Dibs Posted December 30, 2006 Posted December 30, 2006 Yes, I'm sure about that. He's been calling that "regression toward the mean" for the past two months, and it's not. If you're just reading this thread, you don't have the whole absurd story. But my point is that I care not for what he calls the premise he put forward......just whether the premise is correct or not. If it is correct, it should in all fairness be acknowledged & only after that should one move forward to correct the semantics involved. As a stupid example...... If I were to say that big hairy spiders are a common cause of phobias in society.....it would be wrong(yet semantically right) to say the statement is incorrect. Spiders do not have hairs....only mammals have hairs. There is no doubt that what is meant by the statement, is correct.....yet a stickler in the terms of biology could argue the statement is wrong. Is this not exactly what you are doing? BTW, what's the answer to his question?.....post #441
Bungee Jumper Posted December 30, 2006 Posted December 30, 2006 But my point is that I care not for what he calls the premise he put forward......just whether the premise is correct or not. If it is correct, it should in all fairness be acknowledged & only after that should one move forward to correct the semantics involved. As a stupid example...... If I were to say that big hairy spiders are a common cause of phobias in society.....it would be wrong(yet semantically right) to say the statement is incorrect. Spiders do not have hairs....only mammals have hairs. There is no doubt that what is meant by the statement, is correct.....yet a stickler in the terms of biology could argue the statement is wrong. Is this not exactly what you are doing? No, that is not what I'm doing. The main difference between his example and yours is that the words "big hairy" are not material to the main point of conjecture "spiders are a common cause of phobias in society". But when HA says "error causes regression toward the mean", the misuse of the words "error" and "regression" are material to the topic at hand. BTW, what's the answer to his question?.....post #441 Go back and look it up. I see no need to repost it AGAIN.
Dibs Posted December 31, 2006 Posted December 31, 2006 No, that is not what I'm doing. The main difference between his example and yours is that the words "big hairy" are not material to the main point of conjecture "spiders are a common cause of phobias in society". But when HA says "error causes regression toward the mean", the misuse of the words "error" and "regression" are material to the topic at hand. ""Main point of conjecture."" I believe that what you believe is the main point of conjecture is nothing to do with the main point that HA was establishing. As I mentioned in earlier posts......what difference does it make to the actual concept HA is making if he titles it correctly, or incorrectly. It makes no difference at all. He could have said "What a mean regression".....or...."Errors are mean, but I regress." It makes no difference to the validity of the concept he is talking about. The example he gives is either correct or it's not. The fact that he missuses words in trying to describe the concept has NOTHING to do with whether the concept is correct or not. Go back and look it up. I see no need to repost it AGAIN. I do. You expect me to go through hundreds of posts to find a previous basic yes or no response from you when you could easily & simply type it again? That smacks of you knowing the importance of your response in relation to the relevance of your semantic arguments on HAs initial premise.
Orton's Arm Posted January 3, 2007 Author Posted January 3, 2007 Well...no. The crux of the argument is and has been that he's specifically manufacturing a specific effect and calling it something else, and pretending it's rigorous math. That fails on three points: 1) His example is completely manufactured. I'm sorry, but there's no room for intelligent and informed disagreement here. The effect is real and common. Read this quote from Duke: Regression to the mean is an inescapable fact of life. . . . Your score on a final exam in a course can be expected to be less good (or bad) than your score on the midterm exam. A baseball player's batting average in the second half of the season can be expected to be closer to the mean (for all players) than his batting average in the first half of the season. And so on. The key word here is "expected." This does not mean it's certain that regression to the mean will occur, but that's the way to bet! (More precisely, that's the way to bet if you wish to minimize squared error.) The phenomenon is relevant to medical studies (reread the Tufts quote), it's relevant to I.Q. and testing, it's relevant to any other situation which involves testing and retesting. 2) He's grossly misusing terminology. To the point where his premise is not correct (specifically, he's confusing two very different things - error distribution and population variance). This accusation is absurd. I've been very clear about the fact that regression toward the population's mean results whenever there is a test/retest situation, and where the correlation coefficient between the two tests is less than 1. There are a number of potential reasons for a less-than-1 correlation coefficient between test and retest; one of which is measurement error. ) He's pretending to be mathematically rigorous without being the least bit mathematically rigorous. By way of analogy, he's pointing at a dog and calling it a cat because it happens to be a domestic pet with fur and four legs, so what's the difference? It's a semantic argument because in this case the semantics are actually important. In this case, "regression" has a specific and definite mathematical meaning - either linear regression or multiple regression, take your pick in this context. Yes, it has those meanings, but in addition it has the meaning I've been using, ever since Sir Francis Galton started describing "regression toward mediocrity." (Today the phenomenon is known as "regression toward the mean" or "the regression effect.") You seem to think that if "regression" is understood to imply A under some circumstances it cannot also imply B under others. Well, maybe it shouldn't also imply B, but guess what? it does. If you're semantically rigorous, that's easy to explain. If you're a complete dunce like HA, and keep using inexact and colloquial terms like "luck", no amount of explanation is going to help.The people who wrote the Stanford and Duke articles also used the word "luck." Maybe they're also complete dunces unable to receive help from any amount of explanation. But have you considered an alternative view? I know I'm going out on a limb here, but just maybe the people at Stanford and Duke a) actually know what they're talking about, and b) want to express this knowledge in as clear a way as possible, and c) have the self-confidence and humility to use words like "lucky" and "unlucky."
