"Statistically, a baserunner bunted over to 2nd is more likely to score."I don't know if there's any data to specifically support this.What the tables show, all things being equal:
1) If you sac bunt a guy from 1st to 2nd, you will score less runs in that inning
2) If you sac bunt a guy from 1st to 2nd, you will be less likely to score 1 run in that inning
I read these as both contradicting the quoted statement.
The problem is, in game situations all things aren't equal.If you approach this from an EV (expected value) point of view, situations involving the current batter, current pitcher, and next batter will tweak the numbers somewhat.If someone hits bad, the bunt may improve either 1), 2) or both.The thing we DON'T have is context specific information about this batter/this pitcher/next batter.I wouldn't be a bit surprised if the analytics department HAS calculated these and hands them to Molitor.
It's pretty rare you see 2/3/4 bunt; if it's someone that is a worse batter, it may be the best play you have.
I'm the first to say, in general, all the bunting is bad.I am willing to consider that analytics Molitor has that we don't make it the best option.