The Scottish Government

« Previous | Contents | Next »

Listen

Chapter Four: Research on Family Goup Conferencing

Introduction: The nature of the available FGDM research

1.4.1 Lupton & Nixon (1999) state that, to date, FGDM studies have been characterised by a focus on solving organisational problems and documenting the views of professionals and agencies rather than children and families. Indeed, although many schemes have been carefully evaluated, the majority of studies have focused upon the satisfaction and involvement of various actors with the process of conferencing. In spite of the fairly extensive literature available, a lack of material on long-term outcomes means that FGDMs effectiveness has not been clearly evidenced.

1.4.2 The dearth of research on long-term outcomes may be the result of political, theoretical, methodological and practical issues. Firstly, political pressure for early results on programme effectiveness may partly explain a tendency to emphasise short-term findings demonstrating participant satisfaction, rather than rigorously scrutinising comparative long-term outcomes. Turning to theoretical issues, Bazemore & Stinchcomb (2000) argue that FGDM 'purists' may view attempts to define, map and monitor outcomes as contradicting the creativity and adaptability at the heart of the initiative. Hence, there may be philosophical reservations preventing rigorous evaluation of the approach.

1.4.3 In terms of methodological considerations, it is extremely difficult to isolate the effects of FGDM from the influence of other services which are typically offered alongside this decision-making approach. In addition, achieving consensus on the aims of conferencing, defining standards and measuring the associated outputs are also problematic. In a practical sense also, efforts to robustly appraise the comparative effectiveness of FGDM using randomised controlled trials have so far been stymied by operational difficulties. A planned comparative study by Lupton and Brown (Brown: 2001) did not go ahead because current national guidance did not allow the participating local authority to create a comparison sample by using FGDM in some cases referred to child protection services. Crampton (2006) also reports that similar problems caused considerable difficulties in Thomas et al's (2003) recent Californian study.

1.4.4 In the absence of robust evidence of beneficial long-term outcomes for children, the basis for implementing FGDM schemes may seem uncertain. However there can be other reasons for implementing FGDM. Responding to a survey by Brown (2003), 19% of UK councils stated that the aim of FGDM was to promote partnership working with parents. For a further 14% the aim was to enhance family involvement in decision-making, re-focus services towards family support or maintain kinship care. These 'process' objectives would be expected to produce better outcomes for children, so demonstrating that they have been met might well indicate that children's needs are being effectively catered for. Nevertheless, more direct evidence on how children fare over the longer term remains a significant gap in current knowledge.

1.4.5 This chapter provides a review of the available literature which attempts to measure the success of FGDM. First research in the fields of child welfare and protection is examined, followed by short discussions of FGDM in juvenile justice and education. The chapter concludes with a discussion of the available research on the cost of FGDM schemes.

FGDM in Child Welfare and Protection -Evidence of Process Satisfaction

1.4.6 Virtually all evaluations of specific FGDM schemes provide details of professional and participant satisfaction with the process. Elements of the available literature are presented below, grouped on the basis of participant type. Within each of the following sections, efforts are made to present information about multiple aspects of the process (e.g. satisfaction with individual FGC components, ability to have a say, involvement etc.) in as far as the relevant literature is available.

Family

1.4.7 In contrast to generally negative perceptions of child welfare and protection processes, much of the literature reveals high levels of satisfaction among family members participating in FGDM. Although the published research strongly tends toward positive findings, it must be borne in mind that findings may vary between different cases and fields where the model is applied (so far, no research using a consistent methodology applied across a number of fields and/or risk types has been conducted). Also, because of the processes of referral and high rates of plan agreement, most respondents are likely to conclude their encounter with the model in a relatively successful fashion. Taking these factors into account, the absence of co-existing research on the satisfaction ratings of the (relatively small) proportion of participants who did not agree a plan or who withdrew from the process altogether would permit a fuller examination of the relevant issues.

1.4.8 Merkel-Holguin et al (2003: 8) state that there is 'considerable across-studies agreement that families are highly satisfied with most aspects of FGC'. The authors state that a number of studies based on surveys, focus groups and interviews report high levels of satisfaction among family members with the process, their influence on decision-making, a high level of willingness to undertake the process again and a preference for FGCs compared to other decision-making forums. Families, they conclude, state that their feelings are respected, influential and they regard the decision-making process as fair.

1.4.9 In Canada, Burford & Pennell (1995) also report high mean satisfaction ratings among family participants across a range of issues. Satisfaction with the way the conference was run ranked highest (mean 3.87 out of 4), while satisfaction with their ability to 'have their say' (mean 3.78) and the appropriateness of decision-makers (mean 3.73) were ranked relatively low yet remained very positive findings. Routhier (2006), reporting on Manitoba's FGC project, found that 76% of family respondents found conferences very helpful, while 75% stated they would recommend the process to other families. Two-thirds of families participating in the Toronto project (Mandell et al: 2001) indicated that they felt fairly treated during and after the conference, although the remainder indicated that professionals were initially cold and arrogant at the initial conference. Despite this, all family members stated that the process was beneficial for them and that others should participate in FGDM.

1.4.10 Similarly high levels of satisfaction have been reported in Europe (Sundell & Haeggman:1999; Rasmussen & Hansen: 2002 (cited by Rasmussen: 2003)) and in the UK (Lupton & Stevens,1997; Holland et al, 2003; Gribben,2005). In Scotland, Murray et al (2001) found family members to be generally positive about FGCs, emphasising procedural informality and their opportunities to speak as particularly beneficial elements of the process. Hamilton (2005) argues that 95% of family members attending Scottish FGCs stated they comfortable at least part of the time, while two-thirds (65%) stated they felt comfortable throughout (4% said they felt uncomfortable throughout). She argues that this compares favourably with findings from O'Brien's (2002) Irish study, where only 36% stated they were comfortable throughout the meeting, while a further 14% were comfortable some of the time. Cashmore & Kiely (2000) and Murray et al (2001) both identified a significant proportion of family respondents who complained about the duration of the conference.

