The Future of Children, Princeton - Brookings: Providing research and analysis to promote effective policies and programs for children.

Although no databases provide reliable prevalence estimates, research findings to date underscore that domestic violence occurs in large numbers of households with children. Existing data sources, including national crime reports and population-based surveys, are limited in a number of ways. Crime reports, though theoretically important vehicles for verifying the occurrence of domestic violence, are subject to differences in legal definitions for domestic violence, police protocols for reporting, and the training and technological sophistication of the police officer handling the call. Population-based surveys use a clear set of definitions to collect data, but are limited by their reliance on retrospective accounts of the violence by survey participants.¹² The Spousal Assault Replication Program, though not a national research effort, addresses some of the weaknesses of these other data sources by providing substantiated data collected by law enforcement officers, and using explicit definitions for domestic violence and child exposure to it, as well as standardized research methods. This study holds promise as a model for how the field can move toward building a more credible national prevalence database.

Domestic violence is a crime as well as a public health problem. Criminal codes have been revised in recent years to broaden the categories of activities that are considered domestic violence and to strengthen the authority of police officers to intervene in violent or potentially violent situations. All states have passed some form of domestic violence legislation providing civil as well as criminal penalties for acts of violence within the home.¹³ (See the article in this issue by Matthews for a discussion of such legislation.)

Since 1929, the Uniform Crime Reports (UCR) system has required local and state law enforcement authorities to aggregate the number of criminal incidents by offense type and to report these totals to the FBI. Information on eight broad categories of crime—homicide, forcible rape, robbery, aggravated assault, burglary, larceny-theft, motor vehicle theft, and arson—is collected. Unfortunately, the UCR system does not provide specific information on domestic violence or detailed demographic information on victims and offenders. For example, 1997 data from Philadelphia indicated that there were 1.6 million 911 calls; approximately 200,000 indicated a possible domestic violence situation. However, the limitations of this database make it impossible to verify that these were in fact domestic violence incidents, or to distinguish the characteristics of the incidents.¹⁴

A new system, the National Incident Based Reporting System (NIBRS), was designed to provide more detailed national crime data. The NIBRS contains data on 57 types of crimes, including domestic violence related crimes.¹⁵ The crime categories that pertain to domestic violence include: (1) assault offenses (aggravated assault, simple assault, and intimidation); (2) forcible rape; (3) nonforcible rape; (4) disorderly conduct; and (5) family offenses, nonviolent. The NIBRS also collects detailed information on the particular incidents, as well as victim-offender demographics, victim-offender relationship, time and place of occurrence, weapon use, and victim injuries. Under the NIBRS, law enforcement personnel are required to use standard forms to collect information on each crime occurrence and its surrounding circumstances. This promising system does not yet provide national data on reported domestic violence crimes because it is fully operational only in Austin, Texas. In addition, NIBRS collects data only on crime victims; it does not provide information about other individuals who were nearby, or exposed to, the violent incident, unless those individuals are also victims of the crime. Only children who are physically victimized by a domestic violence incident are considered victims under this system. Therefore, although NIBRS will contribute to knowledge about the prevalence of domestic violence, it will not provide information about the exposure of children to domestic violence.

The National Family Violence Survey (NFVS) and the National Crime Victimization Survey (NCVS), national telephone surveys of representative samples of households, are currently the primary sources of information on violence between intimate adults in the United States. The NFVS, conducted in 1979 and 1987, used the Conflict Tactics Scale (CTS), an 18-item questionnaire that asks respondents to indicate the number of times in the previous year that an intimate partner committed a particular verbal or physical action against them during a conflict in their relationship.¹⁶ The CTS measures verbal aggression and physical violence, but does not identify sexual and psychological abuse, which have been found to be very important aspects of domestic violence. The CTS has other limitations as well.¹⁷ For example, the questionnaire does not inquire directly about children's exposure to the violence.

The 1987 NFVS sampled 6,000 households by telephone interviews. This survey indicated that 116 per 1,000 women reported experiencing some form of physical or verbal aggression by an intimate partner in one year, and 44 per 1,000 women reported that they had engaged in some form of physical or verbal aggression toward their male partners in that year.¹⁸ Thirty-four per 1,000 women surveyed reported that they had experienced severe violence at the hands of their intimate male partners. Although the NFVS did not gather data on child exposure to domestic violence, survey results have been used to estimate the prevalence of child exposure to be at least 3.3 million annually.¹⁹ This figure has been cited by researchers and policymakers as if it were a fact, even though the data on which it is based are 20 years old, and the study sample did not include families with divorced parents or children under age three.²⁰

