measurement in science and psychology

["Allen Flanigan." <Allen.Flanigan@USPTO.GOV>]

Art writes,

There is no monolithic "theory of intelligence" built on those assumptions.

Au contraire.  Any claims that psychology (and its branch, psychometrics) is
a "science" are based on the assumption that, as in science, theories are
developed which are used to explain phenomena (in this case, human
behavior).  Whatever the "measurement", be it answers on a standardized test
or locations and magnitudes of bumps on a person's skull, interpretations
drawn from these measurements are necessarily based on some hypothesis or
theory.  If I perform measurements and interpret the results, my
interpretations (or assumptions, to use your phrase) are necessarily based
on some underlying theory which supports my interpretations of the
measurements.  Arguing that my assumptions are not tied to a "monolithic
theory" is ludicrous.

For example, if I hold a balloon in my hand, measure its circumference and
look at a thermometer, I can infer the number of molecules of gas it
contains.  This assumption is based on the kinetic theory of gases.  If
Victor were to point out that there were flaws in the results of my
measurements which didn't jibe with the theory, it would be silly for me to
say, "well, my assumptions about the amount of gas in the balloon (i.e. the
meaning I ascribe to my measurements of temperature)  are not tied to some
monolithic theory of gases".

As for the question of how consistently data matches theoretical
predictions, in real sciences like physics, even one verified instance of a
result which does not match the prediction of the hypothesis is enough to
shoot the hypothesis down and send scientists off in search of a better
hypothesis.  Attempting to minimize aberrant results which don't fit the
theory by saying, well, that's only one or two instances, makes it sound
like practitioners in the field of psychometrics and psychology are willing
to ignore the precepts of science when it is convenient.

We probably never would have gotten to the moon if scientists like Kepler
had clung to accepted theories of planetary motion by rationalizing, "well,
not every measurement of planetary motion fits the theory, but that's not
reason enough to discard the theory".


-----Original Message-----
From: aburke@VANSD.ORG [mailto:aburke@VANSD.ORG]
Sent: Thursday, March 14, 2002 11:26 AM
To: ARN-L@listsrva.CUA.EDU
Subject: Re: Can a 298 IQ be real? No way.


There is no monolithic "theory of intelligence" built on those assumptions.


>>> Victor.Steinbok@VERIZON.NET 03/14 8:06 AM >>>
At 7:55 AM -0800 3/13/02, Art Burke wrote:
>The kid was coached?  No way.  Can't be.  How could anybody even
>suspect that?  Just 'cause his IQ was 298?  I would never have
>guessed.  I am totally surprised.  Oh, I am so naive.  This must be
>my first rodeo.
>
>A couple guys take IQ tests over and over and get different scores
>and that demonstates "conclusively the bankruptcy of measurability
>of intelligence theories."  Now, wasn't that easy?
>
>Art

Art,

It is foolish to base conclusions on a couple of individual
stories... most of the time. There are exceptions. The manner in
which IQ theories are constructed does not allow for variation, so it
is much easier to bring down this house of cards. The general premise
is 1) intelligence is innate and cannot be changed, in principle
(although IQ scores are age dependent and do rise with education
level), and 2) coaching cannot significantly alter the
outcome--progress can only be achieved via long-term education. Any
significant violation of this assumptions, especially one that is
reproducible, would undermine the whole theory.

Although the first point is plausible to some extent--there may well
be some natural foundation to intelligence--the variability is far
greater than any measurement-based theory can account for. The most
identifiable differences are culturally and educationally based and
are far better approached qualitatively than quantitatively--the
measurement error in a qualitative comparison may exceed any value
derived from such a measurement.

        VS-)

--------------------------------------------------------------------------
To unsubscribe from the ARN-L list, send command SIGNOFF ARN-L
to LISTSERV@LISTS.CUA.EDU.

--------------------------------------------------------------------------
To unsubscribe from the ARN-L list, send command SIGNOFF ARN-L
to LISTSERV@LISTS.CUA.EDU.

--------------------------------------------------------------------------
To unsubscribe from the ARN-L list, send command SIGNOFF ARN-L
to LISTSERV@LISTS.CUA.EDU.