• Family Surveillance: Police and the Reporting of Child Abuse and Neglect

Police are responsible for producing about one-fifth of all reports of child abuse and neglect investigated by local child welfare agencies, and low-level interactions with police often result in the initiation of a child welfare investigation. Because police contact is not randomly or equitably distributed across populations, policing has likely spillover consequences on racial inequities in child welfare outcomes. This study shows that police file more reports of child abuse and neglect in counties with high arrest rates, and that policing helps explain high rates of maltreatment investigations of American Indian– Alaska Native children and families. The spatial and social distribution of policing affects which children and families experience unnecessary child protection interventions and which children who are victims of maltreatment go unnoticed.

Keywords

child protection, surveillance, policing, family, child abuse, neglect

Police routinely interact with families and children and have exceptionally intimate access to the interactions of parents and children. Unlike doctors, educators, or social service providers, police can gain access to observe the daily lives of children and families at home with or without the consent of a subject family. Whereas other street-level bureaucrats use passive surveillance of children and families, police can engage in an active and coercive manner to monitor and regulate family life (Lipsky 1980). State and federal policymakers have long recognized the capacity of the police to engage in intensive family surveillance. In all U.S. states, police are required to report suspected child abuse and neglect to local child protection agencies. They do so quite frequently. In 2015, police originated about four hundred thousand reports to child welfare agencies alleging abuse or neglect, nearly one-fifth of the national total (Children’s Bureau 2017).

This study describes the interactions between police and child welfare agencies and explores whether exposure to policing helps explain how and why certain children enter the child welfare system. After describing the kinds of cases police report to child welfare agencies and the distribution of police reporting of child abuse and neglect across U.S. counties, it constructs a series of regression models to evaluate [End Page 50] whether variation in police activities is predictive of the intensity of maltreatment reporting. It then evaluates whether racial inequalities in exposure to policing contribute to racial inequalities in contact with the child welfare system (Roberts 2002; Wulczyn et al. 2013). These analyses use restricted data from the National Child Abuse and Neglect Data System (NCANDS), and provide the first systematic analysis of police involvement in the child welfare system across nearly all U.S. counties.1

Contact with the criminal justice system has a host of consequences for families (Comfort 2008; Braman 2007; Wildeman and Muller 2012; Roberts 2012; Wildeman and Wang 2017). The incarceration of a family member strains the emotional and material resources of children’s caregivers in ways that can have complex and disruptive effects on families (Wakefield and Wildeman 2014; Wildeman 2014; Turney 2014; Foster and Hagan 2015). The arrest or incarceration of a parent or caregiver may present both an immediate and a long-term crisis for the care of children, demanding that either kin, fictive kin, or the state step in to provide care for children (Andersen and Wildeman 2014; Berger et al. 2016; Comfort 2008, 2016; Roberts 2012). We know that parental incarceration has detrimental impacts on children and families. What this study illustrates is that even low-level contact with the criminal justice system exposes children and families to the risk of serious disruption through the deep interconnection of policing and child protection.

Policing likely provides a partial explanation for racial inequalities in child welfare system outcomes (Roberts 2002; Kim et al. 2016; Wildeman et al. 2014). As a key component of American family surveillance systems, local police agencies play a role in shaping the composition of the population of children and families singled out for maltreatment investigations. Agency and officer decisions about where to patrol, what to enforce, who is suspicious, and whether to make an arrest all play a role in determining which families are subject to surveillance and which are not.

the intersections of criminal justice and child protection

Contact with the child welfare system is incredibly common. About 37 percent of children in the United States will experience a child welfare maltreatment investigation during their childhood (Kim et al. 2016). About 12 percent will experience a confirmed case of child maltreatment before they turn eighteen (Wildeman et al. 2014). The likelihood of interacting with child welfare systems is dramatically higher for children of color. About half of all African American children will experience a child welfare investigation before their eighteenth birthday (Kim et al. 2016). In 2015, about four million children were reported to local child welfare agencies, of whom more than three million were screened in and received some form of agency response. About 5 percent of the U.S. child population was the subject of a report to child welfare agencies at some point in 2015 (Children’s Bureau 2017). The prevalence of arrest follows a strikingly similar distribution: about 30 percent of Americans but 49 percent of young black men will experience an arrest by age twenty-three (Brame et al. 2012, 2014). The FBI Uniform Crime Reports show a national arrest rate in 2015 of about forty-five arrests per thousand adults, an incidence rate quite similar to the per capita rate of child abuse and neglect reporting (author’s calculation).

Criminal justice and child welfare systems are likely to be most active in similar communities and neighborhoods, and overlapping contact with criminal justice and child welfare systems within families is common (Berger et al. 2016; Roberts 2012). Using administrative data in Wisconsin, Lawrence Berger and his colleagues find that 28 percent of children involved [End Page 51] in the child welfare system between 2004 and 2012 in Milwaukee County had a parent in jail or prison within a year of their contact with the child welfare system (2016). They further find that 18 percent of incarcerated eighteen-to twenty-one-year-olds in Wisconsin were involved with child welfare agencies as adolescents. Ethnographic work suggests that the communities in which child protection and police departments are most aggressive and most active often overlap (Fernandez-Kelly 2015; Roberts 2008).

Police agencies have deep institutional ties to child protection agencies. Child welfare agencies routinely conduct joint investigations with police, many police departments have created special units directed at child abuse and neglect, and police themselves handle non-criminal maltreatment investigations in some jurisdictions (Cross et al. 2015; Cross, Finkelhor, and Ormrod 2005). Regardless of jurisdiction, however, police play a fundamental role in child protection: they conduct front-line surveillance of children for signs of abuse and neglect; they produce information about the fitness of an adult to parent through the application of criminal stigma; and they create both short-and long-term crises of care when they incapacitate caregivers.

Police suspicion is likely to affect bureaucratic appraisals of the incidence of abuse or neglect within a family through the application of criminal stigma. Criminal records and arrests convey a powerful social signal to street-level bureaucrats and other community members. Places with more aggressive police forces mark larger proportions of their population with racialized and gendered criminal stigmas connoting irresponsibility and dangerousness (on racialized stigmas, Asad and Clair 2017; Harris, Evans, and Beckett 2011; on gendered stigmas, Rios 2011; Haney 2010). These stigmas likely affect child welfare system decision-making about the fitness of parents (Vesneski 2012).

