Child Welfare Worker Turnover: Understanding and Predicting Who Actually Leaves
Nancy Dickinson, PhD, University of North Carolina at Chapel Hill, John Painter, University of North Carolina at Chapel Hill, and Jung-Sook Lee, MSW, MA, University of North Carolina at Chapel Hill.
The persistent shortage of a competent child welfare workforce has hampered the ability of agencies to provide effective services to families and children. Recruitment, selection and retention issues faced by child welfare agencies are numerous and readily acknowledged throughout the professional literature. Yet, there is a lack of rigorous research about the nature of the child welfare workforce crisis and initiatives to address the issues. Data for this presentation are from an experimental study about child welfare worker retention in a mid-sized state. All child welfare workers (785) from 33 randomly selected county departments of social services were invited to complete a web-based survey about perceptions, characteristics and attitudes shown to be related to retention. Workers were also asked how likely they were to leave their positions within the next six months (“intent to leave”). With a response rate of 45%, 356 surveys were suitable for analysis. The survey measured workers' perceptions about the domains of their work, supervisor, and agency, all shown in the literature to be related to intent to leave. These domains were in turn defined by twelve scales designed to measure specific aspects of three domains. The work domain is represented on the survey by the scales of depersonalization, desire to help, self-efficacy, match of skills to work, and pre-employment job portrayal. Supervisor domain is represented by scales of practice support, mentoring, and emotional support. Agency domain is represented by scales relating to affirmation and recognition, salary and benefits, concern for worker safety, and learning organization. Scales were validated at the item level using confirmatory factor analysis (CFA), and composite scores were created for each scale. Two path models tested the relationship of each domain to “intent to leave.” Each path analysis model displayed adequate fit (GFI>.95 and/or RMSEA <.05), and two were significantly related to intent to leave. Cluster analysis was then performed using the twelve subscales. Three groups accounting for over 98% of the sample were identified and labeled “strong,” “average,” and “weak” stayers. ANOVA with preplanned comparisons found that overall the groups were significantly related to intent to leave (p<.001, R-SQ = .21).and were also significantly different from one another with respect to intent to leave (p<.001). Finally, surveys were linked to administrative data from county human resources databases used to track actual child welfare worker turnover. The relationship between “intent to leave” and actual turnover during the next two years was verified, providing an estimate of the external validity for the measure of “intention,” a key contribution to retention research. Further research will focus on cross validating results of the CFA and cluster analysis and conducting multigroup path analysis to understand more clearly the differences between cluster groups. Findings from this research further clarify methods to measure the complex set of factors contributing to retention. Findings have implications for measuring the mechanisms and outcomes of such interventions as supervisor training, improved selection methods, and changing agency culture to better support worker retention.