Incorrect studies is one thing. The site I linked to uses a model to calculate finger distance traveled - here's how they calculate it and here are some other parameters. He runs all layout through about 15 books of typing to see what takes longer. I don't think Dvorak is great, but it is less effort / time to type. It's not good enough to warrant relearning typing from scratch for me, but that's a different discussion.
Yeah, I know. I don't believe in the site you linked. Ask a biomechanicist how many degrees of freedom the human hand has and you'll get a number between 21 and 27... and you have two of them. Throw arms in there and you've got another 6 or 7... x2. Your website only accounts for 16 parameters in its math. Statistically, they've oversimplified the model by a factor of five at least. A system that complex is best developed empirically, not theoretically, which means iterative testing. Iterative testing developed the QWERTY keyboard, less iterative testing developed the Dvorak keyboard. Empirical testing bears out that the QWERTY layout is better.