Err America files Chapter 11

[Bungee Jumper]

uh...you're BJ (AKA CTM, DCT, etc...)

832087[/snapback]

 

I am? Oh, yeah...

[EC-Bills]

uh...you're BJ (AKA CTM, DCT, etc...)

832087[/snapback]

 

Maybe Tom's uncomfortable with the use of T-Bone and BJ in the same sentence?

[Bungee Jumper]

discuss:  dp x dx > h / (2 x pi) = Planck's constant / (2 x pi)...

831911[/snapback]

 

And that doesn't cause regression toward the mean, either.

 

But what the !@#$ do I know?

[Nervous Guy]

Maybe Tom's uncomfortable with the use of T-Bone and BJ in the same sentence? 

832095[/snapback]

 

he lacked forethought of initials...

[Nervous Guy]

And that doesn't cause regression toward the mean, either.

 

But what the !@#$ do I know?

832098[/snapback]

what about "measurement error"?

[Bungee Jumper]

no not unclear...4 dice example would be better.

 

and BTW I'm actually 76.004124 inches...close enough though....

831908[/snapback]

 

100 dice would be better...but 2 illustrates a probability distribution well enough without having to figure binomial factors like (100!)/(63!)(47!)

 

BTW...tell T-Bone I respect his scientific ability, so he's allowed to agree with me. He's just a loser in every other aspect of his life.

[Bungee Jumper]

 

 

he lacked forethought of initials...

832099[/snapback]

 

I was just regressing toward the mean...

[Orton's Arm]

Poor choices of words, again?

Had you understood the underlying phenomenon, you would have known that in the examples you cited my choice of words was not poor.

[Orton's Arm]

And that's a perfect example of regression toward the mean...of the friggin' error.  You take a test, you score X.  The test has no error.  You take it again, you score X.  If the test has error, normally distributed with a standard deviation of s, you have roughly a 68% chance of scoring within s of X, a 95% chance of scoring within 2s of X...

 

Now postulate two different but related tests measuring the same thing, one exact and one with inherent normally distributed error.  On the exact test you score X.  You take the "errored" one.  You score X + 2s.  You take it again.  There is a 95% chance that you will score within 2s of X...which means there's a 98% chance that you will score less than X + 2s.  This is because the error is normally distributed with an associated probability of occurrence.  It is not because X is normally distributed.  It's not.  It's fixed by the first test.  The regression toward the mean is the regression toward the mean of the normally distributed paramater - the error.

 

It's as I've been saying; he doesn't know what he's actually measuring.  He set up this wacky simulation, measured the regression of a variable, then proceeded assign said regression to the wrong !@#$ing measurable.  THAT'S why regression disappears if he eliminates error...because when he eliminates error, he eliminates the regression of such.  It's because he can't even begin to understand that he's confusing two different variables that he thinks error is "causing" regression.

832090[/snapback]

I agree with all of your first paragraph, and large portions of your second. Suppose you could magically look beneath the surface to see someone's true I.Q. A man with a true I.Q. of 150 sits down to take an I.Q. test. He gets unlucky on the test, and scores a 140. Now you watch him sit down to take a second I.Q. test. If you had to guess in advance what his second score will be, the best guess is 150; because that's his true I.Q. The error term, as you point out, is expected to regress toward its mean of zero.

 

But in the real world, there is no way of magically looking beneath the surface to see someone's true I.Q. The man who scored a 140 on the first test might be an unlucky 150, a luck-neutral 140, or a lucky 130. The unlucky 150s who retake the test are expected to, on average, get a 150 the second time around. The luck-neutral 140s who retake the test are expected to get an average of 140 the second time. And the lucky 130s are, on average, expected to get a 130 the second time they take the test.

 

Suppose the underlying population had equal numbers of true 130s, 140s, and 150s. Someone who scored a 140 on an I.Q. test could just as easily be an unlucky 150 as a lucky 130. If someone scored a 140 the first time they took the test, their expected score the second time would be 140. This is because the 140 score is just as likely to indicate an unlucky 150 as it is to indicate a lucky 130.

 

Now imagine there are 100 times as many true 130s as there are true 150s. Therefore, a score of 140 is 100 times more likely to signal a lucky 130, than it is to signal an unlucky 150. Imagine a room with people who scored a 140 on an I.Q. test. There are 100 lucky 130s, one unlucky 150, and a number of luck-neutral 140s. The average score for someone in this room the second time around will be less than 140. This is another way of saying that someone who scores a 140 on an I.Q. test the first time around will generally score somewhat lower the second time.

