IQ Tests

[Beth]

It is correlation that we don't see any other possible factors to explain.

A nice succinct explanation. Consider the example of the price of rum and clerical salaries. There is likely a factor that can explain both, presumably some sort of overall economic indicator. It seems to be that g, being an attempt to measure some sort of overall intelligence indicator is more analogous to the overall economic indicator used to predict clerical salaries. Bpesta making a lot of sense to me in this thread.

 

[benji]

I'm saying I've seen no evidence that it's real. Personally, I believe that it's not real because I've seen no evidence that it's real, and it ends up getting filed in the same drawer as dragons, unicorns, and the Easter Bunny.

No, you decide to ignore the evidence. The evidence is the same as for "gravitation". You see its effects (the correlation of effects with the existence of gravitation), you don't have a better explanation based on already known constructs, and you don't find any contradicting evidence. The same goes for "g". Both are constructs.

Factor analysis can certainly find real things. But it's also well-known as being able to "find" things that don't really exist.

True.

Hume answered this one long ago -- "correlation whenever we check" doesn't work. The cock's crowing does not cause the sun to rise no matter how many times we check. There are a lot of formal philosophical definitions of "causation," none of which are entirely satisfactory.... but they're a lot better than naive "correlation."

Of course you could come up with a theory that says that the rum prices in havanna influence the salaries of ministers in London. However, I would probably be able to come up with a better theory including inflation at some point. Let's use Ockham's Razor here: If you find evidence for a Factor that explains all the above (by bpest22) mentioned findings and some more by an "external" explanation (meaning that the differences are not to a substantial part based on individual abilities), then the g-model is dead. Until then, I am accepting it.

There's an obvious, if somewhat facetious, analogy to the g-model of "intelligence." Perhaps what we measure as IQ is in fact a failure of mental process; the actual human mind/brain is ideally capable of infinite performance, but various aspects limit it. Perhaps we should be measuring a "stupidity quotient" and searching for ways, not to enhance intelligence, but to limit (aspects of) stupidity.

I don't think you understand what "g" really means. "g" means "general factor", and it's just that. A strong first factor invariably found by analyzing the result of different tasks that are supposed to measure facetts of intelligence. By not accepting g as a predictieve variable, you would have to argue why this factor shares substantial variance with test-tasks, but doesn't so with "real-life" tasks that are similar to those test-tasks.

The g-factor-model doesn't say anything about how "g" (or "intelligence") "looks" or "tastes" in reality. If you want define IQ as a measure of failing or not failing mental processes - fine. As long as you aknowledge that some people fail more often than others (wich we call having "a lower IQ"), that's completely consistent with the g-factor- model. Other theories postulate different conductivities of nerves.

Similarly, perhaps the neurologists are wrong (and the esoteric idealists were right) all along, and "intelligence" is a dualistic process tapping like a radio receiver into the "mind of God," where IQ is simply a measure of how little interference is present.

Again - if some people are consistently better at "tapping into the mind of god", that woul dbe perfectly consistent with the g.factor model. You are not presenting alternatives to g, but other concepts if "intelligence".

Either of these cases (which I admit I find improbable) would suggest that the g-factor model is accurate, but ontologically incorrect.

Not at all. See above.

Similarly, there may be some process ("oxidation") that produces behavior ("heat"). This is one of the central questions behind much theoretical IQ work -- does the g-factor model measure the process capacity for intelligence, or does it measure the behavioral results of the processs? But of course, if the g-factor itself does not exist, then the question itself reveals a fundamental misapprehension caused by the reification of 'g.'

How can you question the existence of the "g-factor"? It is a statistical reality. This has nothing to do with correlations to external data at all.

But to cut directly to the chase, what test could you perform to falsify the first hypothesis above in favor of a 'real' g-factor model?

To falsify the "g-factor-model", you would have to run FA's on data that is known to measure facetts of intelligence, and repeatedly NOT find a strong first factor. That's all.

 

[drkitten]

How can you question the existence of the "g-factor"? It is a statistical reality.

I don't question its existence at all. I question whether the g-factor itself describes anything real, or whether it's just a "statistical artifact."

Which gets back to my original question to bpesta -- have you any evidence at all that 'g' is not purely a statistical artifact?

In my considered opinion, 'g' is the modern-day psychological equivalent of 'phlogiston.'

