Clemens Stats: A Cautionary Tale on Numbers

Roger Clemens and his agents have taken their fight against the Mitchell Report to the internet. On www.rogerclemensreport.com, the Rocket's people have a compiled an 18,000 word report attempting to prove that Clemens' late career success wasn't so atypical. Of course, their goal is to prove that Clemens' success wasn't due to outside factors (i.e. steroids or HGH), and simply due to Clemens learning how to pitch better.
Clemens' camp uses comparisons to pitchers such as Nolan Ryan, Randy Johnson and Curt Schilling to prove their point. But the numbers can be twisted any way you choose. The New York Times is running a piece today that provides a different slant on the same stats.
"Yet such comparisons tell an incomplete story. By comparing Clemens only to those who were successful in the second act of their careers, rather than to all pitchers who had a similarly successful first act, the report artificially minimizes the chances that Clemens’s numbers will seem unusual. Statisticians call this problem selection bias."

Instead, the Times article, written by a group of professors from Wharton's School of Business, looked at the statistics of all pitchers since 1968 who enjoyed the longevity that Clemens had. Certainly should give you a better comparison, right? Well, the professors reached a different conclusion:

"Relative to this larger comparison group, Clemens’s second act is unusual. The other pitchers in this durable group usually improve steadily early in their careers, peaking at around age 30. Then a slow decline sets in as they reach their mid-30s.

Clemens follows a far different path. The arc of Clemens’s career is upside down: his performance declines as he enters his late 20s and improves into his mid-30s and 40s."

And their conlusion:
"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."

Statistics alone cannot prove or disprove the allegations that Clemens is a cheater. But the unusual career pattern noted here certainly provides some damning circumstantial evidence against Clemens' innocence pleas.

But the real lesson learned from these two differing opinions is that, in this age of baseball analytics, statistics can be found to back up any opinion you would choose. I'll give you an example. Say I wanted to prove that Whitey Ford is the worst pitcher in postseason history (an asinine opinion to be sure.) Well, it took me all of two minutes to find that Ford has by far the most losses in World Series history - he lost 8 times. It takes further digging to find out that Ford also won 10 World Series games, made 22 WS starts, and threw 146 innings - all also are all-time highs.

Stats can be twisted to fit what we want them to. You have to take a real close look at what stats the writer is using to back up his opinion.