CONTEXT. Witnessing domestic violence increases a child's chance of emotional or behavioral problems during childhood and entering abusive relationships in adulthood, even without co-occurring child maltreatment.
OBJECTIVE. Our goals were to estimate the prevalence of reported violent disagreements in the homes of US children and to assess prevalence differences by race/ethnicity, residence, and reported parenting stress.
PATIENTS AND METHODS. Data were drawn from the 2003 National Survey of Children's Health. Case subjects with unknown gender, race/ethnicity, or residence were excluded, yielding 99660 observations. Disagreements were classified on the basis of how the family deals with serious disagreement. If disagreements involved hitting or throwing, even rarely, the household was categorized as having violent disagreements. Households reporting heated argument and shouting were classified as having heated disagreement.
RESULTS. Nationally, 10.3% of children lived in homes with reported violent disagreements. Violent disagreements were most prevalent among black households (15.1%), followed by “other” (12.1%), Hispanic (11.3%), and white (8.6%) households. Urban areas had higher prevalence (10.7%) than did small through large rural counties (8.3%–9.9%). In multinomial logistic analysis, parents living in rural counties were less likely to report violent disagreements compared with those in urban. Black children were more likely to be exposed to both violent and heated disagreements than were white children. Parents reporting high parenting stress had higher odds of violent and heated disagreement than parents reporting less stress.
CONCLUSIONS. A substantial number of children are exposed to violent disagreement. Although demographic and cultural factors may also influence disagreement style, parental stress seems instrumental in the development of violent disagreements. Parents who experience difficulty with parenting constitute a high-risk population. Helping parents understand and address child behavior may reduce such stress.
Witnessing domestic violence increases a child's risk for emotional and behavioral problems1–4 and may be associated with adverse physical outcomes.5 Assessing the prevalence of childhood exposure to violence in the home, however, is difficult. The National Survey of Adolescents found that about one third of white children (34%), 57% of black children, and 50% of Hispanic children had witnessed violence in their lifetimes, whether in the home or in the community.6 Among a small sample of inner-city 6- and 7-year-old children, nearly one third reported having seen adults in their home “hit each other.”4
The lifetime estimated prevalence of violence against women found by the National Violence Against Women Survey (NVAWS) was 25.5%.7 Annual prevalence rates for interpersonal violence, defined as reported physical harm, are relatively low, ranging from ∼2%7–9 to 6% in population-based surveys.10 Similarly, a survey of urban mothers found a domestic violence prevalence of 3.7%.11
Development of national estimates of the prevalence of childhood exposure to violence is complicated by the variety of methods used to study violence. Many studies, such as the NVAWS, link violence to injury or examine children's exposure to stabbing, shooting, and killing.12 Research focusing on injury may be hampered by underreporting and neglects symbolic violence, such as threats or violence against objects. Symbolic violence has been related to subsequent physical violence.13 In addition, children can perceive symbolic behaviors, such as shouting, as a form of violence.14 Development of estimates is further complicated, because many studies have been conducted in patient care settings in small geographic areas. 15–18Diversity across instruments, settings, and populations makes it difficult to draw accurate conclusions.
Our research sought to develop national estimates of the prevalence of violent disagreements in the homes of US children and to assess prevalence differences by residence, race/ethnicity, and parental stress. Rural residence was studied because access barriers for domestic violence services among rural women may increase violence exposure among rural children.18,19 Racial minorities could be similarly vulnerable. Parental stress was conceptualized as potentially modifiable through community or practitioner intervention.
Data were drawn from the 2003 National Survey of Children's Health (NSCH), a telephone survey assessing physical, emotional, and behavioral health indicators and measures of children's experiences with the health care system.20 Children with unknown gender, race/ethnicity, or county of residence were excluded, yielding 99660 observations. To maintain confidentiality when using small geographic units, all analyses were conducted at the Research Data Center of the National Center for Health Statistics in Hyattsville, Maryland. This study was granted exempt status by the University of South Carolina's Institutional Review Board.
Definition of Variables
Violent disagreements were identified by using 3 questions addressing conflict style: “When you have a serious disagreement with your family members, how often do you (1) … discuss your disagreements calmly; (2) … argue heatedly or shout; (3) … end up hitting or throwing things.”21 We defined 3 levels of disagreement violence:
If a respondent indicated that disagreements involved hitting or throwing, even rarely, the household was categorized as having violent disagreements. We used an inclusive cutoff, reasoning that even a single incident in which a child experiences or witnesses violence contains the possibility of harm.
