BF00985658

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Journal of Family Violence, Vol. 4, No.

1, 1989

The Effects of Observing Conjugal Violence on


Children: A Review and Analysis of
Research Methodology
John W. Fantuzzo 1,2 and Carol Ummel Lindquist

This review specifically addresses the effects on children of observing conju-


gal violence. A research of six major computer databases revealed 1764 refer-
ences on family violence o f which 29 articles dealt directly with the effects
on children. These articles were judged by independent raters using a 56-item
rating system designed to analyze methodological and assessment issues of
empirical investigations. This analysis summarized these research efforts and
indicated the need for more careful consideration of definitional, subject
variable, familial stress, and reactive vs long-term effect issues. Addition-
ally, specific proposals for future research were outlined," suggestions related
to these issues, and some new directions for investigation were provided.
KEY WORDS: Conjt~galviolence; observation of family violence; research methodology.

INTRODUCTION

Spouse abuse is a provocative social problem that deserves serious at-


tention from mental health professionals. Over the past 15 years, there has
been a dramatic increase in public awareness of this serious and neglected
problem. A representative national U.S. sample of over 1000 women sur-
veyed by Straus et al. (1980) and a survey of over 1000 women in Kentucky
(Schulman, 1979) revealed that approximately 20-30% of women reported
that physical violence occurred at some time in their marriage. Additional-
ly, their findings indicated that repeated physical abuse takes place in about

ICalifornia State University, Fullerton, California 92631.


zCurrently a faculty member of the Graduate School of Education at the University of
Pennsylvania.
77
0885-7482/89/0300-0077506.00/0 9 1989 Plenum Publishing Corporation
78 Fantuzzo and Lindquist

10% of couples and that physical abuse is more common among younger
couples in the 20-30 age range who are more likely to have small children.
Most of the efforts in this area have been directed toward designing
and mobilizing various community resources for these women (e.g., legal
services and shelters). More recently, the focus has expanded to include at-
tention to the needs of the unintended v i c t i m s - t h e children who witness the
physical abuse between their parents. Social workers, shelter staff, and
researchers first called attention to the symptoms that they observed in chil-
dren of violent homes. They reported symptoms of internalizing problems
such as enuresis, nightmares, depression, and psychosomatic complaints (e.g.,
headaches, ulcers, asthma). They also reported symptoms of externalizing
problems, such as temper tantrums, aggressive disruptive behavior with si-
blings and schoolmates, delinquency, and violence (Hilberman and Munson,
1977; Levine, 1975; Rounsaville and Weissman, 1977).
Another troubling aspect of these reported behaviors was that several
studies suggested that the violence observed at home in childhood was repeat-
ed later in life (Herrenkohl et al., 1983). Straus et al. (1980) estimate that
sons who witness their father's violence have a 1000% greater likelihood of
violence against their own partners than sons who do not. Studies of adult
spouse abusers documented that they were more likely than their nonviolent
cohort to have observed violence between their parents in their family of origin
(Rosenbaum and O'Leary, 1981; Telch and Lindquist, 1984). These studies
suggest that the effects of observing spouse abuse may persist into adult-
hood and perpetuate an intergenerational cycle of violence.
Emery's (1982) review of this literature helped clarify some distinction
and pointed out some salient variables. He made an important conceptual
differentiation between open interparental conflict and marital dissatisfac-
tion. His review suggested that open marital conflict (i.e., amount and type)
is a primary determinant of childhood problems in discordant families. He
also identified parental pathology and parental buffering, age, and sex as
variable critical to understanding these deleterious effects.
Despite the recognition that witnessing conjugal violence is harmful to
children, we know very little about (a) the unique effects of observing con-
jugal violence, (b) the nature and severity of the effects on children, and (c)
variables that directly influence these effects. In the past 5 to 7 years a small
number of exploratory research studies have been conducted. With this small
but growing data base, we are now ready for a careful scrutiny of the emerg-
ing methodology in this area. This examination will help us to: (1) accurate-
ly qualify our existing findings, and (2) prescribe more precise methods for
the next wave of investigations. In order to develop an effective prevention
strategy, we need to know which factors in the environment or specific child
characteristics are related to a child's vulnerability to deleterious effects of
witnessing violence, and how to prevent this harm.
Conjugal Violence and the Child 79

The purpose of this paper is to review the literature that specifically


addresses the issue of harm to children observing conjugal violence. The ob-
jectives of this review are threefold: (a) summarize the current findings; (b)
provide a comprehensive critique of the investigations, highlighting methodo-
logical problems with the present literature and assessment concerns; and
(c) propose directions for future research.

