[EMAIL PROTECTED] wrote: >> Well Koufax, Bob, a pretty knowledgeable baseball guy, > said that Pedro was better than he was. That's worth > something too, don't you think?
He's just being modest. But yes I would take that very seriously. > Bob, I have some idea of what a phenomenally > accomplished doctor you are. Don't believe everything you hear from impressionable young men. It is all smoke and mirrors. I'm just asking that you > to apply the same sort of rigorous thinking to > something that is much easier to analyze - if you put > your emotions aside. > > Let's say I was a pharma rep for GSK trying to sell > you on Zocor. If I came to you and told you how great > Zocor was, I'm guessing that you would demand the > clinical data. If I hemmed and hawed for a while, and > then finally admitted that, well, the clinical data > says that Lipitor is stronger, what would you say? If > I told you about how these great doctors (from before > Penicillin was invented, or the role of cholesterol in > heart disease was discovered) all thought Zocor was > stronger, that might impress you a little bit, I > guess. And I could tell you stories about that time > Lipitor didn't do anything for my friend's cholesterol > problem, but Zocor cleared it right up. But if the > M&M data said that Lipitor has better life-extending > results (which I think it does) and the clinical data > said that it was stronger at lowering LDL and raising > HDL (which I'm pretty sure it is) then would you > prescribe Zocor to your patients just because I told > you it was wonderful? I hope not. You raise an interesting point; one that goes beyond the fun of two bull headed people arguing for its own sake. What is the nature of proof? Now clearly anecdotal evidence is not as good as quantitative measure but the difficulty is in determining what you are trying to quantify. The drug analogy is edifying. It is the best case scenario for this sort of comparison. it is relatively easy to set up an experiment where the effects of a drug can be measured objectively. In your example we would use cholesterol level as our primary outcome. But this would actually be just a surrogate for our real outcome, reduction of heart attacks and strokes. Since measuring the true outcome is trickier more expensive and too time consuming we use surrogates. That is fine but this requires a judgement on what that surrogate should be. In this case in addition to primary outcome measure we would need to have secondary measures (e.g side effects). We would need to make some subjective judgements about which outcome is most important. Things are even more complex in my field where it is difficult if not impossilbe to measure some outcomes. Diagnostic efficacy sensitivity specificity positive and negative predictive value are all used to assess the value of diagnostic imaging tests. But I remain deeply skeptical that these tools tell us much that we don't know from daily clinical experience. Most of the science I have done might best be described as the art of medicine. I use statistics in my work but I know that sometimes they fail to provide clear information. Several years ago I reviewed a very complex paper on imaging of Multiple Sclerosis submitted to the New England Journal of Medicine. It concluded that MR was not all that useful in detecting and characerizing MS when compared to clinical evaluation. They had the stats to prove it. But my own experience told me this was simply wrong. I understood the data and knew why the authors had come to an erroneous conclusion but the fact of the matter was that the paper did not reflect clinical reality and subsequent experience showed this to be correct. I am no genius nor am I someone who automatically trusts my judgement above others but I knew that the conclusions of the paper were wrong because of my direct experience in interpretting studies and dealing with neurologists. > > You said that Pedro and Koufax both had the best ERA > possible. But that's not really true, is it? Gibson > had a better ERA than Koufax at least once - much > better. Gibson had the single greatest season a pitcher can have (68). Is ERA was about one run difference from Koufax. So my point is I think correct. 1.5-2.0 is about the best you can do. Rarely you can do a bit better. Since I may time out on gd aol I'll continue in the next post There's one yardstick for you right > there. No pitcher has put up numbers that even > vaguely resemble Pedro's at his peak during the last > few years. But there were pitchers who put up numbers > that were comparable to (or better than) those of > Koufax. Gibson, IIRC, won 26 games in 1968. Now, W-L > for pitchers aren't particularly informative, but, > well, how often did Koufax do that? > > Now, here is the player page for Koufax at the > Baseball Prospectus Web Site: > http://www.baseballprospectus.com/cards/koufasa01.shtml > > And here is the player page for Pedro: > http://www.baseballprospectus.com/cards/martipe02.shtml > > You tell me what those numbers suggest. I'd point out > that "Stuff", which is a rough statistic that BP uses > for dominance, has Pedro as considerably better than > Koufax in his best seasons. > > If we use your metrics - that is, just against the > other players of his time, ignoring park effects, > difficulty, everything - then why isn't Gibson the > best ever? His 1968 season was better than anything > Koufax ever did, phenomenal though Koufax was. If > Koufax had five seasons so much better than everyone > else that they automatically qualify him as the most > dominant pitcher ever - why didn't he win five Cy > Youngs? Randy Johnson has five. Clemens has six. > Maddux won _four in a row_. Pedro won three in a row, > and probably deserved more. > > You mentioned postseason performance. The first > question, of course, is how many Division Series did > Koufax have to pitch his team through? How many > League Championship Series? So yes, he did very well > in the World Series. But in terms of pure postseason > performance, did he do anything as impressive as Randy > Johnson last year? Mike Mussina in 1997? Lots of > people claimed that Barry Bonds couldn't "hit in the > clutch" because of his poor postseason performance. > Do you still think so after last year? Willy Mays, I > would point out, _sucked_ in the postseason. Does > anyone blame him for it? No, of course not. Players > who people like are clutch players, and players who > people don't like aren't, and that's as far as it > goes. > > The same thing with injuries. It's true that Maddux > has much better medical care available to him than > Koufax did - not that he's ever needed it, but > certainly it's true. But Koufax had better medical > care than Walter Johnson. Which one was more durable? > Koufax was legendarily fragile during his own era. > If you're right, and we only count players against > their contemporaries, what does that tell us? > Furthermore, Koufax had what Maddux and Pedro don't - > a high pitching mound, and the chance to take it easy > against at least half the batters in the other teams > lineup. Don't you think that decreased his chance of > injury? > > If statistics only told us what we "know" to be true, > then they would be useless anyways. It's only when > they tell us something that is contrary to our > perceptions that they are useful. In this case, the > statistics are saying something that you don't like, > Bob, but that doesn't mean they're wrong. Now, if > they declared that Andy Pettite was the greatest > pitcher ever, then clearly we'd have to cook up some > new statistics. That would be absurd. But it's > certainly reasonable to say that Pedro's 1999 season > was the most dominant ever. It's also reasonable to > say that Gibson's 1968 season was. Or one of Koufax's > great ones. It just so happens that Koufax's don't > seem to quite make the grade against Pedro's best, and > Koufax's career clearly doesn't quite make it against, > say, Seaver or Clemens. That doesn't make him > anything less than a phenomenal pitcher - one of the > best of all time. Just not _the_ best. > > ===== > Gautam Mukunda > [EMAIL PROTECTED] > "Freedom is not free" > http://www.mukunda.blogspot.com > > __________________________________ > Do you Yahoo!? > SBC Yahoo! DSL - Now only $29.95 per month! > http://sbc.yahoo.com > _______________________________________________ > http://www.mccmedia.com/mailman/listinfo/brin-l
_______________________________________________ http://www.mccmedia.com/mailman/listinfo/brin-l
