| Abstract: | Abstract Using administrative records, this paper tracks for up to four years using Cox survival methods the retention of 839 public child welfare workers who began child welfare work in 1999. It also examines, using ordinal logistic regression, supervisory performance evaluations of 382 of these workers. Participation in a IV-E-funded social work educational program predicted better retention. In particular, risk of termination decreased by 52% during the mandated contractual employment period in which the educational stipend was “worked off”. Other predictors of longer retention included prior non-child welfare employment at the public agency and working in the state office setting. Temporary job classification predicted higher risk of termination. With temporary classifications excluded, female gender predicted better retention. In exploratory Cox frailty regressions, neither county of employment nor supervisor explained significant variance in retention. Involvement with the IV-E funded social work program was not associated significantly with supervisory ratings. Regression analyses revealed an association between an ethnic group's representation in the population of child welfare workers and supervisory evaluation; the greater that representation, the higher the overall evaluation for the group. |
