G x E

4 replies [Last post]
Dorothy Bishop's picture
Offline
Joined: 02/04/2010

I'm trying to understand the analysis used in a paper that has recently created a splash in Science:
Taylor, J., Roehrig, A. D., Hensler, B. S., Connor, C. M., & Schatschneider, C. (2010). Teacher quality moderates the genetic effects on early reading. Science, 328, 512-514.
They have used a moderator variable analysis described by:Purcell, S. (2002). Variance components models for gene-environment interaction in twin analysis. Twin Research, 6, 377-382.
This is in old Mx.
Am trying to track it down, but wondered if there is anything comparable in OpenMx?

neale's picture
Offline
Joined: 07/31/2009
Hi Dorothy Yes, I put a

Hi Dorothy

Yes, I put a script for this together some months ago when we were first testing definition variables. It is attached. You can probably find similar on the Boulder workshop website http://ibgwww.colorado.edu/workshop2010.

It can be a bit dodgy using moderators that may be caused by, rather than cause of, the variable of interest. If the moderator is age, sex, or measured genotype, I'm ok with it. Otherwise I worry.

AttachmentSize
UnivariateTwinAnalysisModerated_Matrix.R 8.82 KB
Steve's picture
Offline
Joined: 07/30/2009
I agree with Mike's

I agree with Mike's cautionary note. Remember that using a definition variable as a moderator makes the assumption that the definition variable was measured without error. This may or may not be the case. If your moderator has error that needs to be accounted for, you are better off with a mixed effects model. These take longer to fit, but can be done in OpenMx.

Dorothy Bishop's picture
Offline
Joined: 02/04/2010
Interesting. The moderator

Interesting. The moderator variable in the Science paper was definitely not measured without error - they estimated 'teacher quality' by taking the amount of improvement in reading shown by a child's classmates (using regression to predict gain while controlling for initial score). I also have some concerns that the classroom a child ends up in might not be independent of their genetic characteristics. But it has made an almighty splash in the media - although greeted with cynicism by some. I picked up the paper via a tweet from a colleague who wondered why Science was publishing a paper showing good teachers are better than bad teachers.....

neale's picture
Offline
Joined: 07/31/2009
It is perhaps time to try

It is perhaps time to try alternative models to the one they used. Moderating the regression may be reasonable when the source only affects one trait, because the factor variance and the regression are confounded, so the moderated regression is a proxy for change in the latent factor variance (heteroscedasticity). However, mechanistically, one might expect either the variance of the latent variable to be increasing, or the sensitivity of the target variable to be increasing. If the latent common factor variance (A C or E) changes then both the target trait and the moderator should show variance increase. If the sensitivity of the target variable is changing then both the common factor and the specific components should show increased regressions, and perhaps by the same amount.

LOL re: the tweet you received. The paper doesn't show this, it says "good genes will out" - once the teacher quality is high, what is left is individual differences in cognitive ability, which are largely genetic in origin. When the teacher quality is low, genetic factors don't get so much of a say in the matter. Possibly, variation in the domestic environment has more impact, although there's not much sign of C or E increasing at the low end of the distribution of teacher quality.