The aim of this paper is to provide an overview of some of the trends and issues in the identification and assessment of children by child protection services where there have been allegations or a suspicion of child abuse and neglect. More importantly, a wider context is provided in which the various approaches and constituent elements of ‘gatekeeping’ or screening and the assessment of risk may be discussed.

It should be noted that while I am is cognisant of the difficulties of child protection work and various stresses and pressures placed upon statutory child protection services to ‘get it right’, as a researcher and systems observer or ‘analyst’, I do not face these pressures directly. Similarly, the views and proposals I outline in this paper are not constrained by budgetary restrictions and legislative requirements. Therefore, while I have attempted to take into account the realities of current child protection practice, my intention is to extend the boundaries, to provide ‘ideal’ models, possible system structures and to critique current practice. Although some of the proposals suggested may be unobtainable in the short term, my intention is to provide a framework to inform future action. Or at least, one which can be used as a ‘benchmark’ when considering future directions.

Since the 1970s child protection services in Australia have been marked by continuous change as a result of, and punctuated by, a series of Government inquiries in most States. Concomitantly, the social welfare departments which have provided statutory child protection services have undergone extensive restructuring, often accompanied by a number of name changes (e.g. Goddard, 1988b).

Based on the work of Stevenson (1992), six system issues which have significantly affected practice are identified: the hostile scrutiny of child protection work; radical organisational change in the health, education and family support sectors; the significant rise in the number of child maltreatment reports and the increased demands on the child protection system; the effect of external factors, such as Compulsory Competitive Tendering (CCT), privatisation and ‘fees-for-service’ on family support work; shifts in the threshold for intervention; and a lack of resources to adequately assist ‘at risk’ and maltreating families.

The media coverage of child maltreatment, and child deaths or other child protection scandals in particular, has played a significant part in the instigation of many of the more recent changes to the child protection system (Goddard & Liddell, 1993; Goddard & Liddell, 1995; Scott, 1995; Goddard, 1996b; Loane, 1997). Media reports have resulted in ‘a "moral panic" about child protection and the channelling of resources into investigation and detection at the expense of supportive and preventative services (Scott, 1995)’ Markiewicz (1996:30).

In the 1990s, statutory child protection services throughout Australia and across the western world, have been struggling to cope with ever-increasing numbers of reports of suspected child maltreatment. Concomitantly, resources have been diverted from wider child and family services, services essential for the prevention of child abuse and neglect and the effective support of families requiring assistance to cope with issues impinging on their wellbeing (Dartington Social Research Unit, 1995; Little, 1995; Tomison, 1996; Armytage, Boffa & Armitage, 1998).

What is also clear, is that a substantial proportion of the reports made to child protection services are inappropriately labelled as allegations of child maltreatment by those who have made the referral (Clark, 1995; Little, 1995; Tomison, 1996; Department of Human Services, 1997). Some of these cases are assessed as inappropriate referrals and screened out immediately, others may appear serious enough to warrant an investigation and an assessment of the need for protection.

Often such cases involve families who are not maltreating their child but have more generic problems, such as financial or housing difficulties, an incapacitated caregiver, or serious stress problems. Such cases may need assistance but do not require child protection intervention. Their labelling as cases of child abuse or neglect further taxes generally limited child protection resources (Clark, 1995; Little, 1995; Tomison, 1996).

A research program implemented by the United Kingdom’s Department of Health (DoH) and coordinated by the Dartington Social Research Unit, University of Bristol (1995), has produced a substantial amount of evidence about the definitions of maltreatment, what happens to children and families caught up in child protection systems, and case outcomes (the Messages from Research report).

Of fundamental concern to statutory child protection services are those children who are thought to have experienced, or be at risk of, significant harm. In the child protection systems of most western countries, the occurrence of child maltreatment is only substantiated in 40 to 50 per cent of notifications which are directly investigated; with only a small proportion of child abuse and neglect notifications result in legal intervention to protect children. It has been argued that over the last decade the style of intervention for all families has become ‘forensically driven’, despite the fact that legal action will not be taken for the majority of families with whom protective services is involved (Little, 1995; Tomison, 1996; Armytage et al., 1998). This approach results in the shifting of scarce resources away from cases of significant harm to the investigation of notifications, and interventions which are less effective in terms of meeting family needs.

It also raises questions in relation to both protective service screening or ‘gatekeeping practices’ and the nature and availability of broader primary and secondary prevention services in the community. Within this, the dilemma is described as one of distinguishing child protection problems from broader welfare concerns, and, in all instances, delivering an appropriate response matched to client need. Sound risk and family assessment skills are seen as fundamental to resolving this dilemma.

Research has further identified distressing and alienating consequences for families of child protection investigations which affect family functioning and relationships, including their preparedness to seek help in the future, long after child protection services have ended their involvement. While detrimental secondary consequences of investigations should be minimised for all clients, such traumatic effects are particularly concerning in relation to those groups of clients not found to require ongoing protective services involvement following initial investigation.

Another consequence of the ‘forensic’ approach has been to ‘cast child protection services as the "expert" and to alienate essential community professionals from a partnership approach to the prevention, support and protection of children’ (Armytage et al., 1998:2).

In contrast, good case outcomes in child protection are found to be strongly associated with the quality of the relationship between the professional and the family; sensitive and informed case practices; and a wide perspective on child protection; and supervision emphasising practice development and interagency collaboration. Using the principles of a child centred family focused or solutions focused approach, emphasis is placed on positively engaging and involving families in all decision making and action from the point of initial contact and throughout the case career. The balance of power between professionals and family members is viewed as critical and emphasis is on partnership based relationships and recognition of family strengths (Dartington Social Research Unit, 1995; Little, 1995; Tomison, 1996; Armytage et al. 1998).

In the United Kingdom, Messages from Research proved to be a watershed in the development of policy and practice in child protection and child welfare (Parton, 1996).

The recommendations from Messages from Research were translated into two major areas for action: enhancing the investigation and screening of child maltreatment reports (including re-balancing in favour of a family support, rather than a forensic approach); and the assessment and treatment of families’ ‘needs’, rather than ‘harms’.

The difficulties of having to deal with a large number of reports, and the recognition that many of the families in unsubstantiated maltreatment cases require support in order to deal with other social and financial problems, has led some Australian State and Territory governments to consider the adoption of ‘new’ models of child protection and family support (Tomison, 1996; Mackieson, 1997). As Armytage et al. (1998) note, such approaches are not new, but are often a re-visiting or recapitulation of solutions previously tried and tested.

Under such an approach cases are no longer seen in purely child protection terms but in the overall context in which the alleged maltreatment has occurred, informed by a comprehensive assessment of the family. Thus, the balance between child protection and the role of family support is altered such that child protection no longer drives the system but becomes merely one important facet in an overall welfare assessment. Good practice and adequate protection thus both emerge from adopting a wider perspective on child protection by means of which underlying problems in the family that may put a child ‘at risk’ or have a detrimental effect on the child’s long-term welfare are addressed (Little, 1995; Tomison, 1996).

The overall intention of the new approach, as envisaged by the Dartington Social Research Unit, was to prevent maltreatment and protect children by addressing family problems holistically. This was to be done in ways which promoted cooperation between workers and families, achieving greater levels of parental cooperation and, subsequently, a better outcome for children and families.

Ideally, the benefits of such a system would be that families would not be unduly stigmatised or traumatised by inappropriate or unnecessary protective investigations, and would therefore be more likely to accept assistance. In addition, family problems would be comprehensively assessed and appropriate services put in place to address them, thus preventing the development of maltreating behaviour, or reducing conditions detrimental to a child’s long-term development.

