State Policies on Child Maltreatment and Racial Disproportionality

Abstract

Over the past several decades researchers have documented disproportionality for Black families across multiple decision-making points within the child welfare system. Yet, few studies have examined how specific state policies may impact disproportionality across decision points. The racial disproportionality index (RDI) was calculated for Black children in each state and Washington DC (N = 51) based on the proportion of children who were received a referral to CPS, a substantiated investigation, or entered foster care. A series of bivariate analyses (one-way ANOVAs; independent sample t-tests) were used to explore the relationship between the RDI and these decision points. Further analyses were conducted between the RDI and state policies (e.g., child maltreatment definitions, mandated reporting, and alternative response). Our results suggest there is an overrepresentation of Black children in CPS across the three decision points. This overrepresentation continues with specific state policies such as a state using harsh punishment in their definition of child maltreatment. Recommendations are provided for policy and research, including a suggestion for further exploration of state policies and county-level disproportionality indexes.

Over the past several decades, researchers have documented racial disproportionality and racial disparities across decision-making points within the child welfare system (AUTHORS, 2021; Dettlaff & Boyd, 2020; Harris & Hackett, 2008; Harris & Skyles, 2008). Racial disproportionality in child protective services (CPS) is defined as the difference in the proportion of children in a certain racial/ethnic group as compared to the proportion of children from that same racial group in the general child population; whereas racial disparity involves disparate outcomes for children of color compared with White youth. Racial disproportionality begins with rates of initial child maltreatment referrals; recent studies have found that over half of Black children will be referred to CPS for alleged maltreatment in their lifetime compared with one in four White children (Kim et al., 2017). Disparities continue with foster care entries (AUTHORS, 2021), service utilization (Osterling et al., 2012) and parental engagement once in foster care (Cheng & Lo, 2012), and permanency outcomes (i.e., reunification and adoptions; AUTHORS, 2021). Although a robust body of literature has explored disproportionality and disparity rates, fewer studies have examined how specific state policies or initiatives may worsen or reduce disproportionality across key decision points (e.g., referral to CPS, substantiation of CPS investigations, and entries into foster care).

Different hypotheses have emerged to explain racial disproportionality and disparities in child welfare. Some experts have argued that families of color, in particular Black families, face systemic barriers and discrimination that increase risk factors for child maltreatment and CPS entry, such as higher rates of single or early parenthood, diminished educational attainment, and criminal justice involvement (Kim & Drake, 2018; Maguire-Jack & Font, 2017; Maguire-Jack et al., 2020). There has also been evidence to suggest that neighborhood conditions of concentrated poverty and other forms of disadvantage increase the likelihood of contact with CPS (Freisthler et al., 2006; Fong, 2019; Dettlaff & Boyd, 2020). Black families are disproportionately exposed to these harsh neighborhood conditions due to racial segregation that is still prevalent in many neighborhoods (Reardon et al., 2015; Owens, 2017). In parallel, other researchers have examined the role of different dimensions of racism and bias among educators, healthcare workers, or police who are often mandated reporters (Dettlaff et al., 2011; Rivaux et al., 2008). For example, Hymel et al. (2018) found that reports of abusive head trauma among children of color, in particular Black and Hispanic children, were two times as likely to be reported to CPS by healthcare professionals than White families whose children presented similar injuries. Another recent study found a higher incidence for CPS referrals, substantiations, and foster care entries among Black children than their White peers (Edwards et al., 2021). These discrepancies may result from the intersection of different levels of racism, such as internalized or implicit bias, interpersonal, institutional, and structural discrimination, in which institutional policies, professional attitudes, and implicit bias may intertwine to perpetuate and maintain racism (Duerme et al., 2021). Given these different levels of racism, it is important to examine the impact of structural policies on racial disproportionality and disparities for children involved in CPS.

Definitions of Child Maltreatment

At the policy level, there is variation in how states define child maltreatment, which could contribute to subjectivity in the CPS decision making process from referrals to substantiations and entries into foster care. At the federal level, the most recently authorized Child Abuse and Prevention Act (CAPTA) of 2010 (P.L. 111–320) defines child maltreatment as “any recent act or failure to act on the part of a parent or caretaker which results in death, serious physical or emotional harm, sexual abuse or exploitation; or an act or failure to act, which presents an imminent risk of serious harm.” These minimum federal standards provide the basis through which states establish their own legal definitions of child maltreatment. All states recognize four main types of child maltreatment: physical abuse, sexual abuse, emotional/psychological abuse, and neglect (Child Information Gateway, 2019). However, states vary widely in how these types are defined with some states taking a broader approach (i.e., not specifying types of neglect such as medical, educational, etc.) while others are narrower and more specific (Coleman et al., 2010; Rebbe, 2018). For example, Rebbe (2018) found that 12 states specified prenatal substance exposure as a form of neglect; however, some national data such as the National Incidence Study (Sedlak et al., 2010) do not classify this as a form of neglect. Discrepancies were also highlighted in how states lacked consistent definitions to differentiate corporal punishment from physical abuse; Coleman and colleagues (2010) reported, “decision making about whether an injury or incident remains in the realm of family business or has crossed the line into the impermissible varies, reflecting a multiplicity of purely personal viewpoints, religious and political ideologies, and academic or disciplinary training and requirements” (p. 108). Therefore, variation in state policy may result in different acts or omissions (e.g., failures to act) being classified as maltreatment in some states, but as reasonable punishment or failing to meet the threshold for maltreatment in others.

In general, broader definitions of maltreatment may capture a wider potential of harm to children but can also lead to more false positives, whereas narrower definitions may lead to false negatives (Liu & Vaughn, 2019). Broad definitions can also increase the likelihood of implicit bias (Beniwal, 2017). Implicit biases are unconscious evaluations (positive or negative) of one group relative to another that are beyond our awareness. These biases are involuntarily activated through situational clues that exert influence on our perceptions and behaviors (Beniwal, 2017; Blair et al., 2011). Broader definitions rely more heavily on the discretion of the worker, and in turn implicit biases may be increased leading to racial disproportionality and disparities in various decision points along the child welfare continuum.

