  |
|
|
||||
|
||||||
| Journal of the New Zealand Medical Association,
27-January-2006, Vol 119 No 1228
|
||||||
|
Domestic violence as witnessed by New Zealand children
Judy Martin, John Langley, Jane Millichamp
Domestic violence or partner violence (i.e. physical or
emotional abuse directed from one adult in a partnership to another) has been
recognised as a major social and economic burden in New Zealand for more than a
decade.1 An increasing number of epidemiological studies in New Zealand show a
pattern of domestic violence that parallels that shown in other countries.
Milestone research includes:
Over the previous decade, much
discussion has taken place on the nature and prevalence of domestic violence,
particularly on the substantially different rates and patterns that result from
community surveys conducted in the context of relationship conflict, and from
studies that emphasise violence victimisation.
In the partner conflict studies, women are often identified
as more frequent instigators of acts of physical violence,5 whereas men are more
often reported as carrying out the violence in victimisation surveys.7 A
consensus is slowly emerging that at least two broad types of relationship
violence are being measured, which partially accounts for differences between
studies.9,10
First is what has been called “common couple” or
situational violence arising in the context of a mutual disagreement, during
which one or other of the partners may verbally or physically attack the other.
This is the most frequent type of violence measured in community samples by
multi-item measures such as the Conflict Tactics Scale (CTS), and is as often or
more often carried out by female partners.11 Such studies usually result in a
high prevalence rate for violence by and to both genders.
Second, attempts have been made to operationalise a more
severe form of partner violence, more likely to be reported in victimisation
surveys. Johnson and Ferraro, for example, have focussed on control as the
central issue in severe violence, defining “intimate terrorism” as
attempts by one partner to control another’s behaviour by a range of
methods including physical violence.9 In contrast, Ehrensaft and colleagues used
a range of outcome variables to separate out what they call “clinically
abusive” partner violence. These comprised injury and medical treatment;
and police involvement and related help-seeking, including use of refuge and
restraining orders.10 They found a 9% prevalence rate of such violence between
ages 24 and 26 in the relationships of the members of the Dunedin
Multidisciplinary Health and Development Study. The nature and overlap of the
two broad typologies and their relative distribution and effect on participants,
victims, and witnesses is a matter for ongoing research and debate.
Maxwell’s data from 528 incidents reported by the
Hamilton Abuse Intervention Pilot Project gives an indication of the distress
felt by children who are witnesses to their parents’ violence.12 The
potentially damaging effect on children of witnessing of violence from one
parent to another has long been recognised,13 although there is no clear
indication of the number of children thus affected.
Figures from the Department of Justice website14 report
30,000 children in families affected by domestic violence protection orders
between 2000–2003, and 9241 children using Women’s Refuge services
with their mothers 2000–2001, but the under-reporting of family violence
makes it certain that this under-represents the proportion of children affected.
A local study by Fergusson and Horwood explicitly assessed
exposure to violence between parents;15 nearly 40% of the Christchurch-born
longitudinal sample reported at least one violent act by one or more parent. The
majority of these were from acts of psychological aggression (name-calling and
criticism) and symbolic violence (property damage and threats of violence), but
approximately 6–10% of 18 year olds reported parents pushing and shoving
or hitting or punching the other partner. They found an increased level of
psychosocial adjustment problems in sample members reporting parental violence,
although a follow-up study on panic disorder suggests that the levels of anxiety
measured may be partially attributable to exposure to coexisting forms of
violence in childhood.16
The current study was designed as part of the Dunedin
Multidisciplinary Health and Development Study (DMHDS) to gain retrospective
information not gathered when the study members were children, of their
experiences and reactions to violence directed from one parent (or step-parent)
to the other. This paper reports on the prevalence and nature of this witnessed
violence, and some of the family and social contexts in which it occurred.
MethodBackgroundThe Dunedin Multidisciplinary Health and Development
Study (DMHDS) is a longitudinal study of the health and development of 1037
children born in Dunedin’s only obstetric hospital between 1 April 1972
and 31 March 1973, and still resident in Otago at 3 years of age. Study members
have been assessed every 2 years between the ages of 3 and 15; and since then,
every 3 to 6 years.