1.4.11 Moving onto the functionality of FGCs as decision-making fora, Sandau-Beckler et al (2005) report that 99% of New Mexico families reported effective communication, all felt their concerns were taken account of and 99% felt these concerns could be expressed to the facilitator. Cashmore & Kiely (2000) reported that most family members (70%) felt able to express themselves during their conference, however, around a quarter felt constrained by the presence of so many other people and/or by the perceived domination of proceedings by particular family members. Across a range of studies which address this issue, 70-80% of family members expressed satisfaction with being listened to, with remainder feeling their views were not heard or accorded little weight (Lupton et al 1995; O'Brien: 2002; Hamilton : 2005). Lupton & Nixon (1999) contrast these findings with those from case protection conferences where only half felt adequately informed, less than a quarter attended, only one third felt listened to and one half felt their views had no weight.

1.4.12 In relation to specific elements of a FGC, Marsh & Crow (1998), evaluating the UK pilots, found that the majority of family members thought that the information-giving element of their conference was handled well (86% felt they could speak, 80% they felt could ask questions and 17% felt uncomfortable). However, there was some early criticism from the pilots that reports were too long or used jargon, a finding also reported by Murray et al (2001). Hamilton reported that 81% of family members participating in Scottish FGCs thought that they were provided with sufficient information to enable them to make a family plan (Hamilton: 2005). Burford & Pennell (1995), evaluating FGCs in Newfoundland, also found that despite this stage of proceedings often being painful for families, they nevertheless benefited from it and reported high mean satisfaction ratings in relation to whether the requisite information was provided (mean 3.83 out of 4).

1.4.13 In relation to private family time, Hamilton (2005) states that 93% of family members expressed approval of this aspect of FGDM without reservations. The concerns expressed by the remainder focused on potential conflict, the effect on young people and the family's ability to work out plans. Other studies have also reported that around 90% of families appreciate the use of private family time and feel they van express themselves freely in this part of the conference (Marsh & Crow,1998; Sandau-Beckler et al, 2005). However, Holland et al (2005) state that at least one member of 41% of families interviewed clearly stated that they wished that the professionals had not left the room during private family time.

1.4.14 A number of studies have found that family members often experiences meetings as emotional and stressful, but most said they would choose to undertake a FGC again (Lupton et al: 1995; Murray et al: 2001). In Cashmore & Kiely's Australian study (2000), 90% stated they would recommend the process to other families in similar circumstances. Hamilton (2005) reports that most (91%) believed that the meeting was appropriate and/or useful, describing it as an improvement on traditional decision-making forums. Contrasting FGCs with traditional child welfare meetings, Walker (2005) states that satisfaction with the child protection system was higher among families whose case was dealt with using FGDM than among those dealt with using traditional processes (41% compared with 23%). Furthermore, research discussed by Helland (2005, citing Titcomb et al: 2003, Kelso: 2003) states that participants with experience of the child welfare system were also cautiously optimistic about the new approach.

Children

1.4.15 Children and young people occupy a unique position within the FGDM approach and their conference experience is likely to reflect a range of factors such as their perception of the problem, reactions of other family members and outcomes from the planning process. Helland (2005) discusses a range of studies reporting reactions such as confusion, hope and anger (Heino: 2003), relief, amazement and gratefulness among children in care who were involved in a FGC (Velen & Devine: 2005). Meanwhile, Horan & Dalrymple (2003) comment on the possibility of confusion and distress arising among children who experience conflict during a FGC. Certainly FGDM can be a particularly intense and emotional experience for children and therefore their satisfaction with the experience deserves close attention.

1.4.16 Children and young people attend most FGCs and there is some evidence (Lupton & Nixon: 1999) that they are generally satisfied with and feel involved in the process. Velen & Devine (2005) found that 83% of children felt they had a real voice and 92% said they felt safe at their conference. Holland & O'Neill's (2006) evaluation of a child-focused FGDM project in Wales specialising in accommodation issues also revealed some positive findings. They found that most young people felt that they did have their say at their conference and that despite the possibility of conflict or risk, their experience of FGDM was a positive one. However, a significant minority felt that their opinions were not respected or understood and the majority felt they exercised only a little or no influence over proceedings.

1.4.17 As this last point demonstrates, the mere presence of children and young people does not ensure that they will feel involved and able to 'have their say' at a conference. Evidence indicating that children are able to cope successfully with a FGC is contradicted by a further body of evidence suggesting the opposite view. Morris & Maxwell (1998) state that less than half of the young people they interviewed felt involved in the FGDM process. Walton et al (2003), evaluating Utah's FGDM project, state that participating children felt ill prepared for meetings and possessed a limited understanding of their purpose. Horan & Dalrymple (2000) cite Clarkson & Frank (2000) and Beecher et al's (2000) findings that children generally felt they were not heard during FGCs and did not participate due to inhibitions, a lack of confidence or understanding of proceedings or because they felt no-one was listening. Rasmussen (2003) also argues that Danish children found FGDM participation difficult.

1.4.18 Further research is needed to ascertain how key characteristics relating to the child, family and FGC practice influence positive or negative experiences among children.

Social workers

1.4.19 Merkel-Holguin et al (2003) state that process evaluations indicate widespread satisfaction with FGDM among social workers and other social care professionals. Measured on the basis of influence, perceptions of conference effectiveness and satisfaction with plans, social workers were generally positive about experience of FGCs. Sundell & Haeggman (1999), Rasmussen & Hansen (2002) and Marsh & Crow (1998) report satisfaction ratings of 76%, 74% and 73% respectively in the Swedish, Danish and UK pilots (cited by Rasmussen: 2003). Velen & Devine (2005) also contribute to this positive body of evidence, reporting similar findings.

1.4.20 As discussed above, while there is support for the FGDM model among social workers, there is also scepticism regarding its capacity to accommodate high-risk cases and manage the possibility of conflict. Alongside other factors (e.g. staff turnover), such concerns may contribute towards low referral rates. Marsh & Crow (1998) state that research from New Zealand and Australia reveals that practitioners support the approach and distinguish it from other methods, but find it more time-consuming and tiring than traditional procedures. Marsh & Crow's (1998) evaluation of the UK pilots found that 92% of social workers thought FGCs were different from other meetings (in terms of principles, roles, power) and two-thirds thought they could empower families. Their main concerns focused on whether families could participate amicably because of conflict between themselves and conflict with social workers. However, most wanted FGDM to remain available, while over half favoured expansion.

