Jeffrey L. Edleson, Lyungai F. Mbilinyi, Sandra K. Beeman, and Annelies K. Hagemeister
MINCAVA - the Minnesota Center Against Violence & Abuse
School of Social Work, University of Minnesota
The David & Lucile Packard Foundation and the Minnesota Agriculture Experiment Station (MIN-55-019) funded this research. The authors wish to thank the staff of the Domestic Abuse Project, Inc. in Minneapolis, Casa De Esperenza and La Opportunidad in St. Paul, the Women’s Center and Shelter of Greater Pittsburgh, the Support Network for Battered Women in Mountain View, CA, and the Family Place in Dallas for their collaboration in this study. Reprint requests may be sent to MINCAVA - the Minnesota Center Against Violence & Abuse, University of Minnesota, 105 Peters Hall, 1404 Gortner Avenue, St. Paul, MN 55108 (tel: 612.624,0721, or email).
Prepublication draft; not yet peer reviewed.
Please do not circulate or cite without the written permission of the authors.
Under review, Journal of interpersonal violence
Temporarily available at http://www.mincava.umn.edu/link
(February 15, 2001)
Abstract
How children are involved in adult domestic violence: Results from a four city telephone survey
Method
Results
Discussion
Endnotes
References
Appendix A
Children's exposure to adult domestic violence and our society's response to it have become major policy and practice debates in recent years. Yet there is limited research available on children's involvement in incidents of adult domestic violence. Earlier studies have often relied on children’s reactions to simulated conflict and focused on studying predictors of child involvement that are associated with a limited number of demographic variables. This study collected direct reports on real-life events and was designed to go beyond earlier research by eliciting information on a larger array of family and contextual factors that may account for predict variation in children’s responses. Anonymous telephone interviews with 114 battered mothers in four metropolitan areas across the United States elicited detailed information from women on their children’s observations and responses to the violence being committed against the mothers. Results indicated that a large proportion of the children in this study were exposed to their mothers’ experience with violence, and a quarter of the mothers reported that their children were physically involved in the events. In addition, children mothers with less stable financial, social, and living situations at the time of the interview reported their children to have been more exposed and to have intervened more during the past violent incidents. In this sample were reported to be exposed and intervene less (during the abuse) violent incidents when their mothers’ financial, social and living situations at time of interview were reported to be more stable. The article concludes with recommendations for a greater emphasis on careful assessment of children’s involvement in domestic violence incidents and on assisting mothers to achieve economic stability as well as safety.
Children’s exposure to adult domestic violence and our society’s response to it have become major policy and practice debates in recent years. While evidence of children’s exposure to and involvement in adult domestic violence has long existed (Edleson, 1991; Gordon, 1988; Pleck, 1987), only recently have state legislatures taken action to revise their definitions of maltreatment to include childhood exposure (see Minn. State Ann. 626.556) or to make such exposure a separate chargeable offense (Utah Code Ann. 76-5-109.1). This interface between childhood exposure, child maltreatment, and adult domestic violence are also the subject of initiatives by the U.S. Departments of Justice and Health & Human Services as well as several national professional organizations and private foundations.
These legislative and programmatic initiatives respond to mounting evidence that child maltreatment and adult domestic violence co-occur. Over 30 studies reveal a link between child maltreatment and adult domestic violence, showing a 40% median co-occurrence in families studied (Appel & Holden, 1998) with the majority of studies ranging from a 30% to 60% overlap depending on the families studied (Edleson, 1999a). Separately, some children are not directly maltreated but are exposed to adult domestic violence in their homes. Child problems associated with exposure to adult domestic violence are also the subject of a growing body of literature that has been reviewed extensively in recent years (see Edleson, 1999b; Fantuzzo & Lindquist, 1989; Fantuzzo & Mohr, 1999; Holtzworth-Munroe, Smutzler & Sandin, 1997; Jaffe & Sudermann, 1995; Kashani, Daniel, Dandoy & Holcomb, 1992; Margolin, 1998; Peled & Davis, 1995; Rossman, in press). Overall, existing studies reveal that some children exposed to adult domestic violence show a host of greater behavioral, emotional, attitudinal and cognitive difficulties when compared to those not so exposed. These problems are also thought to continue into early adulthood for some exposed children.
Children’s presence during domestic violence incidents, even when they were not themselves abused, has been less studied than either the co-occurrence or the impact of exposure. For example, Hughes (1988) found that all 40 child witnesses she studied were “either present in the same room and saw the fighting or were in an adjacent room and heard the physical conflict” (p. 80). In another study, 77% of children reported witnessing domestic violence even when one or both parents reported that their children had not seen it (O’Brien, John, Margolin & Erel, 1994).
How do children immediately respond when they observe marital conflict or violent incidents between parents? Appel and Holden (1998) suggest a “family dysfunction model” where children become involved in violence along with their parents. This is possibly supported by the results of the 1985 National Family Violence Survey which revealed that children in homes where both child and spouse assault had occurred were 18 times more likely than other children to have used violence against their parents (Hotaling, Straus & Lincoln, 1990). While these data reveal greater aggression by children in these homes, they do not clarify the ways in which children are involved in violent incidents and what predicts their involvement.
Children’s involvement in violent situations has been shown to vary from their becoming actively involved in the conflict, to distracting themselves and their parents, or to distancing themselves by leaving the room (Margolin, 1998). Garcia O’Hearn, Margolin, and John (1997) studied 110 families and found that parents whose conflict was often characterized by physical violence reported that their boys were significantly more likely than other boys in the study to respond to conflict by leaving the room or appearing sad or frightened. This difference was not significant for girls.
