A few years ago, I did an extended dive into the statistics of manager ejections in the major leagues. One of the strongest correlations I found was that, as managers age, they tend to be thrown out of games less often. Long-serving managers almost always suffered fewer ejections as their careers proceeded, with the historically volatile Bobby Cox being the highest-profile exception.
It takes two people to make an ejection, and I came to wonder about the other side of the equation. Umpires’ personalities can contribute to the confrontations that lead to managers, or others, being banished from a game. Certain umpires work on a hair-trigger; others strive to still troubled waters.
Do these personality traits evolve with age and experience? Does getting a couple thousand games under your belt make you better able to defuse a clash, or make you obstinate and quicker to flex your ultimate authority? Does advancing age dim your argumentative fires, the way it does with managers, or from the opposite side of the confrontation does it have the opposite effect?
I took an extensive look at ejections over the last five seasons, 2012 to 2016, to find some answers. Along the way, I also found a couple extra questions.
I combined two sources for collecting data on umpire ejections. The first was Retrosheet, for easy ordering of umpires, base assignments, and so forth. The other was a more obscure site, though one I’ve used before, called the Umpire Ejection Fantasy League, which is pretty much what the name implies. It tracks ejections throughout the season, and made for an excellent check on Retrosheet’s data. Where the two sources disagreed, I made deeper searches into the games in question to settle which one was more likely correct.
A discovery I made quite early in my data gathering, one that will be little surprise to any serious observer of baseball, is that the base the umpire is covering has a huge effect on the chances he’ll be tossing somebody. Home plate umpires make the great majority of ejections, including thumbs for balls and strikes and for pitchers throwing at batters. First base comes in a distant second place, with third and second bases bringing up the rear.
Umpiring assignments are rotated to generally give umpires equal time at each base. There are exceptions to this statistical rule: most just random fluctuations, but some with margins wide enough to make one suspect intent. For example, in 2016 relatively new umpire Gabe Morales had 37 games at first base, but only 29 at home plate. Veteran Laz Diaz worked 30 games behind the plate, but 25 at first base.
(And no, they weren’t on the same crew, Diaz taking Morales’ turns at home. In fact, the concept of a four-umpire crew working steadily together throughout the season has weakened greatly in recent years. We can probably credit, or blame, the expansion of instant replay in 2014 for this turn. I shall return to this theme later.)
These differences in assignments have the potential to skew the results, especially if less-experienced umpires are being put more at some bases and less at others. I therefore weighted games at various bases by how often in a season ejections were made by umpires working those bases. The resulting Adjusted Eject+ metric is how I will be judging the quickness of the umps’ thumbs.
Also, I kept postseason umpiring separate from the regular season. Umpires with differing experience levels don’t have equal access to postseason umpiring opportunities (which is not a bad thing, really). I looked at ejection rates both with and without the postseason, but almost all the results I give will be without it.
My previous work was only about managers being ejected, but in this piece I am counting everyone the umpires eject who is part of the teams on the field. This encompasses players, managers and coaches, while leaving out the rare cases where an umpire might expel a fan or ball-girl or other such supernumerary. (Primarily because my sources did not track such instances.)
One borderline case occurred in last year’s playoffs, when Tom Hallion ejected Alex Wood of the Dodgers. The hitch is, Wood was not on the Dodgers’ roster for that postseason round, though he was in the dugout and in uniform. Retrosheet does not list the ejection, while the UEFL did. I decided to leave Wood’s ejection out of my spreadsheets.
While doing this spade-work, I was reminded of a prediction I made in my old ejection series. The new replay challenge system was coming into use just as the final installment reached the homepage, and I took a crack at predicting how this would affect the frequency of managers being thrown out. After the season, I wrote a new piece examining why my prediction was so wrong.
To summarize: I foresaw manager ejections dropping somewhere around 30 percent as expanded replay arrived and replaced futile jawboning with an appeal to the New York video room. Instead, they rose about six percent in 2014, as managers argued the results of the reviews with on-field umpires who hadn’t even made them.