C R I T E R I A F O R I N C L U S I O N OF S T U D I E S

This review was limited to publications that address the effects of wit-
nessing conjugal violence on children. Five well known, on-line databases,
PSYCHINFO, ERIC, SWAB (Social Work Abstracts), NCFR (National
Center for Family Resources), and NIMH (National Institute of Mental
Health) were used to locate relevant articles. These databases were searched
covering a time span from 1967 (or the inception of the database) through
July 1987. Articles were obtained from the databases if they contained any
of the following major descriptors: (a) family violence, (b) family adjust-
ment and violence, (c) marital conflict, (d) marital adjustment and conflict,
(e) marital conflict and child rearing practices plus specific reference to chil-
dren or adolescents. Dissertations and foreign language publications were
not included in the search process. In addition to these on-line databases,
a specialized family violence database (with over 500 references) developed
and maintained by the Family Violence Research Project (FVRP) at the
University of Texas at Tyler was searched for articles pertaining to children
and adolescents.
The search produced approximately 160 references that mentioned chil-
dren or adolescents (out of 2254 listings under the above descriptors). Two
raters searched the abstracts of these publications (and the full article where
necessary) and identified 29 articles that dealt directly with the effects of con-
jugal violence on witnessing children. Twenty-one of the 29 articles were em-
pirical studies and the remaining eight articles were expository papers with
no data presented (Alessi and Hearn, 1984; Carlson, 1984; Elbow, 1982;
Hughes, 1982; Lystad, 1982; Sadoffs, 1976; Star, 1981).

RATING PROCEDURE

A standardized rating procedure was developed to rate the articles that


met the inclusion criteria. Fifty-six separate items under the following six
major categories of data were evaluated for each article: (a) results of the
investigations by relevant domains of child functioning, (b) determination
of violence, (c) definition of the violence that the children observed or heard,
80 Fantuzzo and Lindquist

(d) experimental design issues, (e) parameters of the child data, and (f) demo-
graphics of the parents and children involved in investigations. Each item
required raters to either: (1) abstract information from the article (e.g., "Num-
ber of children in the experimental group "or "Mean age of experimen-
tal group "), (2) answer a "yes" or "no" type of question (e.g., "Were
follow-up data collected?" or "Did the researchers assess for the occurrence
of child abuse?"), or (3) select a response from a list of choices. Table I con-
tains some examples of multiple choice items included in the rating procedure.
Two raters independently rated each of the articles included in this study
on all 56 items. Inter-rater reliability was calculated using a point-by-point
method (Kazdin, 1982). The formula involved dividing agreements by

Table I. Examples of Multiple Choice Items from Rating Procedure


3. Type of Article Reviewed
1 = experimental group design with manipulation
2 = experimental group design without manipulation
3 = experimental single-subject design
4 = non-experimental group design
5 = case study
6 = expository paper without data
5. Types of Data Source for Determination of Violence
1 -- shelter residence vs no shelter residence
2 = clinical impression/interview
3 = questionnaires
4 = archives
5 = standardized instrument
13. Presence and Types of Control Group
1 = no control group
2 = not matched or compared on any dimension
3 = only matched on demographics
4 = only matched on level of distress
5 = matched on both demographics and distress
6 = only compared on demographics
7 = only compared on distress
8 -- compared on both demographics and distress
40. Source of Report of Child Data
1 = child
2 = mother
3 = friend/neighbor/relative
4 = medical/social service/law enforcement personnel
5 = school personnel
6 = independent rater/researcher
7 = multiple sources
8 = other source or report
41. Types of Data Source for Child Rating
1 = clinical impression
2 --- questionnaires/nonstandardized ratings
3 = archives
4 = standardized instrument w i t h / w i t h o u t normative
sample
Conjugal Violence and the Child 81

agreements plus disagreements and multiplying the product by 100. The mean
agreement score across the 56 items was 97.3% with a range of 83.0% to
100%.

SUMMARY OF E M P I R I C A L F I N D I N G S

Across the 23 data-based investigations included in this review, a total


of 1069 children from violent homes were assessed. The average size of the
experimental group was 56.26 children with a range of 15 to 212 children
assessed at one time. The average age of children evaluated was 8.03 years
and the range was 4.50 to 10.3 years. Table II displays a summary of the
empirical investigations. The findings were summarized according to problems
evident in five major domains of child functioning: (a) externalizing behaviors
(or conduct), (b) internalizing behaviors (emotion), (c) social, (d) intellectu-
al/academic, and (e) physical.