While Garcia O’Hearn et al. (1997) found children avoiding situations or responding emotionally, other studies have found a wider variety of responses by children. For example, Peled’s (1998) qualitative study of 14 preadolescent children exposed to domestic violence revealed children’s use of two primary strategies when adult domestic violence occurred: distancing oneself from the event or intervening directly in it. Children who distanced themselves found places in or away from the home where the incidents were not heard nor seen. Some used televisions or loud music to distract themselves, while others “willed away” their feelings and thoughts about the events. Those children who got involved in events took sides in arguments, protected their abused mothers by “jumping in the middle of it” or called the police.
Children of different ages also show some variation in their responses to violent conflict at home. In one of the earliest studies on this subject, Cummings, Zahn-Waxler and Radke-Yarrow (1981) examined mothers’ reports of the responses of 24 children between the ages of one and two and a half years. They found that even children this young responded to angry conflict that included physical attacks with negative emotions and efforts to become actively involved in the conflicts. In a later study, Cummings, Pellegrini, Notarius and Cummings (1989) found that as children aged they showed increasing evidence of a variety of responses. Forty-eight children between the ages of two and six were studied and as they got older they increasingly observed the conflict, expressed concern, sought social support, and intervened to protect or comfort their mothers. This effect was greater among children whose parents were engaged in physical conflict when compared to others and among boys when compared to girls. The study examined children’s responses to only simulated conflicts between the children’s mothers and a researcher, no real-life conflict was observed or reported.
In a more recent study, Adamson and Thompson (1998) examined children’s response strategies when they compared 40 children from homes in which there was a history of domestic violence to 72 other children from homes in which there was no evidence of violence. Children in this study reacted to scenarios describing verbal conflict between parents. The most dramatic difference between children’s responses was in their use of aggressive intervention. Children from homes in which there was violence were eight time more likely to use verbal or physical aggression to intervene in parental conflict than were children from violence-free homes (27% vs. 3%).
Overall, these studies show children responding in a variety of ways to real and simulated violent conflict between their parents. One limitation in several of these studies is that they have examined only simulated conflicts that are either told in a story or enacted in the laboratory. Another limitation of these earlier studies is that very few factors outside simple demographic and parent relationship variables have been examined. Studies that also examine real-life events and the variety of contextual factors at play in these families are needed.
The study reported here aimed to add to this limited literature on children’s involvement in incidents of adult domestic violence by collecting direct reports from mothers on real-life events. It also was designed to go beyond earlier research by eliciting information on a larger array of family and contextual factors that may help to explain variation in children’s responses. Through anonymous telephone interviews with battered mothers in four large cities across the United States, this study elicited detailed information from women on their children’s observations and responses to the violence being committed against the mothers. The study sought to systematically survey a large number of battered mothers regarding the degree and ways in which their children observed, responded to and intervened in the violence they were experiencing. The primary research questions that guided the study included:
In the sections below, we describe the study methods, report the results of data analyses, and discuss their implications for policy and practice.
Sample recruitment
Domestic violence programs in four metropolitan areas assisted in recruiting mothers for this study. The areas included Dallas, Minneapolis/St. Paul, San Jose, and Pittsburgh. Due to the sensitive nature of the interview topic - children’s exposure and involvement in adult domestic violence - and the different requirements of the child maltreatment reporting laws in each of the states involved in the study, sample recruitment posed ethical dilemmas. After extensive consultation with a university Institutional Review Board and experts in child maltreatment reporting, a decision was made to conduct anonymous interviews with mothers who volunteered to participate in our study. In addition, interviews were focused only on incidents of violence that occurred at least 12 months prior to the interview date. All of the women who volunteered for this study were currently connected to domestic violence service agencies, which provided crisis and support services to the women and their children. These features of our sampling and recruitment procedures allowed us to interview women about their children’s exposure and involvement in incidents of adult domestic violence without putting them at risk for a report to child protection services for an incident that, in some states, could have occurred a decade or more in the past.
To recruit study participants, flyers and brochures describing the project and the interviews were sent to participating agencies. The flyers contained a toll-free number for interested women to call to find out more about the study. Phone lines were staffed with at least two interviewers between 9:00 a.m. to 9:00 p.m. (Central Time) on Mondays, Wednesdays, and Fridays, during the seven-month study period. A bilingual Spanish-English interviewer staffed 20 of those hours. This procedure required considerable resources and time to complete interviews, as there were many hours when phone lines were staffed but no one called for an interview. The benefit of this recruitment procedure was, however, that women could participate in the study without revealing their identity. Women who met study criteria and consented to participate in the interview were given a code number so that if the interview was interrupted, she could call back to complete the interview and still maintain her anonymity. Upon completion of the interview, women were given a second code number. This code number was faxed to the participating agency at the end of the interview so that the agency could release a $20 payment to the respondent.
Sample description
Anonymous telephone interviews were completed with 114 battered women using the methods described above. Table 1 summarizes the demographic characteristics of the sample. Information on the woman’s job status, income, transitional housing, and education were all reported for the time of the interview, which was at least twelve months after the violent incidents she described during the interview. Other independent variables, such as the woman’s relationship to abuser, her experience of physical and emotional abuse and its effects on her and her children were only reported retrospectively for the period during which abuse was occurring. These differences in reporting periods are important to keep in mind in interpreting the results of the study. We cannot determine, for example, if a woman’s current economic status at the time of the interview reflected her earlier economic status at the time the abuse was occurring.
The average age of the women at the time of the interview was 34 (sd = 7.70), and at the onset of abuse was 25 years (sd = 6). Caucasian women constituted 45.6% (n=52) of the sample, African-American women 33.3% (n=38), Latina women 11.4% (n=13), and other groups (Native American, Asian, and biracial/multiracial) constituted 9.6% of the sample (n=11). At the time of the interview, 44% of the women (n=50) reported completion of high school or GED as their highest education level, 40% (n=46) had received vocational training or some college, the remaining 9% (n=10) had completed a college degree or higher.
