Wait, how can the strikeout rate have ANYTHING to do with the HR rate after contact? Isn’t contact, by definition, excluding strikeouts? One problem might be the counterfactual — what is the comparison?
Along these lines, here’s this story: After the height of the steroids era, players started “swinging for the fences” more often. This has two effects — it increases the strikeout rate as they have a harder time adjusting, and it increases the rate of HRs per contact. In this case, you could argue that the true HR after contact rate would have been declining after the late 1990s IF PLAYERS USED THE SAME (on average) APPROACH now versus the late 1990s. The higher strikeout rate already suggests that the players are likely using (on average) a different approach now versus then (yes, it could be a change in pitching, too, but still).
Dave’s point is that the gap in home runs between 1998 and 2012 (which actually isn’t all that big) has to do with players making more contact in 1998 (thus more chances to hit home runs), not with more balls in play turning into home runs.
I get Dave’s point. My comment is really just that a straight comparison of the HR rate in 1998 and 2012 implies that the two are comparable along every dimension except (to Dave’s point) for PED usage. This seems to be false given the exceptional rise in Ks during that period. It is possible that, if players were approaching their PAs now the same way that they did in 1998, we’d see a MUCH lower HR rate in 2012, but that it has been buoyed by a change in PA approach.
On the other hand, if you look at this metric on a rolling average or other longer-term basis it was quite clearly lower from 2007-2012 (when it was never above 3.7 until this year) than it was from 1998-2004 (when it was never below 3.7). A one year blip in 2012 does not obliterate a clear trend in the numbers YOU cited.
Dave, you are correct that the world is not entitled to its own facts. You ARE entitled to selectively choose your facts, as you have done here rather egregiously, but the rest of us are entitled to call you out for it.
I think the point that should be made is that the rise in home runs in baseball in the late 90s-early 2000s was the result of a lot of different factors. Using “the players were juicing” as the end-all, be-all argument is both intellectually dishonest and historically inaccurate.
But doesn’t this somewhat assume that everyone in 1998 was using PEDs and everyone now is not?
While I’m not sold on PEDs actually making people better baseball players, I do think there are a few important factors to tease out:
1) Did PED use create an even wider gap for those using them and those who did not? Basically, when McGwire faced an non-PED, league average player, did this increase his chances of hitting baseballs long distances?
2) What explains the drop at the top end of the distribution scale for home runs over the last five or six years (since ’08 we’ve seen one player hit more than 50 home runs. Between ’97 and ’01 we saw 13 and almost all those guys are admitted/strongly suspected PED users).
3)I haven’t really looked into this, but from what I’ve read, total home runs are roughly the same, home run per contact is about the same, so are home runs more evenly distributed these days (aka are the top home run hitters fewer standard deviations from the mean than in, say, 2004)?
Yeah, exactly. Dave is using league data to pretend that steroids didn’t have a dramatic effect on home run rates, but we already know from individual cases that it absolutely did. Not only did home run frequency drastically increase for players later proven to have cheated, but the distance of their home runs also jumped tremendously. I forget the exact numbers, but Bonds had something like 3 homers of more than 450 feet before 1998 and 20-something afterwards.
There were more homers and more contact plays regardless of the percentage of homers per contact plays. They made contact more often and hit more homers. Steroid use may not be as bad as it once was, but it’s still a problem today.
I have a tidy explanation for those years, those were the years that 26-40 AAAA pitchers were added to rotations, just like 94 was a big jump over 93 if you add a bunch of pitchers that weren’t good enough to be in MLB the year before then you let premier hitters like Bonds feast(BTW this is also why premier pitchers like Pedro and Clemens did well)
I think Dave is cherry-picking the two data points from the graph that support his point while ignoring the larger trend. In that data, I see an increase from 3.0 to 4.0 between 1993 and 1999, then a plateau just below 4.0 from 1999 to 2006, then a decline from 4.0 to 3.5 between 2006 and 2011.