If the respondent did not hit or throw, but reported heated argument and shouting sometimes, usually, or always, the household was classified as having heated disagreement. This differs from previous research, which did not categorize shouting as violence.22
If the respondent did not hit or throw and only rarely reported shouting, the household was classified as calm.
Following the National Center for Health Statistics, we used 4 categories of race/ethnicity: Hispanic, non-Hispanic white, non-Hispanic black or African American, and non-Hispanic other. To address acculturation differences across immigrant families, we included language used at home (English/other).
Residence was defined at the county level by using Rural Urban Continuum Codes (RUCCs) from the 2003 Area Resource Files. Counties were classified as urban (RUCCs 1–3), large rural (RUCCs 4–5), medium rural (RUCCs 6–7), and small rural (RUCCs 8–9).
The NSCH parental stress questions were derived from the Parental Stress Index23 and the Parental Attitudes about Childrearing scale.24 Three questions asked how often during the past month the parent had felt the child “was much harder to care for than children his/her age,” “did anything that really bothers you a lot,” and “felt angry with him/her.” Responses used a 4-point scale from never through always. Scores ranged from 3 through 12, and the distribution was positively skewed. Respondents were divided into high versus low stress, dichotomized at a score of 5, the 75th percentile.
A number of demographic characteristics potentially associated with exposure to violence were included in the analysis as possible confounders. Demographic characteristics of the child included age, gender, and reported health status. The child's health insurance status was included, because access to care is associated with parent stress.25 Characteristics of the parents included education, employment, and parental health. Perceived neighborhood support for parenting was measured with 4 items pertaining to neighbors’ willingness to help.26 Parents who responded negatively to ≥2 statements were classified as lacking support. The respondent's relationship to the child was held constant, because it was linked to reporting of violent disagreements. Characteristics of the family included language (noted earlier), income, region, family structure, number of children, and family mobility (number of household moves divided by child's age). Ecological factors included percent poverty and percent owner-occupied housing, measured at the county level.
We used SAS Callable SUDAAN 9.027 for all analyses, reflecting the complex sampling design. χ2 analyses were used to test bivariate associations between independent variables and disagreement style. Multivariable generalized logistic regression models28 were conducted in which the risks of violent and heated disagreement were compared with calm discussion.29 All testing was 2-sided (α = .05).
US Children, By Residence
Because little published information describes rural children, complete descriptive statistics on risk factors by residence are provided (Table 1). Most (79.5% [56.1 million]) US children lived in urban areas, with 6.3% (4.5 million) residing in large rural counties, 11.6% (8.2 million) in midsized counties, and the remaining 2.5% (1.8 million) living in the smallest rural counties. Nonwhite children and children living in families where the primary language was not English represented a larger proportion of the population in metropolitan than in rural counties. Most other indicators of disadvantage, including poverty, lack of private health insurance, low parental education, unemployment, and fair-to-poor parental health were higher among rural children than their counterparts.
Prevalence of Violent Disagreements in the Home
Nationally, 10.3% of children lived in homes with violent disagreements. An additional 31.5% of children lived in homes where disagreements entailed heated argument and shouting (Table 2). Potentially violent disagreement was slightly more common in urban than in rural counties. White parents were least likely to report violent or heated disagreement styles, whereas black parents had the highest unadjusted prevalence of these behaviors (Table 2).
Parental stress and disagreement style were closely related, with high-stress parents being markedly more likely to report hitting and throwing or heated arguing and shouting (Table 2). The proportion of parents reporting high parental stress was slightly lower in rural counties than in urban. High stress was more common among minority than white parents, with Hispanic parents being most likely to report high stress (33.3%), followed by black parents (29.3%), parents of other racial groups (27.6%), and white parents (22.6%).
Multiple factors were associated with disagreement style in bivariate analysis (Table 2). Children reported to be in fair-to-poor health were markedly more likely to live in households with violent disagreement than those in better health (17.1% vs 10.1%). Similarly, the prevalence of violent disagreements was higher among families in which 1 or more parents reported fair or poor health (15.0%) than others (9.6%). Fathers were less likely to report the presence of violent or heated disagreements than were mothers. Fathers responded in 15.4% of all surveys, mothers in 80.2%, and other guardians in 4.4%.
Family structure and size were linked to disagreement style. Single mother homes, had a markedly higher likelihood violent disagreement (odds ratio [OR]:1.95, calculated from values in Table 2) than 2-parent (biological) homes. When the number of children in the household increased, so did the reported prevalence of violent disagreements. Resources, both educational and financial, were inversely related to the prevalence of violent disagreement.