Table 1. Description of sample demographic characteristics
Total: N = 114
| Demographic | Percentage, Mean (SD) |
|---|---|
| Average Age at interview (SD)
Average Age at onset of abuse (SD) Education at time of interview
Housing during the interview at time of interview Annual Income at time of interview Average length of abuse Race
Relationship to abuser during abuse Number of children during abuse |
34 years (7.70)
25 years (6) 44% (n=50) with High School/GED 40% (n=46) with vocational training/some college 44% (n=47) employed 45% (n=51) lived in a shelter, 10% (n=11) lived with family, 8% (n=9) lived in subsidized apartment, and 24% (n=35) lived in a rented or owned home. 47.5% (n=48) <= $10,000 annual household income Six years (6) 45.6% (n=52) Caucasian; 33.3% (n=38) African-American; 11.4% (n=13) Hispanic/Latina; 9.6% (n=11) 'other' 77% (n=88) with abuser 45% (n=51) married to abuser 84.5% (n=97) had 1-3 children |
At the time of the interview, 41% (n=47) of the women were employed at least part time and 52% (n=59) were not employed. Fifty-four percent (n=62) of the women were in transitional housing (shelter, other temporary housing, or living with family/friends and 46% (n=52) were renting or owned an apartment or home. Also at the time of the interview, 42% (n=48) of the 114 women had an annual income of less than $10,000, and 32% (n=36) had an income between $20,000 and $30,000 annually.
Forty-five percent (n=51) of the women identified their abuser as their former or current spouse and 39% (n=44) identified him as a former or current boyfriend. The average length of the relationship with the abuser for the women was eight years (SD=6), and the average length of abuse was six years (SD=6). Seventy-seven percent (n=88) of the women lived with their abuser at the time of the abuse, while 23% (n=26) did not. Out of the 114 women, 12.7% (n=14) had one child, 47.3% (n=52) had two children, and 24.5% (n=27) had three children, with a total of 285 children in the 114 households. At the time of the interview, the children ranged in age from two years old to 32 years, averaging 10 years old (SD = 5 years).1 Forty- one percent (n=117) of the children were female. Among the 285 children, 34% (n=97) were Caucasian, 38% (n=108) African-American, 12% (n=34) Latino, 12% (n=34) mixed or no primary racial group, and 5% (n=14) other (including Native American, Asian, and Pacific Islander).
Data collection methods
The telephone survey had several sections, each developed under consultation with the Minnesota Center for Survey Research (MCSR) and staff from the collaborating agencies. Graduate students in social work were hired to conduct the telephone survey. They already had experience in the field of domestic violence and were trained by the project staff.
The interview guide focused on the development and interaction of violence in a family with children, along with the responses of formal services and informal networks that may become involved in families involved in domestic violence. The first set of questions gathered data on the duration, frequency, and severity of abuse on the mother. Questions on the mother’s emotional abuse used the short version of Tolman’s (1989) Psychological Maltreatment of Women Inventory. Physical violence questions were drawn from the Statistics Canada Violence in Marriage Scale (Section L, see Johnson & Sacco, 1995). Finally, Straus, Hamby, Boney-McCoy, and Sugarman’s (1996) Conflict Tactic Scales 2 (CTS2, Injury Scale only) was used to ask the women about the physical effects of their abuse. Finally, demographic and family of origin data were collected from each woman.2 Interviews lasted an average of 75 minutes.
Data analyses
Statistical analyses progressed in several phases. First, descriptive statistics were generated and reviewed. Then scales were developed by recoding conceptually related variables in the original data set (see Appendix A). For example, ‘children intervened’ included questions such as ‘how often did any of the children holler or yell at you or your partner while in the same room during an incident?’, ‘how often did any of the children get physically involved in an incident?’. The scales for children’s exposure and for their physical involvement in abuse were used as the dependent variables in this study.
Most demographic variables were originally provided in ordinal form and recoded into dichotomous variables to be used in multiple regression analyses (as shown in Table 3). A woman with a job at the time of the interview was given a value of ‘1’ for bivariate and multivariate analyses, and one without a job was given a value of ‘0’. The woman’s highest level of education achieved was recoded to be ‘1’ if she had at least some college degree or above, and ‘0’ if she had high school or GED as her highest education level. A woman residing in a shelter or living with friends would be assigned a value of ‘1’ for the variable ‘transitional housing’, while a woman renting or owning a home would be assigned a ‘0’ for this variable. Furthermore, a woman with an income of more than $10,000 a year received a value of ‘1’, and those with incomes less than $10,000 were given a value of ‘0’ in the analyses. A woman’s relationship to abuser was recoded into a value of ‘1’ if she was married to him, and a value of ‘0’ if she was not married to him. In addition, if the child(ren) were biologically related to the respondent or to the abuser, they were given a value of ‘1’, and they were not biologically related to him they then were given a value of ‘0’. The woman’s, children’s, or abuser’s racial groups remained categorical, as did the child’s living situation at the time of the abuse (see Table 3).
Comparisons were carried out within different groupings of demographic variables (e.g. relationships between women’s demographic variables, relationships between men’s demographic variables, relationships among children’s demographics variables). In addition, bivariate analyses were conducted between groupings of demographic variables (e.g. women’s education with children’s ages). Finally, comparisons of these demographic variables with dependent variables were conducted (e.g. comparing women’s, men’s, and children’s demographics to children’s exposure and physical involvement in abuse directed at their mothers).
