Striking a Better Balance between Child Safety and Parental Rights

By J. Michael Tower

News Type
Advocates' Forum


Recent court cases have questioned whether the use of safety plans by the Illinois Department of Children and Family Services (DCFS) violates parents’ rights to custody of their children.  However, little attention has been paid to a key determinant that initiates safety planning – risk assessment.  DCFS utilizes a model of risk assessment known to have serious accuracy flaws, which may lead to inappropriate custody interference by the state. This paper links research on risk assessment to the use of voluntary safety plans by DCFS, and considers implications for the rights of parents, children, and the state.

Child protection services are one of the most important and controversial functions of the state.  State interventions in private family life can bring the safety of children and the rights of parents to care for their own into conflict.  Each state’s child welfare system involves both social work and legal systems.  In Illinois, social workers in the Department of Children and Family Services (DCFS) implement the day-to-day child protection casework, including home visits and family assessment.  Cases of abuse or neglect are referred to the courts, which oversee a child’s removal, reunification or other custody arrangement.  In these situations, DCFS and parents are often on opposing sides of a child welfare case; a judge hears the facts of the case and determines the appropriate custody, guardianship, and parental rights for a child. 

When a child protection case becomes court-involved, three critical sets of rights must be considered: the rights of parents, the rights of children, and the rights of the state. Parents have a fundamental right to the care and custody of their children, as located in the Due Process Clause of the Fourteenth Amendment.  In Troxel v. Granville (2000), the Supreme Court determined that a parent’s care and custody of their children is, “perhaps the oldest of the fundamental liberty interests recognized by this Court.”  Fundamental rights are not absolute, but present a demanding “strict scrutiny” standard the state must overcome in order to intervene.  In Illinois, the state is required to provide a hearing within 48 hours of taking temporary custody of a child, during which they must prove that there is a reason to believe the child is in “imminent danger” of harm, according to Illinois Compiled Statutes, Chapter 705, Section 405.  Additionally, parents are entitled to “reasonable efforts” by the states to reunify before parental rights are terminated permanently.  While children have the right to be cared for by their parents without third party interference, as established in Smith v. Org. of Foster Families for Equality  & Reform (1977), they also have the right to be free from abuse and neglect by their parents (Planned Parenthood v. Danforth 1976).  As for the state, it has a parens patriae interest, defined as “a profound interest in the welfare of the child, particularly his or her being sheltered from abuse” (Tenenbaum v. Williams 1999).  In this context, two casework practices of DCFS have come under scrutiny for their impact on parents’ fundamental rights: risk assessment and voluntary safety plans.

In cases where neglect or abuse is suspected, DCFS casework protocol requires a risk assessment be conducted during the initial investigative contact with the family.  The risk assessment attempts to determine the likelihood of imminent harm to the child, and, if harmed, the severity.  The legal standard for child removal is “imminent danger.”  Social workers utilize clinical skills to assess the situation and make a determination regarding the risk of imminent harm.  If a child is deemed “unsafe,” removal is not the only course of action; a social worker can also develop a voluntary agreement with a caregiver to prevent the child’s immediate removal.  Safety plans might involve actions such as temporarily removing suspected perpetrators of abuse from the home or asking that children stay with relatives until an investigation, which can take weeks or months, is concluded.  

This paper examines research on risk assessment methods in child welfare, as it relates to safety planning in DCFS.  Risk assessment practices have become increasingly accurate at identifying a child’s risk for abuse or neglect.  However, use of a substandard risk assessment protocol, which may not accurately identify imminent harm, has the potential to misguide intervention, including the use of safety plans, and to infringe on parents’ rights to the care and custody of their children without accurate cause.  This paper will link the latest research on risk assessment practices to the legal controversy over the use of safety plans within DCFS and will conclude with recommendations for public policy and social work practice. 

Risk Assessment Practices in DCFS: Lagging Behind the Research

Implemented in 1994, the DCFS Child Endangerment Risk Assessment Protocol (CERAP) is 15-question “yes/no” checklist of risk factors for re-abuse, mitigating circumstances, and family strengths.  .“Re-abuse” is defined as the recurrence of abuse or neglect within 60 days of the start of the investigation.  Mitigating circumstances are conditions that reduce the chances for abuse or neglect and family strengths reflect the psychological or relational resources a family can draw upon for support.  After filling out the 15 questions and narrative components, a worker uses clinical judgment to check one of two boxes: “safe” or “unsafe.”  “Unsafe” means that a child is in imminent danger of moderate to severe harm. 