The DMHDS sample and study has been fully described
elsewhere.17 The sample as a whole is more exclusively white and more
economically advantaged than the New Zealand average, but a high retention rate
has ensured the inclusion of a full cross-section of social classes. The data on
domestic violence on which this article is based was collected when the study
members were aged 26.
The Family Violence Interview was a 10-minute
face-to-face interview carried out as part of a day-long assessment. It included
questions asking for retrospective reports about parents’ potentially
violent behaviours to the study member as well as to each other. Such questions
had not previously been asked of study members directly because the earlier need
for parental consent for participation restricted the use of sensitive questions
that might lead to withholding of consent. Before the assessment, pre-testing on
the suitability of the topic among a random sample of both study members and
their parents established that such questions would be acceptable at this stage
of the study members’ lives, and a pilot questionnaire among a volunteer
sample tested the suitability of specific questions.
Domestic violence questionsStudy members were asked about interparental violence
in the context of parents’ arguments. The questions came after a sequence
on discipline in the family, and were introduced as follows:
We’ve talked about
what happened when your parents were mad with you. Now I want to ask you about
what happened when your parents used to argue with each other.
All study members were asked first how their parents
got on, with a range of possible answers from “really well” to
“lots of rows.” They were then asked whether they had seen or heard
of their mother being hit or hurt by their father. As a separate question they
were asked whether the mother had been threatened with harm. These two questions
were repeated with relation to the father being hit or hurt or threatened by the
mother. This briefer screen was used instead of a multi-item checklist such as
the CTS because of the short time available.
We focussed on physical or threatened acts as being
behaviours more easily identifiable by witnesses than the subtleties of purely
psychological violence, although several study members questioned our omission
of this aspect, believing it central to their parents’ experience.
All study members who gave a positive answer to any of
the four prevalence questions were asked to provide a brief description of what
happened, including specific details:
If the study member had lived with a
step-parent at any time, the same screening questions were repeated for the
subsequent relationships, and brief contextual details then collected.
The interviewers (both health professionals) noted any
signs of distress or comments the study members made about the interview content
and process. A psychologist was available as part of the Unit team if
participants needed to discuss issues raised by the interview. Study
members’ descriptions, typically two or three sentences, were noted on the
interview form and used to validate study member’s responses to the
prevalence questions. They were later transcribed and content-analysed to
identify patterns of assaultive behaviours, injury levels, and alcohol
involvement.
A subgroup who reported obviously severe physical
violence was identified. These study members either volunteered information
about one parent being injured by the other, or used descriptions such as
punching, “beating up”, choking, throwing down, threatening with a
weapon, kicking, or hitting with something likely to cause injury, and rape.
These are the behaviours that rate as severe violence in the revised CTS,18 and
were also designated as severe violence by Fanslow and Robinson.8
Categorisation of family violence experiencesOn the basis of their responses to the assessment,
study members were placed in one of four categories:
The two dimensions of physical versus
threatened violence and frequency of physical violence were chosen because they
were the qualities that child witnesses could most easily identify and report
on. Because of the varying age of the children as witnesses, as well as the
environment of secrecy in which most family violence occurs, it was thought that
study members would not be in a position to reliably assess the degree of injury
involved. As explained above, many study members did in fact describe injuries
and severe assaults, but because this information was volunteered, rather than
asked systematically, neither was used as a measure of severity in our
statistical analysis.
The following variables were used in the
cross-sectional analyses (brackets indicate study members’ age at the time
specific information was obtained):
Demographic information
Study member mental health
At age 21,
study members participated in the DIS, a structured psychiatric interview based
on DSM-IV criteria.20 The multiple categories of anxiety disorders were combined
into a single binary variable for the purposes of this analysis, identifying
those who had experienced any anxiety disorder in the previous 12 months, and
the same process was used to identify those experiencing any depressive
disorder.