The latter bivariate analyses (t-tests, ANOVA, and chi-squares) were reviewed to determine variables that were independently associated (p < .15) with types and levels of children’s involvement. This process was conducted in order to determine what other factors seem to be associated with children’s exposure and involvement with abuse towards their mothers. This cut off point was used following Hosmer and Lemeshow’s (1989) suggestion to be more liberal with bivariate significance levels of predictor variables to be included in later multivariate modeling using logistic regression analyses. Finally, the associated variables were included in multiple regression analyses to determine which elements statistically predicted variance in the multiple ways that children were exposed to and became involved with abuse inflicted upon their mothers.
The results discussed below are separated into two major categories: children’s exposure to violence and children’s intervention in incidents of violence. In each of the two sections, we present descriptive statistics followed by the results of bivariate and multivariate analyses.
Children's exposure to violence
Data on children’s level of exposure to violence was reported by their mothers, with each mother having on average two or three children. Overall, as seen in Table 2, this sample of battered mothers reported a high level of children’s exposure to violence. Seventy-eight percent (n=89) of the mothers reported that their children saw results of the abuse to their mothers at least occasionally, with 32% (n=36) reporting their children frequently saw results of abuse. Eighty-three percent (n=95) of the mothers reported their children at least occasionally overheard the abuse from another room, with 41% (n=47) reporting their children frequently overheard the abuse. Forty-five percent (n=51) reported their children came into the room where abuse was occurring at least occasionally, while 18% (n=21) reported that their children frequently came into the room. There were also some children who witnessed the beginning of an violent incident, but then left the room - 40% of mothers (n=46) reported their children saw abuse begin, then at least occasionally left the room while 12% (n=14) reported their children did so frequently. Lastly, 44% (n=50) of the mothers reported their children watched the entire abusive incident at least occasionally with 20% (n=23) reporting the children frequently watched the entire event.
Table 2: Descriptive statistics of variables within 'Children's exposure to violence'
| Variable | Value | Frequency | Percent | Valid Percent |
|---|---|---|---|---|
How often did children see results of an abusive incident?
|
|
|
|
|
How often did children hear an incident from a different room?
|
|
|
|
|
How often did children come into the room and watch some of incident?
|
|
|
|
|
How often did children watch an incident begin, then leave?
|
|
|
|
|
How often did the children watch an incident from beginning to end?
|
|
|
|
|
A scale labeled ‘Children Exposed’ was constructed and consisted of the following items that required mothers to rate their children’s exposure to their abuse, ranging from never, sometimes, occasionally, frequently, to very frequently:
Bivariate analyses were also carried out to compare mothers’ experiences of abuse with their children’s exposure to the violence. In general, the higher the level of the the mothers’ emotional abuse (r=.40, p =.001), physical abuse (r=.279, p=.003), physical effects (r=.326, p=.001), and emotional effects (r=.443, p=.001), the more she reported that her children observed her abuse.
Table 3: Comparison of women's, abuser's, and children's demographics with children exposure
Women's Demographics
| Independent Variable | Children Observed Mean (SD) | Statistic | p= |
|---|---|---|---|
| Age | r=-101 | .316 | |
Race/ethnicity
|
|
F=2.415 | .071* |
Relationship to abuser
|
|
t=-1.241 | .217 |
| Length of relationship | r=.024 | .815 | |
Education
|
|
t=-.543 | .588 |
Job Status
|
|
t=-2.216 | .029* |
Transitional Housing
|
|
t=.901 | .369 |
Income
|
|
t=-.572 | .569 |
Children's Demographics
| Independent Variable | Children Observed Mean (SD) | Statistic | p= |
|---|---|---|---|
| Years of Exposure | r=.059 | .572 | |
| Prop. of children who are girls | r=.113 | .243 | |
Race/ethnicity
|
|
F=1.672 | .162 |
Relationship to woman
|
|
t=.086 | .932 |
Relationship to abuser
|
|
t=.086 | .932 |
| Number of children | r=-.118 | .225 |
Abusers' Demographics
| Independent Variable | Children Observed Mean (SD) | Statistic | p= |
|---|---|---|---|
| Age | r=.070 | .491 | |
Race/ethnicity
|
|
F=.542 | .654 |
Children's living situation
|
|
F=1.376 | .257 |
* = significance at the .15 level
Upon completion of the bivariate analyses, multiple regression analyses were conducted to determine the degree to which a set of independent variables together statistically predicted children’s exposure to abuse of their mothers. The independent variables were entered into the regression in steps. In the first step, emotional abuse and physical abuse were entered alone. These two composite scales were entered first as a result of our interest in the variability in children’s exposure that could be accounted for by the mother’s direct experience with emotional and physical abuse.
In the next step, variables were entered into the analyses if they were conceptually related to the question of interest and if, in bivariate analysis, there was a statistically significant relationship at a p-value of at least .15. In the third step, the women’s ‘physical effects’ and ‘emotional effects’ were entered.
The dependent variable (measured by a composite scale) for this first multiple regression analysis was ‘children exposed’ (see Table 4). ‘Physical violence’ and ‘emotional abuse’ of the mother explained 17% of the observed variability in ‘children exposed’. In the second step of this regression model, the following independent variables were entered: (a) ‘woman’s job status’ and (b) ‘ethnic identity’. Separate dichotomous variables were created for women’s race in order to be entered in the multiple regression analyses, as the original variable was ordinal. Dichotomous dummy variables were created for ethnic identity to indicate the race of the woman, using Caucasian as the reference group (e.g. African-American: 1=yes, 0=no). Adding these two variables increased the percentage of explained variance to 23%. Lastly, ‘physical effects’ and ‘emotional effects’ increased the prediction to 30%. Overall, all the independent variables accounted for 30% of the variability in ‘children exposed’.
A mother being employed was associated with a decrease in the predicted value of ‘children exposed’ (positive ß value). In addition, being an African-American mother was associated with a decrease in children’s exposure to domestic violence relative to children of Caucasian mothers (negative ß value).