Many states include failure to provide education in their neglect definitions without providing specific details on how this should be determined. Although definitions of educational neglect may vary from state to state, it often focuses on the failure of a caregiver to enroll or send a school-age child to school or to provide homeschooling (Child Welfare Information Gateway, 2018). Although few studies in the U.S. have examined educational neglect at the national level, Van Wert and colleagues (2018) published their Canadian study that found that distinctions are not made between failure to comply with compulsory attendance laws and truancy, thus making it difficult to discern child protection issues and child behavioral challenges. Variation in policy related to absenteeism and truancy may also contribute to racial disparities, as one recent study found that disproportionality in unexcused absences explained disparities in truancy court referrals among a nationally representative school district sample (McNeely et al., 2021). Specific to educational neglect and child welfare, one Canadian study found that CPS referrals due to educational neglect were more likely to be substantiated than other types of neglect (Van Wert et al., 2018).

Another dimension of the definition of child maltreatment that varies by state is whether exposure to domestic violence is considered maltreatment. Roberts (2012) argued that the inclusion of exposure to domestic violence as a type of child maltreatment punished abused woman by criminalizing victims of domestic violence rather than providing services (i.e., affordable housing) that could allow the mother to leave the abuser. It is possible that inclusion of exposure to domestic violence as child maltreatment could also increase racial disproportionality, as Black women are more likely to experience physical violence from an intimate partner during their lifetime compared to other racial groups (DuMonthier et al., 2017). And because Black women also make up one of the largest populations of caregivers in the child welfare system (Putnam-Hornstein et al., 2013), it is possible these types of policies may disproportionately impact poor, Black mothers.

In parallel, there is state variation in whether corporal punishment or other harsh punishment is defined as maltreatment. For example, some states’ definitions for physical abuse include language prohibiting excessive corporal punishment or discipline and excluding reasonable corporal punishment (Coleman et al., 2010), or harsh punishment. Indeed, as of 2021, corporal punishment was lawful in all states, although in many, punishment is differentiated from abuse in that “the force used against the child is reasonable,” and does not count as abuse if the force is related to the “minor’s misconduct” (End Corporal Punishment, 2021). Evaluating what is excessive or reasonable discipline invites considerable subjectivity into decision making. Evaluations of what is reasonable or excessive may be clouded by personal or cultural values, religious or political views, and pressures to meet organizational demands (Coleman et al., 2010; Beniwal, 2017)— all of which may exacerbate bias and disproportionality.

State Mandated Reporter Policies: Who Is Required to Surveil and Report?

Often, the first individuals who must decide whether an act or failure to act constitutes reasonable suspicion of maltreatment are mandated reporters. Most states have mandated reporter laws that require professionals (i.e., education personnel, daycare providers, social workers, legal and law enforcement personnel, and medical personnel) who have regular contact with children to report suspected instances of child maltreatment. Nationally, in 2019, approximately 69% of child maltreatment reports were made by mandated reporters (USDHHS, 2020), with the remainder of reports coming from non-professional reporters, such as family members or neighbors. Yet, as with definitions of maltreatment, there is also variation in mandated reporting laws across states (Child Welfare Information Gateway, 2019).

Notably, some states (n = 16; 31.4%) have universal mandated reporting laws in which all adults are required to report suspected maltreatment. Findings have been inconsistent on the relationship between mandated reporting laws and child maltreatment referrals; some studies have found that universal mandated reporter laws increase rates of maltreatment reports (Palusci & Vandervort, 2014; Palusci et al., 2016), whereas other studies have no relationship between universal mandated reporting and reports or substantiations (Krase & DeLong-Hamilton, 2015). There are also variations in mandated reporting laws related to circumstances under which they must make a report, and whether their identity can remain anonymous. Some researchers have raised concern about class-based visibility bias (also known as exposure bias) that may exist in mandated reporting, in that low-income families may be more visible to mandated reporters due to more involvement with other social services, public schools, or after school centers whose employees are required by law to report suspected maltreatment (e.g., Ards et al., 2003; Slack et al., 2003). However, findings on mandated reporting and class-based visibility bias have been refuted in more recent literature (Drake et al., 2017; Kim et al., 2018). For example, Kim and colleagues (2018) used individual-level data from 4 states and national county-level data to determine if there was an increase in CPS reports from professionals as poverty rates increased. At both the individual and county-level, they found a consistent trend of a lower proportion of reports coming from professionals as poverty rates increased and this was true across racial/ethnic groups and maltreatment subtypes, which suggests that class-based visibility bias may not be a primary factor in CPS reporting disparities. These mixed findings highlight the need for more research to clarify the relationship between mandated reporting and disproportionality in child welfare referrals.

Policies That Impact Decision-Making in CPS after Referral

Once cases are referred, variation in state policy and procedure may also impact decision-making related to cases screened in for investigation, substantiations, and foster care entries. Thus, some policies such as mandated reporting might influence the number of initial referrals, while other policies such as screening and threshold for substantiation may impact subsequent substantiation and entries into foster care. Biases may occur at different points in the decision-making process, highlighting the importance of identifying how policies may impact disproportionality across the diverse decision points. One model that guides decision-making for cases referred to CPS is Structured Decision Making (SDM). SDM consists of “safety and risk assessments” that guide caseworker decision making related to risk in cases referred to CPS for alleged maltreatment (Nash, 2017). Some researchers have found that SDM has good predictive accuracy in detecting maltreatment risk (Coohey et al., 2012; Johnson et al., 2015), including among diverse racial groups (Coohey et al., 2012). As of 2019, 16 states (31.4%) had implemented SDM.

Once the initial level of risk is assessed, another CPS approach that offers a different pathway from the traditional investigative process is alternative response ). Alternative response is different from a traditional investigation or traditional response in that caseworkers provide a family-centered, preventative approach (Fluke et al., 2019; Merkel-Holguin et al., 2015) where engagement and targeted service referral are the focus rather than the collection of forensic evidence to support or refute an allegation of abuse or neglect (Hughes et al., 2013; Marshall et al., 2010). What is unique to alternative response is the focus on low-to moderate-risk families (i.e., severe physical and sexual abuse allegations are regularly excluded; Loman & Siegal, 2013) and the lack of formal entry into the system (i.e., there often is no required investigative finding; Fuller et al., 2013). The belief is by removing the coercion families are more likely to engage with CPS (Merkel-Holguin et al, 2015).

