TobiasFunke said:
dparker713 said:
TobiasFunke said:
Love Pos. Great read as always.
My two biggest issues with Walker's study and conclusion, admittedly after only one read:
1. His statistical analysis is focused entirely on baseball-wide data. That seems faulty to me, as it also includes steroid-using pitchers and non-steroid-using batters. I mean there's obviously some value in it, but it seems to me that the more relevant data if you're considering the impact of something vs. a control would be to measure only the numbers of confirmed/strongly suspected users vs the control. If you do so, they idea that 1995-2004 was nothing more than a minor blip goes out the window. The unbelievably disproportionate number of
these seasons that are attributable to confirmed or strongly suspected steroid users between 1995 and 2004 demands to be addressed in any analysis of whether steroids help you hit HRs.
2. A large part of his argument is that the smallest tinker can make a big difference in HR numbers- like for example minor changes in the ball. Another large part of the argument is that steroids mostly help you build upper body strength, and while they help you build lower body strength, it's not nearly as much as they help you build upper body strength, and HRs come from lower body strength. But if even the smallest tinker can make a big difference in HR production, and steroids admittedly help you at least a little to build lower body strength, and HR production comes from lower body strength ... how in God's name can you conclude that they can't possibly make a big difference in HR production?
1. That would only succeed in introducing confirmation bias into the study.2. He based his findings on calculated increases in bat speed due to increased musculature. His numbers did not show enough of an increase in bat speed to account for the observed increase in HRs.
Not sure what you mean by #1. I was bringing up a criticism of his methodology, not suggesting a foolproof methodology of my own. I don't think baseball-wide numbers are the best arbiter of the impact of steroids on home run hitting, because pitchers usedsteroids too, and some hitters did not use them. It seems to me that since nobody is claiming that all of baseball took steroids, nor are they claiming that steroids were only taken by hitters or only helped hitters, it's the elite power hitting seasons throughout baseball history that should be the more important data point in evaluating whether or not steroids impact power numbers, not game-wide numbers.While seasons put together by hitters generally, and confirmed or strongly suspected (with evidence to back the suspicion) steroid users specifically, between 1994 and, let's revise it to say 2007 (since the benefits of steroid use may last for months or years after usage ends) are only a tiny, tiny percentage of the total seasons played by major league baseball players in the "live ball" era, they are a HUGE percentage of the elite HR-hitting seasons. The variance in total home runs hit per year, while it obviously peaked in the steroid era, can't possibly account for numbers that out of whack, yes? So that means any alternate theory as to why HRs peaked during that time frame, by itself, can't explain the extraordinary differences in the elite seasons during that era, right?
I'm no statistician, just thinking intuitively. If you can straighten me out here, I'm all ears. I mean, I get some of what Walker is saying- we shouldn't just assume steroids helped, we should look at the data from a detached, scientific viewpoint. But I'm wondering if he's trying too hard and missing some obvious stuff. For example, he trashes the work of a physicist, Roger Tobin, who found that the 10% increase in muscle mass could produce a 50% increase in home run production simply by noting that there was no such increase. But isn't he in the process of telling us that such a simplistic data evaluation or HR increases or decreases is invalid without considering other possible mitigating and aggravating factors. Like, say in this instance how steroids may have helped pitchers too during that era?
When making a subset of data, you need to make choices based on assumptions. Here, you want to use your assumptions of those you suspect were aided by steroids to hit HRs to determine if steroids aided them to hit HRs. This means of selection leads to erroneous confirmations of the underlying assumptions. Basically, its cherry-picking data. In the present case, steroids are believed to have been a systemic problem for many years in baseball and that steroids were the primary cause of increased power numbers in the majors during those years. If those assumptions are true, then there is no issue using league-wide statistics. This is helpful because it means that no speculation is needed as to who was using steroids and who was not.
Also, the common wisdom on pitchers and steroids surmises that pitchers primarily benefitted from greater fastball velocity due to steroids and that steroids were not significant in increasing the effectiveness of breaking pitches. There is a positive correlation between fastball velocity and effectiveness, however, there is also a negative to higher velocity from a pitchers' perspective. Simple physics shows that the more velocity an incoming pitch has the farther it will travel in the opposite direction if struck. When you watch Strasburg pitch this reality becomes fairly obvious. By using league-wide statistics, both of these competing factors are included in the analysis.
As for elite season variance, I don't recall if he looked to see if there was a different distribution of HRs per player then in years past. But even if there were, that would be inconclusive at best. Historically great players are exceedingly rare and also an extremely small segment of the MLB population at any given time. For instance, Clemens, Maddox, and Randy Johnson can all be legitimately considered among the top 10 pitchers of all time and all of them played at roughly the same time. Same can be said of contemporaries Walter Johnson, Grover Cleveland Alexander, and Christy Mathewson. This clustering of talent is actually more likely with such a limited sample than a more even distribution. So its quite possible that many of the greatest peak and career HR hitters of all time were playing all during the same era. However, most people intuitively expect a random distribution to spread things out rather evenly because randomness violates human sensibilites. Aside from the possibility of a collection of extraordinary HR hitting talent, smaller strike zones, expansion, smaller ballparks, pitch counts, differing ball construction, de-emphasis on defense, de-stigmatization of the strikeout, improved training without steroids, and video study are all potential contributing factors to extreme HR rates.
I think the main point to take away from the article is that anyone claiming to KNOW that steroids were the primary factor in the power increases in MLB really is talking out their butt. Its a plausible, maybe even probable contributing factor, but there are just too many variables to isolate the effect of steroids, and given what we know of swing physics there does not seem to be enough extra bat speed generated through increased muscle mass to create a spike in power anything like we observed. (As for Robin Tobin's work, Im pretty sure its been thoroughly panned. When I read that article I thought the methodology was pretty poor)
