Family Group Decision Making: Measuring Fidelity to Practice Principles in Public Child Welfare

Background and Purpose: Increasingly public child welfare systems in the United States and other countries seek to engage family groups in decision making about their young relatives. Although family group engagement is compelling both ethically and conceptually, additional research is needed to conclusively establish its efficacy in terms of child safety, permanence, and well-being. A critical element is to first determine model fidelity in order to make the connection between programs and their outcomes. The current study aimed to revise and to examine the factor structure of a self-administered version of the Achievement of Family Group Conferencing Objectives Scale. It fills a critical gap in the field of child welfare practice by testing both the psychometric qualities of the scale as well as the feasibility of broader monitoring.

Methods: A short, self-administered, 17-item fidelity measure was developed and used to capture the perspectives of a range of individuals involved in FGDM. Participants included 5,456 parents (n=923), youths (n=314), relatives (n=1751), friends (n=507), professionals (n=1435), and others (n=526) who participated in FGDM conferences in Pennsylvania between 2010 and 2011. The sample was randomly and evenly divided into two subsamples of 2,728 participants. The first half was subjected to an exploratory factor analysis (EFA) using an oblique rotation. After developing an initial factor structure based on the EFA for the first half of the data, a confirmatory factor analysis (CFA) was used on data from the second half of participants in order to compare model fit for the two competing models suggested by the exploratory factor analysis.

Results: Results of the EFA suggested either a two- or three-factor solution, with ten items loading on Factor 1, three items loading on Factor 2, and four items loading on Factor 3. Results of the CFA mirrored the results found under the three-factor solution produced by the EFA. The data were not multivariate normal, z=86.88, p<.001. Maximum likelihood with robust adjustments was used. Findings from the CFA supported a three factor model that includes sets of practices yielding: (1) Productive Decision Making; (2) Family Group Inclusion; and (3) Professional Supportiveness, with the highest correlation observed (.81) for Factor 1 and Factor 2 and the lowest (.55) between Factors 2 and 3.

Conclusions and Implications: These results provide direction to public child welfare practitioners and policymakers about the importance and the feasibility of attending to multiple perspectives in assessing program delivery. Rather than the professionals making assumptions, this measure offers a clear way of asking participants what happened before, during, and at the end of the meeting, who was there, and how the professionals engaged with the family and interacted with the group. This study also illustrates how to refine an instrument originally based on principles so that it can be understood by a range of participants and broadly distributed and managed with limited administrative support. Results provide valuable guidance about the importance of attending to multiple perspectives in assessing program delivery and support the feasibility of incorporating fidelity measurement into standard practice.