Today, we’re going a level deeper. We’ll look at how ejection rates correspond to other vital statistics for managers, both collectively and as individuals. We’ll find out whose relations with the umpires were most affected by success and time on the job. And we will see how George Steinbrenner helped make for some strange bedfellows.
My first point of interest is whether being a winning manager correlates with one’s ejection rate, one way or the other. I took every manager whose full career began in 1889 or later, to avoid including anyone who managed when umpires did not have the power of ejection. I also set a minimum of 100 games managed. Both criteria will hold for this entire section.
A positive correlation between winning and ejection rate is there, barely. The coefficient of determination, R^2, is less than one half of one percent, effectively zero. Also, there are factors not measured in the graph boosting that tiny connection to the level we see.
One of these factors is that many of the lowest winning percentages, including the six worst, come from the 19th century, when ejection rates were extremely low. Weaker competitive balance in those early days made such abysmal records likelier and thus creates some false correlation. Also, winning managers tend to manage more games, meaning the top of the trendline has a stronger sample behind it than the lower-left end. That probably bends the correlation up, as well.
Perhaps a better way to examine the question is to measure ejections against career games above or below .500. This should knock out the worst of the first bias, and at least some of the second.
The positive correlation remains, but as suspected, the relationship becomes even weaker than before. Any tendency for winning managers to be oft-ejected managers is minimal, easily washed out by random noise.
A few trivia points of note on this graph. You’ll note Bill Dahlen is the highest dot, with over five percent ejections and more than 100 games below .500. Down on the baseline, the man with no ejections and over 400 games clear of .500 is Frank Selee. His counterpart to the left, over 200 games under water, is Connie Mack, pretending to be immaculate with his 0.01 percent ejection rate. Way over to the right, the high man is John McGraw, and the low man is Joe McCarthy. We’ll see them again soon.
So if winning means next to nothing to one’s ejection rate, how about sheer tenure? Do the long-timers go after the umpires harder, trusting that this won’t poison them with the front office? Or do the flameouts-to-be have a lack of control with the men in blue that foretells a lack of control of their clubs and a quick pink slip?
The answers are no and no. That trend line looks pretty solid in its upward march, but the correlation is again very weak, the R^2 just over one percent. One could think it an optical illusion, because the data points are squeezed so heavily onto the left side of the graph. A logarithmic scale for games managed should make things look clearer.
Things are far clearer, showing a far more even distribution. You can even see the curving lines, arcing from the center going left and upward, of the sets of managers with one career ejection, two, three, and so forth. (Not to mention all of the angels down at zero, plus that pretender, Mack. You don’t fool us, Connie.)
The result, though, is the same. Long-term managers do get ejected a bit more often, but the relationship is faint. This really should not surprise anybody. Managers are such a diverse group of personalities that single factors like these ought not have that much effect. If anything, the surprise is that on-field success has less of an influence than success in hanging onto one’s job on the ejection stats.
One at a Time
But if managers are such unique individuals that you cannot judge them in a group, it may make more sense to judge them against themselves in differing circumstances. Do managers get thrown out more often when their teams are doing poorly, taking out their frustrations on the umpires? Or could it be that fire in the belly brings both victory on the field and getting shown off the field?
To examine every manager in the ejection era this way would be extremely time-consuming, so I needed to take a sample. I chose criteria I thought interesting and relevant, and came up with four categories of manager to look at.
1. Managers with the 20 most games managed
This is obvious, on baseball and statistical levels. You want the broadest swath of baseball history you can get, and long-time managers give it to you. You want as much data as you can get at once, and ditto.
Managers included in this list are, from most to 20th-most: Mack, Tony LaRussa, McGraw, Bobby Cox, Bucky Harris, Joe Torre, Sparky Anderson, Gene Mauch, Casey Stengel, Leo Durocher, Walter Alston, Bill McKechnie, Lou Piniella, Jim Leyland, McCarthy, Ralph Houk, Dusty Baker, Bruce Bochy, Tommy Lasorda, and Dick Williams. (Bochy is still active; Dusty Baker could possibly fetch up with another team after his firing from Cincinnati, but the odds are against it.)
