Methods: We used publically available City of Chicago Police Department data locating police responses to 11,177 child abuse incidents that occurred between 2001 and 2015. Geographic “shapefiles” of Chicago Census tracts were used to aggregate incidents at the tract level. Demographic and economic characteristics of the Census tracts were obtained from the 2000 and 2010 dicenniel U.S. Censuses. We employed a longitudinal multi-level model to estimate the growth trajectory of the rate of child abuse per 1,000 children across Census tracts from 2001 to 2015. Our main predictor was the average rate of child abuse during the prior year for all Census Tracts sharing a boundary with the focal Census Tract. Neighborhood income, residential instability, concentration of African American residents, population density, and child population density were included as covariates in the analysis.
Results: The results indicate a contagion effect such that the rate of child abuse in a neighborhood (Census tract) had a statistically significant, positive association with the rate of child abuse in the surrounding neighborhoods during the prior year (𝛽 = 0.125, p < .001), even after controlling for the effect of time and the covariates. Consistent with the tenets of social disorganization theory, household median income (𝛽 = -0.063, p < .001), percent of Black population (𝛽 = 0.012, p < .001), and residential instability (𝛽 = 0.019, p < .001) were also significant predictors of higher rates of child abuse.
Conclusions and Implications: These findings indicate that police-investigated child abuse is more common in neighborhoods that are in proximal spatial contexts with other neighborhoods that had high rates of child abuse during the prior year. This provides preliminary evidence that child abuse, like some other problematic behaviors, may ‘spread’ across space and time. Future research should use social network analysis to confirm whether this spread is a function of the social transmission of parenting norms among neighbors. Additionally, our findings support the utility of mapping past child abuse incidents to help social service and law enforcement agencies predict the locations of future incidents, i.e. ‘smart policing’ or social service delivery.