Multivariable Analysis: Effects of Race/Ethnicity, Residence, and Parental Stress
Generalized logistic regression was used to distinguish the effects of residence, race/ethnicity, and parental stress on the likelihood of violent disagreements in the home while holding other factors constant (Table 3). All children living in non-metropolitan statistical area counties were considered rural in this analysis.
Rural households, holding other characteristics constant, had reduced odds for violent disagreement when compared with those in urban areas (OR: 0.86; 95% confidence interval [CI]: 0.77 to 0.95). There was no rural effect regarding heated argument. Effects did not vary across levels of rurality; when the analysis was restricted to rural counties alone, residence was not statistically significant (data not shown).
In adjusted analysis, black and “other” children remained significantly more likely than white children to live in homes with reported violent disagreements (black children, OR: 1.73; other children, OR: 1.38; Table 3). Similarly, black and other children were at increased risk, compared with white children, of experiencing heated argument and shouting.
Hispanic households did not differ from white households. Low English fluency, positively associated with violent disagreements in bivariate analysis, was not significant with other family characteristics held constant and was negatively associated with heated argument (OR: 0.85).
The factor most closely associated with violent disagreement in the home was parental stress (Table 3). Parents reporting high stress had >3 times the odds of reporting violent disagreements than parents reporting less stress; high-stress parents also had higher odds of disagreements involving arguing or shouting.
Other Characteristics of Child and Family
Poverty, strongly associated with both violent disagreements and parenting stress in bivariate analysis, had only modest levels of association with violence when parenting stress, as well as other factors, were held equal. The child's health and health insurance status were not significantly related to the likelihood of violent disagreements. Very young children (aged 0–5 years) were less likely to live in households with violent or heated disagreement than were children 12 to 17 years of age.
Several characteristics of the parent and the household were closely associated with violent disagreements. As in bivariate analysis, single-mother households were more likely than 2-parent biological families to report both violent disagreement and heated disagreement. However, 2-parent stepfamilies were less likely to report violent disagreement than 2-parent biological families.
Families in which 1 or both parents experienced fair-to-poor health were more likely to report both violent and heated disagreement than other households. The association of violent disagreement with larger families, single-mother families, and families in which 1 or more adults have health problems is consistent with a conceptual model that links stress to violent disagreement. However, the number of children in the home was not meaningfully associated with level of parental stress reported (data not shown).
County characteristics, percent poverty, and percent owner-occupied housing did not contribute to the prevalence of violent or heated disagreement. However, parental perception of community, specifically neighborhood support for parenting activities, was linked to disagreement style. Parents who did not perceive a supportive neighborhood were more likely to report both violent and heated disagreement. The proportion of parents reporting low neighborhood support was highest in urban areas (14.8%), declining steadily with increasing rurality (Table 1).
In 2003, 1 of every 10 US children lived in a household where adults expressed disagreement in a potentially violent manner through hitting or throwing. Nearly one third were exposed to yelling and shouting, behaviors that children may consider violent. The prevalence identified through this survey was considerably higher than that for physical violence against women found in earlier research.7 In total, violence exposure, with potential adverse psychological2 and physical outcomes5 in adulthood, was experienced by 41.8% of children.
Although rural children were hypothesized to be at higher risk for violence exposure based on lower access to interventions for rural women, adjusted prevalence was slightly lower among rural children. Racial differ ences in the prevalence of violent disagreements, however, emerged as anticipated. Even with poverty, parental stress and other factors held constant, black and other children were significantly more likely than white children to live in homes with violent and heated disagreement styles. Given the relationship between violence witnessing and social and academic problems,6 race-based differences may exacerbate eventual health and economic disparities experienced by black and other children.
Parental stress was closely associated with violent and heated disagreements in both unadjusted and multivariable analysis. Similarly, individuals who perceived low support for parenting in their neighborhoods had an elevated risk for both forms of disagreement. In each case, the direction of the relationship cannot be inferred; violence may lead to stress as well as arise from it. However, practitioners may wish to ask parents about stress and their community as potential links to identifying families with violence.
In general, the risk and protective factors included in our model failed to identify strong correlates of violent disagreement. Parenting stress (OR: 3.17) was by far the most clinically significant variable, whereas parent's health (OR: 1.48), having ≥3 children (OR: 1.57), and single-parent status (OR: 1.31) also seem to have clinical relevance. However, several factors that might be perceived as stressful by an observer, such as poor child health, parental unemployment, or frequent changes of residence, were not linked to potential violence in multivariate analysis. In addition, when significant associations were documented, ORs were generally small and may lack clinical relevance. It is also likely that cultural and individual attitudes, beliefs, and experiences, not tapped by the present study, are important contributors to violent behaviors among adults.
