Scope of the Problem
From 1976, when the first national figures for child maltreatment were generated, to 1990, the most recent year covered by the National Child Abuse and Neglect Data System, reports of maltreatment have grown from 416,033 per year (affecting 669,000 children) to 1,700,000 per year (affecting 2,712,917 children) (NCCAN, 1981, 1988, 1992). Prominent groups, such as the U.S. Advisory Board on Child Abuse and Neglect, have labeled the increase in reported cases an ''epidemic" of child maltreatment in the United States, calling the problems of child abuse and neglect a national emergency (U.S. Advisory Board on Child Abuse and Neglect, 1990).
The alarming rise in the number of reported cases of child maltreatment is a significant development, but the full dimensions of its meaning are not yet clear. A significant number of cases reported to child protective services (CPS) agencies are not substantiated. The results of the second National Incidence Study, for example, indicate that in 1986 the alleged maltreatment was unfounded for 47 percent of those children reported in CPS cases (NCCAN, 1988). However, the process of substantiating a reported case may be affected by a wide range of social and economic factors within the case investigation system (such as the number of case workers) as well as the characteristics of the case itself. Whether the rise in reported cases of maltreatment genuinely represents an epidemic, or whether, like apparent increases in other social problems, the epidemic is due to increased recognition rather than a true increase in the phenomenon, remains controversial due to methodological problems with the data collection efforts (Daro et al.,
1990; Eckenrode and Doris, 1987; Peters et al., 1986; Straus and Gelles, 1986). For example, recent research on child death rates suggest that child abuse and neglect (CAN) death rates have not changed during the period 1979 through 1988, but 85 percent of CAN deaths have been systematically misclassified as due to other causes (McClain et al., 1993; Ewigman et al., 1993).
The purpose of this chapter is to review the evidence on prevalence of maltreatment from a number of sources: congressionally mandated maltreatment reports, population-based surveys of specific types of maltreatment, surveys of maltreatment in special population subgroups (e.g., disabled children), and cross-national data. Following a review of the information available on the scope of these problems, the panel discusses methodological problems common to these data, including issues of sampling, research design, and measurement. Finally, the panel provides recommendations to improve the quality of research on the prevalence of child maltreatment.
Governmental Case Report Surveys
In 1974, the Child Abuse and Prevention Treatment Act (P.L. 93-247) stipulated that the newly created National Center on Child Abuse and Neglect (NCCAN) investigate the national incidence of child abuse and neglect (NCCAN, 1981). The act authorized reporting of suspected abuse or neglect and required reporting by a range of professionals (including educators, health professionals, and others) who come into contact with children. The American Humane Association (AHA) was asked to determine the feasibility of a national study of child maltreatment reports and, until 1987, AHA collected reports on various types of maltreatment from child protective services personnel in cooperating states. Data were collected in summary form (e.g., the number of cases of physical abuse in a calendar year) and by individual case.
The data collected in these congressionally mandated national studies of official reports show a dramatic rise over this period in the number of reported cases of maltreatment. Comparative analysis of the data across time indicates that between 1983 and 1986 the percentage of neglect cases decreased (45.7 percent to 36.1 percent) and the number of abuse reports increased (27.9 percent to 35.0 percent). Over the decade 1976 to 1986, estimates of the rates of sexual maltreatment rose drastically from 0.86 per 10,000 children in 1976 to 20.89 per 10,000 children in 1986.
The American Humane Association reports are difficult to interpret for
a number of reasons. First, they were never intended to be studies of the true occurrence of child maltreatment but, rather, to reflect reports of maltreatment coming to the attention of child protective services and other designated personnel. Second, they set no admission criteria for cases, relying instead on varying agency definitions of maltreatment and individual caseworker applications of those definitions. Consequently, the precise nature of the incidents included in the reports is difficult to judge and is subject to multiple sources of bias. Third, the studies include all reported cases, including those that are duplicated or unsubstantiated. Fourth, there are historical inconsistencies in the data. For example, a report that is labeled unsubstantiated was either investigated and the evidence for maltreatment was not present, or the report was still under investigation and had not, at the time of the report, been confirmed. Finally, the data base was not generated from the same participants each year. States and territories voluntarily participated in the project and, over the years, the participants changed (American Humane Association, 1979, 1981; American Association for Protecting Children, 1986, 1987; NCCAN, 1981, 1988, 1992).
National Incidence Studies
Recognizing the need to estimate the true occurrence of child maltreatment rather than simply relying on reported cases, in 1976 NCCAN contracted with Westat to design and implement a study of the incidence and severity of child abuse and neglect. The first National Incidence Study (NIS-1) was conducted in 1979-1980, guided by the concept that the cases reported to state authorities represented only "the top of the iceberg" (NCCAN, 1981). To expand knowledge about incidence of maltreatment, reports were gathered from investigatory agencies (e.g., courts, police) and professionals from other community institutions (such as physicians and educators). In a probability sample of 26 counties, data regarding reported and unreported cases of maltreatment were gathered from child protective services agencies, investigatory agencies, and professionals in other community agencies. In addition to employing an operational definition of each type of maltreatment, NIS-1 collected information on a number of child and parent-related characteristics as well as two other important characteristics, the nature of maltreatment (omission or comission) and the effect of maltreatment. The injury or impairment resulting from the maltreatment must have been rated as moderate or severe for the act or omission to be classified as maltreatment. These data do not represent true occurrence of maltreatment but, once again, reports of maltreatment known to service providers. NIS-1 estimated the number of maltreated children in a 12-month period (5/1/79-4/30/80) and the incidence was expressed as a rate with the denominator being the number of children in the United States in an age range.
Data from child protective services agencies indicated that the reported incidence was 17.8 per 1,000 children with 42.7 percent of these reports substantiated, for a substantiated incidence rate of 7.6 per 1,000 children per year. Fifty-three percent of these substantiated cases did not meet the NIS-1 maltreatment definition that required moderate or severe harm, for a corrected incidence of 3.4 per 1,000 children per year. When data from other community agencies are added to the data from child protective services, the incidence increased to 10.5 per 1,000 children per year.
The second National Incidence Study (NIS-2) was compiled in 1986. Using the same design as NIS-1, NIS-2 estimated the current incidence of maltreatment and documented changes in the frequency, character, or severity of maltreatment since the completion of NIS-1 (NCCAN, 1988; Cicchetti and Barnett, 1991). NIS-2 employed two sets of operational standards. The first conformed to NIS-1 standards and required identifiable harm to establish maltreatment. The second set of standards was broader and included "endangered" children at risk for harm. When the more restrictive definitions were used, the reported incidence of substantiated maltreatment was estimated at 14.8 per 1,000 children.1 If the broader "endangered" definition was employed, the reported incidence was 22.6 per 1,000 children.2 The more conservative definition represents a 66 percent increase over the NIS-1 figure (NCCAN, 1988). Most of this increase was due to abuse, with physical abuse up 58 percent and sexual abuse up 300 percent (NCCAN, 1988). No increases in emotional abuse or any form of neglect were observed.
Looking at specific categories of maltreatment in corrected estimates of the NIS-2 survey employing the endangerment standard, the overall rates per 1,000 children per year for any type of abuse was 9.4, with a rate of 4.9 for physical abuse, 2.1 for sexual abuse, and 3.0 for emotional abuse3 (Sedlak, 1990). There were 14.6 neglected children per 1,000 children, with physical neglect being most common (8.1 per 1,000). Factors associated with increased rates of maltreatment included older age of the child and lower family income.
National Child Abuse and Neglect Data System
In response to the 1988 Child Abuse Prevention, Adoption and Family Services Act (P.L. 100-294), NCCAN redesigned the national child maltreatment data collection and analysis system. Based on two years of work with national experts and state representatives, the National Child Abuse and Neglect Data System (NCANDS), like the National Incidence Studies, included two parts. The summary data component compiled information from state agencies on reports, investigations, victims, and perpetrators.
