behavior. On
the other hand, to admit one’s own weakness and helplessness contrasts sharply with the
norms established by the subculture in youth prison. Rather, considering prison subculture,
it also seems plausible that social desirability leads to an underreporting of the unrecorded
violent offending. Nevertheless, the effects of victimization on later violence are very con-
servative estimations with respect to officially recorded offenses. Future research should
incorporate outcome measures that go beyond delinquency and that also examine different
aspects of mental health symptoms. Moreover, the influence of various forms of traumas
on different outcome variables should be studied in more detail, thereby using elaborated
instruments to assess posttraumatic stress.
Although results should be interpreted with caution, they underline the importance of
programs focusing on early prevention of violence. Prevention should address both interfa-
milial violence and the school context. Especially regarding persons whose biographies are
characterized by altering experiences of violence, either as a victim or as a perpetrator, the
creation of nonviolent periods, areas, and contexts seems necessary and sometimes even
vital to overcome internalized behavioral patterns. Prevention programs against violence
should address the problem of victim–perpetrator change. As victims and perpetrators
of violence are often congruent, especially in a certain social context and a specific age-
group, it is important to address victimization experiences in treatment approaches for vio-
lent offenders (McMackin et al., 2002). Interventions should consider multiple experiences
Child Maltreatment, Revictimization, and Violent Behavior 331

of traumas throughout the life cycle and their potential impact on emotional, cognitive, and
behavioral reactions. This may lead to decreased avoidance or defense reactions regarding
one’s own violent behavior. The teaching of problem-solving strategies and the widening
of social support networks may simultaneously fulfill a protective function with regard to
victimization and violent behavior.

NOTE

1. Ordinal logistic regression models are based on the assumption of proportional odds. This
means that the effects of each variable on the odds of entering the subsequent response category are
constant at each threshold. According to the Brant-Test (Long, 1997, p. 143), which has a χ² value
(df) of 29.36(16) at p = .031, this assumption is violated. Therefore, determinants of self-reported
delinquency were estimated in a multinominal logistic regression model.

REFERENCES

Arata, C. M. (2000). From child victim to adult victim: A model for predicting sexual revictimiza-
tion. Child Maltreatment, 5, 28–38.
Banyard, V. L., Williams, L. M., & Siegel, J. A. (2001). The long-term mental health consequences
of child sexual abuse: An explanatory study of the impact of multiple traumas in a sample of
women. Journal of Traumatic Stress, 14, 697–715.
Becker-Lausen, E., Sanders, B., & Chinsky, J. M. (1995). Mediation of abusive childhood
experiences: Depression, dissociation, and negative life outcomes. American Journal of
Orthopsychiatry, 65, 560–573.
Chemtob, C. M., Roitblat, H. L., Hamada, R. S., Carlson, J., & Twentyman, C. (1988). A cognitive
action theory of posttraumatic stress disorder. Journal of Anxiety Disorders, 2, 253–275.
Classen, C., Field, N. P., Koopman, C., Nevill-Manning, K., & Spiegel, D. (2001). Interpersonal
problems and their relationship to sexual revictimization among women sexually abused in
childhood. Journal of Interpersonal Violence, 16, 495–509.
Desai, S., Arias, I., Thompson, M. P., & Basile, K. C. (2002). Childhood victimization and subse-
quent adult revictimization assessed in a nationally representative sample of women and men.
Violence and Victims, 17, 639–653.
Dietrich, A. M. (2003). Posttraumatic stress disorder and associated features as predictors of revic-
timization and perpetration with samples of adults abused during childhood. Retrieved June, 6,
2006, from http://www3.telus.net/trauma/Dissertation.pdf
Dodge, K. A. (2003). Do social information-processing patterns mediate aggressive behavior? In B.
B. Lahey, T. E. Moffitt, & A. Caspi (Eds.), Causes of conduct disorder and juvenile delinquency
(pp. 254–274). New York: Guilford Press.
Dodge, K. A., & Schwartz, D. (1997). Social information processing mechanisms in aggressive
behavior. In D. M. Stoff, J. Breiling, & J. D. Maser (Eds.), Handbook of antisocial behavior
(pp. 171–180). New York: Wiley.
Enzmann, D., & Greve, W. (2001). Strafhaft für Jugendliche. Soziale und individuelle
Bedingungen von Delinquenz und Sanktionierung [Imprisonment of juveniles. Social and
personal conditions of delinquency and punishment]. In M. Bereswill & W. Greve (Eds.),
Forschungsthema Strafvollzug [Research theme: Prison] (pp. 109–145). Baden-Baden:
Nomos.
Fishman, G., Mesch, G. S., & Eisikovits, Z. (2002). Variables affecting adolescent victimization:
Findings from a National Youth Survey. Western Criminology Review, 3, 1–20. Available at
http://wcr.sonoma.edu/v3n2/fishman.html
332 Hosser et al.

