We can infer that they are supposed to be generally consistent representations of expected damage (I expect them to represent cRatio rather than actual average pDif, as that's the simplest way of conveying the value), otherwise the point he was trying to illustrate would be meaningless. If the numbers picked were some random sampling out of the full pDif range, there's literally no meaning in saying that defense reduces damage taken, since the numbers could be -anything-.

This is further backed up by Bayohne's post; if the intent is to provide a simple illustration of how defense works to those who may not be as math-savvy, and they choose non-representative samples of the damage done, they are effectively lying to their audience. I doubt they would want the fallout of being caught directly lying about basic game mechanics (as opposed to merely bad testing methodology, as in the TH post, where they didn't assert any actual numbers).

Regardless, it wouldn't be too hard to test for mob level correction with a little level syncing, to verify one way or the other.