An Examination of Class-Based Visibility Bias in Child Maltreatment Reporting

Background/Purposes:

There is a widespread, long-held belief among researchers and policy makers that poverty increases children’s exposure to professional reporters (e.g. social services and mental health personnel) causing more reports to be made. This is sometimes called “Visibility Bias” (VB) and the literature is consistent in placing this effect entirely among professional reporters. Surprisingly, there is virtually no empirical evidence bearing on the issue of VB. We seek to fill this gap. As VB is operative, there will necessarily be a much higher proportion of reports from professional reporters as poverty increases. We examined this relationship using national data.

Methods:

We linked National Child Abuse and Neglect Data System data (2009-2013) to county-level Census data to determine (1) whether counties with higher child poverty rates had higher maltreatment report rates and (2) whether larger proportions of reports were from professional reporters in counties with higher child poverty rates. The linked data covered 59,105,280 children (80.0% of US children) in 752 counties (23.9% of US counties). We further constructed race/ethnicity-specific data sets through this linkage: 27,519,794 White children (70.2% of US White children) in 657 counties, 7,871,410 Black children (74.2% of US Black children) in 209 counties, and 14,196,730 Hispanic children (81.9% of US Hispanic children) in 227 counties. We used additive mixed modeling to handle both the nested (state/county) data structure and curvilinear relationships.

Results:

As expected, county maltreatment report rates increased as county child poverty rates increased. We did not find, however, that poorer counties had higher proportions of reports from professionals. There was a tendency towards a slightly lower proportion of reports from professionals as poverty increased. While county child poverty rates increased from 3.6% to 54.2%, report rates increased from 1.8% to 5.9%, but proportions of reports from professionals decreased from 68.4% to 52.9%. We found this trend consistently across all racial/ethnic groups. With an increase of race/ethnicity-specific child poverty rates (1.5%-33.7% for Whites, 11.1%-61.7% for Blacks, and 10.1%-56.7% for Hispanics), race/ethnicity-specific report rates increased (0.4%-6.8% for Whites, 3.8%-8.9% for Blacks, and 1.9%-4.5% for Hispanics), but race/ethnicity-specific proportions of reports from professionals always decreased (69.2%-48.6% for Whites, 67.6%-50.0% for Blacks, and 68.1%-54.7% for Hispanics). This trend was also consistent across maltreatment subtypes (i.e., neglect, physical abuse, and sexual abuse).

Conclusions/Implications:

Contrary to expectations, proportions of reports from professional sources decreased with increasing poverty, holding true across all groups, showing that this unexpected relationship is not due to a racial/ethnic confound. We see two plausible explanations for these effects. First, VB may be present both among professionals and nonprofessionals being slightly *stronger* among nonprofessionals. This invites a reworking of existing VB theory, as there is no currently postulated rationale for why VB would apply to nonprofessionals. The second possible explanation is that VB is not a primary driver of official maltreatment reports, and that concerns about VB effects have been overblown. Given the recently emerging scientific consensus that the poverty/maltreatment relationship is largely real, and not a function of reporting bias, the second explanation may be more plausible.