CERAP is known as a consensus-based model because risk factors are derived from child welfare expert consensus, rather than evidence-based findings from research.  Commonly used consensus-based models also include California Family Assessment Factor Analysis, known as the “California model,” and Washington Risk Assessment Matrix, referred to as the “Washington model.”

A second type of risk assessment is called an actuarial model, which differs from CERAP in important ways.  Actuarial models include survey items that empirical evidence suggests are correlated with re-abuse.  Each risk factor is statistically weighted to produce a high/moderate/low risk indicator.  Actuarial models are not meant to replace clinical judgment altogether, but provide a structure for decision-making that counteracts cognitive errors and biases inherent in clinical judgment (Dawes 1996).  A commonly used actuarial model is Structured Decision Making, referred to as the “Michigan model.”

Actuarial Approaches Consistently Outperform Consensus-Based Models

There is much research on the accuracy of various risk assessment models in predicting re-abuse.  It is important to understand the conclusions of this research in order to better understand its ongoing role in the creation of safety plans. 

From 1994-2000, the incidence of re-abuse in Illinois within the first 60 days of initial investigation fell from 2.7% to 1.3% (Garnier and Nieto 2001).  This is a substantial reduction in re-abuse rates.  Researchers credit the positive impact of CERAP for the drop (Fluke, Johnson and Edwards 1997).  However, it must be asked if the CERAP led to more accurate risk assessments and therefore to more effective interventions on behalf of at-risk children, or if CERAP inflated the measure of risk, resulting in greater – and unnecessary – interference in parental rights.  Removing substantially more children from the homes of parents in that time period would produce the same effect of lowering re-abuse rates.  The true measure of a risk assessment instrument is not only that it reduce re-abuse rates, but that it accurately distinguish families who present a high risk for re-abuse from those at low risk.  So, what does the research conclude about risk assessment validity? 


Baird and Wagner (2000) conducted the only nationally representative validity evaluation of risk assessment models.  They compared the two most widely used consensus-based models, California’s and Washington’s, with Michigan’s widely used actuarial model.  In the study, experienced child welfare workers applied each model to real case facts in order to assess the models’ accuracy in predicting re-abuse.  Though the study did not include the Illinois CERAP, the two consensus-based models provided a suitable reflection of the CERAP design. 

If a risk assessment model is accurate, “high risk” determinations will show the highest rates of re-abuse, “moderate risk” determinations will show lower rates of reabuse, and “low risk” cases will show the lowest rates of re-abuse.  The study found that only the actuarial assessment was able to differentiate between the three risk levels accurately (Baird and Wagner 2000).  When case workers used the California consensus model, children rated at “moderate risk” and “high risk” had identical re-abuse rates.  When the case workers applied the Washington consensus model, children estimated at “low risk” and “moderate risk” had identical re-abuse rates.  In Illinois, “moderate risk” is often the minimum for immediate intervention to protect children and yet the findings of the study suggest that a risk assessment done with a consensus model cannot accurately distinguish between children at “low risk” and “moderate risk” of re-abuse.  

The conclusions of the study are clear: actuarial models proved vastly better at identifying the true level of abuse risk to children.  Further, the consensus-based models consistently overestimated the level of risk to children, while underestimating the cases in which children were actually at high risk of harm.  Baird and Wagner’s conclusions about the superiority of actuarial models are supported by other research on the topic.  A meta-analytic review of 136 studies testing the two approaches indicated the superiority of actuarial models in nearly all of the studies (Dawes, Faust and Meehl 1993).  A study of the same two models of risk assessment in the New York child welfare system in the late 1990s led to a state mandate for the use of actuarial approaches based on their superior performance in assessing risk (Falco, George and Salovitz 1997).


Another critical part of measuring the effectiveness of risk assessments is inter-rater reliability: when completing a risk assessment instrument, would two workers conclude similar risk levels on the same case (Rossi, Scheurman and Budde 1996)?  Kang and Peortner (2005) conducted an inter-rater reliability study of CERAP.  DCFS workers reviewed records of real cases and conducted a CERAP risk assessment for each case.  The researchers report that the reliability of CERAP was “weak.”  For the same cases, workers identified a wide range of risk factors and recommended very different interventions.  Other studies have shown that actuarial models have much stronger inter-rater reliability. 

Baird, Wagner, Healy and Johnson (1999) tested the two consensus-based models mentioned previously, those from California and Washington, with Michigan’s actuarial model. The study concluded that the actuarial approach was significantly more reliable than consensus-based approaches.  The authors concluded the consensus-based approaches tested were “well below adequate.”  Case workers rarely reached similar risk levels when given the same case facts (Baird et al. 1999, 743).