This trend continued into the following season, manager ejections rising again. Then in 2016, the plunge I had foreseen finally happened, two years past the prophesied date.
Ejections for managers and non-managers alike jumped in the first year of replay challenges. (I didn’t observe the latter at the time.) Non-manager thumbs have held roughly at the same level in the two following years, and did not mirror the drop for managers.
What brought on the ejection bump, and the later drop for managers? A plausible answer to the first question is the very phenomenon I thought would cause a decrease: the replay challenge system. More precisely, a change in umpiring that came along with that system.
Umpires are not assigned permanently to replay duties, but are cycled through as the season progresses. This applies also to the postseason. For the League Championship Series and World Series, there are now seven-umpire crews. One ump starts the series at the replay center, then goes onto the field, taking the place of a crewmate who now handles the replays.
Naturally, this necessitates more umpires being on the payroll, to have people handling replay duties. MLB added six umpires to its roster for this purpose. At the same time, a liberalization of vacation policy led to more frequent call-ups of Triple-A umpires as fill-ins. In 2014, there were 13 umpires who got their first taste of major league work. In the rest of the 2012-2016 range, the highest number of debuting umpires was four.
This meant there was a sudden fall in the average major league experience of umpires. Given the circumstances, this isn’t surprising, but what may be a surprise is the quick rebound from that drop.
|Avg. games umped||1760||1777||1570||1670||1760|
Average games computed at end of given season.
It’s a quirk of the numbers that the mean of games umpired is exactly the same in 2016 as in 2012, but an illustrative one. The experience level took a sharp temporary hit, but built back up to pre-replay levels in two years.
Part of the rebound came because MLB was willing to drop some new umpires. Five of the 13 new umps in 2014 did not call a game in 2016. The washouts were exactly matched by the five new umpires who got their shot in the majors those two years, while all the other umpires kept gaining experience and re-balancing the books. My hat’s off to MLB if this was part of the plan, but intended or not, it worked out nicely.
Not that one could prove this by the ejection numbers. Non-manager ejections have remained elevated even as umpire experience returned to pre-replay levels. The manager ejections took such a sharp drop from 2015 to 2016 that it’s difficult to claim umpire experience as a primary cause.
Were I to offer a guess—and given my track record with predictions, this is pretty hazardous terrain, so add salt to taste—I’d suggest that the front offices of MLB sent managers some kind of advisory before the 2016 season. Changes in replay procedure, involving things like the take-out slide, the “neighborhood play,” and how fast they had to notify umpires of a challenge, made it a good time for such a circular. If it included a strengthened warning about not arguing a replay call once it is made, and if managers heeded this call, that could explain the quick 25 percent drop.
Aside from that speculation, we’re left with some evidence that umpires gave out more overall ejections when a large batch of rookies joined their ranks, which receded somewhat once the newcomers got a couple seasons under their belts. This definitely hints that umpires’ age or experience affects how often they make ejections.
As the Calendar Turns
This brings me, finally, to my original aim of measuring ejection rates against umpiring experience. I produced charts both for games umpired (by end of season) and for age (on June 30, just like for players) as the measure of experience, since I was open to a change coming either from having handled more games or just getting older. I made ejections into a rate stat, Eject+, with 100 being league average and higher numbers meaning more thumbs, and with the aforementioned adjustments for base assignments.
I ran the numbers both for regular-season work and for all games including the postseason. The results were the same in any combination. Experience was irrelevant.
I would show the scatter-plot charts here, except they don’t really matter. The coefficients of determination (or r-squared) are microscopic in both cases, 0.0008 going by games, 0.0033 going by age. That was for regular season work: the r^2 slid downward when playoff games were included, if you can believe there was room for the numbers to go down. The numbers show no meaningful correlation.