2012 is an oddball that might be a blip in a downward trend or the start of a new upward one. We can’t know until more data is added to the right side of the graph.
And 2001 was the year Bonds hit 73 home runs. Why no mention of 2001 in the article? Didn’t fit the narrative?
Dave’s point is valid – people do choose their own facts, just like Dave did here.
It is a fact that HR’s spiked dramtaically during the steroid era. What is not a fact is to what extent steriods contributed to the spike.
On the other hand, to say they were responsible for 100% of the spike doesn’t make sense when there were about 4000 HR’s/yr in 1993 and 1995, but in the last 5 years the average has been about 4,800 HR’s/year.
This really only works if we assume that the only value of steroids to power hitters is increased distance on hit balls, and not an increase of rate of hitting balls. I know that this is really an argument against the writers that say “Steroids absolutely made more home runs” – but I don’t know if it is as cut and dry as this posts makes it sound. I think that the large amount of home runs in the era, marked with the potential that more guys could hit balls that they may not have been able to hit made for an increase in both numbers. And I’m no mathmatician, but I bet someone could talk about how big of a difference both contact plays/home run rates during that era do not fit the standard deviation of the rest of the time period. Anyhow, other guys in this thread could probably numerically make that point better, but really I think that other Fangraphs guys could break this apart pretty easily.
And on that note, any “Hall of Fame” that doesn’t have Barry Bonds in it, on the first ballot, PED usage or not, isn’t a Hall of Fame. Basically, not putting Bonds in makes the Hall completely irreverent in my book.
We can argue about the Tim Raines, Sosas, Jack Morris’ and Schillings of the world… but Bonds (and Clemens) deserve to be in there no questions asked.
Right. Or to put it another way, what does the list look like when you rank HR per Contact % historically for people that have over say, 450 HRs? I haven’t done the math, but gut feel is the top 10 is going to be littered with PED guys from the time period in question here. It’s just too coincidental to put aside the PED issue when talking HOF, even if you assume there was some increase in bad pitching in the league.
Nice work! I did the home runs/PA thing a few months back and found they were significantly higher, but your way adds so much more to the analysis! It would be nice to see the change in strikeout rates over this time just to make it easy to comprehend.
Here’s a tidy explanation, since 2002, curveball usage has steadily gone up. Pitchers are smarter than they were in 1998-2002 and throw fewer fastballs, in addition to fooling more batters, fewer fastballs means less ball velocity approaching the bat which=diminished batted ball distance.
Is there anything worse than when somebody makes a terrible argument and then is pompus about it. Dave, please understand that your simples facts don’t demonstrate any of your conclusions and that I can still pretty easily argue PEDs helped players hit more home runs when they made contact.
I was thinking this as well. Maybe players swing for the fences more now, need to swing harder than they did in 1998 to get it out.
Or maybe they are still taking PEDs and just got better at masking it.
As for strikeouts, just looking at that, perhaps advanced metrics have convinced teams getting more homers is worth more Ks, and/or the old idea of “pitching to contact” wasn’t optimum.
It’s interesting, if you go back and look at successful pitchers from earlier era, like say Early Wynn, his k rate was low, not to mention going back even earlier. Batters really struck out a lot less “back in the day.”
This is spot on. I usually appreciate Dave’s articles, but in this case he was swinging for the fences and totally whiffed. Needs to rethink his approach here.
Comment by Sidney Ponson — January 9, 2013 @ 12:52 pm
Exact quote from Dave’s Carlos Ruiz article 6 weeks ago:
Last year, his HR/FB rate was 15.1%, and his ISO was .215. Or, if you prefer pictures, here’s Carlos Ruiz’s slugging percentage, by year, compared to league average.
Yeah. It’s not hard to figure out which of these is not like the others. And so, today, no one is really shocked that Carlos Ruiz has been suspended 25 games for failing a drug test, specifically one for amphetamines.
