Orton's Arm

You might want to go back and reread that thread. Syhuang threw a number of accusations against me. One of his complaints was that I was comparing stats from Losman's performance in his second stint from last season to his stats from this season to see if there'd been an improvement. In addition to that, he complained because there were some scoring drives from Losman's second stint of last year which syhuang wanted me to throw out. Of those three drives, one included a touchdown pass; and I remember that the same or a different drive had a really long Losman run. Anyone who thinks those drives should have been thrown out obviously hasn't looked at them very closely.

You are confused. Nobody came to my defense for that misremembered formula. Wraith--who works with statistics for a living--came to my defense when Bungee Jumper and Ramius tried to deny that those who obtain extreme scores tend to obtain somewhat more average scores upon being retested. Then there are those who, like yourself, have felt the need to participate in this discussion without fully understanding the material. These people have generally made the incorrect assumption that Bungee Jumper and Ramius must have at least a vaguely correct idea as to what they're talking about. Have you ever heard the expression that something was so stupid only a college professor would believe it? In this discussion, Bungee Jumper has exhibited a college professor-like ability to confuse himself and others about things which ought to be perfectly obvious. Someone who scored a 140 on an I.Q. test is more likely to be a lucky 130 than an unlucky 150. There are more 130s available for getting lucky on the test, than there are 150s available for getting unlucky. Ramius's problem is that he has no idea what he's talking about, but he thinks he does.

In a discussion about whether Losman's performance was better this year than in his second stint from last year, syhuang called me out for comparing Losman's stats from this year to those from his second stint last year. That's even stupider than any stats-related claim I've seen you make, which is saying something. Insofar as that research plan required original thought, I'm sure such thought was supplied by your professor. You are incapable of analytically rigorous, original thought.

Proven? Proven! Ha! You haven't proven a single thing. Hey, you haven't even attempted to prove anything at all. No links, no logical discussion, nothing. Nothing but insults that is, for an endless number of pages. Do you actually think people on these boards are stupid enough to believe you know the first thing about what you've been discussing? Don't you think that at some point, they'll notice you haven't exhibited the slightest degree of understanding about statistics, or even about "heritability?" That all you're doing is aping the insults Bungee Jumper's been throwing at me--and aping very poorly I might add. Other than the fact that you're doing some grunt work for some hapless professor, you've given nobody even the slightest reason to take you seriously. Not when you backed up the idiotic statistical claims being made by syhuang. Not when you failed to understand the simple fact that people who obtain exceptionally high scores on tests are disproprortionately lucky (assuming there's measurement error). I haven't once seen you display an original thought, or an intelligent insight, or anything really. You make up for your clear lack of talent by throwing insults at anyone who disagrees with you. Just like a four year old.

Could your four year old cousin act as childishly and stupidly as you've been acting?

No need to thank him for incorrect information. Identical twins raised apart show very strong correlations in their I.Q.s--far higher than the 0.4 Bungee Jumper is claiming. In contrast, unrelated children raised together show no correlation between their I.Q.s--at least not by adulthood. Genetics play a far greater role in explaining differences in people's intelligence than does the environment.

On the contrary, the definition for lower-case h^2 said exactly what heritability is. You just conveniently "forgot" that little point.

I don't "think" I won the heritability debate. I know I won that debate. First, I showed you a website which gave specific, mathematical definitions of the word. The definition you were using was capital H^2. The definition I was using was lower-case h^2. When the American Psychological Association declared that heritability for intelligence was about 0.75, they specifically said that by "heritability" they meant lower-case h^2. I couldn't possibly have asked for a more clear-cut victory.

We've already had the "heritability" debate, which I won hands-down. Why you insist on bringing it up again is beyond me.

The correlation between parents' and childrens' I.Q.s is around 0.75; and it goes up as children get older. The correlation between a person's score between different I.Q. tests is somewhere between 0.8 and 0.9 (I don't remember where, specifically.)

Wrong from start to finish. This isn't something I "came up with." It's an observation of fact. Do the math. Create a Gaussian population. Give each member an I.Q. test, with a normally distributed error term with a mean of zero. Select out some subset of the population that did the best on the first test, and have them retested. Compare the group's average score for the first test, versus the second one. Any time you select a group of people who did the best on an I.Q. test, you're selecting a group that's disproportionately lucky on that test. The parents or children of this group will, on average, have been luck-neutral on the test; and will have had lower I.Q. scores.

New York State apparently feels it hasn't driven away enough good people yet, so now it's driving out the Amish. Terrible.

Merciful death? Are you kidding me? Every football discussion board should have its own politics section. Every politics section needs at least a ten page thread on regression toward the mean.

In this discussion, your attempts at rebuttals have never been particularly strong. But this is one of the weakest ones yet. One of the interesting things about the phenomenon I'm describing is that it works the other way too. Suppose you were to select children with the highest I.Q. scores. On average, their parents will tend to have somewhat lower scores. Why? Because in selecting children with the highest scores, you selected a group that was disproportionately lucky on the test. Their parents were, on average, luck-neutral; and hence obtained somewhat lower scores.

Suppose you have a man and a woman who each scored a 140 on an I.Q. test. Those two people are more likely to be lucky 130s than unlucky 150s. Let's say their expected score on a retest is 133. They decide to have kids, and the kids score an average of 133 on their I.Q. tests. On the surface, it seems like the kids are closer to the population mean than their parents. But that's not necessarily the case. In searching for the most intelligent parents, you selected those with the highest I.Q. scores--and hence, a group of people that was disproportionately lucky on the test. This test taking luck won't be passed onto the next generation, so the kids will score somewhat lower on their I.Q. tests; even if they're just as smart as their parents.

If I'm the White House press secretary, Ramius is the White House janitor.

This from someone who: a. Has not supplied any data with which to support his points. b. Has not supplied any links with which to support his points. c. Has not supplied any logical thought with which to support his points. d. Has not supplied any statistical insight with which to support his points. e. Has not supplied any statistical knowledge with which to support his points. You have, however, supplied plenty of insults with which to support your points, so I guess that cancels out the above.

The fact that you're still questioning the arbitrary threshold this late in the game is absolutely absurd. Of course you need an arbitrarily defined threshold. The point--which I've been trying to drive into your skull for quite some time now, is that people who do exceptionally well on their first I.Q. test tend to do somewhat less well on the second test. The arbitrary threshold is necessary to give a mathematical definition for "people who did exceptionally well on the first test." Or exceptionally badly. If 100 people got a 140 on an I.Q. test, and if those people are sitting down to retake it, the average score of that group will be less than 140. Yes, you need an arbitrarily-defined threshold (in this case, 140) to prove that. Why on earth you think the presence of such a threshold make this phenomenon not "real" is completely beyond me.

Unfortunately for your argument, you've yet to present a single meaningful datum for me to discard. True 130s outnumber true 150s. Therefore, in a group of people who scored 140 on an I.Q. test, lucky 130s will outnumber unlucky 150s. Have the group retake the test, and the group's average score will be closer to the mean than that original 140. What part of this don't you understand?

He's the perfect coach to quote. He's pithy, and often tells it like it is.

Well, if you're going to annoint him anyway, you may as well do it right. Don't pinch pennies on the cheap oil--get the good stuff. Wear proper robes or something. Properly plan and execute the ceremony. Be sure to send a DVD of said ceremony to Bill Parcells, just to get a reaction.

My statement that I'm right isn't a claim. It's an observation of fact. The fact that you're accusing me of flip-flopping shows a complete absence of even a rudimentary understanding of my post. What a surprise. I'll spell things out more clearly, in terms hopefully even you can understand. Suppose you take a group of people who scored a 140 on an I.Q. test, and ask them to retake the test. That group will have a greater number of lucky 130s than unlucky 150s. You'd expect a lucky 130 to have his or her error term regress to zero on the retest. In other words, if your real I.Q. is 130, and if you got lucky and scored a 140 on the first test, your expected score on the retest is 130. The unlucky 150 is expected to score a 150 on the retest. Because individual people's error terms are expected to regress to a mean error of zero upon being retested, and because there are more lucky 130s than unlucky 150s, those who got a 140 on an I.Q. test will tend (on average) to score lower than 140 upon retaking the test.

You make a convincing case that Losman has been playing well. But don't get the annointing oil out just yet.

Have you ever tried shooting yourself in the head with a .45? Didn't think so. Get back to me when you have some firsthand experience in that area, and then we'll talk.

Make your face redder too, just to have all the bases covered.