Peter Kabai wrote:

>

> > Mark Macnair wrote:

> >

> > The standard model is additive:

> > > P=A+D+E+G*E+I etc

> >

> > I guess explanation of why it has to be multiplicative is

> > not sufficient. Perhaps someone can explain why it has to

> > be additive.

>

> This basic model is additive, because variances are additive. Quite simple.

THis is backwards. The variances are additive because the model is additive.

> Realistic models are far more complicated than the basic model, mainly

> because the interactive variance cannot be neglected in most cases.

That is true. Even using a simple additive model the "interactive"

variance has to be left out. But the interaction itself has to be

modeled as multiplication also, and it comes out multiplicative even

if the basic linear/additive model is used.

> The interpretation of heritability is also simple: the fraction of genetic

> variance (specifically, the additive variance in the narrow sense of

> heritability) of the total variance in a given population for a given trait

> measured in the given way. So when discussing what heritability really means,

> besides being critical about the methods of measuring the trait, one also

> should keep in mind, that the estimate is restricted to the population in

> which it was measured.

If the true state of IQ measurement is a multiplicative function of E and

G, then the heritability as measured actually is not the contribution of

genes but instead the environment. IT is easy to show, and it can be

found in my paper, the latest version of which I just put at my website.

The calculations are in the conclusion section.

--

Sincerely,

M. Hubey

Dept of Computer Science, Montclair State University

hubeyh@... http://www.csam.montclair.edu/~hubey