Statistical shenanigans.
Even if you skew that data the other way it's going to show the same thing.
But if you include the the 96-99%, I have a hunch the lines won't cross. And if you wanted to, you could compare the top 1% to the bottom 25% the lines would flop and never touch.
