As many of you know, this offseason proved monumental for the site as we added a wide array of evaluative metrics, becoming one of the primary sources for player valuations. One of these additions, UZR, the fielding metric designed by Mitchel Lichtman, enabled analysts and readers alike to incorporate the fielding aspect of baseball into discussions. Several aspects of fielding combine to provide the final UZR figure, and two, range runs and error runs, are of particular interest given their reputations in the world of conventional wisdom.
The conventional wisdom goes that the better a player’s range, the more likely it is that he will commit errors. The underlying reasoning is that the player will be able to get his glove on more balls, thereby not only giving himself a chance to make more plays, but also the chance to mess up on more plays. I like to refer to this as ‘The Abreu Complex’ as Bobby Abreu used to be considered a solid fielder by many fans because he rarely made errors. The issue of course is that his limited range prevented him from covering more ground: he didn’t bobble many balls but he couldn’t get to balls that others would catch and that he might then bobble.
With the different components of UZR freely available on the site, I decided to see if the conventional wisdom held true – does more range really translate to increased errors? I pooled every player with at least 100 innings at a position over the last three years, removed catchers, and wound up with 722 player position seasons. Correlations were then run for infielders and outfielders with regards to both range run and error runs. A correlation is basically a statistical test that measures the lack of independence of two random variables; in this case, do range and errors relate strongly to one another in the sense that as one goes so too does the other?
For two variables to be considered to have at least a moderately strong relationship, a correlation coefficient of at least 0.40 would be needed. Among infielders, range runs and error runs produced a 0.10 correlation, while outfielders featured only a slightly stronger relationship at 0.15. Neither group of fielders exhibited anything close to a moderately strong relationship between range and errors, leading the conventional wisdom astray: more range does not necessarily result in more errors, no matter how much sense the statement might make from an intuitive standpoint.
Even when I restricted the data to at least 800 innings at a position, the correlations remained virtually the same–0.16 for OF, 0.11 for IF. Based on this data it seems that there are certainly cases where range and errors relate to one another, but it is in no way a foregone conclusion that more range results in more errors.
Print This Post