Of course, they can. There might be a question as to what extent they'd want to (as useful that information might be in assessments for a fighting force), but they clearly can.
How? Nobody sits around a pulls a rank amateur out and faces him off against a goblin one after another to see how long he lasts. Then suddenly one day he hits level 2 and they're like, "He's lasting 6 seconds longer on average now!"
1. Fights happen infrequently and people aren't timing them to see how much longer they are lasting.
2. Hit points don't look like hit points to anyone. They look like skill(the fighter parries a thrust and takes damage even though he's not hit). They look like luck(The fighter goes to parry one goblin and accidentally also blocks a second thrust he didn't see coming and takes damage even though he's not hit). They look like divine intervention(An arrow shot from behind thunks next to the fighter's ear instead of going through his head, even though he was unaware of the shot).
3. What the party can see are power improvements. The group of goblins ended quickly to a fireball, where before they would have been more of a challenge.
So to sum up, the party can see skill, luck, divine intervention(what hit points are), but not actual hit points, and they can see power levels. They also are not running scientific experiments to see if they can discover hit points.
With assessment in a context like boxing it would become ever clearer that there was a 10% difference between the two participants.
Yes. SKILL matters. Skill =/= hit points. It can be hit points sometimes, but often does not.
Gamers also know hit points as they are a player mechanic.
And are wrong so often they might as well not matter. Oh, wait...
Either way, the watchers can be aware of any flow of "lucky punches" one way or another and still make a clear assessment of an issue such as the durability of boxers.
Skill, yes. Durability, no. Durability would be punch each of the boxes while they are defenseless with exactly the same amount of force and see which one folds first. Durability is a matter of constitution, not hit points.