Prevention Resources, Community Capacity and Child Maltreatment: A Hierarchical Linear Modeling Analysis

Background: Prevention of child maltreatment (CM) remains a challenge for communities faced with increasingly limited funding and support for local child welfare programs. Identifying how communities and states can build capacity for proactive engagement with families and children who are at risk for child maltreatment may help to meet the challenge of increasing resources for prevention work while not reducing funding for vital protective services. In our previous longitudinal work, we identified a number of clusters where maltreatment rates were significantly high throughout a 13-year period. Building on this work, our current study seeks to examine how county level prevention resources, such as staffing for prevention activities, may have an impact on the variability in CM rates.

Method: We derived our dataset from a number of publically available data sources, which included state prevention survey data and CM rate data from the Virginia Department of Social Services (VDSS), crime data from the Virginia Police Department and poverty data from the Census Bureau. CM rate data included the calculated rate of founded maltreatment for each county. To assess prevention activities throughout the state, we used the state prevention survey data collected by VDSS from local county agencies (N=111) in 2011. Three prevention-related variables were coded in MaxQDA 12, including types of prevention services provided, staffing available for prevention activities, and types of prevention funding sources. Hierarchical linear modeling (HLM), repeated measures nested within counties design, was conducted in SAS 9.0 (MIXED procedure). Independent variables at level 1 included time, poverty rate and crime rate. Level 2 independent variables included three prevention variables such as types of prevention services, staffing available for prevention activities, and prevention funding sources. A spatial power covariance structure was used, as it provided the best model fit.

Results: The present study found that 54% of the variance in CM rates was found between counties, while 46% of the variance in CM rates was found within counties. Our analyses also found that that CM rates have decreased over time (β = -0.47, SE =0.09, t(551) = -5.13, p < .001). Poverty was positively, significantly related to CM rates (β = 0.04, SE = 0.02, t(551) = 2.22, p < .05]. Additionally, the interactive effects of poverty and staffing available for prevention activities were found in that the absence of dedicated prevention staff was positively related to CM rates when poverty rates were high (β = 0.02, SE = 0.01, t(551) = 1.95, p<0.05]. Finally, we also found that interaction between universal prevention services with poverty rate (β = -0.04, SE = 0.02, t(551) = -1.96, p < 0.05) were significantly, negatively related to CM rates.

Conclusion: Our study indicates that availability of universal prevention services and the availability of dedicated prevention staffing may buffer the negative impacts of poverty on child maltreatment. Enhancing access to community resources for prevention activities, such as universal prevention programs and providing workforces dedicated to local prevention activities may be an important step in proactively reducing the incidence of CM.