Does Cognitive testing and training work?

[dann]

I tend to find Bpesta's obvious racial bias very disturbing:
http://www.internationalskeptics.com/forums/showthread.php?postid=3232120#post3232120
http://www.internationalskeptics.com/forums/showthread.php?postid=3232580#post3232580

 

[dann]

In the context, I think it's significant. The was a blind reviewer selected by the best journal in this subfield to critically review a submitted manuscript.

The comment was made as part of the review.

In this case I am not impressed by a reviewer from "the best journal in this subfield", and no matter how blind, I don't think that he would have any difficulties recognizing a quotation from Stephen Jay Gould.

About the controversy surrounding The Mismeasure of Man.

 

[bpesta22]

re: bell curves

Been awhile since I thought about this issue, so I looked it up.

There are practical reasons for making IQ tests that produce the BC and theoretical reasons to expect the population distribution of g is bell shaped.

Practical: One can determine the shape of the distribution just by weighting the amount of difficult versus easy items that appear on the test. For any item that exists, the test maker will have data on it. The most basic stat is the item's p value-- it's simply the proportion of test takers that get the item right.

With adaptive testing today, the first question we'd ask is one with a p value of .50 (exactly half the population gets it right). If you get it right, the next item might have a p value of .25 (only 25% of the population gets it right). If you get this one wrong, the next one might be p=.37; if you got it right instead, the next one might be p=.125.

By doing it this way we can pretty quickly hone in on your true IQ.

What does this have to do with bell curves? At a practical level, most IQ tests are not computer administered like in the example above. Instead, I just scatter the paper and pencil test with items of various p values. If your IQ is really high, lots of those items will be a waste of time; way too easy for you. The opposite is true if you're IQ is way too low.

To create the biggest market possible for your IQ test, you want the p values to produce a normal distribution in the population. So, most IQ tests are structured to measure the fat part of the curve, because that's where the vast majority of people are re IQ.

In fact, I remember making this point before, but a typical IQ test is likely very good at distinguishing middle scorers from each other (the difference between a 98 and a 104 is likely meaningful) but very poor at distinguishing people at either end from each other (the difference between a 140 and 150 is likely not meaningful).

So, the practical reason to force the IQ test into producing a bell curve is to maximize the number of people the test can differentiate (rank basically if Claus will let me use that word).

I'm not sure if this point is correct, but I suspect the way to do it would be to just create a bell curve of items based on their p values. So, you have lots of p=.50 items; many p=.25 and p = .75 items and fewer p = .125 and p = .875 items.

This by definition I think would ensure that IQ scores are normally distributed.

The theoretical reasons are more complex. One is the central limit theorem, the other appeals to Item Response Theory. I've posted a bit about IRT here before. The explanation for why it predicts normality is something I don't want to spend time writing about here. Jensen covers it well in the g factor (pp 100-104, but read only if you really, really want the answer to the question) and here's a wiki link to the basics of IRT:

http://en.wikipedia.org/wiki/Item_response_theory



Claus; it's getting hard to follow the replies and re-replies. If you would produce a classic larsen list o questions that you think haven't been addressed, I'll try to answer them. If you think further debate here is futile, that's cool too.

 

[CFLarsen]

re: bell curves

Been awhile since I thought about this issue, so I looked it up.

There are practical reasons for making IQ tests that produce the BC and theoretical reasons to expect the population distribution of g is bell shaped.

Practical: One can determine the shape of the distribution just by weighting the amount of difficult versus easy items that appear on the test. For any item that exists, the test maker will have data on it. The most basic stat is the item's p value-- it's simply the proportion of test takers that get the item right.

With adaptive testing today, the first question we'd ask is one with a p value of .50 (exactly half the population gets it right). If you get it right, the next item might have a p value of .25 (only 25% of the population gets it right). If you get this one wrong, the next one might be p=.37; if you got it right instead, the next one might be p=.125.

By doing it this way we can pretty quickly hone in on your true IQ.

What does this have to do with bell curves? At a practical level, most IQ tests are not computer administered like in the example above. Instead, I just scatter the paper and pencil test with items of various p values. If your IQ is really high, lots of those items will be a waste of time; way too easy for you. The opposite is true if you're IQ is way too low.