Finally the model recognised that child protection services are part of the wider child and family support system and reinforced the need for effective collaboration between child protection services and other family support agencies in order to more effectively assess family needs and to provide a response that could positively affect family wellbeing and ensure the protection of children from abuse and neglect. Such models, if appropriately resourced, would enable child welfare and family support agencies to regain prominence in child abuse prevention and the early detection of ‘at risk’ children, a role which many services have been unable to perform as substantially in the 1990s because of a lack of resources, exacerbated by the high demands for services that accompanied the recession of the late 1980s and early 1990s (Tomison, 1996; Armytage et al., 1998).

Although the adoption of a ‘family support’ model of child protection can be beneficial for many children and families, it can also have potentially negative consequences. If inadequate resources are put in place to enable agencies to cope with the demand for services which results from such an approach, then families, particularly families labelled as being ‘at risk’, or as having a generic welfare or family support concern, are no more likely to receive support or remediation of their problems than they would under the current ‘forensic investigation’ models of child protection (Tomison, 1996).

That is, case screening and risk assessment may more effectively target protective investigations and legal intervention to those families where children are ‘at risk’ or are being maltreated. But without the resources to treat and support these families and those screened out with more generic family problems, nothing may be done to alleviate the concerns which led to a notification in the first place. This issue will be discussed in more detail below.

In addition, the interprofessional and interagency communication and coordination problems that have beset child protection systems since the modern discovery of child abuse in the 1960s (Kempe et al., 1962), may be exacerbated by a ‘family support’ model. With child protection services reducing their role as primary or coordinating agencies, family support services will have greater responsibility for case coordination.

It is therefore of paramount importance that adequate means are put in place to ensure adequate resourcing of the family support system and the development of a structure to enhance comprehensive interprofessional involvement in caseplan discussions. If not, it is likely that history will repeat itself, with cases slipping through cracks in the system or families being inadequately serviced (Tomison, 1996).

Variations of this new approach is currently being adopted in the United Kingdom and across a number of Australian States. In the UK, Parton has noted that:

The primary response to Messages has been the restructuring of case intake procedures; specifically, gatekeeping or screening and risk assessment methods. In Australia, it would appear that Western Australia has also focused predominantly on better gatekeeping or screening, in South Australia, there has been the development of a comprehensive system of screening and resource targeting, using a serries of structured risk assessment checklists. Victoria, in contrast, has adopted a differentiated response to notifications (screening), in combination with the development of a risk assessment guide and the statewide implementation of the Strengthening Families program, designed to provide support and treatment to those ‘at risk’ families who do not require a protective response.

The professional context of child maltreatment ‘is of a social and legal pendulum which oscillates somewhat unpredictably between poles of Prevention/Family Support on the one side, and Child Protection/Rescue on the other’ (Dale, 1998:1).

The direction of the swing is determined by whether a non-interventionist approach, where there is subsequently a perceived failure to remove children in need of protection (‘false negatives’), or an emphasis on legal action and child removal (which may lead to some overzealous actions and the inappropriate removal of a child ‘false positives’) is more perilous politically at that time (Scott, 1995; Dale, 1998). The intervention pendulum may swing in different directions for different types of abuse (e.g. fear of under-intervention for physical and sexual abuse cases; over-intervention in cases of neglect or emotional abuse) (Scott, 1995). The criteria for determining which cases will be accepted or investigated is, at its core, a moral and political matter, which should be made explicit’ (Wald & Woolverton, 1990; Dale, 1998).

The high numbers of notifications received by child protection services, and in particular, the high number of ‘inappropriate referrals’ have helped to swing the pendulum towards non-intervention, that is, to target protective investigation, assessment and casework strictly to those families where there appears to be strong cause to suspect that child maltreatment is occurring.

In 1995, Western Australia set up a new model of case intake, where cases were classified into either a generic ‘child concern report’ or ‘child maltreatment allegation’ stream. Regardless of the stream into which a case fell, all cases were to undergo a full assessment and receive supports where necessary.

Under the West Australian approach, an experienced child protection worker decides at the time of report whether a case requires a protective assessment by the child protection team, thereby being designated as a ‘child maltreatment allegation’, or whether it can be managed as a ‘child concern report’ and can therefore be referred to welfare services for an assessment and the provision of services.

In this model, only severe incidents of maltreatment are initially referred to the child protection team: for example, where an illegal act has been committed, where there is evidence of severe or persistent harm or where a significant history of child maltreatment in the family exists. The majority of cases are expected to be assessed as ‘child concern reports’, but regardless to which stream a case is assigned, appropriate services were meant to be provided.

However, it appears that inadequate resources were put in place to enable agencies to cope with the demand for services. Families, particularly families labelled as having a generic welfare concern, appear to be no more likely to receive support or amelioration of their problems than they would under the ‘forensic investigation’ models of child protection.

Significantly, this situation occurs in a system that streams cases. Thus, as a consequence, there has been a substantial decrease in the number of official child maltreatment cases, as many cases have been labelled as a generic concern rather than as a child maltreatment allegation. Overall then, with many cases no longer recorded as child maltreatment, and a failure to increase family support resources, there is a danger that the incidence of child abuse and neglect is reduced by definition rather than by a reduction in actual maltreatment. It could therefore be argued that such a system is a cosmetic reduction of concerns which, through inaction and the failure to provide adequate support, may contribute to the development of more serious familial problems, or the maltreatment of children in the long term.

Conversely, it could be argued that such streaming results in an more accurate record of the extent of child maltreatment issues in Western Australia. However, families without adequate support, particularly in harsh economic times, cannot fulfil the requirement of the UN Convention on the Rights of the Child of providing ‘an atmosphere of love and understanding’ (Rayner, 1995) and may be more at risk of maltreating their child. Child maltreatment may therefore be a particularly sensitive marker of the strength of the social fabric. Although it may not be politically palatable, I would argue that an accurate record of all reports or requests for service to child protection services, in conjunction with a depiction of the proportion of cases who actually require investigation and those that are subsequently substantiated, provides a valuable indication of the size of both child maltreatment and generalised family dysfunction in the community.

An alternative to the case streaming system applied in Western Australia, is a differentiated response system, where the intention is to recognise the varying degrees of risk and diversity of needs within the client population at intake. Victoria is currently implementing the Enhanced Client Outcomes (ECO) system within child protection services. The aims of ECO are:

• Reaffirming the importance of risk assessment as the basis of decision making in child protection and the key to discriminating between different client needs, including high risk and urgency issues

• Providing access to a range of differential response options ensuring sensitive and flexible responses to the full range of presenting problems.

• Promoting interagency relationships that ensure maximum knowledge of local resources and networks and use of advanced collaborative practice

• Builds on the principles of child centred family focused practice to ensure that the principles of partnership, strengths, sensitivity and respect underpin all transactions with families and other professional service providers

ECO practice incorporates the principles of:

• Greater attention to developing practice skills and strengthening supportive structures to enable greater linkages and collaboration between statutory child protection and other service providers

• The articulation and ongoing development of child centred, family focused practice principles and strategies for use by the protective worker in their work with children, young people and their families.

Put simply, ECO provides workers with an opportunity to use a graded scale of assessment and investigation, tailoring responses to identified concerns, such that families where it appears that there is a generalised family dysfunction, rather than a child protection issue, are provided with a less intrusive response, possibly involving family support agency workers making an assessment, rather than having two child protection workers conduct an unannounced home visit.

