Discretionary bias is still possible if you attempt to correlate the race stastics with the racial group as a whole while ignoring other factors such as social and economic status. For example the majoriry of murders that occur during a robbery are commited by poor people. A group made up of mostly middle/upper class citizens would mostly exclude those type of murder cases. Gang violence is another big driver of murder and also closely correlated to socioeconomic status. The middle class and up don't join gangs and therefore aren't part of organizations where murder is part of operations.

The discretion is on if you choose to rely on a correlation without looking at the actual causation to make conclusions.