Hi
James M. Clark
Professor of Psychology
204-786-9757
204-774-4134 Fax
[EMAIL PROTECTED]
>>> <[EMAIL PROTECTED]> 25-Mar-08 8:24:27 PM >>>
OK, I seem to be taking a beating on this one, and perhaps I deserve it,
as I have discovered that my thinking on the matter is muddled. Yet in my
own defense, I should say that I asked Mike for examples. I didn't say
there weren't any (although, truth be told, I thought there were only a
few good ones).
Martin's cite is a good one. I've tried to clarify the issue by dipping,
once again, into Judy Harris' book _No Two Alike_. I find there that she
agrees that certain kinds of gene-environment interaction do occur quite
commonly (see around p. 58), as per her example of the child with musical
genes who is born into a musical environment and becomes a Mozart, but a
child without those musical genes born into the same family who grows up
to have no facility for music. Then she mentions the Caspi study,
describing it as showing that a child with genes predisposing to
depression brought up in a stressful environment becomes depressed, but
not a child who lacks those genes.
She calls this a "sensitivity" type of interaction, because there are
always main effects as well--that is children with musical genes become
more musical overall than children without; children with depression
genes become more depressed overall.
She has no problem with the sensitivity type of interaction. Referring to
a talk by a critic of her theory, she says "Maccoby had hit upon the
three areas--IQ, criminal behavior, and mental illness--where reliable
interactions have been demonstrated. None of these interactions, however,
are found in the absence of main effects". (p. 59)
It's the claims for a gene-environment interaction effect in the absence
of a main effect (which, she says, would disconfirm her theory) that get
her goat. These claims are often thrown at her, she says, but when she
goes in search of them, they disappear. She calls them "vaporware". My
muddle comes in in not sufficiently appreciating the difference between
the two types of interaction, at least when I'm left without a minder.
Like Barbie, I have to say "This stuff is hard!"
JC:
I haven't read Harris on this, but I am confused why anyone would expect the
diathesis-stress kind of interaction to occur in the absence of main effects?
The prototypical pattern would be something like following, where ND = no
disorder and D = disorder.
Stress
No Yes
Diathesis No ND ND
Yes ND D
Since the main effects represent effect of each factor aggregated over other
factor, this pattern would necessarily produce main effects, as shown below.
Stress
No Yes Diathesis Main Effect
Diathesis No ND ND (ND+ND)/2
Yes ND D (ND+D)/2
Stress MainEff (ND+ND)/2 (ND+D)/2
The only way you could get no main effects in the presence of interaction would
be following.
Stress
No Yes Diathesis Main Effect
Diathesis No D ND (D+ND)/2
Yes ND D (ND+D)/2
Stress MainEff (D+ND)/2 (ND+D)/2
But no rational person would expect No Diathesis + No Stress to lead to a
disorder.
So on what grounds is this potentially huge class of interactions deemed
irrelevant to Harris's thesis?
Take care
Jim
---
To make changes to your subscription contact:
Bill Southerly ([EMAIL PROTECTED])