Police are not dispassionate or objective instruments of social measurement. The social (and spatial) organization of policing is informed by and reproduces entrenched racialized and gendered inequalities (Beckett, Nyrop, and Pfingst 2006; Haney 2010; Gilmore 2007; Epp, Maynard-Moody, and Haider-Markel 2014; Lerman and Weaver 2014; Soss and Weaver 2017; Roberts 2012). The distribution of policing is not socially uniform (Carmichael and Kent 2014; Capers 2009; Perry 2009b). Further, criminal-legal decision-making is systematically related to race, class, and gender (Harris 2016; Murakawa and Beckett 2010; Haney 2000; Steen, Engen, and Gainey 2005; Rios 2011). These persistent and widespread inequalities in exposure to policing may be responsible for exacerbating racial inequalities in family exposure to the child welfare system.

child protection systems, family surveillance, and family regulation

Child protection systems are responsible for the investigation of alleged child abuse and neglect and are empowered to separate children from their families. Like the police, they are charged with the identification and regulation of unlawful and deviant behavior. However, unlike the police, child welfare agencies are tasked with an explicitly therapeutic and rehabilitative mission. Agencies often help children and families access housing, medical, counseling, and other benefits and services, and children in state custody become automatically eligible for a wide range of state and federal benefits. However, participation in these services is often unwanted and involuntary, because agencies may require compliance with case plans as a condition to allow children to remain or return. American child protection agencies operate in a distinctly coercive and paternalistic manner (Edwards 2016; Gilbert 2012). They require families to pursue what the state determines to be the best interests of children. Child welfare systems ensure that parents comply with agency and court mandates through the implicit or explicit threat of family separation.

However, child welfare agencies lack the direct surveillance capacities required to detect child abuse and neglect in communities. They depend on schools, police, medical professionals, social service agencies, and the community at large to act as their eyes and ears (Wells et al. 2014; Aleissa et al. 2009). This diffuse surveillance system is formalized by mandated reporting laws. In all states, professionals who routinely [End Page 52] interact with families and children are required by law to report suspected child abuse or neglect, and a growing number of states have passed universal mandated reporting laws, which extend this obligation to a state’s entire adult population (Krase and DeLong-Hamilton 2015; Drake and Jonson-Reid 2015; Raz 2017). This dependency on external agencies to originate reports of abuse and neglect is a likely source of variation in the flow of cases into the child welfare system. This institutional feature—a multi-organizational system of mal-treatment surveillance—also creates conditions under which inequalities generated from one set of state actors can cause inequalities in proximate policy areas.

Surveillance is a process that requires a series of interactions and decisions. Prior to the generation of an agency investigation, a child or family must have contact with some professional or community member capable of monitoring the family. That observer must then use cultural scripts and institutional routines to classify a family interaction as normal or deviant. Following this classification, the observer may choose to submit a formal report to the relevant agency, and that agency must decide whether to respond to the allegation. Under this model, exposure to potential reporters, classification and reporting routines, local law, and agency rules for responding to cases all play a role in determining which children and families come under investigation. In this analysis, I direct attention to the first stage of this process by evaluating whether the rate of contact between police and community members is systematically related to the rate at which police report suspected child abuse and neglect.

data and measures

Outcomes for this study are constructed from the National Child Abuse and Neglect Data System, the federal data system responsible for tracking child maltreatment investigations and responses. NCANDS records case-level information on all investigated reports of child mal-treatment annually with data reported from state and local child welfare agencies to the federal government. It is the most comprehensive source for national information on suspected child abuse and neglect, and contains several million records annually.

Child Maltreatment Surveillance

I construct counts of investigated maltreatment reports initiated by police at the county-year level. NCANDS does not capture reports of child abuse and neglect that are screened out as not requiring an investigation by child welfare agencies. Processes for classifying reports of alleged child abuse and neglect as worthy of investigation or response vary by jurisdiction, but are not quantifiable with current federal data. Because NCANDS does not record reports that are screened out and receive no agency response, the rates of police reporting of mal-treatment presented here are conservative estimates.

Although comprehensive, the quality of NCANDS data varies by jurisdiction. Some counties have high levels of missing data on focal variables for this study likely related to agency data collection practices. All county-years in which more than 10 percent of reports are missing data on the original source of the investigated maltreatment report are treated as missing, as are those in which more than 10 percent of reports are missing data on the race of the investigated child. This procedure results in treating about 8 percent of county-year counts of police-initiated maltreatment reports as missing. Multiple imputation models address this and other sources of missing data and measurement error.2 Because they are subject to unstable rate measures, all observations for county-years in which the population of children in the county by race is less than ten are excluded from the regression models.

Restricted versions of the NCANDS data allow for much higher coverage of U.S. counties and the U.S. child population than was possible with alternative versions of the data. Data are included on maltreatment reporting in 3,064 of the 3,142 U.S. counties or county-equivalent [End Page 53] units. By contrast, previous work using these data has been able to include geographic information on about only six hundred counties. In 2015, this sample includes data from counties representing 99 percent of the U.S. Asian– Pacific Islander child population, 96 percent of the American Indian– Alaska Native child population, and more than 99 percent of the African American, Latino, white, and total U.S child population. Descriptive statistics on police-initiated maltreatment reports with Latino/a child subjects are presented, but the role of Latino/a ethnicity is not evaluated in regression models that include criminal justice data because federal criminal justice data systems systematically underreport Latino/a arrests (Nellis 2016).

crime, policing, and arrests

Focal predictors are constructed from the Uniform Crime Reports (UCR) arrests by age, sex, and race annual data for 2009 through 2015 (Federal Bureau of Investigation 2014a). The UCR, collected by the FBI and maintained by the National Archive of Criminal Justice Data, provides the only national time series data on law enforcement activity available at the jurisdiction level and covering the period of interest for this study. This series provides data on arrests by race at the police agency level aggregated to the county level. I also include information on the number of officers employed by police agencies at the county-year level from the UCR Police Employee Data (Federal Bureau of Investigation 2014b).

Using four offense categories, I create county-year sums for all arrests, and for arrests by race. Violent offenses include murder, manslaughter, rape, robbery, and aggravated assault, following the FBI’s classification of violent offenses in the UCR index crime classification system. I also rely on the UCR’s classification of drug offenses, a set that includes either possession or sale of opiates, marijuana, synthetic narcotics, and other dangerous non-narcotic drugs. Quality-of-life policing captures a more diffuse set of offenses that are generally low level and subject to high officer and agency discretion in enforcement. These include vandalism, liquor laws, drunkenness, disorderly conduct, vagrancy, general suspicion, curfew and loitering, and various gambling offenses.

