Orton's Arm Posted November 7, 2006 Share Posted November 7, 2006 No, not at all. You can't assume that the mean of normally distributed error, applied over multiple measurements, decreases. It doesn't, by definition...it's normally distributed. You're saying that your simulation invalidated its own initial parameters that you established. All that proves is that you didn't know what you were doing when you wrote it.Again, no...because you established as an initial fixed parameter normally distributed error. Measured IQs should over- and understate "real" IQs at the same rate. If they didn't...again, you !@#$ed up your simulation. What? Again, you're not even wrong. Ignoring the obvious bull sh-- in that paragraph...how the hell does the error in the test magically disappear in the second iteration? I can tell you why: because you don't know what you're doing. You're not measuring what you think you're measuring. Actually, they all think you've got oatmeal for brains, regardless of insightfulness. 827631[/snapback] The people in your workplace know only those things about me which you choose to tell them. Given that, their views are unsurprising. As for the rest of your post, I finally understand your thought process a little better. Suppose someone with, ah, a touch of overconfidence were to take a surface view of this situation, without giving it serious thought. You start off with a normally distributed error term, with a mean of zero. You're right in saying that the measured I.Q. for the population as a whole ought to be the same as its true mean. Further, you know that any given individual who sits down to take an I.Q. test is just as likely to get lucky as unlucky. Why, then, did the Monte Carlo simulation demonstrate people with measured I.Q.s above the threshold were, on average, not quite as smart as their test results indicated? Some people with true I.Q. values below the threshold value got lucky the first time they took the test, and were placed in the threshold group. Other people had true I.Q. scores above the threshold value, but got unlucky when they took the test. These people weren't placed in the Threshold group. In other words, some of the people who were eliminated were smarter than some others who were included. Now consider the second test for the Threshold group. Some of the people taking that test don't belong there, because their I.Q.s fall below the threshold value. These people will drag down the Threshold group's average I.Q. when that group gets retested. Meanwhile, you have people with above-threshold I.Q.s sitting there not being retested, because they were excluded from the Threshold group based on their bad luck the first time they took the intelligence test. Link to comment Share on other sites More sharing options...
Bungee Jumper Posted November 7, 2006 Share Posted November 7, 2006 The people in your workplace know only those things about me which you choose to tell them. Given that, their views are unsurprising. As for the rest of your post, I finally understand your thought process a little better. Suppose someone with, ah, a touch of overconfidence were to take a surface view of this situation, without giving it serious thought. You start off with a normally distributed error term, with a mean of zero. You're right in saying that the measured I.Q. for the population as a whole ought to be the same as its true mean. Further, you know that any given individual who sits down to take an I.Q. test is just as likely to get lucky as unlucky. Why, then, did the Monte Carlo simulation demonstrate people with measured I.Q.s above the threshold were, on average, not quite as smart as their test results indicated? Some people with true I.Q. values below the threshold value got lucky the first time they took the test, and were placed in the threshold group. Other people had I.Q. true I.Q. scores above the threshold value, but got unlucky when they took the test. These people weren't placed in the Threshold group. In other words, some of the people who were eliminated were smarter than others who were included. Now consider the second test for the Threshold group. Some of the people taking that test don't belong there, because their I.Q.s fall below the threshold value. These people will drag down the Threshold group's average I.Q. when that group gets retested. Meanwhile, you have people with above-threshold I.Q.s sitting there not being retested, because they were excluded from the Threshold group based on their bad luck the first time they took the intelligence test. 827669[/snapback] No, the simulation demonstrated that because, again, you don't understand it. You're still measuring error, not IQ. I can understand why you think you're measuring IQ...but you're not. You're treating IQ as a fixed parameter, and watching the error evolve around it. And - again - you still haven't done the math. Get a pencil and a piece of paper, and go through the equations. That should be much more difficult for you to screw up than a MC simulation. Link to comment Share on other sites More sharing options...
Orton's Arm Posted November 7, 2006 Share Posted November 7, 2006 Actually, as you admit, your example is extremely simplified. But as BJ points out, you have essentially set up your example so that the result you desire is what you will get out of your experiment. The model you put forth will not necessarily tell you anything about how children's IQ's are effected by their parents IQ's, nor will it tell you how much deviation in IQ from 1 to the other are due measurement of error or other sources and factors. 827656[/snapback] Bungee Jumper's criticisms of my model are invalid. My model is only designed to say one thing, but to say it very well: error in measurement can cause the appearance of regression toward the mean. The other factors you mentioned are important to a discussion about intelligence in general, but aren't really what we're arguing about right now. Link to comment Share on other sites More sharing options...
Orton's Arm Posted November 7, 2006 Share Posted November 7, 2006 No, the simulation demonstrated that because, again, you don't understand it. You're still measuring error, not IQ. I can understand why you think you're measuring IQ...but you're not. You're treating IQ as a fixed parameter, and watching the error evolve around it. And - again - you still haven't done the math. Get a pencil and a piece of paper, and go through the equations. That should be much more difficult for you to screw up than a MC simulation. 827690[/snapback] You will have to elaborate more if you want your post to have any meaning. Link to comment Share on other sites More sharing options...