2. Managers with the 15 most ejections
Another obvious choice. You want as much of what you’re studying as you can concentrate into a manageable group.
I gave this list in the first article, but I’ll repeat it. Managers in italics are repeats from above. The top 15 are: Cox, McGraw, Durocher, Earl Weaver, LaRussa, Frankie Frisch, Paul Richards, Leyland, Ron Gardenhire, Torre, Piniella, Clark Griffith, Bochy, Charlie Manuel, Bill Rigney.
3. The 12 best, and 12 worst, managers of all time
I wanted to be able to compare and contrast the highest and lowest echelons of the managerial arts to see what, if anything, makes them different in this respect. But who is the best and worst, and who do you trust to tell you so? For the latter question, I went in house.
I took the rankings from the book, Evaluating Baseball’s Managers, written by long-time THT contributor Chris Jaffe. The book and its ratings are based on the Birnbaum Database, Phil Birnbaum’s massive quantification of managerial statistics. If you want ratings as far from pure subjective judgment as we’re likely to get for a while, this book has it.
(Yes, I am shilling rather blatantly. The book deserves it.)
Jaffe rates the managers from 1896 onward, close enough to our 1889 cutoff not to leave much out. One hiccup was that a few years have passed since the book came out, giving time for some managers to shift in the ratings.
Originally, I was going to use the top and bottom 10. To account for the chances that Cox had climbed into the top 10 with his last years in Atlanta and Clint Hurdle had clawed out of the bottom 10 with his Colorado Rockies pennant and the Pittsburgh Pirates’ success in 2013, I expanded it to 12 apiece.
I will list the managers in both groups, but in alphabetical order this time, so as not to give away too much of Jaffe’s work. Italicized names are repeats.
(Mack, as Jaffe noted, is a unique case. Measured through his second A’s dynasty, Mack has a strongly positive career mark. After the Depression sell-off, and as he aged into his seventies and eighties, he plummeted. Since he was already in the study by career length, it doesn’t affect the roster, but it deserved explanation why a five-time World Series winner was on this list. And I’ve given him enough razzing in this piece already.)
4. Author’s discretion
In studying a particularly interesting case, I added a couple of related managers to help me examine whether certain patterns held up. I want to save mentioning that case for a while, but naming the two added managers—Bill Virdon and Buck Showalter—plus the teaser I gave in the introduction, should clue you in.
I considered including the managers with the highest raw ejection rates, but did not. Since I couldn’t also include the lowest rates—there’s nothing worth measuring in a string of zeroes—it would have left the data pool skewed too heavily toward the hotheads for my comfort. Using the top 15 ejections did enough of that.
That brought me to a grand total of 45 managers to examine. I calculated their Eject+ numbers year by year, tracking them against the winning percentages they put up, as well as by the passage of time. You can do this either going by year, or by seasons managing in cases where managers have breaks in their careers. I did both, with somewhat differing results I will explain.
I must note that I took seasonal win-loss records from Baseball-Reference for ease of compilation, while ejection and other data came from Retrosheet. This makes it possible that there are discrepancies between the two in numbers of games managed.
The table of data I used for this group of 45 is large enough that it would be a black hole in the middle of this article. I have appended it to the bottom of the page.
Does Success Spoil the Skippers?
First I re-examined the relation between ejections and tenure for this group, and the result for individuals was different than for all managers.
We get a notable downward trend, about one career Eject+ point dropped for each 100 games managed. Granted, this is skewed a bit by the extraordinary case of Mack on the far lower right, but I’m not discarding 7,749 games as a fluke. Correlation is still a flimsy 0.0262, however. As this group is skewed more toward successful managers than the overall one, there’s a suggestion that managing success goes with forbearance toward umpires, but the statistical underpinnings just aren’t there.
The relationship between ejections and winning percentage gives hints of being stronger than those we saw above, but only hints. Of the 45 managers, 20 saw Eject+ go up as their teams improved (including the outlying case of Mack’s lone heave-ho), and 25 had it go down as their wins rose. The top 12 managers split 7-5 toward the negative correlation, while the bottom 12 were at 6-6, so there is no strong pattern there.
