Examining California Child Welfare Appellate Case Characteristics and Outcomes

Background and Purpose: In California, parents may appeal a juvenile court ruling on their child protective services (CPS) case to correct certain agency or court errors. The literature is sparse on using court opinion documents to measure the frequency and magnitude of parental appeal success. This study aims to take the first step by identifying which types of recent California child welfare appellate cases are most likely to succeed. Because this project examines novel data, its primary goal is to obtain descriptive detail of case characteristics using the computational grounded theory framework outlined by Nelson (2020) and provide a foundation for further study.

Methods: Data from Nexis Uni consisted of 546 child welfare appeals court opinion documents from 2022 which cite California’s authorizing statute for CPS intervention. Approximately 58% of these appeals were denied, 14% were mostly denied, 22% mostly approved, and 6% were fully approved. A Latent Dirichlet Allocation (LDA) topic model was used to create a distribution across documents to be used as features in an ordinary least squares regression. A seven-topic model was chosen reflecting the most distinction between each topic. All topic labels were synthesized by the author using the first ten words of each grouping. Using the topic weights for each document, an ordinary least squares regression was used to test whether these seven topics were significantly associated with the disposition label. Subsequently, the author conducted deep reading on fully approved appeals by topic.

Results: Several topics were negatively associated with the outcome of the appeal: cases involving domestic violence/drug use by parents (β = -0.92, SE=0.15, p<0.001), custody disputes over minors (β = -0.73, SE=0.30, p=0.015), infants with physical injuries (β = -0.37, SE=0.24, p=0.125), and where parental rights were already terminated (β = -0.05, SE=0.28, p=0.871). The only positively associated topic was whether a case involved provisions of the Indian Child Welfare Act (ICWA) (β = 0.04, SE=0.24, p=0.125). However, most estimated coefficients were not statistically significant at the 5% level except domestic violence/drug use. Subsequent deep reading revealed accurate topic modeling and evidence that the appellate courts corrected consequential errors in cases of full approval. For instance, one judge reversed a lower court’s decision approving CPS’ jurisdiction of an infant who was found to have solely medical reasons for a subdural hematoma and no physical abuse by his mother who filed the appeal. Another appeal decision reversed the juvenile court’s findings that a father was a perpetrator of maltreatment when there was no evidence to support the claim. Yet another judge reversed approval by the juvenile court for CPS to remove children against their wishes from prospective adoptive parents to live with unfamiliar kin in another state.

Implications: Although successful appeals constitute a minority of cases, preliminary evidence of appellate courts correcting harmful systemic errors highlight a promising avenue to bolster support for vulnerable families and children involved in CPS.