1.4.21 Aside from their referral role, social workers must also provide information at FGCs (using clear, jargon-free forms of communication) and follow up plans made during conferences. Marsh & Crow (1998) state that while workers were clear about their 'reporting' role in relation to providing information about problems, they were less familiar with their 'enabling' role in providing families with information on potential resources. Murray et al (2001) found that although most families were neutral about the benefits of the information-giving stage, social workers saw it as beneficial for families.

1.4.22 In relation to private family time, Marsh & Crow (1998) found that professionals were supportive of the concept but often expressed concerns regarding the possibility of conflict. Correspondingly Hamilton (2005) reported that 77% of professionals approved of private family time and had no concerns about it; the concerns expressed by the remainder (particularly prevalent were education and health workers and those attending their first FGC) focused on potential conflict, the effects on young people and the ability of the family to work out plans. In the Scottish study by Murray et al (2001) some professionals raised concerns that some issues relevant to problems being considered were not discussed during conferences.

1.4.23 Overall, professionals are generally supportive of the process of conferencing, but have reservations regarding the potential for conflict and the ability of families to contribute towards effective outcomes. If a means of allaying such concerns without compromising the empowerment of the family can be found then it may assist in promoting the acceptability of FGDM and mainstreaming its use.

Child Welfare and Protection- FGDM outcomes

1.4.24 Evidence on FGDM outcomes varies greatly in terms of quality. While many evaluations provide information on process outcomes (e.g. plan agreement), only a small number provide for long-term outcome measures. This section explores this evidence and provides methodological comment where appropriate.

Plans

1.4.25 Satisfactorily concluding a conference with family and professional agreement upon a plan is a significant measure of the success of the FGDM process. However Lupton & Nixon (1999) point out that the term 'plan agreement' is often ambiguous, as agreement might be reached at different points e.g. following consultation with service providers. They also point out that the extent of professional input and post-conference negotiation is not documented.

1.4.26 Not withstanding some ambiguity about how plans are agreed, the overwhelming majority of FGC plans are accepted by agencies/professionals and are viewed positively by families. Merkel-Holguin et al (2003) state that, on average, only 5% of plans are not accepted by authorities and very few FGCs result in no plan being formulated. Evidence from a range of studies demonstrates a high rate of plan agreement:

• New Zealand - 92% plans agreed (Paterson & Harvey: 1991; Mason et al: 1992);
• New Zealand (youth justice) - 95% plans agreed (Maxwell & Morris: 1993);
• Australia (New South Wales) - 95% plans agreed (Cashmore & Kiely: 2000);
• Canada - 97% plans agreed (Burford & Pennell: 1995);
• USA (Washington state) - 99% plans agreed (Shore et al: 2001);
• Northern Ireland - 100% plans agreed (Gribben: 2005);
• UK - 93% plans agreed (Marsh & Crow: 1998).

1.4.27 Information on the few cases where plans are not successfully formulated is sparse. However, Marsh & Crow (1998) provide some insight into two such instances. In the first the parents denied the factual basis of the case, while in the second, family members were unable to resolve their differences.

1.4.28 The available literature suggests that most families are satisfied with the content of FGC plans. Burford & Pennell (1995), evaluating FGCs in Newfoundland, reported high mean satisfaction ratings among family participants in relation to whether the requisite information was provided (mean 3.83 out of 4) and whether they agreed with the resulting plan (mean 3.79). Overall, they found that only seven out of 293 (2%) participants were not happy with plans and 15 were not sure (5%). Mandell et al (2001), also evaluating a Canadian scheme, found that families were generally happy with all aspects of the FGDM process and the resulting plans. Marsh & Crow (1998) reported that 80% of family members were satisfied with the plans made, but this figure declined over time due to implementation problems. Scottish research has found similarly high levels of satisfaction with plans (Murray et al: 2001; Hamilton: 2005)

1.4.29 FGC plans were also viewed positively by professionals. Lupton et al (1995) found that although a small proportion of social workers were not satisfied with plans because they failed to address all their concerns, these cases were concentrated in the earliest FGCs convened during the UK pilots. Barker & Barker (1995) also found that plans made in some early FGCs were too vague and lacked detail. Overall, Marsh & Crow (1998) stated that professionals were generally very impressed by the plans made in the UK pilots. They acknowledged their creativity and often stated that social care professionals could not have formulated such plans themselves. In terms of protection, 67% of social workers thought that plans were likely to result in better protection for the child than would otherwise have been possible, while the remainder stated that the level of protection would be as good as that afforded by social services.

1.4.30 In terms of content, the literature suggests that plans are positive, robust and utilise a mixture of family and professional resources. Marsh & Crow (1998) state that most FGDM plans included elements of assistance from the family (characterised as wide-ranging and practical in nature) and that 31% involved a relative taking on the care of the child for a period of time. Lupton & Nixon (1999) found that FGCs generally resulted in positive plans which utilised and increased the availability of family resources.

Plan Implementation

1.4.31 Effective plan implementation is evidently one indicator of FGDM effectiveness, but the relationship between carrying out the plan and positive outcomes is not straightforward. Assessing whether plans have been implemented is problematic since individual participants may have differing views about whether or not certain aspects have been carried our as envisaged. In addition, changing circumstances or new information may make it clear that it is in the child's interests to change or omit key elements of the original plan However other less helpful influences may also prevent plans being put into practice. Lupton & Nixon (1999) argue that resources for plans are often not delivered, despite plans being regarding as reasonable. They attribute this to lack of resources and commitment from either professionals or members of the wider family.

1.4.32 In spite of these problems, some commentators regard plan implementation as a useful outcome measure. Lupton & Nixon (1999) reported that plans were implemented with some success but often several components (particularly those involving behavioural change) were not put in place. Pennell & Burford (2000) report that in the majority (69%) of cases they observed in Newfoundland, plans were only partially implemented. However, despite the level of plan implementation, two-thirds (66%) of family members stated that the family was better off because of the conference. Responses indicating that the family was 'the same' or 'worse off' were most likely to originate from families where plans had not been implemented at all. Cashmore & Kiely (2000) state that most of the plans made during New South Wales conferences were only partially carried out (75%). The remainder were thought to have been carried out in full by families, but not by social workers. The main reasons for partial (or non-) completion were failures among family members to change unhelpful behaviours, (particularly in relation to drug and alcohol use), changes in circumstances and statutory agency's slowness or inability to perform their duties. Both social workers and families viewed a joint lack of commitment to plans and the resulting failures in delivery as the main problem with the FGC approach.