In cases where a formal investigation has been conducted, some states will offer a provision of in-home services to unsubstantiated cases to support families and reduce the risk of foster care entry (Chile Welfare Information Gateway, 2021). In states where in-home services are offered for unsubstantiated cases, families are given an opportunity to access services even when there is not enough evidence to substantiate the case. One reason for this is to prevent continued contact with CPS, as prior research has found similar patterns of recurrence among substantiated and unsubstantiated maltreatment cases (Holbrook & Hudziak, 2020). Similarly, some specific models of in-home service programs such as the Youth Villages Intercept ^® Program (Huhr & Wulczyn, 2022), intensive preservation efforts (Bezeczky et al., 2020), and Triple P (Prinz et al., 2009), have been found to reduce the likelihood of foster care entry. Although few empirical studies have examined the impact of in-home services or other programs on racial disproportionality, Jones Harden and colleagues (2021) argued that a more prevention-based focus in child welfare could help increase equity for Black families. They identify primary, secondary, and tertiary prevention programs that can reduce maltreatment, with a focus on programs that target Black families. Indeed, this study reported positive findings for in-home programs such as Nurse Family Partnership, Triple P Parenting Program, and Family Connects, among Black families.

While there have been several evaluations of family or child-level outcomes such as foster care entries, less is known about the structural impact of implementation of policies related to in-home services and subsequent disproportionality. As there is increased focus on reducing racial disproportionality in child welfare, it is important to consider institutional and structural policies that may decrease disproportionality across decision-making points.

The Current Study

This study explored the relationship between state policies related to child maltreatment definitions, mandated reporters, and system responses, and the state’s racial disproportionality index(RDI) across diverse decision points in child welfare. The specific research questions that guided this study were: 1) Do Black children experience disproportionality across child welfare referral and investigation decision points?; 2) What state policies are associated with the RDI for referrals, substantiated cases, and foster care entries of Black children in the United States?; and 3) What state policies are associated with changes in the RDI for Black children between referrals, substantiations, and foster care entries?

Method

This study used state-level data from two national datasets: the National Child Abuse and Neglect Data System (NCANDS) and the State Child Abuse & Neglect (SCAN) Policies Database. NCANDS is an annually released dataset that houses information on all children referred to CPS during the previous fiscal year in the United States. We used data from the FY 2019 NCANDS file (version 2) and merged it with the SCAN policies database. The SCAN policies database was developed by Mathematica and Child Trends in 2018–2019 to compile state data on definitions and policies used in their surveillance of child maltreatment. From the 2019 NCANDS child file, we merged the percentage of referrals, substantiated cases, and foster care entries by child race/ethnicity among states. We also merged data from the 2019 Annie E. Casey Kids Count data book to include the percentage of each race/ethnicity among the general child population of each state. We included all U.S. states and Washington DC, excluding Puerto Rico due to different conceptualizations of race/ethnicity and different policies related to child maltreatment. Thus, we analyzed N = 51 U.S. States and Washington DC.

Measures

Racial Disproportionality Index

The RDI was calculated for Black children in each state and Washington DC (N = 51) based on the proportion of children who were referred (referral RDI), who had substantiated cases (substantiation RDI), or who entered foster care (foster care entry RDI) and were identified as Black, divided by the proportion of children who were identified as Black among the child population per state. Consistent with prior literature, we used the RDI to gauge the overrepresentation of Black youth at different decision points in child welfare (Huggins-Hoyt et al., 2019; Shaw et al., 2008). State-level child population data were obtained through the Annie E. Casey Kids Count data book for 2019 (Annie E. Casey Foundation, 2019). We calculated the frequencies and percentages of children from each racial/ethnic group at each CPS data point (e.g., referrals, substantiations, foster care entries) by running descriptive statistics in the National Child Abuse and Neglect Database (NCANDS) for FY 2019. As an example, in FY 2019, there were 180,442 children who resided in the state of Alaska. Of these, 5,227 were Black (2.89%); in the same fiscal year, of 18,237 referrals to CPS in Alaska, 587 children referred were Black (3.21%). Therefore, the referral RDI for Black children in Alaska in FY 2019 was 3.21/2.89, or 1.11. An RDI of 1.0 reflects the same proportion of a given racial/ethnic group at any particular point of interest in the child welfare system as in the general child population. An RDI below 1.0 reflects a lower proportion of children from a specific racial/ethnic group at a point of interest (e.g., referrals) than in the general child population (e.g., underrepresentation), whereas an RDI greater than 1.0 reflects an overrepresentation of a particular racial/ethnic group. Consistent with prior literature (Morton et al., 2011) we used the RDI as it best aligned with our research question to observe the proportion of Black children and youth at each child welfare data point relative to the proportion of Black children in the population, and to observe changes across data points. Table 1 displays the number of Black children in each state’s child population and percentage of the overall child population, the number and percentage of Black children at each decision point (referral, substantiation, and foster care entry), and the RDI for referrals, substantiations, and foster care entries among states.

Table 1.

Number and Percentage of Referrals, Substantiations, and Foster Care Entries to Child Protective Services for Black Children by State (N=51)