In the first step of the regression model, only ‘emotional abuse’ significantly contributed to the model (p = .002). Physical abuse did not (p = .422) even though it was significantly related to ‘children’s exposure’ in bivariate analyses. In the second step, ‘Latina’ and ‘other race’ did not significantly contribute to the model. In the third and final step (when ‘physical effects’ and ‘interference’ were entered), ‘emotional abuse’, ‘physical abuse’, ‘woman’s race’, and ‘physical effects’ did not significantly contribute to the model, leaving only ‘job status’ (p = .075) and ‘woman’s emotional effects’ (p = .012) as significant statistical predictors of variance in the multivariate model.
Table 4: Summary of Multiple Regression Analysis for Variables Predicting Children's Exposure (N=100)
| Independent Variable | B | SE B | ß | p |
|---|---|---|---|---|
Step 1
|
|
|
|
|
Step 2
|
|
|
|
|
Step 3
|
|
|
|
|
Note. R2 = .17 for Step 1; R2 = .23 for Step 2; R2 = .30 for Step 3 (p < .01).
* p < .15.
**p < .01.
A woman’s job status at the time of interview and the emotional effects of the violence on her during the abuse, were found to be significant statistical predictors of variation in a child’s earlier observation of domestic violence. These variables may, in fact, be proxies for two larger issues operating here. First, job status at time of interview may indicate the woman’s economic status, with greater poverty possibly being associated with smaller living quarters, more transient housing and less stability in the home. Second, as mothers report they were more emotionally affected by violence when it was occurring, it appears that their children were also more likely to be present during past violent events. Perhaps, as in prior studies, children increase their presence by offering support or seeking it as the conflict shows a greater impact on their mothers.
Children’s Intervention in Violence
As shown in Table 5, 52% (n=59) of the mothers reported that their children yelled from another room during abuse towards her at least occasionally, but 24% (n = 27) reported their children frequently yelled from another room. Fifty-three percent (n=60) of the mothers reported that their children at least occasionally yelled while in the same room, while 21% (n = 24) reported they did so frequently. In addition, 21% (n=24) of the mothers reported children called someone else for help during the abuse at least occasionally, and 6% (n=7) reported they did so frequently. Finally, 23% (n=26) of the mothers reported that their children became physically involved during an abusive incident involving the mother at least occasionally, and 8% (n=9) reported that their children physically intervened frequently.
Table 5: Descriptive statistices of variables within 'Children's intervention in violence'
| Independent Variable | Value | Frequency | Percent | Valid Percent |
|---|---|---|---|---|
How often did the children yell something from a different room?
|
|
|
|
|
How often did the children yell at your partner while in the same room?
|
|
|
|
|
How often did the children call for help during the incident?
|
|
|
|
|
How often did the children physically intervene?
|
|
|
|
|
‘Children intervened’, a scale measuring the frequency of children’s physical involvement with the abuse to their mothers, as reported by their mothers, was constructed using the following items:
Bivariate analyses focused on relationships between children’s, mothers’, and abusers’ demographics and how often children were physically involved with the violence towards their mothers (see Table 6). Children whose mothers were not employed at the time of the interview intervened significantly more, at the time of the abuse, than those whose mothers were employed (t = -1.709, p = .090). In addition, children of married mothers (at the time of the abuse) intervened significantly less (t = 2.067, p = .041) than children of unmarried mothers. Children whose mothers were more educated intervened less than those whose mothers were less educated (t = 1.771, p=.079). Furthermore, children of women in transitional housing (at the time of the interview) intervened significantly more (t = 3.561, p=.001). Children who were biologically related to the abuser intervened significantly less than those not biologically related to the abuser (t = 3.093, p =.003), and children living with the mother alone during the abuse or with the mother and abuser intervened significantly more than children who did not live with either the abuser or the mother (F = 2.207, p=.115). Finally, abuser’s age was significantly related to how often children physically intervened (r =.315, p=.001): the older the abuser, the more the child(ren) intervened.
Bivariate analyses were also carried out between the mother’s type of abuse experienced, its effects on her life and children’s intervention in the violence, to determine if the severity of the mother’s abuse and its effects on her were related to the intensity of children’s physical involvement. The higher the level of mothers’ emotional abuse (r=.305, p=.001), physical abuse (r=.542, p<.001), physical effects (r=.575, p<.001), and emotional effects (r=.247, p=.075), the more children’s were reported to be physically involved in violent incidents.
Table 6: Comparison of women's, abuser's, and children's demographics with 'children intervened'
Women's Demographics
| Independent Variable | Children Intervened Mean (SD) | Statistic | p= |
|---|---|---|---|
| Age | r=.110 | .269 | |
Race/ethnicity
|
|
F=1.170 | .325 |
Relationship to abuser
|
|
t=2.067 | .041* |
| Length of relationship | r=-.046 | .648 | |
Education
|
|
t1.771 | .079* |
Job status
|
|
t=-1.709 | .090* |
Transitional housing
|
|
t3.561 | .001** |
Income
|
|
t=-.717 | .475 |
Children's Demographics
| Independent Variable | Children Intervened Mean (SD) | Statistic | p= |
|---|---|---|---|
| Years of Exposure | r=-.008 | .940 | |
| Prop. of children who are girls | r=-.041 | .673 | |
Race/ethnicity
|
|
F=.568 | .687 |
Relationship to woman
|
|
t=-.071 | .943 |
Relationship to abuser
|
|
t=3.093 | .003** |
| Number of children | r=.049 | .606 |
Abusers' Demographics
| Independent Variable | Children Intervened Mean (SD) | Statistic | p= |
|---|---|---|---|
| Age | r=.315 | .001** | |
Race/ethnicity
|
|
F=1.293 | .281 |
All groups
| Independent Variable | Children Intervened Mean (SD) | Statistic | p= |
|---|---|---|---|
Children's living situation
|
|
F=2.207 | .115* |
* = significance at the .15 level
** = significance at the .01 level
The dependent variable for this multiple regression analysis was ‘how often children intervened in abuse directed towards their mother’ or ‘children intervened’ (see Table 7). In the first step, emotional abuse and physical abuse were entered alone because of our interest in the variability in children’s exposure that could be accounted for by the mother’s direct experience with emotional and physical abuse. In the second step of this regression, the folloependent variables were entered: (a) ‘children’s living situation at the time of the abuse’, (b) ‘abuser’s age at onset of abuse’, (c) ‘woman’s job status at the time of the interview’, and (d) ‘children’s relationship to abuser’. ‘Children’s living situation’ was recoded into several dichotos0 flskdkdkder to be entered into the multiple regression (similar to ‘women’s race’ analysis) as it was originally an ordinal variable. Separate dichotomous variables were created for children’s living situations, as the original variable was ordinal. A set of dichotomous dummy variables was created for children’s living situation to indicate with whom the child was living, using the child living with both the mother and abuser as the reference group (e.g. child living with mother only: 1=yes, 0=no).