Externalizing Behavior P r o b l e m s

Conduct was assessed in 62~ of the studies reviewed. The case studies
indicated that the children who witnessed conjugal violence were generally
reported as evidencing behavior problems at home, at school, and in the com-
munity (Levine, 1975; Pfouts et al., 1982; Rounsaville and Weissman, 1977).
Other case studies reported that these children tended to be more aggressive
(i.e., fighting with siblings and school mates, and displaying tempers tan-
trums) than children who were not exposed to violence in the home, with
males showing more aggression than females (Hilberman and Munson, 1977;
Levine, 1975; Westra and Martin, 1981).
Findings from correlational, retrospective, and cross-sectional research
indicate the need to attend more closely to child and family variables. Some
correlational studies found that children from violent homes evidenced sig-
nificantly more externalizing behavior problems than comparison groups
(Jaffe et al., 1986; Wolfe et al., 1986). One retrospective investigation found
that exposure to marital violence was associated with increased levels of ag-
gression for females only (Forsstrom-Cohen and Rosenbaum, 1985). The only
cross-sectional study indicated that school-age boys from violent homes dis-
played more aggressive behavior than same-aged girls and both boys and girls
from the preschool group (Hughes and Barad, 1983). However, others studies
suggest that these differences can be explained by ongoing parental or familial
factors that are related to interparental or familial factors that are related
to interparental violence. Hershorn and Rosenbaum (1985) reported no differ-
ence between violent and nonviolent groups if the families displayed corn-
Table I1. Summary of Empirical Studies of the Impact of Conjugal Violence on Children"
Study and Comparison Age of chil- Domain o f Type of
(design) group dren (N) functioning measurement Major results
Levine (1975) No 6-14 Emotional Interview and Global reports that children exposed to conjugal
(case study) (117) Conduct Case history violence tend to display anxiety and conduct
disorders
Hilberman and No Not reported Physical Case history Global reports indicating that children who
Munson (1977) (209) Emotional conjugal violence tended to have reported difficul-
(case study) Conduct ties at school, conduct problems were more fre-
quently reported for males and somatic and inter-
nalizjng emotional problems were evidenced more
by females
Rounsaville and No Not reported Conduct Interview and Global reports that children exposed to violence
Weissman (1977) (53) case history tended to display some behavioral disturbance,
(case study) trouble with the law and difficulties in school
Jacobsen (1978) No 3-17 Social Interview and standard- The greater the amount of interparental hostility
(correlational) (51) Emotional ized checklist (P) the greater the amount o f child maladjustment
Conduct
Porter and O'Leary No X- 10.45 Social Standardized Overt conjugal violence correlated significantly
(1980) (64) Emotional Checklist (P) with behavior problems for boys only
(correlational) Conduct
Westra and Martin No X = 5.2 Physical Medical exam, standard- Children exposed to conjugal violence showed an
(1981) (20) Intelligent ized checklist (P), inordinate number of physical problems including
(correlational) Conduct standardized hearing and articulation, more aggressive behavior,
Emotional test (C) and lower verbal cognitive, and motor abilities
Ulbrich and Huber No 18-58 Attitudes Survey by telephone Only men were more likely to approve of violence
=_
(1981) (342) about violence interview (C-A) against women if they observed their father hitting
(retrospective- toward women their mothers
o
correlational)
Rosenbaum and Yes b X = 10.04 Conduct Standardized No significant group differences on any
O'Leary (1981) (53) Emotional checklist (P) trends suggesting that the violent group manifested t-*
(correlational) more problems.
Hinchey and Yes X = 4.42 Social Nonstandardized tests Violent group performed poorer than control on
Gavelek (1982) (16) (C), behavior- tests of role-enactment, social inference, and role- 4.
(correlational) al observations taking (girls poorer than boys on role-taking) but
were not different on behavioral observations of so-
cial behavior.
Pfouts et al. (1982) No Not reported Conduct Case histories, interviews Global reports of "acting out school failure,
(case study) (25) Academic with social workers and emotional problems"
Emotional
Hughes and No 2.9-12.7 Conduct Standardized Higher anxiety for girls than boys. More aggressive
Brad (1983) (130) Emotional checklist (P) behavior in school-aged boys than girls and below
(cross-sectional Standardized checklist average self-concept score for preschool children
correlational) (c)
Hershorn and Yes X = 8.46 Conduct Standardized Both violent and nonviolent (NV) marital discord
R o s e n b a u m (1985) (15) Emotional checklist (P) groups evidenced m o r e conduct problems than
(correlational) c~
control. Only NV group showed more emotional
problems than control.
Wolfe et al. (1985) Yes X = 8.9 Conduct Standardized Violence group evidenced more social, emotional,
(correlational) (102) Emotional checklist (P) and conduct problems than controls. However, a
Social family crisis index as a covariate eliminated signifi-
cant group differences
Forsstron-Cohen Yes X = 19.7 Emotional Standardized Exposure to marital violence was associated with
and Rosenbaum (44) Conduct checklists (C-A) higher levels of anxiety for both males and females
(1985) (retrospec- and higher levels of depression and aggression for
tive correlational) females only.
Jaffe et al. (1986) Yes X = 8.9 Emotional Standardized Level of conjugal violence was associated with
(correlational) (58) Conduct checklist (P) greater adjustment problems for only boys. Both
Social boys a n d girls in violent group showed lower levels
o f social competence than control group, girls in
violent group showed more emotional problems,
boys showed more conduct and emotional problems
Wolfe et al. (1986) Yes 4-13 Emotional Standardized Violent group-current residents of shelters showed
(correlational) (40) Conduct checklist (P) significantly lower social competence than former
Social residents or nonviolent groups. No group differ-
ences on ratings of conduct or emotional problems
~(P) = assessment instruments combined by parents; (C) = instruments completed the children; and (C-A) = instruments completed by the children as adults.
~In each case where a "yes" is indicated the studies represent quasi-experimental designs with nonequivalent control groups (Campbell and Stanley, 1963).
84 Fantuzzo and Lindquist

parable levels of marital discord. Wolfe et al. (1985) found that if you
accounted for concurrent maternal psychological adjustment and family stress
and crises, the differences in the magnitude of externalizing problems be-
tween children from violent and nonviolent were eliminated.

