The Drubbing

Would this man force more errors if he could make contact with the ball?

Perhaps more than any other sport baseball is subject to the cliché. With so many games a year and so many hours of radio/television coverage for each of those games, things get repeated a lot. As with anything else, once something gets repeated enough people begin to accept it as true. It is an amazing thing about the human mind, once we hear something a few times, as long as it seems logical, we accept it. One of them has been bothering me for a while, and until now I didn’t have a way of looking at it. Thanks to the magic of baseball-reference.com, now I can.

Whenever someone like Ichiro, Torii Hunter, or Jose Reyes reaches first on an infield error one of the announcers will inevitably say “That shows you how speed forces defenses to rush which creates more errors”. This sounds perfectly logical. If I am at shortstop and Usain Bolt is running to first I know that I have to hurry the throw. However, because I tend to overthink things I always wondered if it was true.

Before I explain the methodology I used, let me first admit that it is not perfect. I took two different sets of data and looked at the statistics as best as I could. It is rudimentary statistical analysis at best, and would not qualify to be published in any scholarly journal. However, it can give us some illustration of what is going on So here goes.

First, I needed a way of measuring the number of times a player has reached on an error. As I said earlier, baseball-reference.com now has this data. With that data, I had to find a metric that measured speed, and I used stolen bases. It is not a perfect measurement of speed, but I think we can agree that to accumulate a large number of stolen bases a player would have to possess some decent wheels (triples also seems possible, and maybe I will find the time to do that analysis later).

Using the all-time stolen base leaderboard, I made a chart of the total reached on error (ROE) as well as total at-bats for the top 50 players we have data for. This list did not include anyone whose entire career wasn’t post 1954 (that is when the data for ROE begins). I also eliminated any current players from the list to ensure that we had a complete data set for a career. Lastly, to account for differences in the length of a player’s career, I made their career ROE a function of their career total AB’s. That left me with a number of AB’s per ROE. Then, I did the same thing for 50 players much further down the list I started with number 769 on the list Don Kessinger and took the next 50 players eligible using the same criteria I used above (and since I am sure you are wondering, I picked number 769 on the list because he had exactly 100 career steals and I needed to start somewhere). So, with the methodology out of the way, let’s look at some numbers.

The top 50 stolen base leaders averaged 69.05 AB/ROE. The lower 50 players averaged 78.27 AB/ROE. So, over the course of a career, the lower ranked players averaged 13.3% more AB/ROE. Over 5,000 AB’s that is 8.53 more ROE for the “faster” guys. Put another way, assuming 500 AB’s a year (which is quite low for the faster guys) that is .853 additional ROE a year for the faster guys. If this cliché were true, I think we would expect to see it happen more than once a year.

On the chart below you can see all of the AB/ROE for the players I looked at. In it, we can see some oddities. Kevin McReynolds has a lower AB/ROE than Lou Brock, Rickey Henderson, Tim Raines, Vince Coleman, and Kenny Lofton. If you look at the list long enough, there are a lot more oddities in it.

So, what does all of this mean? Well, it doesn’t definitively mean anything. This is not a perfect statistical analysis. Stolen bases are not a perfect predictor of speed. The sample size could be larger. However, despite those shortcomings, I will draw some speculative conclusions.

This cliché is overplayed and does not have much to it. If you chart this data, neither one shows much of a relationship. In fact, for the lower players, it seems that the lower they get on the list the lower their AB/ROE is (even though its not a statistically significant relationship).

In short, we hear this statement so much because it appears to sound right. It makes sense to us that this should be true. But like many other things in life, our brain isn’t always right. My guess is that if we did a perfect statistical analysis we would see the same result, ROE is not a product of speed. More likely, it is probably a product of division, league, the time period played in, turf/grass, strategies employed at the time (like a shift), and last but not least, luck.