1.4.33 Marsh & Crow (1998) present evidence from a number of studies (Lupton et al: 1995, Barker & Barker: 1995, Rosen: 1994) stating that between 43 and 83 per cent of family members viewed FGDM plans as being successfully implemented. Marsh & Crow themselves rated 78 plans, concluding that in 75% the general aims had been achieved. Grimshaw & Sinclair's (1997) review of 180 plans indicated that 68% had been implemented. In comparison, (Lupton et al : 1995) report social workers' assessments that 66% of child protection plans were fully implemented.

Relationship-oriented outcomes

1.4.34 Evidence suggests that relationships between family members and between the family and social care professionals may improve following a FGC. A number of researchers argue that mere participation in FGDM often brings about unintended and unplanned positive outcomes for families in terms of improved family functioning, renewed contact with relatives and more open communication (Barker & Barker:1995; Marsh & Crow:1998; Lupton & Nixon :1999; Sandau-Beckler et al :2005, Routhier : 2006) However, Mandell et al (2001) and Cashmore & Kiely (2000) report that while half of family participants reported such improvements, half experienced no change, with only a few reporting deterioration. Pennell & Burford (2000) contend that conferences helped by strengthening positive ties among participants and enhancing their sense of family unity. In the minority of cases where families were worse off, conflict had increased and families were separated by attempts to implement FGC plans.

1.4.35 There is also some evidence of improvement in the relationship between family members and social workers. Just over a third of family respondents in Routhier's study (2006) identified a better working relationship with social care workers, while Nixon et al (2005) found that social workers reported less conflict with families following a conference. Half of social workers participating in the UK pilots (Marsh & Crow: 1998) stated that FGDM helped them improve their partnership practice and think more positively about families, as did two-thirds of workers participating in Mandell et al's (2001) Toronto study. Interestingly, Cashmore & Kiely (2000) report that three quarters of social workers but only half of family members stated their relationship had improved or remained positive.

1.4.36 It appears then that around half of family members discern some positive change in terms of familial relations and some also identify improvements in relationships with social workers. By re-establishing links between relatives and reinvigorating networks of support, FGDM may help to demonstrate the efficacy of family decision-making and thereby restore social worker's confidence in them as viable partners in social care.

Medium -Long-term Outcomes

1.4.37 As noted previously, three is relatively research based evidence on whether the use of FGDM helps achieve better outcomes for children in the medium-long term. Marsh & Crow's (1998) evaluation of the UK pilots provides follow-up data for 64 children (80% of their sample) over a six-month period. According to their circumstances, the children formed three sub-groups: 'accommodation request/risk', 'protection' and 'looked after'. A high degree of placement stability is reported across all groups, with the 'accommodation request/risk' group most likely to experience a change in placement after 6 months (8/21). Plans for all but two of the 'protection' group were successful although the most successful group were 'looked after' children who, in many cases, gradually moved back to the family over time. Social workers thought that about half of children would have had the same outcome, but may have experienced less family support and that 20% of children would have been worse off, with an increased likelihood of being accommodated away from home.

1.4.38 Marsh & Crow try to assess the influence of FGDM by comparing outcomes for these children with those reported in other studies of children returning from care (Bullock et al: 1993) or involved in the child protection system (Thorburn et al: 1995; Levy et al: 1995). In terms of avoiding reception into care, court proceedings and reducing re-referral the FGDM outcomes seemed favourable. However the comparisons were with quite different studies and samples, so these findings can only be tentative. A follow-up period of six months is also too short in which to assess lasting benefits.

1.4.39 Drawing on the research available at the time, Lupton & Nixon (1999) measure outcomes across a number of stages: immediate, intermediate and long term. Immediate outputs and outcomes relate to aspects of the FGDM process. Despite a lack of evidence on monitoring and culture recognition and the difficulties of measuring concepts such as 'responsibility', they argue that 'process' findings are generally positive and indicate that FGDM mobilises family support effectively.

1.4.40 Examining intermediate outcomes, they found that while most plans were put into practice, often multiple plan components were not implemented. Evidence on improved family functioning was found to be equivocal. With regard to retention within, or return to, the family network, Lupton & Nixon state the evidence available from New Zealand is problematic due to the lack of available comparators. Attempting to fill this knowledge gap, they compare Marsh & Crow's (1998) FGC data with those of Lupton & Stevens (1997) who examined placement following child protection conferences. This comparison indicates an increased likelihood of family network accommodations and a decrease in the use of foster care among FGC families. However, the use of non-family care was not reduced nor did the likelihood that children would remain with the immediate family increase. Lupton & Nixon (1999) argue that, although generally positive, the findings may be influenced by the characteristics of the children involved.

1.4.41 In terms of long-term outcomes, Lupton & Nixon examine criteria relating to whether children are protected from abuse and neglect, their welfare is promoted and enhanced, a communal sense of responsibility is promoted, the needs of the main parent/carer are met and families are empowered via the successful resolution of problems. In relation to their first criterion, outcomes are measured in terms of continuing registration with relevant child protection authorities. Here, a lower rate of 'at risk' registration and a lower re-abuse rate (6% compared with around 30%) was observed among FGC children after twelve months when compared with data from non- FGC children. In addition to this, professionals thought that FGC children were better protected. Finally, they argue that in over half of cases a demonstrable link existed between FGC plans and positive outcomes (Lupton et al:1995) and that outcomes were generally more positive than either families or professionals expected.

1.4.42 In summary, Lupton and Nixon's approach is a rigorous analysis of the findings then available from other FGDM evaluations. The result is a complex picture of diverse approaches to research, measures and constructs. Their key findings are that, when not curtailed by professional interests, FGDM often leads to positive outcomes for families. They argue that FGDM is marginally less effective at resolving problems than traditional approaches and that the safety of children is no less likely to be compromised after a FGC. However, they concede that little is known about enhancements to child safety. While no robust conclusions are forthcoming, the general picture provides some support for the FGDM approach in the sense that it is successful in empowering families and results in outcomes which are no worse than those associated with traditional approaches.

