Economic and Environmental Correlates of Public Child Welfare Worker Turnover
Gingi Fulcher, MSW, University of California, Berkeley and Richard Smith, MFA, MSW, University of California, Berkeley.
Background and Purpose: This study examines the relationship between county-level environmental and economic factors and the turnover rate of public child welfare workers in California counties (N=58). This relationship is important because in a review of 62 family reunification cases, Hess, Folaran, & Jefferson (1992) found that turnover and caseload size were factors in explaining 42 case failures. GAO (2003) also suggests that turnover hinders the formation of quality, client-worker relationships, which are needed in making appropriate child permanency decisions under mandatory timelines (PL 105-89) (GAO, 2003). The literature documents individual, organizational, economic, and environmental factors that are determinants of turnover (e.g. Cotton, 1986; Price, 2002; Steel & Griffeth, 1989). In particular, the construct opportunity, sometimes operationalized by the unemployment rate, moderates individual and organizational variables. Yet in Mor Barak's (2001) meta-analysis, none of 55 studies social worker turnover studies investigate economic or environmental factors. Similarly, the child welfare turnover articles reviewed by Zlotnik et. al. (2005) excluded such factors. Only recent studies in social work consider the economy (e.g., Barth, 2003; Weaver & Chang, 2004). This study helps close this large gap in child welfare literature by testing two hypotheses: 1) Do neighboring counties share similar turnover rates? and 2) Is child welfare worker turnover higher in counties with a good economy? Methods: Unlike previous studies, ours models the dependent variable turnover rate at the county-level (number of child welfare workers who left in 2004/total number of child welfare workers) rather than individual exits. The turnover data is from a workforce study on child welfare agencies (Clark & Fulcher, 2005) Independent variables are from the US Census and county child welfare indicators, included as a proxy measures for environmental factors because earlier research found them associated with neighborhood poverty and unemployment rates (e.g., Coulton et. al., 1995; Steinberg, et. al., 1981). Data are analyzed with exploratory spatial data analysis (ESDA), bivariate correlations, multivariate regression, and spatial statistics (Freisthler, et. al., In Press). Results: First, our study found that eleven counties in the San Francisco Bay Area, which is well connected by rail and bus transit, had low child welfare turnover rates and were near other counties with low rates, a phenomena known as “low-low local spatial autocorrelation” (Moran's I = 0.0324; p < .05). Second, we found bivariate correlations between the turnover of child welfare workers and the following economic and environmental factors: (1) population density, (2) percent employed in manufacturing, and the (3) education level of the county population. Finally, we found an association between child welfare worker turnover and county-level measures of child maltreatment. Conclusions and Implications: At a given point in time, child welfare worker turnover can be correlated with the level of other work opportunities in the commuting area. Future research should utilize a longitudinal, time series design in order to make attempts at inferring causality. Furthermore influential factors and interactions need to be estimated in a in a multilevel model. Because this study relied on previously collected data, analyses were limited.