Reporting to the UCR is voluntary. Some police agencies faithfully report data annually, some do so intermittently, some never do, and some submit reports subject to clear error (these limitations are described in more detail in the discussion). For agencies with intermittent reporting, I replace all unreported years with linear interpolations constructed from the 2002 through 2015 data.3 I treat counties as missing that include agencies with known or likely reporting error, including all counties with more than one thousand adults that report no arrests.4 I construct imputation models to address both these explicit and identified implicit missing cases (Honaker, King, and Blackwell 2011; Honaker and King 2010).

demographic data and measures

Of course, rates of reported child abuse and neglect are sensitive to actual rates of child abuse and neglect. No direct measures of the incidence of child abuse and neglect are available, and few measures of child well-being, child injury, and family stability are available for all U.S. counties by race in reliable time series. Child poverty is widely regarded by child maltreatment researchers as among the best predictors of abuse and neglect (Sedlak et al. 2010). Although the measure has many known flaws, it provides a crude indicator of child and family well-being available for all U.S. counties annually by race (Brady 2009). I include measures of child poverty per capita by race between 2009 and 2015 using the American Community Survey (ACS) five-year data (Ruggles et al. 2010). Because many U.S. counties have very small populations, ACS five-year estimates are far more reliable than three-year or single-year [End Page 54] estimates. Five-year estimates are available only beginning in 2009, however, so I limit the regression models to the years 2009 through 2015.

I also include urban-rural classifications from the National Center for Health Statistics to control for potential differences in policing and child welfare system operations across metropolitan types. The Centers for Disease Control and Prevention produce annual population estimates for counties by race, sex, and age through the Surveillance, Epidemiology, and End Results Program, providing a reliable time series of adult and child populations by race for all counties and years available in the NCANDS data.

analytic strategy

I first describe the kinds of cases police report to child protection agencies, the kinds of children and families police report, and the outcomes of police-reported maltreatment cases. I also evaluate national temporal trends in police-reported child maltreatment. I compare patterns of police-originated maltreatment reports to all maltreatment reports to explore whether police are more likely than other kinds of reporters to capture particular kinds of mal-treatment, to report children of color, or to generate substantiated cases of child abuse and neglect.5

Next, I evaluate whether the activities of local police agencies are systematically related to police-originated child maltreatment reporting. I construct a series of multilevel regression models that evaluate the relationships between rates of arrest by offense and race and the volume of police-originated maltreatment reports by race at the county-year level. These models include child poverty rates, measures of county racial composition, urban-rural classification codes, police staffing levels, county-level random intercepts, and a national linear trend. Arrest, poverty, and composition measures interact with race, allowing for varying linear relationships between police reporting of mal-treatment and regression predictors for all children, African American children, Asian– Pacific Islander children, American Indian– Alaska Native children, and white children. The outcomes for all regression models are counts of investigated reports of child abuse or neglect initially filed by police offset by the size of the focal child population.

Different kinds of enforcement may be more likely to result in police interaction with families and are subject to varying degrees of agency discretion. Police responses to violent and property offenses are largely reactive, in response to public calls or complaints. By contrast, police departments have more flexibility in deciding whether, where, and how to enforce drug laws (Beckett, Nyrop, and Pfingst 2006). They also have considerable flexibility in deciding whether to aggressively police low-level violations, in strategies frequently described as broken windows or quality-of-life policing (Fagan and Davies 2000; Gelman, Fagan, and Kiss 2007). We are more likely to capture discretion in law enforcement by comparing rates of drug and quality-of-life arrests across jurisdictions. I therefore separately model the relationships between police-initiated child protection reports and all arrests, for arrests for violent offense arrests, for drug offenses, and for quality-of-life offenses.

Because both policing and child protection are deeply racialized and gendered legal and administrative practices, heterogeneity may be substantial in the relationships between policing and family surveillance for families of color relative to white families (Roberts 2012; Rios 2011; Soss and Weaver 2017). Ideas about parental fitness are deeply intertwined with race, class, and family structure in ways that may affect the likelihood of a maltreatment report (Roberts 1997, 2012; Masters, Lindhorst, and Meyers 2014). Police may assess families of color as more dangerous to children’s well-being than similar white families and may be more likely to have routine contact with families of color than with white families. Racial heterogeneity in risk of maltreatment surveillance and reporting would produce varying relationships between policing and reporting even after accounting for racial inequalities in rates of police contact. To account for these [End Page 55] varying relationships, I model race-specific relationships for arrests, child poverty, and racial composition.

Fixed-effects approaches to multilevel data have the advantage of purging unmeasured heterogeneity in individual units from models, enabling close evaluation of longitudinal within-unit trends, but they also inhibit cross-unit comparisons and provide little insight into time-stable features or slow-moving processes. I pursue a random-effects approach that estimates parameters for both average values of county-level covariates and for annual mean-difference values for county-year level covariates (Bell and Jones 2015). Evidence for short-term within-county relationships may provide stronger evidence of a potentially causal relationship; cross-county relationships identify whether variation in policing regimes across places helps explain the high geographic variation in maltreatment reporting.

For county i, year j, race k, child population m, and predictors X, I estimate multilevel Poisson models for counts of investigated maltreatment reports Y. These models estimate county-level intercepts, observation-level intercepts to model overdispersion, and race-specific intercepts and slopes for focal variables. Additionally, they estimate parameters for relationships between police maltreatment reporting and both county-average values (cross-sectional variation) and annual changes (longitudinal variation) for focal measures. I separately estimate models for each offense category. These models take the following general form:

Y ~ Poisson

inline graphic

findings

Police are less likely than others are to file mal-treatment reports involving physical abuse (see table 1). In 2015, about 17 percent of all investigated maltreatment reports filed by police involved physical abuse, versus 22 percent of reports from all sources. Given the central role of police in responding to family violence, this is surprising. Police were more likely than other reporters to file reports involving allegations of psychological maltreatment or sexual abuse of children. As with other reporters, neglect is the primary form of suspected maltreatment in most police-filed reports. More than 70 percent of police maltreatment reports involving American Indian– Alaska Native children in 2015 centered on allegations of neglect.

Children are far more likely to be classified as victims of child abuse and neglect following a child welfare agency assessment when they are reported by police than when reports originate from another source. Although 22 percent of all investigations result in a conclusion that a child was a victim of abuse or neglect, 39 percent do when police file the initial report. Substantiation rates for children of all racial and ethnic groups are similar, concentrated at around 40 percent for police-filed reports and 22 percent for reports from all sources (see table 2).

Rates of Police Reporting

Police filed at least 394,482 reports of child abuse and neglect in 2015, about 19 percent of the more than 2.2 million investigated reports that year (see table 3).6 Among professionally mandated reporters of child abuse and neglect, teachers and police have nearly identical rates and most frequently report maltreatment to local child welfare agencies (Children’s Bureau 2017). About 30.9 reports of child maltreatment per thousand children in 2015 were investigated and about 5.8 were initiated by police.

Racial and ethnic inequalities in the rates at which child abuse and neglect are reported and investigated are substantial (see table 3). In [End Page 56]

Table 1. Reports by Type of Alleged Maltreatment in 2015
Click for larger view
View full resolution
Table 1.