Bungee Jumper Posted November 7, 2006 Share Posted November 7, 2006 You will have to elaborate more if you want your post to have any meaning. 827697[/snapback] If I have to elaborate more, you'll never understand it. You fixed everything but the error in your simulation, and you think one of your fixed parameters is what you're simulating. You're not. And no matter how many different ways I explain it, you just won't get it. Link to comment Share on other sites More sharing options...
Bungee Jumper Posted November 7, 2006 Share Posted November 7, 2006 Bungee Jumper's criticisms of my model are invalid. My model is only designed to say one thing, but to say it very well: error in measurement can cause the appearance of regression toward the mean. The other factors you mentioned are important to a discussion about intelligence in general, but aren't really what we're arguing about right now. 827693[/snapback] We're not arguing. You're pretending you know what you're talking about, and I'm laughing my ass off. Link to comment Share on other sites More sharing options...
Orton's Arm Posted November 7, 2006 Share Posted November 7, 2006 If I have to elaborate more, you'll never understand it. You fixed everything but the error in your simulation, and you think one of your fixed parameters is what you're simulating. You're not. And no matter how many different ways I explain it, you just won't get it. 827709[/snapback] You're wrong. I was simulating whether the existence of measurement error in an I.Q. test can cause the appearance of regression toward the mean. It can. Link to comment Share on other sites More sharing options...
Bungee Jumper Posted November 7, 2006 Share Posted November 7, 2006 You're wrong. I was simulating whether the existence of measurement error in an I.Q. test can cause the appearance of regression toward the mean. It can. 827728[/snapback] No, you were trying to. You failed miserably. Link to comment Share on other sites More sharing options...
meazza Posted November 7, 2006 Share Posted November 7, 2006 No, you were trying to. You failed miserably. 827742[/snapback] Wow, this thread is still going on. Link to comment Share on other sites More sharing options...
EC-Bills Posted November 7, 2006 Share Posted November 7, 2006 I understand. It's the government's job to protect us from bad lasagna... 827654[/snapback] Damn straight, I have been a victim of bad lasagna. Link to comment Share on other sites More sharing options...
Ramius Posted November 7, 2006 Share Posted November 7, 2006 But as BJ points out, you have essentially set up your example so that the result you desire is what you will get out of your experiment. 827656[/snapback] Gee, HA has never done this before in any statistical assessment. Link to comment Share on other sites More sharing options...
Ramius Posted November 7, 2006 Share Posted November 7, 2006 I just finished creating a Monte Carlo simulation. In this simulation, measurement error led to the appearance of regression toward the mean. In other words, I was right. Methodology 1. I created a population of 1000 members. The members were randomly assigned I.Q.s from a normal distribution with a mean of 100 and a standard deviation of 10. I set trueIQ =norminv(rand(),100,10) 2. To obtain a given population member's measured I.Q., I applied an error function to each member's true I.Q. The error function was random and normally distributed, with a mean value of zero and a standard deviation of 2.5. I set measured IQ =trueIQ+norminv(rand(),0,2.5) 3. I arbitrarily defined a threshold of 115. Those members with measured I.Q.s below the threshold were ignored. =if(measuredIQ>threshold,1,0) 4. Members with measured I.Q.s above the threshold were subjected to a second measurement. The second measurement was the original I.Q. plus a random, normally distriubted error function with a mean of zero and a standard deviation of 2.5. =trueIQ+norminv(rand(),0,2.5) 5. I compared the results that threshold members obtained from the second test with those obtained from the first test. I pressed F9 at least ten times. Each time, the average score for threshold members was worse the second time they "took the test" than it was the first time. In other words, those who did well on an I.Q. test appeared to regress toward the mean when tested a second time. The sole source for this regression toward the mean was measurement error. 827404[/snapback] WOW! i mean WOW! in between the fits of laughter, and going to the bathroom to clean my jeans (this pathetic attempt at a statistical simulation made me piss myself laughing), i managed to read this giant statistical cluster!@#$. How you took something as simple as a monte carlo simulation and managed to !@#$ it 40 ways from sunday is beyond me. Actually it isnt. Its right in line with all of your previous analyses. I'd gladly lay into it, but what you have done right and wrong has been pointed out numerous times in the previous few pages. My god you are stupid. Link to comment Share on other sites More sharing options...
Bungee Jumper Posted November 7, 2006 Share Posted November 7, 2006 WOW! i mean WOW! in between the fits of laughter, and going to the bathroom to clean my jeans (this pathetic attempt at a statistical simulation made me piss myself laughing), i managed to read this giant statistical cluster!@#$. How you took something as simple as a monte carlo simulation and managed to !@#$ it 40 ways from sunday is beyond me. Actually it isnt. Its right in line with all of your previous analyses. I'd gladly lay into it, but what you have done right and wrong has been pointed out numerous times in the previous few pages. My god you are stupid. 828177[/snapback] Technically, it's not even a Monte Carlo simulation. Though that's only a minor matter of verbage. Link to comment Share on other sites More sharing options...