Well, PEDs could very easily lead to more contact.
Just simply: they make you stronger, enabling you to get around quicker, this in turn gives guys a split second longer to analyze a pitch, allowing them to not just make more contact, but better contact.
Further, the somewhat ignored benefit of PEDs, is they can help artificially lengthen a player’s career/prime. A 23 year old has physical gifts a 37 year old does not; while a 37 year old has mental gifts (largely in the form of experience, but that counts for a lot) that the 23 year old does not.
You take the mind of a 37 year old and put it into the artificially created body of a 23 year old and you have a really, really excellent combination – an unnatural one, really.
And, in general, PEDs probably are more advantageous at the fringes. They’ll take a AAAA player and give him just enough extra “something” to make him into a big leaguer, and they’ll take an all-star and give him the extra “something” to turn him into a once in a lifetime legend. ie, it raised a level 1 talent to a level 3 and a level 8 to a level 10. The rest of the guys went from 4 to 6, or 5 to 7, which is an improvement, but not necessarily something that sets off bells and whistles.
This isn’t to say that PED usage should keep a guy from the HOF, because, honestly, the pitchers likely saw a great deal of benefit as well, and it’s too murky to me to keep the all-time greats out.
Comment by MrKnowNothing — January 9, 2013 @ 12:53 pm
Dave makes a good point. Even if you take the high average of 4, and compare it with the low recently of 3.4, or even 3, you get a difference of 15-30%, which means say Mike Schmidt who hit 38HR in the 70’s and led the league might have hit 6-12 more HR’s in 2000.
A lot of this was due to expansion, as good hitters feast on bad pitching. The expansion of 94 and 98 diluted pitching a lot. I’m sure if you looked at 69 or 78 you would see similar bumps in HR per contact. Of course, 69 also came with the lowering of the mound, so maybe 78 when Foster hit 52 would be a better example.
In fact, without expansion and the DH players like Aaron and Mays would have never hit the all time HR milestone’s they reached as they experienced a mini-renaissance for a few years after expansion.
Steroids also had an effect we cannot completely roll that under the carper, but to me the increase in HR’s from 94-2002 was 40% bad pitching, 40% smaller ballparks and 20% steroids. I would be interested to see say 74-81 HR/CT rates.
IF the argument is that PEDs made the batters stronger and hit the ball harder… then can we filter out ground balls and only look at HR’s as a percentage of line-drives and fly balls? Unless the assumption is that hitting the ball harder also means being able to loft the ball, that is, the PEDs also help you get the ball in the air.
If it looks like a duck, swims like a duck, and quacks like a duck, then it must have not been measured correctly?
I am not personally against putting the PED guys in the HOF, I have no moral beef, it happened and that’s that. But I believe it’s not proper to use this data to downplay PED importance. These guys looked like they could lift cars. Then hit them over the outfield.
Great set of statistics. However, by taking the entire league contact rate through the years aren’t you implying that everybody is doing steroids or not doing steroids? obviously not everybody in the majors was taking steroids in 1998. In fact the majority of MLB probably wasn’t taking steroids then. If there was a better statistic on how many players, or even better the percentage of players, that were believed to be on steroids we could understand just how much steroids contributed to these numbers.
Also this does not account for ballpark changes such as the yankees moving in the fences.
Something that would be interesting to look at is the distribution of HR across the league throught the years. Having 10 guys hit 50 HRs each comes out to less HRs than 20 guys hitting 30 HRs each. Maybe there was a spall concentrated group (possibly steroid users) back in the late 90’s that made up a large percentage of the HRs and now it’s players hitting a few more HRs than average but everybody stepping up their game.
Also, how many players tested positive fore steroids in the last 3 years? Steroids are not out of baseball yet.
This all assumes that PED use is still not rampant. In recent years melky and manny have been caught, and for every rat caught there are hundreds who have not. Old anabolic steroids focused on mass gain, newer cycles of PEDs have focused more on fat burning. And yes, peds help contact rates.