In many ways ECO is a formalisation of practice as it was undertaken in the 1980s, recognising the expertise of other professionals and involving them collaboratively in case assessments and caseplanning, where appropriate. The current evaluation of the ECO approach indicates that the system has resulted in positive changes to gatekeeping for protective workers, non-government agencies and the families. Liz Armitage (DHS) will discuss the ECO approach in detail.

The ECO approach, like other UK and Australian child protection systems that have recently been developed, clearly acknowledges a reliance on effective risk assessment to assist decision making, and the adoption of child centred family focused practice to achieve positive change in families - these are now considered.

The investigation of the decision making of the various professionals who deal with child maltreatment cases has a history almost as long as the re-discovery of child abuse by Kempe and his colleagues (e.g. Delsordo, 1963; Boehm, 1964; Elmer, 1966). Researchers investigating child protection decision making have usually utilised one of two alternative methods, traditionally referred to as the statistical and clinical approaches (Wiggins, 1981; Ruscio, 1998). The statistical approach commonly consists of controlled experimental and quasi-experimental studies that result in the development of a statistical decision model which identifies the factors which account for the variance (or a proportion of the variance) in making a particular decision. It is argued that a statistical decision model provides greater accuracy (i.e. a better ‘hit rate’) and less judgement errors (i.e. ‘false positives’ and ‘false negatives’). It is not claimed that all decision errors are able to be eliminated by statistical modelling of decisions, rather that the ‘levels of accuracy are higher than we could achieve if we did not possess the risk assessment tool in question’ (Johnson, 1996:14).

Such experimental, logical positivist, decision modelling studies were initially used to determine the factors which influence decisions. In the 1980s and 1990s, most modelling studies have been designed to construct structured risk assessment scales designed to predict case outcomes for use in child protection practice.

In contrast, the clinical approach, is associated with a desire to develop causal explanations for decision making, involving ‘nothing more than a human judge evaluating available information and arriving at a decision.’ (Ruscio, 1998:145). The clinical approach generally utilises qualitative-descriptive methods of data collection to describe the decision making process, such as: self-report measures; behavioural observational techniques; case tracking; and the content analysis of case records. Such methods are ecologically valid and their flexibility enables their application to a variety of research questions. However, the generalisability of their results and their ability to empirically test cause and effect relationships are hampered by their lack of experimental control (Ruscio, 1998).

In an excellent conceptualisation of the realities of human decision making, Dalgleish (1997) notes that the relationships between the indicators or factors, the worker’s judgement and the actual reality of the family situation is ‘inherently probabilistic’.

He argues that statistical models focus on the ecological environment (to an extent - mainly as it relates to case factors!) and the reality of a family’s situation, while clinical judgement studies focus on the factors and influences affecting the decision maker, in this case, the child protection professional. The ‘irreducible uncertainty’ present in both actuarial and clinical models leads inevitably to some degree of error.

The assessment of risk is intrinsic to the child protection role; beginning with Kempe et al.’s (1962) paper, which discussed the decision to return an abused child to her/his family and the inherent risks involved:

Kempe and his colleagues also anticipated a time when a

In the early 1980s research utilising statistical approaches to child abuse decision making began to shift focus away from the identification of the factors influencing decisions, to develop ‘risk assessment’ models which would enable the prediction of future risk to children (Jones, 1996). In the 1990s risk assessment became the primary area for decision making research.

Risk assessment can be defined as

Risk assessment has several objectives: to help protective workers identify situations where children are at risk of maltreatment; to improve consistency in service delivery; and to help protective services determine the appropriate priorities within protective services caseloads (Browne & Saqi, 1988; English & Pecora, 1994). The instruments which have resulted, known as structured risk assessment measures, organise information related to risk (Schene, 1996). Specifically, they ‘comprise risk factors selected for assessment and forms designed to capture the procedures and calculations needed to determine risk’ (Saunders & Goddard, 1998:16).

Developed at a time when economic rationalism was beginning to have a significant effect on resources while the demand for services was increasing, the introduction of structured risk assessment measures was due partly to the need for services to screen out inappropriate reports, or cases where the maltreatment was suitable for a community-based caseplan without the involvement of protective services (Wald & Woolverton, 1990; Doueck, English, DePanfilis & Moore, 1993a; English & Pecora, 1994; Parton, 1996c; Tomison & McGurk, 1996; Saunders & Goddard, 1998).

Another motivation was to improve the ability of workers to detect high risk cases prior to the child suffering some form of injury. Child death inquiries in Britain and Australia have identified errors in worker judgements, their relative inexperience, lack of knowledge of risk factors and ‘danger signals’, failures in interagency coordination and communication and a lack of ‘rigour’ and ‘consistency’ in the management of cases (Jones et al., 1987; Armytage & Reeves, 1992; Howe, 1992).

In the U.S. child protection services had been criticised for ‘irrational decision-making, subjective and inconsistent investigations, ineffective interventions and inefficient resource allocation (Cicchinelli, 1995; English, 1996)’ (Saunders & Goddard, 1998:21). Child protection services had

Since 1987 at least 42 U.S. States have adopted some form of structured risk assessment system (English, 1996). In most of those states, statutes governing child protection services have meant that risk assessment procedures are only allowed to be used after a child has already been defined as a substantiated case of child maltreatment (English & Pecora, 1994). Risk assessment is generally used as a tool to determine the appropriate levels of service to provide to the child and family, based upon an assessment of severity of the maltreatment (English & Pecora, 1994).

The speedy adoption of structured risk assessment measures by U.S. child protection services has recently begun to be repeated in Australia. However, unlike in the U.S. where the use of risk measures has been somewhat restricted by legislation, some Australian State/Territory child protection services have explicitly or implicitly developed screening and/or risk assessment tools for use at intake, as well as at later stages of the child protection case management process (Tomison, 1996; Department of Family & Community Services, 1997; McPherson, Macnamara & Hemsworth, 1997).

When considering structured risk assessment systems in the context of the overall field of child protection decision making research, there are two important questions:

• Does structured risk assessment offer the means of enhancing professionals’ decision making in child protection cases?
• Is the decision making research focus on structured risk assessment systems misplaced?

When evaluating the utility of risk assessment tools, it is important to acknowledge that no decision model will be 100 per cent accurate (Dalgleish, 1997), and that this has real life consequences. Hammond (1996) proposes that depending on which errors are present in a decision, that such ‘inevitable error’ will lead to ‘unavoidable injustice’ for the various stakeholders in the decision, that is, the child, the family, the worker, the worker’s team, the agency, the local community and society as a whole. ‘No matter what [system] threshold is set, someone is going to suffer injustice’ (Dalgleish, 1997:10).

However, it has been generally recognised that the use of statistical or actuarial procedures to inform judgements is more accurate than the reliance on unassisted clinical judgment (Dawes et al., 1989), and that

English and Pecora note that

In contrast, Schene (1996) contends that,

Similarly, Wald and Woolverton caution that structured risk assessment measures should not be seen as a panacea for an ailing child protection system:

DePanfilis (1996), utilising the work of Cicchinelli (1990) and Hornby (1989), identified some of the potential benefits of risk assessment procedures. These included:

• good casework practice in operation;
• providing a base for the allocation and prioritisation of cases in risk-related groups;
• having case information readily available via the case record;
• broadening workers’ knowledge and investigation of relevant child maltreatment ‘risk’ factors; and,
• providing a basis for worker training and supervision.

In addition, because structured risk assessment measures are designed to promote consistency in worker decision making and subsequent service provision (English & Pecora, 1994; Schene, 1996), the instruments can reduce arbitrary case classification and management practices (Stone, 1993).