Family-relatedFamily Environment
Scale, (age 7)—This validated interview (measuring aspects of
overall family functioning) was given by the primary caregiver.21
Subscales used in this analysis were:
Questions and answers
related to the family as a whole, not to specific family members. Higher scores
indicate more favourable conditions—i.e. high cohesion, high expression
and low conflict.
Results962 of the 980 study members participating in phase 26
completed the interview on family violence—equivalent to 94% of the
original sample of 3 year olds and 96% of the living cohort. Three-quarters
(n=726) reported no violence between parents or step-parents; 55 (6%) reported
threats as the most severe form of violence; 86 (9%) reported 1 to 4 incidents
of physical violence; and 95 (10%) reported 5 or more such incidents.
Descriptions provided showed that “threats”
included a range of verbal and non-verbal behaviours, including “constant
put-downs”, “ranting and raving”, threatening with a wide
range of objects, throwing and smashing objects, and threats of injury or death.
Physical assaults included examples of slapping, hitting, shoving or pushing
over, as well as the more severe forms of assault such as punching, kicking,
beating up, hitting with something hard, throwing around, and sexual assault.
Table 1 shows demographic, family and individual
characteristics associated with these categories of violence. For 90% of the
study members witnessing parental violence, the first violence reported was
between natural parents.
Table 2
(link including Table 1) shows a
summary of the characteristics and contexts of the parental violence reported by
study members in response to direct questioning. Of the 236 study members who
reported parental violence, 69% witnessed it directly. The others either heard
the sounds of fighting (18%) or were told about it afterwards (13%). Eighty
percent of the study members reported violence beginning before they reached the
age of 11 (mean age: 7.5). Police were involved at sometime in 12% of the
non-physically violent families and only 6% of families with infrequent
violence, but 45% of those involving frequent assaults. Two-thirds of this last
group had one partner leave at least temporarily as a result of the
violence.
Study members were asked in separate questions about their
mothers being “hit or hurt”, or threatened and then their fathers
“hit or hurt” or threatened by the other partner. Of the 236 exposed
to violence, 171 (73%) reported mothers being threatened, and 158 (67%) reported
mothers being physically assaulted by the male partner.
One-third (33%) of those exposed to violence reported
threats to the father, and 29% reported that the father was physically assaulted
by the female partner. Altogether, 39 (16%) of the exposed group reported
violence by the mother only, 67 (28%) by both partners, and 130 (55%) by the
father only. When physical assaults only were considered, 23/181 (13%) were by
women alone, 46 (25%) involved both, and 112 (62%) were by men only.
Characteristics of families were compared for the four
categories of violence exposure identified. (See Table 1 for details.) Domestic
violence was less common in families who remained intact. Only 37% of families
with frequent physical assaults between the birth parents had parents who
remained together, compared to 75% of the families where no violence was
reported (Chi squared=49.65; df=3; p<0.001).
Study members reporting violence were much less likely to
say that the parents got on well together. Violence was related to
father’s socioeconomic-status, with more study members reporting repeated
violence if their fathers were in SES groups 5–6 (manual and unskilled).
In families reporting violence (especially physical violence) both the mothers
and fathers were significantly younger than the mean age. Families where
violence occurred rated lower on the family cohesion scale and higher on the
family conflict scale. Rather than demonstrating a “dose-response
effect”, it was families with infrequent violence that reported slightly
worse cohesion and conflict.
There was no significant relationship between gender of the
study members and of the type of conflict they witnessed, although there was a
trend for men to report relatively more infrequent violence and less threatened
and frequent violence (p=0.114). Violence was not associated to the
mother’s employment status (age 7), mother’s education (at age 13),
or the level of family expressiveness reported.
When the study members were asked to reflect on whose fault
they thought the fighting was, 46% said just or mostly father, 45% attributed
blame to both father and mother, and 7% to just or mostly mother. Four study
members said “neither”. When only the father was hitting or
threatening, 61% considered he was solely or mostly to blame. When the mother
was reported as threatening or hitting the father, whether or not he was violent
as well, the proportion blaming the father more fell to 27%. There were gender
differences between attribution of blame, with female study members more likely
than males to hold the father responsible overall (52% compared to 40%), and 54%
of men thinking the violence was a joint responsibility compared to 39% of
women, (Chi-squared=32.24; df=6; p <0.001).
