Bridging Disciplinary Boundaries (January 11 - 14, 2007)

The national problem of low employee retention rates in child welfare (CW) is well documented. High turnover creates difficulties in employee recruitment and retention, and providing quality of services for children and families. Though several large studies of CW employee retention have been completed, this study is the only known application of hierarchical linear modeling (HLM) to better understand this national problem.

Purpose: To examine the extent to which CW employee intentions to remain employed in CW can be explained by county office (organizational level) versus personal (individual level) variables.

Methods: A 198 item, statewide survey of all public CW staff in one large state was completed (fall 2002) that included original measures of intent to remain employed (IRE) in CW; self-efficacy (SE) and efficacy expectations (EE) about work tasks; human caring (HC); job satisfaction (GJS); work morale (WM); and professional organizational culture (POC). 1423 surveys (63.2%) were returned. Alpha reliabilities for factored dimensions of these measures ranged from .52 to .94.

A two-level HLM approach was used in the data analyses (Raudenbush & Bryk, 2002). The IRE measure was used as the dependent variable. A level-1 model regressed individual IRE scores on seven individual level (independent) variables: age; education level; years of work experience; SE; HC; EE; and GJS. A level-2 model assessed influence of county level variables (WM and POC) on the regression slopes for IRE and each individual level variable for the total sample, and African and White American sub samples.

Research Questions: 1) Are there differences in IRE scores among CW county offices? Do… 2) county level variables explain differences in IRE scores? 3) individual level factors explain differences in IRE scores? 4) county level variables influence the magnitude of the effects of individual level variables on IRE?

Results: The results showed statistically significant (p<.001) variation among CW county offices in IRE for the total sample and both sub samples. Maximum likelihood estimates showed African Americans were more variable in IRE than the total sample and the White American sample. Coefficients for the model intercepts of IRE and the independent variables showed statistically significant (p<.01) effects except for SE for the total sample, SE and EE for the White sample, and age for the African American sample. Cross-level interactions showed that HC and GJS varied significantly among county offices and that WM clearly influenced relationships between IRE and the individual level variables.

Implications for Practice: The results have implications for CW employee selection, supervision, and retention. The finding that organizational variables influence linkages between individual variables (e.g., HC) and IRE has implications for educating CW leadership staff to carefully select new employees who possess strong personal characteristics (e.g., HC, SE), and to mentor and support new employees as a means of strengthening job satisfaction and work morale. Such efforts are important keys to strengthening employee retention. These perspectives and additional implications of the findings for practice will be discussed in the complete paper.