With the data provided from Dave, I calculated the standard deviation (or more specifically google docs did) as .24%, with an average HR per contact of 3.69%. Add them together and you get 3.93% as the max HR per Contact that is within “average”. Which means 1999-2001 were the only years that were significantly higher than average. A few theories that could explain a higher home run rate in the late 90’s.
1) League expansion. The main theory being that with more teams, the talent pool of good pitchers goes down. The corresponding drop in hitter talent could be explained by the prevalence of AAAA type hitters, who can hit AAA pitching but not good mlb pitching.
2) Hitters approach and training methods advanced faster than the pitchers, who eventually caught up. The pitch selection numbers show pitchers have changed their approach around the same time.
3) Stadium selection. Have more pitcher friendly parks replaced hitter friendly ones than vice-versa? Are teams playing in pitcher friendly stadiums unable to land power hitters? Do teams in HR friendly parks now built out a rotation of ground ball pitchers more?
4) Roids. Not the, roids are magic fairy dust that that make homeruns happen, but more from the standpoint that they enable older players to swing like they were young again. So they get the lower strike out rates from old man plate discipline and young man bat speed and peak power (and then some maybe).
5) Random chance as it pertains to which side of the hitter/pitcher roles where a generation of great athletes wanted to be sluggers vs pitchers or SB kings.
6) More HRs, combined with the lower strike out environment lead to more PA opportunities, so pitchers tired out more and hitters got better pitches to hit off of the tired pitchers. –This is stretching so I’ll stop here, but it would be interesting to see what the average inning HR’s were hit in per year. Also the Rsquared between PA and HR/Contact is only .3 (after removing 94/95 because they were strike shortened), so this is probably not a useful one.
I think in our hearts most people would love to punish the steroid users by keeping them from the Hall for a while (or forever) IF we knew who used PEDs. We don’t and never will. To guess based on appearance and heresay is not feasible.
I say let the best players from the generation get into the Hall and put an asterisks for steroid era next to them all, including Pedro and griffey and guys who aren’t suspected. It’s not a judgement on guilt but a label that they played when a large percent of the league used drugs and stats are not necessarily comparable to other eras.
Keep suspected users out a couple years, though; make them squirm a little at least.
The premise of the article is that players were making more contact and more contact created more opportunities for home runs. I disagree with the thinking that performance enhancing drugs don’t help players make more contact. Although i do think a lot of the home run explosion can be tied to expansion teams – rockies, dbacks, marlins, rays. The talent in the majors became diluted and elite perfomers could feast on subpar talent.
I couldn’t agree more with this. And to add some, when you don’t have that extra strength, presumably from PEDs, you need to swing “harder”, especially to get around on pitches with the kind of increased velocity pitchers have these days. Swinging harder, as we all know, would certainly lead to the drastic increase in batter K’s as well as the much lower contact plays shown in the last 5 years. But it also would explain why the HR per contact play remained consistent with the ’98 rate. Swinging harder means similar bat speed so when players do hit the ball it will still travel HR type distances. In addition, with pitcher velocities increasing this would also account for some of the additional power generated to produce a HR.
Data doesn’t lie but interpretations are notoriously subjective. What was the distribution among players like? In 1998 the top 10 players hit a total of 517 HRs, with #10 at 44. In 2012, the top 10 players hit a total of 396 HRs, with #1 at 44 HRs.
Even if steroid use was widespread in 1998, it doesn’t mean that it was equally effective among all users. Have training regimens improved overall? Is the average player a bit stronger and faster than 12 years ago? If the high end has come back to earth and the low end improved, would this explain the data?
I don’t know, but the bridge from data to interpretation seems a bit narrow here.
Let us not forget the postage stamp sized strike zone in play through the late 90s and early 21st century. When a hitter can zone in on a smaller area, it’s a lot easier to make that sort of contact.