Internalizing Behavior Problems

Data on internalizing behavior problems were assessed in 73.9% of the


studies. Case studies reported depression, suicidal behavior, specific fears
and phobias, tics, enuresis, and insomnia in children and adolescents who
had witnessed conjugal violence. One study reported that female children
were more likely to present as withdrawn, passive, and clingy (Hilberman
and Munson, 1977). The correlational studies again were somewhat equivo-
cal. They ranged from general reports of no differences for the violent group
(Hershorn and Rosenbaum, 1985; Wolfe et al., 1986), to reports of group
differences indicating that the violent group evidenced more emotional
problems (Wolfe et al., 1985), to specific reports that girls from the violent
group showed more internalizing behavior problems than boys.
One retrospective effort (Forsstrom-Cohen and Rosenbaum, 1985)
reported that increased levels of anxiety were evidenced for both males and
females who were exposed to conjugal violence. However, only the females
who were exposed to violence in the home displayed higher levels of depres-
sion. The cross-sectional study indicated higher anxiety for girls than boys
and an indication of below-average self-esteem for the preschool group (Hughes
and Barad, 1983).

Social Functioning

Only 43.5 % of the studies reviewed provided assessments of social func-


tioning. The correlational studies indicated that both boys and girls from
violent homes evinced lower levels of social competencies (Jaffe et al., 1986;
Wolfe et al., 1985). Hinchey and Gavelek (1982) found that children from
violent homes performed poorer than a nonviolent comparison group on non-
standardized measures of role enactment, social inference, and role-taking.
However, there were no significant group differences on a direct observa-
tional measure of social behavior.

Intellectual/Academic and Physical Functioning

Only 17.4% and 9~ of the investigations provided any information


regarding intellectual/academic, and physical functioning, respectively. One
Conjugal Violence and the Child 85

case study (Hilberman and Munson, 1977) reported that children exposed
to conjugal violence appeared to have impaired concentration spans and
difficulty with school work. Westra and Martin (1981) found that children
from violent homes were significantly lower than the standardized popula-
tion on measures of verbal, motor, and cognitive abilities from the McCar-
thy Scale of Children's Abilities.

METHODOLOGICAL PROBLEMS

Definitional Issues: Determination of Violence

In the majority of the investigations reviewed (66.7%), the victim was


the sole source of the report of conjugal violence. No studies included reports
f r o m more than one source, thus making it difficult to assess reliability and
validity. In only 34.8% of the studies, a standardized instrument was used
to ascertain that violence had occurred. This device was the Conflict Tactics
Scale developed by Straus (1979). The remaining 65.2% of the studies relied
on less rigorous mean of determination of violence (e.g., nonstandardized
questionnaires, clinical impressions, or a decision based on only shelter
residence).
In 89.5% of the cases the rater or data collector was not blind regard-
ing abuse status of the women when making the determination of violence.
In 11.8% of the articles, an independent rater established whether violence
had taken place. Most commonly (29.4% of the articles), however, the
researcher collected these data and made the determination. In 11.8% of the
studies, the adult reported on his or her own childhood. In the remaining
articles, the researchers relied on information from social services workers
(5.9%) or other raters involved with the cases (29.4%) for this determination.

Definitional Issues: The Violence Children Observed

The majority of studies failed to precisely describe the essential


parameters of violence to which the children were exposed. Frequency and
type of violence data were not reported in 77.3~ and 65.2070 of the studies,
respectively. The extent of the violence that the children were exposed to was
not reported in 69.6% of the cases. And in 73.9% of the cases, nothing was
reported about the recency of the last violent episode. In fact, no statements
as to whether the children directly observed overt violence were made in 56.5 %
of the articles.
86 Fantuzzo and Lindquist

General Design and Experimental Control Issues

These issues involve the nature of the experimental designs employed


and the description and selection of targeted violent families and nonviolent
comparison families. The majority of the empirical studies were either
retrospective (47.8% of the studies) or correlational (43.6O/o). There was only
one investigation (4.3%) that used a cross-sectional design (Hughes and
Barad, 1983). Further, there were no longitudinal, cross-sectional/longitudi-
nal, or designs that involved a planned experimental manipulation. Addi-
tionally, no studies provided any follow-up data, and 39.1% of the
investigations failed to report any specific selection criteria for the violent
families that were examined.
Two essential components of experimental control involve careful
description of the subject sample and control for potent family factors that
may also adversely affection children. Inspection of family data revealed that
39.1% of the studies reported the ethnic status of family, 69.9% the socioeco-
nomic status, 9.1%o the employment status, 43.5% the marital status, and
43.5% the number of children in family. Twenty-six percent of the studies
reported age of mother. Only 56.5% of the studies used control groups for
comparison purposes. Of this number, only 20~ reported reliability and 10%
reported validity data for the control group selection criteria. No studies
matched violent and control families on demographic and distress variables.
Only 43.4% of the studies made comparisons post hoc on demographic vari-
ables and only 17.3% compared on distress variables. Apart from matching
or comparison factors, data on parental pathology, parental stress, stability
of home environment, and parental skill were collected in only 8.7%, 17.4%,
8.7%, and 4.3% of the studies, respectively.