State	N (%) of Child Population	N (%) Referrals		RDI Refe rrals		N (%) Subst.	RDI Subst.	N (%) Removals	RDI Removals
Alabama	315167 (28.94)	14183 (33.41)		1.15		3680 (29.69)	1.03	1202 (30.36)	1.05
Alaska	5227 (2.89)	587 (3.21)		1.11		71 (1.85)	0.64	36 (2.41)	0.83
Arizona	82195 (5.00)	10820 (11.90)		2.38		1989 (11.71)	2.34	1119 (12.87)	2.57
Arkansas	124938 (17.81)	12443 (18.44)		1.04		2710 (19.92)	1.12	582 (19.02)	1.07
California	444334 (5.00)	52729 (13.79)		2.76		10477 (13.77)	2.75	4561 (16.56)	3.31
Colorado	54174 (4.31)	5371 (10.11)		2.35		1380 (10.32)	2.39	445 (10.20)	2.37
Connecticut	83282 (11.45)	4721 (23.01)		2.01		2043 (23.15)	2.02	493 (24.71)	2.16
Delaware	51730 (25.32)	6289 (44.80)		1.77		944 (47.20)	1.86	153 (51.17)	2 02
DC	6 80 1 7 (53.15)	8415(84.87)		1.60		2047 (85.32)	1.61	292 (82.95)	1.56
Florida	851392 (20.10)	105841 (32.65)		1.62		12891 (30.72)	1.53	4510 (29.12)	1.45
Georgia	842024 (33.60)	84020 (46.66)		1.39		7472 (44.37)	1.32	2459 (38.03)	1.13
Hawaii	5246 (1.75)	78 (1.76)		1.01		15 (1.03)	0.59	17 (1.38)	0.79
Idaho	3736 (0.83)	172 (1.04)		1.25		25 (1.26)	1.52	21 (1.68)	2.02
Illinois	431439 (15.31)	64202 (32.90)		2.15		16957 (34.54)	2.26	2295 (34.55)	2 .26
Indiana	178375 (11.36)	34035 (16.89)		1.49		5325 (18.20)	1.60	1763 (19.22)	1.69
Iowa	40296 (5.53)	7757 (14.39)		2.60		2025 (14.15)	2.56	470 (13.20)	2.39
Kansas	43321 (6.17)	5001 (11.44)		1.85		337 (9.96)	1.61	491 (11.65)	1.89
Kentucky	92186 (9.17)	11301 (12.47)		1.36		2490 (10.82)	1.18	791 (12.47)	1.36
Louisiana	397922 (36.50)	14272(49.02)		1.34		4144 (47.79)	1.31	1174(43.85)	1.20
Maine	7532 (3.01)	444 (2.92)		0.97		82 (1.83)	0.61	33 (3.29)	1.09
Maryland	410238 (30.65)	14097(50.19)		1.64		1826 (45.68)	1.49	955 (56.74)	1.85
Massachusetts	120073 (8.87)	11505 (15.93)		1.80		4869 (14.96)	1.69	788 (15.12)	1.70
Michigan	343201 (16.00)	53225 (25.90)		1.62		5964 (21.81)	1.36	1565 (27.23)	1.70
Minnesota	134403 (10.31)	7996 (18.29)		1.77		1252 (17.28)	1.68	898 (15.20)	1.47
Mississippi	291386 (41.62)	20802 (44.95)		1.08		4472 (41.10)	0.99	848 (36.10)	0.87
Missouri	185187(13.47)	14976 (18.99)		1.41		733 (15.07)	1.12	1072 (16.72)	1.24
Montana	1475 (0.64)	209 (1.49)		2.33		52 (1.55)	2.42	25 (1.32)	2.06
Nebraska	28694 (6.02)	4277 (14.89)		2.47		747 (19.51)	3.24	355 (16.28)	2.70
Nevada	73391 (10.56)	9837 (30.14)		2.85		1840 (27.71)	2.62	1008 (31.30)	2.96
New Hampshire	5068 (1.98)	559 (4.02)		2.03		32 (2.58)	1.30	43 (5.60)	2.83
New Jersey	259376 (13.34)	28350 (31.05)		2.33		1879 (30.96)	2.32	1037 (38.95)	2.92
New Mexico	8465 (1.77)	1114 (3.56)		2.01		303 (3.28)	1.85	65 (4.68)	2.64
New York	594438 (14.74)	70312(26.55)		1.80		23913(28.64)	1.94	3183 (38.30)	2.60
North Carolina	517194(22.44)	42577 (37.79)		1.68		1758 (29.67)	1.32	1462 (26.79)	1.19
North Dakota	7761 (4.29)	715 (10.40)		2.42		155 (7.79)	1.82	81 (8.54)	1.99
Ohio	3 9 1 497 (15.16)	39781 (29.81)		1.97		5607 (27.07)	1.79	3796 (30.86)	2.04
Oklahoma	72744 (7.62)	6970 (10.53)		1.38		1662 (9.79)	1.28	410 (9.06)	1.19
Oregon	20045 (2.31)	2844 (5.26)		2.28		622 (4.64)	2.01	98 (3.21)	1.39
Pennsylvania	340981 (12.93)	8680 (21.19)		1.64		1055 (21.48)	1.66	2972 (31.58)	2.44
Rhode Island	14996 (7.35)	1321 (13.95)		1.90		401 (12.91)	1.76	165 (14.98)	2.04
South Carolina	326283 (29.29)	41366 (40.65)		1.39		9969 (43.17)	1.47	1423 (37.88)	1.29
South Dakota	6560 (3.01)	156 (3.49)		1.16		64 (3.71)	1.23	56 (4.62)	1.53
Tennessee	285751 (18.91)	21703 (25.33)		1.34		1296 (20.37)	1.08	1540 (23.51)	1.24
Texas	890541 (12.02)	61257 (20.28)		1.69		15380 (20.47)	1.70	3625 (19.90)	1.66
Utah	1 1 1 1 9 (1.19)	1041 (3.36)		2.82		377 (3.14)	2.64	73 (3.63)	3.05
Vermont	2041 (1.78)	68 (1.37)		0.77		6 (0.66)	0.37	27 (3.77)	2.12
Virginia	370953 (19.85)	13713 (26.01)		1.31		1698 (26.04)	1.31	759 (27.46)	1.3 8
Washington	70930 (4.27)	4510 (7.76)		1.82		358 (7.04)	1.65	443 (8.34)	1.95
West Virginia	12759 (3.53)	2402 (4.22)		1.20		358 (3.86)	1.09	201 (3.98)	1.13
Wisconsin	112760 (8.89)	11526 (28.36)		3.19		1048 (20.36)	2.29	1191 (26.33)	2.96
Wyoming	1214 (0.90)	151 (2.61)		2.90		36 (3.27)	3.63	51 (5.05)	5.61
AVERAGE				1.78			1.68		1.92

Changes in RDI

To assess changes in RDI across data points, we computed a variable to measure the difference in RDI between: 1) substantiation and referrals; 2) foster care entries and substantiations; and 3) foster care entries and referrals. For each of these variables, we subtracted the RDI at the later data point from the initial referral RDI. Therefore, 0 reflected no change in the RDI of a particular state, whereas negative numbers reflected a reduction in the RDI over time, and positive numbers reflected an increase in the RDI.