The NCVS, conducted by the U.S. Department of Justice, is designed to provide annual estimates of crimes experienced by the public at large. Telephone survey data on the frequency, characteristics, and consequences of criminal victimization are collected from a sample of approximately 49,000 households.¹⁵ The NCVS provides data on domestic violence and includes questions about whether children are living in the victim's household. Respondents are first asked if they experienced a major crime during the previous year. If they have, they are then asked for details, including household demographics and information about the perpetrator and the perpetrator's relationship to the victim, where the crime occurred, whether a weapon was used, and what actions the victim took subsequent to the crime. The household demographics specify who was living in the household at the time of the crime and the relationship of those persons to the victim. These are the only national data available to help determine if a child was present in the victim's household when domestic violence occurred. The data, however, do not describe the nature of the child's exposure to the crime. Based on these data, the NCVS reports that the annual rate in 1993 of physical attacks against women by intimate partners was 9.3 per 1,000, and that children under the age of 12 reside in slightly more than half of the affected households.¹⁵

In 1995 and 1996, the National Institute of Justice and the Centers for Disease Control and Prevention co-sponsored the National Violence Against Women Survey. A sample of 8,000 men and 8,000 women was drawn from random-digit telephone dialing to households in all 50 states and the District of Columbia.²¹ This survey posed questions to respondents about experiences with violent victimization, using a modified version of the Conflict Tactics Scale that included items regarding physical assault as children by adult caretakers, physical assault experienced as adults, and queries about rape and stalking. Findings indicated that in the United States approximately 1.5 million women and 834,700 men are raped and/or physically assaulted by an intimate partner annually.

One major multicity research effort sponsored by the National Institute of Justice, the Spousal Assault Replication Program (SARP), addresses some of the weaknesses of other data sources by providing substantiated data on children exposed to family violence, collected by law enforcement officers and researchers using standard methods.⁵ The SARP database was derived from investigations, in several U.S. cities, of carefully selected misdemeanor domestic violence cases.⁵ These investigations represent a partnership between law enforcement and university researchers, who collected detailed information about the violent incidents, persons present in the household at the time of the incident, who placed the police call, and who else was assaulted. Data relevant to risk factors associated with domestic violence (for example, substance use and poverty) were obtained as well. A subsequent analysis of the SARP database examined information relevant to children's exposure to domestic violence and addressed the following questions: (1) Are children disproportionately represented in households with substantiated cases of adult female abuse? (2) Are younger children disproportionately present in households in which domestic violence occurs? (3) Do other factors that pose developmental risks to children occur disproportionately in these households? (4) To what degree are children who live in households with domestic violence involved, in some way, in the incidents of violence?

Findings revealed that, in all five cities studied, children were present in the households of the domestic violence group at more than twice the rate they were present in comparable households in the general population.⁵ In Milwaukee, Wisconsin, for example, 81% of the households in the SARP database had children present, whereas only 32% of the comparison households included children. Moreover, children under the age of five were more likely to be present in the homes in which domestic violence occurred; in Milwaukee, 48% of the SARP households with children had children under age five, whereas this was true for only 31% of the comparison households with children. Furthermore, children ages five and under were more likely than older children to be exposed to multiple incidents of domestic violence over a six-month period, and to parental substance abuse. Of the 633 children included in the Charlotte, North Carolina, data, 42% of those ages five and under had experienced multiple incidents of domestic violence, compared to 27% of the children ages 6 through 11, and 21% of the children ages 12 through 18. In Omaha, Nebraska, 14% of the children five years old or younger had experienced parental substance abuse, compared to 10% of the children ages 6 through 11, and 6% of the children ages 12 through 18. Other well-known risk factors, such as poverty, status as a single-parent household headed by a female, and a primary care provider with a low educational level, were also more likely to be present in the homes in which domestic violence occurred. For example, data from Atlanta, Georgia, revealed that 79% of the children in the SARP households were living in poverty, whereas this was true for only 16% of the children in comparison households; and 51% of the SARP households were single-parent households headed by a female, compared to 24% of the comparison families.⁵ Overall, these data suggest that those children who are most dependent on their caregivers are most vulnerable to witnessing serious domestic violence, and are additionally threatened by a host of other developmental risk factors. These co-occurring risk factors also complicate efforts to identify the unique developmental consequences of exposure to family violence.

Many children in these violent households appeared to be involved in the abuse incidents in various ways. For example, adult victims indicated that children somehow influenced the onset of the violent incident in about 20% of the households. In the two cities that tracked the identity of the persons placing the 911 call, children were found to have placed 10% of the calls. In the two cities that recorded the co-occurrence of child physical abuse with a domestic violence incident, child abuse was present in 6% of the incidents. These findings indicate that children in households with domestic violence are not just "witnessing" a tragedy; often they are a part, or are perceived by the adults to be a part, of the incidents in some way.⁵

The SARP study provides a unique and promising approach to collecting data on children's exposure to domestic violence. The collaboration between police and university researchers allows for a direct sampling of substantiated cases. The data are collected at the time of, and immediately following, the incident, thus avoiding the problems of retrospective reports. Participants are not selected from limited specialized settings like domestic violence shelters. In addition, identical data collection methods are used across cities, facilitating cross-site comparisons of data. The study is limited, however, by its focus on a small number of cities that may not have child populations generally representative of children in the United States. In addition, felony domestic violence cases were not included in the study, and there were many gaps in the data collected on children.²²