The detailed case component compiled case data to investigate trends and issues in the field.
The 1990 NCANDS information was gathered on reports from 49 states, the District of Columbia, 1 territory, and all branches of the armed services. In all, there were 1.7 million reports of maltreatment involving 2.7 million children. Forty-five percent of the reports were of neglect, 47 percent of the reports were of abuse: 25 percent of physical abuse, 16 percent of sexual abuse, 6 percent of emotional abuse. The rates of substantiation for reported cases remained reasonably constant at 40 percent (NCCAN, 1992).
Private Sector Studies Using Case Reports
In addition to the governmental data collection activities that rely on state reports, the National Committee for Prevention of Child Abuse (NCPCA) conducts an annual 50-state survey by contacting child welfare administrators in all 50 states and the District of Columbia to track trends in child abuse reporting, child abuse fatalities, and child welfare policy (Daro et al., 1990). Beginning in 1972, the committee conducted a telephone interview with each state's federal government liaison officer for child abuse and neglect. Trends indicate that reports increased from 1980 to 1985, stabilized between 1985 and 1987, and began to rise substantially in 1988. In 1989, reports involving approximately 2.4 million children were filed, up 10 percent from 1988. This increase in reports of maltreatment has been accompanied by a growing number of child abuse fatalities. Fatalities attributed to reports of abuse have grown from 899 in 1986 to 1,237 in 1989.4 These data, like other data collection efforts based largely on official or unofficial reports, are problematic to interpret because of the noncomparability of case definitions across states and the inclusion of duplicated cases.
The National Committee for the Prevention of Child Abuse also estimated the scope of child neglect. On the basis of its 1990 50-state survey, the committee estimated that 46 percent of reported children (or 2 percent of all children) are reported for child neglect each year (Crittenden, 1992; Daro and McCurdy, 1991).
It is important to note that definitions of child neglect vary over time and circumstances as standards governing the acceptable level of adequate care of children have expanded from covering nutrition, shelter, and other material necessities to include medical, educational, and emotional care. Helfer (1987) has stated that "the true incidence of child neglect is unknown, since no consensus on the definition exists." Although extreme forms of maltreatment may be easy to recognize, gradations between satisfactory care and borderline neglect are difficult to measure, and variations within and between social service and child welfare agencies in handling reports of child neglect are common.
A number of investigators have attempted to estimate the prevalence of child maltreatment through standard survey techniques rather than relying on official reports or reports of cases known to certain professionals (Gil, 1970).
After reviewing the incidence data available in the 1970s, Straus and Gelles (1986) concluded that little was known about patterns of violence against children. In 1975 they surveyed a national probability sample of 2,143 families consisting of a married couple or a man and woman who identified themselves as living together as a conjugal unit (Gelles, 1978). Violence was operationalized using the Conflict Tactics Scales, an instrument with 18 items in 3 categories: use of rational discussion and argument, use of verbal and nonverbal expression of hostility, and use of physical force or violence as a tactic in a conflict (Straus, 1979). Using a physical abuse index that included hitting a child with an object (such as a hairbrush or belt), the survey found a prevalence rate of 140 per 1,000 children ages 3-17 per year. If hitting with an object is omitted so that the index includes acts that are almost universally accepted as physical abuse (such as kicking and punching a child), the rate was 36 per 1,000 children.
Straus and Gelles (1986) repeated their survey in 1985 (with the same instrument) with 3,520 families using telephone interviews. The 1985 sample included children under three and single parents; after removing them for purposes of comparison with the rates generated by the 1975 survey, Straus and Gelles (1986) found a statistically significant decrease in the index of severe violence against children (from 140 to 107 per 1,000 children), a significant decrease in the very severe violence index (from 30 to 19 per 1,000 children), but no significant difference in the rate of minor violence, i.e., corporal punishment.
Straus and Gelles must be commended on two counts. First, they used a nationally representative sample and, second, they used an instrument with known reliability in determining whether abuse occured. The work is limited by its attention only to physical abuse and verbal forms of emotional maltreatment.
Studies of the incidence and prevalence of sexual abuse, while initiated early in this century (Hamilton, 1929), became more common in the 1980s. Prior to 1976, surveys inquiring about sexual abuse were largely undertaken
with volunteers or with college students (Peters et al., 1986). These studies mainly collected data on lifetime prevalence of noncontact (exhibitionism, solicitation) and contact sexual abuse. Depending on both the population used and the definition employed, lifetime rates of sexual abuse have ranged from 6 percent to 62 percent for females and 3 percent to 31 percent for males (Peters et al., 1986), with most investigators concluding that there has been a rise in the reported rates of being a victim of sexual abuse (Leventhal, 1988; Peters et al., 1986; Wyatt and Peters, 1986a,b).
Sexual abuse by adolescents has received increased attention. One study indicated that 56 percent of child molestations are committed by someone under age 18 (Fehrenbach et al., 1986), and the majority of adult sex offenders indicate that the onset of their deviant sexual behavior occurred before age 18 (Abel et al., 1988). Until recently, however, adolescent sexual offenders have been neglected in clinical and research literature. There are no empirically derived and tested models to explain why adolescents commit sexual crimes or develop deviant sexual interest patterns (Becker, 1991).
Prevalence of Abuse and Neglect in Disabled Children
Although early work in child maltreatment research pointed to a link between disabilities and maltreatment, studies in this area are limited. The chief problem, in addition to identifying the number of disabled children, is the temporal ordering of the two phenomena, determining whether disabled children are more likely to be abused or whether abused children are more likely to become disabled through abuse (Groce, 1988). Furthermore, most studies examining the relationship between disabilities and maltreatment have been inadequate due to small sample size, lack of appropriate comparison groups, and selection bias (Bertolli et al., 1992). In general, studies in this area find a higher rate of maltreatment of disabled children than of children without disabilities but the causal sequence is unknown (Groce, 1988). The possibility of disaggregating cause and effect by a comparative study of maltreatment in children whose disabilities are largely genetically determined, children whose disabilities could be caused by maltreatment, and children without disabilities has not been explored, even though it could provide significant insights in this area (Garbarino, 1987).
Cross-national comparisons of the prevalence of child abuse and neglect are difficult because of anthropologists' focus on normative cultural behavior, the low base rate of the phenomenon, and the lack of a universal
standard for what constitutes abuse and neglect (Korbin, 1987). Although abuse and neglect statistics are not collected across countries, the World Health Organization collects data on causes of death by age (Bertolli et al., 1992). Analyses of these data indicate that child homicide is more common in the United States than in comparison countries for every age group (Williams and Kotch, 1990; Division of Injury Control, 1990).
This review of the research on the scope of child maltreatment reveals important methodological problems that limit the usefulness of these data for drawing conclusions about both the dimensions of the problem and the factors that cause it. Such problems include definitional issues, confusion of prevalence and incidence, the source of maltreatment data, sampling and design considerations, the paucity of psychometrically sound instruments, and the impact of mandatory reporting requirements on the reliability of survey respondents' reports. Although some of these issues have been documented by investigators in the field, the panel believes their importance merits emphasis in this report because they must be remedied through research to improve the quality of data (Bertolli et al., 1992; Cicchetti and Barnett, 1991; Fromuth and Burkhart, 1987; Haugaard and Emery, 1989; Leventhal, 1982; Mash and Wolfe, 1991; Peters et al., 1986; Widom, 1988; Wyatt and Peters, 1986a,b).
