Tuesday, March 23, 2010

Fisking Howard Megdal and the Greenberg Conspiracy of '38

Sometimes smart people get ahold of statistics and do dumb things. Yesterday, at the New York Times, on Jack Marshall’s Ethics Alarm, and on NBCSports.com, Howard Megdal, author of The Baseball Talmud, explored the long-discussed theory that Hank Greenberg was pitched around in 1938, and therefore denied a shot at the home run record, because he was Jewish. In the comments of HardballTalk, Megdal writes that the central question is, “Was Hank Greenberg treated differently than others sluggers chasing Ruth’s record. And the answer, thanks to the historical record, is yes.” It’s a line Megdal repeats on the New York Times, when he writes, “the statistical record stands as evidence that Greenberg’s religion might have been an additional barrier.” Here’s the problem: the statistical record says no such thing.

Megdal comes to this conclusion because Retrosheet and BaseballReference.com have now provided access to game results and box scores for all of 1938, and indeed for most of Greenberg’s career. His central argument is that Greenberg’s walk rate was 20.4 percent in September of 1938, while Megdal’s math has him at 15.9 percent of plate appearances at the end of August. In addition, Greenberg was walked three times in a game three times during September of 1938, after having just two three-walk games over the course of the rest of the season. Just to be clear, Megdal is talking about a 4.5 percent change in walk rate over a full month and three games in particular in September as evidence of an Anti-Semitic desire to keep Greenberg from breaking Ruth’s record.

Now, a couple of points. First, Hank Greenberg undeniably faced a great deal of prejudice as he chased this mark: he received hate mail and death threats. This was definitely a serious consideration and one that may have played a role in Greenberg falling short if he was distracted or stressed in the season’s final weeks. Second, it is absolutely true that Greenberg’s walk rate in September of 1938 (20.4) is higher than his career rate (13.9).

However, there are significant problems with Megdal’s analysis. The first is that, according to The Common Man’s math, Greenberg didn’t walk in 15.9 percent of his plate appearances through the end of August in 1938. According to BaseballReference.com, Greenberg walked 91 times in 544 plate appearances from April through August. That works out to a walk rate of 16.7 percent. How this slipped past Megdal and the New York Times fact checkers is beyond The Common Man. Really, then, we’re talking about a difference of 3.7 percent between Greenberg’s non-September and September walk rates. Given the same number of plate appearances in September and October, we are talking about an actual difference of five walks between Greenberg’s September rate and the rest of his season. Five walks. That’s what Megdal’s argument comes down to. Five walks spread across 32 games. Statistical blip doesn’t even come close to describing how flimsy this data is.

What’s more, thanks to the work done by two intrepid bloggers at MLB Expert Analysis Blog back in February (which Megdal apparently never saw), it’s clear that, while Greenberg’s walk rate did rise in September, it had been even higher in April and May. Indeed, September and October represent the third highest BB% for Greenberg in 1938. Did pitchers presciently begin pitching around Greenberg in April and May because they knew he would break the mark? Of course not. Again, what you’re seeing is a statistical blip.

This article is also important since it points out that, going into September, Greenberg had 46 homers, and would need 14 to tie the Babe, 15 to break the record. Greenberg’s high for homers in a month in 1938 was 15 in July, which suggests that Greenberg was always a long shot to break the record, until he got hot again in September (and hit 12). Why is this important? Because two of the three three-walk games that Megdal identifies as so important occur on September 3 and 5, when Greenberg was still sitting on 46 homers.

Now, let’s look deeper into those three games that play such a pivotal role in Megdal’s argument. On September 3, Greenberg came to the plate 5 times against the White Sox. Jack Knott, on the mound for the Sox, was about a league average pitcher. He walked Greenberg twice in seven innings (in which he gave up 11 hits and 8 runs, but no other walks). Harry Boyles, who pitched 29 innings that year, with a 5.22 ERA and 7.7 BB/9 (8.5 career mark). Boyles also walked Hank once.

On September 5, the Tigers were playing the Browns in a double-header. In the first game, Greenberg walked three times against Lefty Mills in four plate appearances. Mills was finishing up his only season as a full-time starter, and walked 116 batters in 210 innings. He had a 5.31 ERA and a 5.0 walk rate. However, this was the lowest rate of his career, as he finished above 7.0 in every other season in which he pitched. His career BB/9 was 6.2. In addition to Greenberg (who batted cleanup), Mills walked Dixie Walker (2nd), Charlie Gehringer (3rd) and Rudy York (5th). To accept the notion that Mills was intentionally putting multiple runners on base in a tie game (the score was 2-2 going into the bottom of the 8th, when the Browns pushed across the lead run. Given that he was skating on thin ice as a major league starter, TCM has a hard time believing that Mills would jeopardize his personal ambitions to stop Greenberg from achieving a record he had little chance, at that point, in breaking. In the second game of the double-header, which the Tigers won, Greenberg went hitless in four appearances with just one walk. Aparently starter Bobo Newsom didn’t get a team-wide memo.

