Family group conferencing in child welfare: responsive and regulatory interfaces.

A regulatory approach compels the child welfare worker to make decisions according to set procedures and prevents responding flexibly to families. Differential response is a way that child welfare is departing from legal formalism. One means is convening a family group conference (FGC) to develop a plan. John Braithwaite's regulatory pyramid assists in conceptualizing differential response. This article reports a factor analysis of data on achievement of FGC objectives to elaborate three interfaces for fostering responsive regulation. Each interface keeps the family group at the center of planning while firmly maintaining their connections with community and government programs.

Key words: family group conferencing, responsive regulation, child welfare, differential response

**********

A well circulated cartoon shows a child welfare worker being lynched by an angry mob. The caption for one frame reads, "Social worker who took child into care," and the caption for the other reads, "Social worker who did not take child into care." No matter what the social worker decides, the populace is provoked to take the law into its own hands. Such vigilante justice is in reaction to seemingly arbitrary authority. It condemns the social worker and in all likelihood the child's parents and leaves the child in need of protection. This gallows humor will continue to resonate as long as public child welfare is defined solely as saving children from their parents. Some children need such rescuing but far more need supports and protections that safeguard them and their families.

The doctrine of parens patriae, however, obligates the state to substitute as parent when the child's own parents fail to protect because of their personal limitations or those of the wider society. This same doctrine has raised legitimate concerns about the abrogation of the rights of parents and children and has led to an emphasis on due process through the courts. The result is heightened legalism interacting with the state's liability for the child's safety. This combination chokes off opportunities for child welfare to join with the family and community in forming partnerships of caring.

John Braithwaite's (2002) theory of responsive regulation points to a viable alternative for child welfare. He posits a regulatory pyramid with a broad base of responsiveness to offenders underneath an apex of legal regulation. Because the state is charged to care for children in need of protection, child welfare must maintain a firm interface between responsiveness and regulation. This article examines how family group conferencing (FGC) achieves this interface. First, an overview is provided of the movement toward responsive regulation in U.S. welfare, the role that FGC can play in promoting responsive regulation, and its key practices in child welfare settings. Then, utilizing a factor analysis of findings from a FGC study, three interfaces are elaborated--family leadership, cultural safety, and community partnerships. In conclusion, a model for interfacing responsiveness and regulation in child welfare is presented.

Responsive Regulation and Child Welfare

Child welfare in the United States has a lengthy history of swinging between a priority of child safety or family support (Jimenez, 1990). Child safety stresses the state's responsibility to regulate the child's care and ensure that it meets adequate standards of protection; family support urges a responsive approach to children and their caregivers to promote healthy families. While both regulation and responsiveness are necessary for safeguarding children, neither approach alone is sufficient for an effective child welfare system (Pecora, Whittaker, Maluccio, & Barth, 2000). Family support is limited when caregivers will not, or more often cannot, change their practices on their own; community services are lacking, inaccessible, or under utilized; and the broader economic and political systems undermine families (Pecora, Reed-Ashcraft, & Kirk, 2001).

The prevailing approach, though, is reliance on regulation, that is, forensic investigations, court hearings, strict timelines for termination of parental rights, registering abusers, and so forth. With precedence given to following legal procedures, workers are preoccupied with defensible rather than helpful implementation, parents feel stigmatized by service, and the effects on children are deleterious (Parton, 1997). Significantly, children of culturally marginalized groups in the United States are disproportionately represented in the foster care system (U.S. DHHS, 2000). All of this serves to alienate workers from their clients and communities and to increase the workload and responsibility of public child welfare beyond its capacity.

As a counter measure, some U.S. states have passed legislation permitting a differential response, variously known as dual track, multiple response, and alternative response (National Child Welfare Resource Center, 2001). This allows child protective services to adopt more than one method or "track" of handling reports of child abuse and neglect. States have at least two tracks for responding: an investigation track for substantiating child maltreatment in the more severe situations and mandating interventions and an assessment track for determining need and involving families from the outset in finding solutions. All cases screened-in meet statutory definitions of child maltreatment--in other words, child protective services cannot "walk away" from these families (North Carolina Division of Social Services, 2002, p. 7). Differential response is not intended as a means of widening the net of child welfare cases. It is intended to concentrate legal interventions on the cases that truly warrant such a response and in the other cases to engage families in services and foster supportive community networks.

