You are on page 1of 9

Accelerat ing t he world's research.

Adverse Behavioral and Emotional


Outcomes from Child Abuse and
Witnessed Violence
L. Amaya-jackson, Wanda Hunter

Child Maltreatment

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

Parent -Yout h Discordance about Yout h-Wit nessed Violence: Associat ions wit h Trauma Sym…
T isha Wiley

Fat her Involvement and Children's Funct ioning at Age 6 Years: A Mult isit e St udy
Des Runyan

At t achment Behaviors, Depression, and Anxiet y in Nonoffending Mot hers of Child Sexual Abuse Vict i…
Linda Lewin
JohnsonMALTREATMENT
CHILD et al. / ADVERSE OUTCOMES
/ AUGUST 2002

Adverse Behavioral and Emotional Outcomes


From Child Abuse and Witnessed Violence

Renee M. Johnson
Jonathan B. Kotch
Diane J. Catellier
University of North Carolina School of Public Health
Jane R. Winsor
Rho, Inc.
Vincent Dufort
Northeast Health Care Quality Foundation
Wanda Hunter
University of North Carolina Injury Prevention Research Center
Lisa Amaya-Jackson
Duke University Medical Center

This article examines mental health outcomes of children who abuse to be associated with adverse behavioral and
have witnessed violence in their social environment and/or emotional outcomes in children, including internal-
have been physically abused. Participants (n = 167) come izing problems such as withdrawal, anxiety, and de-
from a longitudinal study on child maltreatment. Out- pression, and externalizing problems such as con-
comes—including depression, anger, and anxiety—are mea- duct disorders, aggression, and delinquency (Boney-
sured by the Child Behavior Checklist and the Trauma McCoy & Finkelhor, 1995; Malinosky-Rummell &
Symptom Checklist for Children. The authors used adjusted Hansen, 1993; Panel on Research on Child Abuse and
multivariate analyses to test the statistical significance of as- Neglect, 1993). Witnessing violence in the home
sociations. The majority of children were female (57%) and and/or neighborhood is also associated with height-
non-White (64%). One third had been physically victimized; ened levels of adverse behavioral and emotional prob-
46% had witnessed moderate-high levels of violence. Results lems (Carter, Weithhorn, & Behrman, 1999; Hughes,
confirm that children are negatively affected by victimization 1988; Osofsky, 1995; Socolar, 2000). Children’s pro-
and violence they witness in their homes and neighborhoods. longed exposure to chronic neighborhood violence is
Victimization was a significant predictor of child aggression associated with post-traumatic stress disorder, emo-
and depression; witnessed violence was found to be a signifi-
cant predictor of aggression, depression, anger, and anxiety.
Authors’ Note: This study is based on a poster presented by Vincent
Implications will be discussed. Dufort, Ph.D., at the 125th annual meeting of the American Public
Health Association in Indianapolis, IN, November 12, 1997. This
research was supported in part by Grant 1 RO1 HD39689, “Neglect
A lthough child abuse and children’s witnessing of and Adolescents: A Multi-Site Longitudinal Study,” from the
USDHHS’s National Institute of Child Health and Human Devel-
violence are highly comorbid (Edelson, 1999; opment, and Grant 90CA1677/01, “Longitudinal Studies of Child
Shipman, Rossman, & West, 1999), they are generally Abuse and Neglect,” USDHHS/ACF/ACYF/CB/Office on Child
investigated separately. Studies have shown child Abuse and Neglect. Requests for reprints should be sent to Dr.
Jonathan B. Kotch, Department of Maternal and Child Health,
CHILD MALTREATMENT, Vol. 7, No. 3, August 2002 179-186 Rosenau Hall, CB# 7445, University of North Carolina at Chapel
© 2002 Sage Publications Hill, Chapel Hill, NC 27599-7445.