Definitions of Child Maltreatment

We included three items that reflected specific child maltreatment definitions that were not consistently present in state policy: 1) harsh punishment/corporal punishment; 2) educational neglect; and 3) exposure to domestic violence. Each of these items was recorded in the SCAN Policy Database as a dichotomous variable. Items were worded such as: “the definition of child maltreatment includes exposure to domestic violence, intimate violence, or household violence”. For each item, states were coded as 1 if the state definition of child maltreatment included each dimension of maltreatment and were coded as 0 if state policy did not include each type of maltreatment.

Universal Mandated Reporting

Universal mandated reporting was a dichotomous variable in the SCAN policy database that reflected whether the state had a requirement for all adults to report suspected child maltreatment. States that had universal mandated reporting laws were coded as 1; those that only listed specific professions as mandated reporters were coded as 0.

Structured Decision Making

Structured Decision Making was measured with one item in the SCAN policy data base that asked whether each “state offers a structured decision making approach.” States could be coded as 0 (no), 1 (yes—implemented statewide), or 2 (yes—implemented in specific counties or regions). We dichotomized the variable to compare states with no structured decision making ^® (0) to states with any (1) SDM.

Alternative Response

Alternative response was measured with one item in the SCAN policy database that asked whether each “state has alternative or differential response.” States could be coded as 0 (no), 1 (yes—implemented statewide), or 2 (yes—implemented in specific counties or regions). We dichotomized the variable to compare states with no alternative response (0) to states with any alternative response (1).

In-Home Services for Unsubstantiated Cases

State utilization of in-home services for unsubstantiated cases was measured with one item in the SCAN policy database that asked whether each state “has in-home services for unsubstantiated cases” of child maltreatment. States were coded as 0 (no), 1 (yes), or 2 (unknown).

Data Analysis

We first conducted descriptive statistics to examine frequencies, proportions, means, and standard deviations across the entire sample to better understand the RDI for Black children across the country. There were no missing data in the sample. Some variables of interest (e.g., in-home services) had an “unknown” category, which we retained in the bivariate analyses. We subsequently conducted a series of bivariate tests (one-way ANOVAs and independent sample t-tests) to examine differences in the average RDI at each decision point, including referrals, substantiations, and entries into foster care for Black children among states based on statewide enactment of policies related to child maltreatment reporting, definitions, and service provision. For the final analyses, we ran a series of one-way ANOVAs and independent sample t-tests with each policy and the changes in RDI between each data point. All analyses were conducted in the Statistical Package for Social Sciences (SPSS), version 28.

Results

Research Question 1: Disproportionality across Decision Points

Figure 1 displays the average referral RDI for Black children by state. As seen in the map, there were some similarities by geographic region.

Of the 51 states plus Washington DC, only two had an average referral RDI for Black children that was less than 1.00, while five had a referral RDI between 1.00 and 1.15. Therefore, in 45 states, Black children experienced disproportionality and overrepresentation in referrals to CPS.

Table 1 displays the average (RDI) and policies by state. Across the entire sample, the average RDI for Black children was 1.78 for referrals, 1.68 for substantiated cases, and 1.92 for foster care entries. Overall, only two states had an RDI of 1.00 or less for referrals, five states had an RDI of 1.00 or less for substantiated cases, and three states had an RDI of 1.00 or less for foster care entries.

Table 2 displays the policies by state child welfare system. Almost all states (n = 48; 94.1%) included harsh punishment in their definitions of child maltreatment. Fewer (n = 37; 72.5%) included educational neglect in their definition of child maltreatment. The majority (n = 41; 80.3%) included exposure to domestic violence in the state definition of maltreatment. Only 16 states (31.3%) had universal mandated reporting. Less than one-third (n = 16; 31.3%) used SDM for screening. Approximately half of states had implemented a statewide alternative response (n = 25; 48.1%), while n = 5 states implemented alternative response in some counties or regions (9.6%), and 21 states had no alternative response (41.1%). Most states provided in-home services for unsubstantiated cases of child maltreatment, either statewide (n = 37; 71.2%) or in specific regions or counties (n = 1; 1.9%); eight states did not provide in-home services for unsubstantiated cases (15.4%), and there was not information for the remaining five (9.8%).

Table 2.

Child Maltreatment Policies by State (N=51)

