Discrepancies in longitudinal self-reported racial categorization: Implications for mental health services research

Background: Racial categories are typically treated as mutually exclusive in mental health research. Contrary to this standard, racial categorization is coupled with dynamic socio, economic, and political factors resulting in unstable self-reported race data that can potentially misrepresent the meaning of race and how it relates to mental health services and treatment outcomes.

Methods: Longitudinal survey data were collected from 661 adults at 13 rural community mental health agencies in the Midwest. Race was self-reported on the Behavior and Symptom Identification Scale-24 (BASIS-24), which participants completed three times over six months. Crosstabs, Lambda, and Chi-square tests were used to analyze the data.

Results: Crosstabs established that 77 (11.6%) of the 661 participants changed their self-reported race. Lambda confirmed that race at the first measure was a strong, though not very strong or perfect, predictor of race at the second and third measures (λ=.76; .75). Chi-square tests established that there were statistically significant variations (p<.001) when comparing participants' first to second and first to third self-reported race. Individuals identifying as American Indian (n=10), Multiracial (n=21), Native Hawaiian (n=3), White (n=29), and Black (n=14) changed their identity in a wide variety of patterns.

Conclusions: Mental health researchers should not assume that race data are stable across multiple measures. Nearly 12% of participants changed their self-reported race in an unpredictable array of variations in our sample. Possible solutions include resolving participants' race in follow-up phone or face-to-face interviews, or treating these participants as a unique race group in statistical analyzes.