179
180 Johnson et al. / ADVERSE OUTCOMES

tional distress, depression, fear, somatic complaints, in 1985-1987 from North Carolina hospitals and
memory problems, aggression, and social withdrawal health departments. Four out of every 5 infants
(Fitzpatrick & Boldizar, 1993; Osofsky, Wewers, recruited had at least one risk factor qualifying them
Hann, & Fick, 1993; Richters & Martinez, 1993). for North Carolina’s High Priority Infant Program,
Children who have witnessed domestic violence ex- such as low birth weight (< 2,500 grams), young
hibit symptoms similar to those of children who have maternal age (younger than 18 years), congenital
been abused or exposed to neighborhood violence abnormalities, birth defects, maternal impoverish-
(Carter et al., 1999; Martinez & Richters, 1993; ment, substance abuse, or other significant medical
Socolar, 2000). or social problems. The remaining 20% of the infants
Research exploring child outcomes resulting from had none of the above risk factors.
witnessing violence or being abused tends to have When the participants were 4 years of age, a subset
cross-sectional study designs and methodological of the original group of 788 was selected for contin-
weaknesses including small sample sizes and biased ued follow-up as the Southern site of the Longitudinal
sampling procedures, all of which preclude determin- Studies of Child Abuse & Neglect (LONGSCAN) con-
ing whether the relationship is causal. Particularly sortium, which is a collaborative of five longitudinal
problematic, however, is the lack of statistical control studies of child maltreatment (Runyan et al., 1998).
for confounding variables. For example, although it Of those SSS children who, by their fourth birthday,
has been theorized that child psychological outcomes had come to the attention of the State Division of
attributable to witnessing neighborhood violence Social Services for suspicion of being maltreated, 70
depend partly on levels of parental well-being and were randomly selected and then matched with two
family functioning, these variables generally have not nonreported controls (also randomly selected) based
been accounted for in analyses. These limitations are on age, socioeconomic status, race, and sex. Maltreat-
understandable given the ethical considerations that ment status was determined by a regular review of the
arise in the process of conducting family violence state’s Central Registry of Child Abuse and Neglect,
research. Nevertheless, the findings that witnessing which contains all state documentation on child mal-
violence and being abused are associated with varying treatment reports. As part of LONGSCAN, partici-
degrees of child psychopathology need to be con- pant children and/or their primary caregivers were
firmed by more rigorous scientific investigation. interviewed in person at child ages 4, 6, and 8 years.
The present study examines the extent to which The interview contained items about the caregiver,
children respond negatively to witnessing violence in family, and social domains of the participant child’s
their social environments and to victimization at the environment, as well as about the child’s own experi-
hands of caregivers. In particular, this investigation ences, behaviors, and socioemotional well-being. The
seeks to predict mental health status at age 8 from data for this study came from the age 6 and age 8 inter-
exposure variables related to reports of physical views and the Central Registry review. The Southern
abuse, witnessing domestic violence, or observing vio- site of LONGSCAN has been approved by the Insti-
lence in children’s neighborhoods prior to age 8. tutional Review Board for the Protection of Human
Internalizing and externalizing behaviors will be Research Subjects at the University of North
examined as consequences of exposure to neighbor- Carolina at Chapel Hill School of Public Health.
hood and domestic violence on one hand and physi- The research activities of the Southern site of
cal child abuse on the other. We expect both victim- LONGSCAN have been approved by the Institutional
ization and witnessing of violence to be associated Review Board at the University of North Carolina at
with internalizing and externalizing problems, even Chapel Hill School of Public Health. Ethical issues
when controlling for confounding variables. concerning the investigations are described in detail
elsewhere (Knight et al., 2000; Kotch, 2000; Runyan,
2000).
METHOD
Mental Health Outcome Variables
Sample
Outcome variables can be conceptualized as
The participants for this analysis were drawn from externalizing and internalizing problems. The only
a sample of participants in the Stress, Social Support externalizing behavior—aggression—was assessed
and Abuse & Neglect in High Risk Infants Study using the Aggressive Behavior construct of the Child
(SSS), which is focused on determining the anteced- Behavior Checklist (CBCL). The CBCL is designed to
ents of child maltreatment reports. Seven hundred obtain reports of children’s competencies and behav-
and eighty-eight mother-infant dyads were recruited ior problems, as reported by caregivers, teachers, or

