But Walker is not a lone voice out in the Washington wilderness; many credible statistical analysts are similarly skeptical about how much steroids and other drugs may have distorted modern ballplayers’ records. Regarding Bonds, for example, they note that, yes, his peak home run rates came at 36 through 39 years old, when most players are in decline. Then again, another slugger three decades before enjoyed almost the same late-30s surge: a fellow named Hank Aaron.

This old chestnut again. Aaron at the age of 37 beat his previous best slugging percentage (at age 25) by .033, having moved from County Stadium (park factor of 95 in that year) to the Launching Pad, which had a park factor of 108 in his best year there. Bonds's best SP (at age 36) jumped by .186 over anything he posted at a younger age. Using Walker's own pet tool, the Power Factor (TB / H), Bonds's best previous 6-year stretch was from 1993 to 1998 (age 28-33), when it was 2.01. He then for the next six years (age 34-39) took that number up to 2.55, a leap of over 25%. The park factor can only go so far in explaining that jump away, and from 34 to 36 was still in Candlestick anyway.

That said, it's too bad that Schwarz drags this argument into the mix, because the Walker website is far and away the best case I've seen written on the "steroid skeptics" side. There are several questions I'd ask that he doesn't go into, but the most interesting counterpoint he raises is the question of the effect of steroids on upper body strength (substantial), but along with that the effect of upper body strength on power (not much at all). I think he minimizes (or doesn't have enough reverse skepticism about) Bonds's late career power spikes, but I'm going to have to read that part through again before I'd say that for sure. What I wrote above was in reaction to Schwarz's unnamed other "credible statistical analysts."

All in all it's an impressive piece of work, and I'll be interested to see if and how the thread on it develops. But again, it's Walker's website (and its own sublinks to his main page) that's interesting and well reasoned, not so much the Schwarz piece by itself. I doubt it's the final word on the subject, but it certainly can't be ignored.

There are several questions I'd ask that he doesn't go into, but the most interesting counterpoint he raises is the question of the effect of steroids on upper body strength (substantial)

This one is a big sticking point for me. It seems to me that if you were doing steroids, whichever muscles or muscle groups you exercised would be effected. So if you did steroids and nothing but, say, neck bridges, you would wind up with a bull neck but could still have arms like pipe cleaners.

"This old chestnut again."

Actually, quite a number of men, many well known, reached their career peak in home runs in their late 30s. Here is a list of at least some of those who peaked at age 37 (or, in one case, age 38): Hank Sauer; Hank Aaron; Rico Carty; Carlton Fisk; Gary Gaetti; Chili Davis; Edgar Martinez; Rafael Palmeiro; and Barry Bonds. Only the last two are under any cloud of suspicion. And there are many more men who, though they didn't peak in their later 30s, were still performing quite well then--among them, just from the all-time top-10 HR hitters, Aaron and Frank Robinson (30 HR, above his career average, at age 37). Or perhaps someone wants to tell me (much less Robby) that his spike of 49 HRs at age 30 was "artificial" in some way? But if he did it today, lots would.

In general, player power performances really only signify when compared to the then-current average for at least their league, if not all of MLB. I have a set of graphs of the top-10s' career relative PFs, and hope to be posting a page with them and some commentary/analysis tonight or tomorrow (if I can find time from answering the many blog responses now cropping up).

"It seems to me that if you were doing steroids, whichever muscles or muscle groups you exercised would be effected. [sic]"

No. Different muscle beds have different numbers of, and different ratios of, androgen receptors. Here (as presented on the site, and linked there), from Recent Progress in Hormone Research 57:411-434 (2002):

Androgen receptors are present in skeletal muscle of every mammalian species (Sar et al., 1990; Takeda et al., 1990). Levels of expression differ from muscle bed to muscle bed in a manner consistent with reported AAS effects on muscle strength in different tasks. For example, human muscle beds differ from each other, with expression higher in the muscles of the neck and chest girdle, in comparison to the limbs (Kadi et al., 2000).

That means that it's just as the site says: steroids affect musculature differentially, no matter the degree of exercising involved.

This old chestnut again."

