- Overlook important variation among states in the demographics of the children and families served.
- Fail to account for systemic state differences in caseload inclusion criteria, and the inherent practice and policy conflicts between measures.
- Draw on data derived from a database (AFCARS) that was not designed to measure longitudinal performance, and is still not of the quality to justify imposing fiscal penalties.
- Count/weight states equally despite enormous differences in the size of the child population.
- Employ a complicated statistical method, principal components analysis (PCA), in the absence of any evidence that such a method is in any way required or superior to simpler and more-transparent approaches to measurement.
- Make many arbitrary and statistically inappropriate decisions in the use of the PCA procedure, thereby undermining the ranking of states that the method produced.
- Arbitrarily set the national standard at the 75th percentile, and then rely on ill-conceived rules that adjust the standard to a different level.
State policymakers can use their own policies for holding their child welfare systems accountable.
No comments:
Post a Comment