In sum, the research literature indicates that actuarial approaches succeed in evaluating risk and that consensus models have “serious problems” (Baird et al. 1999, 846).  Therefore, it can be concluded that CERAP shares features of unreliability with the consensus-based models evaluated by Baird et al.  Furthermore, the CERAP does not assist the worker in summarizing the information gathered and calculating risk, a function which would simplify the assessment process, reduce unintentional bias, and improve decision-making in time-pressured workplaces (Baird et al. 1999, 743).

Voluntary Safety Plans

When a child has been indicated as “unsafe” using CERAP, the child welfare worker will often offer a safety plan to the family in lieu of the immediate removal of the child.  Safety plans are tailored to the specific circumstances of each case.   For example, in a sexual abuse investigation, a social worker and family can agree that the parent accused of perpetrating the abuse live elsewhere until the investigation is complete.  DCFS protocol indicates that cooperation from the family should be enlisted when developing the terms of a safety plan.  Because a child or children have been deemed “unsafe” according to the risk assessment done through the CERAP, the social worker must explain to the parents that if there is refusal to sign or follow through with an appropriate safety plan, the child or children may then be removed from the home, according to Title 89, Chapter III of agency regulations.

The Role of Voluntary Safety Plans

Safety plans have several useful purposes in the child welfare system.  First, they offer an intermediary step between unrestricted custody by parents and protective custody by the state, an approach that provides stability in the child’s life and respects the custody rights of non-offending parents.  Second, a safety plan offers a child welfare worker the opportunity to establish a therapeutic alliance with a family by jointly planning for a child’s safety.  According to Dore and Alexander (1996), the therapeutic alliance is a positive relationship between worker and client that serves as a vital resource for beneficial client change.  When utilized properly, safety plans allow the worker and family to collaborate in meeting the goal of child safety.  

The Legal Controversy with Risk Assessment

Again, according to Title 89, Chapter III of DCFS regulations, social workers are required to notify parents of the possible consequences of refusing a safety plan, which can include protective custody and/or a referral to the State’s Attorney’s Office for a court order.  Plaintiffs in a recent Seventh Circuit Court Appeals case, Dupuy v. Samuels (2006), argued that though parents are officially said to choose participation in a safety plan, in practice, their participation comes through coercive tactics.  Plaintiffs presented evidence that social workers, in practice, threatened the removal of children if a voluntary safety plan were not signed.  Social workers’ promises of child removal constituted a coercive threat in which social workers overstepped their authority because, in fact, only a judge may authorize a removal. 

Plaintiffs argued more broadly that the use of safety plans allows DCFS to circumvent court oversight by having parents “voluntarily” relinquish custody of their children.  In one example cited by plaintiffs, a father was prohibited from living with his wife and children for six months while DCFS investigated an abuse allegation. Without clear regulations indicating the length of time a safety plan can separate a parent from his or her children, and without court involvement to ensure that parents’ due process rights are considered, plaintiffs argued that safety plans represent a fundamental infringement on parents’ rights.  The court in Dupuy v. Samuels decided against the plaintiffs, arguing that safety plans are purely voluntary and that coercion only occurs when a social worker uses illegal means to obtain agreement to a safety plan, such as making physical threats against a parent. 

“Coercion” in Context

While the coercion standard established in Dupuy v. Samuels is that there is no infringement on parents’ rights in safety planning so long as the worker does nothing illegal, research on the function of power and reliance in the worker-parent relationship casts doubt on this contention (Bundy-Fazioli, Briar-Lawson and Hardiman 2008; Smith, 2008; Payne and Littlechild 1999).  According to Smith (2008), child welfare workers wield an immense amount of power during their interactions with parents.  Workers come equipped with an in-depth knowledge of the child welfare system and conferred status as government representatives while parents often have limited knowledge of the child welfare system and their legal rights.  One recent study of the social worker-parent relationship in Britain found pervasive feelings of powerlessness among parents (Bundy-Fazioli et al. 2008).  The parents in that study felt that workers had control over them, and some felt workers were unfair or abusive towards them.  Certainly not all worker-parent relationships are so negative, but the prevalence of “hierarchical and imbalanced power” is common (Bundy-Fazioli et al. 2008).

Not only is there a power imbalance in the worker-parent relationship, but the parent must rely on the worker’s judgment of their parenting ability, the safety of their children, and the presence of distressing socioeconomic circumstances versus neglect.  The parent must also rely on the worker for access to knowledge about the investigation process and for access to procedural rights.  It is within this context, with its imbalance of power, that a parent may hear a DCFS worker say: “We ask that you participate in this voluntary safety plan.  If you do not, the state may seek to take protective custody of your children.”

