Child Welfare Worker Characteristics That Predict Service: Examination of the Substantiation Decision

Background and Purpose: Child welfare workers are the conduit between the system and families served by the state. It is assumed that child welfare workers have certain characteristics, yet there is a dearth of literature about what skills, knowledge, and experiences child welfare workers require to provide effective services. In an effort to build the body of literature that connects worker characteristics and service, this study examines what clinical and worker factors predict the substantiation decision.

Methods: Secondary data analysis of the Canadian Incidence Study of Reported Child Abuse and Neglect (2008) was used to answer the questions: What clinical and worker factors predict the substantiation decision?; and Does the substantiation decision differ among workers? Multi-level logistic modeling was used to examine which clinical (i.e., child, caregiver, household, case) and worker (i.e., education – level and field, age, training, experience, gender, caseload, ethnicity, position) predicted the substantiation decision.

Results: Results of the final two-level logistic model indicate that the majority of factors that predicted the substantiation decision were clinical. Children and caregivers with functioning concerns were more likely to be substantiated than those with no functioning concerns. Caregivers with social supports and no previous involvement, households who regularly run out of money, and who were deemed not cooperative were significantly more likely to be involved in substantiated allegations than their respective counterparts. Reports from a professional referral source were significantly more likely to be substantiated than referrals from a non-professional source. The only worker factor that significantly predicted the substantiation decision was their caseload size; workers with a caseload between zero and 13 were significantly more likely to substantiate allegations than workers with more than 13 cases. Goodness of fit test results suggested that the two-level model fit significantly better than the single level, indicating dependence within the data. The intraclass correlation coefficient score suggested variance at the worker level, indicating differences between workers’ decision to substantiate maltreatment allegations.

Implications and Conclusions: Findings that suggest caseload size may have an impact on the decision to substantiate allegations are consistent with the literature. This study, however, is unique in its use of multi-level modeling and consideration of clustering in the data. Non-significant results may illustrate similarity in decisions based on characteristics typically assumed to be unique, such as education field and level, and years of experience. Further research linking worker characteristics to their decisions and service outcomes are required. This study illustrates similarity in the substantiation decision between workers with different education fields and levels, years of experience, and training; areas assumed to be unique in their ability to equip workers for effective practice.