The measure of potentially violent disagreement used in our study was broad, asking about behaviors (hit, throw) rather than about injury or other outcomes. However, it may still underestimate violence. First, violent behaviors are generally underreported, as documented in a meta-analysis by Archer.30 Second, fathers, 15% of respondents, were significantly less likely to report hitting or throwing than mothers or other respondents, possibly reflecting a general tendency of men to underreport violence to a greater extent than women.29 Third, the phrasing of the disagreement questions is ambiguous. Respondents may have interpreted “do you” as singular, that is, only the respondent, or as plural, meaning all members of the respondent's family. Thus, behaviors of other family members may not be captured.
Given that violent and heated discussions may have adverse psychological outcomes for children and that these behaviors are witnessed by an estimated 42% of all children, what should health care providers do? The American Academy of Pediatrics has recommended screening for domestic violence during routine visits.31 Screening during pediatric visits is acceptable to most mothers32 and has been demonstrated to markedly increase case identification for potentially violent homes.12 The more crucial question concerns what should be done to intervene in homes where either frank violence or heated argument are present. A recent review concluded that there is little evidence to support the effectiveness of screening, or currently available interventions, at reducing domestic violence.33
Nonetheless, primary care providers are unlikely to be satisfied with recognizing but ignoring potential harm to children. In this article, we found that parental stress may be instrumental in the development of violent disagreements in the home. Numerous educational programs, such as the Bright Futures curriculum developed by the American Academy of Pediatrics and the Maternal and Child Health Bureau, are available to assist practitioners in providing anticipatory guidance to parents in an attempt to reduce stress and to channel disagreements into calmer styles. In addition, all practitioners should be knowledgeable regarding referral resources in their communities and be prepared to follow-up with patients about actual implementation of referrals.
This study was supported, in part, by grant 1 U1CRH 03711–01 from the Office of Rural Health Policy, Health Resources and Services Administration, US Department of Health and Human Services.
- Accepted September 15, 2006.
- Address correspondence to Charity G. Moore, PhD, MSPH, Department of Medicine, Division of General Medicine and Epidemiology, University of North Carolina, CB 7005, Chapel Hill, NC 27599. E-mail:
The authors have indicated they have no financial interests relevant to this article to disclose.
- ↵Carlson BE. Children's observations of interpersonal violence. In: Roberts AR, ed. Battered Women and Their Families. New York, NY: Springer; 1984:147–167
- ↵Bair-Merritt MH, Blackstone M, Feudtner C. Physical health outcomes of childhood exposure to intimate partner violence: a systematic review. Pediatrics.2006;117 (2). Available at: www.pediatrics.org/cgi/content/full/117/2/e278
- ↵Tjaden P, Thoennes N. Extent, Nature and Consequences of Intimate Partner Violence. Washington, DC: US Department of Justice, Office of Justice Programs, National Institute of Justice; 2000
- ↵Holtrop TG, Fischer H, Gray SM, Barry K, Bryant T, Du W. Screening for domestic violence in a general pediatric clinic: be prepared! Pediatrics.2004;114 :1253– 1257
- ↵Johnson RM. Rural health response to domestic violence: policy and practice issues. Available at: http://ruralhealth.hrsa.gov/pub/domviol.htm. Accessed May 23, 2006
- ↵Blumberg SJ, Olson L, Frankel M, Osborn L, Srinath KP, Giambo P. Design and operation of the National Survey of Children's Health, 2003. Vital Health Stat 1.2005;(43) :1– 124
- ↵Abidin RR. Parenting stress index: a measure of the parent-child system. In: Zalaquett CP, Wood RJ, eds. Evaluating Stress: A Book of Resources. Lanham, MD: Scarecrow Press, Inc; 1997:277–291
- ↵Fields JM, Smith KE. Poverty, Family Structure, and Child Well-being: Indicators From the SIPP. Washington, DC: US Bureau of the Census; 1998. Population Division Working Paper 23
- ↵Research Triangle Institute. SUDAAN Language Manual, Release 9.0. Research Triangle Park, NC: Research Triangle Institute; 2004
- ↵Stokes ME, Davis CS, Koch GG. Categorical Data Analysis Using the SAS System. Cary, NC: SAS Institute Inc; 1995
- ↵Archer J. Assessment of the reliability of the conflict tactics scales. J Interpers Violence.1999;14 :1263– 1289
- ↵Parkinson GW, Adams RC, Gmerling FG. Maternal domestic violence screening in an office-based pediatric practice. Pediatrics.2001;108 (3). Available at: www.pediatrics.org/cgi/content/full/108/3/e43
- Copyright © 2007 by the American Academy of Pediatrics