Concerns frequently raised are that workers will place families in the wrong track, not transfer cases to the investigation track when necessary, and, thus, endanger the safety of children. The preliminary findings are that approximately one-quarter of families are placed in the investigation track and about three-quarters in the assessment track and that cases do not usually change tracks (National Child Welfare Resource Center, 2001). Social services workers and collateral agencies prefer this way of working with families (Virginia Department of Social Services, 1999). Reports and re-reports of child maltreatment decrease while reported incidents for which action is taken increase, children are removed from their homes at the same rate but spend less time in placement, and the safety of children does not appear to be compromised despite workers' large caseloads and limited resources (Institute of Applied Research, 1998).

Family Group Conferencing on Multiple Tracks

Braithwaite's (2002) theory of responsive regulation was developed by integrating his work on business regulation and criminal justice. With some qualification, this theory provides a way of conceptualizing and expanding differential response in child welfare. The base of his regulatory pyramid is about responding flexibly to all cases, and this is achieved through using restorative justice processes. The aim is to repair the harm caused by the offense, involve the key stakeholders in deliberations, and transform community and government relationships (Schiff & Bazemore, 2002). Except in extreme cases, Braithwaite recommends that restorative practices are first applied and then only if offenders refuse to comply, is recourse to the law and courts sought. Once offenders begin to comply, the approach moves down the pyramid from legal interventions to restorative processes.

Braithwaite's prescription would raise grave fears in child welfare, or for that matter in domestic violence, where the victims reside with their abusers and may require immediate interventions backed by the force of the law. In these contexts, the tension between regulation and responsiveness cannot be relaxed. Nonetheless, his notion of applying restorative justice processes early in serious as well as more moderate cases is worth pursuing. If child welfare clients in whatever service track have a voice regarding their plans, interventions are more likely to respond to their conditions and cultures and regulate their actions and interactions.

One way to promote families' voices in child welfare is by using family group conferencing. The "family group" is composed of the family members along with their relatives, friends, and other close supports, and the "conferencing" refers to holding a decision-making forum to resolve areas of concern. This restorative justice process is now applied in child welfare as well as other arenas including schools, youth justice, and adult services in many countries and cultures (Burford & Hudson, 2000). Like restorative processes generally, family group conferencing (FGC) reflects traditional practices and, in particular, decision making among the New Zealand indigenous people the Maori (Love, 2000) and, more broadly, South Pacific islanders (Shook, 1985). The approach was first legislated in the New Zealand 1989 Children, Young Persons and Their Families Act. This law emphasizes the family group's responsibility for their young relatives, children(s safety and rights, respect for cultural diversity, and community-government partnerships for the benefit of children and young people (Hassall, 1996).

FGC is not uniformly implemented in child welfare (Merkel-Holguin, 2000), but desirable practices can be specified (Burford, Pennell, & MacLeod, 1995; Marsh & Crow, 1998; North Carolina Family Group Conferencing Project, 2002; Paterson & Harvey, 1991). A family is referred to a FGC usually by their social worker. The referral is sent to a FGC coordinator, who is responsible for organizing and convening the conference but who does not assume case-carrying responsibility. By only having one role with the family, the coordinator avoids confusing the family group and helps them focus on how to plan the conference rather than negotiating other services.

In advance of the conference, the FGC coordinator explains the process to the family members and emphasizes that they have the option of whether to attend or not; collaboratively develops an invitation list of who are family and 'like family'; checks out the family group's wishes on the meeting including its place, timing, food, and opening; assesses potential risks for participants and builds in needed supports and protections; arranges travel, child care, and other logistics; and generally prepares family group and service providers to take part safely and effectively.

The service providers often need help in understanding their roles. The FGC coordinator stresses that these do not change at the conference. The child protection workers retain their responsibility for child safety and have the final say over whether the FGC plan goes into effect. The other service providers, such as a substance-abuse counselor or domestic violence advocate, are there to share information that will help the family group produce the best plan possible. The coordinator assists the service providers on how to present information in a way that is understandable and respectful of family group.

