Dishonest coverage, or just sloppy?

"Canseco said that he injected Rafael Palmeiro with steroids. Palmeiro's performance improved the next year. That proves that Canseco was telling the truth!"

I've seen a lot of this sort of nonsense in the last week, people looking at Palmeiro's raw home run numbers and saying "AHA! This PROVES that Canseco's right!" Here's a typical example, from Skip Bayless at ESPN.com.

Remember, Canseco and Palmeiro were born in Cuba and played youth baseball together in Miami. They were reunited when Canseco was traded to Texas midway through the 1992 season. That season, Palmeiro hit 22 homers.

But from '93 through '03, Palmeiro's homers jumped to 37, 23 (in just 111 games), 39, 39, 38, 43, 47, 39, 47, 39, 47, 43 and 38. Coincidence? I think not.

Pretty damning evidence, huh?

But let's look at it a little bit differently. With some context. For which, we need a couple more pieces of information.

1. Baseball hitters typically peak at around age 27, reaching basically full physical maturity. Some do it earlier, some later, but that's the average.


2. Raw home run totals can be skewed by opportunity. We really need to look at HR/PA to determine whether there was some nefarious performance level change.


3. League performance levels change, too. 20 home runs in 1966 is not the same 20 home runs in 1986 or 1996. Different ballparks, umpires, baseballs, bats - all of these things affect the ease with which HRs are hit.

So here are Palmeiro's rate stats, compared to the leagues in which he was playing:

Rafael Palmeiro HR Rates vs. League HR Rates

Year	Age	PA	HR	Palmeiro	League	Palmeiro Rate/League HR Rate
1986	21	78	3	0.038	0.022	1.771
1987	22	244	14	0.057	0.026	2.211
1988	23	629	8	0.013	0.019	0.686
1989	24	632	8	0.013	0.02	0.635
1990	25	651	14	0.022	0.021	1.033
1991	26	714	26	0.036	0.022	1.628
1992	27	701	22	0.031	0.02	1.534
1993	28	686	37	0.054	0.024	2.277
1994	29	498	23	0.046	0.028	1.626
1995	30	624	39	0.063	0.027	2.275
1996	31	732	39	0.053	0.031	1.74
1997	32	692	38	0.055	0.028	1.945
1998	33	709	43	0.061	0.028	2.128
1999	34	674	47	0.07	0.03	2.326
2000	35	678	39	0.058	0.03	1.895
2001	36	714	47	0.066	0.029	2.286
2002	37	663	43	0.065	0.028	2.28
2003	38	654	38	0.058	0.029	2.027
2004	39	651	23	0.035	0.03	1.192

Then let's look at it graphically.



Jose Canseco joined the Texas Rangers towards the end of the 1992 season, Palmeiro's 4th season in Texas. During the previous two seasons, Palmeiro's HR rate vs. the league's had taken two large jumps. It drifted down slightly in 1992, then jumped again in 1993, the year which is presumably, according to the storyline, the first "dirty" season for Palmeiro. When we look at the chart, 1993, the age 28 season, does show a big improvement for Palmeiro. That's the increase that people are pointing at and yelling "proof!"

But is it? While there was a big jump in his rate during 1993, it was his age 28 season. It's not unusual for players to have a peak season at age 28. And it was a jump to a rate that he's already demonstrated over 244 plate appearances at age 22. And, furthermore, it was a jump that was relatively smaller than the jumps he'd made in the two seasons before Canseco arrived.

Then, in 1994, his home run rate dropped back to about where it have been in 1991, at age 26. If it was meeting Jose Canseco and getting "juiced" that was responsible for Palmeiro's 1993 performance increase, you'd expect it to have continued, or even improved in 1994. But that's not what happened. While the rates of home runs throughout the Major leagues were increasing from 1993 to 1994, Palmeiro's dropped.

Of course, he wasn't playing with Jose anymore. Maybe it just took him a year to find a supplier in Baltimore, because his numbers went up again in 1995. Then, in 1996, they fell. Then, in 1997, they rose, and in 1998 rose further, not quite back to where they were in 1987, when he was 22, several years before he played with Jose Canseco.

So, is the "spikyness" of his home run rate chart the result of taking and stopping steroids? Or is it the normal kind of random performance variation that you could expect from anyone? And is that jump in 1993 something out of the ordinary, or something that happens to anyone?

Let's look at a few other players for comparison's sake.



There are a couple of things that jump out here.

1. Palmeiro's peak home run rate, compared to the league, was not as high as any of the others.


2. The up-and-down of league-normalized home run rates is visible for all 6 hitters. Palmeiro's chart doesn't look any more suspicious than the others.


3. The 1993 theoretical "Canseco effect" isn't the biggest relative increase on Palmeiro's chart, and several of the others showed increases as big or bigger. Palmeiro's age 28 increase pales in comparison to Yastrzemski's age 29 and Schmidt's age 27, and is smaller than Aaron's age 37 season, just to pick a few.


4. Palmeiro did have more seasons close to his peak than several of the other hitters. This could be a steroid effect, and it could also be the ramifications of never having that one "great" season.

In other words, it's simple and easy for someone like a Skip Bayless to look at the raw home run numbers and say "Look - Palmeiro's home runs went up when he started playing with Canseco! That PROVES that Palmeiro's dirty!" In reality, it "proves" no such thing. Players numbers bounce up and down, randomly, throughout their careers. There's no sudden spike to a previously unheard of performance level for Palmeiro.

None of this means, of course, that he's not "guilty," that he hasn't used steroids. I didn't believe him in March, I don't believe him now. But the numbers that people are pointing to are not only not proof, they barely qualify as evidence. And it's sloppy and dishonest to trot 'em out the way that people are trotting 'em out...