Reports by Type of Alleged Maltreatment in 2015

Table 2. Substantiation Rates for All Cases and for Police-Initiated Maltreatment Reports
Click for larger view
View full resolution
Table 2.

Substantiation Rates for All Cases and for Police-Initiated Maltreatment Reports

Table 3. Investigated Child Abuse and Neglect Reports in 2015
Click for larger view
View full resolution
Table 3.

Investigated Child Abuse and Neglect Reports in 2015

[End Page 57]

2015, about 9.7 police reports of child maltreatment were investigated per thousand African American children, about 4.3 per thousand American Indian– Alaska Native children, about 1.2 per thousand Asian– Pacific Islander children, about 4.7 per thousand Latino/a children, and about 5.1 per thousand white children. Black children were subject to 1.9 times more police-initiated maltreatment investigations than white children; American Indian– Alaska Native, Latino/a, and Asian– Pacific Islander children were all subject to a lower rate of police reporting of maltreatment than their white counterparts. Although the share of maltreatment reports filed by police appears relatively constant by race in 2015, the increase in both the share of maltreatment reports filed by police and the volume of police reporting of mal-treatment between 2002 and 2015 for most children has been notable (see figure 1).

The rate of police reporting of maltreatment increased for three groups—African American, Latino/a, and white children—between 2002 and 2015. Black families saw a 60 percent increase, Latino/a families a 23 percent increase, and white families a 39 percent increase. Rates of police reporting on Asian– Pacific Islander and American Indian– Alaska Native families remained relatively stable between 2002 and 2015. The proportion of all investigated mal-treatment reports filed by police increased for all groups except Asian– Pacific Islanders between 2002 and 2015. This increase was most pronounced for American Indian– Alaska Native families. Although the rate of police reporting of maltreatment remained relatively stable over the period, police have been responsible for initiating a growing proportion of cases involving native families.

Spatial and Temporal Heterogeneity

Policing varies dramatically across U.S. counties. Table 4 presents the coefficients of variation by race for between and within-county rates of police reporting of maltreatment and rates of arrest.7 Police reporting varies more between counties than it does within counties for all groups. The rates at which American Indian– Alaska Native and Asian– Pacific Islander families are reported show exceptionally high between-county variation.

Variation between counties in rates of arrest is also extreme. For all arrests, the standard deviation of county-average arrest rates is twice as large as the national average. Variation in arrest rates within counties is much lower than variation between counties. However, within-county variation in arrests is on the same order of magnitude as within-county variation in police reporting of maltreatment.

As shown in figure 1, national trends in police reporting of maltreatment over time are clear. Asian– Pacific Islander families experience particularly high degrees of fluctuation in reporting rates within counties over time. Although most of the variation in police reporting of maltreatment is between counties, within-county variation is substantial.

Regression Results

The results of multilevel regression models of police reporting of maltreatment show that policing is closely tied to the intensity of family surveillance at the county level. I present results for focal regression variables in table 5, a full table of regression parameter estimates in tables A1 and A2, and expected values for marginal changes in focal variables in figure 2. Both between-county average levels of arrest and annual within-county differences in arrests are significantly related to the number of police reports of child abuse and neglect for all categories of offenses and for nearly all racial groups. The expected change in police reporting of maltreatment for a marginal increase in within-county arrests is small, but the expected rate of police abuse and neglect reporting for a county with high county-average arrest rates is substantially greater than the expected rate for a county with national mean arrest levels.

County-average arrest rates are positively and significantly associated with rates of police reporting of maltreatment for all offense categories. The magnitude of this positive relationship is greatest for models including all arrests and smallest for models of quality-of-life arrests. In the model of total arrests, a county with average arrest rates at one standard deviation [End Page 58] above the national mean is expected to have 8 percent more police reports of suspected child abuse or neglect. Within-county changes in arrest rates are significantly and positively associated with police reporting of maltreatment for all offenses, drug offenses, and quality of-life offenses, but not for violent offenses. These associations have a relatively small magnitude; a one standard deviation increase in year-to-year total arrest rates is expected to correlate with an increase in police reporting of maltreatment rates of about 2 percent.

Figure 1. Police-Initiated Reporting Rate and Proportion of All Police-Initiated Investigated Reports Source: Author’s calculations based on NCANDS data, 2003–2015 ().
Click for larger view
View full resolution
Figure 1.

Police-Initiated Reporting Rate and Proportion of All Police-Initiated Investigated Reports

Source: Author’s calculations based on NCANDS data, 2003–2015 (Children’s Bureau 2016).

[End Page 59]

Table 4. Coefficients of Variation for Police-Initiated Reporting Rates and Arrest Rates
Click for larger view
View full resolution
Table 4.

Coefficients of Variation for Police-Initiated Reporting Rates and Arrest Rates

Table 5. Parameter Estimates and Significance of Focal Regression Predictors
Click for larger view
View full resolution
Table 5.

Parameter Estimates and Significance of Focal Regression Predictors

I next model rates of police reporting of mal-treatment by race of the child reported as a function of rates of arrest by race and category of offense. These models specify interactions of race, arrest, child poverty, and population composition. County-average arrest rates are positively and significantly associated with police reporting of maltreatment for all groups and for all offense categories. Within-county annual changes in arrest rates are positively and significantly associated with police reporting of maltreatment for all offenses and groups save Asian– Pacific Islander drug arrests, though again, the magnitude of this positive association is relatively small. I illustrate the expected rate of police reporting of maltreatment for marginal increases in both the county-average arrest rate and for marginal increases in the year-to-year within-county arrest rate by category of offense and race in figure 2. I also plot expected values for marginal increases in child poverty for comparison.

I expect police in a county with a cross-period average arrest rate of Asian– Pacific Islanders at one standard deviation above the national mean observed value to generate 36 percent more investigated reports of child abuse and neglect involving Asian– Pacific Islander children than a county with arrest rates at the national average. For African American [End Page 60]

Figure 2. Expected Changes in Police-Initiated Maltreatment Reporting for Marginal Changes in Focal Variables Source: Author’s calculations based on NCANDS data, 2009–2015 (), and UCR data, 2009–2015 ().
Click for larger view
View full resolution
Figure 2.

Expected Changes in Police-Initiated Maltreatment Reporting for Marginal Changes in Focal Variables

Source: Author’s calculations based on NCANDS data, 2009–2015 (Children’s Bureau 2016), and UCR data, 2009–2015 (Federal Bureau of Investigation 2014a).

