Perhaps another way to look at this is who gives you a greater chance for a runner in scoring position (RISP). In the example, above, Trout (32/39) puts a RISP 32 times -- Cabrera does this 3 times. So in terms of giving your team additional chances to score, Trout is + 29; relative to Cabrera, Trout is 9.67 times more likely to put a RISP. Now, of course, putting a RISP and the probability of the RISP scoring is dependent on the rest of the lineup behind the player. Say, for example, Trout and Cabrera are followed by a .300 hitter in the lineup; the probability of Trout or Cabrera producing a run is a function of stolen bases and the average of the guy behind them. So for Trout, we can say the probability of his stolen base turning into a run scored is 29 * .3 = 8.7; Cabrera's probability would be 3 * .3 = 0.9. SO, while Trout produces a much greater likelihood of scoring, the difference between Trout and Cabrera is 8.7 - 0.9 or 7.8 runs spread out over the course of a year.
Now, in sabermetrics, there is a calculation for runs created by stolen bases:
RC = [ (hits + walks - stolen bases) * (Total Bases + {.55 * Stolen Bases}) / At Bats + Walks
Assuming I didn't muck up the equation of the math, Cabrera's RC with the stolen base effect is 146, and Trout's is 161.
We can also express this per game or as RC/27; for Cabrera this is 5.41 and Trout, 5.96. OF course, since a player averages
about 4.2 at bats, the real effect on average is (RC/27)/4.2, which for Cabrera is 1.29 and for Trout 1.42.
The difference in the stolen base effect can then be determined by comparing Cabrera's and Trout's no-stolen base-adjust RC to their basic RC metric.
For Cabrera, this is 145.6, so the difference in stolen bases for Cabrera is 0.4 RC. For Trout, the effect is 30. That means that at this point in the year, Cabrera has produced 0.4 more runs through steals, while Trout has produced 30 more runs through steals.
As you can see from the stolen base runs created formula, the average base stealing effect is used (.55). This, of course, may vary across players, so that the average may not represent the actual performance of the player.