Organizational Factors in Using Child Welfare Data

Purpose: As evidence-based practice has been promoted in the child welfare field to produce better outcomes for children, child welfare agencies have recognized the importance of using data for decision making (Shackelford, Harper, Sullivan, & Edwards, 2007). A data-driven decision-making process is also crucial to effective implementation of new practices (Nation Implementation Research Network, 2008-2012). However, little is known about what organizational factors influence child welfare workers’ data use and whether or not they actually use data in their day-to-day decisions. Therefore, the current study aims to determine which organizational factors are related to child welfare workers’ data use.This study adopted the PRECEDE portion of the PRECEDE-PROCEED model to identify organizational factors that are associated with child welfare workers’ data use in their day-to-day work, and to test the hypothesis that perceived importance, familiarity, support from colleagues and supervisors, allowed time, resources, accessibility, and usefulness are predictive of actual data use in child welfare workers’ day-to-day work. According to the PRECEDE model, which is useful in examining factors that predispose, reinforce, and enable the targeted behavior (Doughty, 2011), the variables that would explain data use behavior change were constructed into three factors: (1) predisposing factors; perceived importance and familiarity, (2) reinforcing factors; support from colleagues and supervisors, allowed time, and resources, and (3) enabling factors; accessibility and usefulness.

Methods: The Child Welfare Data Use Survey was administered in 2011 as part of the data training element of the North Carolina Reaching Excellence and Accountability in Practice (NC-REAP) project. Data from 237 respondents were analyzed using binominal logistic regression. Five-point Likert scale items were used to assess the independent variables. The dependent variable, child welfare data use in day-to-day work, was dummy coded into non-use and use categories.

Results: The overall model was significant (X²₍₈₎ = 169.36; p <.001). Analysis of parameter estimates showed that familiarity (Exp(B) = 18.97, p <.001, 95% CI: 8.31, 43.34) and support from supervisor (Exp(B) = 2.29, p=.007, 95% CI: 1.25, 4.20) were significant predictors of data use. Child welfare workers who are familiar with data were more likely to use data in their day-to-day work. When child welfare workers received more support from their supervisor, they were more likely to use data in their day-to-day work.

Conclusions and Implication: The results underline that the PRECEDE portion of the PRECEDE-PROCEED model is useful in identifying organizational factors that support data-driven decision-making among child welfare workers. These findings are also consistent with established principles of implementation science related to systems change, including the importance of facilitative administration and the need to provide training to increase familiarity (NIRN, 2008-2012). Child welfare organizations should devote attention to methods that make workers increasingly familiar with data, and focus on training or educating supervisors so that they can support and facilitate workers’ use of data.