ASSESSMENT ISSUES

Sources and Types of Child Data

The majority (48%) of cases data on children were provided by mothers.


Six studies collected data obtaining retrospective data from the child as adult.
The predominant mode of data collection was a standardized dating instru-
ment. Nonstandardization questionnaires, clinical impressions, and archival
data were used in 21.7%, 13%0, and 4.3% of the studies, respectively. Only
one investigation included behavioral observations of children. The primary
assessment data involved parental checklists of internalizing and externaliz-
ing behavior problems. Social, intellectual/academic, and physical function-
ing were assessed in less than 50% of the studies. Additionally, there were
Conjugal Violence and the Child 87

no data presented assessing the effects across multiple domains of function-


ing where the assessment was derived from a specific theoretical model.

CONSIDERATIONS FOR CONDUCTING FUTURE RESEARCH

Since this area of investigation is in early stage of development, research-


ers should be cognizant of both the hypotheses generated and the methodo-
logical shortcomings of this small but growing data base and carefully
consider these issues in planning future research.

Definitional Issues

Central to any empirical investigations of the effects of conjugal vio-


lence on witnessing children is a careful detailed description of the family
violence and the extent of the child's exposure to the violence. Previous studies
have not adequately addressed this fundamental issue. Most of the investi-
gations relied on vague reports of violence or used, as their sole criterion,
shelter residence.
In order to establish the impact of violence, the source of the report
should be specified and the reliability of the report clearly established.
Detailed descriptions of parameters of violence as well as an indication of
precisely what the child observed should also be obtained. Ideally there should
be multiple raters blind to the nature and amount of conjugal violence.
Minimally, standardized instruments should be routinely used to standardize
reports for comparison purposes and to improve validity. At present, the
Conflict Tactics Scale (CTS) (Straus, 1979) is the only standardized measure
of family violence available. The CTS is a list of items developed by Straus
and his associates that describe violent and nonviolent actions that may be
used in resolving family conflict. Straus (cf. 1982) has provided reliability
and validity information supporting the utility of this measure. However,
the CTS recently has been reevaluated (Rhodes, 1985, Straus, 1982) and limi-
tations of this instrument have been identified: (1) it is restricted to violence
that occurs over a conflict; (2) it includes a limited list of violent acts; (3)
here are some questions regarding the accuracy of self-reports, especially using
a 1-year time frame; (4) the CTS equates acts that differ greatly in serious-
ness; (5) the context of the violence or reasoning behavior is not identified
with the CTS; and (6) the CTS does not provide data on the antecedents or
consequences (e.g., injuries, court action) that resulted from the violence.
A new instrument should be devised that operationalizes both "violence"
and "witnessing" and also addresses some of these CTS limitations. This meas-
ure should include ratings of: type of violence, rate of violence (frequency
88 Fantuzzo and Lindquist

over time), recency (last episode prior to assessment), and antecedents and
consequences of violence. It is also critical to verify that the child witnessed
violent episodes and to detail the type of exposure (observed, heard, inter-
vened, etc.) and the frequency and recency of exposure prior to assessment.
Also, child self-report measures should be developed to assess children's per-
ceptions of the conjugal violence that they witnessed. These types of instru-
ments would facilitate the precise investigation of relationships between
violence and negative sequence.

Child and Family Variable Issues

The exploratory studies reviewed in this paper, in conjunction with other


empirical investigations of child maltreatment and dysfunctional families,
call attention to the importance of careful examinations of relevant child and
family variables. Sex, age, and general intellectual functioning are child fac-
tors that should be included in any thorough investigations in this area. Seven
of the studies reported here showed differential findings for the two sexes
indicating that males were more likely to evidence externalizing behavior
problems as a result of exposure to conjugal violence and that females were
more likely to display internalizing behavior problems (Forsstrom-Cohen and
Rosenbaum, 1985; Hilberman and Munson, 1977; Hinchey and Gavelek,
1982; Jaffe et al., 1986; Porter and O'Leary, 1980; Westra and Martin, 1981).
Two studies (Hughes and Barad, 1983; Porter and O'Leary, 1980) showed
differential effects of age suggesting a different pattern of problematic be-
haviors for younger and older children. Hughes and Barad (1983) found a
pattern of aggressive and anxious behavior for school-aged children and sig-
nificantly lower self-esteem scores for preschool children. Porter and O'Leary
(1980) found in indication of more conduct problems for boys 10 years and
younger and higher scores on inadequacy-immaturity scales for older boys.
None of the investigations reviewed included a 14- to 18-year age group. Also,
none of the studies assessed children's IQ or commented on the relevancy
of this variable in the interpretation of the effects of witnessing violence.
Future studies should be responsive to those preliminary findings and
the growing mandate in child clinical psychology to consider subject varia-
bles more carefully (Harter, 1985). A cross-sectional design blocked by sex
and age would be a more precise step toward determining the differential
effects of these variables.
In addition, sociocultural differences warrant careful consideration. This
review indicated that family variables such as economic and ethnic status
were not adequately evaluated. Recent research directs attention to the im-
portance of these variables indicating that they affect both how families
respond to problems and how social service agencies respond to families
Conjugal Violence and the Child 89