[End Page 61]

children and families, police in a county with high average arrest rates are expected to produce 27 percent more reports of child abuse and neglect. For American Indian– Alaska Native children, counties with high average rates of native arrest are predicted to have 72 percent more police-investigated police reports of child maltreatment. For white children and families, counties with high average arrests are expected to have 32 percent more police maltreatment reports than those with average white arrest rates. Although within-county changes in arrest are significantly associated with police reporting of maltreatment rates, the magnitude of the relationship is relatively small. A within-county increase in arrests is predicted to increase reporting rates on average by about 3 percent for white children, 6 percent for Asian– Pacific Islander children, 4 percent for American Indian– Alaska Native children, and 2 percent for African American children.

Average arrest levels are incredibly strong predictors of the volume of police maltreatment reports involving American Indian– Alaska Native children and families. Although the magnitude of the relationship between average arrests and police reporting of maltreatment is relatively stable across offense types, for American Indian– Alaska Native children and families, drug arrests have an especially powerful association with police reporting. Counties with high average rates of American Indian– Alaska Native drug arrests are expected to have 82 percent more police reports of child maltreatment involving native children than counties at the national mean.

For white children, the magnitude of the estimated relationship between child poverty and police reporting is nearly identical to the magnitude of the estimated relationship between arrest rates and police maltreatment. A county with high white child poverty is expected to have about 35 percent more reports of child abuse and neglect filed by police than a county with average levels of white child poverty. However, for children and families of color, arrests are a far stronger predictor of the volume of maltreatment reports filed by police than child poverty is. For American Indian children, the expected rate of police reporting for a place with higher than average native child poverty is 13 percent higher than a county with average child poverty. For Asian– Pacific Islander children and families, high child poverty is not significantly associated with police reporting of maltreatment. For black children and families, counties with high black child poverty are only expected to have 8 percent more police mal-treatment reports than counties with average rates. The relationships between child poverty and family surveillance appear to be highly sensitive to race.

Although not illustrated in figure 2, racial composition also has a powerful relationship to police reporting of maltreatment (see table A2). Counties with larger than average white populations are expected to have 50 percent more police maltreatment reports involving white children than counties with average proportional white populations. By contrast, population composition is negatively associated with police reporting of maltreatment for children and families of color. Counties with a greater than average share of Asian– Pacific Islanders are expected to have 18 percent fewer maltreatment reports than an average county. For American Indian children, high population composition predicts a 16 percent lower rate of police reporting of maltreatment, and for African American children and families, 42 percent fewer reports than would be expected in an average county.

discussion

Police are central components of local regimes for family surveillance. Contact with police is a key vector through which children and families come under the scrutiny of child welfare systems. These results show that average levels of arrest are tightly associated with the rates at which police report child abuse and neglect across counties. The results further show that within-county changes in arrest rates are associated with small changes in rates of child abuse and neglect reporting by police. But further research—ideally with micro-level data—is needed to investigate a possible causal relationship between police contact with families and the reporting of child abuse and neglect. These results suggest that involvement with the child [End Page 62] welfare system is a spillover consequence of arrest, particularly for American Indian– Alaska Native children. Contact with police is a common precondition to a child welfare investigation, and opens the possibility of a child’s separation from their family through entry into the foster care system.

Race plays a powerful role in explaining the geography of family surveillance (Roberts 2008). For children and families of color, population composition and policing powerfully explain the intensity of family surveillance, whereas child poverty—typically considered a key correlate of child abuse and neglect (Sedlak et al. 2010)—is only weakly associated with the rate of police reporting of maltreatment for children of color. Although reporting rates are certainly associated with the actual incidence of child abuse and neglect, reports and investigations are organizational artifacts. Reporting is contingent on the observation of a family by a street-level bureaucrat, the cognitive classification of a child as a potential victim or a parent as unfit, the decision to file a report, and a child welfare agency formally classifying a report as credible and deserving of a response. Race, gender, and entrenched ideas about the family have central roles in structuring both the infrastructure of family surveillance and the micro-level interactions that lead to the decision to file a report.

The racial politics of policing and crime, driven by complex dynamics of threat, control, and predation (Soss and Weaver 2017; Smith and Holmes 2014; Carmichael and Kent 2014; Capers 2009), result in a distinctively punitive style of policing in many communities of color: a simultaneous overpolicing of perceived black and brown criminality and underpolicing in response to victims of color (Rios 2011; Perry 2009a; Beckett, Nyrop, and Pfingst 2006). Tight symbolic and legal associations among race, gender, ethnicity, criminality, and parental fitness inform both cognitive and institutional classification routines that lead officers, caseworkers, and agencies to view poor women of color as unfit parents who may pose a danger to their children (Roberts 2014; Haney 1996). These processes lead to a racialized spatial distribution of both the quantity of police and the qualitative character of their interactions with community members.

Surveillance infrastructure is likely related to both the production of excess child welfare cases and the underdetection of child maltreatment. Because surveillance is not equitably distributed, we should expect some communities to experience a high volume of false or needless reports, and others a high volume of false negatives (nonreports). The decoupling of offending and arrest has likely expanded contact between families and police over time (Weaver, Papachristos, and Zanger-Tishler 2019). It may also be a source of racial inequity in exposure to family surveillance, the generation of excess maltreatment reporting in communities of color, and underreporting in communities where police are less active or aggressive. Disparities originating in criminal justice may drift across organizational boundaries to reinforce the deep racial inequalities that are a defining feature of American child protection (Roberts 2002; Jacobs 2014).

Although it is reasonable to assume that the distribution of surveillance has a significant effect in generating the over-and underreporting of child abuse and neglect, no reliable data on actual maltreatment incidence across places currently exists. Using data on childhood injuries may provide some insight, but the overwhelming volume of maltreatment cases involve neglect, which is subject to tremendous discretion in identification and classification. Because many types of maltreatment are not cleanly demarcated, it is unlikely that the development of an objective surveillance procedure is possible. Street-level bureaucrats enforce maltreatment statutes by leveraging formal information (including stigmas such as criminal or arrest records) and informal biases in an always subjective process that classifies parents as abusive or neglectful and classifies children as victims or nonvictims.

The geography of policing has likely spill-over consequences on child protection beyond the direct reporting of child abuse and neglect by police. Legal cynicism resulting from direct and vicarious experiences of negative interactions with police may lead to a generalized cynicism that extends to other coercive state institutions, [End Page 63] such as the child welfare system (on police, Geller and Fagan 2019; on other institutions, Lerman and Weaver 2014; Fernandez-Kelly 2015). Legal cynicism may lead to a reduction in reporting of suspected child abuse and neglect by community members, suspicion of the motives of child welfare agencies and family courts, and avoidance of institutions that interface with either law enforcement or the child welfare system (Fong 2017).

