If the team sacrifices when behind by one run, then instead of increasing the chance of winning, it decreases the win expectation by 3.6%.
Now, some of you are probably saying things like “Yeah, but what about Jason Kendall? He sucks anyway.” Well, we have research on that too. With a runner on first and no one out, a bunt makes sense (meaning the team’s chance to score at least 1 run will increase) if the batter’s line is worse than .177/.192/.153.* Jason Kendall is bad, but his .216/.307/.265 still makes swinging away the correct strategy.
I recommend reading the whole chapter and the whole book.
*If you’re wondering how slugging can be lower than BA, the numbers were calculated independently.