I'm trying to tease out the different ways to understand racism through a series of thought experiments. 

Here's the first one. 

# The Experiment

A shopkeeper knows that a certain percentage of all his customers are thieves. 

Could be 5%, could be 10%. Let's just call it _percentTotal_, or __pT__.

He keeps record of all the thefts. Looking over, he notices the following.

For a certain group of customers, who all share the same race X, the percentage of thefts differs from entire customer base as a whole. 

It might be higher than __pT__, or it might be lower. We're not gonna say. 
We're just gonna call it _percentageRace_, or __pR__.

__The question__ is this: Could __pR__ _possibly_ be so much bigger than __pT__ that it would allow the shopkeeper to ban everyone from race X from his shop without being __racist__?

__If so,__ _how much_ bigger does _pR_ have to be than _pT_? 
If 50% of customers of race X where thieves compared to only 5% of the total, would it be racist to ban them from the shop? 

__And,__ are there _other factors_ that would change the conclusions? 

What do you think? 
All considerations and reflections appreciated :)