Lee, S. J. (2013). Paternal and household characteristics associated with child neglect and child protective services involvement. Journal of Social Science Research, 39, 171-187.
There exists a demonstrated gap in the literature investigating paternal characteristics associated with child neglect or Child Protective Services (CPS) involvement. This article attempts to address this dearth in research by performing secondary analyses of a subsample of the community-based Fragile Families and Child Well-Being Study (FFCWS), which follows a cohort of nearly 5,000 children born between 1998 and 2000 from 20 large U.S. cities. Through a developmental-ecological lens, Lee examined individual-level (paternal) and family-level (household) characteristics and their association with dependent variables of child neglect and household CPS involvement. These measures were obtained when the index child was 5 years of age or younger. A subsample of 1,089 biological and residential (i.e. living with index child) fathers was used for the current analysis.
This study showed that variables of paternal psychosocial functioning were found to be significantly associated with neglect and CPS involvement, even after accounting for family and household factors. Paternal depression was identified as a strong risk factor for both variables, with doubled odds of child neglect observed in this sample. Other paternal psychosocial characteristics, like parenting stress and alcohol use, were linked to elevated risk for child neglect but not CPS involvement. One compensatory factor, father involvement in common child-care activities, was found to be protective against child neglect. Data on demographic and household characteristics were also collected and analyzed, resulting in the detection of three factors related to CPS involvement: more children under 18 living in the home, financial assistance, and low parental education.
Several of the methodological concerns stem from the use of data obtained from a study with research goals that differ from the author’s. Lee identified her use of the developmental-ecological framework, a model that examines child maltreatment etiology as influenced by individual-, family-, and macro-level factors. However, this study was restricted to data containing only two of the factors applicable to the guiding framework, and therefore failed to include potential macro community-level effects. Furthermore, the FFCWS sought to collect data specifically on unmarried parents, resulting in a clear sampling bias. As a result, within the current study’s subsample, almost two-thirds of the fathers were not eligible for inclusion since they did not continue to live with the family when the index child was 5 years old. In addition, the FFCSW sample from 20 large U.S. cities has claimed to be nationally representative. Upon further examination of the primary data’s study design, it is noted that a large city is classified as having a population of 200,000 or more. Generalizability of findings may therefore be limited to urban versus rural populations.
Another important sampling bias becomes apparent upon close consideration of the inclusion criteria. As previously mentioned, fathers had to be biological and either married or co-residential partners in order to participate in the study. This means that high-risk groups, such as single mother-headed households or mothers in relationships with men that are not the biological father, are excluded. The author acknowledges that this sample is more likely to be advantaged and predicted that this would generate lower-than-average levels of neglect and CPS involvement in the subsample. Indeed, comparisons of the subsample with the entire original sample revealed 4% less CPS involvement. This expectation also lead researchers to employ dichotomous variables to the dependent measures of child neglect and CPS involvement in order to capture any relationship, since these conditions were considered so infrequent within this type of sample. In this study, dichotomous variables include options of 0 or 1, with 1 indicating involvement in even one instance of child neglect in the past year. Similarly, a unit of 0 or 1 was assigned to indicate whether CPS has contacted the family since the birth of the index child. This simple measurement system was deemed necessary in order to detect a relationship in the sample; however, it also disregards possible complexity in the variables. For example, for household CPS involvement, the measure does not specify which child the CPS contact was regarding, who the perpetrator of suspected maltreatment was, the timing of maltreatment, or type of maltreatment. Rather, it only provides a general and ambiguous indicator of household risk.