1.4.43 Six to twelve months after their conference, two thirds of children in Cashmore & Kiely's (2000) New South Wales study were living at home or within the family group, three had entered out-of-home care and out-of-home placement orders were being sought for a further four children due to the breakdown of plan arrangements. In addition, three children had changed their out-of-home placement. On balance, these findings are generally positive, particularly taking account of the fact that most children originated from families at the 'hard' end of the child welfare spectrum. Overall, workers believed that risk to children had been addressed and reduced as a result of FGC plans, but had significant concerns about the long-term wellbeing of children for a significant minority of the families involved.

1.4.44 Pennell & Burford (2000) attempt to measure specific outcomes using indicators associated with levels of family violence before and after a FGC (the Manitoba project they evaluated was specifically set up to deal with such issues). Despite problems with data validity and under-reporting, they utilise information from progress reports (158 completed by 115 individuals attending 28 of the 32 project families) and a child protection events checklist (verified against case files for all 32 FGC and 32 comparison group families) to quantify levels of violence ²⁴.

1.4.45 Pennell & Burford found that families undertaking FGDM exhibited more child protection events than comparison group families in the pre-test period (mean 7.28 per FGC family compared to 4.16) yet exhibited fewer at the end of the twelve-month test period than comparators (mean 3.66 per FGC family compared to 5.32). Further subdivisions of the items on the child protection checklist suggest that in relation to child safety, service agency activity (including emergency visits and apprehensions) and substantiations of abuse or neglect fell among FGC families but rose among comparison group families during the test period. Members of the relevant agencies reported fewer visits to FGC families and an enhanced knowledge and understanding of their circumstances. Interviews also confirmed these trends, with less abuse and neglect and better provision of care apparent in FGC families.

1.4.46 Unlike other researchers, Pennell & Burford are not solely concerned with child welfare-related outcomes. Examining mother and spouse abuse, they found that associated events declined substantially for FGC families over the test period while rising for comparison families. Indeed, the only area in which negative outcomes were observed for families was in relation to child-adult abuse (normally of the mother), which grew in prevalence among FGC families and was strongly associated with child self-harm. Although incidents of child self-harm also rose among FGC families, this was a low level occurrence across the sample as a whole.

1.4.47 Pennell & Burford's robust findings provide an endorsement of the use of the FGDM approach in high-risk cases involving family violence. Although it is hard to ascribe direct causality to the FGDM approach, findings on the reduction of violence and abuse among FGC families are persuasive. However, as the focus of their analysis is fairly narrow, families were randomly assigned to experimental and control groups and the period of analysis was only of a moderate length; further research is required in order to verify the generalisability of these findings in relation to other problems (e.g. accommodation) and in other jurisdictions.

1.4.48 Stevens' (2003) longitudinal study of family's views on FGDM used audio diaries and repeated interviews with a small number of participants (social workers, children and families). Tracing participants over 18 months, the study aimed to inform the theoretical understanding of outcomes of child welfare intervention and contribute to the development of outcome measures. The work lays primary emphasis on the narratives of the families concerned and draws together issues around the ownership and resolution of problems. It is argued that, at best, FGDM plays a steering role, establishing a family's ability to care for their children, influencing more powerful decision-making fora and strengthening family support. In a similar fashion to Lupton & Nixon (1999) they found that such influence was contingent upon strong professional and family commitment to FGDM and a willingness of professionals to work with family decisions in a multi-intervention context. Without the presence of these factors, the role of FGDM was short lived, minimal or both. This suggests a strong interaction between level of control, degree of public concern, level of family and professional commitment to the process and the impact of FGCs.

1.4.49 Shore et al (2001) report the findings of a retrospective study of conferences in Washington State (held between 1998 and 2001). The study examines outcomes for 138 children (participating in 70 FGCs) who were at least six months post- FGC. Information relating to 114 of these children (about whom information regarding pre-conference living arrangements was available) indicated that the proportion of children living with parents increased after a FGC and the proportions living with relatives and non-relatives both decreased. Permanent placements (defined as the least restrictive (and most-family like) setting available and in close proximity to the parents home, consistent with the best interests and special needs of the child) occurred as a result of FGCs for 82% of the children involved, though it was not considered in all cases. There was also a high rate of placement with paternal relatives (24% with father, 10% with a paternal relative).

1.4.50 In terms of safety, a very low rate (6.8%) of substantiated allegations of abuse or neglect was observed among the sample (133 children). The overwhelming majority of children (90%) were stable in the placements identified in their FGC plan. Only 10% of children (14 of 137) experienced difficulties and were subsequently located in out-of-home placements. For children in sexual abuse cases, findings were very similar with a high proportion of stable placements and no substantiated post- FGC child protection re-referrals. Cases were closed (indicating that the problematic situation the conference had been convened to address had been resolved to the satisfaction of the social worker) in over half (54%) of the cases where the FGC had been held two or more years before. On the surface, Shore et al's findings are very encouraging for FGDM, but the absence of a comparison group is extremely problematic, there being no evidence that the same results would not have occurred had the families been dealt with using traditional decision-making processes.

1.4.51 Sundell & Vinnerljung's (2004) Swedish study compared 97 children involved in 66 FGCs with 142 children from a random sample of 104 traditional child protection investigations conducted by Child Protective Services ( CPS). All children were tracked for three years in order to compare future maltreatment episodes. After controlling for age, gender, family background and type and severity of problems, analysis found that FGDM children exhibited higher rates of re-referral to CPS. FGC children were re-referred more often because of abuse, were more likely to be re-referred by extended family members and spent longer in out-of-home placements than children dealt with using traditional processes. However, over time, the type and level of support they received from CPS was less intrusive. Involvement in a FGC was not related to re-referrals on the basis of neglect, case closure after three years or the number of days services were provided for.