CHILD MALTREATMENT / AUGUST 2002


Johnson et al. / ADVERSE OUTCOMES 181

the child. We used the parent report version. The Exposure to Violence
CBCL is the most commonly used measure of child
We determined the level of violence witnessed by
psychopathology. Evidence for content, construct,
the child through both child and caregiver reports.
and criterion-related validity is well documented
The child’s report of witnessed violence was mea-
(Achenbach, 1991; Achenbach, Edelbrock, & Howell,
sured by the “Things I’ve Seen and Heard” survey,
1987). administered at the age 8 interview (Richters & Marti-
Two internalizing behavior problems—anger and nez, 1990). Things I’ve Seen and Heard assesses expo-
anxiety—were assessed using the Anger and Anxiety sure to violence throughout a child’s lifetime. A child
subscales of the Trauma Symptom Checklist for was considered to have witnessed minimal levels of
Children (TSCC). The TSCC is a widely used and rep- violence if she or he did not endorse any of the follow-
utable measure of psychological symptomatology ing items: (a) seeing a person arrested more than
intended for use in the evaluation of children who once, (b) seeing someone beaten up more than once,
have experienced traumatic events (Briere, 1996). It (c) ever having heard grownups in the home yell at
assesses the effect of trauma through the child’s self- one another, (d) ever having seen grownups in the
report. The internal consistency of the TSCC has home hit each other, (e) ever having seen someone
been reported as high; Cronbach’s alpha =.95 (Evans, get stabbed, (f) ever having seen someone get shot,
Briere, Boggiano, & Barrett, 1994). Results of TSCC (g) ever having seen a dead body (not in the context
are congruent with those derived from similar mea- of a funeral or wake), (h) ever having seen someone
sures, including the CBCL, indicating good concur- pull a gun on someone, (i) ever having seen someone
rent validity (Lanktree & Briere, 1990, 1995). shot in the child’s own home, and ( j) ever having seen
someone pull a knife on someone. If the child
A third internalizing behavior, depression, was
endorsed any of the first four items but none of the
measured using the Depression subscales of both the items (e) through ( j), she or he was considered to
CBCL and the TSCC. We followed the instrument have witnessed a moderate level of violence. If the
developer’s recommendation for nonclinical samples child endorsed any of the last six items, she or he was
and thus used raw scores, rather than T scores, in ana- considered to have witnessed a high level of violence.
lyzing the CBCL (Achenbach, 1991). All outcome Caregiver report measures of violence witnessed by
variables were kept continuous and were measured at the child were assessed using items from the age 8
the age 8 interview. administration of the Child Life Events survey. This
project-developed survey is based on an existing sur-
Victimization
vey (Sarason, Johnson, & Siegel, 1978) and was
Child victimization status was determined using designed to track events that occurred in the past year
both Central Registry data prior to the age 8 interview that may have had an impact on the child’s psycholog-
and caregiver responses to the Conflict Tactics Scales ical well-being. A child was considered to have wit-
nessed a high level of violence if the caregiver
(CTS) at the age 6 interview (Straus, 1979). The CTS
reported that the child observed any of the following
measures the extent to which caregivers use reason-
events in the home or neighborhood environment:
ing and nonviolent discipline, verbal aggression, or
someone getting threatened with a weapon, someone
physical aggression in response to their child’s behav- being stabbed, someone being killed or murdered, or
ior. Validity for the CTS has been demonstrated in a someone getting kicked, hit, or otherwise physically
number of studies (Straus & Hamby, 1997). If a care- harmed. If the parent reported that the child was
giver’s report on the CTS did not imply physical exposed to loud arguments but not to any of the other
aggression, and if there had not been a report of child indicators, the child was considered to have been wit-
abuse, victimization was considered not present. The ness to a moderate level of violence. For cases in which
child was considered to have experienced a moderate parents did not endorse any of the above-listed indica-
level of victimization if the caregiver reported throw- tors of violence, the child was considered to have been
ing, smashing, hitting, or kicking an object, or push- exposed to a minimal level of violence.
ing, grabbing, shoving, or slapping the child as means
Control Variables
of discipline and punishment. If there was a report of
physical abuse (either with or without parental To reduce the chances of finding a spurious associ-
endorsement of physical aggression), the child was ation, as well as to show the adjusted effects of victim-
considered to have experienced a high level of ization and violence exposure on adverse outcomes,
victimization. we controlled for the following variables in our study:

CHILD MALTREATMENT / AUGUST 2002


182 Johnson et al. / ADVERSE OUTCOMES

race of caregiver (non-White, White), sex of child, Taylor, 1987). Further analyses were to be conducted
number of siblings (0, 1, 2+), caregiver level of depres- with each predictor variable that is significantly associ-
sion (continuous), age of caregiver (continuous), ated with all five outcome variables.
whether caregiver has a high school education, We also conducted MANOVA tests to identify
whether caregiver is the child’s biological mother, potentially confounding variables. Using a signifi-
presence of father figure in the home (no, yes: biolog- cance level of .10, we first assessed the interaction
ical father, yes: caregiver’s boyfriend, partner, or hus- effect of caregiver level of depression with each pre-
band), child’s health status as reported by the care- dictor variable on the outcomes. These sets of interac-
giver (fair/poor, good, or excellent), caregiver tions were hypothesized because caregiver depression
perception of social support (continuous), socioeco- has been found to be a strong effect modifier in previ-
nomic status (continuous), and the number of signifi- ous analyses involving this study population (Kotch
cant life events experienced by the child in the past et al., 1999). Next, we fit successively reduced
year (continuous). Maternal depression and social MANOVA models by discarding the least significant
support were included as control variables because control variables at each step until the remaining con-
they have emerged as strong predictors of child mal- trol variables met the criterion of p = .10.
treatment in previous analyses involving this popula- If the overall multivariate test of the effect of the
tion (Kotch et al., 1995). Maternal depression was victimization or witnessed violence variable was statis-
measured by the Depression subscale of the Brief tically significant, we went on to perform analysis of
Symptom Inventory (Derogaitis & Spencer, 1982), in variance (ANOVA) tests of the predictor variable on
which respondents rate statements about their nega- each individual outcome variable, adjusting for con-
tive feelings on a 5-point Likert-type scale. Scores for trol variables. This approach allowed us to interpret
the subscale can fall within a range of 0 (not at all the findings by identifying the specific dependent
depressed) to 30 (extremely depressed). The Social Provi- variables that contributed to the significant overall
sions Scale was used to examine the degree to which effect. Analyses were performed using SAS software
respondents’ social relationships provide various (SAS Institute, 1990).
dimensions of social support (Cutrona & Russell,
1987). Possible scores range from 4 to 96; a higher
RESULTS
score indicates a greater perception of social support.
To capture family socioeconomic status, we com- The sample for this investigation is limited to the
puted the family’s percentage of the poverty level by 167 caregiver-child pairs who completed both age 6
dividing the total household income by the federal and age 8 interviews. A comparison of the study sam-
poverty level, taking family size into account. The ple with those pairs who only completed the age 6 (n =
number of child life events was assessed using the 222) or age 8 (n = 180) interview showed that there
Child Life Events survey. This 31-item survey covers were no significant differences across groups in terms
events relating to changes in the participant child’s of the predictor variables or with regard to levels of
household composition, serious illness or injury victimization and witnessed violence.
among the child’s household members, child’s wit- Forty-four percent of the children in the sample
nessing of violence, and involvement of family mem- were male, and 64% were non-White. Only 7% were
bers with the legal system. Social support and care- reported by their caregivers to be in fair or poor
giver depression were measured at the age 8 health; all others were reported to be in good or excel-
interview. All other control variables were measured lent health. Fewer than a quarter of the participants
at the age 6 interview. (22%) were the only children living in their house-
holds; 38% had one sibling, and 40% had two or more
Analysis
siblings. Children had an average of 1½ significant life
Initially, we conducted a descriptive analysis, gen- events in the past year according to the Child Life
erating frequencies, means, and standard deviations Events survey (SD = 1.4). The mean score for care-
as appropriate. Then, to test the global effect of each givers on the Social Provisions Scale was 76, indicating
predictor variable on all outcome variables (internal- that parents in the sample perceived themselves as
izing and externalizing behavior problems), we fit having above-average levels of social support. The
multivariate analyses of variance (MANOVA) models. mean score for caregivers on the Brief Symptom
MANOVA models compare the mean values of multi- Inventory was 2.1, indicating that respondents were
ple outcome variables across different levels of the not experiencing high levels of depressive symptom-
predictor variable while maintaining a Type I error atology. Sixty-one of the caregivers reported that they
rate below the prespecified alpha level (.05) (Hand & had finished high school, and 80% reported that they

CHILD MALTREATMENT / AUGUST 2002


Johnson et al. / ADVERSE OUTCOMES 183

were the biological parents of the participant child. TABLE 1: Prevalence of Victimization and Exposure to Vio-
lence Among Participants (N = 167)
The average age of caregivers was 32 years (SD = 8.7,
range = 21-63). There were biological fathers residing N %
in 28% of the participant children’s homes, and care-
givers’ partners resided in 19% of the children’s Level of victimization
homes. The average income for families in this sam- None 117 70.9
Moderate 24 14.5
ple was 98% of the poverty level (SD = 74.1, range =
High 24 14.5
11%-417%). Child self-report of witnessing violence
More than one quarter (29%) of the children were Minimal 38 22.8
categorized as having been victimized (Table 1). Moderate 77 46.1
According to the child self-report of exposure to vio- High 52 31.1
Caregiver report of child’s witnessing violence
lence, more than three quarters of the participants
Minimal 90 53.9
had witnessed violent events. Caregivers tended to Moderate 52 31.1
underreport children’s exposure to violence when High 25 15.0
compared to the child self-report measure. According NOTE: The total N for some variables does not add to 167 due to
to caregiver report of exposure to violence, slightly missing data.
fewer than half of the children had witnessed moderate-
high levels of violence in their home or community.
In the first series of MANOVA analyses, all three
predictor variables (victimization, witnessed 7
6.65
violence–child report, and witnessed violence– 6
caregiver report) emerged as significant predictors of

Means of Child Depression


High Caregiver Depression
5
all five outcome variables (p < .05). The following con- 4.69
4
trol variables survived the covariate selection proce-
3
dure: sex of child, caregiver education, age of care- 3