 

[mumblethrax]

In my considered opinion, 'g' is the modern-day psychological equivalent of 'phlogiston.'

And what of Mendel's gene or Dalton's atom? It seems we can only pillory theoretical constructs with the benefit of hindsight.

Phlogiston strikes me as an excellent example of a model that was rejected precisely because it failed to accurately predict experimental results. Of course, it was replaced by another mistaken theory of substance of heat which works better as an analogy here, as it did have considerable predictive power.

But as I understand it, psychometricians are not attempting to identify some substance g but to view g as a property, of brain or otherwise. It seems ungenerous (and damaging to science) to a priori reject this model before we have the observational power to falsify it.

 

[Ducky]

Allow me to introduce myself. My name is Wile E. Coyote, Super-genius.

 

[WanderinWTF]

i want to take a IQ test can anyone help me do this

 

[drkitten]

Phlogiston strikes me as an excellent example of a model that was rejected precisely because it failed to accurately predict experimental results.

Not quite. If you look at the historical record, people looked at the expected properties of phlogiston and tried to run experiments on them. (Heck, the original report of the discovery of oxygen reported it as "dephlogistinated air." Of course, if you remove negative oxygen, you're actually making an oxygen-enriched mixture.)

But no one's even trying to devise a test to see if 'g' is merely a statistical artifact or not.

Or if they are, I've never read about it.

But as I understand it, psychometricians are not attempting to identify some substance g but to view g as a property, of brain or otherwise. It seems ungenerous (and damaging to science) to a priori reject this model before we have the observational power to falsify it.

It seems unparsimonious (and equally damaging to science) to a priori accept the model before we have any observations to support it.

 

[ynot]

The more I do IQ tests, the higher I score. Does this mean IQ tests make me more intelligent? Or is it that not doing them makes me dumber?

Didn't Marilyn Monroe have a higher IQ than Al Einstein.

 

[Hindmost]

i want to take a IQ test can anyone help me do this

Hi,

try the following link for a reasonable test...you typicallyhave to submit info on yourself at some point--and give an email address to send the results.

http://www.mensa.org/index0.php?page=12

A long time ago I took a British mensa test online...it was a reasonable test for an IQ test. I can't seem to find a link that has an online variety. They have a pdf variety

http://www.mensa.org.uk/


There are a variety of all sorts of tests on the following site.

http://www.2h.com/iq-tests.html


always remember: Standard IQ tests do not measure ambition, imagination, artistic ability etc.

glenn

 

[Hindmost]

Allow me to introduce myself. My name is Wile E. Coyote, Super-genius.

And yet he never learned how to order ACME take-out.

glenn

 

[bpesta22]

I don't question its existence at all. I question whether the g-factor itself describes anything real, or whether it's just a "statistical artifact."

Which gets back to my original question to bpesta -- have you any evidence at all that 'g' is not purely a statistical artifact?

In my considered opinion, 'g' is the modern-day psychological equivalent of 'phlogiston.'

I think perhaps we differ on what we'd accept as evidence. What types of evidence would prove in your mind that g is real? Obviously, just showing that it correlates with some things but not others, doesn't seem good enough.

So, could you give me an example that would do it for you:

"If we could show that g does what...then that would be evidence that g is real."

Complete the sentence!

***

Btw, I pulled out Jensen's The g Factor-- the culmination of 40 years of research for him when he wrote it.

He spends a fair amount of time talking about the reality of g (p. 545 on), and how one can get at cause and effect using techniques other than random assignment (which is impossible here, as with many important variables social scientists study).

I'm not sure it's worth reading a 600 page book to settle an internet debate, but if one were to read it, one might get the impression that the level of scientific sophistication in this area goes well beyond the basic intro to r&m issues we're discussing here.

Here's what he says about g in his forward:

"I have come to view g as one of the most central phenomena in all of behavioral science, with broad explanatory powers....the g construct extends beyond its psychometric origin and definition.

The g factor is actually a biologically based variable, which is....necessarily a product of the evolutionary process. The human condition cannot be adequately described without taking into account the powerful explanatory role of g...students...need to grasp the essential psychometric meaning of g, its basis in genetics and brain physiology, and its broad social significance."