The conference can be viewed as having distinct stages: the opening, information sharing, family's private time, finalizing the plan, and closing. To signal that the conference belongs to the family group, it is commonly held in a community center and opens in a way fitting the family's traditions. This may mean a welcome by a senior family member, a prayer, or simply choosing one's own seat. The FGC coordinator ensures that all participants are introduced and informed about the purpose of the conference and its format. Then the service providers and family group overview the situation along with possible resources for remedying it. If the child welfare worker has "bottom lines" or matters that are not open to negotiation such as keeping a sexual abuser away from a child, then these should be clearly stated at this time. Once the family group has sufficient information about the concerns to be addressed, the service providers including the FGC coordinator leave the room.

This is the start of the family group's private time to formulate a plan. During this period, the family group often lends support to members and challenges them to change their behaviors and almost always comes up with a plan. Typically the family group members select professional services but also offer to contribute their homes and other resources. After developing a plan, the family group invites back the service providers. The child welfare worker and other involved authorities review the plan to ensure that it addresses the safety and care issues, approve the action steps, and authorize the allocation of public resources. In closing, conference participants may say their good-byes or have a more elaborated ceremony. After the conference is the work of carrying out the plan. This is facilitated if the FGC plan includes a clear system of monitoring and evaluating implementation and reconvening the group as needed.

As can be seen, FGC is labor intensive in its preparations and deliberations and intrusive in the sense of conveying so much confidential information to a larger group. Given the nature of FGCs, workers should have very solid reasons for making a referral. Although usefully applied to plan family supports, FGCs tend to be convened even more frequently in the difficult situations where workers are uncertain about how to proceed and important decisions must be made such as whether to place a child outside the home (Hudson, Morris, Maxwell, & Galaway, 1996; Marsh & Crow, 1998; Pennell & Burford, 2000; Trotter, Sheehan, Liddell, Strong, & Laragy, 1999). These cases are usually substantiated and involuntary and already involved with the legal system.

FGC is a restorative process that is readily applicable to multiple tracks in a differential response system in child welfare. By amplifying the voice of the family group, more responsive interventions can be designed on whichever track the family is assigned. If the family is involved with the courts, the FGC plan can be referenced by the judge at the time of disposition or sentencing. If the family requires assistance from relatives, the community, and public agencies, the plan serves to coordinate these services according to the expressed wishes of the family group. Because the plans must be approved before they are implemented, the mandatory authorities retain their legal role while responding to the family group and community.

Repeated studies have clearly demonstrated that FGC participants like the process (Cashmore & Kiely, 2000; Marsh & Crow, 1998; Pennell, 2002a; Pennell & Burford, 1995; Trotter et al., 1999; Unrau, Sieppert, & Hudson, 2000; W. R. McDonald, 1999). They are satisfied with how the conferences are run, the decision process, and the resulting plans. Although both family group and service providers rate the process highly, the former are more enthusiastic and prefer FGC to other child welfare decision approaches (Marsh & Crow, 1998; Trotter et al., 1999). Some child welfare workers are more skeptical about the appropriateness of FGC plans as compared with those generated at other child welfare meetings (Trotter et al., 1999). The workers' position reflects more general questions in child welfare circles about family group dysfunction and the worker's and agency's liability if the plans go awry (Lupton & Nixon, 1999; Sundell, Vinnerljung, & Ryburn, 2002).

Despite some workers' trepidations, the outcomes of FGC appear promising. The preliminary findings indicate a greater likelihood of children staying with their parents or kin, siblings kept together, placements stabilized, child maltreatment and domestic violence reduced, and a sense of family pride enhanced (Crampton, 2001; Marsh & Crow, 1998; Pennell & Burford, 2000; Shore, Wirth, Cahn, Yancey, & Gunderson, 2002; Sundell, 2000; Walter R. McDonald, 2000).

