19 out of 30...cripes, I guess Duke Snider was right when he said that Jackie Robinson wasn't really that fast.

In real terms, though it can't be "less than a single game's difference", can it? Either it never made the difference in a single game (in which case the answer is "0") or it won at least one game (in which case the answer is "1 or more games"). I think the scale is too small to just take the average.

Well, I don't know that this is really true. Suppose for instance:

Tie game, bottom of the ninth, 2 outs. Jackie Robinson steals home and wins the game. Did the steal of home make the difference? We don't really know -- the batter might have gotten a hit, or the Dodgers might have won in extra innings, or whatever.

Similarly,

Dodgers down by 1, bottom of the ninth, 2 outs. Jackie Robinson is caught stealing home and the game is over. Once again, we don't really know if he cost them the game.

Actually, by its very nature, a stolen base or caught stealing will never change a loss to a win, only a loss to a tied position or a tied position to a win, so really there is no such thing as a stolen base directly leading to 1 more win. Maybe half a win, but even that's a stretch -- and of course a more textured analysis (e.g. WPA) will give you fractional wins if you are looking for microstructural wins.

Even in less ridiculous situations, we still don't really know: runner steals second with 1 out and scores on a hit by the next batter. Next two batters are retired. Well, was the steal run-producing? Maybe the runner would have gone first to third on the hit and scored on an ensuing out. If the team ends up winning by one, maybe the pitching in the following innings would have been different if the game had been tied (maybe the other team would, or wouldn't, have bunted; maybe IBBs, infield in, etc.., let alone pitch selection.)

I considered looking at it via win probability, but decided to stick with run expectancy (and 9.7 runs/win in the 1947-56 NL) so I could address all 30 tries in the story. The attempts didn't seem to come disproportionately in either high- or low-leverage situations, so I suspect the result would be pretty similar.

The failed squeeze bunt was his last recorded attempt, with the bases loaded and 1 out against Hoyt Wilhelm in 1956.

It seems to me a very big item is left out of this analysis. A run scored in the bottom of the 9th in a tie game translates to a win. My guess is that there were more than a couple of high leveraged steals in his list. I'd like to see the same analysis using WPA. Though I was a young man then and don't remember them well, it just seems to me these attempts weren't random 9.7 runs per win events - they were saved for times with the game on the line.

Robinson’s final total was 8.30 runs. That mark hardly justifies the fame he garnered for stealing home.

I guess there are more and less sympathetic ways of interpreting this claim - but I'd say that this would be a very weird and far less fun world if fame tracked exactly to run expectancy added.

The problem is that the empirical numbers available at winexp.walkoffbalk.com have questionably small sample sizes for many specific situations (~200 games), and that's the only access I have to WPA data. I can provide a complete account of every attempt here if someone else wants to crunch the WPA (it would have to be customized for a 4.49 R/G environment).

I'm almost flabbergasted that Rosenheck doesn't even mention one of the huge side benefits of Robinson's attempted steals of home, both the successful and the unsuccessful ones. And that's the disruptive effect that they often had on the pitcher.

You have to remember that in Robinson's day the full windmill windup was in effect whenever there was a runner on third, or with runners on second and third, or with the bases loaded. Robinson may have attempted only 30 steals of home, but there were countless hundreds of other times when he would take off down the third base line screaming like a maniac, often with a telling effect on the pitcher. There are no numbers I see on the number of balks that this caused, but whatever it was, that would be only the tip of the iceberg.

Robinson's steal of home in the first game of the 1955 World Series is remembered today, but the Dodgers lost that game. What is far less remembered by those who didn't see the third game of that Series, when the Dodgers began their comeback, is the way that his baserunning completely spooked out the Yankees, in particular their starter Bob Turley and their rookie leftfielder Elston Howard.