Chef Jim Posted November 7, 2006 Share Posted November 7, 2006 Technically, it's not even a Monte Carlo simulation. Though that's only a minor matter of verbage. 828193[/snapback] Actually it's more of a Monte Cristo simulation if you will. Link to comment Share on other sites More sharing options...
Ramius Posted November 7, 2006 Share Posted November 7, 2006 You're wrong. I was simulating whether the existence of measurement error in an I.Q. test can cause the appearance of regression toward the mean. It can. 827728[/snapback] No, it cant. And why in the hell did you eliminate all IQ scores below 115? I am convinced that you think you can just eliminate any data, for whatever reason. And you wonder why your posts have no credibility around here. Link to comment Share on other sites More sharing options...
Orton's Arm Posted November 8, 2006 Share Posted November 8, 2006 No, it cant. And why in the hell did you eliminate all IQ scores below 115? I am convinced that you think you can just eliminate any data, for whatever reason. And you wonder why your posts have no credibility around here. 828463[/snapback] Wow! You managed to take a break from insulting me to ask a legitimate question! That's better than your usual level of self-control. Look at the group who scored 115 or better the first time they took the test. That group was selected mostly based on true intelligence, but also partly based on their luck when they first took the test. While lucky and unlucky members presumably balanced each other overall, the Threshold group contained a disproportionate number of lucky (on the first test) members, while the non-Threshold group contained a disproportionate number of people who got unlucky on the test. When the Threshold group was asked to take the test a second time, its average score was a little lower than it had been the first time. This second score was more indicative of the group's true I.Q. This example illustrates that those who obtain exceptionally high scores on I.Q. tests are, on average, a little less intelligent than their scores would indicate. Conversely, those who obtain exceptionally poor scores on I.Q. tests are a little smarter than their scores make them seem. Measurement error causes the population to appear to be more spread out than it really is. Link to comment Share on other sites More sharing options...
Bungee Jumper Posted November 8, 2006 Share Posted November 8, 2006 Wow! You managed to take a break from insulting me to ask a legitimate question! That's better than your usual level of self-control. Look at the group who scored 115 or better the first time they took the test. That group was selected mostly based on true intelligence, but also partly based on their luck when they first took the test. While lucky and unlucky members presumably balanced each other overall, the Threshold group contained a disproportionate number of lucky (on the first test) members, while the non-Threshold group contained a disproportionate number of people who got unlucky on the test. When the Threshold group was asked to take the test a second time, its average score was a little lower than it had been the first time. This second score was more indicative of the group's true I.Q. This example illustrates that those who obtain exceptionally high scores on I.Q. tests are, on average, a little less intelligent than their scores would indicate. Conversely, those who obtain exceptionally poor scores on I.Q. tests are a little smarter than their scores make them seem. Measurement error causes the population to appear to be more spread out than it really is. 828834[/snapback] That makes no sense. I'm not surprised, because it's you after all. But it makes no sense. You're using a defined normally distributed error to prove the error's not normally distributed. That is all kinds of !@#$ing stupid. Link to comment Share on other sites More sharing options...
GG Posted November 8, 2006 Share Posted November 8, 2006 Hey, I don't know much about statistics, but I know how to call bullcrap: ... Look at the group who scored 115 or better the first time they took the test. That group was selected mostly based on true intelligence, but also partly based on their luck when they first took the test. 828834[/snapback] Is the 115 score based on their "intelligence" or based on this: QUOTE(Holcombs_Arm @ Nov 6 2006, 07:34 PM)3. I arbitrarily defined a threshold of 115. Those members with measured I.Q.s below the threshold were ignored. =if(measuredIQ>threshold,1,0) So, what's so special about 115? Why not 125? 150? Geez, even when people who don't know statistics can pick holes in your argument it's time to let it go. Link to comment Share on other sites More sharing options...
Bungee Jumper Posted November 8, 2006 Share Posted November 8, 2006 Hey, I don't know much about statistics, but I know how to call bullcrap:Is the 115 score based on their "intelligence" or based on this: So, what's so special about 115? Why not 125? 150? Geez, even when people who don't know statistics can pick holes in your argument it's time to let it go. 829084[/snapback] In his defense (don't ask me why I'm defending him, I don't !@#$ing know), 115 would be 1.5 standard deviations away from the mean, which would mean he's safely selecting from within the "gifted" category. Which doesn't explain why he picked a standard deviation of 10 and a selection criteria of 115 out of his ass, rather than go with the standard 30 and 145 in a normal IQ measure. It's not really material to the math, since a gaussian's a gaussian's a gaussian...but it's pretty funny that, when attempting to measure something related to IQ, he can't even pick the correct numbers to describe it. Link to comment Share on other sites More sharing options...
IBTG81 Posted November 8, 2006 Share Posted November 8, 2006 And this thread still goes on... So, what has this proven? That HA knows his stats? ::ducks for cover:: Link to comment Share on other sites More sharing options...
Recommended Posts