Methods/Statistics
Constructing a multidimensional measure of structural racism: A balancing act between simplicity and complexity Lauren E. Barber* Lauren Barber Maret L. Maliniak Leah Moubadder Jasmine M. Miller-Kleinhenz Jeffrey M. Switchenko Michael R. Kramer Lauren E. McCullough
Structural racism is a complex multidimensional construct. However, prior studies have failed to incorporate this multidimensionality. In this study, we describe methods for operationalizing a multidimensional structural racism measure and consider its challenges.
County-level data on 24 indicators representing six domains (criminal justice, education, employment, healthcare, housing, and political participation) were obtained from sources including the American Community Survey. Indicators were measured separately for Black and White populations in 115 Georgia counties with available data. For each indicator, a racial inequity ratio was calculated such that higher values indicated White advantage. Domain-specific principal component analyses (PCA) were performed to identify indicators for inclusion in the multidimensional structural racism measure. Indicators with loading values >0.4 were selected from each domain and included in a cross-domain PCA. Loading values for the first principal component were assessed and used as weights to derive a multidimensional measure.
Domain-specific PCAs identified 12 indicators for inclusion in the multidimensional measure such as incarceration, household income, mortality, and voting ratios, and school dissimilarity and Thiel’s H indices. The first principal component from the cross-domain PCA, which summarized the structural racism construct, explained only 29% of the variation. Employment, healthcare, and housing indicators had the highest loading values (0.32-0.39).
Providing a transparent method for constructing a multidimensional structural racism measure can advance the study of structural racism as a determinant of health. However, researchers may be challenged with balancing potentially oversimplifying the structural racism construct and adequately capturing its underlying complexity. Varying the type and number of indicators may help strike a balance. Ongoing work to develop a valid structural racism measure is critical.