As pointed out in the previous chapter, child maltreatment is not a uniform entity. Failure to recognize this stifles work on etiology, treatment, and prevention. While nominal definitions of maltreatment (e.g., physical, sexual, and emotional abuse) are common across studies, operational criteria, when available, differ greatly (Aber and Zigler, 1981; Bertolli et al., 1992; Cicchetti and Barnett, 1991; Peters et al., 1986; Wyatt and Peters, 1986a). Operational definitions can be based on the act itself (e.g., its form, intensity, duration, frequency), its consequences (e.g., bruising, developmental delay), the perpetrator's intent, the child's age, age differences and relationships between victims and perpetrators, and social norms on appropriate behavior (e.g., physical punishment as appropriate parental behavior) (Bertolli et al., 1992). Inconsistencies among definitions in these aspects, as pointed out in a rigorous comparison and adjustment of the operational criteria of sexual abuse by Wyatt and Peters (1986a), appear to account for much of the difference in prevalence estimates across studies.
Given that only the most serious episodes of abuse are reported, such as those observed by outsiders such as physicians, neighbors, and teachers, child maltreatment is most likely underreported and underestimated in official records (Nagi, 1975; Widom, 1988; Wilbur, 1985). The 1975 National Family Violence Survey, for example, found a prevalence rate for physical abuse of 140 per 1,000 children per year, which is 21 times greater than the 6.8 per 1,000 for cases reported to child protective services in the United States, and also many times greater than the rates from the National Incidence Studies of cases known to all human service professionals (Straus and Gelles, 1986). Many severely abused children are brought to emergency rooms by parents or relatives who blame injuries on accidents that, in reality, never occurred. Children are reluctant to reveal abuse for fear of removal from their homes, feelings of shame, protection of their parents, selective inattention, or loss of memory of the event (Berger et al., 1988; Della Femina et al., 1990; Frischolz, 1985; Herzberger and Tennen, 1983; Lewis et al., 1991; Sullivan, 1956; Wilbur, 1985).
Few pediatricians and child psychiatrists are trained to elicit abuse histories, although the American Medical Association has recently recommended that these kinds of questions should be part of routine adolescent, pediatric, and psychiatric evaluations. Such procedures as looking at a child's or adolescent's back, inquiring about the origin of visible scars, and asking, "Are there any scars I can't see?" are rarely included in evaluations (Stein and Lewis, 1992).
Confusion of Prevalence and Incidence
Measuring the occurrence of child abuse and neglect has been difficult not only because of a lack of definitions but also because of confusion of what measure of occurrence is under investigation. To count an outcome under study, epidemiologists commonly use two measures: incidence and prevalence. Incidence, the number of new cases detected in a defined time period divided by the number of children who are eligible to become new cases, is the ideal measure for identifying etiologic factors. By using new cases, investigators can determine which features are related to occurrence of maltreatment rather than trying to disentangle features related to occurrence from those related to the continuation of maltreatment. In contrast, child protection services often rely on total occurrences, a different measure of incidence: the numerator is total occurrences of abuse and neglect during the defined period and the denominator is all children, whether or not they had previously been maltreated. Previously maltreated children are included in the denominator because they are at risk for new occurrences (Bertolli et al., 1992).
Prevalence consists of the number of existing cases of child abuse and neglect at any point in time divided by the total population from which the cases are identified. Use of prevalence data causes confusion between risk factors that are associated with new cases of abuse and neglect and those that influence continuation of the problem (Bertolli et al., 1992).
Research in this area does not always distinguish between incident and prevalent cases. It is sometimes difficult to distinguish between new and existing cases because the exact time of the occurrence must be arbitrarily determined. One approach for dealing with the confusion between incident and prevalent cases is to use a measure called period prevalence. In it the numerator is individuals in the population who, at any time in a defined period, experienced abuse or neglect divided by the total population. This strategy is commonly used by researchers in this field when they estimate the proportion of adults in a particular age group who have been maltreated at any time during childhood (Bertolli et al., 1992).
Use of Administrative Data
With the exception of a relatively small number of primary data collection efforts, the vast amount of information on the occurrence of child abuse and neglect is based on state child protective services records. The accuracy with which these data reflect the true incidence or prevalence of maltreatment is questionable for a number of reasons. First, although all states have mandatory reporting laws, the definitions of abuse and neglect vary across and within states chiefly as a result of the absence of structured assessment tools, the casework burdens of social services personnel, inaccurate or missing information, and the limited availability of services for families who are identified (Hoaglin et al., 1982; Leventhal, 1990). Second, the segments of the population at risk are subject to unequal surveillance. Specifically, families who, because of their demographic characteristics (e.g., poverty, unemployment, single parent), have more frequent contact with public sector services (e.g., welfare, housing) are more often exposed to mandated reporters and receive closer scrutiny. Consequently, the overrepresentation of families who are poor and members of minority groups in child protective services caseloads may be less likely to be due to poverty being a risk factor for maltreatment and more likely to be due to undetected abuse and neglect in more affluent families (Gelles, 1982; Newberger et al., 1977).
Conflicting evidence exists as to whether differences occur in rates of maltreatment by culture or ethnicity. The two National Incidence Studies did not find a significant relationship between race5 and the incidence, type, or severity of child maltreatment (NCCAN, 1981, 1988). Some studies have reported that ethnic minorities are disproportionately reported for child
abuse and neglect (e.g., Jason et al., 1982; Lauderdale et al., 1980). Others indicate that such overrepresentation is equivalent to the representation of ethnic minorities among the poor, thus supporting the confounding nature of ethnicity and poverty in child abuse and neglect reports (Horowitz and Wolock, 1981). Cultural and ethnic groups that are at the greatest risk of poverty, then, appear to have disproportionate incidence and prevalence rates of child maltreatment. Convincing evidence that disaggregates socioeconomic status and cultural or ethnic identity in rates of reported child abuse and neglect has not been developed, although socioeconomic status may be strongly associated with some forms of child maltreatment. For example, child neglect occurs among the poorest of the poor (Giovannoni and Billingsley, 1970), and physical abuse is more severe among the poorest families (Straus, 1980).
Surveillance bias further complicates the issue. Poor and ethnic minority children are more likely to be identified as maltreated than more affluent white families (e.g., Newberger et al., 1977). A reanalysis of the first National Incidence Study (1981) found that class and race were the best predictors of whether an incident was reported by hospitals, with impoverished black families more likely to be reported than affluent white families, regardless of the severity of the incident (Hampton and Newberger, 1985). Furthermore, service availability and severity of worker caseloads may affect reporting by ethnicity (Light, 1973; Wolock and Horowitz, 1979).
Finally, use of substantiated administrative data is problematic because all cases of abuse and neglect are not pursued with equal rigor (Eckenrode et al., 1988a,b; Giovannoni, 1989). The availability of resources within individual state child protective service agencies clearly affects which cases are investigated and confirmed.
Rate of Occurrence Versus Rate of Reporting
Much of the tremendous increase in maltreatment rates is probably due to increased reporting (Straus and Gelles, 1986). Recent work by Leventhal and colleagues (1993) suggests that for two cohorts of adolescent mothers, one whose children were born from 1967-1970 and the other whose children were born 1979-1981, no differences in rates of maltreatment were detected when injury events were rated using predefined criteria by physicians who were blind to the social characteristics of the mothers. However, over the 14 year time span, reports of maltreatment to the Connecticut Department of Children and Youth Services tripled indicating little real difference in rates of maltreatment but considerable differences in reported rates of maltreatment. A recent study of the classification of child death rates also suggests that although the number of deaths from child abuse and neglect has been consistently underestimated because they are often misclassified
as due to other causes6, even expanded interpretations of child death reports suggest that the rate of child abuse and neglect deaths among children ages 0 to 17 has remained relatively stable (around 2,000 deaths per year) from 1979 through 1988 (McClain et al., 1993).