I also like to think it wasn’t only the players that were juiced in the late 90s, that MLB tweaked the baseballs to try bring fans back. Which largely worked. Pure opinion, obviously, but MLB has played with the ball before, and in both directions.
The very broad trendline is that Contact / PA has been falling since 1919 while HR / Contact has been rising. The more recent trendline is that HR / Contact shot up in 1993 and has roughly plateaued since then. The Contact rate dipped, plateaued, then got back to dipping in that period (and it’s currently the lowest it’s ever been).
That’s a pretty rough-and-ready look at the issue, but if nothing else I think it shows that chopping off home run data at 1993 is not going to provide you with a very complete view of things.
Comment by Rex Manning Day — January 9, 2013 @ 2:02 pm
One year blip means nothing… especially if you’re assuming 2012 is more clean than 2001.
It might also be worth looking into strike-zone changes in the years following the HR/contact peak (1999-2001), given MLB’s edict to umpires in 2001 to shift their strike zones to better align with the official definition, and the subsequent introduction of QuesTec.
If not mistaken, research here has indicated that the strike zone did became larger post-2001, which would presumably require greater plate coverage on the part of hitters, and the need to make contact on pitches along edges less easy to drive for HRs than would be the case with a more squeezed zone.
The premise of PED usage is that players had unnaturally high strength, among other benefits. Stronger players have better bat speed. Hitters with better bat speed are going to make more and better contact on a wider assortment of pitches than weaker hitters.
Almost every major league hitter can belt a 75mph meatball down the heart of the plate. It’s the hitters who can go deep on a 95mph heater up in the zone that hit more total homers.
While this is very interesting, you could say that the steroid usage wasn’t rampant in 1998, and only a couple handfuls were using (McGwire, Sosa for example) and they benefitted. McGwire had a 19.77% HR/Contacted Ball rate in 1998, Sosa a 13.98% rate. The highest rate in 2012? Adam Dunn with a 12.93% rate (next highest 2 are Josh Hamilton (10.75%) and Granderson (10.72%)). McGwire’s rate is such an extreme outlier, you could definitely argue it had to do with PED’s.
McGwire went through his injuries in 93/94, when he admitted he began using PED’s. Pre-PED use, his best HR/Contacted Ball rate was 11.6%. Post-PED use (if you throw out his small sample injured years of 93/94), his worst HR/Contacted Ball was 15.2%. Even if you include 93/94, only 1994 was he ever below his best pre-PED year.
I’m no statistician, but I really wonder about the significance of these numbers. If as a back of the envelope guesstimate you say introducing the humidor at Coors took 50 hr’s out of the total from 2002 on, that’s about a third of the difference from the peak range to the current range. But Coors goes back to 95 and the Rockies to 93, meaning some portion of the historical spike is simply that stadium, and some of the downward trend in the last decade could be the stadiums subtracted and added since. But Safeco opened in 99, right at the start of the peak, so what the hell do I know.
Still, what’s weird to me is all this moralizing about PEDs and next to no mention of rabbit balls. Steroids have been in popular culture since the 70s, MLB players have been taking them since the 70s. It does seem likely their efficacy was highly improved at some point in the 90s, but did everyone take them in 87, then agree not to again until 94?
No it doesn’t prove anyone is clean, it just shows how gnarly the HR per Contact % is for the guys leading the HR charge in the roid era.
Another interesting study would be – individual player HR per Contact % / League avg HR per Contact %. Wouldn’t that help clarify some of the AAAA pitcher and juiced ball arguments? Again, gut feel is the top 10 would be dominated by the major PED guys.
You may have made an inadvertent point, which is Bagwell HR per Contact % may be in line with the games’ immortal power guys…and well below the leaders of his era. I wish I had the numbers, it’d be interesting to see.
Dave is making a point about the so-called “steroids era” that has tainted (supposedly)the candidacies of players like Biggio, Bagwell, and others. He’s not addressing anyone’s specific candidacies.
