a critical look at the child welfare system

Decision Making in Child Welfare

Since the late 1950s, the suggestion has been made that research in child welfare should focus on the decision-making process, so that the variables entering into any given decision can be defined and a decision-making framework to guide child welfare caseworkers could be devised.[1]

In 1983, Theodore Stein and Tina Rzepnicki examined the literature on decision-making by child welfare workers, explaining:

For over two decades, researchers have made efforts to discern the processes that child welfare workers engage in when making decisions for children. This research proceeded on the assumption that if key decision-making points could be identified and the decision-making process described, a decision-making framework to guide child welfare staff in making critical choices could be developed inductively from practice knowledge.

“Unfortunately, consistent decision-making principles have not been identified,” they concluded. [2]

In 1994, Duncan Lindsey, Associate Professor of Public Policy and Social Research at the University of California, again reviewed the literature, concluding that “the child welfare field does not possess an adequate scientific knowledge base for determining which cases are best served in-home and which need out-of-home care.”[3]


As Stein and Rzepnicki explored the literature on decision-making, they identified a number of critical questions which investigators over the years had sought to answer.

Among these fundamental questions were “How do workers make intake decisions?”, “What is the basis for deciding whether children should be placed in out-of-home care?”, “Should children be placed in institutions or foster family homes?”, “When should children who are in placement be reunited with their biological parents?”, “What are the criteria for selecting foster parents and adoptive applicants?”

Significantly, Stein and Rzepnicki noted that at the time of their study, that in more recent years the question had turned to one of how to determine whether or not a child had been abused or neglected.[4]

“Upon what evidence did child welfare professionals decide to place these children in foster care? What factors influenced the decision? Were the decisions reliable, consistent, fair?”

The answers to such questions are the key to evaluating the direction, fairness, and effectiveness of the child welfare system, Lindsey explains. Reliable and valid decisions thus represent the linchpin of the current child welfare system. The decision-making process involved in determining which children are removed from their parents is central to the operation of the child welfare system. Decision-making must have a scientific basis to ensure that the decision to remove children is not biased or prejudicial.[5]


A 1995 study of child protective services caseworker decision making conducted in Texas found that “no clear pattern regarding the decision to provide services or to remove the child from the home emerged.”

Admission or denial on the part of the alleged perpetrator, the capacity of the caregiver as represented by the absence of parenting skills, unrealistic expectations of the child, as well as the refusal to accept parental responsibility were among some of the placement factors the researchers identified.[6]

Bernice Boehm studied caseloads in an effort to identify the criteria used by caseworkers in making the decision to place a child out of the home or to provide at-home services, finding that workers had a “tendency to use a model family as a frame of reference, and to evaluate problem behaviour by the extent of deviation from ideal parenting.”

She concluded that all families deviated from the model, whether or not the children were placed, and suggested that the problem at hand was to distinguish between placement and non-placement families.

Boehm went on to ask workers to submit lists of “behavioural itmes” they used to evaluate family adequacy. Of the items dealing with the father, only his “interest in and affection toward the child” was related to placement decisions. Boehm ultimately concluded that “the decision for placement is based largely on an evaluation of maternal care.”[7]

Similarly, Alfred Kadushin reported a study by Eugine Shinn which pointed to “maternal pathology” as the most frequently cited reason for placing a child in state care.[8]

Let us compare these findings against a study conducted by Michael Phillips, Ann Shyne, Edmund Sherman, and Barbara Haring. Phillips and colleagues examined intake decisions for 309 children in three state agencies located in two eastern cities. No single item or combination of factors pointed strongly to a placement decision.

In contrast to the findings of Boehm and Shinn, they noted that in two-parent families, “the traits of the father were the single most important cluster of variables” in placement decisions, and that neither the “mother’s functioning nor her relationship to the child were important considerations in reaching a decision.”[9]

Edmund Mech reported a study that sought to delineate factors involving whether a child should be maintained in custody or returned to the parent. Differential factors related to either decision could not be identified. Mech concluded that no framework for decision-making existed, and the development of one was vital.[10]

In one of the most comprehensive of the early studies of decision-making criteria, William Ryan and Laura Morris studied the total intake (N = 683) of thirteen public, voluntary, and sectarian agencies constituting the basic child welfare system in metropolitan Boston. An intensive case reading analysis was then conducted of the 265 cases ultimately accepted for services.