Changes in the proportion of child maltreatment reports that are substantiated may provide evidence as to whether the increased reported incidence involves changes in the rate of occurrence as well. One study conducted by Eckenrode and colleagues (1988a,b) observed that the percentage of substantiated reports declined from 1974 to 1984, although the number of reports filed increased significantly during the same period. The authors concluded that the reporting and investigation process was ''cutting further into the tip of the iceberg of child maltreatment, but proportionally more social resources are being used to uncover new cases" (Eckenrode et al., 1988b; p. 9). Their findings are consistent with the summary findings of the NIS-2 study, which reported that although little change was reported between 1980 and 1986 in the severe categories of maltreatment (such as fatalities and serious injuries), dramatic increases occurred at the level of mild to moderate injury and impairment, where there was greater potential for improved recognition of cases (NCCAN, 1988).
In summary, the increase in reported incidence of child maltreatment is probably the result of a combination of factors: expanded definitions, increased recognition and reporting, and increases in the rate of occurrence. Increases in rate of occurrence cannot be ruled out since the number of children living in poverty has increased significantly since the 1970s and poverty is thought to be significantly associated with child maltreatment (see discussion in Chapter 4) (National Commission on Children, 1991). Changes over time in the proportion of reports for each type and severity of abuse, the source of the complaints, and the characteristics of perpetrators and victims can provide insight into whether significant increases in reported rates of maltreatment reflect increases in the actual occurrence of maltreatment (Knudsen, 1992). However, data for each of these factors are so limited that it currently provides no insight into the relative contributions of increased rates of reporting and increased rates of occurrence of child maltreatment.
Sampling and Design Issues
The design of a study is the blueprint through which the investigator achieves specific research objectives. These objectives determine what groups are to be compared, how subjects are selected, what factors should be measured, and what conclusions can be drawn (Bertolli et al., 1992; Leventhal, 1990).
A number of investigators have pointed out the inadequacy of the sam-
pling strategies and study designs commonly used in child abuse research (Leventhal, 1982, 1990; Peters et al., 1986; Widom, 1988). The use of volunteers and college students in cross-sectional studies has produced disparate estimates of the problem, an inconsistent picture of the etiologic factors associated with these problems, and an inability to generalize findings to other population groups.
An ideal design for investigating etiology and sequelae of maltreatment would be a large, prospective cohort study in which random samples of children from different types of communities, and ethnic and sociodemographic groups would be followed from birth to adulthood. The study would measure all known and hypothesized risk factors for abuse and neglect at birth and at prespecified follow-up interviews, and all new cases of abuse and neglect would be identified as they occur with no loss to follow-up (Bertolli et al., 1992).
Although cohort studies are being conducted in this area, key factors of the ideal design, such as ascertainment of representative samples of cases of abuse and neglect, are missing due to reliance on administrative data for cases. Incidents of child maltreatment may also remain undetected and unreported if subgroups of the population at risk for abuse and neglect are not recruited into longitudinal studies.
In the area of abuse and neglect, cross-sectional designs are more common than cohort studies. These studies interview respondents about past experiences of abuse and neglect and risk factors of interest. While quicker, less costly, and easier than cohort studies, cross-sectional designs are not usually used to estimate incidence and, furthermore, because exposure variables are measured after the abuse or neglect occurred, the temporal ordering of cause and effect is problematic. Finally, retrospectively collected data are often subject to recall errors (Briere, 1992; Widom, 1989; Wolfe and Mosk, 1983).
Two other study designs are used less frequently in this area of research but should be mentioned because they are useful for studying phenomena with low base rates. The case-control study is a design in which two groups of individuals are selected separately (one with the outcome, the other without it) and then compared to determine the effects of hypothesized risk factors. To draw accurate conclusions about the effects of the risk factors, the controls should be representative of the population from which the study cases came (Bertolli et al., 1992; Kleinbaum et al., 1982; Leventhal, 1982). This design is far more efficient than prospective cohort studies for studying statistically rare outcomes like abuse and neglect but it is subject to measurement error because exposure variables are measured after the outcome of interest has occurred.
The last study design is considerably different from the other designs in that the areas of analyses are not individuals but rather geopolitical areas
(e.g., census tracts). In this type of design the investigator does not know the joint distribution of exposure and outcomes but simply the number of exposed and unexposed cases and noncases. These studies are quick and inexpensive but commonly subject to a problem called ecologic bias (Morgenstern, 1982), whereby correlations between outcomes and observed characteristics of the area may be the spurious result of unmeasured characteristics. Furthermore, because they employ routinely collected data, they may be subject to problems resulting from the use of administrative data.
Research on child abuse and neglect has been severely hampered by the lack of instruments to measure the phenomena. Relatively few instruments have reported reliability and validity.7 The lack of instrumentation, whether on a state or an individual level, leads to misclassification bias, a systematic error introduced into studies through mistakes in measuring or classifying subjects on variables to be used in the analysis of the study (Bertolli et al., 1992; Cicchetti and Barnett, 1991; Eckenrode et al., 1988a,b). Furthermore, inconsistency in the design of questions about sensitive social topics (such as child sexual abuse) contributes to extreme differences in estimates of the problem (Peters et al., 1986).
One recent study of methodological issues in the measurements used in U.S. programs for tracking the use of licit and illicit drugs concluded that nonsampling components of error and bias in measurements can dwarf the sampling variance in research on socially sensitive topics (Turner et al., 1992). This same study noted that significant improvements in survey instruments can be achieved by focusing on the cognitive demands that individual survey questions make on respondents, and by relying more extensively on self-administered forms rather than interviewer-administered formats when research subjects can complete self-report questionnaires (Turner et al., 1992).
Effect of Mandatory Reporting
In some cases, previously undetected or imminent incidents of abuse or neglect may be detected during an investigation. As part of the informed consent procedure required in federally supported studies, participants in most child maltreatment studies should be informed of their rights and duties as research subjects (see Chapter 9 for a full discussion of ethical issues in child maltreatment research). In addition, every state has adopted legislation that requires research investigators (and other professionals who have contact with children) to disclose to child welfare officials reports of suspected child abuse or neglect that have not been recorded.8 The basic
requirement is that professionals must report any cases in which there is "reason to believe" or "reasonable cause to suspect" that child abuse, past or present, has occurred. Whether such ''reason to believe" refers to a clinical hunch or firm evidence is open to conjecture, but a professional who fails to disclose suspected child maltreatment may be charged with criminal action (Sieber, 1992).
The requirements of mandatory reporting vary from state to state; in some cases, disclosure to child welfare authorities is not required if the individual who makes the disclosure is receiving therapeutic services (such as family counseling). There are also variations in the level of endangerment that constitute child maltreatment, as well as variations in evidentiary standards that warrant a report to child welfare authorities.
Empirical research studies on the conditions and circumstances that affect professional reports of privately disclosed incidents of child maltreatment are rare. But one survey study has indicated that a significant amount of "discretionary reporting" occurs, whereby professionals who come into contact with children may report some suspected cases of child maltreatment but not others (Zellman, 1990). Professionals who have made reports of child maltreatment cite various reasons for doing so, including stopping maltreatment, getting help for the family, helping the family to recognize the seriousness of their problems, and complying with legal requirements (Zellman, 1990).
Estimates of failure to report cases of suspected abuse range from about a third of practicing psychologists (Borsig and Kalichman, in press) to 22 percent of a sample of professionals that included general and family practitioners, pediatricians, child psychiatrists, clinical psychologists, social workers, principals of public schools, and heads of child care centers (Zellman, 1990). The issue of failure to report needs to be examined in light of its impact on incidence and prevalence estimates of child maltreatment as well as its significance for the conduct of research in this area. For example, some potential research subjects may decide not to participate in the research project because of mandatory reporting requirements, while others may falsify or distort responses revealing reportable child maltreatment activity (Sieber, 1992).
The interests and safety of the child are obviously paramount to any research objectives. The primary reasons provided by professionals who have indicated failure to report include the lack of sufficient evidence to justify a report, treatment-related concerns (i.e., "I can do better than the system"), and the costs of reporting to professionals (e.g., reports too time-consuming or fear of lawsuit for reporting) (Zellman, 1990).