State	Def-harsh punish	Def-Ed. Neglect		Def-exp. IPV			Structured DM				Alternative Response	Universal mand. Report.		In-home services
Alabama	Yes	No		Yes			No				Yes	No		Unknown
Alaska	Yes	Yes		Yes			No				No	No		Yes
Arizona	Yes	No		Yes			No				No	No		Yes
Arkansas	Yes	Yes		No			No				Yes	No		No
California	Yes	Yes		Yes			Unknown				Yes	No		Yes
Colorado	Yes	Yes		Yes			Unknown				Yes	No		Yes
Connecticut	Yes	Yes		Yes			Yes				Yes	No		Yes
Delaware	Yes	Yes		Yes			Unknown				Yes	Yes		Yes
DC	Yes	Yes		Yes			Yes				No	No		No
Florida	Yes	No		Yes			Unknown				No	Yes		Yes
Georgia	Yes	Yes		Yes			No				Yes	No		No
Hawaii	No	No		No			No				Yes	No		No
Idaho	Yes	Yes		Yes		No					No		No	Yes
Illinois	Yes	Yes		Yes		No					No		No	Yes
Indiana	Yes	Yes		Yes		Yes					No		Yes	Yes
Iowa	Yes	No		Yes		Unknown					Yes		No	Yes
Kansas	Yes	Yes		Yes		Yes					No		No	Yes
Kentucky	Yes	Yes		Yes		No					No		Yes	Yes
Louisiana	Yes	No		Yes		Yes					No		No	Yes
Maine	Yes	Yes		Yes		Yes					Yes		No	Yes
Maryland	No	No		Yes		Unknown					Yes		No	Yes
Massachusetts	Yes	Yes		No		Unknown					No		No	No
Michigan	Yes	No		Yes		Unknown					No		No	No
Minnesota	Yes	Yes		Yes		Unknown					Yes		No	Unknown
Mississippi	Yes	Yes		No		No					No		Yes	Yes
Missouri	Yes	Yes		No		Yes					Yes		No	Yes
Montana	Yes	Yes		Yes		Unknown					No		No	Yes
Nebraska	Yes	Yes		Yes		Yes					Yes		Yes	Yes
Nevada	Yes	Yes		No		No					Yes		No	Unknown
New Hampshire	Yes	Yes		Yes		Yes					Yes		Yes	Yes
New Jersey	Yes	Yes		Yes		Yes					No		Yes	Yes
New Mexico	Yes	Yes		Yes		No					No		Yes	Unknown
New York	Yes	Yes		Yes		Unknown					Yes		No	Yes
North Carolina	Yes	Yes		Yes		Yes					Yes		Yes	No
North Dakota	Yes	Yes		Yes		Unknown					Yes		No	Yes
Ohio	Yes	Yes		Yes		No					Yes		No	Unknown
Oklahoma	Yes	Yes		Yes		Yes					Yes		Yes	Yes
Oregon	Yes	No		No		No					No		No	No
Pennsylvania	Yes	No		Yes		No					No		No	Yes
Rhode Island	Yes	Yes		Yes		Yes					Yes		Yes	Yes
South Carolina	Yes	Yes		Yes		No					Yes		No	Yes
South Dakota	Yes	Yes		Yes		No					No		No	Yes
Tennessee	Yes	Yes		Yes		Yes					Yes		Yes	Unknown
Texas	Yes	No		No		Unknown					Yes		Yes	Yes
Utah	Yes	Yes		Yes		Unknown					No		Yes	Yes
Vermont	Yes	Yes		Yes		No					Yes		No	Yes
Virginia	Yes	No		Yes					Yes		Yes		No	Yes
Washington	Yes	No		No		Yes					Yes		No	Yes
West Virginia	Yes	Yes		Yes		No					No		No	Yes
Wisconsin	Yes	No		Yes		No					Yes		No	Yes
Wyoming	Yes	Yes		No		No					Yes		Yes	Yes

Research Question 2: Relationship between State Policies and RDI

Table 3 displays the average RDI for Black children across three decision points during the referral and investigative process: referrals, substantiations, and entries.

Table 3.

Average Referral, Substantiation, and Entry Disproportionality Indexes by State Policy (N = 51)

	RDI Referrals M(SD)	F	RDI Subst. M(SD)	F	RDI Entry M(SD)	F
Mandated reporting required		0.60		0.20		0.31
None	1.83 (0.57)		1.72 (0.72)		1.98 (0.93)
Some mandated reporters	1.75 (0.60)		1.66 (0.56)		1.89 (0.78)
All mandated reporters	1.83 (0.67)		1.65 (0.80)		1.80 (0.55)
Unknown	1.27 (0.16)		1.35 (0.17)		1.41 (0.17)
Universal mandated reporting		0.68		0.97		1.34
Yes	1.86 (0.53)		1.83 (0.76)		2.15 (1.16)
No	1.75 (0.60)		1.61 (0.60)		1.81 (0.64)
Definition—harsh punishment		1.48		1.46		1.14
Yes	1.80 (0.57)		1.71 (0.65)		1.94 (0.84)
No	1.32 (0.44)		1.03 (0.63)		1.31 (0.75)
Definition—educ. neglect		-0.13		0.41		0.92
Yes	1.78 (0.57)		1.70 (0.70)		1.97 (0.91)
No	1.80 (0.60)		1.63 (0.53)		1.77 (0.63)
Definition—domestic violence		0.01		-0.10		-0.01
Yes	1.78 (0.55)		1.67 (0.60)		1.92 (0.64)
No	1.78 (0.70)		1.71 (0.88)		1.92 (1.44)
Structured decision-making		2.44		2.10		0.63
Yes	1.68 (0.40)		1.57 (0.59)		1.76 (0.62)
No	1.67 (0.70)		1.57 (0.77)		1.90 (1.10)
Unknown	2.07 (0.45)		1.98 (0.46)		2.11 (0.56)
Alternative response		6.62**		3.72*		3.64*
Yes-statewide	1.67 (0.53)		1.56 (0.73)		1.82 (0.96)
Yes-partial county/region	2.58 (0.53)		2.40 (0.31)		2.84 (0.36)
No	1.73 (0.49)		1.66 (0.52)		1.82 (0.63)
In-home services unsub.		0.55		0.73		2.12
Yes-statewide	1.82 (0.60)		1.75 (0.70)		2.05 (0.88)
Yes-partial county/region	2.01 (n/a)		1.85 (n/a)		2.64 (n/a)
No	1.55 (0.41)		1.37 (0.42)		1.31 (0.32)
Unknown	1.81 (0.66)		1.63 (0.64)		1.75 (0.77)

Note.

^*p<.05

^**p<.01

^***p<.001.

RDI subst = RDI at substantiation of maltreatment. Educ. Neglect = policy that includes educational neglect within the definition of child maltreatment.

There were slight differences in the RDI for referrals by mandated reporting requirements and definitions of maltreatment; however, none of these differences reached statistical significance. States that had universal mandated reporting requirements had a slightly higher RDI for referrals (M=1.86, SD=0.53), substantiations (M=1.83, SD=0.76), and entries (M=2.15, SD=1.16) than states without universal mandated reporting. States that had a definition of child maltreatment that included harsh punishment had an average referral RDI of M=1.80 (SD=0.57), while those without had an average referral RDI of M=1.32 (SD=0.44); this trend continued with the average substantiation RDI (harsh punishment M=1.71, SD=0.65 compared to no harsh punishment M=1.03, SD=0.63). States that included harsh punishment in their definition of child maltreatment had an average entry RDI of M=1.94 (SD=0.84), compared to M=1.31 (SD=0.75) among states that did not consider harsh punishment as maltreatment. Although these differences were not statistically significant, they are worth noting given that there were only 51 units of analysis. There was little variation in RDI for referrals between states based on inclusion of educational neglect or exposure to domestic violence in the definitions of child maltreatment.