Finally, risk assessment measures were developed at a time when the demand for services had increased as resource allocation remained constant or had decreased. In consequence not all children or families in need were able to be provided with the services they might require (Wald & Woolverton, 1990; English & Pecora, 1994; Parton, 1996c; Tomison & McGurk, 1996; Saunders & Goddard, 1998). Concomitantly, there was increasing demand for highly accountable service provision (Doueck et al., 1993a; Parton, 1996c). Risk assessment measures could provide ‘a pseudo-scientific, ostensibly rational basis for decision-making’ (Saunders & Goddard, 1998:22).

The overall efficacy of structured risk assessment measures however, is reduced for a number of reasons, leading Goddard et al. to conclude that overall, the

There is little unanimity in terms of defining fundamental terms in the field of child maltreatment (e.g. What is ‘maltreatment’? What is ‘risk’?), (Hutchison, 1990; Zuravin, 1991; Lyons, Doueck & Wodarski, 1996).

Being ‘at risk’ is not an objective state, but a complex, multidimensional concept that is both socially and professionally constructed and whose meaning has evolved over time (Freeman, 1983; Douglas, 1992; Parton, 1996c; Ryan, 1996). It has evolved from a

There remains a clear need to develop a uniform means of quantifying the levels of risk and to establish clear parameters for the appropriate actions to be taken with each level of risk and type of maltreatment (Wald & Woolverton, 1990).

Bateson (1979) identifies one problem with the application of structured decision making systems to child protection case management. He argues that throughout history there has been a constant tension in the use of probability-based statistics to describe individual behaviour based on the behaviour of groups of individuals. Actuarial tools are derived from statistical generalisations believed to be predictive of the behaviour of a group of like individuals. In essence, it is invalid to argue that the risk to an individual child can be predicted from an analysis of variables drawn from aggregated data for groups of children with common characteristics:

Thus, in spite of some general similarities which may be evident between ‘at risk’ or maltreating families, such measures may be of limited utility in child protection work where the task is to predict the behaviour of particular parents and particular children with unique circumstances (Saunders & Goddard, 1998).

Second, Lyons et al. (1996) reviewed the published, empirical literature on 10 risk assessment models currently employed by various U.S. state child protection services, including the CARF system, the Illinois CANTS 17B and a Philadelphia derivative, the Alaska model, the Utah family assessment model, the Washington State Risk Assessment Model (WARM), the Child Wellbeing (CWBS) (Magura & Moses, 1986), and the New York Child Protective Services Review Document (CPRSD). Two additional models, the Alameda County model and the Vermont model, were included in the analysis, despite their evaluations not having been completed by independent evaluators.

They focused their examination on the psychometric qualities of the risk measures, reliability and validity in particular, and the outcomes which resulted from the implementation of the risk models.

Overall, despite many of the models appearing to have acceptable psychometric properties, such as internal consistency, interrater reliability and concurrent validity, the current level of predictive validity for the models

Finally, Lyons et al. note that the evaluation research, published to date, on process evaluation or implementation research is ‘less than adequate’. The available evidence does, however, suggest that: risk assessment models are being imperfectly implemented; that the measures are being completed after the assessment decisions have been made, merely as a means of documenting decisions, rather than as a guide to the decision making process itself; and finally, that they are perceived by some workers as irrelevant to their work with both ‘at risk’ and maltreating families.

Mandel, Lehman and Yuille (1995) contend that statistical models of decision making, such as structured risk assessment systems, fail to take into account implementation issues. That is, the failure of agencies to train their workers to utilise risk factors identified by a particular statistical model, or to routinely collect data on the risk factors. Similarly, most evaluations of risk tools are outcome evaluations which fail to determine the extent to which a tool has been implemented in practice, that is, there is no process evaluation (Doueck et al., 1992). Adequate evaluation requires a focus on the degree to which a risk assessment model is being used as intended and the impact of the model on the case management process, including outcomes for families.

Failure to conduct both process and outcome evaluations may lead to what Scanlon and colleagues (Scanlon, Horst, Schmidt & Walker, 1977; as cited in Doueck et al., 1992) called a Type III error - the evaluation of a program that has been inadequately implemented. Lyons et al. conclude that minimising ‘implementation problems may be as important for model development as concern about sensitivity and specificity currently is’ (1996:153).

Why are risk assessment models inadequately implemented, even under ‘unusually favorable’ conditions, such as was reported for the implementation of the CARF system (Doueck et al, 93b)?

A number of reasons have been proposed:

• there may be problems integrating the risk model into case management practice (Gleason, 1984; Dalgleish, 1997);
• the complicated nature of some models;

• because of the perception of staff that the measures will increase their workload (Hornby & Wells, 1989; as cited in Doueck et al., 1993b; English & Pecora, 1994; Sheets, 1996).

Dalgleish (1997) provides a concise summary of the major methodological issues surrounding and affecting the implementation and evaluation of risk assessment tools:

• conceptualisation and measurement of outcome - there is no single set of outcomes in child protection; current actuarial measures utilise a variety of outcome measures in order to determine the weighting to be applied to various predictive factors. Hence, construct validity is currently less than optimal;

• conceptualisation and measurement of judgement - risk is a complex concept interpreted and utilised in a variety of ways; such a concept is unlikely to be captured by a single numerical value;

• conceptualisation and measurement of the indicators - a specific, relevant set of indicators needs to be developed to address a particular judgement. For example, the indicators developed to inform the decision to substantiate a case will be different from those required when the decision is to remove a child from her/his family;

• representativeness of design for the estimation of weights - the weights applied to a set of indicators are optimal only for a particular outcome, a particular set of indicators and a particular sample; the weights will not generalise from the specific situation for which they are created;

As Doueck et al. concluded in their evaluation of the CARF system, generalisation ‘to other counties with a different mix of populations and different caseloads is particularly hazardous’ (1993b:465). Similarly, there is no reason to believe that the factors which predict the occurrence of child maltreatment will also predict re-abuse. McDonald and Marks (1991) noted that only half of the 88 variables commonly used to predict child maltreatment had been empirically validated, and that there is ‘virtually no research on the correlates of maltreatment recurrence for any type of maltreatment’ (Lyons et al., 1996:144).

• prospective and retrospective validation - at present some studies involve the determination of indicator weights by conducting retrospective analyses of case data, other researchers have developed prospective studies as a means of ascertaining their measure’s ability to accurately discriminate between cases (discriminant validity). The multiple approaches further muddy attempts to compare and contrast measures, and to determine their effects on outcome.

Part of the appeal of risk assessment measures is that they are designed to ensure that workers give consideration to a wide range of factors in a relatively consistent manner (Corby, 1996). By way of illustration, Daro’s (1988) comprehensive guide to the unitary factors which are thought to contribute to child maltreatment, many of which are typically included in risk assessment measures has been provided in Figure 1. [N.B. The list has been further categorised by myself, according to Belsky’s four interactive levels.].

However, the selection of risk factors is typically based on reported cases (Milner, 1995) and retrospective research (e.g. archival analysis); such research is plagued by problems associated with the use of secondary data (Doueck et al., 1992; de Vaus, 1995; Krysik, 1997). If risk assessment is to have validity, ‘it is essential that risk factors are measured accurately’ (Wald & Woolverton, 1990:490), yet it is clear that the level of accuracy required is not always forthcoming.

In addition, no checklist or model can include every possible risk factor; it is possible therefore, that a significant factor, or the significant factor for a specific case may be omitted (Saunders & Goddard, 1998). Similarly, although specific factors, or the combination of a number of factors, may be important in a case, it is the interaction of factors, or ‘volatile combinations’ (Holder & Corey, 1993), that may especially endanger a child. To date however, ‘nobody is able to point out which interaction of factors makes a difference’ (Wald & Woolverton, 1990:495).