The impact of mandatory reporting requirements on research projects has not been studied. Many research investigators believe that the requirements of mandatory reporting may conflict with other fundamental prin-
ciples in the informed consent process, which assure subjects of complete confidentiality of their responses. The dilemma, from the scientist's perspective, is tantamount to saying, "If you respond truthfully, I may report you to the appropriate authority" (Sieber, 1992). Many research scientists are reluctant to report disclosures of child abuse and neglect that occur in a research investigation unless the child is endangered. In the case of prior abuse incidents, research investigators sometimes believe that little benefit will be gained by recording earlier cases of abuse that may be quite dated, especially if the family or the offender is already in treatment for other reported cases, if the child does not want the incident reported, or if the perpetrator is not alive or not in contact with children.
The panel concludes that much of the methodology for prevalence and incidence research in the area of child abuse and neglect is seriously flawed. Definitional variations, variations in state cooperation with national data collection efforts, legal requirements for mandated reporting, scarce funding for methodological work (specifically instrument development), and the paucity of rigorous epidemiologic investigations have retarded progress in this field. However, the limited available evidence suggests that child abuse and neglect is an important, prevalent problem in the United States, involving more than 1 million children each year in case reports and 2,000 child deaths annually. Child abuse and neglect are particularly important compared with other critical childhood health and mental health problems because the consequences of child maltreatment are often directly associated with adverse health and mental outcomes in children and families (Institute of Medicine, 1989). Furthermore, given the prevalence of childhood maltreatment, the level of federal funds expended in this research area is extremely small when compared with the resources allocated for less prevalent childhood mental disorders, such as autism and childhood schizophrenia.9
Specifically, the panel concludes:
Much of the tremendous increase in maltreatment rates is probably the result of increased reporting rather than significant changes in actual rates of occurrence. However, the possibility that rates of occurrence have increased as well needs consideration, given the large numbers of children now living in adverse circumstances.
Neglect is more common than any individual type of child maltreatment and has consistently accounted for approximately half of the cases of maltreatment (NCCAN, 1992). Although reports of physical or sexual abuse of childhood significantly increased between 1980 and 1986, reports of child neglect still account for a large majority (63 percent) of the cases
reported in the 1986 National Incidence Study. Reported cases of child neglect involved 14.6 per 1,000 countable neglected children, or 917,200 children nationwide, identified in the 1986 study of National Incidence and Prevalence of Child Abuse and Neglect (Sedlak, 1990; NCCAN, 1988).
However, these figures may underestimate the prevalence of neglect because cases involving multiple forms of child maltreatment are likely to be reported as abuse rather than neglect; siblings of reported children may experience the same family conditions; and many children in both poor and affluent families are probably underreported (Crittenden, 1992). In addition to unreported cases of physical neglect, the incidence of psychological or emotional neglect (which includes verbal battering, belittling, and terrorism) is probably much greater than that reflected in the NIS reports, since such cases are rarely reported in the absence of other forms of child maltreatment.
The chronicity of child neglect cases needs to be considered in discussions of incidence and prevalence. Incidence figures are more suited to the measurement of maltreatment cases that are specific and short-lived, and prevalence measures may be more appropriate for child neglect cases, since they tend to be chronic and long-term (Polansky et al., 1981).
Total reports of physical abuse increased 58 percent between 1980 and 1986 (NCCAN, 1988), although severe forms of physical abuse may actually have decreased (Straus and Gelles, 1986).
Sexual abuse reported to child protective services has shown the largest reported increase of any form of abuse or neglect, rising from 0.7 per 1,000 children to 2.2 per 1,000 children per year in the National Incidence Studies (1980-1986) (NCCAN, 1988). Using different definitions of sexual abuse and data from community respondents, estimates of prevalence range from a low of 20-24 percent (Hamilton, 1929; Institute for Sex Research, 1953) to a high of 54-62 percent (Russell, 1983; Wyatt, 1985). The high estimates included noncontact exposure.
Emotional abuse is the least studied of all types of abuse. Emotional abuse, across the 1990 Annual Fifty States Survey, the 1986 Second National Incidence Study, and the 1990 NCANDS accounts for approximately 7 percent of the reported cases of maltreatment.10 However, in the absence of operational definitions and standards of severity, the true extent of occurrence of emotional maltreatment is unknown. In addition to existing in its own discrete forms, emotional abuse may be inherent in many reported cases of child abuse and neglect but it may not be recorded as a specific form of maltreatment (American Association for Protecting Children, 1986). No population surveys of emotional maltreatment have been conducted (Knudsen, 1992).
Recognizing that incidence and prevalence data comprise the cornerstone of good etiologic, treatment, and prevention work, the panel proposes strategies for improving research on the occurrence of child abuse and neglect. Although many of the following recommendations can be achieved with little additional expenditure of funds, a well-designed, multiyear plan with appropriate guidance to the states is needed.
Recommendation 3-1: State data systems should be improved so that high-quality research on service systems can be conducted. The range of variation in state definitions of child abuse and neglect as well as verification procedures seriously undermine the quality of existing data. Effort is needed on a national level to:
mandate state compliance with data acquisition and reporting efforts as in other federal efforts like Medicaid and Medicare;
develop uniform case definitions with measurable criteria;
generate risk assessment tools that are sensitive to complex professional and ethical problems to guide protective services workers' decisions;
identify potential sources of bias in current procedures for reporting and investigation of reported cases;
redesign state data processing systems so that uniform individual-level data are available and unduplicated counts of children affected by abuse and neglect are easily obtainable;
establish an expert panel to periodically review the data system, establish quality indicators, and identify key areas for services systems investigation;
make available state-level data as public use data tapes;
conduct ethnographic studies to identify the systems level features that affect reporting and case verification; and
provide sufficient incentive for state child welfare agencies to become equal partners in the research process while acknowledging the problems (e.g., understaffing, management emphasis) of state-level research.
Recommendation 3-2: Standardized measures and methodological research should be developed for use in epidemiologic studies of child abuse and neglect. Unlike the National Institute of Mental Health, the National Center for Child Abuse and Neglect has not mounted an extensive multiyear effort to develop valid and reliable instruments to identify child abuse and neglect in population-based studies. While a few good measures have been developed over time (such as the Conflict Tactics Scales, the Trauma Symptom Checklist, and the Wyatt Sex History Questionnaire),
more attention must be paid to instrumentation and methodological issues to improve the scientific quality of child abuse and neglect research.
The panel recommends:
development of field-tested instruments and interview techniques for identifying physical, sexual and emotional abuse, and neglect. We recognize that many different types of instruments need to be developed, including observational protocols and child self-report instruments, and we encourage the representation of a diversity of perspectives in this process.
dedication of funds to undertake psychometric work to ensure instruments that are both reliable and valid. Instruments should be tested to ensure that they are useful in diverse ethnic groups and incorporate strategies for eliciting information about a sensitive topic. Development of child as well as parent report instruments is needed.
encouragement of methodological studies to improve the quality of child maltreatment research. Such studies should focus on problems in defining populations of interest to be sampled in studies of child abuse and neglect, choosing sources of information, and deciding the time frame to be covered.
Recommendation 3-3: Data collection efforts should capitalize on future national survey efforts to include questions on child abuse and neglect. Currently, little information is gathered about child abuse or neglect in many national surveys (such as the National Health Interview Survey on Child Health, the National Survey of Children, and the Child Supplement to the National Longitudinal Survey of Youth)11. Although the inclusion of questions on child maltreatment may raise issues of cost and administrative burdens, past surveys and secondary analyses of existing data sets represent research opportunities that could provide further insights into the nature and frequency of child abuse and neglect.
The panel recommends:
establishing an expert panel to resolve issues around mandatory reporting requirements and the legal status of certificates of confidentiality.
including questions on child abuse and neglect and key covariates, such as family violence, on national survey efforts such as the National Survey of Children.