The average trendline slope for the managers is -105 Eject+ points per 1000 points of winning percentage. That scale is a bit tough to grasp, so look at it this way: for every 10 points a manager’s winning percentage rises, his Eject+ numbers fall by 1.05 points. That isn’t a strong trend, at least not in the aggregate.
The overall correlation backing it up is minuscule. Averaging the correlation coefficients—R rather than R^2 this time, since R^2 is always positive regardless whether the correlation is positive or negative—gives a mean of -0.045, which to compare with the figures above would produce an R^2 of about a fifth of one percent. That is almost literally next to nothing. Again, hopes for a useful connection are thwarted by tiny correlation numbers.
Trends for individual managers are more trustworthy, at least sometimes. Going by the correlation coefficient R rather than R^2 for now, 29 of the 45 managers have a value of at least 0.2, positive or negative, which is an accepted cutoff for significant, if low, linear relationships. Of those, 12 are positive and 17 are negative correlations. Three of the positive and five of the negative scores break 0.5 R–plus or minus, respectively–which begins to show strong correlation.
For examples of managers who got thrown out more the more they were winning, the two best are probably Jimmie Wilson and Bell. Wilson is arguably the most woeful manager the majors have ever seen. He helmed the Philadelphia Phillies for five years in the 1930s, when they were at the nadir of their sustained awfulness, then latched on to the second-division Chicago Cubs for three years and small change.
That’s the strongest correlation number, positive or negative, you’ll find in this group, but it comes with a mild caveat. The dot at .100 winning percentage and 0.0 Eject+ is his 1944 year with the Cubs, when he was fired after a 1-9 start. Take that out, and the R^2 is .5785 instead of .6959, which is still the highest in the survey. It also changes the trendline slope: instead of Eject+ rising 54.3 points for each 100 points of higher winning percentage, it would be 80.6.
Bell had the Detroit Tigers for three years in the late 1990s, two-plus years of the Colorado Rockies (with whom he posted his best record, 82-80), and nearly three years of the Kansas City Royals in the mid-2000s. His correlation isn’t as tight as Wilson’s, but the slope puts him at 91.9 Eject+ points gained for each 100 win percentage points he rose.
A third manager, Meyer, had correlation scores in this general range, but none of the other 45 managers had an R^2 above 0.2 for a positive ejection/win relationship. These three, plus Clint Hurdle as a distant fourth–all in Jaffe’s 12-worst list–suggest that it’s the bad managers who get more ferocious with umpires the more they’re winning. Then again, three of the leading R^2 scores for negative relationship also belong to bottom-12 managers, so I won’t make a claim just yet.
To see the negative correlation in action, we go back to the early 20th century and manager Clarke. He had the Pittsburgh Pirates for essentially the whole career of Honus Wagner. You don’t need Wagner to get you onto the top-12 list of history’s best managers, but it sure didn’t hurt.
Clarke had just six of 19 seasons above average in ejections. Five of them were his five losing seasons, at the start and end of his career. His trend line gives a ratio of 59.2 Eject+ points dropped for every 100 points of winning percentage gained. There are stronger negative ratios, and stronger correlation numbers than his .3073 R^2, but I will get to them later.
The overall results on ejections versus time are, finally, more robust. Going by seasons managed, 33 of the 45 managers came out less prone to ejections as time passed. Using age, it was 35 of 45. The connection by age had a slightly stronger mean r^2 correlation, and a slightly steeper mean slope, -4.5 Eject+ points per year of age against -3.9 points per season. The relation held with medians, also, so I will use the ejection-to-age connection as the more established one from here on.
The median R value among all 45 managers is a -0.356, putting R^2 at 0.126. This was for Baker, setting him as the benchmark average for getting ejected less as one ages. This is a moderately low group correlation, but after the microscopic scores earlier, it stands out. Just three managers had R values above +0.5 (two with short careers), but 15 came in under -0.5, 12 of them managing at least 1000 games. That’s a definite split in strong correlations.
