Shocking numbers, but the study you linked did not find that 1 in 1000 black men are killed in the US by police.
What they found was
“Our models predict that about 1 in 1,000 black men and boys will be killed by police over the life course....”
In their methods, they state that 9% of fatal encounters had no race data, so they guessed based on the surname of the victim:
“Between 2013 and 2018, about 9% of FE cases are missing data on race–ethnicity (SI Appendix, Table S1). We use multiple imputation by chained equations (48) to address missing data for observations between 2013 and 2018. Imputation models include victim age, sex, race, cause of death, and the racial/ethnic composition of the county in which a death occurred. We also include surname-specific estimates of the probability of racial/ethnic group identification on the US Census compiled by Imai and Khanna”
They then took their “guesses” aka “imputed data” and assumed that the 6 years of data of fatal encounters would apply for an entire lifetime (ignoring long term trends of declining police force deaths and decreasing inequality in deaths between races see below):
“We use these imputed data to construct multilevel Bayesian count models of mortality risk that allow us to directly estimate uncertainty driven by small annual age–race–sex-specific death counts for some groups, by variation in underlying risk over the 6 y of FE data, and by missing data...Because we lack sufficient data to track a birth cohort over the life course, we rely on synthetic cohorts to estimate lifetime risk (31). Period life tables allow us to estimate deaths over the life course within a compressed period by tracking age-specific mortality risk over hypothetical cohorts”
Also
“Also note that while black people remain disproportionately more likely than white people to be killed by police, the share of white deaths has been increasing in recent years...Prior research suggests that despite high contemporary rates, the risk of being killed by police was higher in decades past.”
So, interesting study. However, I think the journalist who reported it didn’t really understand what it meant...which is:
If US police violence that has happened over the past 6 years remains the unchanged, a black man born today will have a 1 in 1,000 chance of being killed by the police; give or take over the course of his life.