An Intersectional Analysis of American Indian/Alaska Native Youth Aging out of Care

American Indian and Alaskan Native (AI/AN) children continue to be overrepresented in public child welfare systems. However, depending on data protocols, AI/AN children may be under-counted in child welfare data, thus perpetuating institutional bias. AI/AN children may be more likely than children from other underrepresented groups to be combined into “two or more races” categories. If an AI/AN child has an additional racial or ethnic heritage that is also disproportionate (such as also being Black), the possibility of negative outcomes could compound. Drawing from a theory of indigenous erasure, this study tests whether deconstructing a governmental “two or more races” category improves understanding of outcomes for AI/AN youth in national child welfare datasets and highlights compounding risks for youth with complex intersectional identities.

Method

This secondary analysis utilized data merged from the Adoption and Foster Care Analysis Reporting System (AFCARS) and the first two waves (age 17 and 19) of the National Youth in Transition Database (NYTD), 2011 cohort. 542 youths are identified as having AI/AN heritage, of whom 210 are considered solely AI/AN and 332 are “two or more races” according to the government algorithm that forces every subject into one race option. Participation rates at age 19 reduced this sample of 542 to 292 youths.

McNemar tests were used to determine the statistical significance of the loss of AI/AN-identified youths through the use of the “two or more variables” category

Dependent variables studied at age 19 using the expanded race categories included incarceration, having children, homelessness, current educational enrollment, having a diploma or GED, and attainment of part-time employment. Logistic regression taking risk and protective factors into account from age 17 was completed using STATA 15SE.

Results

The decrease in number of youths from 542 to 210 is a loss of 61.25% by using the forced-choice option instead of the multiple identity option. By contrast, the loss of Black youths into the “two or more” category is 15.67%. Both McNemar tests were significant at the .001 level.

Across all six dependent variables, only homelessness was further informed by adding additional race categories into the analysis. Black AI/AN youth (OR: 3.402) were more likely to be homeless while Latin youth (OR: .442, were less likely to be homeless, though this was only marginally significant (p<.10). The protective factor of being Latin for homelessness is attenuated by remaining in foster care at age 19.

Implications

Given the lack of difference for multiracial AI/AN youths, it may be that AI/AN identity profoundly shapes youths’ experiences in a way that subsuming youth into a “two or more races” category may hide in statistical analyses.  The need for state and federal agencies to change their policies and procedures for data collection to ensure full inclusion of American Indian and Alaska Native identifies are highlighted in this data.