In the bottom of the second Robinson got a one out single, moved up to second when Sandy Amoros got hit by a pitch, and went to third on a bunt hit by Johnny Podres. Then with Junior Gilliam at bat, and Turley in his double pump windup, Robinson repeatedly raced halfway down the third base line, screaming at the top of his lungs. Turley got so rattled he nearly threw a wild pitch, and wound up walking Gilliam to force home a run. Stengel had seen enough and yanked Turley out of the game. He never pitched again in the Series until the eighth inning of game seven, when the Yanks were trailing by two runs, in spite of the fact that he was the Yankees' righthanded ace in a year when the Dodgers had but one lefthanded batter and one switch hitter in their regular starting lineup, and in spite of the fact that in the World Series Stengel regularly used his starters in relief roles on short or no rest.

Later in the seventh inning, Robby led off with a double, and then pretended to stumble while rounding second. But when Elston Howard took the deke and threw behind him, he immediately righted himself and reached third easily. Two more runs followed.

None of Robinson's antics like this ever showed up in a box score. There is no real way to quantify them. But that's the fault of statistics, not of Jackie Robinson. Like a pistol flashed but not fired by a bank robber during a holdup, the mere threat of the "gun" was often just as effective as the actual "shooting" in achieving the purpose.

Ah, what I can do is include the leverage of each attempt and multiply it by the run expectancy. I Remember When, your instinct was exactly right: not only did Robinson's attempts occur in situations of above-average importance (the average LI was 1.3), but his success rate was better in higher-leverage situations and worse in lower-leverage ones. Accounting for the effects of leverage, his steals of home had the same impact on his team's likelihood of victory as 15.7 randomly distributed runs--so nearly double the raw total. I don't think this changes the fundamental conclusion of the piece--two peanuts (0.8 wins plus 0.8 wins) are still peanuts (1.6 wins in a 10-year career). But in the interest of accuracy, the story would have been better if I had included this factor (although I would have had to have cut something else to make it fit in my word count).

Jolly Old St. Nick, sure, they can be quantified: you would look at the aggregate performance of opposing P when Robinson was on third (controlling for the quality of the opposing batters) compared to their performance with all other runners on third. I don't have the statistical chops to do this, but a Retrosheet-meister certainly could.

Jolly Old St. Nick, sure, they can be quantified: you would look at the aggregate performance of opposing P when Robinson was on third (controlling for the quality of the opposing batters) compared to their performance with all other runners on third. I don't have the statistical chops to do this, but a Retrosheet-meister certainly could.

That's more work than I would wish on my worst enemy, but if RS-M'r would do this, he should also make note of the number of games like that World Series game I mentioned above, where the overall effect of Robinson's disruptiveness was so immediate.

BTW Turley also suffered an even more humiliating loss in a 1955 contest against the Tigers, when Earl Torgeson stole home in the tenth inning to end the game. I heard that game on the radio in Detroit, and the Tiger announcers all but predicted it, after having seen how exaggerated Turley's windup was. Once Turley switched to the no-windup motion in late 1956 (along with Don Larsen) he became a much more effective pitcher.

I'm not convinced that "baserunning dancing" ever shows up as a tangible benefit.
But if it ever does, this might be the guy.

That said, kudos on writing this in the midst of the wave this week.
These details, by any account, obviously don't diminish Jackie Robinson's remarkable legacy.

If anything, they allow him to be treated as more of a true equal - which he and his predecessors always have deserved.

At the Hall of Merit, we've collectively discovered African-American pioneers that even the HOF's recent laudable effort didn't quite grasp. And we've also dismissed entirely some of earlier Negro Leagues HOF electees, while finding others deserving yet overrated.

No racism. No pandering. That's the goal.

1.6 wins from steals of home alone in a 10 year career seems like a lot, but maybe it's just me.

Is there a record of the number of times Jackie Robinson scored on a balk?

Robinson repeatedly raced halfway down the third base line, screaming at the top of his lungs.

Oh sure, but when ARod tries it...

Jason Kendall--perhaps it is, compared to our expectations of what one could generate via the play. But 1.6 wins over 10 years from *any* play is still little more than a rounding error. Would our assessment of Robinson as a player (Hall credentials, all-time rank among 2B, etc.), be any different if he had hit one more HR in each of his seasons (1.8 wins)? I hardly think so...

I just posted a follow-up at http://bats.blogs.nytimes.com/2009/04/22/robinson-on-the-bases.