However, we cannot expect the sample population participating in such a test to be representative of much else. Keyboard lovers may be faster typists in general, and they are also more likely to try something hard like changing layout just to see if it works. Also, Ryan Heise's test has gotten much attention from this board. So the Dvorak and Colemak users will be overrepresented in the base population of the test as well.

Furthermore, the samples are too small to allow us to make predictions. It's really not easy. I agree with Jammycakes though, that it is pleasing to see such a strong presence from both Dvorak and Colemak there. And it was kinda nice to see Colemak beat Dvorak - which as we all know was the fastest layout in the world up to 2006 (based on a couple of 1-shot case studies, hehe).