Low Caregiver Depression


giver, presence of father figure in home, health of 2
2.22

child, caregiver level of social support, and number of 1

major life events for the child. Tests of the product 0


terms of caregiver depression by each of the victimiza- Minimal/Moderate High

tion and witnessing variables showed that caregiver Caregiver's Report of Witnessed Violence

depression interacted with caregiver report of child’s FIGURE 1: Effect of Caregiver Depression × Witnessed Violence
witnessed violence to affect child depression (p < .01), Interaction on Child Depression
indicating caregiver depression to be an effect modi-
fier. As no other interaction terms were significant,
their product terms were subsequently dropped from adjusted scores showed that levels of aggression and
further analyses. Surviving covariates were adjusted depression jumped sharply from no victimization to
for in ANOVA models. any victimization, with only a slight difference
The effect of the Caregiver Depression × Witnessed between moderate and severe levels of victimization.
Violence interaction term was apparent when wit- Child report of witnessed violence was significantly
nessed violence was a three-level variable. We then associated with all five outcome variables (p < .05).
collapsed the two lower levels of exposure to ease The least-square means for the child-reported vari-
interpretation. When caregivers exhibited high levels ables (depression, anger, and anxiety) increased lin-
of depression, the children’s depression scores on the early as the amount/severity of violence increased.
CBCL were linearly related to their exposure to wit- Caregiver report of witnessed violence was a signifi-
nessed violence. In contrast, when caregivers exhib- cant predictor of caregiver report of depression and
ited lower levels of depression, the child’s depression child reported anxiety, but not of anger or aggression
scores were higher when they witnessed minimal or or child’s report of depression. Levels of child depres-
moderate levels of violence than when they were wit- sion, as measured by caregiver report, increased sig-
ness to severe violence (Figure 1). nificantly from minimal to moderate levels of wit-
The results of the final multivariable regression nessed violence, declining somewhat from moderate
analyses (ANOVA) are shown in Table 2. Child victim- to severe. However, more severe anxiety was reported
ization had significant effects on caregiver report of at the lowest level of witnessed violence and the least
both aggression and depression (p < .05). The corre- severe anxiety was reported at moderate levels of wit-
sponding patterns of least-square means of the nessed violence.

CHILD MALTREATMENT / AUGUST 2002


184 Johnson et al. / ADVERSE OUTCOMES

TABLE 2: Least-Square Means for Predictors From Multivariable Regression Analyses for Each Outcome (N = 167)

Outcomes
Caregiver Report: Caregiver Report: Child Report: Child Report: Child Report:
a a b b b
Predictor Aggression Depression Depression Anger Anxiety

Victimization
None 8.56 3.49 6.28 5.56 7.54
Moderate 11.83 6.32 7.90 8.01 8.11
Severe 12.10 5.23 5.45 4.09 6.56
p .0189 .0014 .1921 .0910 .6367
Caregiver report of witnessed violence
Minimal 9.12 4.51 7.08 6.00 8.74
Moderate 12.23 5.50 5.90 5.88 5.75
Severe 11.14 5.02 6.65 5.78 7.71
p .1686 .0041 .3057 .4483 .0451
Child report of witnessed violence
Minimal 12.84 5.98 4.52 3.67 4.67
Moderate 9.59 3.79 6.50 6.01 7.23
Severe 10.09 5.26 8.61 7.98 10.29
p .0443 .0372 < .001 .0029 < .001
NOTE: Least square means estimates are Type III sum of squares and control for caregiver depression, child sex, maternal education, care-
giver age, father figure in home, child health status, caregiver social support, and significant life events.
a. Measured by the Child Behavior Checklist.
b. Measured by the Trauma Symptom Checklist for Children.

