Yeah anecdotes don't tell you much. You may have noticed I was also replying to an anecdote.
What tells you more is that the diversity statement exists and they say it's used as part of scoring. Therefore, unless the amount of score it counts for is infinitesimally small, some people win/lose based on the content of their diversity statement.
Was that me? Who knows. But unless the whole thing was just busy work for no reason, it was probably a bunch of people.
How many? Who knows. I'm sure you'd agree that it would be interesting if somebody published that data! Maybe the new NSF will be more transparent than the old one.