Further Complicating “Voluntary”: Disproportionate Minority Contact

Issues of coercion and “voluntariness” must also be considered in light of racial disparities in the implementation of child welfare law.  In Dupuy v. Samuels, the court stated that if a parent is indeed not abusing or neglecting children, the parent can freely refuse a safety plan, because safety plans are “optional” and “impose no obligations on anyone.”  If DCFS removes the children in response to a parent’s refusal to cooperate with a safety plan, the parent is entitled to a judicial hearing on the merits of the removal within 48 hours.

The court makes the critical assumption that the child welfare system and juvenile courts are neutral bodies where fair and equitable adjudication of abuse and neglect cases occur.  A wealth of research on disproportionate minority contact in child welfare indicates that in child welfare, all are not equal before the law (Chapin Hall 2009; Harris and Hackett 2008; Hill 2006; Roberts 2002).  Black children are three times more likely than White children to be removed from their families of origin and placed in foster care (U.S. Department of Health and Human Services 2000).  Higher rates of poverty among people who are Black only account for a small portion of this difference.  The majority of disproportionate minority contact in the child welfare system is due to institutional and individual biases at all levels of the child welfare system (Roberts 2002).  Even when caseworkers are given identical vignettes of child welfare cases, those families described as Black are far more likely than those described as White to be judged as abusive (Roberts 2002, 5).  A report by the Children’s Hospital of Philadelphia (2002) found that children with accidental injuries were three times more likely to be reported as abused if they were African American or Latino than if they were White.  In another study, race was shown to be the only explanatory variable in the higher reports of abuse and neglect for Black families (Eckenrode et al., 1998).

For Black and Latino families, there is no guarantee of equitable treatment by social workers, attorneys, or judges.  Within this context, when a social worker asks a parent to leave the home voluntarily, according to a safety plan that restricts parents’ custody of their child, the difference between “voluntary” safety plans and court-ordered removal may become blurred.

Implications for Courts, Public Policy, and Social Work Practice

Workers in child welfare are often thought of as “street level bureaucrats” because of the wide discretion they have in the daily implementation of public policy (Lipsky, 1980).  This discretion can lead to abuses of power in the worst cases.  But discretion can also harness the positive potential of street level bureaucracy to enact viable solutions to the current problems with safety plans.

The first change social workers can make at the street level is to divide safety plans into those that separate children from parents and those that do not.  For example, many safety plans call for parents to refrain from using physical discipline.  These plans do not infringe upon fundamental rights and should be kept as voluntary agreements.  However, safety plans that call for removal of parents or children from the home should be procedurally reclassified by child welfare agencies and given the same due process hearings as protective custody orders. 

A second solution, more specific to DCFS, is to alter the risk assessment tool in use.  There is precedent for such change.  In 1994, the Illinois legislature mandated that DCFS devise a new risk assessment tool that would reduce errors in risk estimation.  Given that the CERAP is now known to be less effective in estimating re-abuse risk, administrators in DCFS have cause to revisit CERAP and implement an evidence-based actuarial model.

A final aspect of improving the balance between ensuring child safety and respecting the rights of parents involves enhanced training for social workers.  “Child protection” is often a synonym for the child welfare system; however, it is critical that training of workers include not only attention to the protection of children, but also additional attention to knowledge of parental rights and an appreciation for the importance of the parent-child bond.  A more complete education in these areas is needed to ensure the child welfare system respects the fundamental rights of parents as it seeks to ensure the safety and well-being of children.


Baird, Christopher, Dennis Wagner, Theresa Healy and Kristen Johnson. 1999. Risk assessment in child protective services: Consensus and actuarial model reliability. Child Welfare 78(6): 723–48.

–––. and Dennis Wagner. 2000. The relative validity of actuarial- and consensus-based risk assessment systems.  Children and Youth Services Review 22(11/12): 839-71.

Bundy-Fazioli, Kimberly, Katharine Briar-Lawson and Eric Hardiman. 2008. A qualitative examination of power between child welfare workers and parents.  British Journal of Social Work (April): 1-18.

Chapin Hall Center for Children. 2008. Understanding racial and ethnic disparity in child welfare and juvenile justice. Chapin Hall Center for Children,  The University of Chicago.

Children’s Hospital of Philadelphia. 2002. “Minority children more likely to be evaluated for physical abuse; abuse in white children may be overlooked,” press release, October 1. 