[Bungee Jumper]

I agree with all of your first paragraph, and large portions of your second. Suppose you could magically look beneath the surface to see someone's true I.Q. A man with a true I.Q. of 150 sits down to take an I.Q. test. He gets unlucky on the test, and scores a 140. Now you watch him sit down to take a second I.Q. test. If you had to guess in advance what his second score will be, the best guess is 150; because that's his true I.Q. The error term, as you point out, is expected to regress toward its mean of zero.

 

But in the real world, there is no way of magically looking beneath the surface to see someone's true I.Q. The man who scored a 140 on the first test might be an unlucky 150, a luck-neutral 140, or a lucky 130. The unlucky 150s who retake the test are expected to, on average, get a 150 the second time around. The luck-neutral 140s who retake the test are expected to get an average of 140 the second time. And the lucky 130s are, on average, expected to get a 130 the second time they take the test.

 

Suppose the underlying population had equal numbers of true 130s, 140s, and 150s. Someone who scored a 140 on an I.Q. test could just as easily be an unlucky 150 as a lucky 130. If someone scored a 140 the first time they took the test, their expected score the second time would be 140. This is because the 140 score is just as likely to indicate an unlucky 150 as it is to indicate a lucky 130.

 

Now imagine there are 100 times as many true 130s as there are true 150s. Therefore, a score of 140 is 100 times more likely to signal a lucky 130, than it is to signal an unlucky 150. Imagine a room with people who scored a 140 on an I.Q. test. There are 100 lucky 130s, one unlucky 150, and a number of luck-neutral 140s. The average score for someone in this room the second time around will be less than 140. This is another way of saying that someone who scores a 140 on an I.Q. test the first time around will generally score somewhat lower the second time.

832566[/snapback]

 

You're a !@#$ing retard. My post was not an opinion for you to agree or disagree with . It was mathematical fact.

 

You can stop trying to explain yourself already. I understand perfectly what you're trying to say. Have for about 20 pages. You're just an idiot.

[erynthered]

You're a !@#$ing retard.  My post was not an opinion for you to agree or disagree with .  It was mathematical fact.

 

You can stop trying to explain yourself already.  I understand perfectly what you're trying to say.  Have for about 20 pages.  You're just an idiot.

832633[/snapback]

 

 

I'm on it. Next post he does, this is what I say........

 

 

You're a !@#$ing retard.  My post was not an opinion for you to agree or disagree with .  It was mathematical fact.

 

You can stop trying to explain yourself already.  I understand perfectly what you're trying to say.  Have for about 20 pages.  You're just an idiot.

[erynthered]

I see that you strongly disagree with me, yet are making the effort to be civil. I respect that.

 

To move onto your example, an aptitude test would have understated the intellectual abilities of the Italian immigrants you describe, both because of their lack of schooling and due to language difficulties. The fact they weren't in school very much probably was due to a lack of opportunity, rather than to anything driven by genetics.

 

I myself know of cases where children of immigrants dug themselves out of poverty. Typically they did so through determination, as well as through having the genetic potential to achieve intellectually challenging tasks. I respect such people both for their drive and for their intelligence.

 

A eugenics program would encourage smart people to have more children than less intelligent people do. But it wouldn't address the issue of work ethic/desire, either in a positive or a negative way. I'd like to address it, but I don't see how an institution can measure desire. In this case I'm settling for half a loaf by having programs increase intelligence, while neither raising nor lowering the average level of desire.

 

America's present childbearing incentives encourage lazy people to have more children than more ambitious people do. Suppose there are two women with the same below-average I.Q. One is lazy, the other ambitious. Which do you think will be more likely to go on welfare? Considering the government encourages welfare recipients to have more kids, this represents a problem. If work ethic is genetic, it's a problem. If it's environmental it's a problem, because the natural role models for these kids (their parents) are lazy.

811656[/snapback]

 

 

 

 

From page 8:

 

My (helped) response.....

 

 

 

You're a !@#$ing retard. My post was not an opinion for you to agree or disagree with . It was mathematical fact.

 

You can stop trying to explain yourself already. I understand perfectly what you're trying to say. Have for about 8 pages. You're just an idiot.

 

 

Its all the same, no?

[Ramius]

From page 8:

 

My (helped) response.....

You're a !@#$ing retard.  My post was not an opinion for you to agree or disagree with .  It was mathematical fact.

 

You can stop trying to explain yourself already.  I understand perfectly what you're trying to say.  Have for about 8 pages.  You're just an idiot.

