Here's an interesting datapoint from SABR's Clem Comly which he posted on SABR-L. The question was asked how often a batter is successful in stretching a single into a double. With the bases empty Clem came up with the following numbers:
2B 1B+out Pct
1962 2315 89 95.9%
2003 4962 114 97.8%
In other words batters were successful 97.8% of the time when going for a double with the bases empty. Certainly on a large percentage of the doubles there was no play on the runner but it appears that major leaguers are pretty good at determining when they can make it and when they can't.
Given these numbers one wonders whether runners shouldn't take more chances, since the odds of scoring a run increase from 12 to 19% with a runner on second only as opposed to a runner on first depending on the number of outs.