UCSF

Mounting evidence suggests that law enforcement organizational factors contribute to higher incidence and racial disparities in police killings. To determine whether agency policies contribute to race-specific civilian fatalities, this exploratory study compared fatality rates among agencies with and without selected policies expected to reduce killings. A cross-section of 1085 fatalities in the 2015-2016 The Counted public-use database were matched to 481 agencies in the 2013 Law Enforcement Management and Administrative Statistics (LEMAS) database. Negative binomial regression estimated incidence rate ratios (IRR) adjusted for agency type, number of officers, percent female personnel, median income, percent with a bachelor's degree, violent crime rate, and population size, with inference using robust standard errors. Agencies with greater proportions of full-time personnel (range 43-100%) had lower rates of all (IRR = 0.85; 95% confidence interval [CI] = 0.77-0.93) and non-White civilian killings (IRR = 0.85; CI = 0.73-0.99). Mission statements predicted lower rates of all (IRR = 0.70; CI = 0.58-0.84) and White killings (IRR = 0.60; CI = 0.40-0.90). Community evaluation and more types of personnel incentives predicted lower rates of White (IRR = 0.82; CI = 0.68-0.99) and non-White killings (IRR = 0.94; CI = 0.89-1.00), respectively. Increasing video use predicted higher rates of White killings (IRR = 1.13; CI = 1.01-1.28). No policies were significantly associated with Black civilian killings. Law enforcement policies that help reduce police killings may vary across racial groups with the least benefit for Black civilians. Impact evaluations and meta-analyses of initiatives aimed to mitigate fatalities should be explored, particularly policies to address anti-Black bias. A national registry tracking all police killings and agency policies is urgently needed to inform law enforcement policies aimed to mitigate civilian fatalities.