DISCUSSION moderate-high levels of violence, whereas the care-


giver report indicated that 46% of children had been
Although there has been previous research on neg- exposed to moderate-high levels of violence. This dis-
ative mental health effects subsequent to child abuse, crepancy may be due to several factors. First, the Child
children’s exposure to household violence, and chil- Life Events survey, which was administered to care-
dren’s exposure to neighborhood violence, few stud- givers, inquires about events in the past year. In con-
ies have used longitudinal data and adjusted for trast, the child self-report Things I’ve Seen and Heard
potentially confounding variables. The purpose of survey inquires about the child’s entire lifetime. Sec-
this investigation was to examine the association of ond, the highly personal nature of the questions may
experiencing physical abuse and/or witnessing vio- have resulted in underreporting by caregivers. How-
lence in the household or neighborhood environ- ever, even the findings from the more conservative
ment with mental health outcomes among children. assessment reflect an above-average level of witnessed
We build on prior research by adjusting for a strict set violence. According to caregivers’ reports, 15% of
of control variables and using data collected at multi- children had seen someone severely assaulted or
ple time points. threatened with a weapon, and 31% were exposed to
The children in our sample experienced high lev- loud arguments.
els of violence at a young age. More than one fourth Our findings confirm that children are negatively
(29%) had either come to the attention of the State affected by exposure to violence, both personal vic-
Division of Social Services as being possible victims of timization and what they see around them. When we
child abuse or had experienced physical aggression at ran multivariate statistical models that tested the
the hands of caregivers, that is, they were shoved, effect of each predictor variable (victimization, child
slapped, pushed, or grabbed. (Physical aggression to self-report of witnessed violence, caregiver report of
children by caregivers was assessed through caregiver witnessed violence) on all five mental health outcome
report.) Violence witnessed by children in the home variables (child self-report of depression, anger, and
and neighborhood was assessed through both care- anxiety; caregiver report of child’s depression and
giver and child report. Nearly half (46%) of the chil- aggression), all three were significantly associated
dren reported seeing an arrest, an assault, or loud with negative mental health outcomes. The series of
arguments. According to the child self-report mea- multivariable statistical tests, which were adjusted for
sure of witnessed violence, more than three quarters potentially confounding variables, examined the
(77.2%) of the children had been exposed to effects of victimization and witnessing violence on

CHILD MALTREATMENT / AUGUST 2002


Johnson et al. / ADVERSE OUTCOMES 185

each mental health outcome separately. Victimiza- observations supports our hypotheses that witnessing
tion was significantly associated with increased depres- and experiencing violence each contribute to adverse
sion and aggression among children. Although the psychological outcomes for young children.
caregivers’ report of witnessed violence was related to
significant increases in depression and anxiety, the
CONCLUSION
child self-report of witnessed violence was related to
significant increases in all five mental health This study of the impact of exposure to violence
outcomes. among young children is unusual in three ways. First,
As compared to children who had not experienced the study is based on a sample selected from a commu-
any victimization, those who had been physically nity, not a clinic. Second, the study is longitudinal,
abused exhibited more severe adverse behavioral and examining the consequences of victimization and of
emotional outcomes. The mean scores for aggression, witnessing violence at a point in time after the expo-
depression, anxiety, and anger among nonvictimized sures. Finally, the analyses are well controlled, taking
children were comparable to the scores of children in multiple potential confounders and interactions into
the general population. In contrast, the scores of account. The conclusion, that witnessing violence
those children who had been victimized were higher and being a victim of physical violence each is associ-
than average (Achenbach, 1991; Briere, 1996). Simi- ated with adverse behavioral and emotional outcomes
larly, those children who reported witnessing severe for children, is not surprising. The implications of the
violence also had high anxiety and depression scores findings, however, are not so obvious. Parents, care-
relative to the general population and to those who givers, and human service providers need to consider
reported witnessing a minimal or moderate amount the context of maltreatment, especially any exposure
of violence. Strangely, systematic linear increases in of children to family and neighborhood violence, to
the magnitude of negative mental health outcomes prevent aggression, anxiety, and depression in victim-
were not apparent for additional models. This is ized young children.
partly due to the fact that in some cases, even those
children who experienced the lowest degree of vio-
REFERENCES
lence had above-average levels of negative mental
health outcomes. Achenbach, T. M. (1991). Manual for the CBCL/4-18 and 1991 profile.
Burlington: University of Vermont.
Achenbach, T. M., Edelbrock, C., & Howell, C. (1987).
Empirically-based assessment of the behavioral/emotional
LIMITATIONS problems of 2-3 year old children. Journal of Abnormal Child Psy-
chology, 15, 629-650.
This analysis depends in part on maltreatment data Boney-McCoy, S., & Finkelhor, D. (1995). Psychological sequelae of
provided by a state central registry that includes all violent victimization in a national youth sample. Journal of Con-
maltreatment reports. It is likely that these reports sulting and Clinical Psychology, 70, 726-736.
Briere, J. (1996). Trauma Symptom Checklist for Children: Professional
underestimate the prevalence of maltreatment. Even manual. Odessa, FL: Psychological Assessment Resources.
though a truncated form of the CTS, without the most Carter, L. S., Weithhorn, L. A., & Behrman, R. E. (1999). Domestic
severe forms of physical violence, was used as well to violence and children: Analysis and recommendations. Future of
Children, 9(3), 4-20.
increase the likelihood of capturing all maltreated Cutrona, C. E., & Russell, D. W. (1987). The provisions of social
participants, our main predictor remains reported mal- relationships and adaptation to stress. In W. H. Jones & D. Perlman
treatment rather than maltreatment itself. (Eds.), Advances in personal relationships (Vol. 1, pp. 37-67).
Greenwich, CT: JAI.
Neither is this a study of a representative popula- Derogaitis, L. R., & Spencer, P. M. (1982). The Brief Symptom Inven-
tion. Because participants were initially recruited tory (BSI): Administration and procedures manual—I. Baltimore:
based on risk of maltreatment, the results are not Clinical Psychometric Research.
Edelson, J. L. (1999). The overlap between child maltreatment and
generalizable to all children. woman battering. Violence Against Women, 5, 134-154.
Finally, not all of the hypothesized relationships Evans, J. J., Briere, J., Boggiano, A. K., & Barrett, M. (1994, January).
were significant or in the predicted direction. Per- Reliability and validity of the Trauma Symptom Checklist for Children
in a normal sample. Poster session at the San Diego Conference
haps, for example, witnessing violence has an inhibit- on Responding to Child Maltreatment, San Diego, CA.
ing effect on aggression in young children who report Fitzpatrick, K. M., & Boldizar, J. P. (1993). The prevalence and con-
high levels of anxiety. It is possible that the parents’ sequences of exposure to violence among African American
youth. Journal of the American Academy of Child and Adolescent Psy-
reports in these cases simply do not reflect reality, or chiatry, 32, 424-430.
the instruments and our attempts to create cut-points Hand, D. J., & Taylor, C. C. (1987). Multivariate analysis of variance
along scales that are continuous may not adequately & repeated measures. London: Chapman & Hall.
Hughes, H. M. (1988). Psychological and behavioral correlates of
capture the emotional and behavioral phenomena we family violence in child witnesses and victims. American Journal of
are studying. Nevertheless, the preponderance of our Orthopsychiatry, 58, 77-90.

