Those numbers may be due to a flaw in the methodology. Couldn't they also be due to random variation? Either way, I'm amazed whenever people use the figures from one month, one quarter, or even one year to broadcast that massive change is underway or required.