From: Rafal Smigrodzki (rms2g@virginia.edu)
Date: Thu Oct 03 2002 - 12:28:02 MDT
Lee Corbin wrote:
> Rafal writes
>
>> ### 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,
>> but for any single person the exact combination of such genes is
>> likely to be different than in their super-smart mate. As a result,
>> the offspring will be heterozygous for many of the traits that made
>> their parents super smart, and the IQ will be on average lower than
>> the parent IQ.
>
> Thanks much for that explanation. I had always supposed
> that it was entirely an environmental effect.
### Oops, here we are in difficult terrain. In the most general formulation,
regression to the mean is a feature of all quantitative traits determined by
a large number of independent influences, whether genetic (recessive,
dominant and codominant) or environmental. It will occur in a population of
genetically identical clones reproducing asexually, in genetically
heterogeneous populations in the most stringently controlled standardized
environment, and in traits whose variance is 100% stochastic. The only case
where it will fail to keep the population homogenous is when there is
assortative mating and/or reproductive isolation of some kind, in other
words, when the requirements for speciation are met. I mentioned recessive
traits because they are easier to illustrate (by the way, thanks for working
out the genotypes), and because from other sources we know that in the
specific case of IQ most of the variance in most modern populations is due
to genetic factors. But see below.
-------
Say one had,
> hypothetically, a collection of giraffes and decided to
> perform some eugenic experiments to maximize neck length.
> We can readily suppose that although (perhaps) predominantly
> determined by genes, environment still can play a significant
> role. Some giraffes, for example, will by luck have a superior
> intra-uterine environment in which to develop, and by chance
> receive extra nutrition or exercise at the most fortuitous moments.
>
> To clinch that argument, suppose that all the giraffes of my
> example were identical twins or clones, and that *all* the
> variation in height was due to environment. One would certainly
> expect, then, that regression towards the mean via my mechanism
> would ensue.
### You are partially right but the case with IQ is more complicated. If
indeed regression to the mean in IQ was caused by environmental factors,
then you would have to conclude that the environment provided by poor
parents increases intelligence (since poor parents are frequently low IQ,
the regression would be upwards), while the environment provided by rich
(and frequently smart) parents results in a loss of IQ. This is not
impossible, but unlikely.
This is why regression to the mean is by itself neither an argument for nor
against any IQ hypotheses - it fits in all models.
-----
>
> Therefore, much as I appreciate your argument, (and as much as it
> is very provocative mathematically), I still contend that regression
> to the mean also obtains from environmental effects.
### Yes, you are right, I did some simplifying in my previous post. I like
the fact that you and Damien caught all that, and called me on it.
Rafal
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:17:24 MST