1.4.52 Overall, these findings show that children dealt with using the FGDM model were more likely to experience abuse and less likely to sustain kinship placements than those dealt with using traditional child welfare processes. These findings contradict those of other studies, which have tended to provide evidence of long-term benefits. Moreover, the methodology used by Sundell & Vinnerljung is the most robust of those so far discussed. Therefore, provided that the official CPS measures used in the study are themselves accurate and conditions in Sweden sufficiently similar to those in other jurisdictions, extrapolation from these findings casts some doubt upon the efficacy of FGDM schemes in general.

1.4.53 Gribben (2005) provides some outcome data for children involved in FGCs and school restorative conferences ( SRCs) in Northern Ireland. Here it is argued that 18 FGC children (from a total of 71 engaged) were averted from local authority care and 11 were returned from care. Twenty-two FGC children were returned home or maintained there with the help of extended family or community care, 15 were placed or maintained with family members and 26 experienced improved family relationships through FGCs. While these data are encouraging, again the absence of a comparison group reduces the confidence with which we can assign causality to the conference itself.

1.4.54 On balance, much of the available evidence suggests that FGDM probably does have a beneficial effect on outcomes for children and families. However, much of the available literature is methodologically weak in the sense that it does not include comparison with other decision-making approaches. Somewhat disconcertingly, the sole piece of literature to reach a negative conclusion on FGDM outcomes is also the most methodologically robust. While it would be unwise to lay too much emphasis on this one study, the emergence of more negative results does highlight the need for further critical review of how FGDM both fits with other child welfare processes and impacts on individual children and families.

FGDM in Juvenile Justice

1.4.55 In a number of jurisdictions, FGDM is primarily used within juvenile justice. In New Zealand, FGCs are mandated in order to divert cases from the juvenile courts (Morris & Maxwell: 1998, Doolan: 1999), while Southern Australia uses FGDM to work with young offenders occupying the 'middle ground' in terms of the extent and seriousness of offending. Operationally and philosophically distinct from the mechanisms of police cautioning and the youth court, FGDM aims to help both offenders and victims and redress problems from the perspective of the wider community. Similarly in Thailand, Family and Community Group Conferencing ( FCGC) is used as a pre-court diversion, with a referral to child protection and subsequent evaluation for FCGC suitability following the arrest of every child (Roujanavong: 2005). Within youth justice, the restorative element of FGDM is central, thus raising an additional set of issues which do not apply within child welfare systems.

1.4.56 Pratt (1993) argues that, because of a lack of fit between restorative principles and traditional approaches to crime and punishment, it is impossible to introduce interventions based on restorative justice into existing systems. Notions such as the involvement of victims, allowing all parties to 'have a say', shifting power from the criminal justice system to the community and adopting a non-retributivist stance towards offenders all represent, to some degree, a departure from the traditional adversarial system of justice currently in place. Yet there is evidence of FGDM working well in a youth justice context, so these philosophical tensions do not present an insurmountable obstacle to the incorporation of this approach, though they do highlight some potential difficulties in implementation.

1.4.57 One of the identified benefits of the use of FGDM in youth justice is more potential for benefiting victims of crime. Hayes et al (1998) argue that victim representation differs on the basis of the volume of referrals processed by FGDM schemes. Victims were present at around 50 to 60 per cent of conferences held in high-volume jurisdictions such as New Zealand and South Australia, while in low-volume schemes such as that of Queensland, victims attended over 90 per cent of FGCs. This finding suggests that as volume increases, there is less time available for preparing and involving victims, so fewer attend. This explanation is supported by Morris & Maxwell's (1998) research from New Zealand which found some victims had not been invited, others perfunctorily so, some were not well prepared and there was little attention to their needs. Furthermore, Jackson & Morris (1998) state that a significant minority of victims in New Zealand were dissatisfied and felt they had not been able to express themselves when attending conferences. In contrast, Wundersitz & Hetzel ( 1996) describe practice in Southern Australia in which the attendance and needs of the victim are prioritised and outcomes carefully managed to ensure that they are acceptable to all parties.

1.4.58 Research on outcomes typically focuses on the impact on victims and offenders. Based on a meta-analysis of evaluations of both FGCs and mediation schemes, Williams-Hayes (2002) concluded that, in comparison with other interventions these restorative schemes reduced victim's fears of revictimisation, enhanced offender and victim satisfaction with the process and increased the frequency with which restitution negotiations and agreements were completed. Morris & Maxwell (1998) provide a useful overview of research on justice FGCs in New Zealand in which they argue that although a victim is present at around only half of conferences, this can be attributed to them not being invited or the time and place being unsuitable, rather than lack of willingness to attend. Sixty per cent of victims attending a FGC described it as a helpful, positive and rewarding experience in which they generally felt involved, while 70% of victim's family members expressed satisfaction at FGDM outcomes. About a quarter of victims stated that they felt worse after attending a FGC, mainly because they felt that the young person was not truly sorry for their actions. etc.

1.4.59 Evaluating Winnipeg's use of FGCs in youth justice, Longclaws et al (1996) examine process outcomes for a small sample of eight aboriginal offenders. They found that 'disorganised' families could reach a consensus and develop plans for their children using a culturally appropriate decision-making process. However, the low level of victim participation and the fact that judges largely ignored family plans suggest that the process implemented required substantial modification in order to be effective.

1.4.60 An analysis of reconviction data by Morris & Maxwell (1998) found that the proportion reconvicted within a year of their FGC (26%) was no worse than other disposals (a comparison group was not used, but results were compared with studies of other disposals). They also argue that the probability of reconviction was reduced when restorative elements of FGDM were successfully achieved; persistent recidivists were least likely to complete agreed tasks and were less likely to apologise. However, as well as being sources of potential bias within the criminal justice system, prior offending and being Maori were the most significant predictors of reconviction.

1.4.61 In a similar study, Hayes & Daly (2003) analysed re-offending outcomes from the South Australia Juvenile Justice ( SAJJ) project discussed above. They observed and distributed questionnaires at 89 Adelaide conferences in 1998 and interviewed the relevant offenders and victims in order to record indicators of restorativeness (e.g. accepting responsibility, apologising, expressing defiance/remorse) and procedural justice (e.g. outcome decided by consensus, police treated offender with respect). Variability in these aspects of conferences was compared with variability in post-conference rates of offending, indicating that two indicators of restorativeness (remorse, conference ending on a high) and one indicator of procedural justice (outcome decided by consensus) were associated with a lower likelihood of re-offending. Higher rates of re-offending were closely associated with young people being aboriginal, socially marginalised and a record of previous offending, but even among young people who fitted this profile, there were indications that these key aspects of the conferencing process had an impact on future offending.

