August 16, 2001

Aim For The Head

More Reaching on Errors

by Keith Woolner

My previous column on reaching-base-via-error rates for batters generated the most responses yet in the short life of this column, which I think warrants a follow-up column responding to some of the reader mail on the topic.

By far the most common point made by readers was that I neglected to consider handedness when I presented the list of best and worst ROE%. T. M. was the first to write in on this issue:

I noticed something about the reaching-on-error leaders and trailers. The leaders are all either fast guys who make a lot of contact, which you might expect, or right-handed power hitters, which you wouldn't.

Among the "worsts," they are all left-handed pull hitters except Chad Kreuter, who's a switch-hitter.

It seems to me that the most important variable is hitting grounders to the left side of the infield, where the throws are longer and small mistakes lead to errors.

T.M.--and about a dozen other readers--were right on the mark, and it was careless of me to miss this in my original analysis. In fact, right-handed batters are 29% more likely than lefties--per plate appearance--to reach on an error:

Batter ROE%, 1978-2000
Switch    Left     Right
1.12%    0.95%     1.23%

The next most common question is along the lines of what Cooper Nielson wrote in to say:

I enjoyed the analysis of error rates. One thing I'd like to see included is the batters' groundball/flyball ratio. It stands to reason that players that hit more ground balls would be the beneficiaries of more errors (you rarely reach base because an outfielder drops a ball).

Intuitively, a propensity to hit ground balls does suggest a greater likelihood of reaching by error. I took all batter seasons with 500+ PA from 1978-2000, and using the groundball and flyball data (where available, which was not for all batted balls), broke the seasons into quartiles, and calculated each quartile's ROE%. Higher G/F ratios, which mean that batters are hitting more ground balls, are associated with higher ROE%, as you can see in the following table:

Batter ROE%, 1978-2000
Quartile      Min G/F     ROE%
Q1             0.31      0.95%
Q2             0.75      1.10%
Q3             0.92      1.16%
Q4             1.12      1.31%

A few readers thought that looking at plate appearances as the unit of opportunity wasn't the optimal choice. Dan McLaughlin writes:

[I]f you wanted to use ROE to measure differences in speed and hustle rather than the impact of those differences, one thing to check would be to refine the study to look at ROE as a percentage of balls in play (rather than plate appearances overall), or better still--if the data were available--ROE as a percentage of GROUND balls in play (there's really not much ability in reaching on a dropped fly ball). That might yield statistical evidence that would shed light on similar "intangible" qualities that are even harder to measure.

P.F. echoes Dan's observation:

While the impact to OBP is probably best measured by the ratio of ROE to plate appearances, wouldn't the measure of this as a skill be best calculated as the ratio of ROE to BIP (balls in play)? It seems like the [list of] career leaders in ROE% is populated by a fair share of hack-a-matics.

There are a couple of ways of looking at this. As Dan and P.F. (among others) suggest, in order to reach base on an error you generally have to the ball on the ground, or at the very least put the ball in play. Eliminating the plate appearances in which that doesn't happen yields a "purer" measure of the ability to turn ROE opportunities into actual ROE.

However, this isn't sufficient if we want to know the magnitude of the impact on the game as a whole, or in other words, the actual value of an ROE ability. Only a fraction of plate appearances incur an ROE opportunity, and that fraction systematically varies across players. ROE per ball in play (or per ground ball) is a conditional probability, and we need the probability of the condition itself to complete the sequence:

Prob{ROE in a plate appearance} = Prob {ROE given a ground ball} x Prob {ground ball in a PA}

This is not to suggest that looking at ROE/GB or ROE/BiP is without merit. Splitting the overall measure into component parts has some diagnostic value. It may suggest a coaching strategy to encourage certain players to hit more balls on the ground, if they are particularly good at inducing errors (speedy right-handed batters seem to fit the bill, given our evidence to date).

J.C. offers a few thoughts:

Wow, that low ROE list sure has a theme!

Of course, most errors of the base-reaching type should be made by the third baseman and shortstop. Slugging lefty pull-hitters don't have much truck with those left-side infielders.

I'm guessing that Jack Clark grounded into a lot of double plays. It would be interesting to compare ROE to DP...strong correlation might be found. Perhaps strong enough to merit adjusting the DP factor in offensive formulas?