Ramius Posted January 9, 2007 Posted January 9, 2007 I've done some more thinking on this topic... We were discussing an statistics problem here at work, and someone said "Why don't you ask that idiot on the internet" (Holcombs Arm). This, led to a discussion of dice with a true value of 3.5, and an investigation into the properties of such Holcombian dice, from which we determined this... Prove, using Holcombian statistics, that half the time you don't roll a die, you get a value of 1. Proof: a 6-sided Holcombian die has a Holcombian true value of 3.5. It is trivial to see that a Holcombian die with any other number of N sides will have a Holcombian "true" value of (N+1)/2 (ex: a 7-sided die has a Holcombian true value of 4, a 36 sided die has a Holcombian true value of 18.5, etc.) From this, we can determine that the Holcombian true value of a Holcombian die with zero sides is actually one-half. However, if you try to "measure" the value of the zero-sided die, you will naturally get a result of zero. This represents, of course, an error of -0.5 in your measurement, which therefore means if you measure a zero-sided die a second time, the measurement will regress toward the mean, and you will measure a value of 1. And since rolling a zero-sided die is mathematically equivalent to not rolling an N-sided die, it follows that, because error causes regression toward the mean, not rolling a die gives a value of 1 half of the time. I'm sure there's many, many other interesting properties of Holcombian dice that follow from this...such as: a sixth of the time you roll a six-sided die, you get a one, which is equivalent to not rolling the die at all...therefore, rolling a die will regress to not rolling the die because of error...
Orton's Arm Posted January 9, 2007 Author Posted January 9, 2007 I've done some more thinking on this topic... We were discussing an statistics problem here at work, and someone said "Why don't you ask that idiot on the internet" (Holcombs Arm). This, led to a discussion of dice with a true value of 3.5, and an investigation into the properties of such Holcombian dice, from which we determined this... Prove, using Holcombian statistics, that half the time you don't roll a die, you get a value of 1. Proof: a 6-sided Holcombian die has a Holcombian true value of 3.5. It is trivial to see that a Holcombian die with any other number of N sides will have a Holcombian "true" value of (N+1)/2 (ex: a 7-sided die has a Holcombian true value of 4, a 36 sided die has a Holcombian true value of 18.5, etc.) From this, we can determine that the Holcombian true value of a Holcombian die with zero sides is actually one-half. However, if you try to "measure" the value of the zero-sided die, you will naturally get a result of zero. This represents, of course, an error of -0.5 in your measurement, which therefore means if you measure a zero-sided die a second time, the measurement will regress toward the mean, and you will measure a value of 1. And since rolling a zero-sided die is mathematically equivalent to not rolling an N-sided die, it follows that, because error causes regression toward the mean, not rolling a die gives a value of 1 half of the time. I'm sure there's many, many other interesting properties of Holcombian dice that follow from this...such as: a sixth of the time you roll a six-sided die, you get a one, which is equivalent to not rolling the die at all...therefore, rolling a die will regress to not rolling the die because of error... Drunk already? It's not even noon!
Ramius Posted January 9, 2007 Posted January 9, 2007 Drunk already? It's not even noon! not drunk, but attempting to your use logic on statistics made my brain mush. I really cant comprehend your unrelenting levels of mathematical retardation.
Orton's Arm Posted January 9, 2007 Author Posted January 9, 2007 not drunk, but attempting to your use logic on statistics made my brain mush. I really cant comprehend your unrelenting levels of mathematical retardation. You did get one thing right when you admitted you "cant [sic] comprehend" what I've been saying. The fact that a lot of smart people from Stanford, Duke, Berkeley, and the University of Chicago are saying the same thing escapes you.
molson_golden2002 Posted January 9, 2007 Posted January 9, 2007 Is that french for bump? I think kittens are cute
Orton's Arm Posted January 10, 2007 Author Posted January 10, 2007 I think kittens are cute Women think I am cute.
ieatcrayonz Posted January 10, 2007 Posted January 10, 2007 Women think I am cute. Better check in with the stats prof chief. I think the mean is in for some big time regressioning.
Sketch Soland Posted January 11, 2007 Posted January 11, 2007 It took me a while but I knew this thread reminded me of something
Recommended Posts