CHILD MALTREATMENT / AUGUST 2002


186 Johnson et al. / ADVERSE OUTCOMES

Knight, E. D., Runyan, D. K., Dubowitz, H., Brandford, C., Kotch, J., Renee M. Johnson worked as a graduate research assistant with
Litrownik, A., & Hunter, W. M. (2000). Methodological and eth- Dr. Desmond Runyan’s Longitudinal Studies of Child Abuse and
ical challenges associated with child self-report of maltreat- Neglect project (LONGSCAN) and with Dr. Jonathan Kotch’s
ment: Solutions implemented by the LONGSCAN consortium.
Southern local site of LONGSCAN. She is also an advanced gradu-
Journal of Interpersonal Violence, 15(7), 760-775.
Kotch, J. B. (2000). Ethical issues in longitudinal child maltreat- ate student in the Department of Health Behavior and Health Edu-
ment research. Journal of Interpersonal Violence, 15(7), 696-709. cation at the University of North Carolina at Chapel Hill.
Kotch, J. B., Browne, D. C., Dufort, V., Winsor, J., & Catellier, D.
(1999). Predicting child maltreatment in the first 4 years of life Jonathan B. Kotch, a board-certified specialist in both pediatrics
from characteristics assessed in the neonatal period. Child Abuse and preventive medicine, is a professor and associate chair for grad-
& Neglect, 23, 305-319. uate studies in the Department of Maternal and Child Health,
Kotch, J. B., Browne, D. C., Ringwalt, C. L., Stewart, P. W., Ruina, E., School of Public Health, University of North Carolina at Chapel
Holt, K., et al. (1995). Risk of child abuse or neglect in a cohort Hill. He is the principal investigator of the Southern local site for the
of low-income children. Child Abuse and Neglect, 19, 1115-1130. Longitudinal Studies of Child Abuse and Neglect project
Lanktree, C. B., & Briere, J. (1990, August). Early data on the Trauma
Symptom Checklist for Children (TSCC). Poster session at the
(LONGSCAN) project. In addition to child maltreatment, his spe-
annual meeting of the American Psychological Association, cific areas of research and practice relate to child health policy,
Boston. injury prevention, and health and safety of children in out-of-home
Lanktree, C. B., & Briere, J. (1995). Outcome of therapy for sexu- child care.
ally abused children: A repeated measures study. Child Abuse and
Neglect, 19, 1145-1155. Diane J. Catellier, a research assistant professor of biostatistics at
Malinosky-Rummell, R., & Hansen, D. J. (1993). Long-term conse- the University of North Carolina at Chapel Hill, provides statistical
quences of childhood physical abuse. Psychological Bulletin, 114, data management, quality assurance, and studies management ser-
68-79. vices for a number of multicenter public health and medical studies.
Martinez, P., & Richters, J. E. (1993). The NIMH community vio-
lence project, II: Children’s distress symptoms associated with
She has statistical expertise in the areas of missing data, mixed mod-
violence exposure. Psychiatry, 56, 22-35. els, categorical data analysis, and generalized estimating equations
Osofsky, J. D. (1995). The effects of exposure to violence on young procedures. Currently, she is a statistician for three NIH-sponsored
children. American Psychologist, 50, 782-788. multicenter studies: the Atherosclerosis in Communities Study, the
Osofsky, J. D., Wewers, S., Hann, D. M., & Fick, D. C. (1993). Enhancing Recovery in Coronary Heart Disease Study, and the
Chronic community violence: What is happening to our chil- Trial of Activity for Adolescent Girls.
dren? Psychiatry, 56, 36-45.
Panel on Research on Child Abuse and Neglect, Commission on Jane R. Winsor worked as a data manager and statistical pro-
Behavioral and Social Sciences and Education, National grammer for Dr. Kotch’s Southern local site for the Longitudinal