1.4.62 A number of studies have compared reoffending among young offenders subject to FGDM and non- FGDM interventions. The Reintegrative Shaming Experiment ( RISE) project randomly assigned four groups of offenders (drunk drivers, violent offenders aged up to 29, juvenile property offenders and shoplifters) in Canberra (Australia) from 1995 to 2000 to either a court or conferencing pathway. Sherman et al (2000) report significant reductions in the average post-referral rate of offending for violent offenders undertaking a conference but no differences were observed in other groups. However, Hayes & Daly (2003) point out that their methodology was problematic because the requirement of a guilty plea prior to conferencing meant that groups were likely to be similar in ways that were associated with their re-offending.

1.4.63 Luke & Lind (2002) compared re-offending among first-time offenders appearing in the youth court and undertaking conferencing in New South Wales. After controlling for gender, age, offence type and aboriginal background, members of the conference group were 15 to 20 per cent less likely to re-offend or reappear than members of the court groups. However, using further analysis to evaluate the influence of ethnicity, gender and FGDM involvement on recidivism, Baffour & Doniece (2003) found no statistically significant difference between re-arrest rates of FGC and non- FGC participants.

1.4.64 Aside from Pratt's argument regarding a philosophical lack of fit, there remain some uncertainties regarding the practical place of such initiatives in contemporary justice systems. New Zealand opted for a drastic overhaul of the court process, but Jackson & Morris (1998) argue that FGCs could easily be implemented by undertaking them during periods of deferment following an admission of guilt. In Scotland, as is the case in New Zealand and Southern Australia, Youth Courts (operating in Airdrie and Hamilton (McIvor et al: forthcoming)) could present an ideal mechanism for instigating such proceedings. The introduction of FGDM as a pre-court diversionary mechanism would also be possible. Again, the existence in Scotland of Children's Hearings could possibly act as a model for such an initiative.

FGDM in Education

1.4.65 Alongside developments in child welfare, FGDM is also being used in some educational settings in the UK (notably Hampshire and Northern Ireland). In this context, an education professional working with a child and their family identifies the need for a conference in order to support their attempts to find solutions to issues encountered. Although much appears to be left to the discretion of the professionals concerned, the kinds of problems FGDM tends to be used for in such settings are bullying, non-attendance, behaviour which may lead to a threat of exclusion and other situations which could jeopardise a child's education. Aside from the child, their family, the referrer and the co-ordinator, the child's teacher and other staff they know well may attend the FGC. Also, if the case involves bullying, the other child and their family may be represented. Aside from this, FGCs in education are largely the same as in community settings.

1.4.66 In Hampshire, FGDM has been used in schools since 1998. Crow (1999, 2000, 2001) reports on their implementation, process and outcomes in a series of papers available online. She argues that implementation generally progressed well due to enthusiasm among head teachers and the fact that an FGDM infrastructure already existed in the county. However, implementation was hampered by a lack of involvement from social services, the large number of schools involved and limited funding. Despite these problems FGDM was received well by education professionals with 60 referrals and 44 FGCs during the first 18 months. Crow notes that boys were referred more often than girls (3:1) and more frequently for behavioural problems (girls tended to be referred for attendance problems). The particulars of the process (length of time to arrange, number of attendees) were largely similar to other FGCs. Similar findings were also apparent when participants were asked about their satisfaction with aspects of the FGC process; most were positive about their meeting but a small number expressed dissatisfaction. In particular, 21% of family members stated that they did not feel they could participate fully while others identified a lack of support or respect from professionals and 'hearing the child's voice' as problematic. Overall, most professionals and family participants were positive about the FGDM process and stated they would be happy to take part in a conference again in the future. They were often surprised by the level of commitment shown by others to the process. This was seen as essential to the success of FGDM and to achieving positive outcomes.

1.4.67 In the short-term, beneficial outcomes included an increase in understanding between the family and school, participant satisfaction with plans (agreed in 95% of FGCs) and immediate positive changes at home and at school. In a similar fashion to the high level of social welfare involvement observed in normal FGC plans, two-thirds of plans agreed at education FGCs included actions to be undertaken by authorities (in this case, the school). Crow (2001) discusses outcomes after six and twelve months for 50 children referred during the first year of operation. Boys with behavioural problems were most prevalent within this sample. Half of the schools made positive comments about the effect of FGDM on the children concerned. There was a significant positive effect on outcomes, with over half (52%) improving, 28% staying the same and 20% deteriorating in terms of their behaviour, attendance or both (a statistically significant improvement was observed in attendance). However, in relation to behavioural problems, FGDM resulted in improvements for many, but deterioration (i.e. permanent exclusion) for others. Despite this, there were no significant differences between outcomes for children- much of the difference between groups was thought to be due to age and gender.

1.4.68 The Northern Ireland School Restorative Conferencing ( SRC) scheme began in five post-primary schools in 2000 (expanding to cover 18 schools in 2005). Adapted from the juvenile justice model used in Northern Ireland, it is run along very similar lines to the Hampshire scheme described above. Over the first two years of operation the scheme received 64 referrals, resulting in 27 SRCs and 27 plans (Gribben :2005). The main reason for referral was bullying (followed by aggressive behaviour, drug use and truancy) and Gribben reports that positive outcomes were achieved in most, if not all, cases.

1.4.69 On the basis of the limited evidence available, all of which originates from the UK, FGDM in an educational setting appears to be no less effective than when implemented in a community setting. These studies indicate that they are viewed positively by professionals and families and can result in positive outcomes when both sides are committed to the process. Although no long-term perspective on the issue is forthcoming, referrals appear to be relatively high, perhaps indicating that education professionals may not have the same reservations about using the model as social workers. However, such a finding might also be consistent with the high proportion of low-risk cases encountered in an education setting.