Recommendation 3-4: Research should encourage secondary analyses of existing data available from multiple national surveys for questions about abuse and neglect. Specifically, the panel suggests:
funding secondary analysis of already existing data sets that do not specifically apply to maltreatment but may reveal important information.
expanding the data archives to include studies about children with questions potentially pertaining to child maltreatment.
Recommendation 3-5. After considerable work on instrumentation, including investigations into the most effective questioning strategies, the panel recommends the funding of a series of full scale epidemiologic studies on the incidence and prevalence of child abuse and neglect. Once the methodological work is complete, large scale epidemiologic investigations are feasible. These investigations would provide solid information on the occurrence of these important problems as well as on key etiologic agents. This series should be:
multisite and competitively awarded to ensure that the most qualified investigators participate in the effort.
of sufficient size to characterize families from a variety of communities. Sites must be chosen to: (1) permit adequate representation of all segments of the U.S. population; (2) ensure samples enriched in key predisposing features through the use of multistage probability sampling schemes; and (3) exemplify state-of-the-art field study procedures including skilled and well-supervised interviewers, concurrent data entry, and structured protocols for follow-up of refusals.
inclusive of all ethnic groups represented in the United States. Studies should be undertaken that include ethnic subgroups (e.g., Cambodians, Samoans) rather than large groupings (e.g., Asian/Pacific Islanders).
continued as a cohort study where subsamples of families thought to be at high and low risk for maltreatment are followed over time and assessed at specified intervals to determine the incidence of new cases of abuse and neglect.
mandated to use multiple data sources to verify reports of abuse and neglect, particularly when parents report either medically attended injuries or social services interventions as a result of the maltreatment. These strategies would serve not only as case ascertainment but would also allow the development of realistic indicators of what portion of these problems are brought to the attention of child protective services.
1. This figure is the corrected estimate, revising an earlier 16.3 figure reported in the initial report of the NIS-II survey (see Sedlak, 1990). Technical amendments to the study findingsNational incidence and prevalence of child abuse and neglect: 1988. May 23, 1990.
2. This figure is the corrected estimate, revising an earlier 25.2 figure reported in the initial report of the NIS-II survey (see Sedlak, 1990). Technical amendments to the study findingsNational incidence and prevalence of child abuse and neglect: 1988. May 23, 1990.
3. The NIS-II survey also had a category of "emotionally neglected," which is included in the neglect rather than abuse incidence estimates.
4. This figure compares to about 8,000 deaths per year for children ages 1 to 4 (4 million births per year x 4 years x 50/100,000 death rate for 1-4-year-olds).
5. Broad racial categories often do not accurately reflect cultural or ethnic affiliation. There are multiple types of Hispanics/Latinos (e.g., Puerto Ricans, Mexicans, Mexican-Americans, Guatemalans, Cubans); blacks (e.g., African-Americans; Haitians, West Indians); Asians (e.g., Chinese, Japanese, Koreans, Thais, Cambodians); Native American Indians (e.g., Navaho, Sioux); Pacific Islanders (e.g., Hawaiians, Samoans); and European-Americans (e.g., Italians, British, Germans). Furthermore, within any of these populations, there is substantial intracultural diversity along such dimensions as socioeconomic status, acculturative status, education, and gender.
6. Child deaths are classified according to the International Classification of Diseases, 9th revision, known as ICD-9 (World Health Organization, 1977). ICD-9 includes several categories (E967, E968.4, and E904.0) that explicitly note child abuse and neglect, including child battering and other maltreatment, criminal neglect, abandonment and neglect of infants and the child maltreatment syndrome (McClain et al., 1993). However, other categories in the ICD-9 system may also include deaths from child abuse and neglect, including some deaths recorded as the following: child homicide, undetermined origin, accidental fatalities, or Sudden Infant Death Syndrome.
7. Examples of such instruments include the Conflict Tactics Scales (Straus and Gelles 1986; Straus 1990a,b), the Trauma Symptom Checklist (Briere and Runtz, 1987), the Wyatt Sex History Questionnaire (Wyatt, 1985; Wyatt et al., in press), the Childhood History Questionnaire (Milner, J.S., Robertson, K.R., and D.L. Rogers, 1990), and the Child Abuse Potential Inventory (Milner and Wimberley, 1979; Milner, J.S., Gold, R.G., Ayoub, C., and M.M. Jacewitz, 1984; Milner, J.S., Gold, R.G., and R.C. Wimberley, 1986; Milner, J.S., Robertson, K.R., and D.L. Rogers, 1990).
8. By 1967, all states had adopted mandatory reporting laws, largely in response to the widely publicized cases of child maltreatment following the publication of "The Battered Child" paper by Kempe and colleagues (1962).
9. For example, officials at the National Institute on Mental Health estimate that that agency alone spent $4 million in FY 1992 on research on childhood autism and pervasive developmental disorders.
10. Although most state statutes include some reference to the concept of child maltreatment, emotional maltreatment is seldom defined and few cases are processed as such (Hart et al., 1987).
The validity of the figures in the National Incidence Study are difficult to assess, given the ambiguous diagnostic criteria and the large number of agencies asked to report. In the 1986 National Incidence Study 3.4 per 1,000 children per year were estimated to be victims of emotional abuse or in danger of emotional abuse. These rates are higher than those for sexual abuse and they may reflect the sources of data for the study: schools, day care, and social service agencies (Knudsen, 1992). An analysis of official reports to child protection agencies in 1986 by the American Humane Association indicated that 8.3 percent of all reported cases included emotional abuse or neglect (1.1 per 1,000 children per year were victims of emotional maltreatment). Differences in estimates between the incidence of emotional maltreatment in the National Incidence Study (which suggests that 435,000 children were victims of child maltreatment) and the American Humane Association analysis of reports (which suggests that 28,000 children were identified as victims of emotional maltreatment) primarily reflect the problems with definitions and diagnosis. The differences in these rates may demonstrate, in part, that only a fraction of the cases of emotional maltreatment are reported. Unless it co-occurs with other forms of severe abuse, emotional maltreatment is less likely to be documented, and less likely to receive intervention, than other forms of child maltreatment (Hart and Brassard, 1991).
11. For further information about these and other national child surveys see the following references (Zill and Coiro, 1992):
1) National Health Interview Survey on Child Health (National Center for Health Statistics, 1989; Dawson, 1991; and Zill, Moore, Smith, Stief, and Coiro, 1991);
2) National Survey of Children (Peterson and Zill, 1986; Allison and Furstenburg, 1989); and
3) Child Supplement to the National Longitudinal Survey of the Labor Market Experience of Youth (Baker and Mott, 1989).
Abel, G., J. Becker, J. Cunningham-Rathner, and M. Mittelman
1988 Multiple paraphiliac diagnosis among sex offenders. Bulletin of the American Academy of Psychiatry and the Law 16(2):153-68.
Aber, J.L., and E. Zigler
1981 Developmental considerations in the definition of child maltreatment. New Directions for Child Development 11(1):1-29.
Allison, P.D., and F.F.Furstenburg, Jr.
1989 How marital dissolution affects children: Variations by age and sex. Developmental Psychology 25:540-549.
American Humane Association
1979 Annual Statistical Report: National Analysis of Child Neglect and Abuse Reporting, 1978, Englewood, Colorado.
1981 National Analysis of Official Child Neglect and Abuse Reporting, 1979. Englewood, Colorado.
American Association for Protecting Children
1986 Highlights of Official Child Neglect and Abuse Reporting, 1984. Englewood, Colorado: American Humane Association.
1987 Highlights of Official Child Neglect and Abuse Reporting, 1985. Denver, Colorado: American Humane Association.
Baker, P.C., and F.L.Mott
1989 A Guide and Resource Document for the National Longitudinal Survey of Youth 1986, Child Data. Columbus: The Ohio State University, Center for Human Resource Research.
