> I'm not sure that's true.

It is backed up by the author's analytics, though. Perhaps there are valid criticisms about her sampling methodology and the sample size at the very high end, but it's a good start.