As I think you realize and as others have pointed out, this is a sort of impractical approach to looking at starting hands and there's lots of other things to consider, but I suppose it could be worthwhile as an exercise. After all, what does it even mean to "play" a hand?--do you mean open raise? open limp? complete the small blind after everyone else has limped? etc etc.
Let's set some conditions to make things easier. Assume that you are first to act at a table of n players (you're one of the n) and that starting hands are easily rankable from best to worst--with what hands can you say that you're better than 50:50 to have the best hand at the table? Note that the knowledge of other players entering the pot would affect all of this since they are presumably playing the top X% of hands from their position (whatever they have set to X) and everything dissolves into some game theory type stuff.
Start with the case of 3 players. It would be natural to think that if you have a hand at the 33.3rd percentile of hand rankings, you're a favorite to have the best hand. But that's not exactly the case. Since you're at the 33rd percentile, the chances that the other two players are both below that are 0.67*0.67 -> 44.9%, so there's actually a 55.1% chance that either or both have a better hand. The breakeven point is actually 1 - (.5)^(1/2) -> 29.3rd percentile. In general, the formula would be 1 - (.5)^(1/(n-1)). The graph of this will be a little above a graph of 1/n at the points of interest, but like 1/n it decreases and approaches 0 as n gets higher. It turns out 1/n is actually not that bad as a rough approximation though.
As alluded to earlier, though, this doesn't have any precise application except to show that yes, in general you should have a tighter range with more players involved and a wider range with fewer players, and that the difference between adding a 4th player to a 3-person table should have more of an adjustment effect to your range than adding a 9th person to an 8-person table.