How Racial/Ethnic Composition Obscures the Relationship Between Suicide and Homicide in the United States

Keywords: Suicide, homicide, health, confounding, disease rates, race/ethnicity, geostatistics
Session Type: Virtual Paper Abstract
Day: Tuesday
Session Start / End Time: 3/1/2022 03:40 PM (Eastern Time (US & Canada)) - 3/1/2022 05:00 PM (Eastern Time (US & Canada))
Room: Virtual 56

Abstract

Race/ethnicity plays a critical role in both suicide and homicide mortality. The black homicide mortality rate is nine times higher than the white rate. Conversely, the white suicide mortality rate is nearly three times higher than the black rate. Historically, a clear spatial relationship between these outcomes has been elusive, despite theory suggesting that similar conditions should contribute to both. This paper argues that race/ethnicity confounding obscures the positive relationship between homicide and suicide.

Using county-level mortality data from the Centers for Disease Control, 1999-2019, I calculate crude rates and race/ethnicity-adjusted rates based on widely-used age-adjustment methods. Expected suicide and homicide rates are constructed using indirect adjustment based on racial/ethnic composition. Then, standardized mortality ratios are calculated using the crude rates and expected rates. The standardized mortality ratios are interpreted as the variance in mortality once the effect of race/ethnicity composition has been removed. Relationships among crude, expected, and adjusted mortality rates are explored using correlations and linear regression.

There is no statistically significant correlation between crude suicide and homicide mortality. However, adjusted homicide mortality can explain 20.8% of the variance in adjusted suicide mortality. Once the compositional effect was removed, the contextual effect of each racial/ethnic group was measured. The final regression model uses adjusted homicide and percent American Indian to predict 40.5% of the variance in adjusted suicide rates. This project demonstrates the powerful confounding effect of race/ethnicity, as well as the importance of separating the compositional and contextual dynamics of explanatory variables.