Regardless of who carried out the violence, the study
members remember being upset by the conflict they witnessed; 64% described
themselves as being upset “a lot” or “extremely”, with
another 23% nominating “a bit” upset. Logistic regression modelling
showed that the violence related variable that was most strongly associated with
being very upset was the frequency of violence (adjusted 95%
CI=2.34–8.34), with blame attributed to both parents acting as a slight
protective factor against upset (95% CI: 0.257–0.875).
Levels of distress were not predicted by the other variables
in the model: the gender of either the perpetrator or victim; the age of the
study member when the violence was witnessed; and whether the violence was
physical or threatened.
When less immediate forms of distress (anxiety and
depression diagnoses) were examined (Table 1), differences between study members
exposed to different types of violence were less marked than the contrast
between those who witnessed violence, and those who did not. Neither diagnosis
was itself associated with being very “upset” at the violence in
cross-tabulation (p values 0.986 and 0.755 respectively). Repetitions of the
logistic regression modelling of violence related variables with anxiety and
depression as the dependent variables showed that female gender of the study
member was the only variable that predicted anxiety and depression (adjusted 95%
CIs=1.27–4.61 and 2.15–9.26 respectively).
Study members’ contextual reports on parental violenceStudy members were asked to describe briefly the violence
they witnessed, heard, or were told about. Some gave descriptions that were
general in nature, but patterns emerged from specific information given by the
majority. The most common single element mentioned by study members was the
involvement of alcohol (mentioned by 22% of the reporters [53/236]), most
commonly in relation to father’s drinking. Injury, ranging from bleeding
noses to prolonged hospitalisation, was mentioned in one-fifth of the reports
(n=46), but many others detailed acts such as punching and “beating
up” that carried a high probability of some injury. The element of
controlling or emotionally abusive behaviour was specifically mentioned by only
14 study members, half of whom were commenting on relationships without physical
violence.
About half of the study members who described physical
violence (43% of those reporting infrequent physical violence, and 55% of those
reporting frequent violence [89/181]) gave accounts which either fitted the
revised CTS criteria for severe violence, or referred to some degree of injury.
Forty-three percent of the dyads where the mother alone was physically
assaultive mentioned such severe violence (n=10), as did 48% of those where the
father was responsible (n=54), and 54% of the partnerships where both partners
assaulted the other (n=25).
Narratives by the 46 study members where both father and
mother were reported to have hit or hurt the other partner were examined to see
if the precipitating actions for violence and its relative severity could be
determined. Levels of violence were equivalent or unable to be determined in
over half of the descriptions, fathers were described as responsible for more
severe violence in 37% of the accounts, and mothers in 2%. Mothers were
described as initiating the dispute in some way in 24% of cases, and fathers in
7%, but 46% of the fathers and 17% mothers were identified as being the first to
use physical violence. Two couples were specifically held equally responsible,
and direction was not specified in the other reports.
DiscussionThis New Zealand study
identified a quarter of the participants as witnesses to physical or threatened
violence between their parent or parent-figures. It demonstrated a high level of
family conflict and the potential for upset to younger family members.
The strength of this study is
its ability to identify and describe (in a community sample) the range of
violent acts that children witness; the contexts in which they occur; and
consequences for family members. The quality of the representative community
sample with high participation rates and availability of prospectively gathered
data allows a clearer and more accurate identification of the nature of violent
families.
The information was gained
retrospectively when the study members were adults, and so recall and reporting
cannot be considered completely independent of biases conferred by later
experiences. However, this difficulty must be weighed against the alternative
problem of trying to get sensitive information about parents from children when
they are still dependent on the consent of the people they are reporting on, as
well as the more subtle influences of close physical and emotional ties on
disclosure.
The retrospective nature of
this data could also limit its epidemiological relevance. The cohort were
children during the 1970s and 1980s, before widespread public education on
domestic violence was underway, so it is unwise to project from these figures
what rates of violence may be witnessed by children in today’s families.
