Selection Bias

Two weeks ago YF linked to a piece by Alan Schwarz detailing the work of Eric Walker, who posits that late career performance spikes may have nothing at all to do with PED usage. Eric took time to detail his case further at this site here and here, in addition to his extensive work on his own site.

Today the Times gives space to three Wharton faculty members who examine Clemens’ own camp’s report, a report intended to explain the common-ness of Roger’s geriatric excellence. Unlike Walker’s views of Barry Bonds’ late surge, the Wharton boys cast a skeptical eye over his accomplishments, seeing something “unusual” at work. Though they will not charge PEDs with influence (like the rest of us these Professors, advanced degrees and all, cannot know whether Roger actually used), their skepticism is reasoned and apparently impartial. Click over for some interesting reading.

4 comments… add one
  • Ironically, I found this even less convincing than Walker’s piece, especially in light of the rather strong evidence Walker has presented suggesting the limited benefits of PED use. The Wharton authors present an aggregate of a small sample group, and then compare what is extrapolated from that to the singlular experience of Clemens. The suggestion from their graf is that all pitchers’ perfomances follow a certain curve over time, and that the Clemens curve (or line) is different. But it’s ridiculous to compare the Clemens data to an aggregate; if you look at the individuals who make up the aggregate, they all deviate. The authors write:
    “these comparisons do not provide evidence of his innocence; they simply fail to provide evidence of his guilt. Our reading is that the available data on Clemens’s career strongly hint that some unusual factors may have been at play in producing his excellent late-career statistics.”
    So basically, they don’t provide any actual evidence, and at best “hint” that “unusual factors” which they can’t define “may” (or may not) “have been at play.”
    So really what is the point of this article? Why is the Times printing material that accuses someone of lying and cheating on such thin evidence. In the name of “science,” some bad journalism has slipped into the paper of record. Boo.

    YF February 10, 2008, 3:22 pm
  • I had the same reaction to that article. Even more interesting is the Times had a piece by Schwartz basically saying stats are so variable as to not be helpful when examining PED use:

    A YF February 11, 2008, 8:39 am
  • the wharton gang has indeed missed the point…they accuse roger of “selection bias”…no duh, that was the point of his report…roger is saying that there are other guys with similar career paths, that he is not so unusual after all…to do that he had to select guys most like him for the comparison [and guys not mentioned in the mitchell report]…they readily dismiss his explanation as “smoke and mirrors” that adjustments to his pitching style were the reason his career had a resurgence…they were at least bright enough not to conclude that their analysis proves anything about how roger capitalized on “unusual factors” that were apparently present late in his career…hopefully these guys aren’t as lazy and sloppy in the classroom as they were with this “research”…

    dc February 11, 2008, 9:53 am
  • YF hit the nail on the head here. Walker’s work is far, far superior to the “Wharton Three.” And he didn’t even necessarily convince me (although he came a lot closer than I expected).
    Further, the NY Timnes has really done an abysmal job of printing this kind of bush league analysis of the Clemens scandal. The worst was when they printed the musings of “body language experts” who observed Clemens on 60 Minutes and at the press conference he held (link: The article intimates that the experts believe Clemens is lying, stressed, and hiding something before going on to say, “Even the most skilled body-language experts are right in only about half of all cases.”
    It was good for a laugh, though.

    Mark (YF) February 11, 2008, 11:33 am

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Next post:

Previous post: