In the GWAS on educational fields, the authors said “We note that it is not recommended to base heritability estimates on the former approach (GWASs by subtraction), because environmental variance is undefined.” Does that mean we should not report the heritability from the multivariate GWAS from GenomicSEM?
Replying to “In the GWAS on educational fields, the authors sai…”:
Correct! (Even though it’s tempting… the number isn’t interpretable)
Replying to “In the GWAS on educational fields, the authors sai…”:
From what I recall this is specific to this analysis (trying to control educational fields for EA)
Replying to “In the GWAS on educational fields, the authors sai…”:
… Given the different tutor answers, we should ask Andrew his thoughts!
Replying to “In the GWAS on educational fields, the authors sai…”:
I agree for asking 
Interpretability of heritability is always a tricky one