The argument for why Jensen devoted his life to studying g-- why he believes it's real-- then goes on for 600 pages.

**

Ironically, a 600 page state of the art text by the man in IQ research makes absolutely no mention of Gould or Mismeasure.

Rightfully so, the science has ignored mismeasure as being irrelevant (I've argued before that the worst thing that can happen to a scientific "contribution" is not that scientists attack it-- as they did with the Bell Curve, but that scientists ignore it-- as they did with Gould).

 

[Mercutio]

Jensen...is that 1999? Just for the sake of accuracy, for those following this thread?

 

[Melendwyr]

IQ tests do not measure intelligence. They measure... well, whatever it is that they measure.

At a rough guess, IQ tests are pretty good at measuring the capacity for manipulating minor abstractions. The better you are at drawing conclusions from simple concepts, the higher your IQ will be.

What they don't measure is your ability to integrate those conclusions in complex ways, inhibit existing patterns of learned behavior so to approach things from a 'fresh' perspective, and the capacity for keeping passions from screwing up reasoning.

IQ scores generally are not affected by damage to the frontal lobes - in fact, they survived lobotomies with little if any change. It takes very sophisticated testing that goes far beyond the simple problems poses on an IQ test to even begin to detect the most subtle aspects of thought.

People with frontal lobe damage have severe problems setting and achieving goals, at least partially because they've lost the ability to associate positive and negative emotional responses with complex intellectual concepts. They tend to respond to situations by rote, and have immense difficulty learning new responses to situtations they had learned to deal with before the damage had taken place. They often have difficulty with inhibition, acting on immediate instinct without regard for the complexities of the environment, and they frequently cannot plan ahead.

None of those aspects of mental function are measured by IQ tests. IQ is correlated with success, yes, but it doesn't do to become overly concerned with it. It's not everything. It's not even the greatest part of mental ability, and it certainly doesn't determine to what degree thinking capacity is used usefully in everyday conditions - there are many fools with high IQs that don't apply their mind well to real problems.

And this is from a person whose IQ is probably higher than yours. Believe me, I'm living proof of the limitations of IQ testing.

 

[bpesta22]

Jensen...is that 1999? Just for the sake of accuracy, for those following this thread?

Yeah, so kinda dated now, but I meant state of the art when he wrote it.

 

[benji]

I don't question its existence at all. I question whether the g-factor itself describes anything real, or whether it's just a "statistical artifact."

Which gets back to my original question to bpesta -- have you any evidence at all that 'g' is not purely a statistical artifact?

In my considered opinion, 'g' is the modern-day psychological equivalent of 'phlogiston.'

Again: How do you explain the high correlations between different tasks that are shown to measure facettes of the concept "intelligence"?
Remember: an alternative source for that phenomenon would have to lie "outside" of the subjects and it would need to show stability (from 0.8 upwards), otherwise you are just making assumptions of what "g" could actualy be.

And if you don't find a better explanation for the "experimental" situation above and hence some base level of intelligence can be assumed to affect all tasks in intelligence testing to a certain point, please explain why this shouldn't also be the reason for the correlations found with "real-world" data.

 

[benji]

IQ tests do not measure intelligence. They measure... well, whatever it is that they measure.

No, they measure certain facettes of intelligence. To know which test to use (as a psychologist), you have to know which facettes you are interested in.

At a rough guess, IQ tests are pretty good at measuring the capacity for manipulating minor abstractions. The better you are at drawing conclusions from simple concepts, the higher your IQ will be.

Of course, but why limit it to "simple" concepts? That depends very much on the test. There ARE tests for the higher range of intelligence, and for the lower range as well.

What they don't measure is your ability to integrate those conclusions in complex ways, inhibit existing patterns of learned behavior so to approach things from a 'fresh' perspective, and the capacity for keeping passions from screwing up reasoning.

The last part wouldn't be the goal of intelligence testing, but a question of personality assessment. But of course that depends on your definition of "intelligence". Yours seems to be rather wide.

Your mention of a "fresh perspective" is actually very interesting. It has been shown that schizophrenics and geniuses share this ability. After all, making new connections and having fresh perspectives is the basis of innovation. The difference is: geniuses also know about the "coventional" perspective, while schizophernic do not.
And yes, there are tests which measure exactly this ability.