(Acosta et al., 1982). Parental variables, such as child-rearing style and pu-
nitive parenting, assessed in only one study (Hershon and Rosenbaum, 1985),
were found to affect adjustment. Data should also be collected on the na-
ture and quality of parent-child interactions to determine the effect that posi-
tive or negative patterns of interaction have on the degree of maladjustment
in these violent families.
Other measures that should be included in any investigation are those
tapping child and family stress factors. Marital discord, physical health of
parent, parental psychopathology, single-parent families with higher than
average number of children in the family, employment status, and substance
abuse are all salient variables that should be assessed in future investigations.
Findings by Shaw and Emery (1986) further document the impact of cumula-
tive family stressors on children's maladjustment. Their results showed that
difficulties were directly related to the number of family stressors. Wolfe
et al. (1985) found that differences in the level of externalizing behavior
problems between violent and nonviolent groups could be explained primar-
ily by level of family crises and stress.
Most of the families in the studies reviewed had high unemployment rates,
poverty level incomes, and severe stressors associated with basic survival need
(Hilberman and Munson, 1977). Rounsaville and Weissman (1977) found
20% had depression levels comparable to hospitalized patients and 80% had
moderate depression. Hilberman and Munson (1977) found evidence of psy-
chological dysfunction in more than half of the abused women in their sam-
ple. Substance abuse is another important family stressor that has not been
carefully evaluated in this area. This is a particularly important variable since
data suggest that substance abuse is quite prevalent in this population (Lin-
quist, 1986).
Another important research consideration is that nearly all of the 1069
children that were studied were residents in shelters for battered women. This
raises two questions: First, how representative are these sheltered children
of the general population of children exposed to conjugal violence (includ-
ing children who witness violence but whose mothers never leave the home)?
Second, given the amount of disruption in the home environment experienced
by children living with their mothers in shelters, are the adjustment problems
that are detected merely a result of a "shelter effect" and not uniquely relat-
ed to the degree of family violence? Future studies need to compare shelter
versus nonshelter groups of children who have witnessed conjugal violence
to assess this "shelter effect."
Child abuse and neglect, one of the most provocative childhood stres-
sors, were not assessed in 75% of the articles in this review despite evidence
indicating that children of abused women are at a higher than average risk
of being abused. Straus et al. (1980) found that there was a 129% greater
chance of child maltreatment in a home where conjugal violence was present.
90 Fanluzzo and Lindquist

Giles-Sims' (1985) longitudinal data on 27 battered women with children also


suggest the prevalence of violence in these families that victimize the child.
Additionally, retrospective studies indicate a close relationship between con-
jugal violence and child abuse (Telch and Lindquist, 1984). Given this preva-
lence, future studies should control for the presence of child maltreatment.
Moreover, further investigation is needed to determine the relative impact
of child abuse, witnessing of conjugal violence, and child abuse plus exposure
to conjugal violence.
Future research efforts should eb mindful of these relevant factors and
make careful attempts to control for them. Studies that match a priori on
these relevant demographic and distress variables would clearly advance our
understanding of how exposure to conjugal violence uniquely affects chil-
dren at various ages.

Multiple Sources and Multimodal Assessment

Multiple sources of reports using standardized assessment instruments


are essential to establish better reliability and validity in this area. Current-
ly, abused mothers are the predominant source of child data in these studies.
This is problematic because as Jaffe et al. (1986) have noted, biases may be
affecting the mothers' ratings of their children (males in particular). Hughes
and Barad's (1983) data indicated a pervasive tendency for abused mothers
to rate their children more negatively than did other observers. The collec-
tion of behavior ratings from multiple sources (e.g., shelter staff, other fa-
mily members, teachers, neighbors, or peers) should be an essential element
of future investigations.
Researchers should also make attempts to include observational data
as part of the assessment of children's functioning. School situations would
provide excellent opportunities to estimate on-task behavior, classroom con-
duct, and productivity, as well as the child's level of social functioning. Also,
systematic observational systems [e.g., The Behavioral Coding System (Reid,
1978)] could be employed to assess mother-child interactions. Since it is often
difficult to elicit the father's cooperation, obtaining data on father-child or
family interactions is likely to be problematic.
Because of a growing understanding of the complexity of the socioe-
motional development of children, it is essential that we evaluate children's
functioning across multiple domains. Without a comprehensive multimodal
assessment, we have no way of determining how conjugal violence affects the
"whole" child. We should include assessments of not only emotional adjust-
ment but also physical, social, and intellectual/academic functioning. Cur-
rently, there are a number of excellent standardized questionnaires and
checklists on children's psychological adjustment and social functioning (see
Conjugal Violence and the Child 91

Frame and Matson, 1987). These include measures completed by significant


adults in the child's life (teachers, parents, other relatives), sociometric devices
completed by peers, and self-report measures. Beck (1987), Edelbrock
(1988), Hops and Lewin (1984), and Witt et al. (1988) provide comprehen-
sive reviews of these strategies.

