I'm not claiming that de-escalation implies ignoring antisocial behavior or failing to confront.

I'm claiming that some antisocial behavior is only triggered by some degree of escalation.

I wonder what you base this on. What would statistics backing this up look like?

It's a good question.

It would be violent crime trend correlation with de-escalation training. Or even with complaints of police aggression.

Even more useful would be to separate out assault + battery where the victim is random vs. non-random, IE domestic or gang.

I had a look around and there is very little actual ecidence for detrimental effects. Most things seems to be exaggerations by politicians who want to be tough on crime.