Actually, quite a number of men, many well known, reached their career peak in home runs in their late 30s. Here is a list of at least some of those who peaked at age 37 (or, in one case, age 38): Hank Sauer; Hank Aaron; Rico Carty; Carlton Fisk; Gary Gaetti; Chili Davis; Edgar Martinez; Rafael Palmeiro; and Barry Bonds. Only the last two are under any cloud of suspicion. And there are many more men who, though they didn't peak in their later 30s, were still performing quite well then--among them, just from the all-time top-10 HR hitters, Aaron and Frank Robinson (30 HR, above his career average, at age 37). Or perhaps someone wants to tell me (much less Robby) that his spike of 49 HRs at age 30 was "artificial" in some way? But if he did it today, lots would.

Well and good, Eric. But if you read what I wrote after that one brief sentence you can see that you've replied with a non sequitur.

Because I thought you were saying that it was the Power Factor that was important, not the home run total. And as I've shown, using this number (and also slugging average), the jump in that Power Factor in the Indian Summer of Bonds's career, compared to what he'd already done earlier in his career, far surpassed the jump in Aaron's, especially allowing for the park factor. If that's not true for those others you named, let me know.

I should add that due to the late night effect I used SA for Aaron rather than a strict Power Factor, so I've gone back and looked at that for his earlier peak and his later peak.

In 1959 Aaron had a Power Factor of 1.79. In 1971 that Power Factor was 2.04.

But in 1959 the park factor of County Stadium was 95. In 1971 Fulton's was 108.

Tell me with a straight face that these numbers are in any way comparable to what we have for Barry Bonds. Tell me that what Aaron did in 1971 was seriously analogous to what Bonds did in 2001, using your own favorite measurement, the Power Factor.

IOW if you want to dismiss Bonds's 73 home runs on the grounds that the raw number of home runs isn't important (and in your breakdown of that year you make a very good argument for doing so, the best I've read), then I don't see why you're bringing up Aaron's or Frank Robby's raw home run total to reinforce a point that you're saying is irrelevant in the first place. Give us a comparison of the late career Power Factors (or slugging averages) of all of these players and compare them to their own earlier Power Factors (or slugging averages), as I just did for Bonds and Aaron. At that point you'd be more internally consistent with your own argument. When you talk about spikes, you have to use a player's own (earlier) career as a base. Do that for Bonds and for Aaron, and then do it for those other players you named, use the Power Factor, and then tell us what you come up with.

As I wrote above, your overall methodology and argument is incredibly impressive, so I'm not sure why you're evading this one particular issue by introducing numbers that you seem to have said are irrelevant when used by us poor laypeople.

But Walker is not a lone voice out in the Washington wilderness; many credible statistical analysts are similarly skeptical about how much steroids and other drugs may have distorted modern ballplayers’ records.

And by credible, I guess he means people who don't understand the value of having reliable statistics before one analyzes them.

While this will no doubt escalate into a flame war, I'd like to say I'm glad to see another RSBBer and pre-Moneyball stathead pop up here!

“I’m tired of people saying, ‘This is what happened because I see more home runs,’ ” Walker said. “If you disagree with me, deconstruct the argument; tell me where it’s wrong. If you can, more power to you.”

OK Eric, thanks for the invite. This is how I'll deconstruct your argument.

First of all, we don't really have any idea of how extensive the use is, what the regimens are and how the balance of pitchers and hitters shakes out. There have been historically great seasons in the past 15 years from both the hitter and the pitcher side.

Before you can do a thorough statistical analysis, you have to have reliable data. As of now, we do not. So trying to measure the impact of PEDs from a statistical standpoint and apply a numerical value to it is a fool's errand.

To do what you are proposing to do from a statistical standpoint, you would have to run a double-blind, placebo-controlled clinical study. Good luck slipping that one by an FDA review panel.

While this will no doubt escalate into a flame war,

Be it noted that this is the last thing I want. Asking someone to use a consistent frame of refrence---in this case the very metric that he says is the most relevant one---when making comparisons between players, is not flaming. What I don't understand is why Walker allows himself to get sidetracked on an unsupportble statistical point when he gives a far more complex, and more plausible explanation for Bonds's 2001 season on his own website. I'll leave it at that.

