When there just isnt enough raw data to have a decent sample size , wacky numbers like this appear , theres no way that training passing speeds up with increases in height. as others have said , its more likely due to the people having a high sublevel and submitting very little information to the training analysis , thus giving a inflated or seemingly falsified appearance of this phenomena where passing is claimed to train faster on bigs than guards.
You put two things on the table.
One, that the "0,60-0,65" for his age/height is not accurate because of small sample and possible high sublevels for the players submitted.
Two, that "the taller the player the faster he trains in passing" is inaccurate, because of the aforementioned factors.
Well, a sample of over 3000 training sessions in passing, is not to be dismissed that easily.
But considering that the the sample of a certain age and (especially big)height will be consederably smaller it can produce false results, regarding the accuracy of a PA training session in terms of pop ups per session.
Having said that, the same sample, seems im my mind, much bigger when it comes down to a more general conclusion, such as the height to speed ratio. It is much harder to get that wrong, with a sample of 3000+ submitted PA sessions.