Date of Award

Spring 5-15-2019

Author's School

Graduate School of Arts and Sciences

Author's Department

Social Work

Degree Name

Doctor of Philosophy (PhD)

Degree Type



Objectives: The overall objective of this dissertation is to build knowledge about the phenomena of neglect subtypes to better guide prevention and child welfare intervention efforts in the future. The first aim focuses on how we come to officially identify cases of child neglect and its relationship to policies that set definitions for what is reportable in a given state. The second aim highlights whether or not there appear to be differences in safety and permanency outcomes for children known to CPS for different forms of neglect. The third aim focuses on why families may develop specific neglecting behaviors that may require differing approaches to intervention. This will be a three-paper dissertation and thus the significance section for this proposal is divided by research aim, followed by an overall methods section that describes the data sources and planned analyses.

Methods: Data for the present study will be drawn from five sources. Because of the differences in coverage and ability to track subtypes of neglect, three child level data sources are used in varying combinations to attempt to answer the research questions. This provides a means of triangulating results to help overcome some of the weaknesses in the individual datasets to be used. The fourth source will be a combination of state statute information readily available from the Child Welfare Information Gateway supplemented by a Lexus/Nexus search. The final source will be Social Explorer to get state-level child poverty measures.

Results: In paper one, I found that there is great variability across state statute regarding the definitions of child neglect. For example, emotional neglect was identified specifically in 11 states, and approximately 26 states specified that failure to educate the child is an element of neglect in the law. Moreover, the analyses showed that educational neglect was significantly associated with the percentage of reported neglect. States that include “educational neglect” in their state statutes are more likely to have a significantly higher percentage (%) of child neglect reports. For example, the average neglect reported rates are 69.46% (n=21) for the states include educational neglect in their statue and 60.32% (n=21) for states that did not in 2014. In paper 2, I didn’t find that cases reported for different types of neglect show significant differences in predicting the recurrence outcome. In NSCAW, families reported for domestic violence-related neglect were 4.89 time (OR=4.89, p<.05), 3.89 times (OR=3.89, p<.05), and 5.69 times (OR=5.69, p<.05) less likely to enter the foster care than families reported for physical neglect, supervisory neglect, and prenatal substance abuse. In addition, families reported for substance exposure were 2.14 times (OR=2.14, p<.05) more likely to enter the foster care than families reported for physical abuse. In regional data, families reported for physical neglect were 1.40 times (HR=1.40, p<.05), 1.31 times (HR=1.31, p<.05), and 1.20 times (HR=1.20, p<.05) times more likely to enter foster care than families reported for medical neglect, educational neglect, and supervisory neglect. In paper three, I found that family characteristics differed for physical neglect compared to lack of supervision neglect across a number of dimensions in both datasets in bivariate analyses though this was greatly attenuated in multinomial models for NSCAW data. Also, both bivariate and multivariate models using both data sets indicated a number of practically important (effect size) differences between cases reported for multiple types of neglect and supervisory neglect. On the other hand, the results the LCA showed that a 5-class and 6-class were the best models for NSCAW-II and the regional data. With classes contained families with different subtypes of neglect in the NSCAW data, most of the risk factors didn’t show much variation across the 5 classes. For the regional data, while there were variations between risk factors, most of all subtypes of neglect hung together across the 6 classes.

Conclusions: While studies argued that state-level administrative data are often the most accessible data source for child maltreatment research, it is important to better understand how cases that come to the attention of child protection may vary according to the policy gatekeeping mechanisms. In addition, we found that states identified “educational neglect” in their state statute had a higher percentage (%) of child neglect reports. Not only should we need to examine the effectiveness of the intervention programs in child protective service for this population for these states, but we need to examine whether children with unmet education needs to be ignored in the states that did not identify education neglect in their state statute. Also, in paper two, the results showed that significant variation between types of neglect and foster care entry. While the greater risk of entry associated with physical neglect, it was surprising that the risk was greater than that for supervisory or medical neglect cases. On the other hand, while there is debate whether exposure to domestic violence as a reportable form of maltreatment, we found children who were reported for domestic violence were less likely to enter foster care than other subtypes of neglect. This study highlights the need to examine the trajectory of children as a function of different forms of neglect to child welfare outcomes, suggesting the necessity of addressing the high-risk population in child welfare policy and practice. In paper three, the study did find variation in risk and demographic factors using two different datasets with different forms of data. This was only true, however, for the variable based approaches. The person-oriented analytic models were less informative in regard to subtypes but were consistent with the idea of CPS families facing multiple risk factors- most classes had high probabilities for multiple risk factors in both datasets. It is possible that the “iceberg theory” best captures the dynamics between the risk factors and children reported for different subtypes of neglect. If this is true, then the intervention programs for child neglect may need to focus on the cumulative risk of the family.


English (en)

Chair and Committee

Melissa Jonson-Reid

Committee Members

Adrienne Atzemis, Brett Drake, Derek Brown, Patrick Fowler,


Permanent URL: https://doi.org/10.7936/s80y-s604

Included in

Social Work Commons