Joel Waldfogel notes a Jennifer Brown study showing that golfers do worse in tournaments where Tiger Woods participates than in tournaments where he doesn't, and concludes that the ferocity of Tiger's competition causes players to "quit." I haven't seen the study, so I don't know if it controls for possible ceteris paribus problems (such as the fact that Tiger these days tends to skip the easier tournaments and plays a higher percentage of his tournaments in "major" courses that are "Tiger-proofed" and hurt everyone's score), but, assuming that it does, there is another possible explanation.

Recall the movie "Tin Cup," where Kevin Costner's character insisted on trying to overshoot the water hazard in one stroke rather than laying up and avoiding the hazard with a safer second stroke. Golfers facing Tiger may take similar riskier shots that hurt one's expected score, but increase the variance and the likelihood of an abnormally exceptional score that can beat Tiger. (Poker players are certainly familiar with the phenomenon of the player who is in a hole and starts playing much looser, a strategy that has a lower expected mean performance, but a higher chance of taking down a big pot.) Golfers who are doing this will have mean scores indistinguishable from golfers who lose a stroke or two because they are "quitting," but they are clearly doing the very opposite of quitting.

Update: Ok, now that I've looked at the study, Brown claims to have tested for riskiness, and did control for golf-course difficulty. If Brown's result that superstar presence in a tournament decreases performance by non-superstars is generalizable to other fields, that has interesting implications: for example, it suggests that gigantic signing bonuses for Supreme Court clerks are a bad idea.

From the Economist, a longer-term view.