Deviations from key practices of the model however, can affect its outcomes and capacity to establish a responsive and regulatory approach in child welfare. To assess for such divergences, the North Carolina Family Group Conferencing Project developed a series of key practices or "objectives" to be realized for each conference. These objectives and their measurement are described next.

FGC Objectives

To guide practice, the model was specified as a series of principles and their related steps. The use of principles helps to prevent over-prescription (Henggeler, Schoenwald, Borduin, Rowland, & Cunningham, 1998) that can limit the flexibility necessary in applying FGC in diverse contexts. This practice guidance was translated into a measurement instrument called "Achievement of FGC Objectives" so that adherence to the model can be assessed (Pennell, 2002b). The questionnaire has 25 items that are scored on a scale of "strongly disagree," "disagree," "agree," and "strongly agree" with space for "don't know" and "not applicable." The last two along with no response were coded as missing datum.

The questionnaire was used by the North Carolina Family Group Conferencing Project as a means of assessing model fidelity and a guide for improving FGC implementation, training, and policy. During the course of the Project, the instrument was completed by 151 participants from 30 conferences. These participants, in rounded percentages, were 60% family group members, 23% FGC coordinators, 16% research observers (who observed conferences where permission was granted by all participants), and 1% service provider. The FGC coordinators and research observers filled out the instrument on their own shortly afterwards. On average about one month after the conferences, consenting family group members scored the questionnaire during an interview, usually by telephone.

The majority of respondents in the three categories completed all items but with the family group members having the least number of missing data. This shows understanding of the process on the part of the family group. The research observers were likely to give a "don't know" on conference preparations that they would not have observed. The FGC coordinators circled "not applicable" for items that they thought were not relevant to a particular conference. The family group members reported "don't knows" particularly on items related to the social services' agency or work taking place after the conference.

A detailed report of the findings and their implications for practice can be found in Pennell (2003). The general finding was that for the most part FGC participants saw their conference as achieving its objectives but with some variation in responses. This finding is congruent with both the diligence of the FGC coordinators in carrying out the preparations and the high level of satisfaction expressed on the evaluation forms distributed at the conclusion of the conferences and in the qualitative feedback provided during the after-the-conference interviews. Given that conferences were more or less implemented according to the model, the data from the North Carolina FGC Project provide a means of uncovering the factors or main forces underlying its key practices or objectives.

Underlying Model Factors

What are the model's underlying factors? In order to address this question, a factor analysis was carried out of FGC participants' views of the extent to which the key practices or objectives of FGC were achieved at their conference. The aim was to see if their views on these objectives coalesced into associations and could be reduced to a smaller number of underlying factors. The assumption is that their pattern of correlations can be explained by these factors shared in common by groups of variables. Because the objectives are steps in the model, conceptually they can be viewed as not causing each other and thus are amenable to factor analysis which looks for commonalities among variables rather than causal paths between variables.

In this study, the factor analysis can be characterized as a heuristic device for specifying patterns among variables (Kim & Mueller, 1978). Although the author did not anticipate the pattern of correlations among the objectives, the instrument for measuring achievement of FGC objectives is based on theory and practice guidance developed out of the author's long-term experience with FGC and its study. The author conceptualizes FGC as "widening the circle" of those committed to safeguarding children and other family members (Pennell & Burford, 1994) and theorizes three ways of achieving this end (Pennell, forthcoming 2004):

* Family leadership a relationship in which the family group members are central and their efforts are supported by community organizations and public agencies

* Cultural safety--a context in which family members can speak in their own language, express their values, and use their experiences and traditions to resolve issues

* Community partnerships--a local collaboration in which each partner retains its distinctive role while striving to realize common goals.

A factor analysis requires that correlations have the same sample size. In order to perform the factor analysis, nine questionnaire items with extensive missing data were initially removed. Priority was given to retaining items answered for the most part by family group members. Later two more items were removed which did not pertain to the extracted factors and had missing values. This left a total of 14 variables in the final analysis. After their removal, some cases continued to have missing data, they likewise were deleted with the total sample size reduced from 151 to 111. Despite these reductions, the overall profile of the respondents remained similar to the original sample. The retained cases were as follows in rounded percentages: 59% family group, 28% FGC coordinators, 13% research observers, and 1% service provider. These respondents came from all of the original 30 conferences.