Table 7 reveals that 29% of the observed variability in ‘children intervened’ was explained by ‘emotional abuse’ and ‘physical violence’ directed at their mothers. Adding ‘abuser’s age at onset of abuse’, ‘woman’s job status’, ‘children’s relationship to abuser’, ‘children’s living situation’, and ‘woman’s relationship to abuser’ increased the percentage to 45%. In the final step, ‘physical effects on respondent’ and ‘woman’s emotional effects’ increased the statistical prediction to 47%. Overall, all independent variables accounted for 47% of the variability in ‘children intervened’.
A mother being employed was associated with an decrease in the predicted value of children intervened (positive ß value). In addition, abuser’s age at onset of abuse was positively associated with increased values of the dependent or outcome variable (positive ß value). Children who were biologically related to the abuser, a woman who was married to her abuser, and children living with either their mother only or with neither the mother nor the father (compared to living with both) were all associated with a decrease in children’s intervention in violent events (negative ß value).
In the first step of the model, only ‘physical abuse’ significantly contributed to the model (p. < .001). ‘Emotional abuse’ did not significantly contribute to the model (p = .501), although it was significantly related to ‘children intervened’ in the bivariate analyses. In the second step, ‘job status’, ‘woman’s relationship to abuser’, and ‘emotional abuse’ did not significantly contribute to the model. In the final step (when ‘physical effects’ and ‘interference’ were entered), ‘emotional abuse’, ‘job status’, ‘woman’s relationship to abuser’, and ‘woman’s physical effects’ did not significantly contribute to the model. Variables that remained significant contributors to the model were ‘physical violence’ (p=.012), ‘physical effects’ (p=.132), ‘children’s living situation’ (p=.003), ‘abuser’s age at onset of abuse’ (p=.003), and ‘children’s relationship to abuser’ (p=.039).
Table 7: Summary of Multiple Regression Analysis for Variables Predicting Children's Physical Involvement in Their Mother's Abuse (N=100)
| Variable | B | SE B | ß | p |
|---|---|---|---|---|
Step 1
|
|
|
|
|
Step 2
|
|
|
|
|
Step 3
|
|
|
|
|
Note: R2 = .29 for Step 1; R2 = .47 for Step 3 (p<.01).
* p < .15
** p < .01
These data reveal a cluster of variables that appear to partially predict children’s involvement in violent events. It appears that the greater the violence and its effects on their mothers, the more likely children are reported to intervene. Also, the older the abuser, the more likely the children are reported to intervene. And finally, children living with neither the abuser nor their mother were less likely to be reported to have intervened in the violence while children not biologically related to the abuser were more likely to have intervened.
This study had several limitations. First, most of the women who called our toll-free number were either in a shelter, or affiliated with one. Therefore the sample represents women and children who have most likely escaped severe violence, and who are probably faced with multiple other stressors during this relocation. Secondly, mothers - not the children themselves - reported on children’s exposure and interventions. Previous studies have found that mothers’ reports of their children’s exposures to her abuse are under-reported when compared to their children’s reports (see O’Brien, John, Margolin & Erel, 1994). Finally, this study asked the women about their economic and housing status at the time of the interview, while asking them about abuse that occurred in the past (at least one year prior to the interview). Future research on this subject should be careful to assess women’s income, education, and housing status at both the time of the abuse and after the abuse had ended. It should also attempt to determine if these factors persist over time, from the period of abuse to the period after the mother is no longer in a relationship with the abuser.
The current study used some unique data collection methods that investigated urgently needed knowledge of battered women and their children’s experiences with domestic violence. This study is one of few to have systematically gathered detailed information from mothers about the nature of their children’s observation and involvement in incidents of adult domestic assault. Contrary to earlier research that showed many mothers unaware of their children’s exposure to violent incidents (e.g. O’Brien, John, Margolin & Erel, 1994), almost all of the mothers in this sample reported that children living with them during abusive incidents were highly exposed to the violence. A little over half of the mothers reported that children who heard the violent incident occurring went to the room where it was taking place. A large proportion of the children, who were reported to be in the room where violent incidents began, left the room once it began.
This study also showed some mothers reporting their children intervening in violent incidents. Just under a quarter of the mothers reported that their children called someone for help during the abuse, and a quarter reported their children were physically involved in the abuse directed at her. We also know that, in general, the more the mother was being abused and the more the abuse interfered with her life and her physical health, the more she reported her children were exposed to and intervened in the violence.