The data suggested that intake decisions were made on a very stereotypical basis, largely unrelated to individual case need. The lack of flexibility in intake policies and decision-making processes, combined with the lack of necessary resources, made it impossible for the participating agencies to accept and serve many of the clients referred for services.

Ryan and Morris concluded that diagnostic criteria play no role in critical intake decisions, and that the concept of a comprehensive child welfare service network in metropolitan Boston was essentially a myth.[11]

Theodore Stein analyzed 68 intake and service worker interviews, finding that between 59 to 78 percent of the information gathered during intake the interview was not related to the objective of reaching a placement decision, and between 22 to 90 percent of the information in service interviews did not relate to the interview goals set by the workers.[12]

Similarly, Naomi Golan found that workers gather excessive amounts of information in community health centers. She reported a “lack of uniformity in the data gathered” by the 12 workers studied. There was either much data that appeared superfluous, or omissions in the data which appeared to be vital to decision-making.

Golan concluded that it “appears inescapable” that workers could carry out more economical and productive interviews, and arrive at more helpful decisions, if they knew what to focus on and which areas would yield the most significant data.[13] In one of the earliest studies on the reliability of foster care decision-making, Scott Briar examined the basis for making foster home vs. institutional placement decisions. Forty-three workers participated in his study, using hypothetical case material.

Briar reported that while there was a relationship between the degree of disturbance and the type of placement recommended, the direction of the relationship was variable and unpredictable, and that predictions about a child’s probable future varied substantially from caseworker to caseworker.[14]

In 1980, Donnelly examined foster care placement decisions by asking experienced caseworkers in four California counties to make recommendations on fifteen hypothetical cases. The decision to remove a child from the home varied substantially between the caseworkers. Those in Riverside and Alameda counties were more likely to remove a child than were their counterparts in San Bernadino and San Francisco counties, despite the geographic proximities.[15]

Such variations in decision-making are to be found throughout the states.

A 1995 South Carolina audit notes that while one county office confirmed abuse or neglect in 89 percent of reported abuse and neglect allegations, another county office confirmed only 14 percent.

The variations in substantiation rates were found to be even wider when the broader category of “maltreatment” was alleged. The variations in substantiation rates varied from one county office’s determination that no maltreatment allegations were unfounded to another county office’s determination that 74 percent of the allegations were unfounded. The auditors noted that “the percentage of confirmed allegations among counties varied more than could be explained by outside factors.”[16]

Judges reliability would appear to be equally as variable. Phillips and colleagues compared decisions reached by three independent judges, each of whom had more than five years of experience in child welfare. The judges agreed with each other and with the caseworkers on less than half of the cases, and when they did agree, they did not identify the same factors.[17]

In a follow-up study, Phillips, Haring and Shyne sought to assess the reliability of the decisions made by experienced child welfare caseworkers and judges in recommending for or against removal.

In examining 127 child placement cases, the researchers found considerable disagreement between not only judges and caseworkers, but between judges themselves.

The overall agreement between the six judges examined was less than 25 percent. Of particular significance was the wide variation between two judges, with one being four times more likely to recommend in-home services than another. While the one judge acceded to removal in 17 percent of cases, the other acceded to remove the child in an astounding 72 percent of the cases studied.

Even in those cases in which the judges agreed to remove a child, they varied substantially on the type of case plan or services to be provided.[18]

Nor would the reliability of decision-making seem to improve once a child has been removed and placed into foster care.

Deborah Shapiro reported that caseworkers with more experience were more likely than those with lesser experience to discharge children from foster placement during the first year in care.[19]

What does this suggest in terms of the high turnover in the field today?

A 1996 study conducted by Trudy Festinger sought to identify those factors which would serve as predictors for reentry into foster care.

While the results were inconclusive, six variables relating to the decision were identified: parenting skills, social support, unmet service needs, caregiver problems, organizational participation, and caseworker experience. The results showed the two factors which were the strongest predictors of reentry were “limited parenting skills, such as assessed problems in communicating with their children, understanding child development, and handling discipline” and “a limited level of support from family, friends and neighbors.”

