The Child Well-Being Scales as a Clinical Tool and a Management Information System

Abstract

In June 1992, a family services agency in the Niagara region of southern Ontario began implementing and operating a computerized version of the Child Welfare League of America's Child Well-Being Scales [Magura & Moses 1986]. The scales are completed on all cases at the point of transfer from intake to long-term family service, and then again every 120 days until case closure. The scales have been useful in clinical and administrative practice for case planning in aggregate form, service planning, community profiling, and outcome measurement. Decisionmaking and service planning in child welfare are complex and difficult tasks, made more so by the limited number of empirically tested client outcome measures and structured decisionmaking models available to guide practice [DePanfilis & Scannapieco 1994]. As a result, workers' decisions may become idiosyncratic over time, perhaps leading to biases, culturally insensitive services, and a misunderstanding of culturally specific child-rearing practices [Gray & Cosgrove 1985; Rubin 1992; Stein & Rzepnicki 1983]. In the field of child protection, one approach to this problem has been the development of structured risk-assessment models to facilitate practice. Pecora [1991] delineated four types of risk-assessment model. One is the matrix approach representative of the earliest models developed, which uses a table composed of risk factors that are then rated in terms of their severity (e.g., Coler [1982]). The second is the empirical predictor model [Baird 1988; Johnson & L'Esperance 1984], in which a small set of factors predictive of maltreatment recurrence are provided to Child Protective Service (CPS) intake workers. The third is the family assessment scales model, which tends to use behaviorally anchored scales to assess levels of parent, child, and family functioning, with the Child Well-Being Scales (CWBS) [Magura & Moses 1986] being one of the prime examples of this type. The fourth is the Child at Risk Field (CARF) model, which uses an ecological approach, organized around five force fields, (child, parent, family, maltreatment, and intervention) [Holder & Corey 1987]. A fifth type, computerized expert systems, was recently added to this taxonomy [English & Pecora 1994]. The use of risk-assessment models has engendered much controversy. In general, criticism has been leveled at the widespread use of models because of inadequate design, a lack of validation research, and methodological limitations [Doueck et al. 1992; Lyons et al. 1996; McDonald & Marks 1991; Pecora 1989; Wald & Wolverton 1990]. The Child Well-Being Scales, however, are much more than a risk-assessment instrument. The Child Well-Being Scales The Child Well-Being Scales were originally designed to meet the needs of program evaluation in child welfare services rather than to measure individual case outcomes [Magura & Moses 1986). The CWBS consist of 43 behavior-rating scales that are multidimensional measures of potential child maltreatment situations. The items represent four critical dimensions linked to child wellbeing: (1) parenting role performance; (2) familial capacities; (3) child role performance; and (4) child capacities. They focus on considerations common to a broad range of child- and familyoriented services, with a particular emphasis on problems encountered by child welfare agencies. Each item consists of four or five behaviorally specific response categories, with clearly described anchoring points and examples. In addition, each scale point is weighted in terms of a common dimension, the seriousness of the condition. The weightings for seriousness are based on opinions collected from a sample of several hundred practitioners and administrators of child welfare services. Testing of the subscales indicates that three factors-household adequacy (10 scales), parental disposition (14 scales, including measures of parenting skills, disabilities, problem recognition, motivation, and adult/child relationships), and child performance (4 scales) accounted for 43% of the variance [Magura & Moses 1986]. …