Strengths and Limitations of NCANDS and UCR Data

Like the UCR, the NCANDS data offer the promise of comprehensive national data on a critically important social policy sector. The UCR is the sole longitudinal data on arrests across police agencies, but its limitations are substantial. Because reporting to the UCR is voluntary, the availability and quality of the data varies tremendously across places and within places over time. Many jurisdictions fail to report data, and others report data that are subject to various kinds of measurement error. Some researchers have challenged the validity of the UCR for subnational inferences of the sort presented in this analysis (Maltz and Targonski 2002), though others have suggested that nonreporting may have little impact on substantive conclusions (Lott and Whitley 2003), or have offered imputation and interpolation procedures to address nonreporting (Lynch and Jarvis 2008). Multiple imputation procedures that adjust for the longitudinal structure of the UCR offer an opportunity to quantify the extent to which missing data may affect inferences by introducing reasonable levels of missing data–induced measurement error into regression models (Honaker and King 2010).

The NCANDS data offer valuable insights into the activities of state and local child welfare organizations and afford the opportunity for the systematic comparison of child welfare systems across counties and states. However, as with all administrative data, the NCANDS has several distinct limitations that are a function of the organizational processes that generate the data. Most important, the data do not record cases that child welfare agencies screen out as not warranting an agency response. These screening processes are a function of varying statutes, policies, and routines that agencies use to determine when to respond to a case. Heterogeneity in screening affects the composition of cases that receive agency responses and hence are recorded in NCANDS. The implications of variation in case screening are difficult to estimate with current data but are likely small. Variance in the screening-in of police-initiated maltreatment reports across places is likely low because police tend to be seen as credible maltreatment reporters by child welfare agency staff. Future work could consider whether automatic screening policies, in effect in some jurisdictions later in the period, reveal shifts in the estimated relationships between policing and maltreatment reporting.

Because NCANDS submissions from state agencies to the federal government are all constructed from internal data systems, the quality of NCANDS variables can differ across jurisdictions. Some measures, such as report source, case substantiation, and child race, are recorded well across the data. Others, including service provision and child and parent risk factors, have a much lower quality across jurisdictions. These data quality issues can make individual-level analyses that take advantage of the multilevel structure of child welfare service provision challenging. However, those high-quality variables do offer researchers a unique opportunity to construct comparable indicators of child welfare agency activity across jurisdictions.

conclusion

American child protection systems are deeply multi-institutional. Lacking their own capacity to monitor children and families for signs of abuse and neglect, they depend on police, medical personnel, teachers, and other professionals and community members to leverage their routine interactions with children and families into a broad and diffuse network for maltreatment surveillance. This dependence turns practices and biases from external organizations into key features of the processes through which maltreatment reports are generated. Neither [End Page 64] the geographic distribution of police officers nor the qualitative character of police-public interactions are uniform. The social and spatial organization of policing plays a central role in selecting children and families for scrutiny by child protection agencies. Exposure to policing plays an important role in determining which children do, and which children do not, come to the attention of child protection agencies.

Frank Edwards

Frank Edwards is a postdoctoral associate at the Bronfenbrenner Center for Translational Research at Cornell University and incoming assistant professor of criminal justice at Rutgers University–Newark.

Direct correspondence to: Frank Edwards at fedwards@cornell.edu, Martha Van Rensselaer Hall, Room 1302D, Cornell University, Ithaca, NY 14850.
Table A1. Investigated Police Child Maltreatment Reports, Parameter Estimates and Standard Errors for Multilevel Poisson Regression
Click for larger view
View full resolution
Table A1.

Investigated Police Child Maltreatment Reports, Parameter Estimates and Standard Errors for Multilevel Poisson Regression

Table A2. Investigated Police Child Maltreatment Reports for Multilevel Poisson Regression
Click for larger view
View full resolution
Table A2.

