We continue the earlier post on interpreting data from an asymmetric sample space.
An estimated 9540 non-Hispanic black people died from police violence during 1980-2018, says a study published in The Lancet last year. In the same period, the number of non-Hispanic White people who met the same fate was 15,200. So, whites are more likely to die from police violence in the US. Right?
Yes, if the population of non-Hispanic blacks and non-Hispanic Whites in the US are equal. But that is not true. As per Wikipedia, the former accounts for 12.3% of the US population and the latter 61.5%. If there is no correlation between death and race, you would expect around 12.3% of deaths for blacks and 61.5% for whites. As that is not apparent from the numbers, we will calculate the odds.
The easier way to do this is to divide 9540 with 12.3 and 15,200 with 61.5 and take ratios. The numbers are 775.6 : 247.2 = 3.1 : 1. In plain English, a non-Hispanic black has a 3.1 times more chance to die from police violence than a non-Hispanic White.
Here we only considered how the mind works while interpreting data when the representation of groups is not symmetric. Studying the reason behind the disparity of either behaviour of people of certain races or the reaction of police in response was not the topic. Statistics rarely tell the cause, but it may suggest a problem that requires a solution.
Fatal Police Violence: The Lancet
Demography in the US: Wiki