The point is that the “steroid era” isn’t really allow that different from today. In fact, it’s plausible that we’re still in the “steroids era”. So who gets to decide when it began and ended? Why are players for whom there is absolutely no evidence of steroids use — such as Bagwell, Biggio, Piazza, and Schilling — penalized simply because they played in an era when homers were no more likely to be hit on hatted balls than they are today? An era that is distinguished from the present for no real reason whatsoever!
He’s not making the case for voting for specific players. He’s simply saying that it’s ridiculous to manufacture arbitrary “eras” to simply fit one’s preconceptions.
What was the HR/FB rate for the league in each of those years? What new stadiums opened since then that may have influenced HR rates? How did they average game time temperature influence HR rates over the years? I ask that last one because just read 2012 was by far hottest year on record in US. An average temp of 1-2 degrees higher over a season at stadiums could have a statistically significant influence on what you seek to measure.
Heyyy, really neat trick, Dave: disappear the rates of the _obvious users_ into the aggregate data of the many _non-users_. And of course with many pitchers juicing also at the same time, it’s reasonable to infer as a first hypothesis that the non-users had somewhat suppressed HR rates relative to a putative baseline too, thereby suppressing the _overall_ rates marginally the other way. . . . Which of course gets no mention in the putative analysis. (Think about all of those relievers suddenly throwing 95-96 mph racking up Ks; kinda suppresses those HR rates.)
Then there’s the strained comparison of ‘juice years’ to 2012 as if to show nothing’s changed—and one thing hasn’t: We still have a certain number of players hitting more balls in the seats by using PEDs. I haven’t forgotten Melky, or Braun’s lawyered out positive, or a half dozen other guys with big numbers that seem odd indeed. Guys have gotten better at beating the tests, and there’s some reason to believe that marginal players are doping less, but it’s still going on. So the decision to use 2012 as a clean/cleaner year for a comparison proves nothing beyond what the analyst sets out to prove. Certainly using it as a ‘control year’ without mentioning the problematic nature of current results is an analytic decision to question.
Sure, looking at aggregate rates to see if there is something there is a great place to _start_ a piece of analysis, but no place to end it. The demonstrable fact that this quick and dirty look seems half thought through, with the obvious follow up questions *cough* NOT asked, doesn’t strike me as a coincidence. The next time I want folks to stop looking at something I’d rather they don’t see, I’ll remember to pull out this technique myself.
Oh for god(s) sake, chuckb, please take your foot out of your mouth and put it back on the floor. mcbrown took _all_ years after a certain date, showed a trend, and even smoothed that by rolling year data. There is no way to call that cherrypicking, he was inclusive. Think his timeframe is arbitrarily short and so the trend is questionable? He could have included the putative spike years, and _still_ would have shown a decreasing trend over that entire period. And btw, there is a rolling year _increasing_ trend in to the spice years from the early 90s also. And I’ll bet my wallet against yours that if Dave had done a LEGITIMATE study and included data from the entire 1980-2012 we would have an obvious spike in HR ratess per contact in exactly the late 90s which stood out glaringly. Why do you think Dave cut off the front end of the study _after_ players had ALREADY started juicing in the late 80s?
And as ARB mentions below, using HR per contact is rather a dicey approach to begin with. Contact itself isn’t constant, and there’s reason to at least suspect that contact improved also for those using PEDs. As Dave says, because it is widely known, K rate dipped quite demonstrably during the peak of the steroid years, i.e. contact went UP. If contact goes up, it effectively masks contact _result_ when that result is considered as a rate. A slightly higher rate of a larger number is a larger absolute number relative to a longer baseline without the contact spike. At the very least, this study should have been controlled for the drop in K rate.
While I appreciate the broader point, I think the way the data is presented in percentage terms masks the fact that the HR rate during the ’99-’02 period is 7.5% higher than it has been in the period 2003 (beginning of testing) to date.
