Predicting Turnover

A number of proxies have been used in child welfare workforce research to represent actual turnover; however, there have been no psychometric studies to validate a scale specifically designed for this purpose. The Intent to Leave Child Welfare Scale is a proxy for actual turnover that measures workers’ intention to leave. This scale was validated in the current study by a CFA. The resulting factors were compared to actual turnover. Nearly two in three workers who indicated that they had considered looking for a job in the past year actually left their agencies (60.0%). A confirmatory factor analysis was conducted to determine the validity of the Intent to Leave Child Welfare Scale. The best fitting model consisted of three factors with acceptable fit statistics ( X 2 = 28.6, p = 0.04; RMSEA = 0.05, 90% RMSEA CI = 0.01-0.08; CFI = 0.99; TLI = 0.98). Identified latent factors included “thinking,” which included observed variables related to workers thinking about leaving their current jobs; “looking,” which included observed variables related to workers searching for a new job; and “acting,” which included observed variables related to workers actually taking physical steps to seek a new job. Once a good fitting model was identified, binary logistic regression was conducted to determine odds ratios to predict who actually left their agencies. Each of the identified latent factors was significantly predictive of actual leaving (thinking: OR = 1.24, p = 0.00; looking: OR = 1.25, p = 0.00; acting: OR = 1.28, p = 0.01).