Costs

1.4.70 The true costs of implementing and administering a FGDM project are extremely difficult to determine. In order to make a robust comparison between FGDM and other decision-making fora, start-up and operational costs (the principal expenditure being co-ordinator time) must be incorporated alongside estimates of savings (including actual/possible outcomes and their associated costs) for matched samples of families dealt with using each process. Taking into account the lack of reliable long-term outcome data, the difficulties involved in collating and analysing budgetary data means that existing analyses have tended to provide only a partial view of the costs and savings associated with FGDM.

1.4.71 In common with the paucity of evidence on costs, funders' cost expectations are discussed infrequently within the FGDM literature. Lupton & Nixon (1999) argue that convincing potential funders of the cost-effectiveness of FGDM is potentially problematic. Presumably, this issue is associated with a generalised reluctance to fund new approaches and may not be quite so relevant now (some fifteen years after the introduction of FGDM). Some of the literature suggests that there is also an (somewhat contradictory) expectation that, once in place, FGDM will reduce expenditure on resource provision by promoting care within the family. Overall, the evidence suggests that FGDM plans tend to bring in both family and social welfare resources but do not result in any drastic reduction in demand for the latter. Shore et al (2001) found that in at least 80% of plans analysed, social welfare services were identified (most frequently health services such as counselling). In all of the agreed plans they studied a family-driven resource was also listed. Rather than reducing demand, the nature of the welfare services requested by families tends to shift from interventions focused on out-of-home care (e.g. foster placements) to those aimed at resolving issues encountered by families (e.g. counselling and in-home support). On the issue of cost, Marsh & Crow (1998), evaluating the UK pilots, found that 75% of FGC plans were graded by service providers as being of low to medium cost. Following implementation, it was found that the majority of plans (55%) cost less than had been anticipated while 39% cost the same. Overall, only six per cent of plans resulted in higher than anticipated costs; these were principally associated with the provision of residential care.

1.4.72 On the basis of an overview of the available evidence, Merkel-Holguin et al (2003) state that FGDM is cost neutral or provides cost savings. Evidence from other, more detailed, costing exercises tends to confirm this finding. Burford & Pennell (1995) recognise the problems of achieving robust cost estimates but state that, as those undertaking FGDM were existing clients of social services, no new monies were required for setting up the Newfoundland project. They found that the only additional expenses associated with holding a FGC were travel expenses and co-ordinator time. With regard to the period following a conference, they recognised that costs vary and may initially be higher in the short-term (e.g. because temporary care placements are requested), stay the same or decrease. Overall, they argue that the greatest shift in resources was from the costs of foster care to the costs of in-home support, but due to the lower cost of the latter and the desire of families to accommodate their own kin, the costs of FGDM could be met within existing budgets.

1.4.73 Mandell et al (2001), discussing their evaluation of Toronto's FGDM scheme, also provide some tentative costings. They estimate that the average cost per conference was $1850 (composed mainly of co-ordinator time and expenses), with the additional cost of plans being negligible. Furthermore, they calculated cost savings of between $600,000 and $900,000 based on the avoidance of 'in-care' days for the 20 to 30 children who went to live with family members following their FGC. They postulate that other savings may result from reduced time spent in court, reduced family legal fees and other court costs. Taking these into account, they argue that the FGDM project, costing $225,000, may have resulted in cost savings of double that figure.

1.4.74 Crampton & Jackson's evaluation (2000) of a FGDM project in Kent County, Michigan provides further interesting findings. From a total of 96 families referred over a three-year period from 1995 to 1998, they state that 60% developed a plan resulting in children avoiding foster care or court involvement. Therefore, assuming that children diverted by FGDM would have been removed from home and made court wards (at an estimated cost of around $6,000 for each of the 50 children involved), savings exceed programme costs by around $80,000. Despite the encouraging findings, this analysis is somewhat simplistic in ascribing to all a universal cause and effect without taking due account of intervening factors or differentiating characteristics.

1.4.75 While again it must be stated that the evidence to support any definite conclusions is lacking, the cost evaluations so far undertaken appear to suggest FGDM has little or no impact on welfare expenditure. While FGDM must not be implemented on the basis of anticipation or expectation that it will reduce expenditure, the approach may result in some cost savings. However, attempting to make further savings by cutting corners in the implementation or operation of the approach may have unintended, probably detrimental effect and should likewise be avoided.

Summary

1.4.76 Literature on FGDM demonstrates a number of important points which broadly support the implementation and use of the approach:

• there is a high level of process satisfaction among family members, children and social workers in relation to FGDM;
• participants from among the family generally experience a considerable level of engagement with, and involvement in, the FGDM process;
• family members are supportive of private family time;
• FGDM may assist in improving relationships between family members and with social care professionals;
• a high proportion of FGCs result in plans being successfully agreed and implemented;
• FGC plans are often perceived by social workers as more creative and offering at least as much protection for children as they could offer themselves;
• FGDM leads to long-term outcomes that are, in most cases, no worse than those following traditional decision-making fora;
• victims of crime may benefit from involvement in a FGC;
• applying FGDM in youth justice and education fields brings about outcomes comparable with those achieved in child welfare;
• FGDM maximises resource provision from families; and,
• FGDM is likely to be cost neutral or provide savings.

1.4.77 However, there are some persistent problems associated with the approach and a generalised uncertainty regarding its efficacy in bringing about positive long-term outcomes. Many of these problems allow obvious scope for further research and clarification. Specifically, these are:

• the effect of selectively adhering to certain core FGDM principles is unknown;
• there is some evidence that family members, particularly children, may feel marginalised and detached from the FGDM process;
• FGCs have the potential to be upsetting and negative experiences for some participants;
• professional remained concerned about the potential for conflict during private family time and often doubted whether all pertinent issues were discussed;
• there is considerable ambiguity regarding adherence to plans, but where monitored FGC plans are not adhered to by family members or by social welfare agencies;
• there is a lack of truly robust research evidence stating that FGDM results in positive outcomes for children;
• in common with FGDM in child welfare, there is no robust evidence that FGDM results in positive outcomes for children in the youth justice and education fields; and,
• there is a lack of reliable evidence on the cost effectiveness of FGDM.

1.4.78 The following chapter continues the interpretation of this material, identifying a number of important practical themes about which there is considerable debate. In relation to these issues, effort will be made to examine the relevant literature and, if possible, identify best practice.

« Previous | Contents | Next »