Its all the same, no?

832636[/snapback]

 

For some reason, holcombs arm thinks if he keeps typing the wrong sh--, its going to eventually become correct.

[Bungee Jumper]

For some reason, holcombs arm thinks if he keeps typing the wrong sh--, its going to  eventually become correct. 

832680[/snapback]

 

Well, you see...if you keep repeating incorrect nonsense over and over, eventually the error will cause the nonsense to regress toward the true value of what you're trying to say...

[Chilly]

For some reason, holcombs arm thinks if he keeps typing the wrong sh--, its going to  eventually become correct. 

832680[/snapback]

 

Isn't that the definition of insanity?

[Orton's Arm]

You're a !@#$ing retard.  My post was not an opinion for you to agree or disagree with .  It was mathematical fact.

 

You can stop trying to explain yourself already.  I understand perfectly what you're trying to say.  Have for about 20 pages.  You're just an idiot.

832633[/snapback]

For some unfathomable reason, I was trying to be polite. This time around I'll be more clear: your earlier post demonstated a fundamental incomprehension of the principle of regression toward the mean. You just don't get it. If statistics was astronomy, you'd be Ptolemy. You will continue to fail to understand this concept until you become a little less arrogant, and a little more willing to read and understand either my posts, the articles to which I've linked, or both. Maybe you could even do a little simulation of your own to prove it to yourself. Create a population with normally distributed, randomly assigned I.Q.s. Give each person an I.Q. test based on their true I.Q. plus an element of luck when taking the test. Retest the ones who did the best the first time around, and see how their scores change.

 

I've led you to water. Instead of drinking, you've spat it in my face. Die of thirst then. I've done what I can.

[Ramius]

I've led you to water. Instead of drinking, you've spat it in my face. Die of thirst then. I've done what I can.

832745[/snapback]

 

No. You claim you are leading us to water, but the rest of us can see its just a mirage. You're the only one that actually thinks there's water out there.

[syhuang]

Had you understood the underlying phenomenon, you would have known that in the examples you cited my choice of words was not poor.

832550[/snapback]

You words are not poor only when you OMIT the fact that regression toward mean is NOT caused by error.

 

Regression toward mean is causd by sample and error being normally distributed.

[EC-Bills]

And just when you thought this fire was beginning to wane...

[Bungee Jumper]

For some unfathomable reason, I was trying to be polite. This time around I'll be more clear: your earlier post demonstated a fundamental incomprehension of the principle of regression toward the mean. You just don't get it. If statistics was astronomy, you'd be Ptolemy.

 

I'm published in astronomy, too. An statistical analysis of galactic rotation.

 

You will continue to fail to understand this concept until you become a little less arrogant, and a little more willing to read and understand either my posts, the articles to which I've linked, or both. Maybe you could even do a little simulation of your own to prove it to yourself.

 

I understand you. You're wrong. I've explained it in great detail why you're wrong. You simply can't accept the fact that you don't know sh--.

 

Create a population with normally distributed, randomly assigned I.Q.s. Give each person an I.Q. test based on their true I.Q. plus an element of luck when taking the test. Retest the ones who did the best the first time around, and see how their scores change.

 

And you will see regression OF THE ERROR TOWARD THE MEAN ERROR. There's a simple way to prove it: Instead of retesting the ones that "do best", retest the ones that show the most difference from their "true IQ" (i.e. if your error has a standard deviation of 6, retest those that are 9 or more away from their "true IQ"). That difference will decrease "on average", because the ERROR regresses, because that is what your simulation is actually measuring.

 

Conversely, you could take the error completely out of your simulation. Test your population, then test them again. "On average", people who score very well/poorly will score less well/poorly even in the absence of error, because chance (i.e. the probability distribution of a normal distribution) dictates it.

 

Alternately...take a pair of dice - oh, no, wait, we've already established that you honestly think there's error in dice that cause regression toward the mean.

 

Alternately...stop playing with Excel and DO THE MATH. But you can't...you don't actually know how to do the math, do you? You couldn't even tell me what a gaussian integrated over all space is, or what that means, or why it's relevant to this discussion, could you? Can you "standard error" for me? What's the mathematical definition of "variance" in a continuous distribution?

 

I've led you to water. Instead of drinking, you've spat it in my face. Die of thirst then. I've done what I can.

832745[/snapback]

 

More like pesticide, really. It still amazes me that the entire scientific community can disagree with you...and you're right, and the rest of the world is wrong.