These findings have important implications for domestic violence programs, child protection agencies and the criminal justice system. First, it is clear that child exposure varied and requires careful assessment. Second, the level of physical involvement by some children in violent events raises serious concerns for their safety and that of their mothers. Much greater attention must be given to the safety of both adult victims and children in the home. Family stability was a factor in predicting a child’s exposure to and involvement in violent incidents. Children whose mothers were not employed at the time of the interview were reported to be exposed to more violence and intervene more often in the violence than those whose mothers were employed. In addition, children whose mothers were married, more educated, not living in transitional housing at the time of interview were reported to have intervened in the violence significantly less often than children whose mothers were not married, less educated, and living in transitional housing. The particular sample recruited for this study may have resulted in a disproportionate number of women who required economic assistance. Almost half the women in the study (45%) were resident in a battered women’s shelter at the time of the interview, with another 8% living in a subsidized apartment. This high proportion of shelter residents is most likely due to the fact that local battered women’s programs facilitated recruitment. While our only data on many of these variables was at the point of interview, it does seem that providing services that increase these battered mothers’ stability may have both indirect and direct links to lessening children’s involvement in violent incidents. This points to the possible need for much greater attention to economic assistance for mothers as a way of providing greater stability and, hopefully, safety to their children.
Children who were biologically related to the abuser were reported by their mothers to have intervened significantly less often than children not biologically related to the abuser. This finding may be supported by prior studies that have reported the presence of children fathered by previous male partners put women at greater risk of being abused (Daly, Singh & Wilson, 1993) and the presence of a step-parent put children at greater risk of being abused (Wilson & Daly, 1987). Children and abusive men may both have more of an investment in maintaining their relationships when there is a longer term tie to each other. Finally, the older the abuser, the more the children were reported to have intervened. This may be associated with older abusers being more likely to have older children in the household and older children feeling they have the size and strength to successfully intervene and stop the abuse.
Overall, a large proportion of the children in this study were exposed to their mothers’ experience with violence, and a quarter of the mothers in this sample reported their children were physically involved in the events. In addition, children in this sample were reported to be exposed and intervene less when their mothers’ lives at the time of interview were more stable (i.e. stable employment, housing). If this same situation held true during the abusive period reported by mothers, these findings might have resulted from less contact between an employed mother and her abuser, homes with more generous physical space and stable social networks offering children alternatives for finding safety and privacy, or families with fewer other issues (e.g. need for food, shelter, transportation and housing) complicating their lives.
Ending the violent events in these homes should be our foremost goal. Children and their mothers will be safe if the violence is ended or removed from their environments. But in achieving this goal, the results of this study point to both potential risk factors and interventions to enhance the safety of both children and mothers. First, children’s experiences clearly varied a great deal, indicating that some may have been at risk for greater harm while many others may not have been at risk. Second, unlike prior research on children’s involvement, factors associated with basic economic need appear to play a larger role than earlier thought. The complexity of children’s experiences and the environments in which they respond to violent incidents must be carefully considered and, where possible, corrected. These results point to the need for more careful assessments and a wider variety of services for battered mothers and their children if we are to seriously impact their safety and lessen children’s involvement in incidents of adult domestic violence.
1. It was difficult to include children’s average age during the abuse since women were asked about all children as a group. This made it impossible to distinguish if a certain outcome applied, for example, to a three-year-old or a 15-year-old in a family. In addition, out of 285 children who were present by the end of abuse, 73 were not born yet at the ‘onset of abuse’. Return to your place in the report.
2. Copies of the Interview Guide are available by writing to MINCAVA - the Minnesota Center Against Violence and Abuse, School of Social Work, University of Minnesota, 105 Peters Hall, 1404 Gortner Avenue, St. Paul, MN 55108.Return to your place in the reports.
Adamson, J.L. & Thompson, R.A. (1998). Coping with interparental verbal conflict by children exposed to spouse abuse and children from nonviolent homes. Journal of Family Violence, 13, 213-232.
Appel, A.E. & Holden, G.W. (1998). The co-occurrence of spouse and physical child abuse: A review and appraisal. Journal of Family Psychology, 12, 578-599.
Cummings, E.M., Zahn-Waxler, C. & Radke-Yarrow, M. (1981). Young children’s responses to expressions of anger and affection by others in the family. Child Development, 52, 1274-1282.
Cummings, J.S., Pellegrini, D.S., Notarius, C.I. & Cummings, E.M. (1989). Children’s responses to angry adult behavior as a function of marital distress and history of interparental hostility. Child Development, 60, 1035-1043.
Daly, M., Singh, L.S. & Wilson, M. (1993). Children fathered by previous partners: A risk factor for violence against women. Canadian Journal of Public Health, 84, 209-210.
Edleson, J.L. (1991). Social Workers' intervention in woman abuse: 1907-1945. Social Service Review, 65, 304-313.
Edleson, J.L. (1999a). The overlap between child maltreatment and woman battering. Violence Against Women, 5(2), 134-154.
Edleson, J.L. (1999b). Children's witnessing of adult domestic violence. Journal of Interpersonal Violence, 14(8), 839-870.
Fantuzzo, J.W. & Lindquist, C.U. (1989). The effects of observing conjugal violence on children: A review and analysis of research methodology. Journal of Family Violence, 4, 77-94.
Fantuzzo, J.W. & Mohr, W. K. (1999). Prevalence and effects of child exposure to domestic violence. The Future of Children, 9, 21-32.
Garcia O’Hearn, H., Margolin, G. & John, R.S. (1997). Mothers’ and fathers’ reports of children’s reactions to naturalistic marital conflict. Journal of the American Academy of Child and Adolescent Psychiatry, 36, 1366-1373.
Gordon, L. (1988). Heroes of their own lives. New York: Penguin Books.
Henning, K., Leitenberg, H., Coffey, P., Turner, T. & Bennett, R.T. (1996). Long-term psychological and social impact of witnessing physical conflict between parents. Journal of Interpersonal Violence, 11, 35-51.
