Baseball Musings: Racist Umpires?

August 14, 2007

Racist Umpires?

Time Magazine reports on a finding that there exists a small racial bias among Major League umpires. Of course, being a technologically backward organization, Time doesn't bother to link to the actual report. They also appear to be mathematically challenged:

According to the new study led by Daniel Hamermesh, a professor of economics at the University of Texas at Austin, Major League Baseball umpires tend to call more strikes when the pitcher is of their same race; when they're not, umps call more balls. It doesn't happen all the time -- in about 1% of pitches thrown -- but that's still one pitch per game, and it could be the one that makes the difference.

Starting pitchers are often pulled after 100 pitches, so it should be one pitcher per starter per game. It's not unusual to see 300 total pitches in a game, so we're talking about between two and three pitches per game.

(Correction: It wasn't clear from the Time article, but the 1% represents called pitches, so swings are left out. In that case, one pitch per game is about right.)

Secondly, this is an unpublished paper. In other words, it has not undergone peer review. Now, does this mean it's wrong? No, but other economists haven't had a chance to judge the work yet. So basically, take it with a grain of salt.

I do find this interesting, however:

Table 4 contains the results of estimating (1) separately by the presence of QuesTec in the ballpark. All of the estimated equations presented in Table 4 include controls for inning, pitch count, pitcher score advantage, and top of the inning. The direct effect of being in a QuesTec park is, of course, not directly observable, as it is subsumed in the pitcher-QuesTec fixed-effects terms. The results are remarkable: In ballparks with the umpire monitoring system (Column 1), the coefficient on UPM is -0.21 percentage points and is not significantly different from zero. In parks without QuesTec (Column 2) the same coefficient is 0.66 percentage points per pitch, significantly different from zero at the 1 percent level.
Columns 3 and 4 present the results when QuesTec is interacted with UPM, first without umpire-QuesTec and batter fixed effects and then with these additional controls. Controls for umpires and batters only strengthen the result, so that when the pitcher and umpire match race/ethnicity, being in a QuesTec park reduces the likelihood that a called pitch is ruled a strike by almost 1 percentage point. Columns 5 and 6 reveal that this effect is nearly twice as strong for minority as for White pitchers. As in the first three sets of results in Table 3, however, separating pitchers by race/ethnicity (or even by minority vs. non-minority) has a large impact on statistical significance, impairing our ability to infer much about the relative sizes of the effects of UPM across pitcher groups.

In one of my Baseball Prospectus columns exploring if a game fixing scandal could happen among baseball umpires, I suggest Questec be used in every ballpark to keep the umps honest:

The final piece is QuesTec. While watching a blowout the other day, the announcer was relaying a discussion he recently held with an umpire. The source told him that umpires in QuesTec stadiums could no longer move blowouts along. They would normally call close pitches against the team with the big lead to speed things up, but in QuesTec stadiums, they can't do that because it hurts their grades. If QuesTec is that effective in getting umpires to call games correctly, it should be in every ballpark. And the umpire information system should be used not only as a tool to improve umpire performance, but as another way of looking to see if the fix is in.

Seems QuesTec is a lot more useful than first thought. Finally, I like Hamermesh's overall take on the whole study:

Hamermesh, who has studied discrimination at all levels, says that bias is instilled in infancy -- much like enduring personality traits such as shyness or high self-esteem -- as an essential part of human behavior. "We all have these subconscious preferences for our own group," he says. Ever the economist, Hamermesh adds, "It's important to look at it in baseball because of the amount of money that's being made -- the salary of the umpires, baseball players and the amount of revenue that's being made by the industry. All these things make this important."
But the takeaway message of his study is a hopeful one, says Hamermesh: discrimination can be corrected. "I expect that [MLB] will not be very happy about this, but the fact that with a little bit of effort this kind of behavior can be altered, that's very gratifying. I wish with society as a whole we could reduce the impact of discrimination as easily as it could be done in baseball."

Posted by David Pinto at 10:36 AM | Umpires | TrackBack (0)

Comments

Somebdy alert Gary Sheffield!

Or is he the author?