The first step in the factor analysis is to compute a matrix of correlation values to "load" into the factor analysis. Because the measurement scale for the objectives was ordinal, the polychoric correlation was selected in the Statistical Analysis System (SAS). The polychoric correlation looks at the data as if their ratings were made on a continuous scale, instead of strictly in the ordinal categories of "strongly disagree," "disagree," "agree," or "strongly agree." The polychoric analysis assumes that respondents in one ordinal category had a range in views rather than being tied. From this premise, it asks what continuous distribution would be needed in order to derive the same groupings once all the answers are placed into categories.

From correlations among the 16 objectives, principal component analysis extracted three main factors whose eigenvalues were respectively 7.55, 1.77, and 1.29, and that accounted for 66.3% of the total variance. Although a fourth factor had an eigenvalue slightly above unity, limiting the extraction to three factors was supported by the scree test. Varimax rotation was utilized to ease interpretation. The orthogonal transformation matrix highlighted the loadings of the objectives on each factor. The eigenvalues for the rotated factors were 3.67, 3.48, and 3.46. The relationships were further explored by assessing the contribution of each objective to its factor's total Cronbach Coefficient Alpha, which is usually employed to check the reliability of a measurement tool such as a questionnaire. Two variables were removed because they were shown to reduce the alpha coefficient for their factor and theoretically did not fit well in the factor. The correlations of the 14 remaining variables with their factors are shown in Table I below. Two of the factors have four items, and the third factor has six items. Correlations of the objectives with their factor ranged from .429 to .693. With one exception, all of the items if deleted would reduce the alpha for their factor. The removal of item 17, however, would increase the alpha very slightly from .761 to .763. It is noted that each of the three resulting scales has a Cronbach Coefficient Alpha near 0.8, which is well above the usual norm of 0.6. Thus, the reliability of the scales is assured.

Review of the objectives in each factor shows convergence with the author's FGC theory on "widening the circle" through establishing cultural safety, community partnerships, and family leadership. The four items in the first factor, labeled "cultural safety," each pertain to holding the conference in a way that feels right to the family group. Three objectives refer to where the conference is held, how it held, and who is invited. Their rightness reflects the family group's values and customs and more broadly their culture. The fourth objective is concerned with having sufficient supports and protections and can be viewed as providing the safety necessary for participation. Whether it fits in the factor on cultural safety is debatable theoretically because it does not relate directly to culture and statistically because it has a lower correlation with the factor than the other items and its presence minimally reduces the total correlation. Awaiting further study, all four objectives were kept together because they are seen as contributing to a culturally safe context in which family members can speak in their own language, express their values, and use their experiences and traditions to resolve issues.

The second factor, "community partnerships," includes four objectives relating to both the family group members and service providers being clear about what they are doing at the conference. Two of the objectives concern the family group--whether they understood why the FGC was held and were adequately prepared for it. The other two objectives relate to the service providers--whether they were clear about their role at the conference and prepared to take part. Such clarity on purpose, process, and function promotes the community partnerships necessary for a local collaboration in which each partner retains its distinctive role while striving to realize common goals.

The third factor, "family leadership," encompasses six objectives that all can be viewed in terms of the conference empowering the family group to make a plan. The first two objectives pertain to the FGC coordinator's relationship with the family group, and both foster the family group's decision-making efficacy. If the coordinator respects the family group members and engages with them solely as conference organizer, they are more likely to be clear about the process and feel acknowledged as decision makers. The next two objectives reference the power relationship between the professionals and the family group and encourage the family group to take charge. If the family group members outnumber the service providers, they are more likely to state their views; if the service providers give information and refrain from dictating solutions, the family group members are more likely to come up with their own plans. The fifth objective concerns a distinctive feature of FGC--the family group's private time. With the FGC coordinator and service providers outside the room, the family group usually can express themselves more freely and develop their own solutions. The sixth and last objective is a notable output of conferencing--the inclusion of family group contributions in the plan. This demonstrates a commitment on the family group's part to continue to assist their relatives after the conference. All of these processes advance the family's leadership by generating a relationship in which the family group members are central and their efforts are supported by community organizations and public agencies.

