Someone walking down a lonely alleyway at nighttime would have reason to be more concerned if they noticed that a man was walking toward them, rather than a woman. They would also, presumably, be justified in being more concerned, to about the same degree, if it was a black man than if it was a white man. In fact, in the U.S. (and elsewhere), the black to white crime ratio is about the same as the male to female.
That wariness is now generally considered to be racist. Take the following video:
It’s actually quite perverse. What is accomplished is intimidating women through social pressure to avoid taking precautions that might keep them from being assaulted because it might make them look ‘racist’. Imagine creating videos mocking a woman as ‘sexist’ for taking minor precautions when alone with a man. It wouldn’t happen, and would be considered in bad taste if it was – but take the most criminal class of men, and suddenly it’s a public service announcement.
When people are in situations where just getting a vague feeling that they should get out of it can be the difference between being a victim or not, the attempt to undermine those intuitions as racist or sexist is, in fact, harmful.
It’s not clear whether these people (ie. the kinds of people that make the above video) are actually even aware that the members of certain groups have much higher likelihood of being criminals than other groups. For example, do they have any idea that 1/3 of black men spend time in prison in the U.S.? Would they still make the video if they did? Do they feel that a black man should be treated with no more suspicion than an elderly white lady?
A similar tactic works in guilting people into not opposing immigration from certain places, etc. I suppose prejudice is now considered unwarranted unless every single member of some population can be shown to behave in the undesirable way. To quote Steven Pinker “…according to Bayes’ theorem a decision maker who needs to make a prediction… should factor in the prior probability, such as the base rate of success for people in that group. If races or sexes are different on average, racial profiling or gender stereotyping would be actuarially sound…”