In results somewhat similar to those of Shapiro, Festinger identified the length of experience in the field of child welfare as having some bearing on the decision.[20]

As to decisions involving the selection of foster parents, Martin Wolins found that the amount of information in a case record had a marked impact on the decisions made by judges. Agreement between judges climbed from a low of .47 when the entire case record was used, to a high of .81 when judges were provided only the material most relevant to the decision. Of particular interest is that when 80 percent of the case record was randomly discarded, the level of agreement between judges was as high or better than it was when the entire case record was read.[21]


The choices made at any single point can set the course for future transactions between worker and client, as well as between members of a family.

As Stein and Rzepnicki explain, “the accuracy and quality of data are conditioned by whether workers decide to conduct assessments in the office, in the client’s home, or both; whether they elect to gather information through verbal exchanges during interviews, through client self-reports, through direct observation of family members, through reports by third parties, or a combination of these methods.”[22]

Hence, the questions of how and where such critical decisions are made are significant. Just how and where are they made?

Dennis Lepak, a probation officer for Contra Costa County, California, and vice president of the State of California Organization of Mental Health Advisory Boards, explains that: “Most tragically, children are placed with little or no services to prevent their removal from families. Children are often removed from homes that no representative of the removing agency has even visited.” Instead, says Lepak, decisions to remove children are based on information from old reports, office interviews, and phone calls.[23]

Similarly, sociologist John Hagedorn reports that in Milwaukee, investigations are often conducted and decisions made without the caseworkers so much as leaving the agency lunchroom.[24]

In Texas, an ABC Television affiliate examined child welfare procedures, finding that found that almost 50 percent of Child Protective Services caseworkers never actually leave their office or inspect the conditions or circumstances of the children they are charged to protect. Caseworkers instead often rely on ex parte conversations with attorneys, or the reports of court appointed psychologists, rather than following established procedures.[25]

Some caseworkers do continue to make home visits, while some others extend themselves to calling the accused family in for a face-to-face interview before making a placement decision.

Says Adrienne Cox, assistant director of the Department of Family and Youth Services in the Las Vegas area, her Child Protective Services division prefers to conduct interviews at government offices because resources are more accessible, and the safety of social workers can be more readily ensured.[26]

Decision-making in the area of adoption would similarly appear to require substantial improvement.

In one of the earliest studies of the decision-making process in adoptive placements, Donald Brieland tape-recorded interviews with five couples who were adoptive applicants, playing these tapes for 184 workers in 13 states.

During and after each interview, the caseworkers were asked to rate the suitability of the couples for adoptive placement.

Brieland reported that decisions were based on information contained in the first half on the interview, and that the information in the second half bore no relationship to the decisions the workers made. He concluded that the workers had become “locked into” a decision during the first half of each interview, and that the remaining data did not contribute to the final outcome in any way.[27]

Fiscal incentives contribute significantly to the problem. As Stein and Rzepnicki observed: “The fiscal incentives in federal policy have a direct influence on the policies and actions of state-level administrators, and these, in turn, shape the professional decision-making environment in which child welfare practioners must function.” Despite the value placed on family sanctity, and on the rights of parents to raise their children free of outside interference, federal funds for programs to prevent placement of children or for services to reunify families after children enter care have historically been low.[28]

In contrast, federal funds for foster care placements have historically been high. By the end of the 1970s, nearly three quarters of all child welfare dollars were being spent on foster care supervision and payments.[29]

As a result, argues sociologist John Hagedorn, the investigation of poor families and the removal of children from their homes have crystalized as the core tasks of social work, those functions which define what a social worker does on a day-to-day basis.[30]

With foster care being the primary service child welfare agencies possess, the decision of whether or not remove a child into state care is often the only decision a caseworker worker has to make.

Just how do Child Protective Services caseworkers arrive at their decisions today? Recent research would indicate that they frequently make decisions about cases before assessments and/or paperwork is completed.