Between best and worst managers, we see a little daylight. The top 12 went 2-10 in more/fewer ejections with age, while the bottom 12 went 4-8 by year, and 5-7 by season. That suggests something, especially when you consider that bad managers are likely not to manage long.
Look at members of the survey with the longest and shortest careers, and the distinction grows much sharper. For the 12 short-timers, it’s an even 6-6. For the 12 long-timers, only one grew more ejection-prone as he aged. In fact, expand the group to the 17 managers with the most games at the helm, and there remains only one whose Eject+ trended upward with age. The grumpy old man in question is Bobby Cox.
You’ll see that the coefficient of determination is fairly weak, below average in this category for our group of managers, whichever way their trends point. The high number of years and games Cox managed lets us have confidence that a trend really is there, among everything else moving him one way or another.
For the strongest positive correlations of age with ejections, Frisch and Hurdle are your standards (though Hurdle obviously has time to change that). Tenney and Taylor have steeper slopes and stronger correlation scores, but they also have much shorter careers, in the 600s compared with 2246 games for Frisch and 1645 and counting for Hurdle. There’s too much leeway for Tenney and Taylor to be due to chance to put them on top.
For really strong ejection/age correlations, you have to accentuate the negative. The most noteworthy manager in this respect is McGraw.
McGraw in his early years was something almost unique in baseball history. In his first year of managing, 1899, he personally accounted for half of the manager ejections in the entire 12-team National League. Five years later, as he was winning his first pennant with the New York Giants, he racked up seven thumbs against a total of 10 for the other 15 teams in the majors.
He was a true enfant terrible, in his early 30s in 1904 and still playing occasionally. His highest ejection totals were yet to come, breaking double digits the next year for the first of four times. Umpires must have thought he’d never leave them alone.
But he would. In seven of his last ten seasons, McGraw would go all year without being run out even once. The trend line is so steep, it punches through the floor of zero ejections two years before McGraw retires. By the coefficient of determination, almost two-thirds of this plunge is directly related to his aging.
It’s telling that longevity selects for a downward ejection trend, while short-timers are an even split. By their nature, managers with fewer games aren’t around a long time, so the aging trend doesn’t have a chance to assert itself. That gives clarity to the connection, which I have no problem raising to a general law. Managers will be less prone to being ejected as the years pass.
Before I move on to the case that made me add that mysterious fourth category of managers, I’ll give a few capsule looks at intriguing patterns I found with a few other managers. They’re given in alphabetical order, and with their career Eject+ numbers.
Walter Alston (84 Eject+): Ejections correlated positively with winning, negatively with age, both mildly so. From 1961 to 1964, Alston had Durocher on his coaching staff. This hiring was imposed on him by the front office, not his idea, and Durocher spent much of his coaching time in Los Angeles angling to get Alston’s job. Those four years produce the highest Eject+ stretch of Alston’s career. Durocher had that effect on a lot of people.
Joe McCarthy (45 Eject+): His ejections, few as there were, fell with both wins and age, both moderately. McCarthy had ejections in only eight of his 24 seasons. Not one of them happened the eight times he won the pennant (and usually the World Series) with the Yankees. His earlier stint with the Cubs had been much rockier: ejections in four of the five years.
Bill McKechnie (80 Eject+): McKechnie is a strong example of calming with age. Seven of his first 10 seasons had an Eject+ above 120, while his final 11 all were below 100. In his first managing job, he was tossed three times in 99 games with the Newark Peppers of the Federal League. You may recall from last time, three of the 10 most frequently ejected managers in history managed solely with the FL. Now there’s McKechnie. That was one rough-and-tumble league—and quite possibly their umpires weren’t up to scratch.
Paul Richards (300 Eject+): After stints in the 1950s raising the Chicago White Sox and Baltimore Orioles out of the depths–and racking up 81 ejections–Richards’ career lay fallow for 14 seasons. Bill Veeck then hired him to manage the 1976 White Sox. Richards that year had zero ejections, and zero life, in a disastrous 64-97 campaign. His previous season low was five thumbs. If there is a case for advancing age bringing both weaker on-field performance and disinterest in rhubarbs, Richards is Exhibit A.