Finally, in addition to the various risk factors, operating in isolation or in combination, which may increase the likelihood or risk of maltreatment, unpredictable ‘triggering events’, often significant only to the maltreater (e.g. accidental breakage of crockery, a child returning home with muddy clothes), may determine a child’s safety (Pullan-Watkins & Durrant, 1996).

A critical element frequently ignored when risk assessment is under discussion, is the effect of the decision environment on the use and implementation of the measures. Scott (1993) perceives the child protection field as operating in, what has been termed by the organisational theorists Emery and Trist (1965), a ‘turbulent field’. Child protection by its very nature is ‘dominated by significant moral, emotional and socio-political turbulence’ (McPherson et al., 1997:27).

Under legislation, the aim of statutory protective services is not merely to stop the re-abuse of children, policy dictates that the goal of intervention is to try to prevent maltreatment, while keeping families together, where possible (Wald & Woolverton, 1990; Corby, 1996). Thus, risk assessment, like any case management decision, is not carried out in a ‘scientific vacuum’, but within a ‘socio-legal environment which defines preferred courses of action’ (Corby, 1996:23).

However, because the State is ‘highly ambivalent about setting parameters for intervention into families, [the result is] fluctuating approaches and policy reversals’ (Corby, 1996:14).

Thus, when a child is removed from the family, in most cases the objective is to work for child and family reunification. Risk tools must therefore be able to take account of the effect of particular treatment plans, for example, if a parent attends drug rehabilitation, what is the potential for re-abuse? The question then becomes, under what circumstances is the risk of further maltreatment unlikely?

As Wald & Woolverton note, since

Statistical models fail to take into account the other systemic or organisational factors which may affect decision making, or to allow for workers’ individual differences in decision making. Even if it is assumed that risk can be objectified, individuals respond subjectively, and thus, differently to the same events (Brearley, 1982). Studies of economic decision making support the view that risk taking is not fixed, but a dynamic process dependent on the context of choice and the extent of adversity (Waterhouse & Carnie, 1992).

The model in Figure 2, developed from my doctoral research, I believe provides a more accurate picture of the realities of the influences on decision making. Based on an ecological framework, it takes into account the effects of systemic and professional-related factors, in interaction with aspects of the family system.

A potential benefit of risk assessment measures is their potential to homogenise different levels of practice expertise and qualifications and the possibility of a reduction of the impact of worker idiosyncrasies or biases in decision making (Stone, 1993). Jones and May contend that standardised procedures also ‘reduces the decision options of front-line workers, defines the boundaries of their work [and] minimises [their] discretion’ (1992:491).

Without good quality control and worker supervision, the system can be used to support potentially poor decisions (Doueck et al., 1993a). Workers, particularly the inexperienced, may be lulled into a false sense of security, believing they can reliably predict case outcomes by using risk assessment tools, and thus only focus on the checklist factors when making a case assessment (Reder, Duncan & Gray, 1993). That is, such lists may constrain worker thinking, leading to a heuristical ‘check’ of the variables listed, rather than a comprehensive assessment of all case information and the worker’s own professional observations (Reder et al., 1993; Goddard, 1996a).

Even if it is assumed that all the significant factors for a particular case are included on the risk measure, the information still requires ordering and its significance must be recognised (Goddard, 1996a).

As Sheets notes

Lyons et al. (1996), in their review of the published, empirical literature on 10 risk assessment models currently in operation in various U.S. state child protection services, concluded that the results from the models’ implementations suggested that risk assessment actually demands ‘quality in education, training, and supervision, as well as vigilance on the part of administrators hoping to use it’ (Lyons et al., 1996:154). Thus, there is an obligation upon the test user to have adequate clinical and legal education before attempting to make a child abuse assessment (Monahan, 1993).

Overall, there has been a concerted effort in recent times to focus much of the research investigating aspects of child protection decision making on statistical approaches, and specifically, the development and enhancement of structured risk assessment systems. It is apparent however, that there is currently insufficient information available to determine the efficacy of risk assessment tools for identifying children at risk of serious maltreatment (Wald & Woolverton, 1990; Camasso & Jagannathan, 1995; Lyons et al., 1996; Dalgleish, 1997, Cleaver et al., 1998; Saunders & Goddard, 1998).

While it is generally recognised that the use of statistical or actuarial procedures to inform clinical judgements is more accurate than the reliance on unassisted clinical judgment (Dawes et al., 1989), there has been very little actual evidence derived from the child welfare/child protection field on the extent to which statistical models offer improvements in consistency and accuracy beyond that of child protection workers’ clinical judgements.

Thus, the argument that risk assessment tools (statistical models) should replace clinical judgement is a difficult one to make (Johnson, 1996; Ruscio, 1998). Given the limitations of applying risk measures to individual cases, workers’ professional judgement is needed to fill the gap (Saunders & Goddard, 1998).

Many researchers and practitioners, although acknowledging the importance of assessing risk in child protection practice (Saunders & Goddard, 1998), have therefore maintained an allegiance to clinical decision making research and the use of education and training to improve professionals’ child protection decision making rather than the use of structured decision making tools. This position is aptly summed up by Dr. Pat Cawson, Head of Child Protection Research for the NSPCC, when introducing the new NSPCC risk assessment package in the U.K. (Cleaver et al., 1998):

In spite of their limitations, however, the use of risk assessment systems is becoming more and more common in child protection services in the U.S., Australia and other western countries as governments and bureaucracies continue to see them as a solution when attempting to balance high caseloads, less resources and the need for service accountability (Browne & Saqi, 1988; Wald & Woolverton, 1990; Doueck et al., 1993a; English & Pecora, 1994; Parton, 1996c; Tomison & McGurk, 1996; Saunders & Goddard, 1998). Concomitantly, workers’ clinical experience and intuition is increasingly being undermined (Saunders & Goddard, 1998).

Radical proponents of clinical approaches to risk assessment, that is, a reliance on professional expertise (based on specialist training and experience) argue against the use of any checklist, or guide to risk assessment and child protection decision making. However, the realities of current child protection work are that few workers have received extensive post-qualifying training in child maltreatment and child protection research and theory. What studies that have been done suggest, is that workers do not rely on research and theory to make decisions. In actuality, they appear to be suspicious of research and theory and ignore it (Preston-Shoot & Agass, 1990), remain unaware of it (Stevenson, 1992; Farmer & Owen, 1995), or have few opportunities to acquire it (Carew, 1979; Preston-Shoot & Agass, 1990; Fisher, 1997). Thus, they lack the requisite information and a framework for the organisation of such information.

Field experience, while invaluable, will not provide the grounding in theory and current thinking is assessment and decision making which, it is contended, should be mandatory requirements. In New South Wales new options to access this information are currently being developed, such as the Graduate Diploma in Child Protection Studies at Charles Sturt University, and the current development of a Masters in Child Protection at the University of Western Sydney. However, in general, it is apparent that most workers need some form of additional assistance to inform their risk assessment and decision making.

One approach to this issue, is to support practice experience (and the concomitant provision of supervision), with a guide to inform practice. Such guides are not meant to be used to make the decisions, but to highlight issues for consideration and to provide a framework for conceptualising and justifying a decision.

Although researchers in the United Kingdom have investigated various means of improving the prediction or assessment of the likelihood of future maltreatment (e.g. Greenland, 1987; Browne & Saqi, 1988), Britain has generally been more hesitant that either the United States or Australia in adopting structured risk assessment instruments. In 1988 the U.K. DoH introduced ‘guidelines of extraordinary detail’ (Howe, 1992:501), known as the ‘Orange Book’, which was over 90 pages long and comprised of 167 questions. The Orange Book was developed to enhance the quality of assessments made by social workers and to assist in the identification and prediction of risk of future harm to a child (Howe, 1992). However, the emphasis was on ensuring a thorough assessment was completed, rather than risk assessment per se.

