Innovation in Child Welfare: Factors That Influence the Adoption and Sustainability of Family Group Decision Making

Background and Purpose This paper explores the adoption of Family Group Decision Making in Pennsylvania using a mixed methods design and includes a method of examining "neighborhood effects". Specifically it asks: (1) What local and organizational factors influence the adoption of FGDM in Pennsylvania; (2) What do the individuals using FGDM feel are the factors that create barriers or help to facilitate adopting, implementing and sustaining the practice?

Methods A mixed methods design, using geographic autocorrelation modeling and an analysis of qualitative information about adoption and the implementation was employed. County adoption level, the dependent variable, was operationalized as five levels of practice implementation. Data on the independent variables were obtained from Pennsylvania administrative databases and federal sources (2008-2009) and included measures of population density, child population, agency size, maltreatment reports, and county poverty level. Prior funding for a FGDM start-up grant or a system of care grant was also included. A "neighborhood effect" which measures the adoption level of the neighboring counties was developed through a utilization matrix. The rationale for including a "neighborhood effect" as an independent variable was based on models of social influence (Leenders, 2002) and studies of social comparison (Babcock, Wang & Loewenstein, 1996).

A self-administered, web-enabled survey containing open and closed items was completed by child welfare caseworkers, supervisors and professionals who regularly attend family group meetings. Repeated readings created a list of themes which were given descriptive codes, and responses were independently coded by two raters using the Coding Analysis Toolkit.

Results Six models were tested; maltreatment and poverty rates and the size of the population of children were non-significant predictors of a county using FGDM in all models. Having a FGDM pilot grant, a system of care initiative, the number of caseworkers and a neighboring county implementing FGDM were significant, and population density and number of caseworkers were weaker but significant: population density became insignificant once the two largest counties were removed. In the final model, the neighborhood effect was the most powerful predictor (z = 7.112, p≤0.001.), followed by FGDM pilot (z=3.785, p≤0.001) and the number of caseworkers (z=3.320, p≤0.001). Working in a county with a system of care was weaker but still significant (z=2.424, p≤.05). Limitations are that causality and the nature of the social influence cannot be determined. Individuals from the established and the new adoption counties both felt that FGDM was facilitated through collaborating with other systems and that the strengths-based approach of FGDM helped to facilitate implementation. Established adopters placed more emphasis on the importance of leadership and case worker attitudes, whereas newer adopters felt that additional resources in the form of staff, training and funding were the critical factors in adoption.

Implications To facilitate the adoption of new practices, funders should look to strategically place new programs in close proximity to established programs, provide start up funding and utilize networks established through system collaboration activities.