Surprisingly, though the theory is sensible enough, we don't see strong relationships between hitting into double plays and reaching on errors. The table below shows correlations between seasonal the frequency of double plays and the rates of reaching via error (for the three ways of conditioning ROE opportunities previously discussed). The DP/Opp column is the rate of double plays per opportunity (defined as plate appearances with a runner on first base and less than two outs).

           ROE%    ERR/BIP   ERR/GB     G/F    DP/Opp
ROE%                 0.966    0.856   0.262     0.040
ERR/BIP   0.966               0.909   0.185    -0.017
ERR/GB    0.856      0.909           -0.002    -0.056
G/F       0.262      0.185  -0.002              0.148
DP/Opp    0.040     -0.017  -0.056    0.148

Remember that correlations close to 1.0 imply a strong relationship between the two measures (high values matching high values, and low matching low), and those close to zero indicate no relationship. Negative correlations indicate that the two measures move in opposite directions--if one goes higher, the other tends to be lower.

David Marshall writes:

Just had to point out that the limitations on your data set (1978-2000) probably caused some players to appear on the Low Career ROE% list whose full careers might not have. If the prime years of a player's career fall before 1978, they won't get any credit for their speed and hustle during those seasons. Only their "old and slow" years are being analyzed, but these stats are then labeled "career." So Carl Yastrzemski and Rusty Staub (for instance) are getting unfairly downgraded, while in Darin Erstad's case, his ROE is an accurate reflection of his batting and running style.

This is a fair assessment of the career data I presented, since it doesn't actually represent the full careers of some of the players listed. Additionally, since some current players are still young and haven't started slowing down during their decline phase, they may look better at reaching base now than they will at the conclusion of their careers.

B.K. writes:

Looking at your data on hitters who "create" the most errors and least errors, I saw a very distinct pattern. The hitters who create the most errors were overwhelmingly right-handed hitters (or switch-hitters) and the ones who were the least likely to create errors were left-handed hitters.

If we think about the distribution of errors by position, I would be willing to bet most errors occur at either shortstop or third base, particularly on ground balls.

As we've seen earlier, right-handed hitters do reach via error more often than lefties, so it makes sense that the left side of the infield produces more of the errors. The longer throw to first base gives these positions less time to recover from a bobble or hesitation.

Jeff Hauser thoughtfully muses:

It would seem as if "ROE" could [be subject to] enormous "park effects"--the subjective borderline between what is a hit and what is an error varies significantly between official scorers, and many (all???) teams have regular or semi-regular scorers.

Well, that brings up an "unintended consequence" of inclusion of ROE in OBP: if OBP was recognized as widely as it ought be, and ROE was included, official scorers would be under less pressure from players to grade errors on an artificially inflated/generous curve.

E.F. also suspect parks have something to do with it:

Interesting article about reaching base on errors. You might consider looking at home stadium as well. Being a Braves fan in the '80s I notice that Bob Horner, Rafael Ramirez, and Mike Hubbard are all in the career leaders. We all remember the shape of that infield at Fulton County Stadium!!!

I may address this in a future column, but there have long been suspicions that some official scorers in some parks are more generous than others in charging errors. One reader suggested a hometown bias in charging errors (home players getting fewer errors charged than visiting players).

I'm not sure what I would expect in this situation. Charging an error to a visiting player means taking a hit away from a home player, and I'm not sure that a hypothetically biased official scorer would be any more willing to take a hit away from one of the team's batters. It'll remain a mystery, at least for now.

Bob Evans writes:

I liked this, if only because I can't remember the last time I heard about Billy Sample or Ken Phelps.

I wonder if John Mayberry ever got frustrated that it never seemed to happen to him? Just nine times? He might have had more grand slams! O.K., maybe not.

Maybe more reachable, I wonder what Mo Vaughn would say if you told him of all active players, he's the least likely to reach base because of an error? Would he say, "You know, it always seemed to me like I never catch a break like that," or would he say, "Do you know, I never noticed?" Ask him next time you chat him up, would you?

I'll ask Mo about that next time I visit the Foxy Lady. By the way, I show no grand slams for Mayberry over the span of time.

And to close out this week's column, here's Chuck Nobriga, who puts the whole topic in perspective for us:

Any good statistic showing Johnnie Disaster among the leaders can't be worth a damn.

Keith Woolner is an author of Baseball Prospectus. You can contact him by clicking here.

Keith Woolner is an author of Baseball Prospectus. 
Click here to see Keith's other articles. You can contact Keith by clicking here