To create the biggest market possible for your IQ test, you want the p values to produce a normal distribution in the population. So, most IQ tests are structured to measure the fat part of the curve, because that's where the vast majority of people are re IQ.

In fact, I remember making this point before, but a typical IQ test is likely very good at distinguishing middle scorers from each other (the difference between a 98 and a 104 is likely meaningful) but very poor at distinguishing people at either end from each other (the difference between a 140 and 150 is likely not meaningful).

So, the practical reason to force the IQ test into producing a bell curve is to maximize the number of people the test can differentiate (rank basically if Claus will let me use that word).

I'm not sure if this point is correct, but I suspect the way to do it would be to just create a bell curve of items based on their p values. So, you have lots of p=.50 items; many p=.25 and p = .75 items and fewer p = .125 and p = .875 items.

This by definition I think would ensure that IQ scores are normally distributed.

The theoretical reasons are more complex. One is the central limit theorem, the other appeals to Item Response Theory. I've posted a bit about IRT here before. The explanation for why it predicts normality is something I don't want to spend time writing about here. Jensen covers it well in the g factor (pp 100-104, but read only if you really, really want the answer to the question) and here's a wiki link to the basics of IRT:

http://en.wikipedia.org/wiki/Item_response_theory

Don't you see what you are doing? You are designing the whole concept of IQ to produce a bell curve, based on the assumption that it is a bell curve.

You don't measure and plot the data. You plot the data and then claim it is a measurement. There is no way the bell curve can ever be proven wrong! It can't be falsified, because you are throwing out questions that would wreck the bell curve.

What do we normally say about theories that cannot be falsified? They are not
s _ i _ _ _ _ _ _ c (fill out the word for 5 IQ points).

Claus; it's getting hard to follow the replies and re-replies. If you would produce a classic larsen list o questions that you think haven't been addressed, I'll try to answer them. If you think further debate here is futile, that's cool too.

Delighted to:

• What lab of yours are you talking about?
  
  
• Given that you test university students, why do you expect IQ to be a normal distribution there?
  
  
• Why is IQ assumed to be a normal distribution?
  
  
• Why do you expect not to be questioned, especially about such a controversial field?
  
  
• Do you think that Radin and Schwartz should be allowed to brush Randi off, just because they are scientists and he is not?
  
  
• If 29 % is from 1950 and later, but you accept data from 13 years back, just when does the data become solid and sound?
  
  
• Are you saying that the conclusions in The Bell Curve are correct?
  
  
• Can you please ask the mods to edit your post #53?
  
  
• When do we teach someone rocket science?
  
  
• Do we see an increase in average IQ?

 

[Jeff Corey]

Claus, there is no such thing as the perfectly symmetrical bell curve for IQ scores, as BP knows. I had to test a number of severely developmentally disabled ("retarded") people at one point. There is a huge hump down at the low end , representing people who cannot be tested with the tests like the Stanford-Binet or the Weschler. If you did test them with those tests, you would get scores of 0. So we were told to use the Vineland test, devised by Goddard (see Gould for him). First question, "Please put this block on top of the other block." Some still scored 0.
So the bell curve might look like the upper outline of a Brontosaurus grazing, head facing left toward zero.
Note I say IQ score, not "intelligence". They are not the same thing.

 

[CFLarsen]

Claus, there is no such thing as the perfectly symmetrical bell curve for IQ scores, as BP knows.

Not in real life, no. But the idea behind the Bell Curve is that IQ is distributed on one such - that for every dimbo there is a brainiard, and that the average guy centers around 100.

I had to test a number of severely developmentally disabled ("retarded") people at one point. There is a huge hump down at the low end , representing people who cannot be tested with the tests like the Stanford-Binet or the Weschler. If you did test them with those tests, you would get scores of 0. So we were told to use the Vineland test, devised by Goddard (see Gould for him). First question, "Please put this block on top of the other block." Some still scored 0.
So the bell curve might look like the upper outline of a Brontosaurus grazing, head facing left toward zero.
Note I say IQ score, not "intelligence". They are not the same thing.

