[jkyle]

I do not think most would consider 5 or 6% is "much higher" or that if 90% is "extremely low" that 84% is not also "extremely low".

Particularly when you take variance and margin of error into account.

With such a tight range, I would wager one could only definitely claim they were probably approximately within a percent or two at best.

[I_dont_know]

I don't know if percentages are the right metric to use here. You can certainly make up a scenario which is completely absurd, but still has your percentages. Assume 100,000,000 green people, and 100 blue people on a distant planet. Nine blue criminals kill 9 blue people. One green person kills one blue person. So that's 90% blue-on-blue murder, and an 10% of the blue population are murderers, and there is a 10% murder rate among the blue population. Now lets say there were there were 9 green people murdered by 9 green people, and one 1 green person murdered by 1 blue person. Again 90% green-on-green murder, and a 0.00000001% murder rate.