Long-Term Follow-up

In a comparison of current and former residents of shelters, Wolfe et


al. (1986) found that children recently witnessing violence evidenced lower
level of social competence than the former group. This finding underscores
the importance of long-term follow-up studies. At present, we have no way
to distinguishing between reactive and more long-term negative sequalae of
witnessing conjugal violence. Ulbrich and Huber's (1981) retrospective data
suggest that there is an association betwee parental hitting and later adult
attitudes about violence toward women. They found that parental hitting
was related to men's approval of wife hitting. Also, as children grow and
develop these sequalae may take very different forms as a function of nor-
mal developmental crises. A longitudinal design would allow researchers and
clinicians to investigate these changes with greater precision.

SUMMARY

In summary, this review underscores the contributions of these early


empirical studies and outlines an agenda for more systematic and program-
matic research. Essential to these efforts is the development and validation
of measures that precisely define the nature and amount of violence as well
as violence observed by the child. In addition, the nature and amount of
familial distress should be carefully evaluated. Preliminary findings indicate
that a variety of family and parental stress variables, such as parental pathol-
ogy, parental substance abuse, marital discord, and possibly shelter place-
ment, may have differential effects on children's long- and short-term
functioning. These factors may be confounding the assessment of exposure
to violence per se.
Comprehensive multimodal assessment of child functioning should be
implemented to determine more completely the effects of violence. More im-
portantly, this multimodal assessment is necessary to evaluate competencies
of mothers and children who suffer less detrimental effects with an eye toward
developing prevention programs and social policies which would mitigate the
effects of violence and ensuing family disruption.
92 Fantuzzo and Lindquist

Finally, these early research efforts call for a r e c o g n i t i o n of the i m p o r -


tance o f relevant subject variables a n d d e v e l o p m e n t a l factors in the assess-
m e n t process. A t the very least, we need to evaluate the relative impact of
witnessing conjugal violence for children of different ages, sex, a n d sociocul-
rural b a c k g r o u n d s . F o r example, a cross-sectional sample that w o u l d be fol-
lowed longitudinally to track these effects and competencies over time would
m o r e fully assess how these negative sequalae interact with d e v e l o p m e n t a l
factors. A l t h o u g h this is a complex a n d t h o r n y social p r o b l e m to research,
the stakes are too high n o t to use our full arsenal of research m e t h o d o l o g y
to assess, treat, a n d prevent the h a r m f u l consequences of witnessing violence
in the home.

ACKNOWLEDGMENTS

The authors would like to t h a n k Leslie L a m b e r t , Page H a v i l a n d , Mary


York, a n d K a t h r y n Butler for their assistance in c o n d u c t i n g the review.

REFERENCES

Acosta, F. X., Yamamoto, J., and Evans, L. A. (1982). Effective Psychotherapy for Low-Income
and Minority Patients, Plenum, New York.
Alessi, J. J., and Hearn, K. (1984). Group treatment of children in shelters for battered wom-
en. In Roberts, A. (ed.), Battered Women and Their Families, Springer, New York, pp.
49-61.
Beck, S. (1987). Questionnaires and checklists. In Frame, C., and Matson, J. (eds.), Handbook
o f Assessment in Childhood Psychopathology, Plenum, New York, pp. 79-103.
Carlson, B. E. (1984). Children's observations of interpersonal violence. In Roberts, A. (ed.),
Battered Women and Their Families, Springer, New York, pp. 147-167.
Edelbrock, C. (1988). Informant reports. In Shapiro, E., and Kratochwill,T. R. (eds.), Behavioral
Assessment in Schools, New York, Guilford, pp. 351-383.
Elbow, M. (1982). Children of violent marriages: The forgotten victims. Social Casework 63:
465-471.
Emery, R. E. (1982). Interparental conflict and the children of discord and divorce. Psycholog-
ical Bull. 92: 310-330.
Frame, C. L., and Matson, J. L. (1987). Handbook o f Assessment in Childhood
Psychopathology, Plenum, New York.
Forsstrom-Cohen, B., and Rosenbaum, A. (1985). The effects of parental marital violence on
young adults: An exploratory investigation. J. Marr. Fam. 47: 467-471.
Giles-Sims, J. (1985). A longitudinal study of battered children of battered wives. Faro. Relat.
34: 205-210.
Harter, S. (1985). Commentary on the need for developmental perspective in understanding
the needs of child and adolescent disorders. J. Social Clin. Psychol. 3: 484-499.
Herrenkohl, E. C.~ Herrenkohl, R. C., and Toedter, L. J. (1983). Perspectives on the inter-
generationaltransmissionof abuse. In Finkelhor, D., & Gelles, R. (eds.), The Dark Side o f
Families, Beverly Hills, Sage, pp. 305-316.
Hershorn, M., and Rosenbaum, A. (1985). Children of marital violence: A closer look at the
unintended victims. Am. J. Orthopsychiat. 55: 260-266.
Hilberman, E., and Munson, K. (1977). Sixty battered women. Victimol. lnt. J. 2: 460-470.
Conjugal Violence and the Child 93