Holtzworth-Munroe, A., Smutzler, N. & Sandin, B. (1997). A brief review of the research on husband violence. Part II: The psychological effects of husband violence on battered women and their children. Aggression and Violent Behavior, 2, 179-213.
Hosmer, D.W., & Lemeshow, S. (1989). Applied logistic regression. New York: Wiley.
Hotaling, G.T., Straus, M.A. & Lincoln, A.J. (1990). Intrafamily violence and crime and violence outside the family. In M.A. Straus & R.J. Gelles (Eds.). Physical violence in American families. New Brunswick, NJ: Transaction Publishers.
Hughes, H. M. (1988). Psychological and behavioral correlates of family violence in child witness and victims. American Journal of Orthopsychiatry, 58, 77-90.
Jaffe, P.G. & Suderman, M. (1995). Child witnesses of woman abuse: Research and community responses. In S.M. Stith & M.A. Straus (Eds.), Understanding partner violence (pp. 213-222). Minneapolis, MN: National Council on Family Relations.
Johnson, H. & Sacco, V. (1995). Researching violence against women: Statistics Canada’s National Survey. Canadian Journal of Criminology, 37, 281-304.
Kashani, J.H., Daniel, A.E., Dandoy, A.C. and Holcomb, W.R. (1992). Family violence: Impact on children. Journal of the American Academy of Child and Adolescent Psychiatry, 31, 181-189.
Margolin, G. (1998). Effects of witnessing violence on children. In P.K. Trickett and C.J. Schellenbach (Eds.). Violence against children in the family and the community (pp. 57-101). Washington, D.C.: American Psychological Association.
Minn. State Ann. 626.556
Norusis, M.J. (1999). SPSS 9.0: Guide to data analysis. Upper Saddle River, NJ: Prentice Hall.
O'Brien, M.; John, R.S.; Margolin, G.; Erel, O. (1994). Reliability and diagnostic efficacy of parents' reports regarding children's exposure to marital aggression. Violence & Victims. 9, 45-62.
Peled, E. (1998). The experience of living with violence for preadolescent children of battered women. Youth & Society, 29, 395-430.
Peled, E. & Davis, D. (1995). Groupwork with children of battered women. Thousand Oaks, CA: Sage.
Pleck, E.H. (1987). Domestic tyranny: The making of social policy against family violence from Colonial times to present. New York: Oxford University Press.
Rossman, B.B.R. (In press). Long term effects of exposure to adult domestic violence. In Graham-Bermann, S.A. and Edleson, J.L. (Eds.) Domestic violence in the lives of children: The future of research, intervention, and social policy. Washington, DC: American Psychological Association.
Rossman, B.B.R., Hughes, H.M. & Rosenberg, M.S. (2000). Children and interparental violence: The impact of exposure. Philadelphia, PA: Brunner/Mazel.
Silvern, L., Karyl, J., Waelde, L., Hodges, W.F., Starek, J., Heidt, E. & Min, Kyung (1995). Retrospective reports of parental partner abuse: Relationships to depression, trauma symptoms and self-esteem among college students. Journal of Family Violence, 10, 177-202.
Straus, M.A., Hamby, S.L., Boney-McCoy, S. & Sugarman, D.B. (1996). The revised Conflict Tactics Scales (CTS2): Development and preliminary psychometric data. Journal of Family Issues, 17, 283-316.
Tolman, R.M. & Rosen, D. (2000). Domestic violence in the lives of women receiving welfare: mental health, substance dependence, and economic well-being. Violence Against Women, 7, 141-158.
Utah Code Ann. 76-5-109.1
Wilson, M. & Daly, M. (1987). Risk of maltreatment of children living with stepparents. In Gelles, R.J. & Lancaster, J.B. (Eds.) Child abuse and neglect: Biosocial dimensions (pp. 215-232). NY: Aldine de Gruyter.
Scale guide
| Scale Item | Types of questions |
|---|---|
| Emotional and Verbal abuse | How often abuser called the woman names, swore at her, yelled and screamed at her, treated her like an inferior, monitored her time, used her money or didn’t consult with her in making important decisions, was jealous or suspicious of the woman’s friends, accused her of having an affair, interfered in her relationships with other family members, tried to keep her from doing things to help herself, blamed her for his problems, restricted use of telephone, and tried to make her feel crazy? |
| Physical abuse | How often did the abuser threaten to hit the woman with something, threw anything at her, pushed, grabbed, or shoved her, slapped her, Kicked, bit or hit her with his fist, beat her up, choked her, threatened to use or used a gun or knife, or forced her into any sexual activity she didn’t want? |
| Physical effects on the woman | How often did the woman have a sprain, bruise, cut, pass out from being hit on the head, saw a doctor, have a broken bone from a fight with her abusive partner? |
| Interference with the woman’s life | How often did the abuse affect the woman’s ability to parent the way she would like to, affect her ability to do what was needed around house, and maintain relationships with family and friends? |
| Children intervened | How often did children holler or yell from a different room during an incident, how often did they yell from the same room, called for help, or physically intervene in the abuse towards their mom? |
| Children exposed to the abuse | How often did children see results of abuse, hear the incident from another room, hear and see it from the same room and then stay or leave? |
File Last Modified:
Monday, 14-Mar-2005 11:57:20 CST
© Copyright 2000 - 2004 Link Resource Project,
Minnesota Center Against Violence and Abuse
This website is sponsored by the Link Research Project, a project of the Minnesota Center Against Violence and Abuse. Both the website and the Link Research Project were established with grants from the Allina Foundation (Minnesota) for the first phase of the project, and by The David and Lucile Packard Foundation (California) for the second phase. For more information or to contact project staff, please contact us via the MINCAVA email submission form.