Methods: The sample included children under 6 years old (n=1,628) from the Child Development Supplemental study (CDS I-III) originated in 1997 and their mothers who were subjects of the Panel Study of Income Dynamics (PSID) household survey. For the analysis, family-level data from the PSID household main survey were linked to child-level data from CDS (I-III). Also, census information was linked according to household track codes. To measure CM risk, we used the caregiver-reported proxies for harsh and neglectful parenting prospectively. A CM Risk dummy variable was created using a) the onset of spanking at age one has been found to predict later CPS involvement, b) no medical check-up in the previous year, and c) the lowest 10 percentile HOME score. Multilevel Logistic regression with mixed effects was employed to model children and caregivers clustered in the families. This approach had the advantage of investigating the effects of covariates both on the overall level of the responses and on changes in the responses over time.
Results: The log of Family income was significant at the family level (OR=.65, p=0.001) and at the census-track level log of family median income (OR=.095, p=0.008) - both indicating that higher levels of income reduced the likelihood of any CM risk. In the context of the neighborhood, the model also showed the importance of social cohesion (OR=0.925, p=0.000). In addition, urbanicity measured in percentages at the track level was associated with reduced likelihood for CM risk (OR=0.982, p=0.002). Mothers of female children were less likely to report harsh or neglectful parenting (OR=.55). Children in homes with high levels of family conflict (yes/no) were more likely to be among those subject to harsh or neglectful parenting (OR=1.56, p=0.030).
Conclusions and Implications: The results showed the importance of both family-level and neighborhood socioeconomic resources in predicting a caregiver's self-reporting CM risk. The relationship between CM risk and reported family conflict is consistent with the literature on the overlap between IPV and CM lending some support to the relevance of the proxy measures to maltreating behavior. This adds to the growing literature indicating that poverty may play a causal role in CM risk and suggests that policies targeting child poverty like the EITC or the now-ended COVID may disrupt this path.