In 1998 the NSPCC published Assessing Risk in Child Protection (Cleaver, Wattam, & Cawson, 1998), a report of a research and development project commissioned by the DoH. The project was designed to feed results of research into risk assessment into social work practice, making it accessible to social workers, trainers and students (Cleaver et al., 1998).

The resultant package includes a brief review of salient aspects of the research literature, a discussion of the findings which resulted from a study of workers’ perceptions of assessing risk (based on a series of interviews with workers), and an outline of the subsequent development and refinement of two ‘decision aids’. These were: a chart to assist with the recording of essential information at referral; and a data book to summarise some basic findings to assist with the identification of the full range of services required by the child(ren) and family. They were designed to

Another aid, a ‘manual’ which could give specialist knowledge on topics such as parental mental illness, substance abuse, disabilities and domestic violence and their relationship to child maltreatment and child protection was met with considerable enthusiasm, but was beyond the remit of the project (Cleaver et al., 1998).

The assumption underlying the research and the resultant package was that evidence-based practice, not the use of structured risk assessment measures, should form the basis of effective child protection and child welfare practice. Specifically, that there is no substitute for rigorous observation and assessment of the child and family, and the subsequent development of an action plan which can deal effectively with the child maltreatment concerns and other family issues which may increase the likelihood of maltreatment in the future.

Such an approach would appear to provide some of the uniformity and rigour of structured risk assessment measures, providing a useful guide to some of the salient factors (risk and protective) which may affect an assessment of risk, in conjunction with an overall clinical approach. Such a ‘collaborative path’ has been supported by Webster & Cox (1997), in a discussion of the use of risk assessment when making decisions as to the mental state or ‘dangerousness’ of particular individuals. They argue for a system where clinicians engage in practice supported by statistical evidence (e.g. risk assessment tools), where possible, but that statistical approaches are not seen as almighty without reference to clinicians’ reality (‘the overgeneralization of research findings without due heed to case particulars is inappropriate and misleading’ [Stricker & Trierweiler, 1995:997]). ‘Nomothetic material and ideographic material must be approximately balanced and alternated so as to avoid undue reliance on one kind or the other’ (Webster & Cox, 1997: 1246).

This gets back to the idea of using structured assessment materials as guides to practice, however they should not be limited to case-related factors (see Cleaver et al., 1998). A generic (across maltreatment types), matrix approach to risk assessment was introduced into Victorian statutory child protection services two years ago (McPherson et al., 1997) in order to guide risk assessment. Training was also provided to staff from other agencies frequently involved in child protection cases, such as the police, sexual assault counsellors, and social workers from a variety of family support agencies. The intention was to develop a common conceptual framework to aid the assessment and decision making of various professionals who had some involvement in the management of cases of children at risk of child maltreatment. It was hoped that this would result in the creation of more ‘consistent decision making across workers and, with the same worker across cases’ (McPherson et al., 1997:22). The development of the model is ongoing, with Department of Human Services staff in the process of releasing an upgraded risk assessment tool in the next 12 months.

Although gatekeeping and risk assessment may not require a strong therapeutic stance, the development of effective child protection practice and the prevention of maltreatment (and thus, the reduction in family dysfunction and the number of notifications made) is reliant on the effective support of families in need.

The second component of the new models of service delivery is the adoption of child centred, family focused practice principles. These principles affirm the primary importance of ensuring the safety and wellbeing of children; recognise the mutual significance of the child and family to each other; and, promote the importance of service professionals developing a strengths based partnership with client families (Tomison, Burgell & Burgell, 1998).

Pioneered in the early 1960s by Otto (eg. 1963, as cited in DePanfilis & Wilson, 1997), the underlying tenet of a strengths perspective is that all families have strengths and capabilities (De Jong & Miller 1995). If practitioners take the time to identify and build on these qualities, rather than focusing on the correction of skills deficits or weaknesses, families are more likely to respond favourably to interventions and thus the likelihood of making a positive impact on the family unit is considerably enhanced (Dunst, Trivette & Deal, 1988).

The practice philosophy of a strength-based approach, particularly as it is applied to ‘at risk’ and abusive populations (Saleebey, 1992, as cited in De Jong & Miller, 1995), can be summarised as:

• all people and environments possess strengths that can be marshalled to improve the quality of clients’ lives; these strengths and the ways in which clients choose to apply them should be respected by workers;
• client motivation is fostered by a continued emphasis on client-defined strengths;
• discovering strengths requires a cooperative exploration between clients and workers;
• a focus on strengths reduces the worker temptation to ‘blame the victim’ and enables the discovery of the means by which clients have survived in even the most inhospitable of circumstances; and,
• all environments, even the most bleak, contain resources.

A focus on the positive aspects of family functioning does not imply that family problems and/or the protection of the child are forgotten. The child centred family focused philosophy ensures that the protection and care of the child remain paramount, while maintaining a focus on building family members’ competence and self-esteem in order to tackle protective concerns and other family issues effectively.

The objective is to develop a true partnership between family members and workers, involving the family as much as possible in case management decision-making and encouraging families both to set their own goals and to take responsibility for achieving them. Such competency-based, family-centred practice is not a denial of a family’s problems or shortcomings but a focus on client strengths is perceived to be a more fruitful means to address issues and achieve positive change.

Giving more credence to clients as participants in the helping process increases the likelihood of a worker facilitating positive change in the lives of family members, and helps to ensure the safety of the child through better engagement with the family. The approach is not a cure-all, rather, it is designed to assist families to achieve a level of functioning above the threshold for protective intervention.

Some researchers and practitioners however, have raised concerns that child centred family focused, and solution focused work may not adequately protect children, because they focus too much on the family as a whole rather than being primarily concerned with the protection of the child. The fear is that the primary role of the assessment and intervention process, ensuring a children’s safety, may not remain the priority in a system focused on family support as a whole. It is feared that with generic family support cases in particular, family needs will be addressed at the expense of child protection. The shift to ‘family support’ is thus perceived by some professionals as representing a shift towards family preservation, even in cases where this may be detrimental to the child. It is therefore vital that for families entering the child welfare system as a result of child maltreatment allegations, care is taken to ensure that the best interests of the child are not sacrificed when they conflict with the best interests of the family.

Acknowledging the realities of current child protection practice, this paper has provided an overview of some of the issues in the identification, assessment and management of suspected child maltreatment cases. It is contended that system restructuring, based predominantly on the UK experience and Messages from Research, has involved the implementation of changes to case screening or gatekeeping, and the acknowledgment of the requirement to adequately address family needs, rather than focusing merely on significant harm. However, for the most part, these changes have been translated into better identification and assessment practices, and the development of new risk assessment tools or guides. Few child protection systems, including the UK Social Service Departments (Parton, 1997) have enacted policies to develop better partnerships between workers and families, or to ensure all families are actually assisted to remedy their problems, whether they be child protection matters, or generalised family dysfunction.

Overall, it could be argued that the role of child protection services is to only deal with families where there are significant issues of child maltreatment or a risk of maltreatment. Generalised family support needs are the prevail of the child welfare and family support system. While this demarcation may be appropriate, the substantial increase in the number of reports to child protection services experienced across the nation in the past few years has seen the bulk of the resources allocated for family support and child welfare allocated to the child protection system at the expense of child abuse prevention. As a result, many ‘at risk’ families, or those with more general social problems have struggled to obtain assistance through either system.