Hinchey, F. S., and Gavelek, J. R. (1982). Empathetic responding in children of battered mothers.
Child Abuse Negl. 6: 395-401.
Hops, H., and Lewin, L. (1984). Peers sociometric forms. In T. Ollendick, T., and Hersen, M.
(eds.), Child Behavioral Assessment, New York, Pergamon Press, pp. 124-147.
Hughes, H. M. (1982). Brief interventions with children in a battered women's shelter: A model
preventive program. Faro. Relat. 31: 495-502.
Hughes, H. M., and Barad, S. J. (1983). Psychological functioning of children in a battered
women's clinic. Am. J. Orthopsychiat. 53: 525-531.
Jacobsen, D. (1978). The impact of marital separation/divorce on children: i1. lnterparent hostil-
ity and child adjustment. J. Div. 2: 3-19.
Jaffe, P., Wolfe, D., Wilson, S. K., and Zak, L. (1986). Family violence and child adjustment:
A comparative analysis of girls' and boys' behavioral symptoms. Am. J. Psychiat. 143:
74-77.
Kazdin, A. E. (1982). Single-Case Research Designs, Oxford University Press, New York.
Levine, M. B. (1975). Interparental violence and its effects on the children: A study of 50 fami-
lies in general practice. Med. Law 15: 172-176.
Lindquist, C. U. (1986). Battered women and alcoholics: Treatment implications and a case
study. Psychother. Theory Res. Pract. 23: 622-627.
Lystad, M. (1982). Violence in the home: A major public problem. Urban Social Change Rev.
15: 21-25.
Pfouts, J. H., Schopler, J. H., and Henley, H. C., Jr. (1982). Forgotten victims of family vio-
lence. Social Work 27: 367-368.
Porter, B. K., and O'Leary, K. D. (1980). Marital discord and childhood behavior problems.
J. Abnorm. ChiM Psychol. 8: 287-295.
Reid, J. B. (1978). A Social Learning Approach to Family Intervention: Vol. 2. Observation
in Home Settings, Castalia, Eugene, Ore.
Rhodes, N. R. (1985). The assessment of spousal abuse: An alternative to the Conflict Tactics
Scale, Doctoral dissertation, Fuller Theological Seminary, 1985. Diss. Abstr. Int.
5: 2076.
Rounsaville, B., and Weissman, M. (1977). Battered women: A medical problem requiring detec-
tion. Int. J. Psychiatry Med. 8: 191-201.
Rosenbaum, A., and O'Leary, K. D. (1981). Children: The unintended victims of marital vio-
lence. Am. J. Orhtopsychiat. 51: 692-699.
Sadoffs, R. L. (1976). Violence in families: An overview. Bull. Am. Acad. Psychiat. Law 4:
292-296.
Schulman, M. A. (1979). A survey o f spousal violence against women in Kentucky, U.S. Depart-
ment of Justice: Law Enforcement Assistance Administration.
Shaw, D. S., and Emery, R. E. (August, 1986). Spouse abuse, parentalseparation, and school-
age children's adjustment, Paper presented at the 93rd Annual Meeting of the American
Psychological Association, Washington, D.C.
Star, B. (1981). The impact of violence on families. ConciL Courts Rev. 19: 33-40.
Straus, M. A. (1979). Intrafamily conflict and violence: The conflict tactics scale. J. Marr. Faro.
41: 75-88.
Straus, M. A. (1982, July). A reevaluation o f the Conflict Tactics Scale violence measures and
some new measures, Paper presented at the National Conference on Family Violence
Research, University of New Hampshire, Durham, N.H.
Straus, M. A., Gelles, R. J., and Steinmetz, S. K. (1980). Behind Closed Doors: Violence in
the American Family. Doubleday Press, Garden City.
Telch, C. L., and Lindquist, C. U. (1984). Violent vs. nonviolent couples. Psychother. Theory
Res. Pract. 21: 242-248.
Ulbrich, P., and Huber, J. (1981). Observing parental violence: Distribution and effects. J.
Marr. Faro. 43: 623-631.
Westra, B. L., and Martin, H. P. (1981). Children of battered women. Matern. Child Nurs.
J. 10: 41-51.
Witt, J. C., Cavell, T. A., Heifer, R. W., Carey, M. P., and Martens, B. K. (1988). Child
self-report: Interviewing techniques and rating scales. In Shapiro, E., and Kratochwill,
T. R. (eds.), Behavioral Assessment in Schools, New York, Guilford, pp. 384-454.
94 Fantuzzo and Lindquist

Wolfe, D. A., Jaffe, P., Wilson, S. K., and Zak, L. (1985). Children of battered women: The
relation of child behavior to family violence and maternal stress. J. Consult. Clin. Psy-
chol. 53: 657-665.
Wolfe, D. A., Zak, L., Wilson, S. K., and Jaffe, P. (1986). Child witnesses to violence between
parents: Critical issues in behavioral and social adjustment. J. Abnorm. Child Psychol.
14: 95-104.

You might also like