We have conflicting signals. The anecdotes suggest more ejections with less experience, but the analysis shows no meaningful correlation. However, scatter plots may have been the wrong way to analyze the numbers.
I tried another method: separating umpires’ seasons into buckets for games and for ages, and seeing whether ejection rates show a clearer trend that way. I used increments of 500 career games and five years of age. Also, I weighted the umpires’ seasons based on how many games they worked that year, which I couldn’t do for the scatter plots.
We see more of a pattern here, especially with age. Umpires 35 and under were ejecting more people, then the rates fell steadily below average for the next quarter-century. There’s a spike with the over-60 umps, but this is the smallest cohort, counting 1,498 games umpired compared to the largest bucket, the 46-50 group at 10,994 games*. A single umpire could have a sizable effect on the oldest bucket, skewing its results. (Yes, Joe West, I am looking in your general direction.)
The pattern is much less clear using games worked. The general trend is downward, but a few spikes break it up. The sharpest spikes, at 1,001-1,500 and 3,001-3,500 games, do come from smaller buckets, each fewer than 3,000 games umpired against more than 11,000 for the largest bucket (2001-2500). The 4,001-plus bucket, another spiker, is the smallest of all, easily swayed by a single person. (Yes, I’m looking due West again.)
* Bucket sizes are given in an addendum at the end of the article.
I would not swear to there being a correlation between ejection rate and experience by games umpired, but when it comes to age and ejection rate, I am more confident. Umpires have this in common with the managers they argue with: as they age, they are less likely to end such a confrontation with an ejection.
The reason for this seems apparent. Aging brings a lowering of the hormonal fires ignited in the human body. Not only are we less likely to begin a confrontation, we’re less likely to feed a flame that someone else sparks. This could also come from the accumulated wisdom of the years, but it amounts to the same thing.
It isn’t as clear whether it’s the accumulation of baseball experience, rather than overall life experience, that has this effect. The relationship between games umpired and ejection rates is the wobblier of the two examined. Given the strong but not linear correlation between age and games umpired, I would say that most of the shaky relationship we see derives from the linked age effects.
Whys and Wherefores
I think the effects on umpire ejection rates aren’t as strong or clear as for manager ejection rates because of the natures of the respective jobs. If a manager wants to get himself ejected—and there are times when this is his intention, or at least a preferred result to having a player thrown out—he will find a way. If an umpire is somehow itching to chuck someone, he’s hamstrung by the need for that someone to give him the opportunity. Also, an umpire’s boss is much likelier to look askance at him ejecting people on a whim than a manager’s boss is for him getting himself tossed for transient causes.
This carries over to how players act, and umpires react. Take the notorious José Bautista–Rougned Odor game from May 15, 2016. Four players, three coaches, and one manager were ejected in that game, only half in the main brawl itself. There was little that Dan Iassogna (three ejections) and Dale Scott (five) could do to change that number: the teams were just spoiling to fight.
It’s a similar case when a beanball battle breaks out. A great example is the June 11, 2013 D-Backs/Dodgers game (involving the plunking of Zack Greinke, if you recall). Between automatic ejections for headhunting plus a brawl that then broke out, Clint Fagan had to eject six people, a heavy season total for most umpires that he compiled within minutes.
This has been a long way of saying that ejections by umpires are much more prone to random variations than ejections of managers (or of players, for that matter). The correlations will not, cannot, be as strong, and one may thus add a bit of weight to the scales when analyzing them. Realizing and accepting this unavoidable fuzziness makes me more sanguine in my conclusion that age does dampen an umpire’s likelihood of issuing ejections.
So despite what one might sometimes hear on the other side of an on-field rhubarb, umpires are as human as the rest of us. At least they’re as human as managers—regardless of which side would be more mortified to hear that.
|Career Games||Gms. in Season||Career Games||Gms. in Season|
|Ages||Gms. in Season||Ages||Gms. in Season|