Precisely. Because what is "intelligence"? There's "innate cognitive ability". But then, there's "thinking rationally" (hmmm....), "practical sense" (does that mean that plumbers score 200?), "to deal effectively with the environment" (Paris Hilton must be a genius!), "problem solving" (Rainman, you're off the charts!). New definitions, more or less vague, are invented discovered all the time.

I love Turing's: "To respond like a human being". That's saying that the higher IQ, the more of a human being you are. Great.

The problem is that you can use whatever you want, depending on the situation. The only thing that matters is that you get that darn bell curve at the end of the day.

It's like watching astrologers: Oops, here we have a new planet, let's include that. What about some of the asteroids? Use that - if you want to. Is it OK to use imaginary planets, like Vulcan? Sure! But you have to be able to say: "The horoscope was right! Astrology works..."

 

[bpesta22]

Not in real life, no. But the idea behind the Bell Curve is that IQ is distributed on one such - that for every dimbo there is a brainiard, and that the average guy centers around 100.



Precisely. Because what is "intelligence"? There's "innate cognitive ability". But then, there's "thinking rationally" (hmmm....), "practical sense" (does that mean that plumbers score 200?), "to deal effectively with the environment" (Paris Hilton must be a genius!), "problem solving" (Rainman, you're off the charts!). New definitions, more or less vague, are invented discovered all the time.

I love Turing's: "To respond like a human being". That's saying that the higher IQ, the more of a human being you are. Great.

The problem is that you can use whatever you want, depending on the situation. The only thing that matters is that you get that darn bell curve at the end of the day.

It's like watching astrologers: Oops, here we have a new planet, let's include that. What about some of the asteroids? Use that - if you want to. Is it OK to use imaginary planets, like Vulcan? Sure! But you have to be able to say: "The horoscope was right! Astrology works..."

The problem is, anytime you develop a test to measure these different and indpendent aspects of intelligence, the scores all inter-correlate.

Worse, the unique intelligence each test is supposed to measure doesn't predict anything. The variance caused by the tests' inter-correlations, however, (i,.e., variance due to g) predicts everything.

Ask Sternberg or Gardner how much success they've had developing independent measures of their multiple intelligences. Fail.

More later; got real world demands to do today

 

[bpesta22]

It's likely this was a dead thread-- only bumping it because I said I would address the Larsen List (tm)


* What lab of yours are you talking about?

I'm not sure the word I'm looking for-- I thought it was "euphemism," but that doesn't seem right. Lab could mean a physical place with beakers and mad scientists. I think when social psychologists use the term, they mean in whatever room they collected data in (together with whatever equipment they used to collect it).

So my lab this year has been classrooms where I've ran about 900 WPT IQ tests, and a university computer lab where I had 4 computers with software programmed to run my ECTs and collect data.

I didn't mean to imply anything with the word "lab" except to say that I have collected a fairly large amount of data over the past year.

* Given that you test university students, why do you expect IQ to be a normal distribution there?

College students just represent a more restricted range of the population, but one can still expect, and get a bull curve (as I did in the graph posted above). With range restriction, though, one should get a smaller standard deviation. The population SD for the WPT is 7.0. I get about 5.0 for the data I've collected.

I don't know that I could adequately explain central limit theorem here to you, or that it would be worth the few hours it would take me to translate IRT into a post here that makes sense to you. It's not at all controversial in psychometrics to assume normal distributions (doesn't matter what we're measuring; the same applies for personality tests too).

If you think that the bell curve assumption somehow allows you to completely dismiss IQ tests, then you must also toss most personality tests (even sophisticated ones like the MMPI) as they do the same thing.

I know of only one example where social science measures something and it's not a bell curve: a reaction time distribution.

Plus, I like my school, but it's not Harvard. We have a good mix of students and a diverse student body. We are a medium to large sized urban university. I think our students are fairly representative of "real people" (why do we always assume students are not?), and I suspect my results generalize back to the whole population, at least in the fat part of the curve. That said, I'm pretty sure the graph I posted above was based on only white students (gender and race are confounded at my school-- I don't know why, but most white business students are male, whereas, most black students-- by a large ratio-- are female).


* Why is IQ assumed to be a normal distribution?


See above. The typical IQ test does more than make the assumption; it forces the test to produce a bell curve. There is an answer to this question (whether you buy it or not is up to you) and it's in the 4-6 pages above I cited by Jensen in the "g factor".



