From: Rafal Smigrodzki (rms2g@virginia.edu)
Date: Thu Oct 03 2002 - 10:41:36 MDT
Damien wrote:
> At 06:26 PM 10/2/02 -0400, Rafal wrote:
>
>> ### Regression to the mean is a natural by-product of recombination
>> of recessive traits - the super-smart are so smart because they have
>> the luck of being homozygous for some IQ-enhancing recessive genes
>
> I believe I can see some problems with this (which should
> smartness-conducive genes all be *recessive*, for example, which you
> seem to imply?).
### Regression to the mean would occur also in a system with exclusively
dominant traits but the illustration is a bit more difficult - you would
need to postulate a larger number of loci, too difficult to put in an
example, sorry I didn't mention it. Yes, you are right, the smartness genes
do not need to be recessive.
>
> More generally, if the IQ population mean is 100, sigma 15, and
> correlation of parent and child is about 0.5, then I suppose the
> argument is that in our world stupid parents of 70 IQ will tend to
> have kids with IQ 100 - 30/2 = 85, and their kids will have kids with
> IQ 92.5, all the way back up to mu = 100... But if the IQ pool of the
> braindrained town is going down due to leaching, I guess you're
> saying the breeding pool mean is no longer 100. Or is it? All those
> alleles lurking in all those people... I'm out of my depth, as you
> see, but interested.
>
### You got it just right - the regression is to the mean of the
*remaining*, not the initial population. E.g. if you remove persons above IQ
100, the IQ of the breeding population will drop to 89.88 (see
http://www.psychstat.smsu.edu/introbook/normal.htm for the calculation
applet), and the regression will be towards 89.88, for both the remaining
parents with IQ 100 and the ones below 89.88.
Rafal
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:17:24 MST