U.S.
The Breakdown
The city published figures showing a decline. A separate public dataset, covering an overlapping period, shows something closer to flat with two categories rising.
Both are real. The gap comes down to what gets counted and when it gets counted: reported versus verified incidents, and whether a case gets recategorised after review. Those are boring methodological choices that produce very different headlines.
The honest version is that you can support either narrative with a real number, which is exactly why both sides are doing it.
What neither dataset settles is whether people feel safer, which is a different question and is not measured by either one.