"Mark D. Morin" <mdmpsyd at NOSPAMgwi.net> wrote in message
news:3D2C0DBC.5030508 at NOSPAMgwi.net...
> John Knight wrote:
> > "Mark D. Morin" <mdmpsyd at NOSPAMgwi.net> wrote in message
> > news:3D2BA223.4020605 at NOSPAMgwi.net...> >
> >>>>>Most of the corrlations I've gotten are in the range of 0.6, so it
> >>>>
> > would
> >
> >>>>>really be appreciated if you could provide a reference to the above.
> >>>>
> >>>>0.6 is not a low correlation--it explains over one third of the
> >>>>variation in scores. In any other field, an R of this size would be
> >>>>considered robust.
> >>>>
> >>>>my resources are at the office, I'll dig them out today.
> >>>>
> >>>
> >>>
> >>>Yes, 0.6 really is good correlation, but when compared to the 0.8795
> >>>correlation between brain size and GRE Quantitative, you have to wonder
> >>>what's missing from "IQ tests".
> >>
> >>and there still isn't a reliable source for this statistic.
> >>
> >
> >
> >
> > For which statistic? Are you questioning Philippe Rushton's
measurements of
> > brain size,
>> yes
>
Why?
> > GRE Quantitative Scores, or the method for calculating
> > r-squared?
>> yes
> note the measurement error reported.
>
hmm, you may be getting "standard deviation" and measurement error mixed up.
There are standard deviations reported with GRE scores, but not "measurement
errors". Two thirds of the test takers in each group score within + or - of
the standard deviation from the median score. e.g., the median score for
Black women was 408 with a standard deviation of 110, which means that two
thirds of Black women scored between 298 and 518. But the standard
deviation for Asian men, whose median score was 530, was 131, so two thirds
of them scored between 399 and 661. Because of their larger standard
deviation, the highest scoring Asian men scored 143 points higher than the
highest scoring Black women, compared to a difference of only 122 points in
their median scores.
This is the difference between developing the world's most competitive
semiconductor companies and not being able to build a wheel.
>> >
> > Run the data at http://christianparty.net/grebrainsize.htm yourself.
>> There are no data there that are "runnable."
>> Or use
> > the following figures and see what you get for r-squared. The first
column
> > of numbers is brain size in cubic centimeters,
>> which presumes accuracy of measurement. with the reported variability
> (measurement error), by definition, those numbers are not valid--for
> something to be valid, they first need to be reliable. If brain sizes
> can not be measured reliably, your numbers are meaningless.
>>
You're right. If brain sizes can't be measured accurately, then Philippe
Rushtons' numbers are meaningless.
But why would you presume that the nation which managed to complete the
calculations necessary to fire a rocket at the moon, land on it, and come
back, can't measure something as mundane as an empty skull?
Whay you may be saying is that you believe that Professor Rushton LIED about
his figures, but why would he do that? There are plenty of techniques, all
the way from filling it with known sized pebbles, to MRI, so plenty of
people could refute his work if he tried to LIE, don't you think?
And as you've already acknowledged, getting an r-squared of 0.8795 is VERY
good correlation, which suggests that both GRE and brain sizes ARE accurate.
http://christianparty.net/greracesexyear.htm
John Knight