Research Council. (1993). Understanding child abuse and neglect.
Washington, DC: National Academy Press.
Studies of Child Abuse and Neglect project (LONGSCAN) project
Richters, J. E., & Martinez, P. (1990). “Things I’ve Seen and Heard”: and the Stress and Social Support project for 11 years. She has an
An Interview for Young Children About Exposure to Violence. additional 10 years of programming experience in a variety of appli-
Rockville, MD: Child and Adolescent Disorders Research cation areas. She is currently a project manager for a clinical
Branch, Division of Clinical Research, National Institute of research organization, Rho, Inc., in Chapel Hill, North Carolina.
Mental Health.
Richters, J. E., & Martinez, P. (1993). The NIMH community vio- Vincent Dufort is an epidemiologist who has worked in maternal
lence project, I: Children as victims of and witnesses to violence. and child health as a research associate, focusing on child maltreat-
Psychiatry, 56, 7-21. ment and child injuries. He is currently working as an epidemiolo-
Runyan, D. K. (2000). The ethical, legal, and methodological impli-
cations of directly asking children about abuse. Journal of Inter-
gist for a nonprofit organization whose primary goal is to help
personal Violence, 15(7), 675-681. improve the quality of health care for Medicare beneficiaries
Runyan, D. K., Curtis, P., Hunter, W. M., Black, M. M., Kotch, J. B., throughout Maine, New Hampshire, and Vermont. He continues
Bangdiwala, S., et al. (1998). LONGSCAN: A consortium for his collaboration with researchers at the Department of Maternal
longitudinal studies of maltreatment and the life course of chil- and Child Health at the University of North Carolina and is
dren. Aggression and Violent Behavior: A Review Journal, 3(3), 275- involved in studies addressing prenatal care among Medicaid bene-
285. ficiaries and assessments of blood lead levels among their children in
Sarason, J., Johnson, J., & Siegel, J. (1978). Assessing the impact of New Hampshire.
life changes: Development of the Life Experiences Survey. Jour-
nal of Consulting and Clinical Psychology, 46, 932-946.
SAS Institute. (1990). SAS/STAT user’s guide (Vol. 2, Version 6).
Wanda Hunter is a research associate professor in the Depart-
Cary, NC: Author. ment of Social Medicine and assistant director for teaching and ser-
Shipman, K. L., Rossman, B.B.R., & West, J. C. (1999). Co-occurrence vice at the University of North Carolina Injury Prevention Research
of spousal violence and child abuse: Conceptual implications. Center. She has been engaged in research involving at-risk children
Child Maltreatment, 4, 93-102. for 18 years. At the time this article was written, she was co–principal
Socolar, R.R.S. (2000). Domestic violence and children: A review. investigator for the LONGSCAN Coordinating Center.
North Carolina Medical Journal, 61(5), 279-283.
Straus, M. A. (1979). Measuring intrafamily conflict and violence: Lisa Amaya-Jackson is an assistant professor in psychiatry and
The Conflict Tactics Scales. Journal of Marriage and the Family, 41, behavioral sciences at Duke University Medical Center and director
75-88.
Straus, M. A., & Hamby, S. L. (1997). Measuring physical and psycho-
of trauma, evaluation, treatment, research, and preventive mental
logical maltreatment of children with the Conflict Tactics Scales. health services at the Center for Child and Family Health North
In G. Kaufman Kantor & J. L. Jalinski (Eds.), Out of the darkness: Carolina, which is a consortium among Duke University, the Uni-
Contemporary research perspectives of family violence (pp. 119-135). versity of North Carolina at Chapel Hill, and North Carolina Cen-
Thousand Oaks, CA: Sage. tral University.

CHILD MALTREATMENT / AUGUST 2002

You might also like