This happened once, I was white and had misplayed the opening:
[FEN "r1bq1rk1/ppp2p1p/3p2p1/2nPp2n/1PP1P3/2N1bP2/P2NB1PP/R1BQ1RK1 w - - 1 12"]
I thought that Kh1 would lead to a deadly black attack, so I sighed and played 12.Rf2. There followed 12...Qh4 13.Qe1 Qxf2+ 14.Qxf2 Bxf2+ 15.Kxf2, I had lost an exchange, the queens were off and I assumed I would be lost.
I then proceeded to blow black off the board -- black's rooks turned out to be completely useless, my pieces found great squares, and I won in 32. Of course there were also many more mistakes, according to the computer.
So if a completely unintentional sacrifice can work, then so can intentional sacrifices, right?
I believe that a good move is a good move. If a sacrifice gives compensation (in terms of great squares for pieces, or an attack, etc), then it does, regardless of the level of the players.
Yes, weaker players are more likely to make mistakes and lose the compensation. But guess what, they are also more likely to make mistakes without playing the sacrifice! Maybe the sacrifice is the best move, and not playing it is already a mistake.
How well do these tactics work for amateurs? Do many make "correct" sacrifices (from the master point of view) and then lose because of incorrect follow up? Do amateurs sacrifice at the "wrong" times (when positional considerations don't warrant it)? Or is it usually to be the case that if "someone is good enough to sacrifice (correctly), they're good enough to win?"
This is hard to answer because at one hand, yes, of course weaker players are much more likely to screw it up, because that's what weaker players do, that's not really related to the move being a sacrifice. Weaker players will also play a conscious sacrifice less often, because it's easy to understand that you'll lose a piece, and less easy to understand that you get positional features in return that make up for it, or that the tactic actually does lead to a forced mate in 6. So most amateurs play it safe (but not all).
But if we assume the opponent is at the same level, so also an amateur, then I think sacrifices can be very effective at amateur because the opponent is also less likely to see them coming, and more likely to be confused by them.
So I don't think a general answer can be given, and think what matters is that a good move is a good move. Many moves gives up some type of advantage in return for another type, and sacrifices are just one example of that.
I think amateurs don't play them as often as they should, because of unnecessary fear.
Edit: I guess I need to add the whole game
1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 O-O 5.Nf3 d6 6.Be2 Na6 7.O-O e5 8.d5 Nc5 9.Nd2 Bh6 10.f3 Nh5 11.b4 Be3+ 12.Rf2 Qh4 13.Qe1 Qxf2+ 14.Qxf2 Bxf2+ 15.Kxf2 Na6 16.Ba3 Nf4 17.Bf1 c5 18.dxc6 bxc6 19.Nb3 Rd8 20.Rd1 Bb7 21.Na5 Rab8 22.c5 Nc7 23.cxd6 Nb5 24.Bxb5 cxb5 25.Nxb5 Bxe4 26.Nxa7 Ra8 27.N7c6 Bxc6 28.Nxc6 Rxd6 29.Rxd6 Rxa3 30.Rd2 Nd3+ 31.Kf1 Nc1 32.b5 1-0