Not on a small scale. But on a larger-scale discussion, you absolutely should take that into account. Say you're planning to go fight a dragon with a 5-player team. In your planning, you presumably have some idea of roughly how much damage your group can deal per round, on average.
And that average ABSOLUTELY includes the likelihood that you're going to roll one or two crit fails/successes. You'd instinctively take that into account.
A similar principle applies here. "On average I'll do a little bit of damage to my opponent's deck with my Warrior - that average amount is increased somewhat by the possibility that I hit my opponent's Warrior."
The upside is so huge here that the overall "average damage" (as it were) is probably dragged up noticeably even if the likelihood of the event is small.
But this is a small scale. You have one chance of hitting something with Warrior before you want to turn it into a Hero, and in 90-93% of cases, it doesn't hit your opponent's Warrior. So if your opponent willingly gave himself a small disadvantage in order to prepare for his Warrior getting hit, 90-93% of the time he just has the small disadvantage and that's all it did for him (and probably around 7-10% of the time he still loses because he's still one shuffle behind you even if he has a backup Page going on). Obviously if you want the second Page for other reasons, you want it anyway so yeah.
If I was teaching Dominion to sentient computers, I would mention that when the disadvantage from getting a second Page instead of something else is almost negligible, as well as when you're already so much ahead that your opponent's only chance of winning is pretty much to hope for that small chance of hitting your Warrior, you might still want a second Page just in case your Warrior gets trashed. Unfortunately, we are humans and humans are garbage at estimating any probabilities so it's better to forget about these edge cases when you're actually playing the game.