The problem is that you're assuming the system is fair and therefore the statistics that come from it are a reflection of reality. If the system is racist then the statistics will be a reflection of that racism. Until you can demonstrate that isn't the case then using crime statistics as an excuse for police actions will be questionable at best, and actively enabling racism at worst.
What else can be used other than statistics? Is there some other common frame of reference out there?
Ive seen a lot of evidence of black people not being able to fully prove their innocence and taking a plea deal because they would rather spend 6 months in jail over fighting, with the aid of an incompetent public defender, to avoid a 10 year sentence.
With all the corruption caught on officer body cameras in just the last 5 years (planting drugs, killing innocent people, unnecessary excessive force, death penalty/life sentence convictions overturned, prosecutors caught hiding evidence of innocence, etc)... It is safe to assume that this level of corruption (which is in many cases protected by both qualified immunity and prosecutorial immunity) has been going on for decades, heavily skewing black crime stats with false entries.
So there are plenty of ways to acquire statistics and data other than through the biased system you are trying to examine. Think of it like going after a computer system that you suspect has been rooted. Do you want to use the system tools from that system to examine itself? Probably not.
Qualitative research eg talking to people. Use empathy and compassion. Don't try to distill everything down to numbers.
-Mark Twain
https://www.bing.com/search?form=MOZLBR&pc=MOZI&q=blacks+lie...
The problem with statistics is that you need to take lots of time and care to have all sides fully explain their position and be able to rebut counterarguments with more data. If you do this, you will get to the truth, which is why people who are wrong tend to push conspiracy theories in order to dismiss the data instead of putting forward testable arguments. There is no better (or worse) non-argument than the one that you assert can not be falsified a priori.
If I recall correctly, there's a few confounding factors there - one is that survey results for "have you used drugs in the past week" show much higher relative use by African Americans than "have you used drugs in the past year". Another is that in other surveys African Americans were significantly more likely to answer "no" to "are you a convinced felon" given that they were actually convinced felons, so the survey results aren't necessarily accurate anyways.