* Why do you expect not to be questioned, especially about such a controversial field?

I've been posting about IQ here for something like 7 years. I expect to be questioned. I'm just annoyed that people who pride themselves about being skeptics will completely throw out massive amounts of science re IQ to support their pre-conceived notions. We stick to the science like bible for any other topic here, yet we completely disregard it for IQ-- even generate conspiracy theories about the whole field being racist, and then smugly assume that a scientist in some distant other field can step in and clear things up for the dummies who've devoted the last 30 years of their life to researching the issue. This is one of the reasons why Gould failed. He wasn't trained in the area (he obviously has no clue about factor analysis). He brought a knife to a gun fight.


* Do you think that Radin and Schwartz should be allowed to brush Randi off, just because they are scientists and he is not?

No. But in general, I'd trust the scientists who spent their lives researching the area over the outsider who cries foul, with the caveat that show me the data is always the driving force behind who is right or wrong.

* If 29 % is from 1950 and later, but you accept data from 13 years back, just when does the data become solid and sound?

When it's replicated, theoretically meaningful, fruitful (generates new research) and explains. The stuff gould focuses on are the dead ends in the field that no one today reads or cares about. Fine, but to claim that killing a corpse also invalidates contemporary research is stupid.
Plus, ironically, the brain size - IQ correlation that Gould features as pseudoscience has now been well established in the recent literature.

* Are you saying that the conclusions in The Bell Curve are correct?

Haven't read it in about 10 years (?). I think the data patterns they report exist in abundance elsewhere and are empirical facts. People differ in IQ. Groups do too. So do races. These are mean differences with overlapping distributions. But, given the predicative validity of g, the differences will translate into strong group differences also on measures of "success in life".

I think the data patterns in the bell curve are "correct". It's the interpretation that's tricky. I do believe race has a biological basis, and I do believe that environments have little to do with group differences on g. But I think more data are needed, and I personally would need to gain more expertise in genetics before I could increase the confidence level of what I just claimed.



* Can you please ask the mods to edit your post #53?
Ok

* When do we teach someone rocket science?

Don't know-- you ask me to draw a line in the sand. I can't. It doesn't mean I cant make a reasoned guess about cases that wouldn't be close to any line I would draw (hence my claim that teaching a 85 rocket science probably would be unwise).

I can't give any finer grain distinction than that, and I am not endorsing that we bar people from educational opportunities based on their IQ scores.

If you press me for an answer here, then I would use an economic's perspective and say: educate people until the marginal costs of the education exceed the marginal benefit the person gets from that education (the equilibrium point would be determined largely by g).



* Do we see an increase in average IQ?

IQ test scores; yes. Gains in g; no.

I will likely edit this a few times now quickly-- for typos

 

[aries]

Ok, to get this back somewhat on topic, I have to say that I have been a teacher for about 10 years now (though currently unemployed ). During these 10 years of teaching I have found that some students will do better if they hear or draw something while I'm talking to them. So I make sure that in my preparation I always include assignments that reaches these students as well. I'll also try to get the students involved in what they're about to learn physically, so they don't sit all day long. Sometimes say, in a math lesson, I will let them do some practical math, like measuring the school yard or their class room or the like. Most students actually enjoy this, I think -also those whose learning styles are more wordy or logical oriented.

Please remember that the above of course not is a scientific test; it is just observations of mine, based on about 10 years of teaching. (actually it is 9½ years, if we are to be picky). So to the OP: Yes, there are students who learn through either a visual or auditory learning style...

 

[Jeff Corey]

I think that actively engaging students in measuring rooms or round tables and calculating the area would be great.

 

[CFLarsen]

The problem is, anytime you develop a test to measure these different and indpendent aspects of intelligence, the scores all inter-correlate.

If you fit the questions to match the curve for each aspect, no wonder!

Worse, the unique intelligence each test is supposed to measure doesn't predict anything. The variance caused by the tests' inter-correlations, however, (i,.e., variance due to g) predicts everything.

Waaaait a second. Are you saying that there is no correlation between young children with a high IQ and those who later end up with long educations?