IQ scores generally are not affected by damage to the frontal lobes - in fact, they survived lobotomies with little if any change. It takes very sophisticated testing that goes far beyond the simple problems poses on an IQ test to even begin to detect the most subtle aspects of thought.

Again: the "simplicity of problems" depends on the test you are using.

People with frontal lobe damage have severe problems setting and achieving goals, at least partially because they've lost the ability to associate positive and negative emotional responses with complex intellectual concepts. They tend to respond to situations by rote, and have immense difficulty learning new responses to situtations they had learned to deal with before the damage had taken place. They often have difficulty with inhibition, acting on immediate instinct without regard for the complexities of the environment, and they frequently cannot plan ahead.

So it all comes down again to the definitino of intelligence.

None of those aspects of mental function are measured by IQ tests.

No, because most of them do not belong in the traditional "intelligence" concept. They can be measured using clinical tests though...

IQ is correlated with success, yes, but it doesn't do to become overly concerned with it. It's not everything.

No single factor is "everything". It's just one part of the puzzle.

It's not even the greatest part of mental ability,

That again depends on your definition. If you include everything you included above, it probably is a pretty large chunk. But of course there are other "mental abilities" that are usually not included in "traditional" IQ-testing like attentiveness, concentration, retentiveness, problem solving, interest-structures, creative thinking, mechanical and technical understanding, ortography, spatial sense and many others. Alll of them can be tested.

and it certainly doesn't determine to what degree thinking capacity is used usefully in everyday conditions - there are many fools with high IQs that don't apply their mind well to real problems.

True, but see above. IQ-testing does not claim to tell you everything about a person. We are talking about statistical probabilities, not particular cases. There are also many intelligent people with high IQ's that do apply their mind well to real problems. On the other hand, there are probably not too many people with low "IQs" who do extremely well (although I know of at least one, but let's not get into politics here ;o) ). The correlation is never 1.0!

 

[benji]

The more I do IQ tests, the higher I score. Does this mean IQ tests make me more intelligent?

If by training them you actually learn how to calculate faster and more accurately, or you enhance your vocabulary, then yes, you probably enhance those facettes of your intelligence that are part of what is also called "crystalline intelligence". On the other hand - "fluid intelligence" (e.g. problem solving) is a lot harder to train. But in any case, it mostly means that you learn from the tasks and get better at them. This effect is extremely strong if you do the same test more than once, it may exist when doing different tests with similar tasks, and it will probably not exist at all if presented with completely unknown tasks. This only shows that there are other influences on a test-result than just your intelligence. Most if not all psychological concepts can not be measured with absolute precision. That does not mean they are not existing, just that the measurement is not perfect. It's always an estimation of the real value. Comparing a measured IQ of 120 with one of 121 and saying one was higher than the other is absurd. Comparing a result of 100 with one of 130 is justified. The critical difference is a property of the test used, and is a function of its statistical reliability.

 

[drkitten]

I think perhaps we differ on what we'd accept as evidence. What types of evidence would prove in your mind that g is real? Obviously, just showing that it correlates with some things but not others, doesn't seem good enough.

So, could you give me an example that would do it for you:

"If we could show that g does what...then that would be evidence that g is real."

Complete the sentence!

I'd be delighted to. My primary complaint is that no one has been able even to define 'g' other than as a statistical artifact.

So, "if we could show that 'g' has a meaningful definition other than as the outcome of a statistical process, that would be evidence that 'g' is real."

"I have come to view g as one of the most central phenomena in all of behavioral science, with broad explanatory powers....the g construct extends beyond its psychometric origin and definition.

The g factor is actually a biologically based variable, which is....necessarily a product of the evolutionary process. The human condition cannot be adequately described without taking into account the powerful explanatory role of g."

Yawn. I told you that I had seen no evidence that Jensen understood statistics. I rejected the idea that "g must be real because it has explanatory capacity" a page and a half ago.

 

[benji]

Yawn. I told you that I had seen no evidence that Jensen understood statistics. I rejected the idea that "g must be real because it has explanatory capacity" a page and a half ago.

Again (since you continue to ignore it): Prove that gravitation is real without using its "explanatory capacity".

 

[drkitten]

Again (since you continue to ignore it): Prove that gravitation is real without using its "explanatory capacity".

Yawn. I continue to ignore your irrelevant straw men.