Dawes, Robin, David Faust and Paul Meehl.1993. Statistical prediction versus clinical prediction: Improving what works.  In Handbook for data analysis in the behavioral sciences: Methodological Issues, eds, Gideon Keren and Charles Lewis, 351-67. Hove, Sussex: Lawrence Erlbaum. 

–––. 1994. House of cards. New York: The Free Press.

Department of Children and Family Services, Illinois. 2002. Reports of child abuse and neglect: Appendix G – child endangerment risk assessment. August 13. (accessed May 7, 2009).

Department of Health and Human Services, United States Government. The AFCARS Report, for the period ending March 31, 2000. cb/pubs/cwo00/cwo00.pdf

Dore, Martha M. and Leslie Alexander. 1996.  Preserving families at risk of child abuse and neglect: The role of the helping alliance. Child Abuse & Neglect 20(4): 349-61. Dupuy v. Samuels465 F.3d 757 (7th Cir. 2006).

Eckenrode, John, Jane Powers., John Doris, Joyce Munsch and Niall Bolger. 1988.  Substantiation of child abuse and neglect reports.  Journal of Consulting and Clinical Psychology 56(1): 9-16.

Falco, George and Barry Salovitz, Barry. 1997.  Clinical versus actuarial risk assessment in child protective services: Results from resent research in New York.  Paper presented at the 11th Annual CPS Roundtable on Risk Assessment, San Francisco.

Family Defense Center.  Petition for writ of certiorari, Dupuy v. McEwen.

–––.  Thousands of Illinois families – with the support of leading conservative and liberal groups – seek United States Supreme Court’s attention to child welfare agency abuses. stories/dupuy_press_release_2-13-08.pdf

Fluke, John, Will Johnson, Myles Edwards. 1997. CERAP safety assessment technical report April 1997. Children and Family Research Center, School of Social Work, University of Illinois at Urbana-Champaign. pubs/Pdf.files/cerap97.pdf.

Garnier, Philip and Martin Nieto. 2001. Illinois child endangerment risk assessment protocol evaluation: Impact on short-term recurrence rates – year five. Children and Family Research Center, University of Illinois at Urbana-Champaign.

Grove, William and Paul Meehl.1996. Comparative efficiency of informal (subjective, impressionistic) and formal (mechanical, algorithmic) prediction procedures: The clinical-statistical controversy.  Psychology, Public Policy, and Law 2(2): 293-323.

Harris, Marian and Wanda Hackett. 2008. Decision points in child welfare: An action research model to address disproportionality.  Children and Youth Services Review 30(2): 199-215.

Hill, Robert. 2006.  Synthesis of research on disproportionality in child welfare: An update.  Washington DC: Casey – CSSP Alliance for Racial Equity in the Child Welfare System.

Gambrill, Eileen and Aron Shlonsky. 2001. The need for comprehensive risk management systems in child welfare. Children and Youth Services Review, 23(1): 813-37.

Kang, Hyun-Ah, John Poertner. 2005. Inter-rater reliability of the Illinois structured decisions support protocol. Child Abuse & Neglect 30(2): 679-89.

Lipsky, Michael. 1980. Street-level bureaucracy; dilemmas of the individual in public services.  New York: Russell Sage Foundation.

Payne, Helen,and Brian Littlechild. 1999. Ethical practice and the abuse of power in social responsibility.  London: Jessica Kingsley.

Planned Parenthood v. Danforth, 428 U.S. 52. 74 (1976).

Roberts, Dorothy. 2002. Shattered bonds: The color of child welfare.  New York: Basic Civitas Books.

Rossi, Peter H., John Schueman and Stephen Budde. 1996. Understanding child maltreatment decisions and those who make them. Chapin Hall Center for Children, University of Chicago.

Santosky v. Kramer, 455 U.S. 745, 753 (1982).

Smith v. Org. of Foster Families for Equal. & Reform, 431 U.S. 816, 844-45 (1977).  Cited in Alyson Oswalt, They took my child!  An examination of the circuit split over emergency removal of children from parental custody, Catholic University Law Review 53 (2004): 1161.

Smith, Roger2008. Social work and power.  London: Palgrave MacMillan.

Tenenbaum v. Williams, 193 F.3d 581 (2d Cir. 1999).

Troxel v. Granville. 530 U.S. 57 (2000). Cited in Alyson Oswalt, They took my child!  An examination of the circuit split over emergency removal of children from parental custody, Catholic University Law Review 53 (2004): 1161.