Interfacing Responsiveness and Regulation

A regulatory approach to child welfare places the onus on the worker to make decisions in accordance with set procedures. As depicted in the lynching cartoon, this approach damns social workers whatever decision they make because they cannot respond flexibly to family situations and as a consequence, provoke community outrage. Child welfare is seeking to move away from legal formalism to a differential response so that they do not have to treat all families the same way. One means of generating a differential response is convening a family group conference (FGC) to develop a plan. This approach transforms a vigilante mob into a community of concern who participate in making and carrying out plans to safeguard children and other family members.

John Braithwaite's (2002) regulatory pyramid assists in conceptualizing this movement in child welfare. He theorizes a pyramid with a responsive base of restorative practices as the first site of recourse and then only if dialogue fails, escalating to the regulatory formalism of the courts and correctional services but de-escalating back to restorative processes as offenders come into compliance. In child welfare, the tension between regulation and responsiveness has to be maintained throughout because the children may need protection from their caregivers with whom they reside or from whom they must remain separated at least for some time. One way to qualify Braithwaite's pyramid is to convert it into a series of concentric circles in which the child's family group is interfaced with government and community. As depicted in Figure I below, at the center is the child's family group, the outer ring is the community and government who both play crucial roles in safeguarding children and other family members, and between the two are three interfaces promoting responsive regulation.

The three interfaces are based on theory and a factor analysis of data from the North Carolina FGC Project. Family leadership encourages the family group's taking initiative in planning, not in isolation and instead with the support of public agencies and community organizations. Cultural safety fosters a context in which the family group can access their traditions to find solutions; nevertheless a child welfare conference always remains bi-cultural in the sense of including the family's culture as well as community standards and legal processes for protecting children. Community partnerships include the family group in safeguarding their relatives without the public agencies or community services' jettisoning their functions. All three interfaces work together to keep a firm and productive collaboration between the family group and their community and government programs.

Table 1
Correlation of Achievement of Objectives with Family Group
Conferencing Factors and Alpha If Item Deleted (N = 111)

Item                                               Correlation  Alpha *

Factor 1: Cultural Safety                                        .761
     (Conference held in the right way for
     family group.)

10.  The conference was held in a place that          .645       .666
     felt right to the family group.

11.  The conference was held in a way that felt       .660       .658
     right to the family group (ex., right food,
     right time of day).

14.  People at the conference were relatives and      .530       .732
     also people who feel "like family" (ex.,
     old friends, good neighbors).

17.  The conference had enough supports and           .451       .763
     protections (ex., support persons).

Factor 2: Community Partnerships                                 .782
     (Family group and service providers clear
     about what doing.)

 2.  Each service provider was clear about their      .612       .715
     role (ex., child protection, counseling).

 6.  The family group understood the reasons for      .534       .755
     holding the conference.

15.  The family group was prepared for the            .560       .745
     conference (ex., got enough information on
     what happens at a conference).

16.  The service providers were prepared for the      .653       .699
     conference (ex., got enough information on
     what happens at a conference).

Factor 3: Family Leadership                                      .754
     (Family group empowered to make a plan.)

 3.  The FGC coordinator was respectful of the        .511       .722
     family group.

 4.  The only job of the FGC coordinator was to       .481       .723
     organize the conference. He/she did not
     have other jobs to do with the family.

12.  More family group than service providers         .429       .743
     were invited to the conference.

19.  Service providers shared their knowledge         .453       .733
     but they did not tell the family group how
     to solve the problems.

20.  The family group had private time to make        .693       .676
     their plan.

21.  The plan included ways that the family           .501       .717
     group will help out.

* Cronbach Coefficient Alpha (raw) is given for each factor. The value
next to each objective is the value of the coefficient calculated with
that objective deleted.

References

Braithwaite, J. (2002). Restorative justice and responsive regulation. New York: Oxford University Press.

Burford, G., & Hudson, J. (Eds.). (2000). Family group conferences: New directions in community-centered child and family practice. Hawthorne, NY: Aldine de Gruyter.