However the high rates of violence reported by surveys in the 1990s and later
suggest that the prevalence would still be substantial.
The use of reports from child
witnesses to violence is both a strength and a weakness of this study. As a
negative, it probably does not record the full prevalence of domestic violence
in families, by missing violent acts that were not carried out in front of the
children, or not reported afterwards. It is likely that hidden violence occurred
in more families than of the 29 study members who were unaware at the time but
had found out by the age of 26.
On the other hand, the
exposure of children to domestic violence is a serious public and mental health
concern, and questioning relying on the reports of the witnesses themselves is
the most direct measure of impact of this particular facet of violence. Our main
focus was on the experience of the exposed children, and we wanted their reports
to be unprompted where possible. It is probable that some minor incidents,
threats, and psychological aggression passed unnoticed by the children, but we
were surprised at the level of detail and insight provided in the retrospective
narrative comments, all the more compelling for being undirected.
Information volunteered about
a wide range of assaults, injury, and psychological dynamics appeared to
contradict our initial belief that study members would not be able to provide
information about such details. Unfortunately, because the severity measure we
developed from volunteered information was not based on systematic inquiry, we
were not able to use it as an independent variable in analysis. A greater
confidence in our study members would have resulted in a wider range of assault
parameters being measured.
Our study fits more clearly
into the “violence victimisation” thread of domestic violence
research rather than the “partner conflict” thread, despite the
introduction of the screening questions in the context of parental disagreement.
The factors that match the “victimisation” methodology are:
introduction to the study members as an interview on family violence; use of a
short prevalence screen rather than a multi-itemed checklist such as the CTS;
the use of the terms “hit or hurt” and “threatened with
harm” which may have led some study members to screen out minor incidents
they did not consider harmful; and the exclusion of verbal aggression with the
exception of threats.
Thus, it is not surprising
that our witnessed prevalence rate of 24% is closer to that of the national
crime victims survey than the 40% reported using a modified CTS by Fergusson and
Horwood,9 The high prevalence rate for physical violence in Fanslow and
Robinson’s study, which used a multi-item scale in the context of a
victimisation survey suggests that the number and range of screening questions,
rather than the way they are presented, is the dominant influence on prevalence
rates.8
The studies carried out
within the Dunedin and Christchurch longitudinal studies (and the National
Surveys of Crime Victims) are the largest New Zealand community surveys that
have examined men’s experience of domestic violence at the same time as
women’s.4-7,15 Like the crime victim surveys and the assault victimisation
study in the DMHDS at age 21, 4,6-7 our
study found that women were more often victims of partner assault than men.
These three studies also relied on fewer screening questions.
The greater proportion of
male perpetrators is in keeping with police and health figures, but different to
the CTS-based surveys of Magdol et al and Fergusson and Horwood which have a
more equal gender balance.5,15 It is more useful to explain the divergent
findings fully and examine the parameters which contribute to differences than
to question the value or validity of either methodology. While debate continues,
domestic violence research should assess both men and women where
possible.
This study found fewer female
perpetrators, rather than a greater number of assaultive fathers. This may be
because study members disregarded a proportion of violence they witnessed from
mothers as being less forceful and therefore not “harmful”.
Interestingly, although the number of reported women perpetrators in our study
was relatively low, 43% of those cases were rated as serious by our informal
severity measure, which is a similar proportion to the 48% of male-only physical
assaults. In addition, the proportion of study members who described assaults
where both parents hit or hurt as serious rose to 54%.
Whether reciprocated violence
acts as a trigger to greater severity is a potentially important public health
question that deserves further study. In the mutual violence group, although
unwise to interpolate too much from the incomplete self report data, it is
interesting to note that mothers were more often noted as beginning the conflict
(at least verbally), while fathers more often were physically violent first, and
their violence was often described in more severe terms. Of the 14 parents who
were specifically described as being controlling or emotionally abusive, 12 were
fathers.