There are lots of factors that could contribute to that, but I don’t think the discussion is furthered by burying it.
Also – if the HR/Contact rate for pitchers is lower than for position players, mightn’t you expect some modest decrease in HR/Contact starting in ’97, as AL pitchers had to bat in inter-league games? Maybe it’s too small to move the needle, but worth exploring.
Finally – and I think this has the potential to be more meaningful – isn’t there evidence of +positive+ correlation between higher strikeout rates and higher HR rates? (eg, ‘swinging for the fences’). If so, wouldn’t the increase in K rates Cameron cites argue for, all else equal, an increase in HR/C rather than decrease?
Ok chuck, here you go. For each year starting in 1995, I present the HR/contact rates of the prior three year period (inclusive). So for 1995, the rate is the sum of all HR’s from 1993-1995 divided by the sum of all contact plays in the same period:
This is really dreadful. Its true that we aren’t entitled to our own facts, but you are spinning like a top to see these facts as support for your thesis. Plot the last column. It is a curve peaking in 2000! …never mind the logical fallacy that the hypothesis that steroids improves performance predicts that HR/contact rate will increase.
This is advocacy masquerading as analysis. It really is a disservice to the SABR enterprise. You should be modeling yourself on scientists. You get more and more like the scientific equivalent of a climate change denialist or a bigfoot hunter all the time.
While I am one of the first to say Bonds was likely a juicer, the 450+ HR thing could also partly be because of where he played before and after 1998. SF’s new waterfront stadium where he played for several years had a RF wall that didn’t have stands behind it like Pittsburgh did. There was a maximum distance Bonds could hit a ball in Pittsburgh and in Candlestick where eventually a fan would interrupt the ball’s flight. In the new SF stadium (AT&T?) there is an ocean behind the wall, therefore the ball can travel extra distances before it finally hits something for the first time. And we all know Bonds had many splash hits in that stadium, so that MAY play a larger role in his 450+ HR balls than steroids. I don’t have any evidence proving for or against that though, just a thought on how it might help.
I like what the numbers show here, but what I’d like to see is the number of players hitting a significant number of HR’s in 2012 compared to 1998. Hard to pinpoint “significant” I guess, but what about showing the number of players who hit 20, 30, 40 and 50+ for every season. That might show more of a separation between the leagues regular power hitters and the elite power hitters and how many of them there are each season.
Excellent work on this data though, I’m a believer in the steroid era being bad for baseball and I still like seeing these bits of data showed, it helps keep me honest when I debate the steroid era with others.
Dave, I love your work and admire your perspective, but this analysis is fundamentally flawed. Other comments above have echoed my primary complaint, which is that your breakdown of league data cannot survive the transition to a treatment of individuals. There is a macroscopic echo of the uncertainty principle in this point; you may establish limits on the range of possible performance, but you do not have the information with which to interpolate the effect upon an individual element in the set.
To put iit another way: McGwire was a big guy, a weightlifter, with a brother who played professional football. His access to steroids, and perceived utility from them, would have been far greater than that of most of his peers. The same is true, to varying degrees, of other power hitters, whose familiarity with bulking (and direct incentive to bulk) would have made them cluster along the front edge of the PED function line.
The league-wide analysis does not only obfuscate the knows and nearly-knowns in this case (admissions, evasions, regressions, etc…); it also obscures the most likely explanations for the deviance of the era. There is not a smooth aggregate space between the league and the individual; there are clusters and groupings which move together through the years, and any useful path from league-scale to the individual player must first wind down through them.
So mcbrown, thanks for the gruntwork. _This_ is the raw study Dave should have posted, which clearly disconfirms the contention of the post. In fact, quite a lot _did_ change in the peak steroid years as we see here, even allowing for the fact that contact also increased. And if we controlled for contact against a longer test period, the peak of the sterioid years would be far more glaring still.