Before the flame war starts, does anybody here have a copy of The Sinister First Baseman? Is there a synopsis of any sort in another book or on a website? I've read good things about that book, but I can't afford the $100 it would cost me on Amazon.

Dan, that book was published by a niche publisher and had a very small print run. It's like a million other semi-cult books, in that the current low supply drives the price through the roof. That said, if a regular publisher ever tried to reprint it, they'd quickly find a $3.98 remainder on their hands, because the demand is intense but extremely narrow. A reprint would have to be done by either a private niche publisher like McFarland, or a university press with experience in baseball reprints, like Nebraska or Southern Illinois.

To give you an idea of its relative scarcity, in my years as a used bookseller with a semi-specialty in baseball books, I handled a couple of dozen first editions of Spalding's America's National Game, at least 30 or 40 copies of Shirley Povich's The Washington Senators, and yet I doubt if I sold more than two or three copies of the Walker book. Unfortunately it just never got out there in the first place.

Or perhaps someone wants to tell me (much less Robby) that his spike of 49 HRs at age 30 was "artificial" in some way?

Being traded to the lesser league is probably responsible for much of that spike.

i've read sinister first baseman at some point, i can't remember when.

i'd suggest seeing if your campus library subscribes to worldcat (.org), and get it via interlibrary loan.

the closest one i see to utah is la public library and south dakota.


patience and no mad rush to judgment. i think that's what walker advocates. that's what i do, at least.

i think it will unfortunately take many deaths to convince people to stop taking steroids. len bias didn't stop cocaine use, lyle alzado didn't stop steroids...

I didn't RTFA, but does he argue that steroids aren't dangerous, kevin, or that they don't cause a large change in performance?

The latter, Lassus.

Or, at least, he's saying there's little evidence in the statistical record to suggest steroids have any discernible effect on performance.

In warmer weather, the ball carries further. Where are all the global warming hysterics when MLB needs them?

Or, at least, he's saying there's little evidence in the statistical record to suggest steroids have any discernible effect on performance.

And he's right about that.

Dan, that book was published by a niche publisher and had a very small print run. It's like a million other semi-cult books, in that the current low supply drives the price through the roof. That said, if a regular publisher ever tried to reprint it, they'd quickly find a $3.98 remainder on their hands, because the demand is intense but extremely narrow. A reprint would have to be done by either a private niche publisher like McFarland, or a university press with experience in baseball reprints, like Nebraska or Southern Illinois.

To give you an idea of its relative scarcity, in my years as a used bookseller with a semi-specialty in baseball books, I handled a couple of dozen first editions of Spalding's America's National Game, at least 30 or 40 copies of Shirley Povich's The Washington Senators, and yet I doubt if I sold more than two or three copies of the Walker book. Unfortunately it just never got out there in the first place.

This is a really interesting post. Isn't this the kind of book for which Lulu.com was invented?

Because I thought you were saying that it was the Power Factor that was important, not the home run total. And as I've shown, using this number (and also slugging average), the jump in that Power Factor in the Indian Summer of Bonds's career, compared to what he'd already done earlier in his career, far surpassed the jump in Aaron's, especially allowing for the park factor. If that's not true for those others you named, let me know.

We need to distinguish the various sorts of things being attributed to PEDs. One of them is "spikes", of which Bonds's record season is an example. But two things: one, the logic of connecting a spike--a huge jump followed immediately by a return to established levels--does not comport with any theory of PED use by Bonds, who is accused of having started hot and heavy over the winter of 1998 and to have continued long after. Two, spikes happen; if DeVany's paper is too complex, just consider these names: Davy Johnson, Luis Gonzalez, Terry Steinbach, Lonnie Smith, and of course Roger Maris. Whom among them do we accuse of pumping PEDs, much less pumping them for a single season?