It's likely this was a dead thread-- only bumping it because I said I would address the Larsen List (tm)

If it is dead, it was because it was waiting for you to answer the Larsen List you asked for.

I'm not sure the word I'm looking for-- I thought it was "euphemism," but that doesn't seem right. Lab could mean a physical place with beakers and mad scientists. I think when social psychologists use the term, they mean in whatever room they collected data in (together with whatever equipment they used to collect it).

So my lab this year has been classrooms where I've ran about 900 WPT IQ tests, and a university computer lab where I had 4 computers with software programmed to run my ECTs and collect data.

I didn't mean to imply anything with the word "lab" except to say that I have collected a fairly large amount of data over the past year.

If you have done this to conduct a serious study, then you could call your classroom a "lab". If not, then I think you slightly overstated it by calling it a "lab".

College students just represent a more restricted range of the population, but one can still expect, and get a bull curve (as I did in the graph posted above). With range restriction, though, one should get a smaller standard deviation. The population SD for the WPT is 7.0. I get about 5.0 for the data I've collected.

I don't know that I could adequately explain central limit theorem here to you, or that it would be worth the few hours it would take me to translate IRT into a post here that makes sense to you. It's not at all controversial in psychometrics to assume normal distributions (doesn't matter what we're measuring; the same applies for personality tests too).

It's not a question of explaining the theory. Just explain why you don't get an overall higher IQ score.

You can't have it both ways. Unless - of course - you argue that IQ is irrelevant to the intelligence required to get a higher education.

If you think that the bell curve assumption somehow allows you to completely dismiss IQ tests, then you must also toss most personality tests (even sophisticated ones like the MMPI) as they do the same thing.

There's a crucial difference: Personality tests say that you are either introvert or extrovert (as an example, very simplified), but they don't say whether it is better to be one or the other.

IQ tests rank people: The higher you score on an IQ test, the better you are.

I know of only one example where social science measures something and it's not a bell curve: a reaction time distribution.

Plus, I like my school, but it's not Harvard. We have a good mix of students and a diverse student body. We are a medium to large sized urban university. I think our students are fairly representative of "real people" (why do we always assume students are not?), and I suspect my results generalize back to the whole population, at least in the fat part of the curve. That said, I'm pretty sure the graph I posted above was based on only white students (gender and race are confounded at my school-- I don't know why, but most white business students are male, whereas, most black students-- by a large ratio-- are female).

As long as you have some form of entry level limit, you can't expect all students to exhibit a Bell curve. Again, unless you argue that IQ is irrelevant to the intelligence required to get a higher education.

See above. The typical IQ test does more than make the assumption; it forces the test to produce a bell curve.

So, the answer the the question "Why is IQ assumed to be a normal distribution?"

is

"It forces the test to produce a bell curve"

?

Correct me if I'm wrong, but isn't that circular reasoning? It does so because it does so?

There is an answer to this question (whether you buy it or not is up to you) and it's in the 4-6 pages above I cited by Jensen in the "g factor".

I'm sure you can express it in your own words.

I've been posting about IQ here for something like 7 years. I expect to be questioned. I'm just annoyed that people who pride themselves about being skeptics will completely throw out massive amounts of science re IQ to support their pre-conceived notions. We stick to the science like bible for any other topic here, yet we completely disregard it for IQ-- even generate conspiracy theories about the whole field being racist, and then smugly assume that a scientist in some distant other field can step in and clear things up for the dummies who've devoted the last 30 years of their life to researching the issue. This is one of the reasons why Gould failed. He wasn't trained in the area (he obviously has no clue about factor analysis). He brought a knife to a gun fight.

We don't "stick to the science like bible". We use science to separate the wheat from the chaff. We use the exact same reasoning with IQ as we would with any other subject.

We hear precisely the argument as yours above from any other person who wants their particular pet belief to be accepted as real science, with real scientific evidence. Look at this:

I've been posting about MMR vaccines causing autism here for something like 7 years. I expect to be questioned. I'm just annoyed that people who pride themselves about being skeptics will completely throw out massive amounts of science re MMR vaccines causing autism to support their pre-conceived notions. We stick to the science like bible for any other topic here, yet we completely disregard it for MMR vaccines causing autism-- even generate conspiracy theories about the whole field being a conspiracy, and then smugly assume that a scientist in some distant other field can step in and clear things up for the dummies who've devoted the last 30 years of their life to researching the issue. This is one of the reasons why Barrett failed. He wasn't trained in the area (he obviously has no clue about factor analysis). He brought a knife to a gun fight.

