[extropy-chat] morons in office

David Lubkin extropy at unreasonable.com
Wed Oct 27 23:17:07 UTC 2004


Adrian Tymes wrote:

>Closer to the truth: especially at high levels, your
>IQ primarily measures your skill at taking IQ tests.
>(And if the tests aren't calibrated to accurately
>measure high IQ, then anything above 130 or maybe
>above 145 is statistically the same/within the margin
>of error as most other scores within that range, no?)

I don't want to put the time into a detailed response if no one's 
interested in the subject, but: I know psychometricians who are very 
familiar with IQ research. (If the discussion goes beyond my competence, I 
can relay questions to them.)

What I gather from my own study and what they report --

IQ is actually the best predictor available for some purposes. For example, 
the correlation between poverty and IQ is higher than it is for years of 
education or ethnic group.

IQ tests are each designed for a valid range, and most tests cannot 
accurately measure the intelligence of someone smarter than the author of 
the test. For example, analogy problems are commonly seen. W is to X as Y 
is to ?, with a choice of A, B, C, D, or E. The author is looking for 
answer C but a smarter test-taker chooses D, seeing a connection that the 
author was not aware of. I've suggested that tests be modified to allow the 
subject to justify their answer.

The high-end limit of current tests is roughly 165. There are tests that 
purport to measure higher IQs, but there is no consensus that they can.

On the other hand, there's a new category of test, based on research that 
the time to complete elementary tasks correlates well with IQ, that 
directly measures neural characteristics. To the extent that these can be 
developed and validated, we'd have a culturally neutral, broad-range 
measurement tool.

They'd also be pretty cool for optimizing doses of nootropics and 
quantifying the cognitive impact of environmental factors in your life.


-- David Lubkin.





More information about the extropy-chat mailing list