Casey Stengel (85 Eject+): Moderate negative correlation with wins, stronger negative link with age. Casey had season Eject+ scores of 130 or more his first eight years, spent with the Brooklyn Dodgers and Boston Bees/Braves, but without a winning record. His first three years with the New York Yankees, he never got tossed once. The only two years he broke 100 in pinstripes were the two years they failed to capture the pennant.
Zack Taylor (159 eject+): Taylor managed the St. Louis Browns circa 1950. His first two-plus years were quite calm, with one ejection in 337 games. Somewhere around that point, he apparently realized he was managing the Browns and went mad. He got ejected 11 times in his next (also last) two years.
Taylor was the man who, under Veeck’s orders in 1951, sent 3-foot-7 Eddie Gaedel to bat. Five days later, he sat in a rocking chair near the dugout as a bunch of contest-entering fans managed the team for him–and won. Maybe it was working for Veeck that snapped him. Mercifully, he was fired after the season.
Billy-ball, et al
Finally, there is the case of Martin. Possessed of one of the most volcanic temperaments in the game, and half of the most infamous manager-owner combination in baseball history, he seems primed for a huge career Eject+ number, growing positively titanic in his five stints helming the New York Yankees. It doesn’t work out that way.
With 46 ejections in 2267 games during a somewhat above-average era for ejections, Martin comes out to a 156 Eject+. That’s high but not extraordinary: he’s 14th out of the 45 managers I studied. Martin is something like Piniella in this respect, and Piniella may well have emulated his ex-manager. When he got ejected, he made a memorable display of it, and the impression he left exaggerated his reputation.
What really shocked me was Martin’s line with the Yankees. He was in constant friction with Steinbrenner, aggravating an already noteworthy temper. That massive frustration had to go somewhere, and the men in blue were an obvious target.
Obvious, and wrong. In his stints with the Yankees, Martin accumulated just a 114 Eject+ against his lifetime 156. He had five seasons with an Eject+ over 200 in his 16-year career. Just two of them came in his eight years with the Yankees. One was due primarily to his time with the Texas Rangers the year before moving to the Bronx, and both were in his shortest years managing the Yankees, 56 and 68 games. (And the ejection in his 56-game season was for a mound-visitation violation, hardly highlight-reel stuff.)
It was weird to me that the pressure-cooker environment created by Steinbrenner would roll off Martin’s back this way. But it happened again, with the other great success of the Steinbrenner era, Torre.
In his first 14 years of managing, with the New York Mets, Atlanta Braves, and St. Louis Cardinals, Torre got tossed at above-average rates. His Eject+ for that stretch at 145. It was very high with the woeful Mets, plunged with the stronger Braves, then bounced back partway with the up-and-down Cardinals. Going by his whole career to that point, one could have expected pretty strong numbers in pinstripes.
Instead, none of his dozen seasons with the Yankees registered above an 87 Eject+. Every year of his Yankee tenure was below his career average of 94, which almost defies statistics. He wasn’t dealing with the Steinbrenner of old, but it was still Steinbrenner, and Torre had Martin’s knack of not carrying any frustrations with him into his jawbone sessions with the umpires.
I wondered whether this counter-intuitive result carried over to other managers under Steinbrenner. I looked at four other members of George’s managing merry-go-round.
Ralph Houk had a career 116 Eject+. He managed for the first year of Steinbrenner’s ownership in 1973, posted a 104 Eject+, then resigned. Over his career, Houk’s ejection rate tended to rise as his winning percentage fell, and ’73 was a sub-.500 year for his Yankees. By that measure, he ought to have been higher than 104. Of course, he had the emotional escape hatch of resigning, something no other manager did under The Boss. (Okay, it’s debatable whether Martin’s first departure—”One’s a born liar and the other’s convicted.”—was a resignation or a sacking.)
Virdon took over for Houk and went the other way. He managed four teams in his career, and his ejection rate was highest with New York. Lifetime, he had a moderate tendency to be ejected less as he won more. His Yankees tenure bucked this trend. Virdon went .534 managing them versus .519 for his career, but he got tossed more than anywhere else. He’s the first case we’ve seen of ejection rates rising with a connection to Steinbrenner.