Would you be convinced by such an argument?

No. But in general, I'd trust the scientists who spent their lives researching the area over the outsider who cries foul, with the caveat that show me the data is always the driving force behind who is right or wrong.

It isn't just the data, but how the data was found. Your IQ measuring stick will give you the result you expect, because it is rigged to do so.

When it's replicated, theoretically meaningful, fruitful (generates new research) and explains. The stuff gould focuses on are the dead ends in the field that no one today reads or cares about. Fine, but to claim that killing a corpse also invalidates contemporary research is stupid.
Plus, ironically, the brain size - IQ correlation that Gould features as pseudoscience has now been well established in the recent literature.

No, no, no. You miss my point. You dismissed Gould's numbers because they were too old.

So at what time in history did IQ data become solid and sound? It happened somewhere between 1950 and 1995 - but when?

Haven't read it in about 10 years (?).

Go back and check, please. I'd like to know what you think.

I think the data patterns they report exist in abundance elsewhere and are empirical facts. People differ in IQ. Groups do too. So do races. These are mean differences with overlapping distributions. But, given the predicative validity of g, the differences will translate into strong group differences also on measures of "success in life".

No matter how much you mix garbage, you won't end up with a jewel.

I think the data patterns in the bell curve are "correct". It's the interpretation that's tricky. I do believe race has a biological basis

If you also believe that races differ in IQ, then you must be saying that blacks are less intelligent than whites because they were born that way.

Don't know-- you ask me to draw a line in the sand. I can't. It doesn't mean I cant make a reasoned guess about cases that wouldn't be close to any line I would draw (hence my claim that teaching a 85 rocket science probably would be unwise).

I can't give any finer grain distinction than that, and I am not endorsing that we bar people from educational opportunities based on their IQ scores.

Back in post #45, you said the exact opposite:

Should we teach a 70 iq rocket science? I'd argue it's a waste of time.

If you press me for an answer here, then I would use an economic's perspective and say: educate people until the marginal costs of the education exceed the marginal benefit the person gets from that education (the equilibrium point would be determined largely by g).

Once in a while, there's an octogenarian who gets an education from a university. Is that a waste of resources?

IQ test scores; yes.

But then, IQ is not an inherent, constant factor in humans. It makes no sense to condemn someone with an IQ of 70 in 2008, when he can improve his IQ to 120 in time. The problem is, of course, that he is stuck with that 70 and will be judged subsequently because of that.

 

[Fiona]

I know of only one example where social science measures something and it's not a bell curve: a reaction time distribution.

Could this be because this is one of the few areas where there is an objective measure? What this seems to me to suggest is that in some areas human performance is not normally distributed: I therefore ask again why we assume a normal distribution for "g"? You have conceded it is like that because we make sure it is like that and, as CF Larsen points out, this is circular.

 

[CFLarsen]

Could this be because this is one of the few areas where there is an objective measure? What this seems to me to suggest is that in some areas human performance is not normally distributed: I therefore ask again why we assume a normal distribution for "g"? You have conceded it is like that because we make sure it is like that and, as CF Larsen points out, this is circular.

It definitly isn't. While reaction time can be influenced in various ways *, we can objectively measure it.


* One of the greatest ways of showing people just how much alcohol influences their reaction time is to take a large wooden ruler (one of those 1 meter long ones, used on blackboards, or make a stick and put marks on it) and hold it vertically. The test person must almost grip it at "0 cm". When the ruler is let go, he must grip it when he observes it moving. Record where he grips it (maybe 4-5 cm).

Then, have him drink a beer, a shot or a glass of wine. Wait 15 mins, and do the test again. Repeat: Have a drink, wait 15 mins, do the test.

It always amazes people just how quickly and how much their reaction time increases with alcohol intake!

 

[Fiona]

I am not sure if you are agreeing or disagreeing, Mr Larsen, but I think my point was that there are some things we can definitely measure objectively and some of those, at least, are not normally distributed. I see no reason to suppose that things we cannot measure objectively are normally distributed, a priori.