The final three-walk game of 1938 for Greenberg happened on September 27, in another double-header against the Browns. Again, in game one, Greenberg was passed three times in four trips, this time by the stellar team of Jim Walkup and Fred Johnson. In his other at bat, Greenberg singled and had an RBI. Walkup was aptly named. Aside from 8 games in 1939, he was mercifully on his last legs as a major leaguer. Jim Walkup was given a shot as a swingman in 1935 and posted a 6.25 ERA with 104 walks in 181 innings. In 1937, he bested that with a 7.36 ERA and 83 walks in 150 innings. He settled down some in 1938, dropping his ERA all the way to 6.80 with 53 walks in 94 innings (and a 1-12 record). Walkup’s career ERA was 6.74 and his BB/9 was 5.1. In that game, Walkup would dole out 5 walks to the Tigers. Not that Johnson was any better. At the tender age of 44, Fred Johnson got his first extended shot with the Browns in 1938. He would pitch 69 innings with a 5.61 ERA. In his one inning of work, Johnson let an unearned run score, and got hung with the loss, thanks to the two walks he allowed. Given that Charlie Gehringer (who immediately preceded Greenberg) also walked twice, it’s again hard to believe the Browns were deliberately putting multiple runners on base. In Game 2, against Bill Cox and Ed Cole, Greenberg was not walked, and in fact hit two homers in four at bats, giving him 58 for the season. The Browns lost 10-2. Honestly, if there was ever a time to walk Greenberg, surely it was in this game, where the Browns were handily getting slaughtered, had nothing to play for, and Greenberg was destroying them.

The facts are that, aside from Jack Knott, Greenberg faced terrible, even historically bad, pitching in these games. Many of these pitchers had control problems anyway. Without actual play-by-play data, we cannot know exactly the situations in which Greenberg was hitting, but it’s clear he was coming to the plate with men on base on many occasions, a dangerous time to put additional runners on. While we cannot definitively know what’s in mens’ hearts, it seems likely that these pitchers did not intentionally walk Greenberg to keep him from the record (especially as most of them were fighting to hold on to their careers), and that there was no organizational policy against him (particularly against the Browns).

The play by play data that Megdal looks at is compelling only if you pick and choose what you want to make your argument and if you (TCM assumes) incorrectly state the player in question’s walk rate through the season’s first 5 months. In actuality, all Megdal’s argument comes down to is five walks in September. Five walks in 32 games. A blip. A nothing. And for him to suggest otherwise is to ignore context and to misuse the data he has available.

Hank Greenberg did face racism in 1938, and what was said about him and sent to him was shameful. It’s an embarrassment to the country he fought to protect several years later in WWII. And that racism may have played a role in the way he famously, as he put it, “ran out of gas.” But you cannot make a credible case that American League pitchers were in on any kind of effort to deny Greenberg the title, Howard. You can’t make these statistics say whatever you want them to.

15 comments:

Jack Marshall said...

Terrific analysis. I couldn't have done it better myself. In fact, I tried, and didn't. Bravo.

Anonymous said...

Holy cow. Doggone, pesky facts getting in the way of someone's opinion. All you need now is Megdal to say "and I would have gotten away with it if it hadn't been for you meddling kids and that dog!" Also, NYT has a fact checker?

browngoat said...

Great analysis. Thank you for the persepective.

Ziesdov61 said...

I had wondered the same thing when I saw that "analysis," and did a rough count with Megdal's numbers and also came away with a difference of five walks, though I didn't bother drilling down in the excellent detail you did. Great job!

Anonymous said...

Ooouuutstadning, sir!

Mick said...

Excellent piece.

Anonymous said...

Thanks for doing what needed to be done. It was obvious to me when reading that piece in the Times that the difference in walk rates was not statistically significant. What I didn't know than was that Greenberg had just 46 HRs through Sept. 8. So one could use the same sort of superficial statistical manipulation as Megdal's to argue that, by serving up 12 HRs in his last 24 games, pitchers were trying to LET Greenberg get the record.

LargeBill said...

TCM,

As others have said, thanks for debunking this garbage. Racism and anti-semitism do exist. However, it does a disservice to all to pretend that one can look at a small sampling of statistics and ascribe ugly motives to people. Worse yet, this "journalist" is ascribing those ugly motives to people no longer able to defend themselves. Now, if he had contemporary quotes from some of the pitchers Hank faced and coupled that with statistics that validated his original suppositions then that would make for an article.

Bottom line, while I'm sure there were jerks in sports then as there are today, most players played the game straight up. A player gets 600+ PA's during a season spread out over 150+ games against more than 50 pitchers. One pitcher can't overly influence a race for a batting title or a home run record. Sure, final game of the season . . . . thing is the player could have gotten one more hit or one more homer in April or May.

Additionally, from all I've read Greenberg was well liked and respected by most of his contemporaries. At least while he was as a player as a GM it might have been another story.

Anonymous said...

"incorrectly state the player in question’s walk rate through the season’s first 5 months."

I think you have a valid point and are probably right, but you realize that his error(using a walk rate of 15.9% as opposed to 16.7%) would be worth a single walk, right? If you're going to talk about how five walks is insignificant, it's worth noting that his one walk error is also probably insignificant.

The Common Man said...

@Anonymous

TCM realizes the difference we are talking about is essentially a rounding error. But if Megdal is going to make his arguments using percentages, thereby making the difference between what Greenberg did April-August and September appear larger, then it's entirely appropriate to point out that the difference he's talking about isn't as large as he makes it seem. Also, the error in Howard's article points to his sloppy analysis in this case. Both of those seem relevant in this case.

Anonymous said...

I was hoping someone would respond to Megdal's article. The whole thing was clearly built on flimsy analysis. Excellent takedown.

RAPman said...
This comment has been removed by the author.
RAPman said...
This comment has been removed by the author.
RAPman said...

I came up with a comment that I posted to Jack Marshall's Ethics Alarm site (Randall, March 24, 2010 at 7:41 pm):
http://ethicsalarms.com/2010/03/21/unethical-questions-anti-semitism-and-greenbergs-chase/comment-page-1/#comment-2079

He suggested I post it here, too.

My comment is 300+ characters too long for your site so I couldn't post it in all its glorious (if I may say so myself) entirety here.
--

Dan said...

This is wonderful analysis. I hadn't even read Megdal's theory yet, but this was much more enjoyable.