A study that relies on
witness reports provides a different perspective than one relying on
self-reports from either victim or perpetrator, but some differences reported by
male and female study members highlight the possibility of reporting bias. It is
unlikely that the parameters of violence witnessed by male and female study
members varied substantially, so it is probable that differences in violence
details reported by men and women reflect perceptions rather than actual
differences.
Of particular interest is the
tendency of male study members to more often share the attribution of blame
between both parents. Women were much more likely to blame the father
exclusively, but still this was only in about half the instances of witnessed
violence. Study members in general seemed able to take a nuanced view of family
violence, perhaps influenced by their intimate knowledge of family dynamics.
Both males and females were less likely to place exclusive blame on the parent
of their own sex. Interestingly, those who did perceive both parents as at fault
were less upset by the violence.
Study members typically
reported being very or extremely upset at the time by the violence in their
families. This perception of the emotional impact of the conflict was unrelated
to their previous mental health status, as measured by depression and anxiety
scores. Subjective distress and possible mental health outcomes have long been
recognised as separate (but clinically important) consequences of childhood
abuse.22 It is likely that many other coexisting factors influenced the levels
of depression and anxiety in those witnessing violence, but an examination of
those is beyond the scope of this paper.
Those working with all
aspects of domestic violence should note the very high proportion of the study
members who reported being very or extremely upset by the experience regardless
of the type of violence. It has been acknowledged previously that victims of
domestic violence often find non-physical violence equally as distressing as
physical assaults, and the same is true of witnesses in this sample.23 The
finding that the factor that contributed most strongly to being upset was the
frequency of the violence, and not who carried it out, or whether it was
physical or threatened carries a strong message about the potentially damaging
nature of any violence between adults in the home. There has been a tendency to
discount the harm attributed to violence carried out by women (Morris et al have
a useful discussion on this topic in chapter 5 of their report).7 But the
argument of the relative benignity of female violence does not match our data on
distress, nor our informal data on severity.
Another
useful aspect of the study is the ability to distinguish between families where
physical violence was a rare occurrence, and those where it occurred more
frequently. The two groups differ in various ways, including the degree to which
mothers are involved as perpetrators, the proportions of parents who separated
temporarily or permanently and other reports of family functioning.
Johnson stresses the need to disaggregate family violence into two main typologies—common couple violence and “intimate terrorism” where violence is a tactic deliberately used by one partner, in a pattern of control.9 It was not often possible to make this distinction from the brief narratives given by the study members, but those identified as controlling were almost exclusively frequent users of violence as were those who used severe physical violence as a response to everyday situations such as mealtimes and child management. From this evidence, the commonsense approach of using frequency as a marker for severity has some merit, especially considering its contribution to greater distress in the child. A follow-up paper from this
study will provide a fuller examination of the long-term impact of domestic and
other family violence on children in the context of other personal and family
factors, both positive and negative. Future research could investigate possible
associations between children’s attitudes to their parents’ violence
and their own experience of partner violence in later life. In the meantime, the
negative reactions to both the threats and physical assaults the children
witnessed reinforce the potential for harm and distress for all concerned when
adults in families fight.
Author information:
Judy Martin, Senior Teaching Fellow, Department of Psychological Medicine; John
Langley, Professor, Injury Prevention Research Unit, Department of Preventive
and Social Medicine; Jane Millichamp, Lecturer, Department of Psychological
Medicine; Dunedin School of Medicine, University of Otago, Dunedin
Acknowledgements:
This research was funded by NZ Health Research Council grant 98/140. The Dunedin
Multidisciplinary Health and Development Research Unit is funded by the NZ
Health Research Council. The authors thank Deborah Clarke for help with data
management; Peter Herbison for statistical support; Richie Poulton for support;
and the members of the Dunedin Multidisciplinary Health and Development Study
for their ongoing commitment and participation.
Correspondence:
Judy Martin, Dpt of Psychological Medicine, Dunedin School of Medicine,
University of Otago, PO Box 912, Dunedin. Fax: (03) 474 7934; email: judy.martin@stonebow.otago.ac.nz
References:
|
|
||||
| | Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |
|