1991 Treatment Methods for Perpetrators of Child Sexual Abuse. Background paper for the Child Abuse Treatment Working Group of the American Psychological Association.
Berger, A.M., J.F. Knutson, J.G. Mehm, and K. Perkins
1988 The self-report of punitive childhood experiences of young adults and adolescents. Child Abuse and Neglect 12:251-262.
Bertolli, J., H. Morganstern, and S. Sorenson
1992 The Occurrence of Child Abuse and Neglect: Issues, Findings, and Directions. Background Paper prepared for the National Research Council's Panel on Research on Child Abuse and Neglect.
Borsig, C., and S. Kalichman
in press Clinicians reporting of suspected abuse: A review of the empirical literature. Clinical Psychology Review.
1992 Methodological issues in the study of sexual abuse effects. Journal of Consulting and Clinical Psychology 60(2):196-203.
Briere, J., and M. Runtz
1987 Post sexual abuse trauma: Data and implications for practice. Journal of Interpersonal Violence 2:367-379.
Cicchetti, D., and D. Barnett
1991 Toward the development of a scientific nosology of child maltreatment. Pp. 346-377 in D. Cicchetti and W. Grove, eds., Thinking Clearly about Psychology: Essays in Honor of Paul E. Meehl. Minneapolis: University of Minnesota Press.
1992 Preventing Child Neglect. Chicago: National Committee for Prevention of Child Abuse.
Daro, D., and K. McCurdy
1991 Current Trends in Child Abuse Reporting and Fatalities: The Results of the 1990 Annual Fifty State Survey. Chicago: National Committee for Prevention of Child Abuse.
Daro, D., K. Casey, and N. Abrahams
1990 Reducing Child Abuse 20% by 1990: Preliminary Assessment. National Committee for Prevention of Child Abuse. Working Paper No. 843.
1991 Family structure and children's health and well-being: Data from the 1988 national health interview survey on child health. Journal of Marriage and the Family 53:573-584.
Della Femina, D., C.A. Yeager, and D.O. Lewis
1990 Child abuse: Adolescent records vs. adult recall. Child Abuse and Neglect 14:227-231.
Division of Injury Control, Centers for Disease Control
1990 Childhood injuries in the United States. American Journal of Diseases of Children 144:627-646.
Eckenrode, J., and J. Doris
1987 Unreliable Child Maltreatment Reports: Variations Among Professional and Nonprofessional Reporters. Washington, DC: National Center on Child Abuse and Neglect.
Eckenrode, J., J. Powers, J. Doris, J. Munsch, and N. Bolger
1988a Substantiation of child abuse and neglect reports. Journal of Consulting and Clinical Psychology 56(1):9-16.
Eckenrode, J., J. Munsch, J. Powers, and J. Doris
1988b The nature and substantiation of official sexual abuse reports. Child Abuse and Neglect 12:311-319.
Ewigman, B., C. Kivlahan, and G. Land
1993 The Missouri child fatality study: Underreporting of maltreatment fatalities among children younger than five years of age, 1983 through 1986. Pediatrics 91(2):330-337.
Fehrenbach, P.A., W. Smith, C. Monastersky, and R. W. Deisher
1986 Adolescent sex offenders: Offender and offense characteristics. American Journal of Orthopsychiatry 56:225-233.
1985 The relationship among dissociation, hypnosis, and child abuse in the development of multiple personality disorder. Pp. 100-126 in R.P. Kluft, ed., Childhood Antecedents of Multiple Personality. Washington, DC: American Psychiatric Press, Inc.
Fromuth, M.E., and B.R. Burkhart
1987 Childhood sexual victimization among college men: Definitional and methodological issues. Violence & Victims 2(4):241-252.
1987 The abuse and neglect of special children: An introduction to the issues. Chapter in Garbarino, J. et al., eds., Special ChildrenSpecial Risks. New York: Aldine De Gruyter.
1978 Violence towards children in the U.S. American Journal of Orthopsychiatry 48:580-592.
1982 Problems in defining and labeling child abuse. Pp. 1-30 in R.H. Starr, ed., Child Abuse Prediction: Policy Implications. Cambridge, MA: Ballinger.
1970 Violence Against Children: Physical Child Abuse in the United States. Cambridge, MA: Harvard University Press.
1989 Substantiated and unsubstantiated reports of child maltreatment. Child Youth Service Review 11:299-318.
Giovannoni, J.M., and A. Billingsley
1970 Child neglect among the poor: A study of parental adequacy in families of 3 ethnic groups. Child Welfare 49(April):196-204.
1988 Special groups of children at risk of abuse: The disabled. Pp. 223-239 in M. Straus, ed., Abuse and Victimization Across the Life Span. Baltimore: Johns Hopkins University Press.
1929 A Research in Marriage. New York: Albert and Charles Boni.
Hampton, R.L., and E.H. Newberger
1985 Child abuse incidence and reporting by hospitals: Significance of severity, class, and race. American Journal of Public Health 75(January):56-60.
Hart, S.N., and M.R. Brassard
1991 Developing and Validating Operationally Defined Measures of Emotional Maltreatment. Report supported by the National Center on Child Abuse and Neglect Grant 90CA1216.
Hart, S.N., R. Germain, and M.R. Brassard
1987 The challenge: To better understand and combat the psychological maltreatment of children and youth. In M.R. Brassard, R. Germain, and S.N. Hart, eds., Psychological Maltreatment of Children and Youth. New York: Pergamon Press.
Haugaard, J.J., and R.E. Emery
1989 Methodological issues in child sexual abuse research. Child Abuse and Neglect 13:89-100.
1987 The litany of the smoldering neglect of children. Pp. 301-311 in R.E. Helfer and R.S. Kempe, eds., The Battered Child, Fourth Edition, Revised and Expanded. Chicago: University of Chicago Press.
Herzberger, S.D., and H. Tennen
1983 Coping with abuse: Children's perspectives on their abusive treatment. In D. Finkelhor, ed., The Dark Side of Families. Beverly Hills, CA: Sage.
Hoaglin, D.C., R.J. Light, B. McPeek, F. Mosteller, and M. Stoto
1982 Data for Decisions: Information Strategies for Policymakers. Cambridge, MA: Abt Books.
Horowitz, B., and I. Wolock
1981 Maternal deprivation, child maltreatment, and agency interventions among poor
families. Pp 137-184 in L. Pelton, ed., The Social Context of Child Abuse and Neglect. New York: Human Sciences Press.
Institute for Sex Research
1953 Sexual Behavior in the Human Female. Philadelphia: W.B. Saunders.
Institute of Medicine
1989 Research on Children and Adolescents with Mental, Behavioral, and Developmental Disorders. Washington, DC: National Academy Press.
Jason, J., N.D. Andereck, J. Marks, and C.W. Tyler, Jr.
1982 Child abuse in Georgia: A method to evaluate risk factors and reporting bias. American Journal of Public Health 72(December):1353-1358.
Kleinbaum, D.G., L.L. Kupper, and H. Morgenstern
1982 Epidemiologic Methods: Principles and Quantitative Methods. New York: Van Nostrand Reinhold Co.
1992 Child Maltreatment: Emerging Perspectives. Dix Hills, NY: General Hall, Inc.
1987 Child abuse and neglect: The cultural context. Pp. 23-41 in R. Helfer and R. Kempe, eds., The Battered Child. Chicago: University of Chicago Press.
Lauderdale, M., A. Valiunas, and R. Anderson.
1980 Race, ethnicity, and child maltreatment: An empirical analysis. Child Abuse and Neglect 4:163-169.
1982 Research strategies and methodologic standards in studies of risk factors for child abuse. Child Abuse and Neglect 6(1):113-123.