Once cases were referred, there was little variation between states that implemented SDM compared with those that did not; however, states that did not have clear policies outlined about SDM had a slightly higher average referral RDI (M=2.07, SD=0.45), substantiation RDI (M=1.98, SD=0.46) and entry RDI (M=2.11, SD=0.56) than states that clearly did or did not implement it. There were significant differences in the average referral, substantiation, and entry RDI based on implementation of alternative response. States that implemented alternative response statewide had the lowest average referral RDI (M=1.67, SD=0.53), substantiation RDI (M=1.56, SD=0.73), and were tied for the lowest entry RDI (M=1.82, SD=0.96) compared to states that did not implement any alternative response or states that only partially implemented it in some counties or regions. Finally, states that did not offer in-home services for unsubstantiated maltreatment cases had the slightly lower average RDI across points (M=1.55 at referral; M=1.37 at substantiation; and M=1.31 at entry), than states that had partial implementation of in-home services in some counties or regions across decision points (M=2.01 at referral; M=1.85 at substantiation; and M=2.64 at entry), although these differences did not reach statistical significance.

Table 4 displays the average changes in RDI across time points for states based on policies that may influence the decision to substantiate or remove children. As seen in Table 4, none of the changes in RDI across decision points reached statistical significance. Across policies, we observed a slight decrease in the RDI between referrals and substantiation. However, the downward trend did not consistently extend to entries into foster care. For example, all states experienced a reduction in RDI between referral and substantiation, regardless of whether their definition of child maltreatment included harsh punishment, educational neglect, or exposure to domestic violence. Yet, the average RDI for entries increased in states that included harsh punishment, educational neglect, or domestic violence in their definitions of maltreatment; in contrast, the average RDI continued to decrease at entry in states that did not include harsh punishment or educational neglect in their definitions of maltreatment.

Table 4.

Changes in RDI across Decision-Points by State Policy (N = 51)

	Change in RDI Ref-Sub M(SD)	F/t	Change in RDI Sub-Entry M(SD)	F/t	Change in RDI Ref-Entry M(SD)	F/t
Definition—harsh punishment		1.33		−0.43		0.36
Yes	-0.09 (0.28)		0.23 (0.49)		0.13 (0.54)
No	−0.28 (0.19)		0.28 (0.11)		−0.01 (0.30)
Definition—educ. neglect		1 .25		1.01		1.40
Yes	−0.07 (0.29)		0.27 (0.53)		0.19 (0.57)
No	−0.17 (0.24)		0.13 (0.35)		−0.03 (0.36)
Definition—domestic violence		−0.34		−0.01		0.16
Yes	−0.10 (0.27)		0.24 (0.44)		0.13 (0.39)
No	−0.07 (0.32)		0.21 (0.67)		0.13 (0.94)
Structured decision-making		0. 04		0 . 83		0.58
Yes	−0.11 (0.31)		0.19 (0.44)		0.08 (0.30)
No	−0.10 (0.31)		0.33 (0.60)		0.22 (0.74)
Unknown	−0.08 (0.19)		0.12 (0.30)		0.04 (0.34)
Alternative response		0.34		0.67		0.19
Yes-statewide	−0.10 (0.33)		0.25 (0.60)		0.14 (0.65)
Yes-partial county/region	−0.18 (0.42)		0.43 (0.29)		0.25 (0.41)
No	−0.07 (0.16)		0.16 (0.34)		0.08 (0.39)
In-home services unsub.		0.41		1.79		2.31
Yes-statewide	−0.07 (0.31)		0.30 (0.52)		0.22 (0.56)
Yes-partial county/region	−0.15 (n/a)		0.79 (n/a)		0.63 (n/a)
No	−0.17 (0.17)		−0.05 (0.28)		−0.23(0.32)
Unknown	−0.17 (0.06)		0.11(0.21)		−0.06(0.16)

Note.

^*p<.05

^**p<.01

^***p<.001.

Change in RDI Ref-Sub = the change in RDI between referral and substantiation. Change in RDI Sub-Entry = change in RDI between the substantiation and foster care entry. Change in RDI Ref-Entry = change in RDI between referral and foster care enry.

Similar trends were observed with SDM and alternative response; there was a decrease in the RDI between referral and substantiation for all states. However, states that implemented SDM and/or alternative response experienced similar increases in RDI by entry when compared with states that did not implement these initiatives. We observed a decrease in RDI between referral and substantiation regardless of implementation of in-home services; however, the average RDI increased between referral and entry for states with statewide or partial implementation of in-home services, whereas the average RDI decreased in states that had no in-home services for unsubstantiated cases or in which there was no clear policy on in-home service provision.

Discussion

This study explored the relationship between various state child maltreatment policies and racial disproportionality in referrals, substantiations, and foster care entries. Findings from our first research question confirmed an overrepresentation of Black children in CPS across the U.S., from initial referrals to substantiations and foster care entries. Despite decades of research on racial disproportionality in child welfare, Black children and youth continue to face overrepresentation across the data points (Dettlaff & Boyd, 2020; Harris & Hackett, 2008; Huggins-Hoyt et al., 2019).

The analyses from our second research question highlighted some potential variations in RDI by state policy. Notably, the only policy that was significantly associated with RDI was alternative response; however, given the relatively low statistical power, there were several trends and patterns observed that are worth noting despite the lack of statistical significance. For example, states that included harsh punishment in their definition of child maltreatment had a slightly higher RDI than states that did not. This may reflect the subjectivity in determining the “reasonableness” of corporal punishment or other harsh punishment that has been problematized in prior research (Coleman et al., 2010). Building upon prior literature, our findings may highlight the oversurveillance and punishment of Black families when it comes to expected parenting discipline and norms, in which policies may favor White, middle class parenting norms (Mack, 2021). There was little variation in referral and substantiation RDI among states that included educational neglect in their definition of child maltreatment and those that did not; however, states that did include educational neglect had a slightly higher foster care entry RDI. The variation in state policies may give way to more subjectivity in the decision-making process, which in turn may increase the potential for implicit bias in decision-making (Dettlaff et al., 2020). The variation in RDI by harsh punishment could also reflect how Black families may be subjected to biases related to parenting styles (Harris & Hackett, 2008; Roberts, 2002). Therefore, it may be necessary to question norms rooted in Western colonialism and white supremacy (McCoy, 2020). Notably, we did not observe variation among states based on inclusion of exposure to domestic violence in their definitions of maltreatment. One study in Canada found that the involvement of child welfare in domestic violence incidents did not increase foster care entries (Black et al., 2008); however, to date, few studies have examined the relationship between maltreatment definitions including domestic violence exposure and racial disproportionality. Therefore, this is an area that might warrant further study.