Posted by: Mr. Furious at August 14, 2007 11:51 AM

Cool, Mr. Furious, I bet that is the first time Gary Sheffield has ever been mistaken for a professor of economics!

Posted by: James at August 14, 2007 12:11 PM

I always thought the real discrimination was against hitters of all skin shades. Those "Glavine strikes", you know.

In fact, the electronic gizmos the networks use nowadays are kind of embarrassing. That "backdoor slider on the outside corner" turns out to have missed by four inches. I bet this kind of discrimination swamps any racial stuff going on.

In fact, if the gizmos actually called balls and strikes, they would probably have to widen the plate. There's a reason Schilling attacked that QuesTec camera. Hitters would catch on real quick that the called strike zone had shrunk considerably. You might average twelve, thirteen runs a game, which would mean a lot of real long games.

Posted by: Casey Abell at August 14, 2007 12:54 PM

I imagine this mostly comes from the fact that umpires are forced to make calls they're not quite sure about all the time, and have to imagine what probably happened. In this situation, you are going to show a certain bias toward the kind of player you expect to have done better. It's the same thing that results in so many lopsided call-fests in the NBA. Race is only one of a number of things that might impact that, however subtly, but I would imagine if you measured against reputation in baseball, superstar status, and possibly a few other things, you might get similar results.

That doesn't make race unimportant here; clearly that's going to be an issue. And of course, things like reputation in baseball aren't entirely independent of race either, necessarily. But I'm glad to hear it's such a small effect.

Posted by: jw at August 14, 2007 02:31 PM

David, did the article break down the numbers by the race of umpire? I find it interesting that there is a subtle implication that "umpire=white", when in actuality the figure refers to the fact that "umpires tend to call more strikes when the pitcher is of their same race".

When watching games, that's simply not something I pay attention to, so I'd also like to know what the breakdown is between races amongst the umps (if that figure is available).

Also, Daniel Cabrera is clearly throwing off the curve on this one.

Posted by: NoPepperGames at August 14, 2007 05:02 PM

I find it interesting that the authors of the paper never actually watched the games. They relied on the ESPN.com play-by-play. So, they have no idea of knowing if the pitch was a curve in the dirt or a slider that just missed the corner of the plate? How can you make the analysis that an umpire is calling a pitch wrong if you don't know where the pitcher actually threw the ball?

Posted by: johnr at August 14, 2007 09:01 PM

My impression has always been that the umpire establishes 'his' strike zone early in the game and both teams adjust accordingly. What determines that strike zone is a combination of things, but I seriously doubt race is one of them.

The umpire will usually consider the reputation of the pitcher (good control/bad control), whether the pitcher is a veteran or a rookie, the reputation of the batter (good eye, bad eye), and the score of the game (blowout) when deciding on close pitches.

If I was going to spend this much time studying possible bias in umpires, I would rather see if they favor/disfavor certain teams or certain managers. I seem to hear a lot more of the "he's always tough on us", or "he's out to get us" than I do hints of any racial discrimination.

As for the actual finding of 1 more strike being called per game when the pitcher is of the same race, even that only matters when the opposing pitchers are of different races. Don't know what % of the time that is the case.

Posted by: George S at August 14, 2007 11:00 PM

Re the number of pitches in a game, it's not 1 percent of pitches thrown, it's 1 percent of pitches that are called strike or ball. That subset is about 53 percent of all pitches, according to the study, so it's between 1 and 2 pitches per game that's affected, not between 2 and 3.

Posted by: King Kaufman at August 15, 2007 02:11 AM

"How can you make the analysis that an umpire is calling a pitch wrong if you don't know where the pitcher actually threw the ball?"

That's not what this paper is about. Over the course of, I'm guessing, several hundred games, it's not about "a pitch". It's about tendencies that span thousands of pitches and thousands of subjective calls.

The unfortunate thing about this is that it screams out for a companion article that compares the talent level between pitchers of different races. If one "race" is demonstrably better, then of course they're going to have more strikes called for them. And if that group is a larger population that matches up with a large group of umps, then those numbers completely account for this report.

Posted by: NoPepperGames at August 15, 2007 10:47 AM