Measuring the Impact of Public Perceptions on Child Welfare Workers

Background and Purpose: High turnover rates within the child welfare workforce threatens the psychological well-being and permanency outcomes of children and disrupts the quality of care and service delivery (Strolin-Goltzman, Kollar, & Trinkle, 2010). To better understand how the larger social context influences workers and their intention to remain employed in child welfare, the Perceptions of Child Welfare Scale (PCWS) was developed. Psychometric studies have validated the PCWS with private agency child welfare workers (Auerbach, Zeitlin, Augsberger, Lawrence, & Claiborne, 2016); however, the validity of the PCWS has not been assessed with public agency staff.  Because research shows substantive differences between these two types of workers (Kim, 2011), this study addresses the need for measures that are valid with public child welfare staff by examining the psychometric properties of the PCWS with public child welfare workers.

Methods: Data and samples: Data for this study comes from a multi-state survey of public child welfare staff conducted as part of a workforce assessment. A total of 4,250 child welfare staff completed an on-line survey via a secure e-mail link.

Measures: Workers’ perceptions of how the society and family viewed them were assessed using the PCWS. A confirmatory factor analysis (CFA) was conducted to examine the psychometric properties of PCWS. After defining the best fitting model, we tested for criterion validity by seeing how each subscale developed and then the instrument was related to intention to leave the agency.

Results: Results indicated the existence of four latent constructs: stigma, nature of work, respect, and blame. Factor loadings for each exogenous variable were significant at the 0.00-level, and ranged from a low of 0.55 to a high of 0.90. Correlations between latent variables were all also significant at the 0.00-level and indicated good discriminant validity between each of the constructs. While the overall model fit was significant (X² = 869.65; df = 71; p=0.00), the sample size was large. Because of this, supplemental fit statistics were computed to more accurately assess model fit. The Root Mean Square Error of Approximation (RMSEA) was .064 with a 90% confidence interval ranging from .061-.068, the Comparative Fit Index (CFI) was .98, and the Tucker-Lewis Index (TLI) was .97. These values indicated that the identified model with four latent constructs fit the data well.

To assess for criterion validity, five logistic regressions were conducted, one for each of the identified dimensions and one for the overall validated scale. Each of these significantly predicted intention to leave at the 0.00-level with odds ratios ranging from 0.56 to 0.67 for each of the subscales. The odds ratio for the total scale was 0.74.

Conclusions and Implications: This study shows that the PCWS is a valid measure for understanding how public child welfare workers believe society views them. Additionally, it is predictive of workers’ intent to leave. The PCWS can be utilized in workforce studies to better understand the thoughts and beliefs of public child welfare staff.