[extropy-chat] morons in office

[Patrick Wilken]

On 28 Oct 2004, at 16:22, Brian Lee wrote:

> You're right that IQ tests just test what psychologists think is 
> intelligence. But the difference between 130 and 170 is extremely 
> statistically significant (especially with standard deviation of 10).

It all depends on your measurement devices. Standard IQ tests give 
meaningless results past about 150.

>  It measures your percentile so lets say 130 is in the 99% and 170 is 
> in the 99.999%.

Usually SD is of the population IQ is 15 so 130 is 2 SDs or in the top 
95%, and 145 is in the top 99% (which is about as high standard IQ test 
measure). Keep in mind that one of the big problems with designing test 
for v. high IQs is finding a large enough population (min. 200-300 
people??) to standardize them properly.

> That's a big difference (and there's no real "margin of error" in IQ 
> tests like there are in polls as IQ tests simply compare you to the 
> mean and the sample).

I am not sure I understand what you mean. All IQ are estimates, the 
question is what is the standard error of the estimate. I stated in a 
previous email that this was about 15 for IQ test (I was thinking of 
the WAIS(-R?) in particular). However, this was something I remember 
being taught years ago and I haven't been able to confirm it doing a 
quick Google search. I do remember being surprised by how large the 
range of a typically estimated IQ was, but of course my memory might be 
conflating the SD of IQ of the population with the SE of an individual 
test.

> Of course you can probably give someone 10 tests and get 10 different 
> scores out of them, so IQ tests aren't perfect but they're the best 
> test we can perform to get an idea of a person's intelligence.

Well let's say that someone has an IQ of 150 on the WAIS. They 
obviously are pretty smart at the things the test measures: verbal 
skills and spatial reasoning, but they might be very dumb on other 
things we consider smart. For instance, a musical prodigy like Bach 
might have had an IQ <100 since IQ tests don't measure musical 
intelligence (I am also not sure how well a mathematical prodigy would 
necessarily do on the WAIS - and certainly your physical smarts in bed 
are not measured by standardized IQ tests). Even someone with a high 
general IQ might have a relatively low verbal IQ but high spatial IQ or 
visa versa.

Surely in the end IQ doesn't matter. Its what you do that counts. I 
sometimes wonder if the people who care most about their IQ scores are 
those who haven't actually got so much to show for their lives.

> Also, the younger the test is administered the more accurate. Since an 
> IQ is supposed to test intelligence and not knowledge. A knowledgable 
> 10 year old can really clean up on an IQ test but at 30 could get a 
> much lower score.

IQ tests are age normalized. So even though you do more poorly as you 
age (at least for non-crystalline factors) your estimated IQ is 
adjusted for this decline. So in absolute terms an 50 years old with an 
IQ of 150 is not as smart as a 25 year old with an IQ of 150.

best, patrick

-------------------
Measured Intelligence and Education

WAIS Mean IQ

Educational Equivalent

125
Mean of persons receiving Ph.D. and M.D. degrees

115
Mean of college graduates

105
Mean of high school graduates

100
Average for total population

75
About 50-50 chance of reaching ninth grade

Matarazzo, Joseph D.  Wechsler's Measure and Appraisal of Adult 
Intelligence, 5th Edition.
Oxford University Press, 1972.

--------------------
Best Estimate of IQ Differences for Adults in Different Occupations

WAIS-R Mean IQ Range

Occupational Category

110-112
Professional and technical

103-104
Managers, clerical, sales

100-102
Skilled workers

92-94
Semiskilled workers

87-89
Unskilled workers

Kaufman, Alan S. Assessing Adolescent and Adult Intelligence. Allyn and 
Bacon, 1990.