Investigated Police Child Maltreatment Reports for Multilevel Poisson Regression

REFERENCES

Aleissa, Majid A., John D. Fluke, Bernard Gerbaka, Lutz Goldbeck, Jenny Gray, Nicole Hunter, Bernadette Madrid, Bert Van Puyenbroeck, Ian Richards, and Lil Tonmyr. 2009. “A Commentary on National Child Maltreatment Surveillance Systems: Examples of Progress.” Child Abuse & Neglect 33(11): 809–14.
Andersen, Signe Hald, and Christopher Wildeman. 2014. “The Effect of Paternal Incarceration on Children’s Risk of Foster Care Placement.” Social Forces 93(1): 269–98.
Asad, Asad L., and Matthew Clair. 2017. “Racialized Legal Status as a Social Determinant of Health.” Social Science & Medicine. Online pre-print: doi. org/10.1016/j.socscimed.2017.03.010.
Beckett, Katherine, Kris Nyrop, and Lori Pfingst. 2006. “Race, Drugs, and Policing: Understanding Disparities in Drug Delivery Arrests.” Criminology 44(1): 105–37.
Bell, Andrew, and Kelvyn Jones. 2015. “Explaining Fixed Effects: Random Effects Modeling of Time-Series Cross-Sectional and Panel Data.” Political Science Research and Methods 3(1): 133–53.
Berger, Lawrence M., Maria Cancian, Laura Cuesta, and Jennifer L. Noyes. 2016. “Families at the Intersection of the Criminal Justice and Child Protective Services Systems.” Annals of the American Academy of Political and Social Science 665(1): 171–94.
Brady, David. 2009. Rich Democracies, Poor People: How Politics Explain Poverty. New York: Oxford University Press.
Braman, Donald. 2007. Doing Time on the Outside: Incarceration and Family Life in Urban America. Ann Arbor: University of Michigan Press.
Brame, Robert W., Shawn D. Bushway, Ray Paternoster, and Michael G. Turner. 2014. “Demographic Patterns of Cumulative Arrest Prevalence by Ages 18 and 23.” Crime & Delinquency 60(3): 471–86.
Brame, Robert W., Michael G. Turner, Raymond Paternoster, and Shawn D. Bushway. 2012. “Cumulative Prevalence of Arrest From Ages 8 to 23 in a National Sample.” Pediatrics 129(1): 21–27.
Capers, I. Bennett. 2009. “Policing, Race, and Place.” Harvard Civil Rights-Civil Liberties Law Review 44(1): 43–78.
Carmichael, Jason T., and Stephanie L. Kent. 2014. “The Persistent Significance of Racial and Economic Inequality on the Size of Municipal Police Forces in the United States, 1980—2010.” Social Problems 61(2): 259–82.
Children’s Bureau. 2016. National Child Abuse and Neglect Data System (NCANDS) Child File, FFY 2015 [Dataset]. Washington: U.S. Department of Health and Human Services, Administration for Children and Families.
———. 2017. Child Maltreatment 2015. Washington: U.S. Department of Health and Human Services, Administration for Children and Families. Accessed July 26, 2018. https://www.acf.hhs.gov/sites/default/files/cb/cm2015.pdf.
Comfort, Megan. 2008. Doing Time Together: Love and Family in the Shadow of the Prison. Chicago: University of Chicago Press.
———. 2016. “‘A Twenty Hour a Day Job’: The Impact of Frequent Low-Level Criminal Justice Involvement on Family Life.” Annals of the American Academy of Political and Social Science 665(1): 63–79.
Cross, Theodore P., Emmeline Chuang, Jesse J. Helton, and Emily A. Lux. 2015. “Criminal Investigations in Child Protective Services Cases: An Empirical Analysis.” Child Maltreatment 20(2): 104–14.
Cross, Theodore P., David Finkelhor, and Richard Ormrod. 2005. “Police Involvement in Child Protective Services Investigations: Literature Review and Secondary Data Analysis.” Child Maltreatment 10(3): 224–44.
Drake, Brett, and Melissa Jonson-Reid. 2015. “Competing Values and Evidence: How Do We Evaluate Mandated Reporting and CPS Response?” In Mandatory Reporting Laws and the Identification of Severe Child Abuse and Neglect, edited by Ben Mathews and Donald C. Bross. New York: Springer.
Edwards, Frank. 2016. “Saving Children, Controlling Families: Punishment, Redistribution, and Child Protection.” American Sociological Review 81(3): 575–95.
Epp, Charles R., Steven Maynard-Moody, and Donald P. Haider-Markel. 2014. Pulled Over: How Police Stops Define Race and Citizenship. The Chicago Series in Law and Society. Chicago: University of Chicago Press.
Fagan, Jeffrey, and Garth Davies. 2000. “Street Stops and Broken Windows: Terry, Race, and Disorder in New York City.” Fordham Urban Law Journal 28(2): 457–504.
Federal Bureau of Investigation. 2014a. “Uniform Crime Reporting Program Data: Arrests by Age, Sex, and Race, 2012.” Ann Arbor, Mich.: ICPSR— Interuniversity Consortium for Political and Social Research.
———. 2014b. “Uniform Crime Reporting Program Data: Police Employee (LEOKA) Data, 2012.” Ann Arbor, Mich.: ICPSR—Interuniversity Consortium for Political and Social Research.
Fernandez-Kelly, María Patricia. 2015. The Hero’s Fight: African Americans in West Baltimore and the Shadow of the State. Princeton, N.J.: Princeton University Press.
Fong, Kelley. 2017. “Child Welfare Involvement and Contexts of Poverty: The Role of Parental Adversities, Social Networks, and Social Services.” Children and Youth Services Review 72 (January): 5–13. DOI: 10.1016/j.childyouth.2016.10.011
Foster, Holly, and John Hagan. 2015. “Punishment Regimes and the Multilevel Effects of Parental Incarceration: Intergenerational, Intersectional, and Interinstitutional Models of Social Inequality and Exclusion.” Annual Review of Sociology 41: 135–58.
Geller, Amanda, and Jeffrey Fagan. 2019. “Police Contact and the Legal Socialization of Urban Teens.” RSF: The Russell Sage Foundation Journal of the Social Sciences 5(1): 26–49. DOI: 10.7758/RSF.2019.5.1.02
Gelman, Andrew, Jeffrey Fagan, and Alex Kiss. 2007. “An Analysis of the New York City Police Department’s ‘Stop-and-Frisk’ Policy in the Context of Claims of Racial Bias.” Journal of the American Statistical Association 102(479): 813–23.
Gilbert, Neil. 2012. “A Comparative Study of Child Welfare Systems: Abstract Orientations and Concrete Results.” Children and Youth Services Review 34(3): 532–36.
Gilmore, Ruth. 2007. Golden Gulag: Prisons, Surplus, Crisis, and Opposition in Globalizing California. Berkeley: University of California Press.
Haney, Lynne A. 1996. “Homeboys, Babies, Men in Suits: The State and the Reproduction of Male Dominance.” American Sociological Review 61(5): 759–78.
———. 2000. “Feminist State Theory: Applications to Jurisprudence, Criminology, and the Welfare State.” Annual Review of Sociology 26: 641–66.
———. 2010. Offending Women: Power, Punishment, and the Regulation of Desire. Berkeley: University of California Press.
Harris, Alexes. 2016. A Pound of Flesh: Monetary Sanctions as Punishment for the Poor. American Sociological Association’s Rose Series in Sociology. New York: Russell Sage Foundation.
Harris, Alexes, Heather Evans, and Katherine Beckett. 2011. “Courtesy Stigma and Monetary Sanctions: Toward a Socio-Cultural Theory of Punishment.” American Sociological Review 76(2): 234–64.
Honaker, James, and Gary King. 2010. “What to Do About Missing Values in Time Series Cross-Section Data.” American Journal of Political Science 54(3): 561–81.
Honaker, James, Gary King, and Matthew Blackwell. 2011. “Amelia II: A Program for Missing Data.” Journal of Statistical Software 45(7): 1–47.
Jacobs, Margaret D. 2014. A Generation Removed: The Fostering and Adoption of Indigenous Children in the Postwar World. Lincoln: University of Nebraska Press.
Kim, Hyunil, Christopher Wildeman, Melissa Jonson-Reid, and Brett Drake. 2016. “Lifetime Prevalence of Investigating Child Maltreatment Among US Children.” American Journal of Public Health 107(2): 274–80.
Krase, Kathryn S., and Tobi A. DeLong-Hamilton. 2015. “Comparing Reports of Suspected Child Maltreatment in States with and Without Universal Mandated Reporting.” Children and Youth Services Review 50 (March): 96–100.
Lerman, Amy E., and Vesla M. Weaver. 2014. Arresting Citizenship: The Democratic Consequences of American Crime Control. Chicago Studies in American Politics. Chicago: University of Chicago Press.
Lipsky, Michael. 1980. Street-Level Bureaucracy: Dilemmas of the Individual in Public Services. New York: Russell Sage Foundation.
Lott, John R., and John Whitley. 2003. “Measurement Error in County-Level UCR Data.” Journal of Quantitative Criminology 19(2):185–98.
Lynch, James P., and John P. Jarvis. 2008. “Missing Data and Imputation in the Uniform Crime Reports and the Effects on National Estimates.” Journal of Contemporary Criminal Justice 24(1): 69–85.
Maltz, Michael D., and Joseph Targonski. 2002. “A Note on the Use of County-Level UCR Data.” Journal of Quantitative Criminology 18(3): 297–318.
Masters, N. Tatiana, Taryn P. Lindhorst, and Marcia K. Meyers. 2014. “Jezebel at the Welfare Office: How Racialized Stereotypes of Poor Women’s Reproductive Decisions and Relationships Shape Policy Implementation.” Journal of Poverty 18(2): 109–29.
Murakawa, Naomi, and Katherine Beckett. 2010. “The Penology of Racial Innocence: The Erasure of Racism in the Study and Practice of Punishment.” Law & Society Review 44(3–4): 695–730.
Nellis, Ashley. 2016. “The Color of Justice: Racial and Ethnic Disparity in State Prisons.” Washington, D.C.: Sentencing Project. Accessed September 1, 2018. https://www.sentencingproject.org/publications/color-of-justice-racial-and-ethnic-disparity-in-state-prisons.
Perry, Barbara. 2009a. Policing Race and Place in Indian Country: Over-and Underenforcement. Lanham, Md.: Lexington Books.
———. 2009b. “Impacts of Disparate Policing in Indian Country.” Policing and Society 19(3): 263–81.
Raz, Mical. 2017. “Unintended Consequences of Expanded Mandatory Reporting Laws.” Pediatrics 139(4) (April): e20163511.
Rios, Victor M. 2011. Punished: Policing the Lives of Black and Latino Boys. New York: New York University Press.
Roberts, Dorothy. 1997. Killing the Black Body: Race, Reproduction, and the Meaning of Liberty. New York: Pantheon Books.
———. 2002. Shattered Bonds: The Color of Child Welfare. New York: Basic Civitas Books.
———. 2008. “The Racial Geography of Child Welfare: Toward a New Research Paradigm.” Child Welfare 87(2): 125–50.
———. 2012. “Prison, Foster Care, and the Systemic Punishment of Black Mothers.” UCLA Law Review 59(6): 1474–500.
———. 2014. “Complicating the Triangle of Race, Class and State: The Insights of Black Feminists.” Ethnic and Racial Studies 37(10): 1776–82.
Ruggles, Steven J., J. Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B. Schroeder, and Matthew Sobek. 2010. Integrated Public Use Microdata Series: Version 5.0 [Machine-Readable Database]. Minneapolis: University of Minnesota.
Sedlak, Andrea, Jane Mettenburg, Monica Basena, Ian Peta, Karla McPherson, Angela Greene, and Spencer Li. 2010. “Fourth National Incidence Study of Child Abuse and Neglect (NIS-4).” Washington: U.S. Department of Health and Human Services.
Smith, Brad W., and Malcolm D. Holmes. 2014. “Police Use of Excessive Force in Minority Communities: A Test of the Minority Threat, Place, and Community Accountability Hypotheses.” Social Problems 61(1): 83–104.
Soss, Joe, and Vesla Weaver. 2017. “Learning from Ferguson: Policing, Race, and Class in American Politics.” Annual Review of Political Science 20: 565–91.
Steen, Sara, Rodney L. Engen, and Randy R. Gainey. 2005. “Images of Danger and Culpability: Racial Stereotyping, Case Processing, and Criminal Sentencing.” Criminology 43(2): 435–68.
Turney, Kristin. 2014. “The Consequences of Paternal Incarceration for Maternal Neglect and Harsh Parenting.” Social Forces 92(4): 1607–36.
Vesneski, William M. 2012. “Judicial Criteria for Terminating Parental Rights: A Content Analysis of 2010 North Carolina Appellate Court Opinions.” Ph.D. diss., University of Washington.
Wakefield, Sara, and Christopher Wildeman. 2014. Children of the Prison Boom: Mass Incarceration and the Future of American Inequality. New York: Oxford University Press.
Weaver, Vesla M., Andrew Papachristos, and Michael Zanger-Tishler. 2019. “The Great Decoupling: The Disconnection Between Criminal Offending and Experience of Arrest Across Two Cohorts.” RSF: The Russell Sage Foundation Journal of the Social Sciences 5(1): 89–123. DOI: 10.7758/RSF.2019.5.1.05
Wells, Rebecca, Monica Perez Jolles, Emmeline Chuang, Bowen McBeath, and Crystal Collins-Camargo. 2014. “Trends in Local Public Child Welfare Agencies 1999–2009.” Children and Youth Services Review 38 (March): 93–100.
Wildeman, Christopher. 2014. “Parental Incarceration, Child Homelessness, and the Invisible Consequences of Mass Imprisonment.” Annals of the American Academy of Political and Social Science 651(1): 74–96.
Wildeman, Christopher, Natalia Emanuel, John M. Leventhal, Emily Putnam-Hornstein, Jane Waldfogel, and Hedwig Lee. 2014. “The Prevalence of Confirmed Maltreatment Among U.S. Children, 2004 to 2011.” JAMA Pediatrics 168(8): 706–13.
Wildeman, Christopher, and Christopher Muller. 2012. “Mass Imprisonment and Inequality in Health and Family Life.” Annual Review of Law and Social Science 8: 11–30.
Wildeman, Christopher, and Emily A. Wang. 2017. “Mass Incarceration, Public Health, and Widening Inequality in the USA.” The Lancet 389 (10077): 1464–74.
Wulczyn, Fred, Robert Gibbons, Lonnie Snowden, and Bridgette Lery. 2013. “Poverty, Social Disadvantage, and the Black/White Placement Gap.” Children and Youth Services Review 35(1): 65–74.