States that had universal mandated reporting requirements had slightly higher RDI across decision points compared to states without, although this relationship also lacked statistical significance. Prior research has highlighted biases among professional and non-professional referrals to CPS (DeZelar & Lightfoot, 2020). Thus, it could be beneficial to integrate trainings for mandated reporters to reduce bias and have a better understanding of identification of maltreatment (Baker et al., 2021; Castro & Perry, 2020). In parallel, child welfare strategies post-referral to better support families did not uniformly reduce disproportionality. States that implemented SDM had slightly lower RDI for foster care entries than states without; however, there was no variation in the referral or substantiation RDI. This may reflect the stage at which many child welfare agencies incorporate SDM; for example, in addition to initial risk assessments, SDM can also be used for placement decisions (Coohey et al., 2012; Wells & Correia, 2012). Therefore, it is possible that implementation of SDM may help reduce disproportionality in foster care entries.

With regard to our third research question, we found that the average RDI was slightly higher at foster care entry than at referral or substantiation, although this was not statistically significant. This finding aligns with some prior literature that has found Black children referred to CPS to have higher odds of foster care entry than White children who are referred (Cénat et al., 2021; Knott & Donovan, 2010). This is also troubling, as it could signal an increase in racial disproportionality as children and families spend more time in the child welfare system. Indeed, states that did not provide in-home services for unsubstantiated cases saw a decrease in RDI between referral and entry—whereas states that provided in-home services for unsubstantiated cases in some counties or statewide had an increase in RDI between referral and entry. On the one hand, it is possible that this could reflect in-home services that serve as a means of perpetuating surveillance of Black families even when maltreatment has not been substantiated (Mack, 2021). On the other hand, it is also possible that this finding reflects the disconnect between services that can support family preservation and services that are offered to families referred to CPS (Sykes, 2011). Therefore, it is possible that more strengths-based services are needed to provide support and move away from a punitive, deficit-focus (Cénat et al., 2021; Miller et al., 2013; Roberts, 2007).

Limitations

The first limitation of this study is that it was cross-sectional so only assessed RDI among states at a specific point in time. In addition, as this study used state-level policies to assess state-level RDI, we analyzed 50 states plus Washington DC; therefore, it is possible that our study may have been underpowered to detect small or moderate effect sizes. We also excluded Puerto Rico due to differences in conceptualization of race/ethnicity and in policies; therefore, future research could use Puerto Rico as a case study to explore child welfare and race/ethnicity. The third limitation of our study is that it is likely the implementation of some policies (e.g., alternative response or inclusion of educational neglect in state definition of maltreatment) may have followed similar patterns of utilization. Prior comparative research has found child welfare systems to have different foci—such as family-focused or child-focused—with variation on the role of the state in provision of services (Muñoz-Guzmán et al., 2015). For example, it is possible that some states take a more paternalistic approach, whereas others do not have as robust a role in service provision (Gupta, 2018). Therefore, it is possible that policies related to required training for mandated reporters, in-home service provision, and alternative response may reflect states that take a more family-focused approach, which in turn could lead to certain states tending to have similar combinations of policies based on their orientation toward family preservation/support and child protection. Similarly, there could be confounding factors that impact both RDI and state child maltreatment policies; as noted previously in the limitation section, certain states or groups of states may have similar policies (e.g., access to services such as health care, mental health care) that could potentially impact child maltreatment reports, substantiations, and entries into care. Finally, RDI was based on child race/ethnicity as reported by the caseworker or other professional responsible for entering state-level NCANDS data. This can lead to discrepancies between the race/ethnicity ascribed to the child by the caseworker and the child’s accurate racial identity.

Implications for Future Research

In light of our findings and limitations, there are several implications for future research. First, future research could use multi-level modeling to explore the relationship between state policies on individual child outcomes by merging the SCAN policy database with national child-level administrative data. This could allow researchers to examine the likelihood of substantiation or foster care entry based on state policy and include race/ethnicity as a moderator. Second, given the variation in implementation among counties in some states, future research could explore policies and county-level disproportionality indexes. Future research could also examine RDI over time and in function of implementation of new county- or statewide policies such as universal mandated reporting, changing definitions of child maltreatment, and practices such as SDM or in-home service provision. This could help continue to build evidence for practices and policies that may reduce racial disproportionality. Finally, future research could examine policies and RDI among other racialized groups, such as Indigenous children and Latino children, who also face structural barriers and discrimination.

Conclusion

Findings from this study highlight the variation in state policies related to identifying and responding to child maltreatment. Although there are some federally-set policies through CAPTA, states have considerable autonomy to enact policies related to definitions of maltreatment, mandated reporting, and decision-making once cases are referred. In addition to false negatives or positives that prior researchers have examined among broad and narrow definitions of child maltreatment (Liu & Vaughn, 2019), findings from our study suggest that policy variation may also impact racial disproportionality. Therefore, as policies are created and enacted to promote child safety, wellbeing, and permanency, it is also important to evaluate how they may perpetuate racial biases. Furthermore, our finding that disproportionality begins at referral and continues throughout the decision points in child welfare cases highlights a need for more preventive efforts to reduce the disproportionate number of referrals of Black children to CPS. These findings align with recent calls to increase services available at the front end of child welfare cases, and primary prevention efforts, such as community education and mobilization, home visiting, and early childhood programs (Jones Harden et al., 2020; Slack & Berger, 2020).

Highlights.

• Black children face disproportionality in referrals, substantiations, and foster care entries

• The racial disproportionality index is slightly higher in states with policies such as including harsh punishment as maltreatment

• State child welfare policies may contribute to disproportionality, highlighting a need for further policy analysis