I'm new here and I don't know how much we are expected to muzzle our cyber mouths in furtherance of the pretensive doctrine of political correctness.
So I'm just gonna say this: I've cross referenced crime stats with demographics more times than I can count. Political correctness may lie but the crime stats don't. Instead they've always led me to the same conclusion. Members of some demographic groups (one in particular) commit a heck of a lot more crime than members of other demographic groups. That's reality!
Knowing that, I think it would be suicidally foolish to not take that into account. Especially if the subject is dressed in what members of his own group call thug or gangsta attire.
There are plenty of other (and more PC) factors I would also take into account:
How nervous is he/she acting?
Are they waiting for the customers to leave?
Do they have a friend waiting outside in a car with the motor running when the place has a drive thru?
The list goes on and on. But I'm not gonna discount anything because it will get me points for being PC. That could get me killed!
I'd rather be politically incorrect and alive than PC and dead!
Correlation does not mean causation. If variable A and B are correlated, there are at least four possibilities.
1. A causes B.
2. B causes A.
3. Another factor, C is at play. One of the three causes, directly or indirectly, the other two.
4. Coincidence.
Once a correlation is determined, one has to look beyond the numbers to see which one of the four (or something else) is at play.
Very commonly the third is at play. By way of an example, let us assume A is "commits a crime" and B is "is a member of a certain group." The common assumption (1) is that membership in the group causes an increased probability of committing a crime. The second possibility seems ludicrous; committing a crime causes one to be a member of the group. If the sample is sufficiently large, the fourth possibility can be ruled out.
That leaves us number 3 to deal with. Researchers deal with this by "controlling" for other variables. They examine subsets of the data that contain another variable in common, for example, poverty. If we stratify the data based upon socio-economic status, does it change the correlation between membership in the group and criminality? Is it poverty that actually causes the increased propensity to commit crime? A disproportionate level of poverty among members of a certain group, combined with poverty increasing the propensity to commit crime, can create the impression the membership in a certain group increases the chance of one's becoming a criminal.
That is not to say that we should ignore the correlation. We should just not assume that it means causation. The correlation is an appropriate explanation of (and a justification for) a disproportionate number of members of a certain group being incarcerated.