Burford, G., Pennell, J., & MacLeod, S. (1995). Family group decision making: Manual for coordinators and communities. St. John(s, Newfoundland, Canada: Family Group Decision Making Project, School of Social Work, Memorial University of Newfoundland. Available at: http://social.chass.ncsu.edu/ jpennell/fgdm/Manual

Cashmore, J., & Kiely, P. (2000). Implementing and evaluating family group conferences: The New South Wales experience. In G. Burford & J. Hudson (eds.), Family group conferencing: New directions in community-centered child and family practice (pp. 242-252). New York: Aldine de Gruyter.

Crampton, D. S. (2001). Making sense of foster care: An evaluation of family group decision making in Kent County, Michigan. Unpublished doctoral dissertation, University of Michigan, Michigan.

Hassall, I. (1996). Origin and development of family group conferences. In J. Hudson, A. Morris, G. Maxwell, & B. Galaway (Eds.), Family group conferences: Perspectives on policy and practice (pp. 17-36). Monsey, NY: Willow Tree Press.

Henggeler, S. W., Schoenwald, S. K., Borduin, C. M., Rowland, M. D., & Cunningham, P. B. (1998). Multisystemic treatment of antisocial behavior in children and adolescents. New York: Guilford.

Hudson, J., Morris, A., Maxwell, G, & Galaway, B. (Eds.). (1996). Family group conferences: Perspectives on policy and practice. The Federation Press: Annandale, New South Wales, Australia; and New York: Criminal Justice Press.

Institute of Applied Research. (1998, January). Missouri Child Protection Services Family Assessment and Response Demonstration impact evaluation: Digest of findings and conclusions. St. Louis, MO.

Jimenez, M.A. (1990). Permanency planning and the Child Abuse Prevention and Treatment Act: The paradox of child welfare policy. Journal of Sociology & Social Welfare, 17(3), 55-72.

Kim, J.-O., & Mueller, C. W. (1978). Introduction to factor analysis: What it is and how to do it. Beverly Hills, CA: Sage.

Love, C. (2000). Family group conferencing: Cultural origins, sharing, and appropriation--A Maori reflection. In G. Burford & J. Hudson (Eds.), Family group conferences: New directions in community-centered child and family practice (pp. 15-30). Hawthorne, NY: Aldine de Gruyter.

Lupton, C., & Nixon, P. (1999). Empowering practice? A critical appraisal of the family group conference approach. Bristol, UK: Policy Press.

Marsh, P., & Crow, G. (1998). Family group conferences in child welfare. Oxford: Blackwell Science.

Merkel-Holguin, (2000). Diversions and departures in the implementation of family group conferencing in the United States. In G. Burford & J. Hudson (eds.), Family group conferences: New directions in community-centered child and family practice (pp. 224-231). Hawthorne, NY: Aldine de Gruyter.

National Child Welfare Resource Center for Family-Centered Practice. (2001, Spring). Meeting each family's needs: Using differential response in reports of child abuse and neglect. Best Practice/Next Practice, 2 (1), pp. 1-14.

North Carolina Division of Social Services. (2002, August). The NC Child Protective Services multiple response system: Policy and practice manual. Unpublished Document. Raleigh, NC: Author.

North Carolina Family Group Conferencing Project. (2002). Family group conferencing in child welfare: Practice guidance for planning, implementing, training, and evaluation. North Carolina State University, Social Work Program, Raleigh, NC [On-line]. Available: http://social.chass.ncsu.edu/jpennell/ ncfgcp/pracguid

Parton, N. (Ed.). (1997). Child protection and family support: Tensions, contradictions and possibilities. London: Routeledge.

Paterson, K., & Harvey, M. (1991). An evaluation of the organisation and operation of care and protection family group conferences. Wellington, New Zealand: Department of Social Welfare.

Pecora, P. J., Reed-Ashcraft, K., & Kirk, R. S. (2001). Family-centered services: A typology, brief history, and overview of current program implementation and evaluation challenges. In E. Walton, P. Sandau-Beckler, & M. Mannes (Eds.), Balancing family-centered services and child well-being (pp. 1-33). New York: Columbia University Press.