1988 Have there been changes in the epidemiology of child abuse during the 20th century? Pediatrics 82(5): 766-773.
1990 Epidemiology of child sexual abuse. Pp. 18-41 in R.M. Oates, ed., Understanding and Managing Child Sexual Abuse. Sydney, Australia: Harcourt Brace Jovanovich.
Leventhal, J., S. Horwitz, C. Rude, and D. Stier
1993 Maltreatment of children born to teenage mothers: A comparison between the 1960s and the 1980s. Journal of Pediatrics 122:314-319.
Lewis, D.O., C.A. Yeager, C.S. Cobham-Portorreal, N. Klein, C. Showalter, and A. Anthony
1991 A follow-up study of female delinquents: Maternal contributions to the perpetuation of delinquency. Journal of the American Academy of Child and Adolescent Psychiatry 30(2):197-201.
1973 Abused and neglected children in America: A study of alternative policies. Harvard Educational Review 43:556-598.
Mash, E.J., and D.A. Wolfe
1991 Methodological issues in research on physical child abuse. Criminal Justice and Behavior 18(1)(March):8-29.
McClain, P.W., J.J. Sacks, R.G. Froehlke, and B.G. Ewigman
1993 Estimates of fatal child abuse and neglect, United States, 1979 through 1988. Pediatrics 91(2):338-343.
Milner, J.S., and R.C. Wimberley
1979 An inventory for the identification of child abusers. Journal of Clinical Psychology 35(1)(January):95-100.
Milner, J.S., Gold, R.G., Ayoub, C., and M.M. Jacewitz
1984 Predictive validity of the Child Abuse Potential Inventory. Journal of Consulting and Clinical Psychology 52:879-884.
Milner, J.S., Gold, R.G., and R.C. Wimberley
1986 Prediction and explanation of child abuse: Cross-validation of the Child Abuse Potential Inventory. Journal of Consulting and Clinical Psychology 54:865-866.
Milner, J.S., Robertson, K.R., and D.L. Rogers
1990 Childhood history of abuse and adult child abuse potential. Journal of Family Violence 5(1):15-34.
1982 Uses of ecologic analysis in epidemiologic research. American Journal of Public Health 72:1336-44.
1975 Child abuse and neglect programs: A national overview. Children Today (May/June):13-17.
National Center for Child Abuse and Neglect
1981 National Study of the Incidence and Severity of Child Abuse and Neglect. 81-30325. Washington, DC: U.S. Department of Health and Human Services. [NIS 1]
1988 Study Findings: Study of National Incidence and Prevalence of Child Abuse and Neglect. Washington, DC: U.S. Department of Health and Human Services. [NIS 2]
1992 National Child Abuse and Neglect Data System, Working Paper 1, 1990 Summary Data Component. April. Washington, DC: U.S. Department of Health and Human Services.
National Center for Health Statistics
1989 Vital Statistics of the United States, 1987, Vol. 1, Natality. Washington, DC: Government Printing Office, DHHS Publication No(PHS)89-1100.
National Commission on Children
1991 Beyond Rhetoric: A New American Agenda for Children and Families. Washington, DC: Government Printing Office.
Newberger, E., R. Reed, J. Daniel, J. Hyde, and M. Kotelchuck
1977 Pediatric social illness: Toward an etiologic classification. Pediatrics 60:178-185.
Peters, S.D., G.E. Wyatt, and D. Finkelhor
1986 Prevalence. In D. Finkelhor, ed., A Sourcebook on Child Sexual Abuse. Beverly Hills, CA: Sage Publishers.
Peterson, J.L., and N. Zill
1986 Marital disruption, parent-child relationships, and behavioral problems in children. Journal of Marriage and the Family 48:295-307.
Polansky, N.A., M.A. Chalmers, E. Buttenweiser, and D.P. Williams
1981 Damaged Parents: An Anatomy of Child Neglect. Chicago: University of Chicago Press.
1983 The incidence and prevalence of intrafamilial and extrafamilial sexual abuse of female children. Child Abuse and Neglect 7:133-146.
1990 Technical Amendments to the Study FindingsNational Incidence and Prevalence of Child Abuse and Neglect. [NIS 2] 1988. May 23, 1990.
1992 Issues Presented by Mandatory Reporting Requirements. Paper prepared for the National Research Council, Commission on Behavioral, and Social Sciences, and Education, Panel on Research on Child Abuse and Neglect.
Stein, A., and D.O. Lewis
1992 Discovering physical abuse: Insights from a follow-up study of delinquents. Child Abuse and Neglect 16:523-531.
1979 Family patterns and child abuse in a nationally representative sample. Child Abuse and Neglect 3: 23-25.
1980 Stress and physical child abuse. Child Abuse and Neglect 4:75-88.
1990a Measuring Psychological and Physical Abuse of Children with the Conflict Tactics Scales. Family Research Laboratory, University of New Hampshire.
1990b The Conflict Tactics Scales and its critics: An evaluation and new data on validity and reliability. In M.A. Straus and R.J. Gelles, Physical Violence in American Families: Risk Factors and Adaptations to Violence in 8145 Families. New Brunswick, NJ: Transaction Publications.
Straus, M.A., and R.J. Gelles
1986 Societal change in family violence from 1975 to 1985 as revealed by two national surveys. Journal of Marriage and the Family 48(August): 465-479.
1956 Clinical Studies in Psychiatry. New York: W.W. Norton.
Turner, C.F., J.T. Lessler, and J.C. Gfroerer, eds.
1992 Survey Measurement of Drug Use: Methodological Studies. Washington, DC: National Institute on Drug Abuse. Division of Epidemiology and Prevention Research.
U.S. Advisory Board on Child Abuse and Neglect
1990 Child Abuse and Neglect: Critical First Steps in Response to a National Emergency. Washington, DC: U.S. Department of Health and Human Services.
1988 Sampling biases and implications for child abuse research. American Journal of Orthopsychiatry 58(2):260-270.
1989 Does violence beget violence? A critical examination of the literature. Psychological Bulletin 106(1):3-28.
1985 The effect of child abuse on the psyche. Pp. 22-35 in R.P. Kluft, ed., Childhood Antecedents of Multiple Personality. Washington, DC: American Psychiatric Press.
Williams, B.C., and J.B. Kotch
1990 Excess injury mortality among children in the United States: Comparison of recent international statistics. Pediatrics 86(6):982-987.
Wolfe, D.A., and M.D. Mosk
1983 Behavioral comparisons of children from abusive and distressed families. Journal of Consulting and Clinical Psychology 51:702-708.
Wolock, I., and B. Horowitz
1979 Child maltreatment and material deprivation among AFDC-recipient families. Social Service Review 53:175-194.
World Health Organization
1977 World Health Classification: Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death. 9th revision. Geneva, Switzerland: World Health Organization.
1985 The sexual abuse of Afro-American and white American women in childhood. Child Abuse and Neglect 9:507-519.
Wyatt, G.E., and S.D. Peters
1986a Issues in the definition of child sexual abuse in prevalence research. Child Abuse and Neglect 10:231-240.
1986b Methodological considerations in research on the prevalence of child sexual abuse. Child Abuse and Neglect 10:241-251.
Wyatt, G.E., L. Lawrence, A. Vodounon, and M.R. Mickey
1993 The Wyatt Sex History Questionnaire: A structured interview for sexual history taking. Journal of Child Sexual Abuse 48(4) (Summer):51-68.
1990 Child abuse reporting and failure to report among mandated reporters. Journal of Interpersonal Violence 5(1):3-22.
Zill, N., and M.J. Coiro
1992 Assessing the condition of children. Children and Youth Services Review 14:119-136.
Zill, N., Moore, K.A., Smith, E.W., Stief, T., and M.J.Coiro
1991 Life Circumstances and Development of Children in Welfare Families: A Profile Based on National Survey Data. Washington, DC: Child Trends.