Writing in the industry journal Child Welfare, Diana English and Peter Pecora conclude: “At present, caseworkers rely on intuitive processes based on supervision, experience, and training to make decisions.”[31]

Yet it is precisely the lack of supervision, experience, and training that serve as the rules, rather than the exceptions, among far-too-many Child Protective Services caseworkers today. As a recent performance audit in the state of Utah explains: “Better training, more focused supervisory review, and changes in staffing will help ensure that children are protected, families are preserved where possible, and a permanent home is established.”[32]


Add to this the lack of clear definitions as to what constitutes child abuse and neglect, coupled with lack of clearly defined intake guidelines, and it becomes clear why Child Protective Services cannot deliver on its mandate to protect children–even as it continues to needlessly destroy families. As Douglas Besharov explains:

These laws set no limits on intervention and provide no guidance for decision making. They are a prime reason for the system’s inability to protect obviously endangered children even as it intervenes in family life on a massive scale.[33]

As Jeanne Giovannoni and Rosina Becerra explain, “Many assume that, since child abuse and neglect are against the law, somewhere there are statutes that make clear distinctions between what is and what is not child abuse. But this is not the case. Nowhere are there clear-cut definitions of what is encompassed by the terms.”[34]

Stein and Rzepnicki note that the problems created by the absence of guidelines for gathering and using information are sufficient to explain the decision-making problems uncovered by investigators. Without criteria to guide data gathering, the process breaks down in its earliest stages.[35]

Nor are there legal standards to protect a family once it has entered the system, as existing legislation grants judges and caseworkers virtually unrestricted dispositional authority. As a result, “decision making is left to the ad hoc analysis of social workers and judges.”[36]

Existing legislation invests judges and state agency personnel “with almost limitless discretion to act in accord with their own child-rearing preferences in areas generally under the exclusive control of parents.” Identifying children whose placement should be called into question requires more than vague and subjective language, and statutes must be revised in order to “prevent judges, lawyers, social workers, and others from imposing their personal views upon unwilling parents.”[37]

Add to this mix a darker side of human nature in the form of overt bias or prejudice. A recent grand jury investigation in California examining the disproportionate impact of foster care on black children suggests that racial bias may indeed be a significant placement factor.

The jury reported that case records and court reports for white children were “consistently more detailed, better prepared and oriented toward family reunification, adoption or guardianship” than cases involving minority children. While white children’s files had “well-documented” plans for permanent placement, such as adoption or guardianship, minority children’s files did not contain any evidence of a permanent placement plan.[38]

In a New Jersey case, an Hispanic father began beating his children after suffering a head injury at work. When asked why she had removed the children without looking to see if some other treatment methods may have prevented removal, the caseworker reportedly said: “Those people shouldn’t have their kids anyway because those people have too damn many kids.”[39]

The Native American continues to bear the brunt of such prejudice, and the problem is not limited to the caseworkers. Says Art Martinez, Ph.D., Clinical Supervisor with the Tolyabe Indian Health Project, “I’ve had judges tell me, ‘I thought we killed all the Indians.'”[40]

If there is one personal bias which manages to creep across all racial boundaries, it is that which Dana Mack has described as “the pernicious confusion of poverty with neglect.”[41]

There remains an additional placement factor which is exceptionally difficult to quantify–that of defensive social work. Notes Douglas Besharov, founding director of the National Center on Child Abuse and Neglect, there is no denying that it affects all aspects of child protective decision making.[42]

As one former New York City caseworker explains: “If the kid’s safe, leave him. If the kid’s not safe, take him. If you’re not sure, take him.”[43]


In 1994, University Professor Duncan Lindsey revisited the literature on child welfare decision-making, applying a hypothetical model which viewed decision-making as a stochastic (random) model.

In applying a reliability factor of .25 to the model–a reliability factor higher than that found in the studies he had examined–Lindsey concluded that “the low reliability leads to a system that is unable to discern which child should be removed and which child should be left at home.”

The most salient feature of this model is that it does not require assumptions of bias or prejudice on the part of the child welfare caseworker to account for the removal of a great many children not in need of placement or the returning home of a large number of children in true need of placement.

Caseworkers in doubt about a child’s situation make the safe decision to remove a child. Too often, the caseworker is in doubt, thus the caseworker too often places the child in foster care. Lindsey explains:

It is hard to imagine how the results of the stochastic model could be more distressful, in terms of what it suggest for the outcome of children considered for removal. If the level of reliability were to slip much further than .25, all children, except in the most extreme cases, would have an equal likelihood of being placed in foster homes, meaning that the decision-making process would be roughly equivalent to a lottery![44]

What must be remembered is that behind all of these decisions stand very real families–and very real children. They deserve better than the results of a lottery when it comes to decisions which may bear an impact on the rest of their lives.