Win expectancy, but from 2002. Someone else can figure out win expectancy over the past decade in each situation if they want.
http://www.tangotiger.net/welist.html
The situation I'm thinking of is man on, no outs, down one run. If I'm reading this right, if you have that man on first base, you don't bunt because doing so will get him into scoring position, but will cost you an out and about 5% (0.331 to 0.282) in win expectancy...this of course is better than a double play, which would give you bases empty and two outs and drop your win expectancy to 0.042. However, if that man is on second base, with no outs the win expectancy is 0.437. Bunting him to third will cost one out, but the win expectancy only drops to 0.412 which is substantially better than the GIDP situation but still not better than if the guy had just swung away.
If you bunt the guy over from second base with one out though, you're going from 0.282 to 0.170, which would get your tying run 90 feet from home, but with two down, the sac fly is out of play and overall it just looks really stupid.
If the game is tied, though, win expectancies will change with the situation. Bunting the guy over from second base with no outs will actually increase your win expectancy. Generally it's not a bad idea to do this, then, if you are the home team, you know how many runs you need, and the score is tied. So there is a time and place for the bunt.
Again this is a win expectancy table from way back so things might change when a bigger sample size of data is considered, but just an example I thought would help some of you STFU already.