The Link between Assessment and Decision in Child Protection

Background and Purpose

The Decision Making Ecology (DME) suggests that child protection workers’ decision-making is influenced by case, professional, organizational, and external factors. The General Assessment and Decision Making (GADM) model, which has been combined with the DME, assumes: a) thresholds for decisions, as points on an assessment scale, where a decision about taking action is made. b) That these thresholds are determined by professionals’ experiences/history, and organizational/external factors, while assessments are determined by case factors. The study aims to test this assumption and specify the effect of case, professionals, and organizational factors on professionals’ risk assessment and out-of-home placement recommendation.

Methods

The study used a multifactorial experimental vignette design (factorial survey) to test a theoretical decision-making model including case, professional, and organizational factors associated with professionals’ judgement and decision-making in child protection assessments. The vignettes described situations of potential child neglect, based on seven case factors whose levels were experimentally varied. Out of all possible case factor-level combinations, a fractionalized sample of 54 vignettes was drawn using a D-efficiency-optimizing algorithm. To each respondent a vignette deck of three vignettes was randomly assigned. For each vignette, respondents determined a) risk assessment (DV1, 7-point scale), b) the likelihood of recommending an out-of-home-placement (DV2, 6-point scale). Several survey items were used to measure professional and organizational factors.

The study aimed for a full-population online survey of professionals doing child protection assessments in German-speaking Switzerland, given the willingness of agencies/professionals to participate. 542 professionals (63% of invited professionals) from 158 agencies completed the survey.

The hierarchical experimental survey data was first analyzed using various multilevel regression models. Second, statistically significant factors from these analyses were included in Structural Equation Models and Bayesian multivariate multilevel models. These models contained the following paths: a) multilevel factors to DV1, b) DV1 to DV2, c) multilevel factors to DV2.

Results

The influence of the seven case factors on respondents’ subjective likelihood of recommending an out-of-home-placement (DV2) in the depicted vignettes was partially mediated by risk assessment (DV1):  While most of the less severe case factor levels were fully mediated by DV1, more severe case factor levels were often partially mediated through DV1, having also a direct effect on DV2. The study identified several worker level variables predicting out-of-home placement recommendation, partially or fully mediated through risk assessment. After dichotomization of DV2, no direct effects of worker variables on DV2 could be found anymore. No organizational variable was significant in the multilevel models.

Conclusions and Implications

Other than expected by the GADM model, this study found: 1) case factors’ effects on the recommendation to place a child were not fully, but only partially mediated through risk assessment. Especially more severe case factor levels had also a direct effect on the placement recommendation. 2) Worker variables were not only associated with out-of-home placement recommendation, but also directly associated with professionals’ risk assessment. Studies in different contexts and using different study designs should be conducted to determine, whether the GADM model should be modified.