This item is under embargo and not available online per the author's request. For access information, please visit http://libanswers.wustl.edu/faq/5640.
Date of Award
Doctor of Philosophy (PhD)
Objectives: This study seeks to improve our understanding of risk and protective factors for child maltreatment both over time and within an ecological context. First, this study examines longitudinal patterns of child maltreatment reports (CMR) with child age from 1 to 17 years based on various risk and protective factors (Aim 1). This study also examines neighborhood contextual effects on CMR (Aim 2).
Methods: This study used secondary data from a larger longitudinal study which had followed up two samples from the 1991-1994 St. Louis birth cohorts. The CAN sample included all children aged 3 or under with a first-time CMR in 1993-1994 (n = 2,111). The AFDC sample included randomly selected children aged 3 or under receiving AFDC in 1993-1994 with no current or prior CMR (n = 1,923). For Aim 1, this study followed up children from 1995 through 2009 in the secondary data and estimated the CMR likelihood at each age from 1 to 17 years. For Aim 2, only age-year observations on welfare (AFDC/TANF) were selected to trace changes of residential neighborhoods through welfare records. This study does not specifically focus on either onset or first-time recurrence of CMR. Rather, this study estimates the likelihood of any CMR at a given age regardless onset, first-time recurring, or any subsequent recurring of CMR. This study used multilevel logistic growth curve models to estimate the CMR likelihood as a function of various risk and protective factors. Variables were measured by the secondary data which had traced children in various Missouri administrative datasets and Census data.
Results: This study found that 60% to 67% of the variance of the CMR likelihood was between age-year observations and 33% to 40% was between children. Less than 1% of the variance was found between neighborhoods. Analyses for Aim 1 found important observation-level (i.e., time-varying) and child-level predictors. Every one-year increase in child age decreased the CMR likelihood by 13% in the CAN sample (OR = 0.87, 95% CI = 0.86-0.88). While the main term of child age was not significant in the AFDC sample (0.99, 0.96-1.02), child age was associated with CMR through interacting with current welfare (AFDC/TANF) receipt. Current welfare receipt increased the CMR likelihood by 2.32 times in the CAN sample (2.32, 1.98-2.71). This relationship varied by child age in the AFDC sample: current welfare receipt increased the CMR likelihood by 3.62 times at age 1 and by 1.18 times (18%) at age 17. Prior welfare receipt (% of months on AFDC/TANF; 1 unit = 10-percentage point) increased the CMR likelihood by 8% for the CAN sample (1.08, 1.05-1.11) and by 12% for the AFDC sample (1.12, 1.08-1.17) only while not receiving welfare currently. When receiving welfare currently, prior welfare receipt was not significant for both CAN sample (1.00, 0.97-1.03) and AFDC sample (0.97, 0.93-1.02). Compared to Whites, the CMR likelihood for Blacks was 16% lower in the CAN sample (0.84, 0.74-0.95) and 35% lower in the AFDC sample (0.65, 0.53-0.80). Many other predictors including prior CMR, CPS in-home services, child mental health, child injury, child special education, parent criminal issue, parent low education, and maternal foster care placement were associated with CMR in both samples. Child behavioral and health problems were significant only for the CAN sample. Multivariate analyses for Aim 2 revealed that no neighborhood characteristics were significant in the CAN sample, while some were significant in the AFDC sample. Each 10-percentage-point increase in neighborhood poverty rate increased the CMR likelihood by 31% (1.31, 1.05-1.64) for Whites. This relationship was not significant for Blacks (1.01, 0.92-1.10). Neighborhood child/adult ratio (1 unit = 0.1) decreased the CMR likelihood by 10% (0.90, 0.82-0.99). The CMR likelihood for children moving out of St. Louis (i.e., making a long-distance move) was 63% higher than for those staying in St. Louis (1.63, 1.07-2.48).
Conclusions: Results suggest that CMR risks largely varied by time. Current welfare (AFDC/TANF) receipt remained a strong predictor of CMR risks. The strong observed interactions of current welfare receipt with child age and prior welfare receipt suggest the importance of longitudinal approaches in understanding their relationships to CMR. CMR risks were much higher at younger ages. Once risk factors were controlled for, Blacks showed no higher CMR risk than Whites. In fact, Blacks showed a lower risk. Although some neighborhood characteristics were significant, their effect sizes were mostly small in contribution to the overall risk and were less observable among families at a higher risk of future CMR. Implications include the importance of considering longitudinal changes among risk and protective factors over time, the centrality of current family economic conditions (if current AFDC/TANF receipt proxies this) in CMR, the importance of early intervention, and necessity of addressing these critical issues in policy and practice. To lower racial disparity in CMR, addressing differential exposure to risk factors, especially low SES, may be more promising than racial bias interventions. Additionally, this study highlights the utility of cross-sector data in improving our ability to better understand and predict child maltreatment.
Chair and Committee
Melissa Jonson-Reid, Patricia L. Kohl, Douglas A. Luke, Mark R. Rank,
Kim, Hyunil, "Understanding Child Maltreatment Report Risks as a Function of Age, Socioeconomic Status, Race, and Neighborhood" (2018). Arts & Sciences Electronic Theses and Dissertations. 1549.
Available for download on Sunday, April 19, 2020