Comparison of Emotional Maltreatment Rates across U.S. Jurisdictions: State Policy Cluster Analysis

Background/Purpose: Lack of clarity in how Emotional Maltreatment (EM) is defined across jurisdictions may contribute to differences in EM reporting rates. EM may contribute to adverse child well-being in the short- and long-term. This study examines how EM is classified and whether policy variation corresponds to differences in EM rates across evidentiary standards, required diagnoses, and parental intent.

Methods:

• Child-Level Analysis: We used the Report and Placement Integrated Data Set (RAPIDS), which combines NCANDS and AFCARS data. We re-coded the maltreatment variable to create three types of EM reports: 1) EM only, 2) non-exclusive EM, and 3) any other maltreatment types without EM. Bivariate analyses were conducted to examine differences in demographic characteristics across three EM report category counts.
• State-Level Analysis: First, we reviewed the EM policies for 51 U.S. jurisdictions (49 states, Washington, D.C., and Puerto Rico) using the State Child Abuse & Neglect Policies Database and the Child Welfare Information Gateway. States were classified along three policy dimensions: evidentiary standard (1 = probable or reasonable cause, 2 = credible or substantial evidence, 3 = preponderance of evidence); diagnosis requirements (1 = not required, 2 = mental health evaluation, 3 = diagnosis required); and parental intent (1 = not required, 2 = intent considered, 3 = intent required). Next, we merged the total child population data from the Annie E. Casey Foundation with the RAPIDS dataset to create annual state-level EM rates per 1,000 children for three EM categories from 2013 to 2021. With the state-year-level collapsed data, Ward’s hierarchical clustering was used to identify clusters of states with similar policy characteristics.

Results:

• Child-Level: Of the exclusive EM-only report total counts (n = 27,538), the largest group was Non-Hispanic White children (37.5%), followed by Hispanic/Latino children (30.13%). Sex was distributed similarly between males (49.19%) and females (50.81%). The mean age of children with EM-only reports was 8.33 (SD = 5.22) years old. Chi-square tests examining associations between the three demographic variables and three EM categories were statistically significant; however, the effect sizes were very small: age (Cramer’s V = .049), sex (Cramer’s V = .007), and age (ε² = .005).
• State-Level: A three-cluster solution showed the best fit. Cluster 1 included states with the most stringent EM policies across the policy indicators whereas Clusters 2 and 3 represent less stringent aspects: Cluster 1 intent required (10 states; 50.69% of EM only reports); Cluster 2 intent not required but preponderance level of evidence required (27 states; 33% of EM only reports); Cluster 3 generally low requirements (14 states; 16.31% of EM only reports). Planned contrast tests comparing Cluster 1 with the average of Clusters 2 and 3 for each of the three models showed that Cluster 1 had significantly lower non-exclusive EM rates than the combined average of Clusters 2 and 3 (estimate = –1.01, p = .021).

Conclusions/Implications: Social work practitioners and policymakers may use these findings to advocate for clearer, more consistent definitions of EM across jurisdictions to facilitate more accurate identification and effective prevention efforts.