The question may then be, what is the point of assessing risk more effectively, if as Corby has noted, ‘the resources needed to achieve risk reduction are not available’ (1996:27)?

Most children at severe risk of maltreatment, or who are being seriously maltreated, are already known to statutory agencies. Gatekeeping while perceived as inappropriately including cases where there are minimal maltreatment concerns, but some form of generalised family problem, are quite successful at identifying children at serious risk.

For example, although they indicate a failure to adequately assess the level of danger to a child at a particular time, or issues of caseplanning and interagency communication, the vast majority of child deaths were known to statutory child protection services.

It is therefore concluded that for changes to identification and assessment to have meaning for children and their families, there must be changes in the conceptualisation of the roles of child protection and the wider child welfare and family support systems. Focusing on minor adjustments to the ‘intervention pendulum’ without adequate resourcing of the system will significantly reduce any possible benefits for children and families identified by the system, particularly those with generic welfare concerns. The ‘system’ needs to be conceptualised as a prevention-protection ‘continuum of action’, where regardless of the level of protective concerns, children and families receive some form of support to alleviate their concerns. The threshold is then less important than ensuring the protection of children within a process of family support, of true child centred family focused work.

Such a framework will only result if governments and senior departmental managements adequately recognise the importance of preventing social ills, particularly child maltreatment and the social and economic costs which can be saved. Under such an approach, identification and assessment, while important, truly no longer drive the child protection system, but maintain an important position in a model where remedying dysfunction is given primacy. Although it is unlikely that this framework will be adopted in the short-term, I would propose that the ‘continuum of action’ be retained as a benchmark against which future re-structuring and innovation in practice are measured.

Tomison, A.M. (1999). ‘Ensuring the protection of children: The role of child protection services in the identification, assessment and treatment of maltreated children’, invited keynote presentation, ‘Entry parameters to child and family services workshop’, NSW Dept of Community Services, Sydney, 22-23 March, 1999.

Armytage, P. & Reeves, C. (1992). Practice insights as revealed by child death inquiries in Victoria and overseas. In G. Calvert, A. Ford & P. Parkinson (Eds.), The practice of child protection: Australian approaches (pp.122-140). Marrickville: Hale & Iremonger.

Armytage, P., Boffa, J. & Armitage, E. (1998). Professional practice frameworks: Linking prevention, support and protection. Paper presented at the Twelfth International ISPCAN Congress on Child Abuse and Neglect, ‘Protecting Children: Innovation and Inspiration’, Auckland, New Zealand, September 6-9, 1998.

Bateson, G. (1979). Mind and Nature: A necessary unity. New York: Bantam Books.

Boehm, B. (1964). The community and the social agency define neglect. Child Welfare, 43, 453-464.

Brearley, C.P. (1982). Risk and Social Work: Hazards and helping. London: Routledge & Kegan Paul.

Browne, K.D. & Saqi, S. (1988). Approaches to screening for child abuse and neglect. In K. Browne, C. Davies & P. Stratton, (Eds.), Early prediction and prevention of child abuse (pp.57-85). Chichester: John Wiley & Sons.

Camasso, M.J. & Jagannathan, R. (1995). Prediction accuracy of the Washington and Illinois risk assessment instruments: An application of receiver operating characteristic curve analysis. Social Work Research, 19, 174-183.

Carew, R. (1979). The place of knowledge in social work activity. British Journal of Social Work, 9(3), 349-364.

Cicchinelli, L.F. (1990). Risk assessment: Expectations, benefits, and realities. In T. Tatara (ed.), Fourth national roundtable on CPS risk assessment summary of highlights (pp.7-22). Washington D.C.: American Public Welfare Association.

Cicchinelli, L.F. (1995). Risk assessment: Expectations and realities. The APSAC Advisor, 8(4), 3-9.

Clark, R. (1995). Child protection services in Victoria. Family Matters, 40, Autumn, 22-23.

Cleaver, H., Wattam, C. & Cawson, P. (1998). Assessing risk in child protection. NSPCC Policy Practice Research Series. London: NSPCC.

Corby, B. (1996). Risk assessment in child protection work. In H.Kemshall & J. Pritchard (Eds.) Good practice in risk assessment and risk management. (pp.13-30). Good Practice Series 3. London: Jessica Kingsley.

Dale, P. (1998). Protection or prevention? An integrated approach to services and parents. Paper presented to the Kids First: Protecting Children - Problems & Solutions Conference, Melbourne, 2-3 April, 1998.

Dalgleish, L.I. (1997). Risk assessment approaches: The good, the bad and the ugly. Paper presented to the 6th Australasian Conference on Child Abuse and Neglect, Adelaide, 20-24 October, 1997.

Daro, D. (1988). Confronting child abuse: Research for effective program design. New York: Free Press.

Dartington Social Research Unit (1995). Child protection: Messages from research. Studies in Child Protection. London: HMSO.

Dawes, R.M., Faust, D. & Meehl, P.E. (1989). Clinical versus actuarial judgement. Science, 243, 1668-1674.

De Jong, P. & Miller, S.D. (1995). How to interview for client strengths. Social Work, 40(6), 729—36.

Dunst, C.J., Trivette, C.M. and Deal, A.G. (1988). Enabling and empowering families: Principles and guidelines for practice. Cambridge, MA: Brookline Books.

Delsordo, J.D. (1963). Protective casework for abused children. Children, November-December, 6-51.

DePanfilis, D. (1996). Implementing child mistreatment risk assessment systems: Lessons from theory. Administration in Social Work, 20(2), 41-59.

DePanfilis, D. & Wilson, C. (1997). Child protective services: applying the strengths perspective with maltreating families. The APSAC Advisor, 9(3), 15—20.

Department of Family & Community Services. (1997). Child protection in South Australia: A new approach. Adelaide: Department of Family & Community Services.

de Vaus, D. (1995). Surveys in social research (4th ed.). St Leonards, NSW: Allen & Unwin.

Doueck, H.J., Bronson, D.E. & Levine, M. (1992). Evaluating risk assessment implementation in child protection: Issues for consideration. Child Abuse & Neglect, 16, 637-646.

Doueck, H.J., English, D.J., DePanfilis, D. & Moore, G.T. (1993a). Decision-making in child protective services: A comparison of selected risk assessment systems. Child Welfare, 72(5), 441-452.

Doueck, H.J., Levine, M. & Bronson, D.E. (1993b). Risk assessment in child protective services: An evaluation of the Child at Risk Field System. Journal of Interpersonal Violence, 8(4), 446-467.

Douglas, M. (1992). Risk and blame essays in cultural theory. London: Routledge.

Elmer, E. (1966). Hazards in determining child abuse. Child Welfare, 3, 29-51.

English, D.J. (1996). The promise and reality of risk assessment. Protecting Children, 12(2), 9-13.

English, D.J. & Pecora, P.J. (1994). Risk assessment as a practice method in child protective services. Child Welfare, 73(5), 451-473.

Farmer, E. & Owen, M. (1995). Child protection practice: Private risks and public remedies. Studies in child protection. London: HMSO.

Fisher, T. (1997). Learning about child protection. Social Work Education, 16(2), 92-112.

Freeman, M. (1983). The rights and wrongs of children. London: Pinter.

Gleason, J.P. (1984). The use of structured decision making procedures at child welfare intake. Unpublished doctoral thesis. University of Illinois, Chicago.

Goddard, C.R. (1988). Victoria's protective services: Dual tracks and double standards. Melbourne:Victorian Society for the Prevention of Child Abuse and Neglect.