 

[bpesta22]

Asking why test makers force their test to produce a bell curve is like asking why many of Ford's cars are blue.

The answer is they painted them blue. Why. Because blue cars sell more than any other color.

RT is a ratio scale, and using it could prove that IQ is bell curved. That's why Jensen proposes we develop a standardized batter of ECTs (where the measure of IQ is not number correct, but RT).

 

[bpesta22]

Claus, I do get a higher score with my students than the mean in the population. It's not a huge difference (about 24 versus 22, with a population SD of 7).

Also, personality tests do indeed get at "better" in the sense you use it.

Would you rather be conscientious or not?

Would you rather be agreeable; kind; warm hearted or not?

Would you rather be neurotic or not?

Are you open to experience or are you closed minded?

 

[Fiona]

Asking why test makers force their test to produce a bell curve is like asking why many of Ford's cars are blue.

The answer is they painted them blue. Why. Because blue cars sell more than any other color.

That is my point really. There is no reason to suppose that these things are based on any real differences between people at all. So what I have been asking is for the rationale for painting them blue.

 

[CFLarsen]

Asking why test makers force their test to produce a bell curve is like asking why many of Ford's cars are blue.

The answer is they painted them blue.

But the difference is that IQ proponents use the bell curve to prove that IQ is bell curve'd. Car makers don't paint cars in a specific color because they want that to be the most popular color.

Why. Because blue cars sell more than any other color.

No, they don't.

Yeah. I check.

Claus, I do get a higher score with my students than the mean in the population. It's not a huge difference (about 24 versus 22, with a population SD of 7).

Also, personality tests do indeed get at "better" in the sense you use it.

Would you rather be conscientious or not?

Would you rather be agreeable; kind; warm hearted or not?

Would you rather be neurotic or not?

Would you rather address the many points and questions in post #71?

Are you open to experience or are you closed minded?

Please, don't give me that fatuous blabber. It isn't a question of being open to experience, but being open to evidence.

 

[bpesta22]

But the difference is that IQ proponents use the bell curve to prove that IQ is bell curve'd. Car makers don't paint cars in a specific color because they want that to be the most popular color.



No, they don't.

Yeah. I check.





Would you rather address the many points and questions in post #71?



Please, don't give me that fatuous blabber. It isn't a question of being open to experience, but being open to evidence.

Wow, we are discussing things on two different planes here.

Neither of us seems to be getting the other person's point.

I wasn't attacking you-- those personality traits are all measured in the Big-5, and I was using "you" in the general sense (i.e., some personality traits might be valued as better than others, in the same sense that higher IQ might be valued or deemed better).

Also, I used blue as an example; whatever color it is that sells the most is fine by me. My point is, companies often try to maximize the market for their product. Making an IQ produce bell shaped data does just that.

Gotta work today; more later. Have we at least agreed that IQ tests produce bell curve data (because for practical reasons IQ test makers pick and chose items so that the most frequently occurring score in the population is 100, and that 100 is also the 50% in the population).

If we can agree on that, I will try discussing the theoretical reason for why IQ might be a bell curve later.

If we don't agree on that, let me know the problem.

 

[CFLarsen]

Wow, we are discussing things on two different planes here.

Neither of us seems to be getting the other person's point.

I wasn't attacking you-- those personality traits are all measured in the Big-5, and I was using "you" in the general sense (i.e., some personality traits might be valued as better than others, in the same sense that higher IQ might be valued or deemed better).

Also, I used blue as an example; whatever color it is that sells the most is fine by me. My point is, companies often try to maximize the market for their product. Making an IQ produce bell shaped data does just that.

Gotta work today; more later. Have we at least agreed that IQ tests produce bell curve data (because for practical reasons IQ test makers pick and chose items so that the most frequently occurring score in the population is 100, and that 100 is also the 50% in the population).

If we can agree on that, I will try discussing the theoretical reason for why IQ might be a bell curve later.

If we don't agree on that, let me know the problem.

We don't need to discuss the theoretical reason for why IQ is a bell curve - we already know and agree why: The tests are rigged that way.

Just address the many points and questions in post #71.