Hacker News Reader

Comment by mitchellshow | Hacker News Reader

Cool study, but it is my understanding that twin studies have a tendency to erroneously assume genetic causation since they fail to control for conditions the twins share in prenatal environments, which can (and do) affect behavior.

Some context: http://www.robertsapolskyrocks.com/behavioral-genetics.html

View on HN

They controlled for that by having about as many fraternal twins in this study, though.

It's particularly surprising to me that they got very similar results with regards to discrepancy between fraternal and identical twins when testing again at a later age.

I don't get how that could be a control against assuming genetic causation. Fraternal twins still have similar genetics, right?

Fraternal twins and identical twins both have the same environment growing up, they experience the same hormones at the same time during gestation, they live in the same city, sleep in the same bed. They eat the same foods at the same times of day and in the same amounts, excepting personal preference. However, fraternal twins share 50% of their genes (on average) and identical twins share near 100%.

could it be be that such studies underestimate these effects because of this obvious assumption that fraternal twins share 50% of their genes - IMHO they share more than that because for all genes that "matter" their parents genes would be quite correlated.

Contrary to the popular saying, like attracts like. Parents are exceedingly likely to share visible traits and also mental traits - so the correlation between siblings wouldn't be halfway between random member of population and an identical twin, but much closer to the identical twin.

"Like attracts like" is an interesting assumption, there is some evidence that people are attracted to others that have genetic differences rather than similarities. A common misconception, for example, is that the genetic differences between ethnic groups are larger than genetic differences within an ethnic group, in fact, the reverse is true, if we speak of averages. (Think of two overlapping bell curves with means that are very close to each other, if you want to visualize what I'm trying to say.) Or in more concrete terms, if you select a random person with European ancestors and a random person with African ancestors, the expected genetic similarity is quite close to the expected genetic similarity of two random persons with the same ancestry.

Furthermore it's not actually that critical whether fraternal twins share 50% or 75% or 90%, the important part is that identical twins share nearly 100%, and the number for fraternal twins is significantly less, so if you see differences for fraternal twins that you don't see for identical twins, you can make a strong case that genetic differences explain the observation.

And finally, the idea that fraternal twins share ~50% of their genes is a bit of a simplification. On average, two randomly selected humans will have genetic similarity of 99.5% or so. So that "50%" quote is just a ballpark figure scaled to the relative genetic diversity of the human population to begin with.

Or it depends on how you measure "genetic similarity" anyway.

Yes, and in order to state "this is definitely a genetic cause" you would need to rule out other factors. Including fraternal twins is interesting, but unless they were separated at birth (and not even then really) this could just as easily read "How Kids See the World Depends a Lot on How They Were Raised" which is a pretty obvious conclusion to draw and not particularly headline worthy.

That's exactly what is getting ruled out by using fraternal twins and identical twins - you measure the difference of this effect between these groups.

If the outcome was determined by how they were raised (or where they lived, or who their parents were), the link would be equally strong between fraternal twins and identical twins; but if the outcome depends a lot on genetics, then the effect will be much stronger in identical twins than in fraternal twins. This does rule out pretty much all other factors except genetics.

Thank you for clearly sharing how and why scientists can accurately measure genetic effects in twin studies. You are correct in your assertion that mono vs di - zygotic twin studies are usually good measures of heritability.

I would add a slight technical caveat. Twin studies are excellent at measuring additive genetic effects, i.e. when different genes add together nicely to contribute to a trait (height is a good example). Twin studies typically assume that genes do not interact in funny ways (e.g. epistasis). Thankfully, for many, but not all, traits examined the additive genetic effect catches most of the heritability.

If epistatic interactions drive a trait, however, the statistical models typically used in twin studies would actually under-estimate the influence of genetics.

Further reading: http://www.apa.org/monitor/apr04/second.aspx

In this particular study, identical twins are compared against fraternal twins, which do share the prenatal environment.

Reply on news.ycombinator.com