If we look, as I have, at the career power values for the top-10 all-time career home-run hitters, we find a few kinds of graph shapes--but those shapes are distributed indiscriminately among both newer- and older-era players. In a very broad generality, a player's power typically starts at some level, goes up with increasing age to a peak, then declines--nothing controversial there. What is interesting is the age at which different players peak: in some, from Ruth to Killebrew, it is remarkably early; in some, it is a reasonable middle 30s; and in a few, it's the late 30s. In fact, the two men with the oldest peaks are Aaron and Frank Robinson. (Robby was something of an anomaly, in that given annual zigs and zags, his power was relatively constant just about till his retirement.) Bonds actually peaked at 34, and that would be more obvious except for the spike--but, as we know, spikes just happen (Steinbach and Smith at least had theirs in the mid-30s, maybe others: I am getting too woozy at 2 am to remember).

I have been spending so much time answering emails and posted comments around the internet that I haven't had a chance to complete my polishing of the individual-case-studies page I will soon (I hope) be adding to the site, but it supports the things i have just said (I had no idea what I'd find when I looked, but there it is), and will have all the career graphs on display.

To do what you are proposing to do from a statistical standpoint, you would have to run a double-blind, placebo-controlled clinical study.

I'll be hanged if I can see why. What we are looking for is effects--the question of causes comes after, if we in fact first find some effects. As I put it on the site, if you assert that there's a great big bear pwowling the winter woods, you're going to have to show me some paw prints. I looked, with diligence, for paw prints and there are none to be seen. Several other eminently respectable analysts have looked and also seen only virgin snow. Moreover, there are sound bases in both medicine and physics to understand how and why we would be unlikely to see paw prints, so it is not some impossible conundrum.

And so, as Sam'l Pepys famously said, to bed.

I'll be hanged if I can see why.

You a DC construction worker too, Eric?

How the #### are you going to determine what the effects are if you haven't even figured out who the users are and who aren't?

As I put it on the site, if you assert that there's a great big bear pwowling the winter woods, you're going to have to show me some paw prints.

You looking for paw prints, Eric? How does 73 homeruns, a .609 OBP and an .863 SLG% sound? You have enough plaster of Paris with you to cast those prints?

I looked, with diligence, for paw prints and there are none to be seen.

Well, try again. And this time, open your eyes.

Several other eminently respectable analysts have looked and also seen only virgin snow.

Bull. Name them. And by analysts, I don't mean somebody dabbling at home in statistics. I mean somebody who is a professional and who has reliable data.

Moreover, there are sound bases in both medicine and physics to understand how and why we would be unlikely to see paw prints, so it is not some impossible conundrum.

Exactly. And the sound medical way of determining statistical effects is to do a double-blind, placebo controlled clinical study. That's the only scientifically acceptable way of doing it. Until you have control of the data, (and by control, I mean knowing what the expectations are of players before and after loading up, and controlling the administration of the drug regimens so there is some basis to make some pharmacokinetic corrolations to performance), there is no ####### possible way to make statistical calculations.

Really, Eric. This is pretty basic ####. Before you do any statistical analysis, you need reliable and transparent data. Right now, you have enither. Anybody who is involved in drug studies knows you have to run controls and know which subjects are the placebos and which are the positives. Your reply, as well as your original article, betrays a naivete in this area.

But two things: one, the logic of connecting a spike--a huge jump followed immediately by a return to established levels--does not comport with any theory of PED use by Bonds

??????????

You did refer to B-A-R-R-Y Bonds record, didn't you, Eric. You sure you didn't do Bobby's stats by mistake?

You don't see huge spike in his numbers? Ans why do you insist on an immediate decline? If he continues to juice, then you would expect the performace spike to continue as well. Your logic is bizarre.

I have been spending so much time answering emails and posted comments around the internet that I haven't had a chance to complete my polishing of the individual-case-studies page I will soon (I hope) be adding to the site, but it supports the things i have just said (I had no idea what I'd find when I looked, but there it is), and will have all the career graphs on display.

Better get another can of Brasso, Eric. You're gonna need it.

Eric, let me give you a tip, before you waste a whole lot of both your and our time. Before you do any statistical analysis, get reliable data. You don't have good data right now. You don't. Period.