Pecora, P. J., Whittaker, J. K., Maluccio, A. N., & Barth, R. P. with R. D. Plotnick. (2000). The child welfare challenge: Policy, practice, and research (2nd ed.). New York: Aldine de Gruyter.

Pennen, J., with T. Turner & J. Hardison. (2002a). North Carolina Family Group Conferencing Project: Building partnerships with and around families: Final Report to the North Carolina Division of Social Services, Fiscal year 2001-2002. Raleigh, NC: North Carolina State University, Social Work Program, North Carolina Family Group Conferencing Project. [on-line executive summary at http://social.chass.ncsu.edu/jpennell/ncfgcp/NCFGCPExecSummary]

Pennell, J. (2002b). Principles for FGC evaluation. In C. Harper, J. Pennell, & M. Well, Family group conferencing: Evaluation guidelines (2nd ed.) (pp.13-32). Denver, CO: American Humane Association.

Pennell, J. (2003). Are we following key FGC practice? Views of conference participants. Protecting Children.

Pennell, J. (forthcoming 2004). Widening the circle. In J. Pennell, & G. Anderson. Widening the circle: The practice and evaluation of family group conferencing with children, young persons, and their families. Washington, D.C.: NASW Press.

Pennell, J., & Burford, G. (1994). Widening the circle: The family group decision making project. Journal of Child & Youth Care, 9(1), 1-12.

Pennell, J., & Burford, G. (1995). Family group decision making: New roles for "old" partners in resolving family violence: Implementation Report (Vol. MI). St. John's, NF: Memorial University of Newfoundland, School of Social Work.

Pennell, J., & Burford, G. (2000). Family group decision making: Protecting children and women. Child Welfare, 79(2), 131-158.

Schiff, M., & Bazemore, G. (2002). Understanding restorative conferencing: A case study in informal decision making in the response to youth crime. Final report submitted to the National Institute of Justice. Fort Lauderdale, FL: Florida Atlantic University, Department of Criminology and Criminal Justice.

Shore, N., Wirth, J., Cahn, K., Yancey, K, & Gunderson, K. (2002, September). Long term and immediate outcomes of family group conferencing in Washington state. EFORUM. International Institute for Restorative Practices. Available: www.restorativepractices.org

Shook, E. V. (1985). Ho'oponopono: Contemporary uses of a Hawaiian problem-solving process. Honolulu: Institute of Culture and Communication, East-West Center.

Sundell, K. (2000). Family group conferences in Sweden. In G. Burford & J. Hudson (Eds.), Family group conferences: New directions in community-centered child and family practice (pp. 198-205). Hawthorne, NY: Aldine de Gruyter.

Sundell, K., Vinnerljung, B., & Ryburn, M. (2002). Social workers' attitudes towards family group conferences in Sweden and the United Kingdom. International Journal of Child & Family Welfare, 5(1-2), 28-39.

Trotter, C., Sheehan, R., Liddell, M., Strong, D., & Laragy, C. (1999). Evaluation of the statewide implementation of family group conferencing. Victoria, Australia: Department of Human Services.

U.S. Department of Health and Human Services, Administration on Children, Youth and Families. (2002). Child Maltreatment 2000. Washington, DC: U.S. Government Printing Office. Available: ww.acf.hhs.gov/programs/cb/ publications/cm00

Unrau, Y., Sieppert, J., & Hudson, J. (2000). Data collection in a family group conference evaluation project. In G. Burford & J. Hudson (Eds.), Family group conferences: New directions in community-centered child and family practice (pp. 298-311). Hawthorne, NY: Aldine de Gruyter.

Virginia Department of Social Services. (1999, December). Final report on the multiple response system for child protective services. Unpublished Document. Richmond, VA: Author.

Walter R. McDonald & Associates, Inc. (1999). The Santa Clara County family conference model: Strategy for system change. Sacramento, CA: Author.

Walter R. McDonald & Associates, Inc. (2000). The Santa Clara County family conference model year one process evaluation report. Sacramento, CA: Author.

JOAN PENNELL

North Carolina State University