Re: Norms of a IQ-test: percent range (PR) and z-score


Herman Rubin wrote:
In article <1157246214.518067.134480@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
Reef Fish <Large_Nassau_Gr0uper@xxxxxxxxx> wrote:

Herman Rubin wrote:
In article <1157162219.942395.323060@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
Reef Fish <Large_Nassau_Gr0uper@xxxxxxxxx> wrote:

Richard Ulrich wrote:
On 1 Sep 2006 11:08:13 -0400, hrubin@xxxxxxxxxxxxxxxxxxxx (Herman
Rubin) wrote:

In article <1156962274.421938.297290@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,

.................


HR> >> I did not think it was very accurate over about 13 or 14; the
HR> >> original scale was extended by multiple comparisons.

HR> BTW, this was the way many temperature scales were
HR> extended. Also, for scales of length, galactic
HR> distances, and many other.

But that was NOT the way any of the IQ tests are scaled and normed!

But
different IQ tests on her at the same age produced IQ score (on record
at her school) as long <sic for low> as 168.

And how were these IQ tests rated?

Read the Wikipedia webpage I referenced about Marilyn Vos Savant and
her IQ.

Did you?

RF >Of course the notion of an IQ being a QUOTIENT is patently absurd
for
RF >Adults. I used the example of how silly it would be to think that
Einstein
RF >(who IQ was reported to be about 160), when he was 70 years old to
RF >have the intelligence of a 102 year old. :-)

RF >At any rate, the IQ scores a approximately normal (up to about 3
standard
RF >deviation -- thereafter, there is no reliable measure), and the
approximate
RF >normality is accomplished by various methods of NORMING.

HR> >> The normality is accomplished by making it normal. I doubt
HR> >> that normality is valid even to one standard deviation, if
HR> >> a real scale was set up, not using relative frequencies.

That was your unsubstantiated speculation. Different IQ tests have
different methods of NORMING, and they are described in the literature
related to the IQ tests.

RF> >That's why they are NORMED. The cutoff for IQ standards for
RF> >admission to High IQ societies are ALL based on the Normal scale!

HR> >> The central limit theorem for adding up n iid random variables
HR> >> has an error O(1/sqrt(n)), and this requires a huge number to
HR> >> be that small. The various factors are not independent, and
HR> >> some might not even be small, so there is no justification for
HR> >> IQ, or anything similar, to be that close to normal.

RF> >That is NOT the idea used for the Normality of IQ scores. They
RF> >are artificially NORMED.

And as I said before, one can produce a scale by successive
comparison to various standards, but NOT by using quantiles.

But you failed to cite ONE single IQ test that is not based on the
normal
quantiles, or a scale that is convertible to a normal quantile for the
purpose
of admission to High IQ Societies -- about a dozen of them, ALL based
on
quantiles of the NORMAL scale of IQ for admission standards!


Herman, I think you are woefully inadequate in your knowledge about
different ACTUAL IQ tests, how they are scored, or how they are NORMED
to a Normal Scale.

You "theory" of Central Limit Theorem and using the quantile from
samples
are either completely out of date, or fabricated in your imagination
because
you've NEVER taken an IQ test, or read about their construction and
scoring.

I have taken and IQ test, and have read too much about their scoring.

WHICH IQ test? Name it. Then I'll probably know its relation to the

normal scale. I said "probably" because I've not seen ALL the IQ
tests.
Only about a dozen or so.


It is the use of relative scaling which has produced
the abysmal college students which YOU agree should not be there.

But college students are NOT admitted by the IQ scores!

This brings up a different topic, the SAT scores. Those are APTITUDE
scores. They are definitely NOT IQ scores or IQ convertible, simply
because
they are learned aptitude. IQ is a measure of intelligence -- which
does NOT
depend on "aptitude". That is why an adult who cannot read a single
word
of English can be tested on IQ, based on various problems NOT found in
any of the SAT tests.

MENSA and TNS and other high IQ societies have ALLOWED certain
SAT and LSAT and MSAT scores as their ALTERNATIVE admission
standards for the societies.

Almost half of the "tree stumps" I ran across in the classes I taught
have
SAT scores high enough to be admitted to MENSA.

Very few of them ever wanted to spare the testing FEE for taking a
MENSA
IQ test. Of those who did, few of those who qualified for MENSA on
the
basis of SAT scores ever received a high enough IQ score for MENSA
membership.

It's one of these ABUSES of High IQ societies, wanting more MEMBERS so
that they can charge their annual membership FEES, and compromised on
the admission standards on IQ requirements!


There are many situations where normality is unimportant, but
using normality, or converting to normality or approximate
normality, is generally a source of huge errors. In other
newsgroups, it was stated that certain IQ tests could not
get scores reaching 130 because the norming sample was too
small. On a real scale, this would not happen.

That statement is patently FALSE. How else can candidates for
MENSA and TNS or any of the other IQ societies requiring IQ scores
of GREATER than 130, and often substantially greater than 130, if
the certain IQ scores could not reach 130?

The IQ NORMING is NEVER based on a sample that is small.

Few people ever take the Mega Test or wants to be in the Mega Society,
which requires IQ to be in the NORMAL 99.99999999 (not sure exactly
how many nines but can be easily found in IQ webpages) quantile.

Read about HOW Hoeflin normed, and re-normed, and the description
of his SIXTH NORMING of that test in the IQ webpage I cited.

--
This address is for information only. I do not claim that these views
are those of the Statistics Department or of Purdue University.
Herman Rubin, Department of Statistics, Purdue University
hrubin@xxxxxxxxxxxxxxx Phone: (765)494-6054 FAX: (765)494-0558

-- Reef Fish Bob.

.

Relevant Pages

  • Re: Norms of a IQ-test: percent range (PR) and z-score
    ... HR>>> original scale was extended by multiple comparisons. ... at her school) as long as 168. ... RF>normality is accomplished by various methods of NORMING. ... Or some kind of high school Aptitude test? ...
    (sci.stat.math)
  • Re: Norms of a IQ-test: percent range (PR) and z-score
    ... RF>normality is accomplished by various methods of NORMING. ... HR>>> a real scale was set up, ... RF>>admission to High IQ societies are ALL based on the Normal scale! ...
    (sci.stat.math)
  • Re: Norms of a IQ-test: percent range (PR) and z-score
    ... HR>>> original scale was extended by multiple comparisons. ... RF>normality is accomplished by various methods of NORMING. ... comparison to various standards, ... and my mechanical "aptitude" score ...
    (sci.stat.math)
  • Re: Norms of a IQ-test: percent range (PR) and z-score
    ... original scale was extended by multiple comparisons. ... normality is accomplished by various methods of NORMING. ... a real scale was set up, ... Herman Rubin, Department of Statistics, Purdue University ...
    (sci.stat.math)
  • Re: Steam cars
    ... I think of slurries, too, as one of those things that scale up better ... sorts of fuels, especially if you're in no hurry; ... forks into "non petrochemical ways to power something resembling what ... The problem with the latter scenario is that those societies had ...
    (rec.autos.tech)