Bonds actually peaked at 34, and that would be more obvious except for the spike--but, as we know, spikes just happen

Peaked at what? Power? I don't see it. At age 34, Bonds slugged .617 with an IsoP of .355 (both of which exceeded his career totals to that point). From 2002 - 2007 (so, AFTER his 2001 "spike"), Bonds slugged .698 with an IsoP of .373. How is the former of those a peak?

Eric,

First, I should acknowledge an arithmetic mistake I made in my first post. Bonds's overall Power Factor for the 1999-2004 period was 2.30, and not the 2.55 which I had stated above. That's still quite a jump over his previous six years overall 2.01, especially considering the age factor.

Because I thought you were saying that it was the Power Factor that was important, not the home run total. And as I've shown, using this number (and also slugging average), the jump in that Power Factor in the Indian Summer of Bonds's career, compared to what he'd already done earlier in his career, far surpassed the jump in Aaron's, especially allowing for the park factor. If that's not true for those others you named, let me know.

We need to distinguish the various sorts of things being attributed to PEDs. One of them is "spikes", of which Bonds's record season is an example. But two things: one, the logic of connecting a spike--a huge jump followed immediately by a return to established levels--does not comport with any theory of PED use by Bonds, who is accused of having started hot and heavy over the winter of 1998 and to have continued long after. Two, spikes happen; if DeVany's paper is too complex, just consider these names: Davy Johnson, Luis Gonzalez, Terry Steinbach, Lonnie Smith, and of course Roger Maris. Whom among them do we accuse of pumping PEDs, much less pumping them for a single season?

But if Bonds's PF numbers remained at historically high rates, compared to the performance base he had established for himself at a younger age, then why would it be exculpatory that his numbers didn't drop?

Just for the record, here are Bonds's Power Factor numbers for each year from 1993 to 2004:

1993: 2.02
1994: 2.07
1995: 1.96
1996: 2.00
1997: 2.01
1998: 2.28
1999: 2.35 (with an injury that presumably would have hurt his production)
2000: 2.24
2001: 2.63
2002: 2.16
2003: 2.20
2004: 2.24

Well, there's certainly one number in there that bolsters your case: 1998. And there's one number that most people jump at in the other direction: 2001.

You spend a good amount of effort deconstructing that 2001 season, and I have to admit that you do an impressive job. You've just about convinced me that given what Bonds had already demonstrated, the one year spike that resulted in 73 home runs might have been merely a case of the sort of "perfect storm" that many other players have seen in the past. At least in terms of home run totals.

That said, it's equally reasonable to say that his pre-steroid jump from 1997 to 1998 was also one of those random spikes. Agreed?

But what doesn't pass the smell test is that overall PF spike from 1993-1998 to 1999-2004. To have that sort of a prolonged jump, beginning at the age of 34 defies any sort of precedent.

Again, there may be a non-juicing explanation for it, but I don't think you're getting anywhere by invoking a few random Roger Maris or Hank Aaron seasons, which either took place at a much younger age, or weren't really that much of a spike at all, using your own preferred PF metric. To try the sort of quasi-innocence by association tactic you're using for Bonds, you'll have to show us a case of any other player with a similar prolonged PF spike at an advanced age.

Maybe you can show us another example of this. God knows you have the resources to find one. But so far you haven't.

i think it will unfortunately take many deaths to convince people to stop taking steroids. len bias didn't stop cocaine use, lyle alzado didn't stop steroids...

Millions upon millions of deaths later and people still smoke. Laws upon laws and people still do illegal narcotics. You can't stop sin; you can only hope to contain it.

Eric Walker found two substantial and essentially permanent jumps. First was the 1920s, because of the introduction of a livelier ball and the sport’s Babe Ruth-inspired embrace of slugging. The second was in 1993 and 1994, when P.F. suddenly leapt 7 percent to about 1.6, where it has since settled. Walker contends that such a jump is far more indicative of a change made to the ball — which Major League Baseball has long denied — than a steroid power boost, which would have produced an effect far more gradual as the decade progressed.

Of course, if steroids entered the game in the 70s and slowly gathered a clientele over the next 15 years, you'd see a gradual rise during the 80s, not the 90s.