Third thread triggered by yesterday's tweet storm. I've been persuaded to downgrade my "it is almost certain that proportion of variance explained of income will drop from 7%" to "I think it is likely that..." (Ewan waves at friendly geneticists who have hassled him overnight)
This number can go up (better measurement of income, rare variants, which was always out for me) as well as down (broader, less biased ascertainment) + then the whole between society thing walks straight into "how to measure income right". Its...complex
...however...
I think arguing about second significant figures in proportion of <anything> explained by genetics is pretty pointless. As everyone who does this knows, much of this is about ... sample choice, outlier removal/cleaning, phenotype scaling, and covariate usage.
Personally I think genetics on traits and diseases comes in really 4 big buckets
- Predominantly genetic
- Substantial genetics, substantial environment
- moderate genetics, mainly environmental
- no genetics
Examples:
- Predominantly genetic
- Substantial genetics, substantial environment
- moderate genetics, mainly environmental
- no genetics
Examples:
Predominantly genetic: Cystic Fibrosis, Sickle cell disease, Height
Substantial genetics + environment: heart disease, weight, intelligence measures, schzipohrenia
Moderate genetics, mainly environmental: Most cancers, bacterial infections
No genetics: football club support
Substantial genetics + environment: heart disease, weight, intelligence measures, schzipohrenia
Moderate genetics, mainly environmental: Most cancers, bacterial infections
No genetics: football club support
In this "four bucket" view, income for me is in the "moderate genetics, mainly environmental"