Footnotes

1. The data used in this article were made available by the National Data Archive on Child Abuse and Neglect at Cornell University in Ithaca, New York. The data from the Substantiation of Child Abuse and Neglect Reports Project were originally collected by John Doris and John Eckenrode. Funding support for preparing the data for public distribution was provided by a contract (90-CA-1370) between the National Center on Child Abuse and Neglect and Cornell University. Neither the collector of the original data, funding agency, nor the National Data Archive on Child Abuse and Neglect bears any responsibility for the analyses or interpretations presented here.

2. These procedures do not affect the substantive conclusions presented. Parameter estimates are generally of the same direction, significance, and magnitude in models with imputed data and models that exclude these missing cases.

3. I classify arrest reports as missing if an agency reported more than fifty arrests in any included reporting year (by race) and reported zero arrests in an included reporting year. I use all observed years to produce a linear interpolation for these missing values.

4. The New York Police Department’s submissions to the UCR are known to be unreliable, and I identify several other agency-years that are extreme outliers and are likely errors in reporting.

5. Regression models that restrict the analysis to only cases in which a child is substantiated as a victim of abuse or neglect (not presented) yield substantively similar conclusions to models including all investigated cases.

6. All counts of police reports used for these analyses are conservative. NCANDS records only those cases that receive an agency response, and many reports in NCANDS have an unidentified report source. The counts reported here are lower bounds to the numbers of maltreatment reports originated from police. Descriptive counts and rates presented in text are the original values in the NCANDS data.

7. The coefficient of variation is calculated as the ratio of the standard deviation to the mean.

Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.

Additional Information

ISSN
2377-8261
Print ISSN
2377-8253
Pages
50-70
Launched on MUSE
2019-03-23
Open Access
Yes
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.