It's hard to answer your question, since both ethics and rationality are fuzzy, slippery concepts that need to be nailed down firmly to reason meaningfully about.
Or, put less pretentiously, I don't know.
When you say "more dangerous" do you mean "inherently more dangerous" or "more likely to commit crimes per capita"? Because if it's the latter, then depending on how you define "race" some of these statistics are already known.
"Known"is a bit strong, since the source for the statistics can be debated.
Humans are notoriously great at pattern matching, and seeing things that aren't there simply based on a prior mental model. Stereotypes of all kinds are a result of this.
Evolution tends to select for short-term survival over long-term viability. It's absolutely a mistake to believe the two are aligned.
The reality is that humans are (probably) the first Earth species to be capable of abstract non-physical modelling of the future. But we have a huge amount of behavioural baggage which guarantees that we tend to ignore long-term warning signals in pursuit of short-term gains motivated by evolutionary heuristics.
The heuristics work fine until they don't. Species discover this the hard way all the time.
It would be nice to think we're not one of those species, but the jury is still out on that.
we can see this most clearly in our politicians' insistence on their particular model of the economy as their priority, rather than the actual concerns of the citizens they are responsible to.
They don't.
Or at least, I'd be highly amused to see someone attempt to square the Trump platform with Burkean thought, and that's a relatively smallish difference, both evolutionarily and intellectually speaking, compared to others.
And if my group was more aggressive, there are some probability that yours didn't even survived and mine did. Or if I was more "racist", perhaps my genes would spread better. So, some genetic that makes me don't like "different" people might have a role.
If our genes didn't evolve (in that aspect) in some few thousand years, simply because there wasn't enough ambiental/social/whatever pressure in that direction... here we are.
http://www.bjs.gov/content/pub/ascii/vvcs9310.txt
It's also the case that violence is so unlikely at any given encounter that it isn't rational to worry about it (in the US, there are billions of stranger-stranger encounters in a given year, millions of incidents of violence).
I guess someone might argue that arranging for encounters that are 99.9% likely to be violence free compared to encounters that are 99.8% likely to be violence free is worthwhile, but their glances probably aren't powerful enough to actually create that difference.