Follette, V. M., Polusny, M. A., Bechtle, A. E., & Naugle, A. E. (1996). Cumulative trauma: The
impact of child sexual abuse, adult sexual assault, and spouse abuse. Journal of Traumatic
Stress, 9, 25–35.
Ganzeboom, H. B. G., de Graaf, P. M., Treiman, D. J., & de Leeuw, J. (1992). A standard interna-
tional socio-economic index of occupational status. Social Science Research, 21, 1–56.
Greenwald, R. (Ed.). (2002). Trauma and juvenile delinquency: Theory, research, and intervention.
Binghamton, NY: The Haworth Maltreatment & Trauma Press.
Heck, C., & Walsh, A. (2000). The effects of maltreatment and family structure on minor and seri-
ous delinquency. International Journal of Offender Therapy and Comparative Criminology,
44, 178–193.
Heitmeyer, W., Collmann, B., Conrads, J., Matuscheck, I., Kraul, D., Kühnel, W., et al. (1996).
Gewalt: Schattenseiten der Individualisierung bei Jugendlichen aus unterschiedlichen Milieus.
[Violence: The dark side of individualization in juveniles from different social milieus]. 2. Aufl.
[2nd ed.] Munich: Juventa.
Hentig, H. (1948). The criminal and his victim. New Haven, CT: Yale University Press.
Kelley, B. T., Thornberry, T. P., & Smith, C. A. (1997). In the wake of childhood maltreatment
OJJDP Juvenile Justice Bulletin. Washington, DC: Office of Juvenile Justice and Delinquency
Prevention, U.S. Department of Justice.
Krisberg, B. (2005). Juvenile justice: Redeeming our children. Thousand Oaks, CA: Sage.
Loeber, R., Kalb, L., & Huizinga, D. (2001). Juvenile delinquency and serious injury victimization.
Juvenile Justice Bulletin, 8, 1–8.
Long, J. S. (1997). Regression models for categorical and limited dependent variables. London: Sage.
Maker, A. H., Kemmelmeier, M., & Peterson, C. (2001). Child sexual abuse, peer sexual abuse, and
sexual assault in adulthood: A multi-risk model of revictimization. Journal of Traumatic Stress,
14, 351–368.
McMackin, R., Leisen, M. B., Sattler, L., Krinsley, K., & Riggs, D. S. (2002). Preliminary develop-
ment of trauma-focused treatment groups for incarcerated juvenile offenders. In R. Greenwald
(Ed.), Trauma and juvenile delinquency: Theory, research, and intervention (pp. 175–199).
Binghamton, NY: The Haworth Maltreatment & Trauma Press.
Messman-Moore, T. L., Long, P. J., & Siegfried, N. J. (2000). The revictimization of child sexual
abuse survivors: An examination of the adjustment of college women with sexual abuse, adult
sexual assault, and adult physical abuse. Child Maltreatment, 5, 18–27.
Patterson, G. R., Reid, J. B., & Dishion, T. J. (1992). Antisocial boys. Eugene, OR: Castalia.
Pfeiffer, C., Wetzels, P., & Enzmann, D. (1999). Innerfamiliäre Gewalt gegen Kinder und
Jugendliche und ihre Auswirkungen [Consequences of interfamilial violence against children
and juveniles] (Forschungsberichte No. 80). Hannover: Kriminologisches Forschungsinstitut
Niedersachsen.
Smith, C., & Thornberry, T. P. (1995). The relationship between childhood maltreatment and adoles-
cent involvement in delinquency. Criminology, 33, 451–479.
Straus, M. A. (1979). Measuring intrafamily conflict and violence: The Conflict Tactic (CT) Scales.
Journal of Marriage and the Family, 41, 75–88.
Wetzels, P., Enzmann, D., Mecklenburg, E., & Pfeiffer, C. (2001). Jugend und Gewalt. Baden-
Baden: Nomos.
Widom, C. S. (1989). Child abuse, neglect, and violent criminal behavior. Criminology, 27, 251–271.
Widom, C. S. (1992, October). The cycle of violence (Research in Brief). Washington, DC: National
Institute of Justice.
Widom, C. S., & Maxfield, M. G. (2001, February). An update on the “cycle of violence” (Research
in Brief). Washington, DC: National Institute of Justice.
Wolfgang, M. E. (1958). Patterns in criminal homicide. Philadelphia: University of Pennsylvania
Press.
Zingraff, M. T., Leiter, J., Myers, K. A., & Johnson, M. A. (1993). Child maltreatment and youthful
problem behaviour. Criminology, 31, 173–202.
Child Maltreatment, Revictimization, and Violent Behavior 333

Acknowledgments. The study is part of a wider research project “Developmental Consequences


of Incarceration” supported by a grant of the German Research Foundation (GR 1386). We thank
Christiane Bosold, Felicitas Sassnick, and Werner Greve for their assistance in conducting the study.
Grit Risse we would like to thank for her professional assistance in shaping the English version of
this article.

Correspondence regarding this article should be directed to Daniela Hosser, PhD, Kriminologisches
Forschungsinstitut Niedersachsen, Lützerodestr. 9, D-30161 Hannover, Germany. E-mail:
hosser@kfn.uni-hannover.de
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

You might also like