Goddard, C.R. (1996a). Child abuse and child protection: A guide for health, education and welfare workers. South Melbourne: Churchill Livingstone.

Goddard, C.R. (1996b). Read all about it! The news about child abuse. Child Abuse Review, 5, 301—309.

Goddard, C.R. & Liddell, M.J. (1995). Child abuse fatalities and the media. Child Abuse Review, 4, 365-364.

Goddard, C.R., Tucci, J., Liddell, M., Hiller, P. & O’Connell, J. (1996). A provisional analysis of decision-making by statutory child protection workers in a rural context. Melbourne: Child Abuse and Family Violence Research Unit, Monash University.

Holder, W.M. & Corey, M. (1993). Child protective services risk management: A decision making handbook (revised ed.). Charlotte, NC.: ACTION for Child Protection.

Howe, D. (1992). Child abuse and the bureaucratisation of social work. Sociological Review, 40(3), 491-508.

Hammond, K.R. (1996). Human judgment and social policy: Irreducible uncertainty, inevitable error, unavoidable injustice. New York: Oxford University Press.

Hornby, H. (1989). Risk assessment in child protective services in field implementation. Portland, ME: National Child Welfare Resource Centre for Management and Administration.

Hutchison, E.D. (1990). Child maltreatment: Can it be defined? Social Services Review, 64, 60-78.

Johnson, W. (1996). Risk assessment research: Progress and future directions. Protecting Children, 12(2), 14-19.

Jones, A. & May, J. (1992). Working in human service organisations: A critical introduction. Melbourne: Longman Cheshire.

Jones, D.N., Pickett, J., Oates, M.R. & Barbor, P.R.H. (1987). Understanding child abuse (2nd ed.). Basingstoke: Macmillan.

Jones, R. (1996). Decision making in child protection. British Journal of Social Work, 26, 509-522.

Kempe, R. S., Silverman, F. N., Steele, B. F., Droegemuller, W. & Silver, H. K. (1962). The battered child syndrome. Journal of the American Medical Association, 18(1), 17—24.

Krysik, J. (1997). Secondary analysis. In R.M. Grinnell Jr., Social work research and evaluation: Quantitative and qualitative approaches (5th Ed.), (pp.391-406). Ithaca, Ill: F.E. Peacock.

Little, M. (1995). Child protection or family support? Finding a balance. Family Matters, 40, 18—21.

Loane, S. (1997). The view from the other side: a journalist’s perspective. Child Abuse Review, 6, 55—59.

Lyons, P., Doueck, H.J. & Wodarski, J.S. (1996). Risk assessment for child protective services: A review of the empirical literature on instrument performance. Social Work Research, 20(3), 143-155.

Mackieson, P. (1997). Brimbank family outreach program: A research report. Melbourne: Department of Human Services (Western Metropolitan Region).

Magura, S. & Moses, B.S. (1986). Outcome measures for child social services: Theory and applications. Washington D.C.: Child Welfare League of America.

Mandel, D.R., Lehman, D.R. & Yuille, J.C. (1995). Reasoning about the removal of a child from home: A comparison of police officers and social workers. Journal of Applied Psychology, 25, 906-921.

Markiewicz, A. (1996). The child welfare system in Victoria: Changing context and perspectives 1945-1995. Children Australia, 21(3), 32-41.

McPherson, L., Macnamara, N. & Hemsworth, C. (1997). A model for multi-disciplinary collaboration in child protection. Children Australia, 21(1), 21-28.

Milner, J.S. (1995). Physical child abuse assessment: Perpetrator evaluation. In J. Campbell (Ed.), Assessing dangerousness (pp.41-67). Thousand Oaks: Sage.

Monahan, J. (1993). Limiting therapist exposure to Tarasoff liability: Guidelines for risk assessment. American Psychologist, 48, 242-250.

Parton, N. (1996c). Social work, risk and the ‘blaming system’. In N. Parton (Ed.), Social theory, social change and social work. (pp.98-114). London: Routledge.

Parton, N. (1997). Child protection and family support: Current debates and future prospects. In N.Parton (Ed.), Child protection and family support: Tensions, contradictions and possibilities (pp.1-24). The State of welfare series. London: Routledge.

Preston-Shoot, M. & Agass, D. (1990). Making sense of social work: Psychodynamics, systems and practice. Houndsmills, Basingstoke: Macmillan Education.

Pullan-Watkins, D. & Durrant, L. (1996). Working with children and families affected by substance abuse: A guide for early childhood education and human service staff. New York: The Centre for Applied Research Education.

Rayner, M. (1995). Children’s rights in Australia. In B. Franklin (Ed.), The handbook of children’s rights: Comparative policy and practice (pp.188-200). London: Routledge.

Reder, P. Duncan, S. & Gray, M. (1993). Beyond blame: Child abuse tragedies revisited. London: Routledge.

Ruscio, J. (1998). Information integration in child welfare cases: An introduction to statistical decision making. Child Maltreatment, 3(2), 143-156.

Saunders, B. & Goddard, C.R. (1998). A critique of structured risk assessment procedures: Instruments of abuse? Melbourne: Child Abuse & Family Violence Research Unit, Monash University.

Schene, P. (1996). The risk assessment roundtables: A ten-year perspective. Protecting Children, 12(2), 4-8.

Scott, D. (1995). Child protection: Paradoxes of publicity, policy and practice. Australian Journal of Social Issues, 30(1), 71—94.

Sheets, D.A. (1996). Caseworkers, computers and risk assessment: a promising partnership. The APSAC Advisor, 9(1), 7-12.

Stevenson, O. (1992). Social work intervention to protect children: Aspects of research and practice. Child Abuse Review, 1, 19-32.

Stone, M. (1993). Child Protection: A model for risk assessment in physical abuse/neglect. Surrey: Surrey County Council.

Stricker, G. & Trierweiler, S.J. (1995). The local clinical scientitst: A bridge between science and practice. American Psychologist, 50, 995-1002.

Tomison, A.M. (1996). Child protection towards 2000: Commentary. Child Abuse Prevention, 4 (2), 1—3.

Tomison, A.M., Burgell, R. & Burgell, D. (1998). An evaluation of the Brimbank Family Outreach Services. Melbourne: Department of Human Services.

Tomison, A.M. & McGurk, H. (1996). Preventing child abuse: A discussion paper for the South Australian Department of Family and Community Services. Melbourne: Australian Institute of Family Studies.

Wald, M.S. & Woolverton, M. (1990). Risk assessment: The emperor’s new clothes? Child Welfare, 64 (6),483-511.

Waterhouse, L. & Carnie, J. (1992). Assessing child protection risk. British Journal of Social Work, 22, 47-60.

Webster, C.D. & Cox, D. (1997). Integration of nonothetic and ideographic positions in risk assessment: Implications for practice and the education of psychologists and other mental health professionals. American Psychologist, 52, 1245-46.

Wiggins, J.S. (1981). Clinical and statistical prediction: Where are we and where do we go from here? Clinical Psychology Review, 1, 3-18.

Zuravin, S.J. (1991). Research definitions of physical child abuse and neglect: Current problems. In R.H. Starr & D.A. Wolfe (Eds.), The effects of child abuse and neglect. (pp.100-128). New York: Guildford Press.

Return to Papers and presentations

AIFS Home

Email queries to webmaster@aifs.org.au | Copyright, privacy, disclaimer | Site map Australian Institute of Family Studies, Level 20, 485 La Trobe Street, Melbourne Vic 3000, Australia. Tel: (03) 9214 7888. Fax: (03) 9214 7839. URL: http://www.aifs.gov.au/