Piniella had a little under three years managing the Yankees, in two separate stints. This gives him the fourth-best longevity under George. All three of his seasons there beat his lifetime average, the final one spiking to a career-high 317. Even adjusting for a strong mellowing tendency as he aged, life under Steinbrenner was doing Pinella’s ejection rate no favors.
Showalter hauled the Yankees out of their worst funk since their mid-’60s collapse, but he was fired four years into his term for losing a playoff series to the Seattle Mariners (which, given Seattle’s history, sorta made sense). Buck posted an 80 Eject+ for New York against a 64 career mark. He has a slight tendency to get thrown out more when doing well, as he did with the Yankees, but that cancels with a slight tendency toward more ejections as he ages, and New York was his first gig. His higher rate under George stands.
So three of four other notable managers under Steinbrenner got ejected more than we would expect while working for him, and the fourth is both the smallest sample and the man who walked away. So the trend of giving umpires less grief while working under The Boss begins and ends with Martin and Torre. Is there a common cause, or are we seeing a big coincidence?
The parallel doesn’t come from aging patterns. Torre had fewer ejections as he aged, a trend that held in his Dodgers years, but he was about a third of the way down the list on that score, not really noteworthy. Martin had just a slight gentling pattern, well below the median on slope.
The aging pattern for Torre may be partly a mirage. His earliest years with the Mets were clearly his worst in winning percentage, while the Yankee years were a glory run by almost any standards. His record improves with age, producing potential for confounding correlation if there’s a strong winning/ejection link.
And is there ever.
The slope of that trendline is -747, meaning his Eject+ goes down about 75 points for every 100 points of winning percentage Torre gains. That’s the fifth-strongest negative age-to-ejections slope in the survey, but three managers ahead of him were bottom-12s, working about four seasons each, whose lines easily can be ascribed to chance. Torre’s can not, with its coefficient of determination over 0.51. Torre’s link between baseball success and leaving the umpires alone is the second-strongest of the 45, quite possibly the second-strongest ever.
And guess who’s number one?
That’s the strongest negative slope (at -972) of any long career, with the strongest correlation of anybody in that direction. When Martin was losing, he was a burden on the men in blue. When he was winning, he could actually be reasonable. It is no coincidence that his only two season Eject+ scores below 100 came in the years he posted his best winning percentages, 1977 and 1985. Both, of course, were with the Yankees.
In our recollections, there are few managers more different. The live-wire Martin, cap turned back to get that much more in an umpire’s face, and the almost laid-back Torre, sitting still in the dugout with that stone-faced squint that inspired Charles Schulz’s last great “Peanuts” strip.
Yet they are bound, not just by Yankees championships, but by the shared temperament running below the surface: the drive to win, and the proneness to lash out when times were rough. Martin let you see it; Torre made you guess it. They both had it, and it showed through in their relations with the umpires, arguably more than any other managers in history. In that similarity may be the secret of how both men were able to produce winning baseball, time and again, in that singular atmosphere.
Third in the order
In my next and final installment (unless I get some outstanding suggestions for further work in the comments), I’ll go beyond careers and seasons and dig into the games themselves. How badly is the game usually going when managers gets themselves tossed? And does their ejection have a “firing-up” effect on their teams, spurring them on to victory more often that we would expect?
References and Resources
- As usual, Baseball-Reference and Retrosheet were invaluable resources. Inspiration and support also came from the aforementioned Evaluating Baseball’s Managers by Chris Jaffe, as well as The Bill James Guide to Baseball Managers.
- Belated acknowledgment goes to Paul Golba, whose Excel-fu made things much smoother for me in this series. Gotta be a second baseman, Paul: you’re that good on the pivot.
- Finally, as promised, here is the table for the 45 managers I investigated. R values are given where often I used R^2: to get the latter, obviously, square R. Units for the slope of ejection/winning relations are one Eject+ point per 1000 points of winning percentage; for ejection/years or ejection/seasons, it’s one Eject+ point per year/season.